adverse nephrotoxic events in critically ill children: a ... · ii adverse nephrotoxic events in...
TRANSCRIPT
Adverse Nephrotoxic Events in Critically Ill Children: A Pharmacoepidemiologic Evaluation
by
Morgan Brooke Slater
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Institute of Medical Science University of Toronto
© Copyright by Morgan Brooke Slater 2014
ii
Adverse Nephrotoxic Events in Critically Ill Children: A
Pharmacoepidemiologic Evaluation
Morgan Brooke Slater
Doctor of Philosophy
Institute of Medical Science University of Toronto
2014
Abstract
Background: Kidney injury is a common adverse event in critically ill children and has a
complex and multifactorial etiology. Some factors, such as age and sex, are non-modifiable
while others, including medication exposure, are modifiable and present an opportunity to
decrease the risk of kidney injury. The objective of this dissertation was to quantify the risk
of acute kidney injury (AKI) attributable to the initiation of nephrotoxic medications among
critically ill children.
Methods: We utilized a pharmacoepidemiologic approach, using AKI as a model for
studying adverse drug events (ADEs) in children. First, we conducted a systematic review of
the application of an accepted definition of AKI (the RIFLE criteria) in children. Second, we
estimated the incidence and associated factors of AKI in a 3.5-year cohort of critically ill
children. Finally, we conducted a nested case-control study to quantify the risk of developing
AKI after the initiation of one or more nephrotoxic medications.
Results: Our systematic review identified wide variation in the application of the RIFLE.
We found an AKI incidence of 23.7% in our patient population and identified patients at
iii
increased risk for the development of AKI. Administration of nephrotoxic medications
during ICU admission was the single greatest risk factor for AKI. We found a high incidence
of nephrotoxic medication use in the ICU; over 80% of patients were exposed to one or more
nephrotoxic medications. The administration of ganciclovir, furosemide and gentamicin were
significantly associated with increased odds of developing AKI before and after adjustment
for underlying differences in risk factors of AKI.
Conclusion: Acute kidney injury is common among critically ill children. Nephrotoxic
medication exposure preceding AKI is common. The results of this work can help to
optimize prescribing and aid clinicians in making optimal treatment choices, reducing harm
and improving patient outcomes.
iv
Acknowledgments
I would like to extend many thanks and appreciate to all those who have encouraged and
supported me during the course of my graduate studies. First, to my supervisor, Dr.
Christopher Parshuram: thank you for your continued support and guidance. My committee
members, Dr. Andrew Howard, Dr. Gideon Koren and Dr. Paula Rochon, were instrumental
in my research and I thank them for their guidance, insight, encouragement and support
during this long process. Andrew Howard is owed a special mention: he hired me after I
completed my MSc degree and gave me my first opportunity in clinical research. Along the
years he has always been supportive of my career and I am fortunate to call him a mentor.
Dr. Andrea Gruneir, an informal member of my thesis committee, provided much wisdom,
methodological expertise and insight into the arduous process of completing a doctoral
degree.
Thanks to Dr. Parshuram’s research group for their assistance along the way, especially
Jessica Grillo, Kate Byrne, Helena Frndova, Tracy Peggie and Christina Stevancec. You all
lent an ear or a shoulder when I needed it – thank you.
I am fortunate to call Dr. Valeria Rac both a friend and a mentor. Her time, support,
advocacy and expertise have been priceless and her love and enthusiasm for her work is an
example for me. She has been unwavering in her support and without her, this thesis would
not have been possible. She is a brilliant scientist and I can only aspire to be half as talented
as she is.
During my career, I have been involved with a number of research groups across the
spectrum of health care priorities. All of these experiences have helped to shape how I think
v
about research. Two people stand out to me: I owe special thanks to both Dr. Dorcas Beaton
and Dr. Rick Glazier. Both are amazing people and dedicated researchers who have helped
shaped my career goals.
Above all, I am fortunate to have family and friends who have listened to me throughout this
long, long process. To you all, words cannot express my thanks for your love, support and
patience.
I would also like to thank the following scholarship programs for supporting my thesis work:
Canadian Institutes of Health Research Frederick Banting and Charles Best Canada Graduate
Scholarship, the Ontario Student Opportunity Trust Fund – The Hospital for Sick Children
Foundation Student Scholarship Program, and Ontario Graduate Scholarship.
Finally, this thesis is dedicated to my Nono, Aldo Narduzzi, who instilled the importance of
education and life-long learning in me from a very young age. I hope he is proud.
vi
Contributions
I, Morgan Slater, solely prepared this thesis and am the first author of the three manuscripts
directly resulting from this thesis work. All aspects of this work, including the planning,
execution, analysis and writing of all original research and publications, were performed by
the primary author. The following contributions by other individuals are formally
acknowledged:
Dr. Christopher Parshuram, Dr. Andrew Howard, Dr. Gideon Koren, Dr. Paula Rochon and
Dr. Andrea Gruneir all provided mentorship, guidance and assistance in the planning and
execution of the analyses and manuscript/thesis preparation.
Dr. Vijay Anand was the second reviewer for the systematic review (Chapter 3). He also
assisted with the editing of the resulting manuscript.
Elizabeth Uleryk provided expertise and guidance in conducting the literature search for the
systematic review (Chapter 3). She also assisted with the editing of the resulting manuscript.
Dr. Christopher Parshuram, Dr. Gideon Koren, Winnie Seto and Angela Trope provided their
clinical expertise in the Delphi panel selection of defining nephrotoxic medications, utilized
in both Chapters 4 and 5.
An abridged version of Chapter 3 has been previously published. The citation is: Slater MB,
Anand V, Uleryk EM, Parshuram CS. A systematic review of RIFLE criteria in children and
its application and association with measures of mortality and morbidity. Kidney
International 2012; 81(8): 791-798. Permission was received to publish in this dissertation.
vii
Table of Contents Acknowledgments..........................................................................................................................................iv!
Contributions...................................................................................................................................................vi!
Table.of.Contents..........................................................................................................................................vii!
List.of.Tables....................................................................................................................................................xi!
List.of.Figures................................................................................................................................................xiii!
List.of.Abbreviations....................................................................................................................................xv!
Chapter.1! Overview.....................................................................................................................................1!
1.1! Preamble*................................................................................................................................................................*2!
1.2! Research*Objectives*...........................................................................................................................................*3!
1.3! Thesis*Organization*..........................................................................................................................................*3!
Chapter.2! Acute.Kidney.Injury,.Adverse.Drug.Events.and.Nephrotoxicity..............................5!
2.1! Overview*.................................................................................................................................................................*6!
2.2! Acute*Kidney*Injury*...........................................................................................................................................*6!
2.2.1! Etiology!of!AKI!...........................................................................................................................................!9!
2.2.2! Incidence!of!AKI!.....................................................................................................................................!11!
2.2.3! Outcomes!of!AKI!....................................................................................................................................!12!
2.2.4! Defining!AKI!............................................................................................................................................!15!
2.2.4.1! Measuring!Kidney!Function!...................................................................................................................!15!
2.2.4.1.1! Biomarkers!..............................................................................................................................................!16!
2.2.4.2! Current!Definitions!of!AKI!.......................................................................................................................!18!
2.3! Adverse*Drug*Events*......................................................................................................................................*25!
2.3.1! At!Risk!Populations:!Critical!Illness!and!Children!..................................................................!28!
2.3.2! Common!Adverse!Drug!Events!in!Paediatric!Populations!..................................................!31!
2.3.3! Importance!of!Understanding!and!Identifying!the!Contribution!of!Medications!to!
Adverse!Events!......................................................................................................................................................!34!
2.3.4! Drug!Toxicity!as!a!Cause!of!AKI!......................................................................................................!36!
2.3.4.1! Factors!Associated!with!Increased!Toxicity!....................................................................................!36!
2.3.4.1.1! PatientQSpecific!Factors!......................................................................................................................!36!
2.3.4.1.2! KidneyQSpecific!Factors!......................................................................................................................!37!
2.3.4.1.3! DrugQSpecific!Factors!..........................................................................................................................!37!
2.3.4.2! Incidence!of!Drug!Toxicity!as!a!Cause!of!AKI!..................................................................................!40!
viii
2.4! Studying*Drug*Safety*in*Children*.............................................................................................................*42!
2.4.1! Studying!Acute!Kidney!Injury!as!an!Adverse!Drug!Event:!Pharmacoepidemiology!44!
Chapter.3! A.Systematic.Review.of.RIFLE.Criteria.in.Children.and.Its.Application.and.
Association.with.Measures.of.Mortality.and.Morbidity..................................................................46!
3.1! Overview*..............................................................................................................................................................*47!
3.2! Introduction*.......................................................................................................................................................*48!
3.3! Methods*...............................................................................................................................................................*50!
3.4! Results*..................................................................................................................................................................*52!
3.4.1! Application!of!the!RIFLE!....................................................................................................................!57!
3.4.1.1! Missing!Data!..................................................................................................................................................!58!
3.4.1.2! GFR!Criteria!...................................................................................................................................................!61!
3.4.1.3! Urine!Output!Criteria!................................................................................................................................!62!
3.4.2! Associations!Between!the!RIFLE!and!Measures!of!Mortality!and!Morbidity!.............!62!
3.4.2.1! Mortality!.........................................................................................................................................................!63!
3.4.2.2! Length!of!Stay!(LOS)!..................................................................................................................................!63!
3.4.2.3! Illness!Severity!.............................................................................................................................................!64!
3.4.2.4! Measures!of!Kidney!Function!................................................................................................................!66!
3.4.2.5! Risk!of!Kidney!Injury!.................................................................................................................................!66!
3.5! Discussion*...........................................................................................................................................................*67!
Chapter.4! Risk.Factors.of.Acute.Kidney.Injury.in.Critically.Ill.Children................................72!
4.1! Overview*..............................................................................................................................................................*73!
4.2! Introduction*.......................................................................................................................................................*73!
4.3! Methods*...............................................................................................................................................................*75!
4.3.1! Identification!of!Nephrotoxic!Medications!................................................................................!77!
4.4! Results*..................................................................................................................................................................*81!
4.5! Discussion*...........................................................................................................................................................*96!
4.5.1! Limitations!...............................................................................................................................................!99!
4.6! Conclusions*.......................................................................................................................................................*100!
Chapter.5! Acute.Kidney.Injury.and.Exposure.to.Nephrotoxic.Medications.in.Critically.Ill.
Children:.A.Nested.CaseRControl.Study...............................................................................................102!
5.1! Overview*............................................................................................................................................................*103!
5.2! Introduction*.....................................................................................................................................................*103!
5.3! Methods*.............................................................................................................................................................*105!
5.3.1! Case!Definition!....................................................................................................................................!105!
ix
5.3.2! Identification!of!Controls!................................................................................................................!106!
5.3.3! Nephrotoxic!Drug!Exposure!..........................................................................................................!107!
5.3.4! Covariates!..............................................................................................................................................!108!
5.3.5! Data!Analysis!.......................................................................................................................................!109!
5.4! Results*................................................................................................................................................................*110!
5.5! Discussion*.........................................................................................................................................................*116!
5.5.1! Strengths!and!Limitations!..............................................................................................................!118!
5.6! Conclusions*.......................................................................................................................................................*119!
Chapter.6! General.Discussion.............................................................................................................121!
6.1! Overview*............................................................................................................................................................*122!
6.2! Summary*of*Major*Findings*.....................................................................................................................*123!
6.2.1! Summary!by!Thesis!Objectives!....................................................................................................!123!
6.2.1.1! Systematic!Review!of!the!Use!of!RIFLE!Definition!of!AKI!......................................................!123!
6.2.1.2! Determining!the!Clinical!Risk!Factors!of!AKI!...............................................................................!124!
6.2.1.3! Evaluation!of!the!Contribution!of!Drug!Therapy!to!Acute!Kidney!Injury!.......................!125!
6.2.2! Five!Main!Findings!............................................................................................................................!125!
6.3! Strengths*and*Limitations*.........................................................................................................................*127!
6.3.1! Strengths!................................................................................................................................................!127!
6.3.1.1! Data!Sources!...............................................................................................................................................!127!
6.3.1.2! Methodologic!Approach!........................................................................................................................!127!
6.3.1.3! Definition!of!AKI!.......................................................................................................................................!129!
6.3.2! Limitations!............................................................................................................................................!129!
6.3.2.1! Data!Sources!...............................................................................................................................................!130!
6.3.2.2! Medications!................................................................................................................................................!131!
6.3.2.3! Definition!of!AKI!.......................................................................................................................................!132!
6.3.2.3.1! Incidence!of!AKI!Using!the!KDIGO!Definition!........................................................................!133!
6.3.2.4! Study!Period!...............................................................................................................................................!134!
6.4! Implications*and*Future*Directions*.......................................................................................................*135!
6.4.1! Implications!..........................................................................................................................................!135!
6.4.1.1! Pharmacoepidemiologic!Methods!Are!Feasible!.........................................................................!135!
6.4.1.2! AKI!Is!Common!In!Critically!Ill!Children!........................................................................................!136!
6.4.1.3! AKI!Definitions!Are!Used!Inconsistently!.......................................................................................!136!
6.4.1.4! Critically!Ill!Children!at!Risk!for!AKI!Can!Be!Identified!...........................................................!138!
6.4.1.5! Nephrotoxicity!Is!an!Important!Contributor!to!AKI!in!Critically!Ill!Children!................!139!
6.4.2! Future!Directions!...............................................................................................................................!140!
x
6.4.2.1! Expand!Pharmacoepidemiologic!Approach!.................................................................................!140!
6.4.2.2! Improve!Consistency!in!the!Use!of!AKI!Definitions!..................................................................!143!
6.4.2.3! Improve!Ability!to!Identify!At!Risk!Populations!........................................................................!145!
6.4.2.4! Improve!Safety!Profiles!and!Increase!Alternative!Treatments!to!Nephrotoxic!Drugs
! 147!
6.5! Summary*...........................................................................................................................................................*149!
References....................................................................................................................................................151!
xi
List of Tables
Table 2.1: Examples of major causes of acute kidney injury in children ................................ 10!
Table 2.2: Current criteria used for the diagnosis of acute kidney injury ............................... 21!
Table 2.3: Risk factors of adverse drug events in the intensive care unit ............................... 30!
Table 2.4: Common drugs known to have nephrotoxic effects, categorized by mechanism of
injury ........................................................................................................................................ 39!
Table 3.1: RIFLE criteria and suggested paediatric modification (pRIFLE) .......................... 50!
Table 3.2: Search strategy ....................................................................................................... 51!
Table 3.3: Summary of studies ................................................................................................ 55!
Table 3.4: Summary of RIFLE application ............................................................................. 59!
Table 3.5: Summary of associations between the RIFLE and measures of mortality and
morbidity ................................................................................................................................. 65!
Table 4.1: Example of second round of expert opinion survey ............................................... 79!
Table 4.2: Description of study population ............................................................................. 83!
Table 4.3: Effect of varying baseline measurements on the estimated incidence of AKI ....... 91!
Table 4.4: Risk factors for developing acute kidney injury (bivariate analyses) .................... 95!
Table 4.5: Risk factors of acute kidney injury in critically ill children (multivariable analysis)
................................................................................................................................................. 96!
Table 5.1: Demographics and clinical composition of case and control patients .................. 111!
Table 5.2: Medication exposure during ICU treatment for case and control patients for all
identified nephrotoxic drugs .................................................................................................. 113!
Table 5.3: Conditional logistic regression model including all nephrotoxic medications .... 115!
xii
Table 5.4: Multivariable conditional logistic regression model, stratified by nephrotoxic
medication prior to ICU admission, adjusted for matching strategy and underlying risk factors
for AKI .................................................................................................................................. 116!
xiii
List of Figures
Figure 2.1: Kidney anatomy and physiology ............................................................................. 8!
Figure 2.2: Representation of the relationship between adverse drug reactions, adverse drug
events and the intersection between error and harm ................................................................ 27!
Figure 3.1: Flow diagram of included and excluded studies ................................................... 54!
Figure 3.2: Reported incidence and distribution of RIFLE-assessed AKI .............................. 57!
Figure 4.1: Definition of study population .............................................................................. 82!
Figure 4.2: Number and proportion of patients with at least one serum creatinine value
available by day (for the first 14 days of their ICU stay) ........................................................ 84!
Figure 4.3: Number and proportion of patients with at least one urine output measurement
available by day (for the first 14 days of their ICU stay) ........................................................ 85!
Figure 4.4: Timing of the most recent creatinine measurement available within 3 months prior
to ICU admission for patients without a creatinine available during the first 24 hours of ICU
admission (n=221) ................................................................................................................... 86!
Figure 4.5: Histogram of first creatinine measurement values within 24 hours of admission
for patients under 1 year of age (n=1020) ............................................................................... 87!
Figure 4.6: Histogram of first creatinine measurement values within 24 hours of admission
for patients between 1 and 12 years of age (n=1605) .............................................................. 88!
Figure 4.7: Histogram of first creatinine measurement values within 24 hours of admission
for patients 12 years of age and older (n=857) ........................................................................ 89!
Figure 4.8: Severity of the first occurrence of GFR-based RIFLE AKI across different
methods of baseline estimation ................................................................................................ 91!
Figure 4.9: Estimated incidence of AKI based on varying RIFLE criteria components ......... 93!
xiv
Figure 5.1: Incidence density sampling ................................................................................. 107!
xv
List of Abbreviations
ACE Angiotensin-converting enzyme
ADE Adverse drug event
ADHD Attention-deficit/hyperactivity disorder
ADR Adverse drug reaction
AKI Acute kidney injury
AKIN Acute Kidney Injury Network
ARF Acute renal failure
ATN Acute tubular necrosis
AUC Area under the curve
BUN Blood urea nitrogen
CI Confidence interval
CICU Cardiac intensive care unit
CKD Chronic kidney disease
CYP450 Cytochrome P450
eCCl Estimated creatinine clearance
eGFR Estimated glomerular filtration rate
ECMO Extracorporeal membrane oxygenation
ESRD End-stage renal disease
xvi
FeNa Fractional excretion of sodium
FDA Food and Drug Administration
GFR Glomerular filtration rate
HSC Hospital for Sick Children
ICD International classification of diseases
ICU Intensive care unit
IL-18 Interleukin-18
IQR Interquartile range
KDIGO Kidney Disease: Improving Global Outcomes
KIM-1 Kidney injury molecule-1
LOS Length of stay
MDRD Modification of diet in renal disease
MODS Multiple organ dysfunction syndrome
NGAL Neutrophil gelatinase-associated lipocalin
NICU Neonatal intensive care unit
NSAID Non-steroidal anti-inflammatory
OR Odds ratio
PELOD Paediatric logistic organ dysfunction
PICU Paediatric intensive care unit
PIM2 Paediatric index of mortality
xvii
pRIFLE Paediatric RIFLE (Risk, injury, failure, loss of kidney function, and end-stage
kidney disease)
PRISM Pediatric risk of mortality
RACHS-1 Risk adjustment in congenital heart surgery
RIFLE Risk, injury, failure, loss of kidney function, and end-stage kidney disease
ROP Retinopathy of prematurity
RRT Renal replacement therapy
SCr Serum creatinine
SIRS Systemic inflammatory response syndrome
TBSA Total body surface area
UO Urine output
VIF Variance inflation factor
1
Chapter 1 Overview
2
1.1 Preamble
Kidney injury is a common adverse event in critically ill children. The etiology of acute kidney
injury (AKI) is complex and multifactorial; some factors, such as age and sex are non-modifiable
while others, including exposure to medications, are controllable and present the opportunity to
decrease the risk of AKI.
The complexity of AKI is magnified in the critical care unit, where patients are acutely ill and it
can be difficult for clinicians to distinguish drug from disease-related causes of AKI. The goal
of this thesis project was to evaluate the contribution of drug therapy to acute kidney injury
(AKI), a common adverse event in critically ill children. To identify patients with AKI, we
chose to utilize a consensus definition known as the RIFLE. The rationale for this decision will
be presented in Chapter 2. First, we needed to understand how RIFLE-defined AKI is
operationalized in the literature. This informed how we operationalized the definition of AKI in
the remainder of the thesis work. We then needed to quantify characteristics and clinical factors
that place patients at higher risk for developing AKI. Finally, combining these results, we then
evaluated the association between specific medications and the development of AKI, controlling
for underlying differences in risk.
The results of this dissertation will add to the growing literature of AKI in critically ill children.
Through understanding what factors and medications increase the risk of AKI, we can optimize
prescribing and help clinicians make optimal treatment choices, thus reducing harm and
improving outcomes.
3
1.2 Research Objectives
The aim of this thesis work was to evaluate the contribution of drug therapy to acute kidney
injury. The specific research objectives of this thesis were to:
1. Systematically evaluate the published literature describing the use of the RIFLE
definition of acute kidney injury in children;
2. Determine the clinical risk factors of acute kidney injury in critically children; and
3. Evaluate the contribution of drug therapy to the development of acute kidney injury.
Each research objective above corresponds to a specific chapter in the thesis work.
1.3 Thesis Organization
This thesis has been written following a three-paper approach. This chapter provides an
overview of the thesis, states the specific research aims and presents the overall organization of
the thesis. Chapter 2 provides the rationale and context for this work, introducing concepts of
acute kidney injury and nephrotoxicity at a macro level. The three specific research aims stated
above are answered through separate studies and are presented in Chapters 3, 4 and 5,
respectively. Finally, Chapter 6 summarizes the major research findings and implications of this
research, as well as discusses the opportunities for future study.
The contents of Chapter 3 have been published in Kidney International. The contents of
Chapters 4 and 5 are currently under review at Kidney International and Critical Care Medicine,
4
respectively. The formatting of these published or submitted manuscripts has been adapted to
conform to this thesis. A common reference section for all chapters follows Chapter 6.
5
Chapter 2 Acute Kidney Injury, Adverse Drug Events and Nephrotoxicity
6
2.1 Overview
The purpose of this chapter is to provide context for the rationale of the thesis work by reviewing
concepts of acute kidney injury (AKI) and nephrotoxicity.
The etiology, incidence and outcomes of acute kidney injury are outlined, followed by a
description of current definition of AKI. This provides the rationale for the choice of AKI
definition used throughout the remainder of the thesis work. Following this is a discussion of
adverse drug events, including definition, incidence and outcomes. We provide information
regarding two high-risk populations, critically ill and paediatric patient populations, the focus of
this thesis. Finally, we discuss common adverse drug events in children and the importance of
increasing drug safety knowledge in paediatric populations.
The chapter then introduces nephrotoxicity as an important cause of acute kidney injury and
briefly discusses the mechanisms of injury. It should be noted that a detailed description of acute
kidney injury at the cellular level is beyond the scope of this thesis work. Finally, we discuss
methods of studying drug safety in children and present pharmacoepidemiology as a viable
method to study nephrotoxicity, as a model to study adverse drug events, in children.
2.2 Acute Kidney Injury
The kidneys are an essential organ; their main functions include the removal of waste products
from the blood, the regulation of sodium, potassium and other electrolytes, the regulation of fluid
balances and blood pressure, the maintenance of acid-base balance, and the production of various
7
hormones. The function unit of the kidney is the nephron, composed of a filtering unit called the
glomerulus and its associated renal tubule (Figure 2.1). Each kidney is comprised of roughly one
million nephrons. Arterial blood enters the kidney via the renal artery. Blood entering the
glomerulus is filtered across the fenestrated glomerular capillary wall, producing an ultrafiltrate
that crossed into Bowman’s space and then enters the tubular lumen proper. As this ultrafiltrate
traverses the length of the tubule, its composition is modified by reabsorption and secretion of
specific components by the tubular epithelial cells. The end result of this process is the
formation of urine, which is transported to the bladder via the ureters, and the concomitant return
of cleaned blood to the circulation through the renal vein ("Urinary System" 2002).
8
Figure 2.1: Kidney anatomy and physiology As blood passes through the kidneys from the renal artery to the renal vein, waste is collected and turned into urine. Inside each kidney is a network of small tubes (nephrons) which filter blood. Waste products in the nephrons move from the capillaries into the urine collecting tubes and into the ureter to the bladder from the renal pelvis, located in the centre of each kidney. From: http://www.cancerresearchuk.org/cancer-help/type/kidney-cancer/about/the-kidneys
Acute kidney injury (AKI) is defined as an abrupt deterioration in kidney function, which results
in an accumulation of creatinine, urea and other waste products (Bellomo, Kellum, & Ronco,
2012; Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working
Group, 2012; "Renal Failure," 2006). The severity spectrum of AKI is broad, starting with
patients who are at risk for AKI, due to genetic or clinical risk factors, proceeding to renal injury
or dysfunction, and culminating in acute renal failure (ARF) (Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012). It can present as subtle
9
biochemical and structural changes, minimal elevations in serum creatinine, or anuric renal
failure (Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working
Group, 2012). AKI is a common comorbid condition in critically ill children and adults (Basu,
Devarajan, Wong, & Wheeler, 2011; Ricci, Cruz, & Ronco, 2008) and independently predicts
mortality in these patient populations (Akcan-Arikan et al., 2007; D. J. Askenazi, Griffin,
McGwin, Carlo, & Ambalavanan, 2009; Ricci et al., 2008; Zappitelli et al., 2008).
Any degree of kidney injury has significant implications on patient health; even mild, reversible
AKI has important clinical consequences including increased mortality (Eric A. J. Hoste et al.,
2006; Shigehiko Uchino, Bellomo, Goldsmith, Bates, & Ronco, 2006). Unfortunately, mild
injury to the renal system begins long before the loss of kidney function can be measured with
standard tests (Bellomo et al., 2012; Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012).
2.2.1 Etiology of AKI
The etiology of AKI has changed over the past decade from primary renal disease to
complications of other systemic illness (D. Askenazi, 2011; Hui-Stickle, Brewer, & Goldstein,
2005; Symons et al., 2007). In critically ill patients, AKI is commonly multifactorial, resulting
from conditions including hypotension, sepsis, congenital heart disease, ischemic injury,
nephrotoxins or malignancy (Akcan-Arikan et al., 2007; Duzova et al., 2010; Hui-Stickle et al.,
2005; Palmieri, Lavrentieva, & Greenhalgh, 2009; Pannu & Nadim, 2008; Schneider, Khemani,
Grushkin, & Bart, 2010; Symons et al., 2007; Vachvanichsanong, Dissaneewate, Lim, &
McNeil, 2006) (Table 2.1). Regardless of etiology, the manifestations and clinical consequences
of AKI are similar.
10
Table 2.1: Examples of major causes of acute kidney injury in children
Classification Etiology Pre-renal
Decreased intravascular volume
Intrinsic Acute tubular necrosis Hypoxia/ischemia Sepsis Multiple organ dysfunction syndrome (MODS) Uric acid nephropathy Tumor lysis syndrome Interstitial nephritis Glomerulonephritis Renal artery/vein thrombosis Cortical necrosis Haemolytic uremic syndrome
Post-renal
Ureteral or urethral obstruction Solitary kidney obstruction
Information from (Andreoli, 2009; Basu et al., 2011)
AKI is traditionally classified by the location of the pathophysiology relative to the kidney; ‘pre-
renal’ diseases alter the perfusion of the kidney, affecting oxygen delivery to the organ,
‘intrinsic’ diseases cause damage within the kidneys themselves, and obstructions of the urinary
tract are considered ‘post-renal’ injury (Andreoli, 2009; Basu et al., 2011; Zappitelli &
Goldstein, 2009). Each of the conditions related to the development of AKI causes renal injury
through different mechanisms.
Pre-renal AKI results from renal hypoperfusion, due to low intravascular volume caused by
dehydration, fluid shifts outside the intravascular space, or capillary leaks, or low circulating
volume from poor cardiac output or systemic vasodilation (Zappitelli & Goldstein, 2009). Pre-
renal AKI can also occur due to renal artery stenosis or compression, though this is rare.
Regardless of cause, the result is a decrease in the perfusion of the kidney. If renal
11
hypoperfusion is restored quickly (e.g. a rapid replenishment of fluid volume to the kidney),
kidney function may be quickly restored. Otherwise, the process of acute tubular necrosis
(ATN) begins. ATN is most commonly caused by a lack of oxygen (ischemia) to renal tissue;
other causes of ATN are toxic or vascular insults to the kidney or through inflammatory
mechanisms. ATN causes damage to or destruction of the internal structures of the kidneys,
particularly the tubules. Once the pathophysiologic processes of ischemic renal injury have been
triggered, restoration of intravascular volume will not restore kidney function.
Intrinsic causes of AKI refer to any acute reduction in kidney function as a result of direct
damage to renal tissue (Zappitelli & Goldstein, 2009). In the ICU, the most common cause of
this form of AKI is acute tubular necrosis (described above). Post-renal AKI results from
obstructions of the urinary tract. In the paediatric ICU setting, this is a rare cause of AKI and is
generally related to tumors or ureteral obstructions from mass effect, stones or blood clots.
Unappreciated catheter obstructions in children with indwelling Foley catheters may also
contribute to the development of post-renal AKI.
2.2.2 Incidence of AKI
The prevalence of AKI in the paediatric hospital setting appears to be increasing
(Vachvanichsanong et al., 2006) likely as a result of the increased availability of treatment
options for many critical illnesses and advances in paediatric and neonatal critical care
(Devarajan, 2013). The incidence of AKI among children admitted to intensive care units has
been estimated to range from 10% to 35% (Alkandari et al., 2011; Kavaz et al., 2012; Schneider
et al., 2010). In more severely critically ill children, the incidence of AKI has been reported to
occur in 90% of patients with traumatic injuries or those needing vasopressor support and
requiring mechanical ventilation (Prodhan et al., 2012). The incidence of AKI ranges from 30%
12
to 50% in children undergoing cardiac surgery (Aydin et al., 2012; Blinder et al., 2012; Fadel et
al., 2012; Krawczeski, Woo, et al., 2011; Li et al., 2011; Toth et al., 2012). Even among non-
critically ill children receiving nephrotoxic medications, the incidence of AKI has been estimated
to be 34% (Moffett & Goldstein, 2011). The wide variability in AKI incidence may be due to
differences in the definition of AKI used; however, it is reasonable to postulate that specific
populations of critically ill children have higher rates of AKI due to underlying differences in
risk.
2.2.3 Outcomes of AKI
AKI has been associated with increased mortality in adult patients. A meta-analysis of 24
studies involving over 71 000 adult patients reported a pooled estimated mortality rate of 31.2%
in patients with AKI compared with 6.9% in patients without AKI (Ricci et al., 2008). In
children, the short-term outcomes of critically ill children with AKI have been well documented.
High mortality rates ranging from 30% to 40% have been consistently reported in critically ill
children with AKI (Alkandari et al., 2011; Bresolin et al., 2009; Duzova et al., 2010; Kavaz et
al., 2012; Palmieri et al., 2009; Prodhan et al., 2012; Schneider et al., 2010). When compared
with other patients, mortality rates are significantly higher among children who develop AKI: in
a study of critically ill neonates and children who received extracorporeal membrane
oxygenation (ECMO), the adjusted odds of death in patients with AKI was three times higher
(95% confidence interval (CI): 2.6-4.0) than in those without AKI (D. J. Askenazi et al., 2009).
In addition, a study of 430 infants who had cardiac surgery for congenital defects found that
more severe AKI was associated with increased hospital mortality (Blinder et al., 2012). When
compared with patients without AKI, the odds of mortality in those who developed AKI ranged
from 5.1 (95% CI: 1.7 – 15.2) for patients with moderate AKI to 9.5 (95% CI: 2.9 – 30.7) for
13
patients with severe AKI. Acute kidney injury is also related to measures of morbidity in
children. In a multicentre, retrospective analysis of 2 106 paediatric ICU admissions, AKI was
shown to be independently associated with mechanical ventilation and increased ICU stay
(Alkandari et al., 2011). Extended use of mechanical ventilation and increased hospital stay
have been shown to be independently associated with AKI in children undergoing cardiac
surgery (Aydin et al., 2012; Blinder et al., 2012; Li et al., 2011; Toth et al., 2012).
While short-term outcomes of AKI in children have been well documented, little information
exists on long-term survival in paediatric patients who develop AKI. Askenazi et al (2006)
followed a cohort of 174 children who developed acute renal failure during their hospitalization
and survived to hospital discharge. They found a survival rate of 80% three to five years post-
hospital discharge; most deaths (65%) occurred within one year of the initial hospitalization.
It was previously thought that patients who survived an episode of AKI would recover full
kidney function; a recent meta-analysis found that adults with AKI were nine times more likely
to develop chronic kidney disease and three times more likely to develop end-stage renal disease
compared to patients without AKI (Coca, Singanamala, & Parikh, 2012).
In a retrospective study of paediatric inpatients who developed AKI during their hospitalization,
two-thirds had recovered full renal function and 15% had improved renal function by hospital
discharge, though the definition of ‘recovered’ and ‘improved’ renal function was not clearly
stated by the authors (Hui-Stickle et al., 2005). In the same patient cohort, 14% developed
chronic renal failure and 5% who were discharged needed ongoing renal replacement therapy
(RRT) (Hui-Stickle et al., 2005). Among a cohort of paediatric AKI patients who survived three
to five years post-hospital discharge, 29 patients consented to additional renal assessments.
Residual renal injury was defined as the presence of any one of: microalbuminuria, hypertension,
14
hematuria, or an estimated glomerular filtration rate outside a normal range, where normal was
defined as 90-150 ml/min/1.73 m2. Fifty-nine percent of patients had signs of renal dysfunction
at the time of follow-up (D. J. Askenazi et al., 2006).
Treatment of AKI in critically ill children focuses on avoiding or minimizing further renal injury
and preventing life-threatening imbalances of fluids or electrolytes (Zappitelli & Goldstein,
2009). Patients with severe AKI or those with severe fluid overload or electrolyte imbalance
may require renal replacement therapy (Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012). While modern RRT has largely eliminated the
traditional, life-threatening complications of AKI such as hyperkalemia, arrhythmia, and uremic
coma, children with AKI treated with RRT still have a high mortality rate, ranging from 30% to
50% (Devarajan, 2013; Symons et al., 2007). This high mortality rate has remained stable over
the past two decades (Devarajan, 2013; Symons et al., 2007) and is likely due to a number of
factors. For example, interactions between the kidney and other major organs, known as ‘cross-
talk’, result in cycle of AKI and multi-organ failure, which leads to death (Amann, Wanner, &
Ritz, 2006; Basu & Wheeler, 2013; Grams & Rabb, 2012; Lane, Dixon, Macphee, & Philips,
2013). In addition, AKI also impairs the immune system, increasing a patient’s susceptibility to
infection (Singbartl et al., 2011). AKI also contributes to medication failure: critically ill
children are at substantial risk of ADE (Awdishu & Bouchard, 2011; Eyler & Mueller, 2011;
Perazella, 2012) and under-dosing of medications often occurs due to unstable elimination rates
and volumes of distribution and significant clearance by RRT (Awdishu & Bouchard, 2011;
Eyler & Mueller, 2011; Perazella, 2012). Finally, AKI often results in fluid overload, which has
been shown to be an independent risk factor for mortality in AKI (Sutherland et al., 2010).
15
2.2.4 Defining AKI
2.2.4.1 Measuring Kidney Function
As the main function of the renal system is the removal of waste products from the body, the
glomerular filtration rate (GFR), defined as the volume of plasma cleared of a substance per unit
of time, is widely accepted as the most useful overall index of kidney function (Traynor, Mactier,
Geddes, & Fox, 2006). Ideally, the best substance to measure GFR is one that is excreted only
by the kidneys and not reabsorbed. While inulin meets these criteria, it is not naturally available
in the human body. Radioactive markers can also be used to measure GFR (Stevens & Levey,
2009), but these measures are not practical for routine clinical use. Creatinine is the endogenous
substance that is closest to an ideal (Chantler, Garnett, Parsons, & Veall, 1969) and creatinine
clearance, the amount of creatinine cleared from the blood during a given time period, can be
used to estimate GFR. However, measuring creatinine clearance requires a 24 hour urine
collection which can be both difficult and inconvenient for patients (Traynor et al., 2006).
Creatinine clearance also tends to overestimate the true GFR (Proulx et al., 2005; Toto, 1995).
As such, a number of formulae have been derived to estimate GFR based solely on serum
creatinine levels, known as estimated creatinine clearance. In adults, the Modification of Diet in
Renal Disease (MDRD) formula is the most widely used ("K/Doqi Clinical Practice Guidelines
for Chronic Kidney Disease: Evaluation, Classification, and Stratification," 2002). The MDRD
estimates GFR using creatinine, age, sex and race (African-American versus other races). The
formula is not accurate for use in children, the elderly, those with unusual muscle mass and
weight (e.g. morbidly obese patients). In children, the Schwartz formula is widely used
(Schwartz, Haycock, Edelmann, & Spitzer, 1976). The Schwartz formula estimates a child’s
GFR based on the child’s height and serum creatinine.
16
In clinical practice, an abrupt decline in GFR, indicative of kidney injury, is assessed by an
increase in serum creatinine and/or oliguria. While these measures are limited in their ability to
provide early detection of renal injury as changes in serum creatinine may lag behind changes
(either decline or recovery) in GFR by several days, changes in serum creatinine and/or urine
output are the foundation for diagnostic criteria of AKI (Akcan-Arikan et al., 2007; Bellomo,
Ronco, Kellum, Mehta, & Palevsky, 2004; Kidney Disease: Improving Global Outcomes
(KDIGO) Acute Kidney Injury Working Group, 2012; R. L. Mehta et al., 2007). Other
diagnostic tools, including clinical history, physical examination, renal ultrasound, fractional
excretion of sodium (FeNa), fractional excretion of urea, blood urea nitrogen (BUN), and urine
microscopy, can provide insight to the etiology of the renal injury.
2.2.4.1.1 Biomarkers
In order to improve the detection of acute kidney injury, a number of new biomarkers have been
identified. These markers identify early stress responses of the kidney and appear in the urine or
plasma before changes in serum creatinine are evident (Devarajan, 2011). The most widely
studied and validated early biomarker of AKI in children is neutrophil gelatinase-associated
lipocalin (NGAL). Levels of NGAL in the urine and plasma of children undergoing
cardiopulmonary bypass were significantly elevated within 2-6 hours after bypass in children
who subsequently developed AKI (Bennett et al., 2008; Dent et al., 2007; Krawczeski,
Goldstein, et al., 2011; Krawczeski, Woo, et al., 2011; Mishra et al., 2005; Parikh et al., 2011).
Increased NGAL measurements have also been shown to be associated with hospital length of
stay and the duration and severity of AKI (Bennett et al., 2008; Dent et al., 2007; Krawczeski,
Goldstein, et al., 2011; Krawczeski, Woo, et al., 2011; Parikh et al., 2011). Studies in the
paediatric ICU and emergency department settings have shown that increases in NGAL predicted
17
AKI roughly one to two days before corresponding increases in serum creatinine were evident
(Du et al., 2011; Wheeler et al., 2008; Zappitelli et al., 2007). Two recent meta-analyses of
NGAL studies in critically ill adults and children have confirmed the clinical utility of this
marker as an early indicator of AKI and showed significant associations between increasing
NGAL levels and clinical outcomes including mortality, length of stay and the initiation of renal
replacement therapy (Haase, Bellomo, Devarajan, Schlattmann, & Haase-Fielitz, 2009; Haase et
al., 2011).
Serum cystatin C levels have also been shown to be highly correlated to AKI and can be used as
a marker of disease progression after kidney transplantation (Gourishankar, Courtney, Jhangri,
Cembrowski, & Pannu, 2008). In addition, kidney injury molecule-1 (KIM-1), interleukin-18
(IL-18), and liver fatty acid binding-protein have been shown to be associated with renal
ischemia (Devarajan, 2008; Parikh, Edelstein, Devarajan, & Cantley, 2007). However, recent
studies have reported relatively low predictive abilities of biomarkers to predict worsening
kidney function; when combined with clinical markers, their predictive ability increases (Arthur
et al., 2014; Hall et al., 2011; Koyner et al., 2012; Koyner et al., 2010). For example, in a study
of patients with AKIN-defined stage 1 kidney injury after cardiac surgery, IL-18 best predicted
the worsening of AKI or death compared with the other 32 biomarkers assessed, with an area
under the curve (AUC) of 0.74 (Arthur et al., 2014). However, this level of accuracy is moderate
(Fischer, Bachmann, & Jaeschke, 2003). The predictive ability of IL-18 increased when it was
combined with the percent change in serum creatinine (Arthur et al., 2014). The use of
biomarkers is a rapidly evolving field; however, to date, biomarkers have not been incorporated
into clinical practice and thus are not currently a practical method to assess and diagnose acute
kidney injury.
18
2.2.4.2 Current Definitions of AKI
More than 35 definitions of AKI have been used in the literature (R. L. Mehta & Chertow, 2003).
In order to promote a consistent definition of AKI to allow for comparison of study results, the
Acute Dialysis Quality Initiative convened an international consensus panel in 2002 and
proposed the RIFLE criteria for use in critically ill adults (Bellomo et al., 2004). Based on
changes from the patient’s normal (baseline) kidney function, the RIFLE classifies increasing
severity of acute kidney injury into 5 categories: Risk (R), Injury (I), Failure (F), Loss of kidney
function (L), and End-stage kidney disease (E) (Table 2.2). The first three severity levels (Risk,
Injury and Failure) are assessed by changes in glomerular filtration rate from a baseline measure
and urine output. The worst of these measures define a patient’s severity level. For example,
consider a patient whose creatinine has increased 1.5 times from their baseline measure and
whose urine output is 0.4 ml/kg/h x 12h. According to the GFR criteria, the patient’s AKI
severity meets the ‘Risk’ definition, but their urine output meets the ‘Injury’ criteria. As such,
the patient would be classified according to their urine output and be diagnosed as having an
AKI severity of ‘Injury’. The length of renal replacement therapy (RRT) defines the final two
severity levels (Loss and End-stage). When a measure of the patient’s baseline GFR is not
available or is unknown, the panel suggested assuming a GFR of 75ml/min/1.73m2, the lower
limit of normal (Bellomo et al., 2004).
A modification of the RIFLE (known as the pRIFLE) has been suggested for use in paediatric
populations (Akcan-Arikan et al., 2007). The changes are minor and include a focus on the
estimated creatinine clearance, calculated using the Schwartz formula (Schwartz, Haycock,
Edelmann, et al., 1976), as the measure of GFR. In addition, the threshold for the ‘Failure’
category was modified from a serum creatinine ≥ 4 mg/100ml to an estimated creatinine
19
clearance (eCCl) < 35 ml/min/1.73m2, and the time interval for the urine output criteria was
increased from 6 hours to 8 hours (Table 2.2).
In 2005, the Acute Kidney Injury Network (AKIN) reviewed and modified the RIFLE criteria,
recognizing that even very small changes in serum creatinine are associated with adverse clinical
outcomes (Coca, Peixoto, Garg, Krumholz, & Parikh, 2007; Eric A. J. Hoste et al., 2006;
Lassnigg et al., 2004; Nash, Hafeez, & Hou, 2002; Ostermann & Chang, 2007; Shigehiko
Uchino et al., 2006). According to the AKIN, AKI is defined as “an abrupt (within 48 hours)
reduction in kidney function currently defined as an absolute increase in serum creatinine of
more than or equal to 0.3 mg/dl (≥26.4 µmol/L), a percentage increase in serum creatinine of
≥50% (1.5-fold from baseline), or a reduction in urine output (documented oliguria of <0.5
mL/kg/hr for >6 hrs)” (R. L. Mehta et al., 2007) (Table 2.2). Similar to the RIFLE, the AKIN
definition also defines levels of severity, defined as Stages 1, 2 and 3, which correspond to
RIFLE severity levels of ‘Risk’, ‘Injury’ and ‘Failure’, respectively. Also similar to the RIFLE,
the AKIN is based on both changes in creatinine and urine output; however, it also adds a
measure of time. The time constraint of 48 hours for diagnosis was selected based on evidence
that small increases in creatinine within 24 to 48 hours were associated with a threefold increase
in 30-day mortality (Lassnigg et al., 2004). Studies have shown that the RIFLE and AKIN
criteria show concordance in critically ill patients (Bagshaw, George, Bellomo, & Committe,
2008; Lopes et al., 2008). However, evidence has shown that the RIFLE captures more mild
cases of acute kidney injury than the AKIN (Zappitelli, Moffett, Hyder, & Goldstein, 2011) and
the AKIN classification does not improve outcome prediction over the RIFLE (Bagshaw et al.,
2008; Chang et al., 2010).
20
Recognizing the need for a single consensus definition and staging system that could be applied
to both children and adults, the Kidney Disease: Improving Global Outcomes (KDIGO) group
proposed a new definition of AKI in 2012 that incorporates the RIFLE, pRIFLE and AKIN
classifications (Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury
Working Group, 2012) (Table 2.2). Both the binary definition of AKI and the severity scale
feature a 0.3 mg/dl increase in serum creatinine to specifically be applicable to paediatric AKI.
The KDIGO staging also allows for a child with an estimated GFR <35 mL/min per 1.73m2 to be
classified as high severity (Stage 3), in contrast with the adult criterion of ≥4 mg/dl which would
be unrealistic in infants and young children.
21
Table 2.2: Current criteria used for the diagnosis of acute kidney injury
Classification Stage Creatinine criteria Urine output criteria
RIFLE (Bellomo et al., 2004)
Risk Increased creatinine x1.5 or GFR decrease >25%
<0.5 ml/kg/h x 6h
Injury Increased creatinine x2 or GFR decrease >50%
<0.5 ml/kg/h x 12h
Failure Increased creatinine x3 or GFR decrease >75% or creatinine ≥4 mg/100ml (acute rise of ≥0.5 mg/100ml dl)
<0.3 ml/kg/h x 24h or anuria x 12h
Loss Persistent ARF = complete loss of renal function > 4 weeks (defined as the need for renal replacement therapy (RRT) for >4 weeks)
End-stage End-stage renal disease (defined as the need for dialysis for >3 months)
Pediatric RIFLE (pRIFLE) (Akcan-Arikan et al., 2007)
Risk
eCCl decrease by 25%
<0.5 ml/kg/h x 8h
Injury eCCl decrease by 50% <0.5 ml/kg/h x 16h Failure eCCl decrease by 75% or eCCl
<35 ml/min/1.73m2 <0.3 ml/kg/h x 24h or anuria x 12h
Loss Persistent failure >4 weeks End-stage
End-stage renal disease (persistent failure >3 months)
AKIN (R. L. Mehta et al., 2007)
1
Increased creatinine x 1.5-2 or creatinine increase ≥0.3 mg/dl
<0.5 ml/kg/h x 6h
2 Increased creatinine x 2-3 <0.5 ml/kg/h x 12h 3 Increased creatinine x ≥3 or
creatinine ≥4.0 mg/dl with an acute increase of 0.5 mg/dl
<0.3 ml/kg/h x 24h or anuria x 12h
KDIGO (Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012)
1
Increased creatinine x1.5-1.9 or ≥0.3 mg/dl increase
<0.5 ml/kg/h x 6-12h
2 Increased creatinine x2.0-2.9 <0.5 ml/kg/h x ≥12h 3 Increased creatinine x3 or
creatinine ≥4.0 mg/dl or initiation of RRT or eGFR <35 ml/min per 1.73m2 (<18 years)
<0.3 ml/kg/h x ≥24h or anuria x ≥12h
AKIN: Acute Kidney Injury Network; GFR: glomerular filtration rate; eCCl: estimated creatinine clearance; eGFR: estimated glomerular filtration rate; KDIGO: Kidney Disease: Improving Global Outcomes; RRT: renal replacement therapy
22
Little evidence exists in the literature surrounding the benefit of the KDIGO criteria over other
consensus definitions. Some studies have shown differences between the RIFLE or AKIN and
the new KDIGO definition. For example, a study of various AKI definitions in adult patients
admitted with heart failure reported an AKI incidence of 36.7% based on KDIGO compared to
incidence rates of 25.6% and 27.9% according to the RIFLE and AKIN, respectively (Roy et al.,
2013). Despite the differences in estimated incidence, there were no differences between the
three definitions in terms of their predictive ability of 30-day and one-year adverse events,
defined as a composite of heart failure-related hospital admission, renal replacement therapy, and
all-cause mortality. Another study of adult patients with acute myocardial infarction (AMI)
found the incidence of AKI during the first 7 days of hospitalization to be 14.8% and 36.6%
according to the RIFLE and KDIGO definitions, respectively (Rodrigues et al., 2013). Patients
diagnosed as having AKI by the KDIGO definition but not by RIFLE had significantly higher
30-day and one-year mortality. Other studies have shown no significant difference between the
definitions: patients undergoing cardiac surgery had post-surgical AKI rates of 25.9% according
to either the AKIN or KDIGO and 24.9% based on the RIFLE (Bastin et al., 2013). All studies
were performed in adult patient populations and used only the serum creatinine criteria of the
definitions, ignoring any effect of urine output on the estimated incidence of AKI. A recent
study compared the pRIFLE, AKIN and KDIGO definitions in children. Lex et al applied the 3
different definitions in a cohort of 1 489 consecutive paediatric patients who underwent cardiac
surgery and found differences in the estimated incidence of AKI (Lex et al., 2014). According to
the AKIN criteria, 20% of patients had AKI. The estimated incidence increased to 29% and 34%
according to the KDIGO and pRIFLE definitions, respectively. As in adult studies, the presence
and severity of AKI was assessed for each classification using only changes in serum creatinine
23
or estimated creatinine clearance; the urine output component of the definitions was not
considered.
Finally, similar to the idea of cardiac angina, the concept of ‘renal angina’ has recently been
introduced to identify patients who may benefit the most from early treatment to manage or
prevent the development of AKI (Basu, Chawla, Wheeler, & Goldstein, 2012; S. L. Goldstein &
Chawla, 2010). There are few signs and symptoms in the early stages of AKI when interventions
are likely to be the most effective and the definition places children at moderate, high or very
high risk of developing AKI. As the risk for AKI increases, less laboratory evidence of AKI (i.e.
specified changes of serum creatinine) is needed to meet the threshold for a diagnosis of renal
angina. In order to be classified as having a moderate risk of AKI, patients need to be admitted
to the ICU and experience either a two-fold increase in serum creatinine or a fluid overload of
more than 15%. Stem cell transplant patients or those with heart failure who have a serum
creatinine increase of at least 0.3 mg/dl or a fluid overload of more than 10% are classified as
high risk, whereas patients with mechanical ventilation and any increase in serum creatinine or a
fluid overload of more than 5% are consider to be at very high risk for AKI.
Recently, the idea of renal angina has been further developed into an index for use in critically ill
children (Basu et al., 2013). To calculate the renal angina index, a patient is assigned a score
based on their risk of AKI (hereafter referred to as the risk of AKI score); patients who are
admitted to the ICU are deemed to be at ‘moderate’ risk and given a score of 1. Patients who
have undergone stem cell transplantation are at ‘high’ risk and given a score of 3; patients who
are ventilated and prescribed inotropes are deemed to be at ‘very high’ risk, which equates to a
score of 5. Similar scores are given to patients based on the signs of renal injury, assessed by
changes in estimated creatinine clearance (eCCl) and fluid overload. Based on the combination
24
of eCCl and FO, patients are given scores of 1 (no change in eCCl, fluid overload <5%), 2 (eCCl
decrease between 0-25%, fluid overload ≥5%), 4 (eCCl decrease of 25-50%, fluid overload
≥10%) or 8 (eCCl decrease ≥50%, fluid overload ≥15%). The renal angina index is calculated
by multiplying the risk of AKI score and the signs of injury score, resulting in a value between 1
and 40. The renal angina index was validated in three cohorts of critically ill patients by
comparing its predictive performance against the KDIGO definition of AKI and illness severity
scores (Basu et al., 2013).
For the purposes of this thesis, the RIFLE definition of AKI was utilized. Both the KDIGO
definition and the renal angina index were published after the majority of the thesis work was
completed. The RIFLE definition was chosen over the AKIN because, as previously stated, the
RIFLE has been shown to capture more mild cases of acute kidney injury than the AKIN
(Zappitelli et al., 2011) and the AKIN classification does not improve outcome prediction over
the RIFLE (Bagshaw et al., 2008; Chang et al., 2010). Finally, the proposed modification to the
RIFLE for use in paediatric populations (known as the pRIFLE (Akcan-Arikan et al., 2007))
provides limited justification for the changes made to the original RIFLE (Akcan-Arikan et al.,
2007). Modification of the urine output assessment from a 6 hour to an 8 hour interval may be
illustrative of pragmatic customization of the RIFLE due to institutional practice or difficulties
with measurement accuracy. Accordingly, AKI was defined as per the RIFLE definition for this
dissertation.
25
2.3 Adverse Drug Events
Pharmacotherapy plays an integral role in health care. Medications are used to treat primary
disease, to facilitate the initiation and continued use of mechanical organ support, such as
mechanical ventilation, as well as to counteract effects of other medications. However,
medications may also have unwanted side effects that can range from easily recognized physical
symptoms (i.e. rashes) to other reactions that may be difficult to associate with a specific drug
(i.e. confusion or organ dysfunction). In a sample of 30 195 randomly selected hospital records,
the Harvard Medical Practice Study identified 1 133 patients (3.7%) with injuries caused by
medical treatment (Leape et al., 1991). Adverse events have high rates of mortality and
morbidity: while not all morbidity and mortality may be directly attributable to an adverse event,
up to 30% of adult patients with complications die or experience disability more than 6 months
after the adverse event (Brennan et al., 1991). The study found that the most common adverse
events experienced by hospitalized patients were complications from medications, accounting for
over 19% of detected events (Leape et al., 1991). Using similar methodology as the Harvard
Medical Practice Study, the Canadian Adverse Events Study reported a rate of 7.5 adverse events
per 100 hospital admissions (Baker et al., 2004). The authors reported the most common types
of adverse events were related to surgical procedures (34.2% of detected events), followed by
drug- or fluid-related events (23.6%). A recent Canadian study focused on estimating the
incidence of adverse events in paediatric patients (Matlow et al., 2012). The authors reviewed 3
669 admissions to 8 academic paediatric hospitals and 14 community hospitals across Canada
and reported a incidence rate of 9.2%, weighted for the sampling strategy. Adverse events were
most commonly related to surgery (32.9%); however, drugs (13.5%), anaesthesia (2.5%) and
fluids (0.8%) were also attributed to the occurrence of an adverse event in paediatric patients.
26
Unintended drug reactions can be defined in a number of ways. An adverse drug reaction (ADR)
is considered to be an unintended and noxious response that occurs under normal or expected
uses of a drug ("International Drug Monitoring. The Role of the Hospital," 1969). Adverse drug
reactions rank as the fifth leading cause of death and illness in the developed world, estimated to
claim between 100,000 and 218,000 lives annually (Ernst & Grizzle, 2001; White, Arakelian, &
Rho, 1999). However, ADRs consider only episodes in which the use of the drug is appropriate,
whereas many adverse events can be attributed to other factors including, but not limited to, error
(Bates, Leape, & Petrycki, 1993). As such, an adverse drug event (ADE) is defined as any injury
or harm related to the use of a drug ("Ashp Guidelines on Preventing Medication Errors in
Hospitals," 1993). As such, adverse drug reactions are considered a subset of adverse drug
reactions (Figure 2.2); the definition of an ADE includes non-preventable occurrences, such as
unpredictable drug rashes and expected events such as complications of chemotherapy, as well as
preventable incidents caused by errors in prescribing, dispensing or administering drugs. This
broader definition of drug-associated harm will be used in this thesis.
27
Figure 2.2: Representation of the relationship between adverse drug reactions, adverse
drug events and the intersection between error and harm Adapted from the Bates model of medication errors, potential adverse drug events and adverse drug events (Bates, Boyle, Vander Vliet, Schneider, & Leape, 1995). All errors do not result in an adverse drug event, nor are all adverse drug events a result of error. ADEs that are a result of error are known as preventable ADEs. Adverse drug reactions (ADRs) are a subset of ADEs as they consider only events that occur after appropriate drug therapy. ADE: adverse drug event; ADR: adverse drug reaction
It should be noted that while adverse drug events encompass injury caused by error, not all
ADEs are caused by error, nor do all errors result in harm (Bates, Boyle, et al., 1995). Several
studies have suggested that roughly one-third of ADEs are associated with medication errors and
thus are considered to be preventable events (Bates, Cullen, et al., 1995; Bates et al., 1993). The
medication process is comprised of 5 stages: prescription, transcription, preparation, dispensing
28
and administration (Hussain & Kao, 2005); errors can occur at any step in this process. Most
errors occur in the administration stage (53% of all errors), followed by prescription (17%),
preparation (14%) and transcription (11%) (Krahenbuhl-Melcher et al., 2007). While only 10%
of medication errors result in an ADE, these errors have implications for patients and health care
providers (Barker, Flynn, Pepper, Bates, & Mikeal, 2002; Bates, Boyle, et al., 1995; Calabrese et
al., 2001).
Adverse drug events are common in hospitalized patients, occurring at rates of 6.5 per 100 adult
admissions, and result in significantly higher hospital stays and resource utilization (Bates,
Cullen, et al., 1995; Bates et al., 1997; Cullen et al., 1997; Lazarou, Pomeranz, & Corey, 1998).
2.3.1 At Risk Populations: Critical Illness and Children
Adverse drug events are common in the intensive care unit (ICU). Providing a critically ill
patient with a single dose of a single medication requires the health care provider to correctly
execute anywhere from 80 to 200 steps (Hussain & Kao, 2005). During a one-year, prospective
observational study of adult patients, the Critical Care Safety Study found that almost half (47%)
of adverse events observed in the medical intensive care and coronary care units of an academic
hospital were drug-related (Rothschild et al., 2005). Among errors deemed as ‘serious’, defined
as those that caused harm or injury or had the potential to cause harm, medications were
responsible for 78% (Rothschild et al., 2005). Given that patients in the ICU have substantially
higher morbidity and mortality than other hospital inpatients, it is unsurprising that the rates of
ADEs in the ICU are higher than those for other hospitalized patients (Bates, Cullen, et al., 1995;
Cullen et al., 1997). The severity of adverse events is also higher in critically ill patients; a
prospective cohort study of all adult admissions to medical and surgical intensive care units and
29
general care units found that 26% of ADEs occurring in ICUs were classified as life threatening
compared to 11% in non-ICU patients (Cullen et al., 1997).
The ICU environment is complex and brings together high-risk patients and interventions with a
narrow therapeutic index. Critically ill patients are vulnerable to ADEs due to their illness
severity (Valentin et al., 2006), the intensity of their treatment, including complex drug regimens
(Pronovost et al., 2003; Rothschild et al., 2005), sedation and mechanical ventilation (Beckmann
et al., 2003; Hussain & Kao, 2005), as well as altered organ function which can affect the
pharmacokinetics of drugs (Table 2.3). Critically ill patients are prescribed twice as many
medications as other inpatients (Cullen et al., 1997) and usually receive multiple drugs and
complex drug combinations. The risk of an ADE increases with each additional drug used
(Impicciatore et al., 2001; Silva et al., 2013), likely due to drug interactions (May, Stewart, &
Cluff, 1977; Rothschild et al., 2005). Thus, it is far more likely that adverse events will occur in
the intensive care setting compared to other inpatient areas of the hospital.
A wide variety of drugs have been associated with adverse drug events, with no group or class of
drugs reported to cause a disproportionate share of ADEs (Bates, Cullen, et al., 1995; Cullen et
al., 1997). In addition, environmental characteristics of ICUs may lead to higher rates of ADEs
and medication errors, including the fast pace of care, frequent personnel changes and handoffs
of care, communication challenges, loud noise, and frequent work interruptions (Camire, Moyen,
& Stelfox, 2009; Donchin et al., 1995; Donchin & Seagull, 2002; Kane-Gill & Weber, 2006).
30
Table 2.3: Risk factors of adverse drug events in the intensive care unit
Factors Specific risk factors Patient Illness severity Strongest predictor of ADE ICU patients more likely to experience ADE than other
hospitalized patients Extreme ages Increased susceptibility to ADEs Prolonged hospitalization Increased exposure and susceptibility to ADEs Sedation Patients unable to participate in care Medications Types of medications Frequent use of boluses and infusions Weight-based infusions derived from estimated weights or
unreliable determinations Mathematical calculations required for medication dosages Programming of infusion pumps Number of medications Twice as many medications prescribed as for other
hospitalized patients Increased probability of medication error and medication
interactions Number of interventions Increased risk of complications ICU environment Complex environment Difficult working conditions make errors more probable High stress High turnover of patients and providers Emergency admissions Risk of an adverse event increases by approximately 6%
per day Multiple care providers Challenges the integration of different care plans Information from (Camire et al., 2009; Moyen, Camire, & Stelfox, 2008) ADE: adverse drug event; ICU: intensive care unit
Paediatric patients may be at an additional increased risk of experiencing harm from a
medication. A systematic review of studies reporting adverse drug reactions in children found
the incidence of adverse drug events in children was estimated to range from 4.3 to 16.7%
(Impicciatore et al., 2001). The authors reported the pooled average incidence of ADEs, adjusted
31
and weighted according to study sample size, as 9.5% (95% CI: 6.8-12.3). Hospitalized children
may be at higher risk of an adverse event (Gill et al., 1995); 14 to 17% of paediatric
hospitalizations result in adverse drug reactions with roughly 28% of these reactions being
severe, resulting in high rates of morbidity and mortality (Gonzalez-Martin, Caroca, & Paris,
1998; Impicciatore et al., 2001; Martinez-Mir et al., 1999; Smyth et al., 2012). The number of
potential adverse drug events has been shown to be higher in paediatric inpatients than in adult
inpatients (Kaushal et al., 2001). Amongst paediatric inpatients, the greatest numbers of adverse
drug events occur in critical care units (Kilbridge et al., 2009; Le, Nguyen, Law, & Hodding,
2006).
Paediatric patients have an increased risk of adverse drug events for a number of reasons. First,
there are important pharmacokinetic differences during childhood and early adolescence due to
changes in body composition and organ function (Anderson, 2002; Fernandez et al., 2011;
Niederhauser, 1997). In addition, children also pose unique challenges during the ordering,
dispensing, administering and monitoring of medications. For example, as weight-based dosing
is required for most medications in paediatrics, ordering drugs typically involves more
calculations compared to dosing in adult patients. Dispensing drugs is also prone to error as
pharmacists must often dilute stock solutions for use in children (Koren & Haslam, 1994). A
systematic review of medication errors in paediatric patients found that the most common type of
error was a dosing error, many of which involve an incorrect dosage of 10 times more or less
than the required dose, known as a ten-fold error (Ghaleb et al., 2006).
2.3.2 Common Adverse Drug Events in Paediatric Populations
In 2012, a systematic review of observational studies of adverse drug reactions in children was
conducted. The authors reviewed studies of ADRs causing admission to hospital, occurring
32
during hospital stay and occurring in the community setting (Smyth et al., 2012). The authors
reported that incidence rates for adverse drug reactions causing hospital admission ranged from
0.4% to 10.3% of all children. For children exposed to a drug during their hospitalization, the
incidence of adverse drug reactions ranged from 0.6% to 16.8% (Smyth et al., 2012).
To date, the largest prospective evaluation of ADEs in paediatric inpatients was a cohort study of
1 120 children admitted to 2 academic hospitals over a six-week period (Kaushal et al., 2001).
The authors evaluated 10 778 drug orders and found 26 ADEs. However, they did not report
disaggregated results according to the type of ADE, the number of drug administrations
(compared to the number of drug orders) or the proportion of critically ill patients. One study of
357 written medication orders and 236 drug administration observations occurring within 26 12-
hour periods in a paediatric ICU (PICU) found 3 ADEs per 100 orders (Buckley, Erstad, Kopp,
Theodorou, & Priestley, 2007). A 2013 prospective cohort study of critically ill paediatric
patients reported an incidence rate of 35.1%, with 110 ADEs occurring in 84 patients (Silva et
al., 2013).
There is limited literature on specific adverse drug events in paediatric populations as ADEs are
generally reported in aggregate form. Kaushal et al found that hypokalemia (41.3% of all ADEs),
hypomagnesemia (11.9%) and nephrotoxicity (11.3%) were the most commonly found ADEs in
a prospective cohort of hospitalized children (Kaushal et al., 2001). A recent study of adverse
drug events in a paediatric intensive care unit reported that hyponatremia, hyperglycemia,
hypokalemia, and skin rashes were the most commonly observed ADEs (Silva et al., 2013).
However, most studies report the incidence of a specific list of adverse events defined a priori;
thus, studies will only capture the adverse events that were pre-specified and thus may not
capture all possible events associated with a specific drug.
33
While specific adverse events are not widely reported, a wide variety of drugs have been
associated with adverse drug events. A recent systematic review of observational studies
reported that anti-infectives, anti-epileptics, non-steroidal anti-inflammatories (NSAIDs) and
corticosteroids were the most frequent drug categories associated with adverse drug reactions
(Smyth et al., 2012). A prospective cohort study of 1 120 paedatric patients reported that
antibiotics, analgesics and sedatives were the most common drug categories involved in errors
and potential ADEs, followed by electrolytes and fluids (Kaushal et al., 2001). Reactions
involving antibiotics were usually mild whereas anticonvulsants and antineoplastic agents were
associated with more severe reactions. Other studies of adverse drug events in paediatric
populations have shown slightly differing results. A review of voluntary safety reports and
computerized surveillance of medication-related events at a paediatric hospital found that
nephrotoxins, narcotics and benzodiazepines and hypoglycaemia were the most frequent
categories of adverse drug events captured by these surveillance systems (Ferranti, Horvath,
Cozart, Whitehurst, & Eckstrand, 2008). An automated surveillance system developed to
identify adverse drug events in a paediatric hospital found hypokalemia, hypomagnesemia,
nephrotoxicity and opiate overdose identified through naloxone administration among the most
common ADEs detected during the 6-month study period (Kilbridge et al., 2009). The authors
also found that diuretics, antibiotics, immunosuppressants, narcotics and anticonvulsants were
the most common medications implicated in an ADE. These results are similar to those of Silva
et al (2013) who conducted a prospective chart review of consecutive admissions to the
paediatric ICU over a 6-month period and reported that antibiotics, diuretics, anti-epileptics,
sedatives and analgesics, and steroids were the most common classes of drugs associated with
adverse drug events.
34
2.3.3 Importance of Understanding and Identifying the Contribution of
Medications to Adverse Events
While there is extensive literature on drug safety and adverse drug events in adults, little is
known about medication safety in paediatric populations, despite their apparent increased risk.
Safety data describing adverse drug events in paediatric populations are not well described
(Gilman & Gal, 1992; Uppal, Dupuis, & Parshuram). More than 75% of pharmaceuticals
licensed in North America have never been tested in paediatric populations and are used without
adequate guidelines for safety or efficacy (Leeder, 2003; Uppal et al., 2008). Regardless, use of
medications without paediatric indication is common and almost considered to be standard
practice (Bavdekar & Gogtay, 2005; Blumer, 1999; Choonara & Conroy, 2002; Conroy &
McIntyre, 2005; Cuzzolin, Zaccaron, & Fanos, 2003; Dick et al., 2003; Kimland, Bergman,
Lindemalm, & Bottiger, 2007; Pandolfini, Campi, Clavenna, Cazzato, & Bonati, 2005;
Pandolfini et al., 2002). Off-label use of medications in paediatric medicine is reported to
account for approximately 50-75% of medication use in children (Roberts, Rodriguez, Murphy,
& Crescenzi, 2003). A recent retrospective review of administrative clinical data from a
neonatal intensive care unit (NICU) found that the United States Food and Drug Administration
approved only 35% of the medications routinely used in the NICU for use in infants (Hsieh et al.,
2013). Recommended doses of drugs used in children are often based on extrapolations from
adult doses using weight, body surface area and age and often ignore pharmacokinetic and
pharmacodynamics properties of the drug, increasing susceptibility to drug-related adverse
events (Hussain & Kao, 2005; Kaushal et al., 2001; Niederhauser, 1997).
A lack of safety information for children has resulted in the use of a number of therapies and
medications with deleterious effects. For example, oxygen therapy to improve breathing for
35
neonates in incubators was a widely accepted treatment in the 1930s (Zito et al., 2008). In the
1940s, increases in dosage and exposure length were routine practice despite a lack of evidence
supporting this practice. This resulted in a substantial increase in the incidence of retrolental
fibroplasia, now known as retinopathy of prematurity (ROP) (Kinsey & Zacharias, 1949).
Considerable debate followed, which suggested that prematurity was responsible for the disease;
finally in 1952, a study definitively linked increase oxygen with the development of retinopathy
of prematurity and blindness in premature babies (Patz, 1957).
Pemoline, used for the treatment of attention-deficit/hyperactivity disorder (ADHD), is another
example that clearly illustrates the challenges of drug safety for children. Early evidence of
hepatotoxicity was evident in adults; however, the relatively low use in children translated into
usage of the drug for over 21 years before it was associated with liver toxicity and fatalities in
children (Safer, Zito, & Gardner, 2001). Though warnings were added along with requirements
for biweekly liver enzyme monitoring, evidence suggested that prescribers were not compliant
with the safety directives (Willy, Manda, Shatin, Drinkard, & Graham, 2002). The drug was
voluntarily withdrawn in 2005; Health Canada withdrew approval in 1999 and the United States
Food and Drug Administration (FDA) withdrew approval in 2005 (Etwel, Rieder, Bend, &
Koren, 2008).
Extrapolating drug safety data from adult populations to paediatric populations is not
recommended. Key differences between adult and paediatric use include dose, frequency of
administration, route of administration and indication for use (Hussain & Kao, 2005), resulting in
different effects in children and adults. For example, milrinone increases mortality in adult
patients with heart failure, but is both safe and effective in children (Hoffman et al., 2002; Packer
et al., 1991). Numerous examples suggest that while adult safety data may drive drug
36
availability for children, treatment in paediatric populations differs and requires appropriate
safety and effectiveness evaluations. Even within paediatric populations clinically important
pharmacologic differences exist (Hussain & Kao, 2005; Kearns, 2000; Kearns et al., 2003).
Adding to this problem is the challenge that clinicians face in understanding the relative
contribution of drugs versus disease processes to suspected adverse events (Carleton et al.,
2009). Determining the occurrence of adverse events is a complex task as the symptoms of the
event may overlap with the patient’s underlying disease or may be caused by several unrelated
factors including unknown drug allergies of the patient or human error. Drug-related factors
such as toxicity, time of administration, dosage and duration of use can also impact the
probability of an ADE. In general, if ADEs are not specifically looked for, it is unlikely that they
will be identified (Kelly, 2008).
2.3.4 Drug Toxicity as a Cause of AKI
As a major function of the kidney is the excretion of metabolites and drugs, it is unsurprising that
the kidney is vulnerable to toxicity.
2.3.4.1 Factors Associated with Increased Toxicity
A number of factors increase the risk of nephrotoxic kidney injury; these factors can be thought
of as being patient-specific, kidney-specific or drug-specific (Perazella, 2009).
2.3.4.1.1 Patient-Specific Factors
Patients may be predisposed to develop kidney injury due to a number of factors such as age,
gender, body composition and underlying comorbid conditions. The genetic makeup of a patient
may also increase their vulnerability to nephrotoxins (Ciarimboli et al., 2005; Harty, Johnson, &
37
Power, 2006; Ulrich, Bigler, & Potter, 2006). Known as the study of pharmacogenetics, these
genetic differences between patients may explain the variations of responses to medications
across patients. The cytochrome P450 (CYP450) enzyme system has been well studied in
relation to hepatotoxicity. Several polymorphisms are associated with reduced metabolism and
subsequent end-organ toxicity. CYP450 enzymes are also found in the kidney; it is reasonable to
expect that certain gene expressions that reduce metabolism would increase nephrotoxic risk
(Perazella, 2009).
2.3.4.1.2 Kidney-Specific Factors
Kidney-specific factors, such as the high rate of blood delivery (approximately 20-25% of
cardiac output) and the high metabolic rate of tubular cells, promote increased sensitivity to
injury when the kidneys are exposed to a nephrotoxic substance (Cummings & Schnellmann,
2001; Perazella, 2009).
2.3.4.1.3 Drug-Specific Factors
Factors specific to the drug also play an important role in the development of renal injury.
Factors including urine pH, renal excretion of toxic metabolites and rapid parenteral or high
dosing (resulting in high peak serum and urine concentrations) enhance the risk of precipitation
and crystal formation in the tubules (Guo & Nzerue, 2002; Singh, Ganguli, & Prakash, 2003).
Concurrent drug administration may increase the risk of nephrotoxicity. For example,
combinations such as aminoglycosides and cephalothin or cisplatin and non-steroidal anti-
inflammatory drugs (NSAIDs) and radiologic contrast are examples of situations with increased
nephrotoxic risk to the patient (Evenepoel, 2004; Guo & Nzerue, 2002; Singh et al., 2003; Wyatt,
Arons, Klotman, & Klotman, 2006). Even with brief or low levels of exposure, some drugs,
38
such as aminoglycosides, amphotericin B, adefovir and cidofovir, are highly nephrotoxic and
promote renal injury (Alexander & Wingard, 2005; Gambaro & Perazella, 2003; Izzedine,
Launay-Vacher, & Deray, 2005; Perazella, 2005; Rougier et al., 2003; Schetz, Dasta, Goldstein,
& Golper, 2005).
Drugs can cause renal injury through several mechanisms (Table 2.4) (Pannu & Nadim, 2008;
Patzer, 2008; Perazella, 2012; Schetz et al., 2005). Most commonly, drugs that are excreted
through the renal system can directly damage renal tubules, causing cellular injury and death
(acute interstitial nephritis). Other types of nephrotoxic injury include osmotic nephrosis caused
by hypertonic solutions or tubular obstruction caused by drug precipitation (i.e. crystalline
nephropathy). While the specific mechanisms of cellular injury caused by these medications are
beyond the scope of this thesis work, a few examples are provided. For example, the use of
aminoglycosides can result in the accumulation of high concentrations of the drug in lysosomes
within the kidney. Once released into the cells, this interrupts the phospholipid cell membrane
and promotes oxidative stress and mitochondrial injury and results in proximal tubular cell death,
leading to acute kidney injury (Perazella, 2009). Amphotericin B disrupts the cellular
membranes, increasing their permeability, resulting in tubular dysfunction (Alexander &
Wingard, 2005). The metabolism of cidofovir produces a specific metabolite which interferes
with the synthesis and degradation of phospholipids within the cell membrane, resulting in
tubular injury (Izzedine et al., 2005; Perazella, 2005).
39
Table 2.4: Common drugs known to have nephrotoxic effects, categorized by mechanism of
injury
Mechanism of injury Medications
Pre-renal failure
ACE inhibitors cyclosporine A norephinephrine NSAIDs tacrolimus
Acute tubular necrosis
acetaminophen Antibiotics:
aminoglyocosides cephalosporins amphotericin B vancomycin foscarnet
cisplatinum contrast media cyclosporine A dextran immunoglobulin mannitol NSAIDs
Acute interstitial nephritis
acetylsalicylic acid (ASA) allopurinol Antibiotics:
ciprofloxacin methicillin penicillin ampicillin cephalosporins oxacillin rifampicin
contrast media furosemide NSAIDs phenytoin
Tubular obstruction
acyclovir methotrexate methoxyflurane protease inhibitors sulfonamides
Information from (Arany et al., 2008) ACE: angiotensin-converting-enzyme; NSAID: non-steroidal anti-inflammatory
40
Prolonged therapy at high doses may exacerbate any of the mechanisms described above based
solely on increased exposure (Perazella, 2009).
2.3.4.2 Incidence of Drug Toxicity as a Cause of AKI
Nephrotoxicity has been estimated to contribute 8% to 60% of all cases of hospital-acquired
AKI, depending on the patient population and definition of AKI (Davidman, Olson, Kohen,
Leither, & Kjellstrand, 1991; Hou, Bushinsky, Wish, Cohen, & Harrington, 1983; Hui-Stickle et
al., 2005; Kohli et al., 2000; Liano, Junco, Pascual, Madero, & Verde, 1998; Nash et al., 2002;
Sural et al., 2000). More recently, drugs have been estimated to contribute to 19-25% of AKI
cases in critically ill adults (R. L. Mehta et al., 2004; S. Uchino et al., 2005) and over 20% of the
100 most frequently prescribed medications in an adult tertiary care center were considered to be
potentially nephrotoxic (Taber & Mueller, 2006).
In paediatric patient populations, exposure to nephrotoxic medications has been estimated to be
the second most common cause of AKI, accounting for 16% of pediatric AKI (Hui-Stickle et al.,
2005). However, there is lack of paediatric data available regarding the actual incidence of
nephrotoxicity for many medications known to have deleterious effects on the kidneys.
Most studies focus solely on exposure to one medication. For example, Zappitelli et al (2011)
conducted a retrospective cohort study of non-critically ill children treated with aminoglycosides
(gentamicin, tobramycin or amikacin) for five or more days during their hospitalization. They
estimated the incidence of AKI in this cohort of children to be 20% according to the AKIN and
33% according to the pRIFLE definition of AKI. Totapally et al (2013) conducted a
retrospective cohort study of children admitted to the paediatric intensive care unit and received
41
vancomycin for three or more consecutive days. Forty-nine patients (17.2%) developed AKI
according to the GFR criteria of the pRIFLE.
To date, the only comprehensive assessment of nephrotoxic medication-related acute kidney
injury was conducted in hospitalized, non-critically ill pediatric patients (Moffett & Goldstein,
2011). The authors performed a retrospective case-control study; cases were defined as patients
who developed any degree of AKI according to the pRIFLE and were matched 1:1 to controls
according to age, gender and admission diagnosis. The authors collected information regarding
the administration of 34 medications deemed to have nephrotoxic effects including medication
exposure, total number of doses, days of medication therapy, exposure intensity, doses per
therapy day and doses per admission day. Three hundred and fifty-seven case-control pairs were
studied. Patients with AKI had significantly greater odds to exposure to one or more nephrotoxic
medications than patients without AKI (OR=1.7, 95% CI: 1.0-2.9). Patients who developed AKI
had exposure to more nephrotoxic medications for a longer period of time than their matched
controls. The authors found that the use of amphotericin, cefotaxime, cisplatin, ifosfamide,
piperacillin with or without tazobactam and vancomycin were independently associated with the
development of AKI. When all medications were included in a multivariate regression, only
vancomycin had a statistically significant association with the development of AKI. A similar
assessment has not been conducted in critically ill children, and thus, the risk of nephrotoxic
medication-induced AKI in critically ill children remains unknown.
42
2.4 Studying Drug Safety in Children
Before regulatory approval, all prescription drugs must go through rigorous study, including
randomized controlled trials, to ensure efficacy and safety. Only 20% of the drugs used to treat
paediatric patients have gained regulatory approval for use in these populations ("Regulations
Requiring Manufacturers to Assess the Safety and Effectiveness of New Drugs and Biological
Products in Pediatric Patients," 1997).
Until recently, children and pregnant women were generally excluded from drug studies, mainly
for reasons concerning the vulnerability of the patients and questions surrounding their ability to
truly provide informed consent. In 1977, the American Academy of Pediatrics argued of the
importance of studying drug safety in children ("American Academy of Pediatrics. Committee
on Drugs. Guidelines for the Ethical Conduct of Studies to Evaluate Drugs in Pediatric
Populations," 1977). They stated that both scientifically and ethically sound studies of drugs
could be conducted in paediatric populations. In fact, they argued that it was more unethical to
exclude children from study as it meant that each time a child was prescribed a medication,
physicians were essentially conducting an uncontrolled experiment on their patients. These
guidelines were updated again in 1995 ("Guidelines for the Ethical Conduct of Studies to
Evaluate Drugs in Pediatric Populations. Committee on Drugs, American Academy of
Pediatrics," 1995).
Legislation was enacted in the United States to improve drug safety knowledge in children,
specifically to include children in manufacturer-led drug trials. Enacted in 2002, the Best
Pharmaceuticals for Children Act encouraged drug manufacturers to conduct paediatric studies
in exchange for additional patent exclusivity ("About the BCPA" 2011; Breslow, 2003; United
43
States Government Accountability Office Report to Congressional Commmittees, 2007). It also
provides public funding and organizes private funding to conduct research on drug that
pharmaceutical companies choose to not test in children. The Pediatric Research Equity Act was
also enacted; it authorizes the FDA to require paediatric safety assessments of drugs (Thaul,
2012). It has been estimated that these pieces of legislations have allowed more drugs to be
studied in children in the last decade than in the preceding five decades (McMaul & Van Hollen,
2011).
Despite these efforts, there is still a lack of randomized clinical drug trials involving paediatric
patients. This is likely due to a number of factors: for example, children account for only a small
percentage of drug prescriptions so the cost of conducting a randomized trial may be prohibitive
for a drug company (t Jong et al., 2001). Paediatric patients may also be more difficult to recruit,
as parents may be concerned about the potential effects of experimental drug on their child’s
growth and development (t Jong et al., 2001). Even short-term exposure to a drug may have
long-term effects; these long-term drug effects are usually unforeseeable and hard to investigate
(Greenhill et al., 2003).
Even among the drugs that are licensed for paediatric use, fundamental gaps still exist in our
understanding of the safety of these drugs in children. Pre-marketing drug trials, in general,
recruit between a few hundred to a few thousand subjects (Strom, 2012); this sample size is
likely too small to detect rare, but serious, adverse drug events (Strom, 2012).
44
2.4.1 Studying Acute Kidney Injury as an Adverse Drug Event:
Pharmacoepidemiology
Pharmacoepidemiology is the study of the use of and effects of drugs in large patient populations
(Strom, 2012). This field has primarily concerned itself with the study of adverse drug events.
Traditionally, adverse drug reactions have been studied through reports of drug-related mortality
and morbidity; however, determining causation in these cases can be problematic. In the field of
pharmacoepidemiology, controlled studies are performed to examine whether the adverse
outcome of interest (in this case, acute kidney injury) occurs more often in exposed patients than
in those unexposed to certain medications. In the case of paediatric studies, this approach is
beneficial due to the smaller patient populations available for study. As children are generally
excluded from pre-market drug trials, this technique can provide needed safety information
regarding drug use in children.
In comparison to randomized controlled trials, pharmacoepidemiologic studies offer advantages
in studying adverse drug events in children (Luo, Doherty, Cappelleri, & Frush, 2007). For
example, pharmacoepidemiologic studies tend to have large sample sizes, which greatly increase
the change of detecting rare adverse events. In addition, compared to randomized trials, which
have rigorous inclusion and exclusion criteria, pharmacoepidemiologic studies tend to be more
generalizable as they study participants in real-world settings (Luo et al., 2007). The feasibility
of pharmacoepidemiological studies as increased with the development of administrative
healthcare databases, based on billing claims or electronic medical records (Luo et al., 2007).
The use of existing data also eliminates time and resources required for primary data collection.
45
This dissertation follows a pharmacoepidemiologic approach using AKI as a model to study
adverse drug events in children. We will exploit the unique electronic data available from the
Department of Critical Care Medicine at The Hospital for Sick Children where all clinical data
for any child admitted to the PICU or CICU is charted electronically. In addition, information
from other departments in the hospital, such as radiology and microbiology can be linked to the
chart data based on the patients’ unique medical record number. Should this method prove
successful, this method can be expanded to incorporate other ADEs of interest.
46
Chapter 3 A Systematic Review of RIFLE Criteria in Children and Its Application and Association with Measures of Mortality and Morbidity
An abridged version of the following chapter has been previously published. The citation is:
Slater MB, Anand V, Uleryk EM, Parshuram CS. A systematic review of RIFLE criteria in
children and its application and association with measures of mortality and morbidity. Kidney
International 2012; 81(8): 791-798.
Permission was received to publish in this dissertation.
47
3.1 Overview
The overall goal of this thesis was to evaluate the contribution of drug therapy to acute kidney
injury (AKI). To accomplish this, we first needed to understand how AKI is defined in the
literature. As stated in the previous chapter, we chose to focus on a specific definition of AKI
known as the RIFLE. At the time of this work, three different definitions of AKI were used in
the literature: the RIFLE (Bellomo et al., 2004), the pRIFLE (Akcan-Arikan et al., 2007) and the
AKIN (R. L. Mehta et al., 2007). The RIFLE had been shown to capture more mild cases of
acute kidney injury than the AKIN (Zappitelli et al., 2011) and the use of the AKIN classification
was not shown to improve outcome prediction over the RIFLE (Bagshaw et al., 2008; Chang et
al., 2010). The pRIFLE, the proposed modification to the RIFLE for use in paediatric
populations (Akcan-Arikan et al., 2007), provides limited justification for the changes made to
the original RIFLE (Akcan-Arikan et al., 2007). As such, the RIFLE was chosen to define AKI
for the thesis work.
The objective of this chapter is to systematically review the use of the RIFLE as a definition of
AKI in paediatric patient populations. We were interested in understanding how the RIFLE was
operationalized in the literature and apply any lessons learned in our utilization of the criteria as
an outcome. In addition, the evidence of an association between the RIFLE and outcomes of
mortality and morbidity are reported. The outcomes from this systematic review were used to
inform the operationalization of the RIFLE as an outcome for the remainder of the thesis work.
48
3.2 Introduction
Acute kidney injury (AKI) is a common adverse event in hospitalized patients, associated with
increased mortality, length of stay, and resource utilization (Chertow, Burdick, Honour,
Bonventre, & Bates, 2005; de Mendonca et al., 2000; F. B. Plotz et al., 2005; Ricci et al., 2008).
The use of over 35 definitions of AKI has contributed to wide variation in reported incidence
rates and has complicated comparisons between studies (Bellomo, Kellum, & Ronco, 2001;
Bellomo et al., 2004; Ricci et al., 2008). Two main definitions of AKI have been adopted: the
RIFLE (Bellomo et al., 2004) and the Acute Kidney Injury Network (AKIN) (R. L. Mehta et al.,
2007) criteria, both based on common markers of kidney function. Given evidence that the
RIFLE captures more mild cases of acute kidney injury than the AKIN (Zappitelli et al., 2011)
and the AKIN classification does not improve outcome prediction over the RIFLE (Bagshaw et
al., 2008; Chang et al., 2010), the focus of this review is on the RIFLE definition of AKI.
The RIFLE criteria were developed by an international consensus panel and are intended for use
in critically ill adults (Bellomo et al., 2004). The RIFLE classifies increasing severity of acute
kidney injury into 5 categories: risk (R), injury (I), and failure (F), loss of kidney function (L),
and end-stage kidney disease (E). The classification is based on the magnitude and duration of
changes from the patient’s normal (baseline) kidney function, assessed by glomerular filtration
rate (GFR) and urine output, and the length of renal replacement therapy (RRT). When a
measure of the patient’s baseline GFR is not available or is unknown, the panel suggested
assuming a GFR of 75ml/min/1.73m2, the lower limit of normal (Bellomo et al., 2004).
The RIFLE has been widely adopted in adult populations and is now being used as a ‘gold
standard’ to validate new biomarkers of kidney function (Haase et al., 2009). However, a
49
systematic review of the RIFLE criteria and mortality in adults, which found increasing RIFLE
severity was associated with increasing mortality, revealed that the application of the criteria was
inconsistent (Ricci et al., 2008). Variations in the application of the RIFLE, particularly in the
estimation methods for baseline renal function, can significantly affect the reported incidence of
AKI (Siew et al., 2010; Zappitelli et al., 2008; Zavada et al., 2010).
The RIFLE criteria are now being used to describe acute kidney injury in children and a
modification of the RIFLE (the pRIFLE) has been suggested for use in paediatric populations
(Akcan-Arikan et al., 2007). The modifications are minor and include a focus on the estimated
creatinine clearance as the measure of GFR, the threshold for the ‘Failure’ category from a serum
creatinine ≥ 4 mg/100ml to an estimated creatinine clearance (eCCl) < 35 ml/min/1.73m2, and
increasing the time interval for the urine output criteria to 8 hours from 6 hours (Table 3.1). The
increasing use of the RIFLE classification underscores the importance of consistent application
and raises a need to understand if the relationships between the RIFLE criteria and mortality and
morbidity found in adult patients also exist in paediatric patients. We conducted a systematic
review to describe the application of the RIFLE and to assess associations between the RIFLE
and measures of mortality and morbidity, specifically length of stay and measures of illness
severity and kidney function, in paediatric populations.
50
Table 3.1: RIFLE criteria and suggested paediatric modification (pRIFLE)
RIFLE (Bellomo et al., 2004) Modification (pRIFLE) (Akcan-Arikan et al., 2007)
GFR criteria Urine output criteria
GFR criteria Urine output criteria
Risk Increased creatinine x1.5 or GFRa decrease >25%
<0.5 ml/kg/h x 6h
eCClb decrease by 25%
<0.5 ml/kg/h x 8h
Injury Increased creatinine x2 or GFR decrease >50%
<0.5 ml/kg/h x 12h
eCCl decrease by 50%
<0.5 ml/kg/h x 16h
Failure Increased creatinine x3 or GFR decrease >75% or creatinine ≥4 mg/100ml (acute rise of ≥0.5 mg/100ml dl)
<0.3 ml/kg/h x 24h or anuria x 12h
eCCl decrease by 75% or eCCl <35 ml/min/1.73m2
<0.3 ml/kg/h x 24h or anuria x 12h
Loss Persistent ARF = complete loss of renal function > 4 weeks (defined as the need for renal replacement therapy (RRT) for >4 weeks)
Persistent failure >4 weeksc
RRT / Dialysis Criteria
End stage
End-stage renal disease (defined as the need for dialysis for >3 months)
End-stage renal disease (persistent failure >3 months)c
ARF: acute renal failure; eCCl: estimated creatinine clearance; GFR: glomerular filtration rate; RRT: renal replacement therapy a Estimated using the Modification for Diet in Renal Disease (MDRD) formula; if no baseline measurement available, assign a lower limit of normal GFR of 75ml/min/1.73m2; b Estimated using the Schwartz formula; if no baseline measurement available, assign 100ml/min/1.73m2
c The definitions of persistent failure were not described; it is assumed that they are the same as defined by the original RIFLE
3.3 Methods
A literature search was conducted in OVID SP MEDLINE and EMBASE to identify peer-
reviewed publications indexed with kidney injury (or disease) that specifically mentioned the
RIFLE score. The search was restricted to studies involving children (Table 3.2). Studies that
51
applied the RIFLE as a measure of AKI and/or described the reliability, validity, or
responsiveness of the RIFLE were eligible for inclusion. Non-English studies and non-original
investigations including letters, commentaries, editorials, conference abstracts, and literature
reviews were excluded. Two reviewers (MS, VA) screened the abstracts independently to
identify eligible studies and final inclusion was determined by consensus. If consensus could not
be reached, a third reviewer (CP) determined eligibility.
Table 3.2: Search strategy
OVID
1 renal insufficiency/ or kidney failure/ or kidney failure, acute/ or kidney tubular necrosis, acute/ or renal insufficiency, acute/ or Kidney Diseases/
94 748
2 (rifle or prifle).ti,ab. 653
3 1 and 2 204
4 Limit 4 to “all child (0 to 18 years)” 28
EMBASE
1 kidney failure/ or acute kidney failure/ or acute kidney tubule necrosis/ or anuria/ or kidney cortex necrosis/ or kidney tubule necrosis/ or oliguria/ or renal osteodystrophy/ or uremia/ or kidney disease/ or kidney injury/ or (renal adj2 insufficien*).mp. or exp kidney function/ or exp dialysis/ or exp renal replacement therapy/
363 833
2 (rifle or prifle or (risk adj2 injury adj2 failure adj2 loss adj2 end adj2 stage adj2 disease)).ti,ab.
973
3 1 and 2 406
4 Child/ 1 015 978
5 3 and 4 29
TOTAL NUMBER OF STUDIES (excluding duplicates) 45
52
For each article, two reviewers (MS, VA) independently abstracted data to describe: [1] the
study: year of publication, study population (including inclusion and exclusion criteria), number
of children studied, study design, and rationale for inclusion; [2] the application of the RIFLE:
the version of the RIFLE criteria cited (the original RIFLE or the pRIFLE modification), the
criteria used to determine RIFLE classification, the range of RIFLE categories reported, and the
number and proportion of patients in each category; [3] how the RIFLE data were measured: the
definition of baseline GFR estimate, and, if applicable, the method for handling missing baseline
values; and [4] associations between the RIFLE and measures of mortality and morbidity. These
included ICU and hospital mortality and length of stay, accepted illness severity scales or scores,
use of mechanical ventilation, other measures of kidney function (AKIN classification and the
use of dialysis or RRT), and the use of nephrotoxic medications. We did not consider
associations between the RIFLE and novel markers of renal function.
The number and percentages of manuscripts with each characteristic assessed were tabulated and
reported. No measures of association were presented due to significant heterogeneity in both
study design and population.
3.4 Results
The search identified 45 articles, 12 of which met the inclusion/exclusion criteria (Ajami et al.,
2010; Akcan-Arikan et al., 2007; Dennen et al., 2010; Duzova et al., 2010; Manrique et al., 2009;
Palmieri et al., 2009; Frans B. Plotz, Bouma, van Wijk, Kneyber, & Kenkamp, 2008; Schneider
et al., 2010; Washburn et al., 2008; Zappitelli et al., 2009; Zappitelli et al., 2011; Zappitelli et al.,
2007) (Figure 3.1). All studies described the application of the RIFLE criteria in a paediatric
53
patient population and 10 (83%) described associations between the RIFLE and mortality and
morbidity (Akcan-Arikan et al., 2007; Dennen et al., 2010; Duzova et al., 2010; Manrique et al.,
2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al., 2010; Zappitelli et al.,
2009; Zappitelli et al., 2011; Zappitelli et al., 2007) (Table 3.3). The total number of patients per
study ranged from 25 (Dennen et al., 2010) to 3,396 (Schneider et al., 2010), all of whom were
hospital inpatients. Six studies were conducted prospectively (Ajami et al., 2010; Akcan-Arikan
et al., 2007; Dennen et al., 2010; Duzova et al., 2010; Washburn et al., 2008; Zappitelli et al.,
2007); the remaining six were conducted retrospectively (Manrique et al., 2009; Palmieri et al.,
2009; Frans B. Plotz et al., 2008; Schneider et al., 2010; Zappitelli et al., 2009; Zappitelli et al.,
2011). The one multicenter study included 17 centers (Duzova et al., 2010).
Patient populations varied and included, for example, all paediatric ICU admissions (Schneider
et al., 2010), specific ICU patients (i.e. those with respiratory and/or cardiac failure (Akcan-
Arikan et al., 2007; Frans B. Plotz et al., 2008; Washburn et al., 2008; Zappitelli et al., 2007)),
and surgical patients (Dennen et al., 2010; Manrique et al., 2009; Zappitelli et al., 2009). A
RIFLE severity of ‘Risk’ or higher was reported in 10% (Schneider et al., 2010) to 100%
(Duzova et al., 2010) of patients (Figure 3.2).
54
Figure 3.1: Flow diagram of included and excluded studies Forty-five studies, excluding duplicates were identified by the literature search; 14 were excluded as they were conference abstracts. Other exclusions were due to language (1), lack of original research (editorials, 4; systematic review, 1) and 13 did not meet our study objectives. Overall, 12 studies met the inclusion/exclusion criteria and were included in this systematic review.
!28
OVID SP Medline
29 EMBASE
45 Studies (excl. duplicates)
14 Conference abstracts
1 Non-English
1 Systematic review
13 Did not meet objectives
4 Editorials
12 Meeting criteria
55
Table 3.3: Summary of studies
Author Year Population N Retrospective/ prospective
Single/multi- center
Measures of association
Ajami (2010)
2010 Paediatric patients referred for diagnostic or interventional catheterization. Exclusion criteria mentioned but not defined.
80 Prospective Single
Akcan-Arikan (2007)
2007 PICU patients with respiratory (defined as requiring invasive mechanical ventilation) and/or cardiac failure (defined as requiring infusions of vasoactive medications) who had indwelling urinary catheters, excluding patients with known end-stage renal disease at PICU admission or were post-renal transplant status.
150 Prospective Single
Dennen (2010)
2010 Children undergoing scheduled first time cardiopulmonary bypass for repair of congenital heart disease, excluding those with known underlying chronic kidney disease, exposure to nephrotoxins within one week of surgery, proteinuria, urinary tract infection, diabetes, unavailable baseline serum creatinine, inability to obtain consent.
25 Prospective Single
Duzova (2010)
2010 Centers identified any patient with AKI (defined as an absolute increase in SCr by either >0.3 mg/dl or an increase of ≥50% from baseline or a GFR decrease ≥25% from baseline or a reduction in urine output (<0.5 ml/kg for more than 8h) at the time of admission or during treatment or those who were acutely ill.
472 Prospective Multicenter
Manrique (2009)
2009 Patients undergoing congenital cardiac surgery requiring cardiopulmonary bypass.
395 Retrospective Single
Palmieri (2009)
2009 Patients with burn injury (TBSA>10%) admitted to Burn ICU, excluding non-survivable burns (decision for comfort care on admission), admission for non-burn diagnosis, or burn size less than 10%.
123 Retrospective Single
56
Plotz (2008)
2008 PICU patients with respiratory failure (defined as requiring >4 days mechanical ventilation).
103 Retrospective Single
Schneider (2010)
2010 All patients admitted to PICU excluding those with preexisting chronic renal insufficiency, end-stage renal disease, or admission for renal transplantation.
3396 Retrospective Single
Washburn (2008)
2008 PICU patients requiring mechanical ventilation and indwelling bladder catherization, excluding patients with known end-stage renal disease at PICU admission or were post-renal transplant status (same as Akcan-Arikan(2007)). Patients with less than 2 SCr levels or those with no urine specimens were also excluded.
137 Prospective Single
Zappitelli (2007)
2007 PICU patients requiring mechanical ventilation and indwelling bladder catherization, excluding patients with known end-stage renal disease at PICU admission or were post-renal transplant status (same as Akcan-Arikan(2007)).
140 Prospective Single
Zappitelli (2009)
2009 Patients undergoing open chest surgery excluding patients with no SCr measured during their ICU stay.
390 Retrospective Single
Zappitelli (2011)
2011 Patients hospitalized in non-intensive care units and received gentamicin, tobramycin, or amikacin for ≥5 days, excluding patients with primary diagnoses of genitourinary or renal disorders at hospital admission.
557 Retrospective Single
GFR: glomerular filtration rate; ICU: intensive care unit; PICU: pediatric intensive care unit; SCr: serum creatinine; TBSA: total body surface area
57
Figure 3.2: Reported incidence and distribution of RIFLE-assessed AKI
The rates of AKI reported in each study ranged from 10% to 100% of patients. The distribution of severity is also shown and varies across studies. Dennen et al (2010) did not report RIFLE categories separately and thus is omitted.&
3.4.1 Application of the RIFLE
Four studies (25%) cited using the original RIFLE (Ajami et al., 2010; Dennen et al., 2010;
Manrique et al., 2009; Schneider et al., 2010) and the remaining 8 (75%) referenced the
proposed paediatric modification (Akcan-Arikan et al., 2007; Duzova et al., 2010; Palmieri et
al., 2009; Frans B. Plotz et al., 2008; Washburn et al., 2008; Zappitelli et al., 2009; Zappitelli
et al., 2011; Zappitelli et al., 2007). All studies used the GFR criteria to define and classify
acute kidney injury (Table 3.4) and five used the GFR criteria in combination with the urine
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Risk Injury Failure Loss/End stage
58
output criteria (Akcan-Arikan et al., 2007; Duzova et al., 2010; Manrique et al., 2009;
Palmieri et al., 2009; Frans B. Plotz et al., 2008). No study relied solely on urine output to
classify AKI. One study (8%) only reported AKI as defined by a RIFLE category of ‘R’ or
higher (Dennen et al., 2010), 10 studies (83%) reported the first three clinical categories
(Risk, Injury, and Failure) (Ajami et al., 2010; Akcan-Arikan et al., 2007; Manrique et al.,
2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al., 2010; Washburn et
al., 2008; Zappitelli et al., 2009; Zappitelli et al., 2011; Zappitelli et al., 2007) and one study
(8%) used all of the RIFLE categories (Duzova et al., 2010).
3.4.1.1 Missing Data
In nine studies (75%), it was unclear if patients were excluded if they were missing data
necessary to determine RIFLE classification and, in the event data were missing, it was
unclear how this was handled (Ajami et al., 2010; Akcan-Arikan et al., 2007; Dennen et al.,
2010; Manrique et al., 2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al.,
2010; Zappitelli et al., 2011; Zappitelli et al., 2007). Two studies purposefully excluded
patients: Zappitelli et al (2009) stated that patients with no serum creatinine measurements
during the study period were excluded and Washburn et al (2008) excluded patients with less
than 2 serum creatinine measurements. It is unlikely that Duzova et al (2010) purposefully
excluded patients as they state that only 423 (89.6%) of the 472 patients enrolled in the study
had a RIFLE classification available.
59
Table 3.4: Summary of RIFLE application
Author Year Tool cited (RIFLE, pRIFLE)
Criteria used (GFR, UO, both)
GFR Criteria Urine Output Criteria
GFR method (eCCl, Cr)
Definition of baseline creatinine
Method for missing baseline creatinine Timing Urinary
Catherization
Ajami (2010)
2010 RIFLE GFR eCCl (Schwartz)
Not specified Not specified N/A N/A
Akcan-Arikan (2007)
2007 pRIFLE Both eCCl (Schwartz)
Lowest SCr in 3m proceeding ICU admission
Assigned baseline eCCl=100ml/min/1.73m2
x8 hours
Indwelling Foley catheter an inclusion criteria
Dennen (2010)
2010 RIFLE GFR Cr Pre-operative measurement (not defined more specifically)
Patients with missing baseline values were excluded.
N/A N/A
Duzova (2010)
2010 pRIFLE Both eGFR (estimation method not specified)
Not specified Not specified x8 hours
Not specified
Manrique (2009)
2009 RIFLE Both Not specified
Not specified Not specified x6 hours
Not specified
Palmieri (2009)
2009 pRIFLE Both eCCl (Schwartz)
eCCl calculated using SCr within 3m prior to admission
Assigned eCCl=120ml/min/1.73m2
x8 hours
Not specified
Plotz (2008)
2008 pRIFLE Both eCCl (Schwartz)
SCr within 3m of ICU admission
Assigned eCCl=100ml/min/1.73m2
x8 hours
Not specified
60
Schneider (2010)
2010 RIFLE GFR Cr Most recent documented SCr within 6m prior to ICU admission
Assigned an age/gender appropriate high-end normal SCr
N/A N/A
Washburn (2008)
2008 pRIFLE GFR eCCl (Schwartz)
Not specified Not specified N/A N/A
Zappitelli (2007)
2007 pRIFLE GFR eCCl (Schwartz)
Lowest known SCr value during the 3m prior to study enrollment.
For patients with no known baseline SCr and PICU admission eCCL> 120ml/min/1.73m2, baseline eCCl was recorded as admission eCCl.
Assigned eCCl=120ml/min/1.73m2
N/A N/A
Zappitelli (2009)
2009 pRIFLE GFR Cr Pre-operative SCr (within 1w of surgery) or lowest in previous month
Assigned minimum norms for age and gender
N/A N/A
Zappitelli (2011)
2011 pRIFLE GFR eCCl (Schwartz)
Lowest SCr within 3 months prior to treatment initiation
Assigned eCCl=120ml/min/1.73m2
N/A N/A
Cr: creatinine; eCCl: estimated creatinine clearance; eGFR: estimated glomerular filtration rate; GFR: glomerular filtration rate; SCr: serum creatinine; UO: urine output
61
3.4.1.2 GFR Criteria
The GFR criteria were assessed by relative changes in estimated creatinine clearance (n=8)
(Ajami et al., 2010; Akcan-Arikan et al., 2007; Duzova et al., 2010; Palmieri et al., 2009;
Frans B. Plotz et al., 2008; Washburn et al., 2008; Zappitelli et al., 2011; Zappitelli et al.,
2007), changes in serum creatinine (n=3) (Dennen et al., 2010; Schneider et al., 2010;
Zappitelli et al., 2009), or were not specified (n=1) (Manrique et al., 2009). Of those that
used eCCl, 7 (88%) specified that the estimated creatinine clearance was calculated using the
Schwartz formula (Ajami et al., 2010; Akcan-Arikan et al., 2007; Palmieri et al., 2009; Frans
B. Plotz et al., 2008; Washburn et al., 2008; Zappitelli et al., 2011; Zappitelli et al., 2007);
one study did not specify the estimation method (Duzova et al., 2010). The Schwartz method
requires the patient’s height along with a serum creatinine measurement (Schwartz, Haycock,
Edelmann, et al., 1976). No study discussed the availability or accuracy of height
measurements or how the absence of these data was managed.
Definitions of the baseline measurement of renal function varied; 5 studies used creatinine
measured within 3 months of admission or study initiation (Akcan-Arikan et al., 2007;
Palmieri et al., 2009; Frans B. Plotz et al., 2008; Zappitelli et al., 2011; Zappitelli et al.,
2007), one used the most recent measure within 6 months of admission (Schneider et al.,
2010), and two used pre-operative serum creatinine measures: one within one week of
surgery or the lowest measure available in the month prior to surgery (Zappitelli et al., 2009),
the other did not specify the time period (Dennen et al., 2010). Four studies did not specify
the baseline definition (Ajami et al., 2010; Duzova et al., 2010; Manrique et al., 2009;
Washburn et al., 2008).
62
Missing baseline GFR values were managed as follows: five assigned values of either
100ml/min/1.73m2 (Akcan-Arikan et al., 2007; Frans B. Plotz et al., 2008) or
120ml/min/1.73m2 (Palmieri et al., 2009; Zappitelli et al., 2011; Zappitelli et al., 2007) for
estimated creatinine clearance. Two studies used age- and gender-adjusted normal values for
missing baseline creatinine values; one used high-end normal values (Schneider et al., 2010)
and the other used the lower limit of normal (Zappitelli et al., 2009). Four studies did not
specify how missing baseline GFR values were accounted for (Ajami et al., 2010; Duzova et
al., 2010; Manrique et al., 2009; Washburn et al., 2008) and one study excluded patients with
missing baseline values (Dennen et al., 2010).
3.4.1.3 Urine Output Criteria
Of the five studies that applied the urine output criteria (Akcan-Arikan et al., 2007; Duzova et
al., 2010; Manrique et al., 2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008), only one
study included patients with indwelling urinary catheters (Akcan-Arikan et al., 2007). Urine
output was assessed in 6 hour intervals according to the RIFLE in one study (Manrique et al.,
2009); four studies used 8 hour intervals, according to the pRIFLE (Akcan-Arikan et al.,
2007; Duzova et al., 2010; Palmieri et al., 2009; Frans B. Plotz et al., 2008).
3.4.2 Associations Between the RIFLE and Measures of Mortality and
Morbidity
Ten studies assessed relationships between the RIFLE and measures of mortality, length of
stay, and illness severity (Akcan-Arikan et al., 2007; Dennen et al., 2010; Duzova et al.,
2010; Manrique et al., 2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al.,
2010; Zappitelli et al., 2009; Zappitelli et al., 2011; Zappitelli et al., 2007) (Table 3.5). To
63
maintain consistency and comparability of results across studies, only unadjusted analyses are
presented below.
3.4.2.1 Mortality
Six studies evaluated the association between mortality and RIFLE-assessed AKI (Akcan-
Arikan et al., 2007; Duzova et al., 2010; Palmieri et al., 2009; Frans B. Plotz et al., 2008;
Schneider et al., 2010; Zappitelli et al., 2007). Two studies showed that AKI, defined as a
RIFLE severity of ‘Risk’ or higher, was significantly associated with increased mortality
(Frans B. Plotz et al., 2008; Schneider et al., 2010); one study found this relationship to be
borderline significant (p=0.057) (Palmieri et al., 2009). Two studies found a non-significant
relationship when the RIFLE was dichotomized in this manner, but found significant
increases in mortality when using different cut points (i.e. ‘Injury’ or higher) (Akcan-Arikan
et al., 2007; Zappitelli et al., 2007). Of the four studies that assessed mortality across RIFLE
strata (Duzova et al., 2010; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al.,
2010), two showed significant differences (Palmieri et al., 2009; Schneider et al., 2010) while
the other two studies found no significant difference in mortality rates across RIFLE
categories (Duzova et al., 2010; Frans B. Plotz et al., 2008).
3.4.2.2 Length of Stay (LOS)
Seven studies evaluated the association between RIFLE-assessed AKI and length of stay
(LOS) (Akcan-Arikan et al., 2007; Dennen et al., 2010; Palmieri et al., 2009; Frans B. Plotz
et al., 2008; Schneider et al., 2010; Zappitelli et al., 2009; Zappitelli et al., 2011); five studies
showed that AKI, defined as ‘Risk’ or higher, was significantly associated with longer
hospital LOS (Akcan-Arikan et al., 2007; Dennen et al., 2010; Palmieri et al., 2009;
64
Zappitelli et al., 2009; Zappitelli et al., 2011) and longer ICU stay (Dennen et al., 2010;
Palmieri et al., 2009; Schneider et al., 2010; Zappitelli et al., 2009), though two studies
showed no significant relationship with ICU LOS (Akcan-Arikan et al., 2007; Frans B. Plotz
et al., 2008). Two studies assessed the relationship between LOS and RIFLE strata and no
significant association was found for hospital (Palmieri et al., 2009) or ICU LOS (Palmieri et
al., 2009; Schneider et al., 2010).
3.4.2.3 Illness Severity
Five studies compared the RIFLE with a total of four validated illness severity scores
(Akcan-Arikan et al., 2007; Palmieri et al., 2009; Schneider et al., 2010; Zappitelli et al.,
2009; Zappitelli et al., 2007). Three studies reported associations between the Pediatric Risk
of Mortality Score (PRISM) (Akcan-Arikan et al., 2007; Palmieri et al., 2009; Zappitelli et
al., 2007); Akcan-Arikan et al found a non-significant association between a dichotomous
measure of AKI and a significant increase in PRISM scores across RIFLE strata (Akcan-
Arikan et al., 2007) while Palmieri et al found a significant association of AKI with PRISM
but no significant difference across RIFLE strata (Palmieri et al., 2009). Zapitelli et al did
show that PRISM scores increased progressively with increasing RIFLE (Zappitelli et al.,
2007). Pediatric Index of Mortality (PIM2) scores were higher in patients presenting with
AKI at ICU admission (Schneider et al., 2010). Surgical severity, as measured by the
Aristotle score, was significantly higher in patients with AKI, but severity according to the
Risk Adjustment in Congenital Heart Surgery (RACHS-1) score was not associated with
RIFLE (Zappitelli et al., 2009).
65
Table 3.5: Summary of associations between the RIFLE and measures of mortality and morbidity
Akc
an-A
rikan
(200
7)
Den
nen
(201
0)
Duz
ova
(201
0)
Man
rique
(200
9)
Palm
ieri
(200
9)
Plot
z (2
008)
Schn
eide
r (20
10)
Zapp
itelli
(200
7)
Zapp
itelli
(200
9)
Zapp
itelli
(201
1)
AKI (RIFLE severity of ‘Risk’ or higher) Mortality - - + + -
Length of stay ICU - + + - + + Hospital + + + + +
Illness severity
PRISM score + + PIM2 score + Aristotle score + RACHS-1 score - Mechanical ventilation - + - + +
Kidney function AKIN RRT + +
Risk of injury Nephrotoxic medications - + + RIFLE categories Mortality - + - +
Length of stay ICU - - Hospital -
Illness severity
PRISM score + - + PIM2 score Aristotle score RACHS-1 score Mechanical ventilation -
Kidney function AKIN + RRT - -
Risk of injury Nephrotoxic medications + AKIN: Acute Kidney Injury Network; ICU: intensive care unit; PIM: Pediatric Index of Mortality; PRISM: Pediatric Risk of Mortality; RRT: renal replacement therapy Note: Significant association denoted by +; non-significant association denoted by -. Blank means that this association was not reported.
66
Five studies assessed the relationship between RIFLE and length of mechanical ventilation
(Manrique et al., 2009; Palmieri et al., 2009; Frans B. Plotz et al., 2008; Schneider et al.,
2010; Zappitelli et al., 2009); significantly longer lengths of mechanical ventilation in
patients with AKI were reported by 3 studies (Palmieri et al., 2009; Schneider et al., 2010;
Zappitelli et al., 2009); the remaining studies found non-significant associations (Manrique et
al., 2009; Frans B. Plotz et al., 2008). No significant difference across RIFLE categories was
found (Palmieri et al., 2009).
3.4.2.4 Measures of Kidney Function
One study compared the RIFLE GFR criteria with the AKIN measurement for AKI and found
a high level of agreement between the two methods (84.4%) (Zappitelli et al., 2011).
Four studies assessed the relationship between RIFLE and dialysis use (Duzova et al., 2010;
Frans B. Plotz et al., 2008; Zappitelli et al., 2009; Zappitelli et al., 2007). Two studies only
assessed a dichotomous measure of AKI and reported that dialysis use and renal replacement
therapy are significantly higher in patients with AKI (Frans B. Plotz et al., 2008; Zappitelli et
al., 2009). The remaining studies both reported an increase in dialysis use across RIFLE
strata, though these trends were not significant (Duzova et al., 2010; Zappitelli et al., 2007).
3.4.2.5 Risk of Kidney Injury
The relationship between use of nephrotoxic medications and the RIFLE was evaluated in
three studies (Palmieri et al., 2009; Zappitelli et al., 2009; Zappitelli et al., 2011); two studies
showed that patients with AKI were treated significantly longer with aminoglycosides than
patients without AKI (Zappitelli et al., 2009; Zappitelli et al., 2011). Palmieri et al found that
67
the proportion of patients treated with nephrotoxic medications (aminoglycosides or
vancomycin) was similar, though non-significant, in patients with and without AKI; however,
the proportion significantly increased with increasing RIFLE severity (Palmieri et al., 2009).
3.5 Discussion
We performed a systematic review of the application of the RIFLE classification of acute
kidney injury and its association with measures of mortality and morbidity in paediatric
patients. Twelve studies met our inclusion criteria. Overall, we found that the methods used
to determine the RIFLE classification were incompletely described. Only two of the twelve
studies provided sufficient details to confirm the measurement of the RIFLE items were
consistent with the cited method (Akcan-Arikan et al., 2007; Frans B. Plotz et al., 2008).
Second, the application of the individual criteria varied significantly between studies,
including exclusion of the urine output criteria and differences in the operationalization of the
GFR criteria, baseline GFR definitions, and methods for estimating missing baseline values.
These findings suggest that paediatric populations may be vulnerable to the same
measurement-induced variability seen in adult populations (Ricci et al., 2008); for example,
of the 24 studies included in Ricci et al’s systematic review on the relationship between the
RIFLE and mortality, only 12 included both the GFR and urine output criteria to define
RIFLE-based AKI. Consistent application of the RIFLE criteria is crucial to provide a valid
estimate of AKI within the population of interest as the estimated incidence of AKI can be
significantly influenced by various applications of the criteria (Siew et al., 2010; Zavada et
al., 2010). For example, different methods for estimating baseline GFR status have been
68
shown to change the estimated incidence of AKI from 12% to 88% in critically ill children
and 5% to 43% in non-critically ill children (Zappitelli et al., 2008). In addition,
heterogeneity in the application of the RIFLE prevents meaningful comparison of the rates of
AKI between different study populations. The proposed modification to the RIFLE for use in
paediatric populations further adds to this variability, with limited justification provided for
the changes made to the original RIFLE (Akcan-Arikan et al., 2007). Modification of the
urine output assessment from a 6 hour to an 8 hour interval may be illustrative of pragmatic
customization of the RIFLE due to institutional practice or difficulty with measurement
accuracy. Only 58% of studies utilized the urine output criteria, perhaps because of the
difficulty of timed urine collection in patients who are not catheterized. Authors may have
overcome logistic challenges by customizing the RIFLE criteria to better fit with existing
clinical data. While customization may serve the immediate needs of individual studies, it
limits the advantages of a common measure and may illustrate a lack of consensus
surrounding the application of the RIFLE in children. This issue is not limited to paediatric
populations as modifications of the RIFLE have also been proposed for adult patients. For
example, Herget-Rosenthal et al proposed that serum cystatin C may be a useful biomarker to
improve the sensitivity of the RIFLE criteria (Herget-Rosenthal et al., 2004) and Tallgren et
al included it within the GFR criteria when they applied the RIFLE to assess AKI in patients
undergoing open abdominal aortic surgery (Tallgren et al., 2007). The issue of customization
highlights the importance of clearly outlining the definitions and measurement processes used
when defining RIFLE-based AKI.
Third, unlike in adult studies (Ricci et al., 2008) we found inconsistent relationships between
the RIFLE classification and measures of mortality and morbidity in children. Studies
reported discordant results between increasing RIFLE severity and mortality, length of stay,
69
illness severity, and measures of kidney function. These discrepancies may be due to sample
size limitations, heterogeneity between the study populations, or differences in the
application of the RIFLE. Sample size is an obvious limitation given the relative rarity of
AKI outcomes, such as mortality, in paediatric populations. A recent review focusing on
adult studies with a minimum of 1,000 patients showed consistent, significant associations
between the RIFLE and mortality (Srisawat, Hoste, & Kellum, 2010). Two studies described
their methodology in enough detail to determine that they followed the methodology exactly
as cited (Akcan-Arikan et al., 2007; Frans B. Plotz et al., 2008); the reported associations
with mortality were not consistent, though both studies found no significant association
between AKI and increased ICU LOS. Three studies utilized the same patient population but
were not consistent in their application of the pRIFLE (Akcan-Arikan et al., 2007; Washburn
et al., 2008; Zappitelli et al., 2007). Washburn et al (2008) did not report any associations of
interest for this review, however, both Akcan-Arikan et al (2007) and Zappitelli et al (2007)
showed significant increases in mortality in patients with AKI as well as significantly higher
PRISM scores with increasing RIFLE severity. From these limited results, it appears that
similar associations are seen when the RIFLE is applied consistently and it may be that
differences in RIFLE application do not affect measures of association within similar patient
populations. It should be noted that as most studies did not capture the full spectrum of
RIFLE severity we cannot evaluate the associations between measures of mortality and
morbidity within the highest severity levels of the measure (‘Loss’ and ‘End-stage’).
However, it may be that the long-term observation required for these grades of AKI severity
was not applicable in studies focused on hospital outcomes. In addition, some of the
variability in the reported associations may be due to several other confounding factors, such
70
as illness severity. We chose to only report unadjusted results as studies did not control for
the same factors and adjusted results were thus not comparable across studies.
While associations between the RIFLE and measures of mortality and morbidity are useful,
they are not demonstrative of the RIFLE’s ability to accurately reflect acute kidney injury.
Thus, the RIFLE may be a good measure of illness severity, but may not truly capture the
presence of AKI. Ideally, the RIFLE would be compared with a ‘gold standard’ measure of
acute kidney injury. However, unlike chronic kidney disease, no ‘gold standard’ diagnostic
test exists for the diagnosis and severity classification of acute kidney injury; it is based on a
combination of patient symptoms, clinical tests, and clinical judgment. Comparison of the
RIFLE with the AKIN classification confirms these tools are unsurprisingly correlated
(Zappitelli et al., 2011; Zappitelli et al., 2008), a comparison limited in value as both
measures are creatinine-based. Substitute measures of kidney function have been assessed,
such as the relationship between the RIFLE and renal replacement therapy (Duzova et al.,
2010; Frans B. Plotz et al., 2008; Zappitelli et al., 2009). However, the value of these
assessments is limited as the use of renal replacement therapy is captured within the RIFLE
definition and non-renal indications for dialysis also exist.
The widespread adoption of the RIFLE criteria is reflective of its collaborative development
and the pragmatic use of the most common measurements of renal function: creatinine and
urine output. The continued use of the RIFLE classification will be strengthened by a
consistent approach to the application of RIFLE criteria, including the handling of missing
baseline values, the methods used for the GFR criteria, and the duration over which urine
output is measured. It is likely that the same variability in application is also prevalent in the
use of the AKIN staging of acute kidney injury, though this was outside the scope of this
71
review. While customization of the RIFLE may be necessary due to specific needs of certain
populations under study, we suggest that a clear description of all methodology used to
classify RIFLE-based AKI will allow for easier comparison between studies and patient
populations. Further work is also required to fully understand the relationship between the
RIFLE and measures of mortality and morbidity in children.
Overall, our results suggest that the use of the RIFLE to assess AKI in children supports the
concept of a standardized measure of acute kidney injury severity. However, inconsistencies
in the use of the RIFLE have not facilitated the comparison of rates and outcomes of AKI in
children. While no single definition or classification of a complex disease such as acute
kidney injury will be perfect, a consensus must be reached as to its operationalization.
72
Chapter 4 Risk Factors of Acute Kidney Injury in Critically Ill Children
73
4.1 Overview
The overall goal of this work was to evaluate the contributions of drug therapy to acute
kidney injury (AKI). The previous chapter systematically reviewed the use of the RIFLE as a
definition of AKI in paediatric patient populations and was used to inform the
operationalization of the RIFLE as a definition of AKI, the outcome of interest, for the
remainder of the thesis work.
The next step of the thesis work is to quantify the characteristics and clinical factors that
place paediatric patients at higher risk for developing AKI. This chapter presents a cohort
study of critically ill children that estimates the incidence of AKI occurring during ICU
treatment. In addition, this cohort analysis assesses the factors associated with the
development of AKI occurring during ICU admission. These risk factors occur prior to ICU
admission (e.g. age, gender) and during the patient’s ICU stay (e.g. organ dysfunction,
nephrotoxic drug therapy). The results from the systematic review and this study will then be
combined to evaluate the association between specific medications and the development of
AKI, controlling for underlying differences in risk (Chapter 5).
4.2 Introduction
Acute kidney injury (AKI), a common adverse event in hospitalized patients, is associated
with increased mortality, morbidity and resource use (Chertow et al., 2005; de Mendonca et
al., 2000; F. B. Plotz et al., 2005; Ricci et al., 2008; Toth et al., 2012). Recent studies have
estimated the incidence of AKI in hospitalized children to range from 9% to 83% (Hassinger
74
et al., 2012; P. Mehta et al., 2012; Morgan et al., 2013; M. B. Slater, Anand, Uleryk, &
Parshuram, 2012; Toth et al., 2012) and our systematic review of AKI in children (Chapter 3)
reported incidences from 10% to 83% (M. B. Slater et al., 2012). Variability among incidence
estimates can be attributed to differences in both the patient populations studied and in the
definition of AKI (M. B. Slater et al., 2012).
Critically ill patients are at risk of developing AKI (S. L. Goldstein & Chawla, 2010; Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012).
In addition, they may be predisposed to developing AKI based on their underlying medical
conditions or treatments before admission to the intensive care unit (ICU). Conversely,
fluctuating medical status and treatments received during ICU admission may also increase
the risk of developing AKI. Previous studies evaluating risk factors for AKI in children have
been limited by a focus on specific patient populations, such as children (Aydin et al., 2012)
or neonates (Morgan et al., 2013) undergoing cardiac surgery and extremely low birth weight
infants (Viswanathan, Manyam, Azhibekov, & Mhanna, 2012), and by small sample size,
limiting the power of analyses to detect clinically important risk factors. In a 2007 study of
985 critically ill children, older age (greater than 12 years), thrombocytopenia, hypoxemia,
hypotension and coagulopathy were significant independent risk factors of AKI (Bailey et al.,
2007).
Given that patients who develop AKI have an increased risk of mortality (Barletta &
Bunchman, 2004; Bresolin et al., 2009; Ricci et al., 2008), identifying risk factors of AKI
may allow for prospective risk assessment and improved patient outcomes. The objectives of
this study were to estimate the incidence of paediatric AKI occurring during ICU treatment
75
and to determine what factors were associated with the development of AKI occurring during
ICU admission.
4.3 Methods
We performed a secondary analysis of electronic health data from a cohort of critically ill
children admitted to the ICU of a university-affiliated, paediatric academic center. Eligible
ICU admissions were the first ICU admission to the Paediatric Intensive Care Unit (PICU) or
Cardiac Intensive Care Unit (CICU) at The Hospital for Sick Children (HSC) in Toronto,
Ontario between January 2006 and June 2009. Patients were excluded if they had an
admission diagnosis of renal disease (i.e., known primary renal failure, post-renal transplant,
or chronic renal failure), drug overdose, tumor lysis syndrome, haemolytic uraemic syndrome
(conditions with known renal complications) or methylmalonic acidemia or urea cycle defect
(metabolic conditions treated with dialysis). In addition, patients who were less than two
weeks of age were excluded to minimize the impact of maternal creatinine on measured
creatinine levels in the child (Guignard & Drukker, 1999; Miall et al., 1999). Patients who
were admitted to the ICU for less than six hours were also excluded as these patients tend to
be admitted for observation or minor procedures only. In addition, patients with no creatinine
or urine output measurements during their ICU admission were also excluded as they were
missing necessary measurements to assess renal function. All subsequent ICU admissions
were also excluded; thus only one ICU admission for each patient was included in the study
population.
76
The study utilized the unique electronic data available from the Department of Critical Care
Medicine at The Hospital for Sick Children. All clinical data for any child admitted to the
PICU or CICU is charted electronically (CIMS, Sunrise Clinical Documentation, Eclipsys).
Radiology procedures, microbiology data, admission diagnoses and creatinine values and
drug administrations prior to ICU admission were accessed through linkages, based on the
patients’ unique medical record number, to other electronic data systems within the hospital.
The RIFLE criteria (Bellomo et al., 2004) were used to define a dichotomous measure for the
occurrence of AKI: any patient who reached a RIFLE severity of ‘Risk’ or higher (a 1.5
increase in serum creatinine from baseline or a urine output of <0.5ml/kg/hour for 6 hours)
was considered to have AKI. To assess AKI occurring due to exposures within the ICU
environment, the baseline measure of renal function was slightly modified from the RIFLE,
defined as the first serum creatinine measurement within 24 hours of ICU admission. If this
was not available, according to the RIFLE we then used the most recent creatinine
measurement within 3 months prior to ICU admission; otherwise the mean age- and sex-
adjusted norm (Schwartz, Haycock, & Spitzer, 1976) was used.
Possible risk factors of AKI were identified through literature review and consultation with
clinical experts. Candidate risk factors were classified into two groups: baseline risk factors
were those existing at the time of ICU admission and in-ICU risk factors were those
occurring during the ICU stay. The 13 baseline risk factors were: age at ICU admission;
gender; unplanned ICU admission; admission type (surgical admission, categorized as either
cardiac or non-cardiac, or medical admission); ICD-10 admission diagnoses (World Health
Organization, 2010) of circulatory disease (I00-I99), endocrine disease (E00-E99), infectious
disease (A00-B99), injury (S00-T98), neoplasm (C00-D48), nervous system disease (G00-
77
G99), or respiratory disease (J00-J99); and Pediatric Risk of Mortality (PRISM) score
(Pollack, Ruttimann, & Getson, 1988). The administration of nephrotoxic medication prior
to ICU admission was collected for those patients admitted to the ICU from within the
hospital; prior nephrotoxic drug administration was classified as unknown for patients
admitted directly to the ICU from another hospital.
The 13 risk factors during ICU care were: the presence of Systemic Inflammatory Response
Syndrome (SIRS) (B. Goldstein, Giroir, & Randolph, 2005), sepsis (B. Goldstein et al.,
2005), shock (Leteurtre et al., 1999), septic shock, organ dysfunction (cardiovascular,
neurologic, haematologic, respiratory), mechanical ventilation, the use of extracorporeal
membrane oxygenation (ECMO), the administration of intravenous radiologic contrast and
nephrotoxic medication administration. The presence of cardiovascular, neurologic,
haematologic, hepatic and respiratory organ dysfunction was defined as a paediatric logistic
organ dysfunction (PELOD) score (Leteurtre et al., 2003) of one or more for each system.
The administration of a nephrotoxic medication was captured as a binary variable indicating
the administration of one or more medications defined as being nephrotoxic. Each risk factor
was assessed until the development of AKI or death or discharge from the ICU.
4.3.1 Identification of Nephrotoxic Medications
All medications administered in the ICU were assessed for nephrotoxicity through a review
of adverse reactions noted in the drug monographs in the hospital formulary, the
Compendium of Pharmaceuticals and Specialties (Compendium of Pharmaceuticals and
Specialties 2013, 2012), the Pediatric Dosage Handbook (Taketomo, Hodding, & Kraus,
2012). Any medication with explicit or implied renal effects was identified; this resulted in
82 medications categorized as potentially nephrotoxic. To confirm this initial list of
78
medications, we used a Delphi process (Fink, Kosecoff, Chassin, & Brook, 1984) whereby
we consulted with the following clinical experts: a paediatric intensivist, a paediatric
pharmacologist and two paediatric pharmacists. For each medication, the experts were asked
to rate the likelihood and severity of AKI resulting from use of the medication. The
likelihood of AKI was assessed through the question ‘Assuming it is used as recommended,
what is the likelihood of a patient developing kidney injury from exposure to this
medication?’. Experts were provided 7 response options: none; very low (<1%), low (1-5%),
medium (5-10%), high (10-20%), very high (>20%) and unknown. Experts were also asked
‘How would you classify the severity of kidney injury caused by this medication?’ and were
given the following response options: low, moderate, severe, unknown and not applicable
(for medications rated as having no likelihood of kidney injury). The first round of responses
were tabulated and a second round of questionnaires, which incorporated the results from the
first round, were sent to the experts. The responses to the first round of questions were
presented as the frequency of responses, represented by the number of dots beside each
response option (Table 4.1). The results from the second round were tabulated and
medications were considered to be nephrotoxic if at least 50% of responses (i.e. two or more
of the four panel members) indicated that the likelihood of kidney injury was medium or
higher or the severity of AKI was moderate or more severe.
The Delphi process resulted in 29 medications identified as having important nephrotoxic
effects: acyclovir, amikacin, amphotericin, captopril, carboplatin, cidofovir, cisplatinum,
cyclophosphamide, cyclosporine, enalapril, enalaprilat, ethacyrnic acid, flucytosine,
foscarnet, furosemide, ganciclovir, gentamicin, hydrochlorothiazide, ibuprofen, ifosfamide,
indomethacin, ketorolac, methotrexate, penicillin, ramipril, sirolimus, tacrolimus,
tobramycin, and vancomycin.
79
Table 4.1: Example of second round of expert opinion survey
!If the response to question 1 is ‘None’ or ‘Don’t know’, you do not need to complete question 2 (the answer will be assumed to be N/A).!
Medication Name!1. Assuming it is used as recommended, what is the likelihood of a patient developing kidney injury from exposure to this medication?!
2. How would you classify the severity of kidney injury caused by this medication?!
6 MERCAPTOPURINE!
• [ ] None! [ ] Very low (<1%)! [ ] Low (1-5%)! [ ] Medium (5-10%)! [ ] High (10-20%)! [ ] Very high (>20%)!••• [ ] Don’t know!
[ ] Low! [ ] Moderate! [ ] Severe! [ ] Don’t know!••• [ ] N/A!!
ACYCLOVIR!
[ ] None!• [ ] Very low (<1%)!• [ ] Low (1-5%)!•• [ ] Medium (5-10%)! [ ] High (10-20%)! [ ] Very high (>20%)! [ ] Don’t know!
[ ] Low!••• [ ] Moderate! [ ] Severe! [ ] Don’t know! [ ] N/A!!
80
Descriptive statistics, including median and interquartile range (IQR) for continuous
variables and proportions for categorical data, were used to characterize the demographic and
clinical composition of the study population. The proportion of patients with creatinine and
urine output measurements available during their ICU stay were reported, calculated as the
number of patients with at least one measurement available by the total number of patients in
the ICU on a given day. The first ICU day was defined as the 24-hour period beginning at
ICU admission and subsequent ICU days were referenced according to the admission time,
rather than defined according to calendar days. Exploratory analyses were performed to
evaluate the impact of different definitions of baseline creatinine and the exclusion of
components of the RIFLE classification on the estimated incidence of AKI. Agreement
between estimated AKI severity according to varying baseline definitions was estimated
using a kappa statistic. A weighted kappa, which weights agreement according to how close
the categories are (Cohen, 1968), was used in order to better reflect the fact that disagreement
between the adjacent categories ‘Risk’ and ‘Injury’ severity levels is different than
disagreement between ‘Risk’ and ‘Failure’.
Bivariate analyses of the association between risk factors and the development of AKI were
conducted using chi-square tests or, for continuous variables, the Wilcoxon-rank sum test.
Any candidate risk factor with a bivariate significance level of 0.2 or smaller was eligible for
inclusion in the multivariable logistic regression model. Spearman correlations and variance
inflation factors (VIF) were calculated to assess multicollinearity. For variables with a VIF
greater than 5 (O'Brien, 2007) or pairs of variables with correlations greater than 0.5, clinical
judgment was used to determine which were included in the multivariable model. Backward
81
stepwise regression was performed to find the most parsimonious multivariable model. All
analysis was performed using SAS software, version 9.3 (SAS Institute Inc., Cary, NC).
This study received approval from The Hospital for Sick Children’s Research Ethics Board.
4.4 Results
Between January 1, 2006 and June 30, 2009, there were 3 865 children admitted to the ICU at
The Hospital for Sick Children in Toronto, Canada who were eligible for further study
(Figure 4.1). The median age was 4.4 years (IQR=11.1); 30% of patients were under 12
months of age, 23% between 13 months and 5 years, 20% between 6 and 11 years, and 28%
were between 12 and 18 years of age (Table 4.2). Forty-four percent of the admitted patients
were female. Approximately 49% of ICU admissions were unplanned; 55% of patients were
admitted to the ICU post-operatively while 18% were admitted directly to the ICU from
another hospital. The median length of stay was of 2 days (IQR=3) and the ICU mortality rate
was 3.3%.
82
!Figure 4.1: Definition of study population During the study period, there were 6 326 admissions to the ICU; 24.5% of these were repeat admissions and were excluded. Also excluded were 654 patients who were less than 2 weeks old, 64 patients who had an ICU stay of less than 6 hours, leaving 4 056 patients. Of these, 173 had pre-existing renal conditions and a further 18 had no creatinine or urine output measurements available. Once these patients were excluded, the study population consisted of 3 865 patients.!
6 326 Admissions during study period
4 774 First ICU admissions
4 056 ICU stay ≥6 hours
654 (13.7%) Age ≤ 2 weeks
64 (1.6%) ICU stay <6 hours
4 120 Age >2 weeks and ≤18 years
1 552 (24.5%) Repeat admissions
173 (4.3%) Pre-existing renal conditions
3 883 No pre-existing renal conditions
18 (0.5%) No creatinine or urine output
measurements available
3 865 At least one creatinine or urine output measurement available
83
Table 4.2: Description of study population
Characteristic n (%) Age group Term neonate (≤27 days) 68 (1.8) Infant (28 days – 12 months) 1096 (28.4) Toddler (13 months – 2 years) 326 (8.4) Early childhood (2 – 5 years) 547 (14.2) Middle childhood (6 – 11 years) 755 (19.5) Early adolescence (12 – 18 years) 1073 (27.8) Gender Female 1701 (44.0) Male 2164 (56.0) Unplanned admission 1887 (48.8) Admitted to ICU from Emergency department 509 (13.2) Hospital ward 553 (14.3) Operating room 2121 (17.6) Other hospital 682 (17.7) ICU length of stay (median, IQR) 2.0, 3.0 ICU mortality 126 (3.3) ICU: intensive care unit; IQR: interquartile range
On the day of admission, over 90% of patients (n=3 843) had at least one creatinine value
available (Figure 4.2); the percentage then decreases to approximately 80%. A similar trend
is seen when we consider the number of patients in the ICU who have their urine output
documented each day (Figure 4.3). It should be noted that the standard of care in the ICU of
The Hospital for Sick Children is hourly documentation of vital signs (e.g. heart rate,
respiratory rate, blood pressure, etc.). Creatinine is monitored on a daily basis, more
frequently if there are clinical implications. For patients with indwelling urinary catheters,
urine output is documented hourly. The standard of care is to remove the catheter as soon as
possible; however, once the catheter is removed, urine output is still documented with each
void.
84
Figure 4.2: Number and proportion of patients with at least one serum creatinine value
available by day (for the first 14 days of their ICU stay) We assessed the proportion of patients who had at least one serum creatinine measurement during each day (based on 24 hour intervals from the time of ICU admission). During the first 24 hours of admission, 90% of patients had at least one creatinine measurement. This percentage dropped to just under 50% and then increased to 80% as the length of stay increased.! !
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
0
500
1000
1500
2000
2500
3000
3500
4000
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Pro
porti
on
Num
ber o
f Pat
ient
s
ICU Day
85
Figure 4.3: Number and proportion of patients with at least one urine output
measurement available by day (for the first 14 days of their ICU stay) We assessed the proportion of patients who had at least one urine output measurement during each day (based on 24 hour intervals from the time of ICU admission). During the first 24 hours of admission, over 95% of patients had at least one urine output measurement. This percentage dropped to just approximately 55% and then increased to roughly 90% as the length of stay increased.!
The timing of the baseline creatinine measurements in the study population were as follows:
90.1% of baseline values are defined as the patient’s first creatinine measured within 24
hours of ICU admission; 5.7% were based on the most recent creatinine available within 3
months prior to the ICU admission; 4.2% had neither measure available and thus had an age-
and sex-adjusted creatinine measure imputed for a baseline value. Of the 221 patients whose
baseline is defined as the most recent pre-ICU creatinine, roughly 60% of these
measurements are taken on the same day or the day prior to their admission to the ICU
(Figure 4.4). Thus, based on the timing of measurements, it is likely that the pre-ICU
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
0
500
1000
1500
2000
2500
3000
3500
4000
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Pro
porti
on
Num
ber o
f Pat
ient
s
ICU Day
86
creatinine value would estimate a creatinine measured within the first 24 hours of ICU
admission.
Figure 4.4: Timing of the most recent creatinine measurement available within 3
months prior to ICU admission for patients without a creatinine available
during the first 24 hours of ICU admission (n=221) For patients who did not have a creatinine measure available within the first 24 hours of ICU admission, their baseline creatinine measure was the most recent value within three months of ICU admission. We assessed how recent these values were and found that over 40% had a measure taken on the same day of ICU admission (far right bar). Approximately 60% of patients had a measurement taken on the day of or the day prior to ICU admission.!
For the majority of patients (90.1%) whose baseline creatinine measurement is defined as the
first serum creatinine measured during the first 24 hours of their admission to the ICU, it is
plausible that this value could be extremely high. A high baseline measure would make it
highly unlikely to see an increase of 1.5 to meet the RIFLE-based definition of AKI. Despite
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
> - 9 weeks
- 5-8 weeks
- 3-4 weeks
- 8-14 days
-7 -6 -5 -4 -3 -2 -1 Same day
Pro
porti
on
Timing of most recent creatinine within 3 months of ICU admission
87
the skewness of the distribution of first creatinine measurements (by age group, Figures 4.5
through 4.7), only 93 of the 3 483 patients with a measurement available (2.7%) have an
abnormally high creatinine measurement, based on the PELOD criteria (<1 year, ≥55µmol/L;
1-12 years, ≥100µmol/L; ≥12 years, ≥140µmol/L) (Leteurtre et al., 1999; Leteurtre et al.,
2003).
Figure 4.5: Histogram of first creatinine measurement values within 24 hours of
admission for patients under 1 year of age (n=1020) For patients under 1 year of age, the distribution of first creatinine measures is highly skewed. However, few are above the PELOD criteria (≥55µmol/L) defining an abnormally high value.! !
0
50
100
150
200
250
300
350
400
450
Freq
uenc
y
Creatinine value
88
Figure 4.6: Histogram of first creatinine measurement values within 24 hours of
admission for patients between 1 and 12 years of age (n=1605) For patients between 1 and 12 years of age, the distribution of first creatinine measures is highly skewed. However, few are above the PELOD criteria (≥100µmol/L) defining an abnormally high value. !
0
100
200
300
400
500
600
Freq
uenc
y
Creatinine value
89
Figure 4.7: Histogram of first creatinine measurement values within 24 hours of
admission for patients 12 years of age and older (n=857) For patients over 12 years of age, the distribution of first creatinine measures is highly skewed. However, few are above the PELOD criteria (≥140µmol/L) defining an abnormally high value.
Nine hundred and fifteen patients (23.7%) developed AKI at a median of 28 hours
(IQR=43.0) after ICU admission. The severity of the first occurrence of AKI was: 84.8%
‘Risk’, 13.2% ‘Injury’ and 2.0% ‘Failure’. Sixty-six percent of patients with AKI achieved a
maximum severity of ‘Risk’, 27.5% ‘Injury’ and 6.4% ‘Failure’. Patients who developed
AKI had a longer length of stay in the ICU (median 5 days, IQR=8 versus 1 day, IQR=1) and
significantly higher mortality (7.2%, 95% confidence interval (CI): 5.6%-9.1% versus 2.0%,
1.6%-2.6%) than those who did not develop AKI.
As previously stated, 90% of patients in the cohort had a baseline serum creatinine
measurement available within the first 24 hours of ICU admission, 5.7% had a measure
0
50
100
150
200
250
Freq
uenc
y
Creatinine value
90
available within 3 months prior to ICU admission and 4.2% had an imputed baseline value.
To determine the effect of different baseline creatinine values on the estimated incidence of
AKI, the RIFLE-GFR criteria were recalculated using the following estimates of baseline
renal function:
1. The first serum creatinine measured within 24 hours of the patient’s admission to the
ICU, available for 3 483 patients;
2. The most recent serum creatinine measured 3 months prior to the patient’s ICU
admission, available for 2 633 patients;
3. An age- and sex-adjusted normal value (hereafter referred to as the imputation
method), available for all 3 865 patients.
For the purposes of this comparison, only the GFR criteria were used to estimate the
incidence of AKI. The imputation method resulted in the highest number of AKI cases (773
patients, 20.0%) whereas use of the most recent creatinine within 3 months of ICU admission
resulted in an incidence of 16.3% (n=429). Use of the first creatinine within 24 hours of ICU
admission resulted in the lowest incidence (7.0%, n=245). The associated severity levels are
shown in Figure 4.8. However, when both the GFR and urine output criteria were used, we
found no significant differences in estimated AKI incidence when comparing the three
methods of assessing baseline renal function (Table 4.3).
91
Figure 4.8: Severity of the first occurrence of GFR-based RIFLE AKI across different
methods of baseline estimation When three different methods of estimation for baseline serum creatinine is compared, we see that the imputation method (age- and gender-adjusted norms) results in the highest estimated incidence of AKI whereas using the first serum creatinine within 24 hours of ICU admission results in the lowest estimated incidence. Severity of AKI also varies across estimation methods, with the highest proportion of the highest severity level, ‘Failure’, results from the imputation method.
Table 4.3: Effect of varying baseline measurements on the estimated incidence of AKI
Baseline measurement Estimated AKI incidence First SCr within 24 hours of ICU admission 23.2% Most recent SCr within 3 months of ICU admission 28.1% Age- and gender-adjusted norm 28.6%
SCr: serum creatinine
In the 2 412 patients (62.4%) that had both a creatinine measured within the first 24 hours of
their ICU admission and a creatinine measured within the 3 month period prior to their ICU
admission, the AKI severity classifications resulting from the use of each baseline measure
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
First SCr within 24h of ICU admission
Most recent SCr within 3 months prior
to ICU admission
Age- and gender-adjusted norm
Pro
porti
on
Risk
Injury
Failure
92
had an agreement of 89.2% and a weighted kappa of 0.58 (95% CI: 0.54 – 0.62), indicating
moderate agreement (Cohen, 1968). The raw agreement increased to 92.4% when we
dichotomized the RIFLE severity to a measure of AKI versus no AKI (kappa 0.67; 95% CI:
0.65 – 0.73), indicating substantial agreement between the estimated incidence of AKI
according to the two baseline methods.
The RIFLE criteria have two components (GFR and urine output); both are intended to be
used when assessing the severity of acute kidney injury. However, many studies use only the
GFR criteria when defining AKI (M. B. Slater et al., 2012). In our patient cohort, 17.9% of
patients met both the GFR and urine output criteria, 59.8% met only the urine output criteria,
and 22.3% only met the GFR criteria at any point during their ICU stay. Translating this to
estimates of AKI incidence, when we considered only the GFR criteria to define AKI, 368
patients (9.5% of the study cohort) had a RIFLE severity of ‘Risk’ or higher. If we solely
consider the urine output criteria, 711 patients (18.4%) had AKI (Figure 4.9).
93
Figure 4.9: Estimated incidence of AKI based on varying RIFLE criteria components When comparing the estimated incidence of AKI using different components of the RIFLE criteria, we find that using only the GFR criteria (i.e. relative changes in serum creatinine) results in the lowest estimate of AKI (9.5%). Using only the urine output criteria, the estimated incidence of AKI increases to 18.4%. Using both criteria, the incidence of AKI in the study population is estimated to be 23.7%). AKI: acute kidney injury; GFR: glomerular filtration rate!
Unadjusted analyses revealed that the development of AKI was associated with younger age,
unplanned ICU admissions, cardiac surgical or medical ICU admissions, International
Classification of Disease (ICD) admission diagnosis codes of circulatory disorders, endocrine
disorders, infectious disease, neoplasms, and respiratory disorders, and receiving one or more
nephrotoxic medications prior to ICU admission (Table 4.4). During their ICU care, patients
who experienced sepsis, shock, septic shock or organ dysfunction or received intravenous
radiologic contrast, mechanical ventilation, extracorporeal membrane oxygenation (ECMO)
0.0
5.0
10.0
15.0
20.0
25.0
GFR Urine output GFR or urine output
Est
imat
ed A
KI I
ncid
ence
94
or one or more nephrotoxic medications were significantly more likely to develop AKI
(Table 4.4).
Twenty-four risk factors were eligible for multivariable analyses; 4 were excluded due to
high degrees of collinearity (i.e. mechanical ventilation, SIRS, shock and septic shock). After
adjustment in a multivariable logistic regression model, we identified 12 variables for our
final model using backward stepwise regression (Table 4.5). Increasing age (OR=1.02, 95%
CI: 1.00-1.03), unplanned admission to the ICU (OR=1.88, 95% CI: 1.38-2.56), admission
diagnoses of circulatory (OR=1.46, 95% CI: 1.09-1.95) or respiratory system disorders
(OR=1.43, 95% CI: 1.10-1.88), increasing Pediatric Risk of Mortality (PRISM) score
(OR=1.03, 95% CI: 1.02-1.05), in-ICU respiratory dysfunction (OR=2.90, 95% CI: 2.29-
3.67), use of ECMO (OR=2.72, 95% CI: 1.14-6.52) and administration of nephrotoxic
medication(s) both prior to (OR=1.43, 95% CI: 1.12-1.84) and during the ICU admission
(OR=3.37, 95% CI: 2.66-4.28) increased the odds of developing AKI. Non-cardiac surgical
admissions (OR=0.69, 95% CI: 0.50-0.95), use of intravenous radiologic contrast (OR=0.76,
95% CI: 0.59-0.97), and in-ICU neurologic dysfunction (OR=0.62, 95% CI: 0.51-0.76) were
associated with a reduced occurrence of AKI. Cardiac surgical admissions had an increased
odds of developing AKI (OR=1.26, 95% CI: 0.88-1.81) compared with non-surgical
admissions, however this result was not statistically significant.
95
Table 4.4: Risk factors for developing acute kidney injury (bivariate analyses); n(%)
unless specified otherwise
Risk factor AKI (n=915)
No AKI (n=2 950)
p-value
At ICU admission Admission age in years (median, IQR) 2.9, 11.1 4.8, 11.1 <0.0001 Sex (female) 381 (41.6) 1 320 (44.8) 0.0982 Unplanned ICU admission 553 (60.4) 1 334 (45.2) <0.0001 Admission type Surgical, cardiac 318 (34.8) 706 (23.9)
<0.0001 Surgical, non-cardiac 101 (11.0) 996 (33.8) Medical 496 (54.2) 1 248 (42.3) Admission diagnoses (ICD-10 codes) Circulatory (I00-I99) 99 (10.8) 188 (6.4) <0.0001 Endocrine (E00-E99) 13 (1.4) 124 (4.2) <0.0001 Infectious disease (A00-B99) 36 (3.9) 58 (1.8) 0.0007 Injury (S00-T98) 71 (7.8) 210 (7.1) 0.5142 Neoplasms (C00-D48) 29 (3.2) 246 (8.3) <0.0001 Nervous system (G00-G99) 70 (7.7) 224 (7.6) 0.9546 Respiratory (J00-J99) 122 (13.3) 242 (8.2) <0.0001 PRISMa score (median, IQR) 5.0, 7.0 3.0, 6.0 <0.0001 Nephrotoxic medication administration prior to ICU admission
Yes 149 (16.3) 264 (8.9) <0.0001 No 544 (59.5) 2 226 (75.5)
Unknown 222 (24.3) 460 (15.6) During ICU admission SIRS 649 (70.9) 2 007 (68.0) 0.0989 Sepsis 108 (11.8) 221 (7.5) <0.0001 Shock 672 (73.4) 1 709 (57.9) <0.0001 Septic shock 89 (9.7) 188 (6.4) 0.0006 Cardiovascular organ dysfunction 701 (76.6) 1 779 (60.3) <0.0001 Neurologic organ dysfunction 657 (71.8) 1 969 (66.8) 0.0042 Haematologic organ dysfunction 144 (15.7) 265 (9.0) <0.0001 Hepatic organ dysfunction 394 (43.1) 938 (31.8) <0.0001 Respiratory organ dysfunction 732 (80.0) 1 462 (49.6) <0.0001 Intravenous radiologic contrast 132 (14.4) 346 (11.7) 0.0304 Mechanical ventilation 729 (79.7) 1 451 (49.2) <0.0001 ECMO 20 (2.2) 8 (0.3) <0.0001 Nephrotoxic medication administration 802 (87.7) 1 613 (54.7) <0.0001 ECMO: extracorporeal membrane oxygenation; ICD: International Classification of Disease; ICU: intensive care unit; IQR: interquartile range; PRISM: Pediatric Risk of Mortality; SIRS: systemic inflammatory response syndrome a PRISM score ranges from 0 to 48, with a higher score indicating more severe illness
96
Table 4.5: Risk factors of acute kidney injury in critically ill children (multivariable
analysis)
Risk factor Adjusted ORa 95% confidence interval
At ICU admission Admission age (per 1 year increase in age) 1.02 1.00 – 1.03 Unplanned ICU admission 1.88 1.38 – 2.56 Admission type (ref: medical admission) Surgical, cardiac 1.26 0.88 – 1.81 Surgical, non-cardiac 0.69 0.50 – 0.95 Circulatory admission diagnosis 1.46 1.09 – 1.95 Respiratory admission diagnosis 1.43 1.10 – 1.88 PRISM score (per 1 unit increase in score) 1.03 1.02 – 1.05 Nephrotoxic medication administration prior to ICU admission (ref: no)
Yes 1.43 1.12 – 1.84 Unknown 1.22 0.96 – 1.55 During ICU admission Neurological organ dysfunction 0.62 0.51 – 0.76 Respiratory organ dysfunction 2.90 2.29 – 3.67 Intravenous radiologic contrast 0.76 0.59 – 0.97 ECMO 2.72 1.14 – 6.52 Nephrotoxic medication administration 3.37 2.66 – 4.28 ECMO: extracorporeal membrane oxygenation; ICU: intensive care unit; PRISM: Pediatric Risk of Mortality a OR adjusted for all 12 risk factors included in the multivariable model; Hosmer and Lemeshow goodness-of-fit test: χ2=5.6391, p=0.6876
4.5 Discussion
We evaluated the incidence and risk factors for acute kidney injury using an accepted
classification system (Bellomo et al., 2004) in a cohort of 3 865 critically ill children
admitted to a large, academic intensive care unit over a 42-month period. To our knowledge,
this is the largest study of AKI in critically ill children performed to date. We have also
demonstrated the feasibility of using electronic data to assess AKI utilizing the RIFLE
criteria.
97
We found that AKI occurred in nearly one-quarter of all critically ill children. Previous
studies of RIFLE-defined AKI report rates varying from 10% to 83% (M. B. Slater et al.,
2012); no studies were found that applied the full RIFLE criteria in a similar patient
population. Two previous studies in a similar patient population did not evaluate urine output
(Bailey et al., 2007; Schneider et al., 2010) and one used a study-specific definition of AKI
(the doubling of serum creatinine relative to the creatinine level at ICU admission or relative
to the upper limit of normal for age and sex) (Bailey et al., 2007), limiting comparison to this
study. Our results reinforce the clinical relevance of AKI in this population as AKI was
associated with an increased ICU length of stay of 4 days and children with AKI had a
mortality rate roughly 3.5 times that of children without AKI. The increased mortality rate
seen in critically ill children with AKI adds to the growing evidence that the development of
AKI itself contributes to increased mortality, rather than the previous belief that patients with
AKI die because of their underlying disease (E.A.J. Hoste & De Corte, 2011).
Increased risk of AKI was seen with increasing age, ICU admission diagnoses of circulatory
or respiratory disorders, increasing PRISM score, administration of nephrotoxic
medication(s) prior to ICU admission, in-ICU respiratory dysfunction, use of ECMO and
administration of nephrotoxic medication(s) during the ICU stay, consistent with previous
work (Akcan-Arikan et al., 2007; Bailey et al., 2007; Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012; Palmieri et al., 2009; F. B.
Plotz et al., 2005; Schneider et al., 2010). While trauma has been previously reported as a
risk factor of AKI in both adults (Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012; Shigehiko Uchino et al., 2006) and children
(Wohlauer et al., 2012), our findings correspond to those of Bailey et al (Bailey et al., 2007)
and Plotz et al (F. B. Plotz et al., 2005) who also found the association between AKI and an
98
admission diagnosis of trauma to be statistically non-significant in critically ill children. Our
finding that radiologic contrast administration was associated with lower rates of AKI was
surprising (Brasch, 2008; Kidney Disease: Improving Global Outcomes (KDIGO) Acute
Kidney Injury Working Group, 2012; Seeliger, Sendeski, Rihal, & Persson, 2012; Waybill &
Waybill, 2001); however, the association between contrast administration and AKI has
recently been questioned. A recent meta-analysis of observational studies revealed that the
risk of AKI, death and dialysis in patients receiving contrast medium was similar to those
who did not (McDonald et al., 2013). Perhaps contrast use is a marker of relative health as,
according to clinical guidelines (Benko et al., 2007; Solomon & Deray, 2006; Stacul et al.,
2011), contrast is withheld in patients perceived to be at risk for AKI. However, patients in
our cohort who received intravenous contrast did not appear to be healthier than their peers;
they were significantly more likely to have unplanned admissions to the ICU, have higher
illness severity scores (PRISM), longer lengths of stay and higher mortality rates (data not
shown). The perception of risk of kidney damage with contrast administration may be
influential here; increased hydration is recommended to minimize the deleterious effects on
the kidney (Benko et al., 2007; Solomon & Deray, 2006; Stacul et al., 2011) and patients are
closely monitored for signs of kidney injury. Our findings may be explained by timing:
serum creatinine levels have been shown to peak three days after contrast administration
(Marenzi et al., 2004; Solomon, 1998) and the effects of contrast use may not be apparent
until after ICU discharge. Further work is needed to elucidate the causal association between
contrast administration and the development of acute kidney injury in critically ill children.
Of the twelve independent risk factors identified, seven were evident prior to ICU admission
while five occurred during the ICU stay. All in-ICU risk factors are markers of disease
severity (organ dysfunction) or therapeutic intervention (intravenous contrast, ECMO,
99
nephrotoxic medication). The single greatest risk factor of in-ICU AKI was the
administration of nephrotoxic medications, indicating that scrutiny of the drugs used in the
paediatric ICU setting is warranted. Lessons from contrast guidelines (Benko et al., 2007;
Solomon & Deray, 2006; Stacul et al., 2011) and recent recommendations for
aminoglycoside and amphotericin-induced AKI prevention (Kidney Disease: Improving
Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012) should be applied to
all potential nephrotoxic drugs: nephrotoxic medications should be avoided wherever
possible for all patients, especially those identified to be in high-risk populations, such those
urgently admitted to ICU, those who develop respiratory dysfunction, and those treated with
ECMO. Further research into the risks of specific drugs on the renal system is warranted to
better understand opportunities for ‘nephroprotection’ as are investigations of the clinical
outcomes of associated with use of potentially less nephrotoxic alternative therapies.
4.5.1 Limitations
Our study has several limitations. First, while the use of electronic data allowed study of a
large patient population, it limited some of the data elements for use in analysis. Some
important clinical variables of interest, such as cardiac arrest or duration of cardiac bypass,
were not available in our electronic systems. This data availability also affected our definition
of AKI; while a pediatric modification of the RIFLE exists (pRIFLE) (Akcan-Arikan et al.,
2007), it requires the calculation of estimated creatinine clearance. Unfortunately, patient
height is not captured in our electronic data sources and, as such, we were unable to calculate
creatinine clearance. In addition, our electronic sources did not include medication
reconciliation from other institutions and we did not have other access to medical records
from other institutions. Accordingly, we were only able to assess nephrotoxic drug exposure
100
during the same hospital admission prior to admission to the ICU; medications administered
in other hospitals or at home could not be captured. The definition of nephrotoxic
medications used in our analysis was based on adverse reactions noted in drug monographs
and in consultation with clinical experts; however, we may have excluded potentially relevant
medications. In addition, the use of a binary measure of medication exposure may have
oversimplified the association between drug exposure and nephrotoxicity.
Finally, some of the clinical data we were interested in were highly correlated and definitions
of many risk factors of interest were similar (e.g., Systemic Inflammatory Response
Syndrome (SIRS) and sepsis, shock and cardiovascular dysfunction, mechanical ventilation
and respiratory dysfunction). While addressing collinearity issues and excluding variables
from the multivariable model strengthens our analysis, the choices made may limit
comparability to other research. For example, while we chose to include respiratory
dysfunction over mechanical ventilation, or cardiovascular organ dysfunction over shock,
others may have made the opposite decisions.
Despite these limitations, we have demonstrated the feasibility of the use of electronic health
data to evaluate detailed patient level outcomes, such as the utilization of both changes in
serum creatinine and urine output to define AKI using the RIFLE criteria.
4.6 Conclusions
This evaluation of a large cohort of critically ill children found AKI to be a common event in
the paediatric ICU, with complex and multifactorial origins. Seven risk factors associated
101
with the development of AKI were pre-existing at ICU admission while five occurred during
the patient’s ICU care. The single greatest risk factor for AKI was the administration of
nephrotoxic medications during the ICU stay. The identification of risk factors suggests that
prospective AKI risk assessment may be possible and allow for targeted interventions to
reduce AKI and improve the outcomes of critically ill children.
102
Chapter 5 Acute Kidney Injury and Exposure to Nephrotoxic Medications in Critically Ill Children: A Nested Case-Control Study
103
5.1 Overview
The overall goal of this work was to evaluate the contributions of drug therapy to acute
kidney injury (AKI). The previous chapter presented a cohort study that evaluated the
characteristics and clinical factors that place paediatric patients at higher risk for developing
AKI during treatment in the ICU. We found several risk factors associated with the
development of AKI, some pre-existing at ICU admission while others occurred during the
patient’s ICU care. However, the single greatest risk factor for AKI was the administration
of one or more nephrotoxic medication during the ICU stay.
Accordingly, the next step of the thesis work is to evaluate the association between specific
nephrotoxic medications and the development of AKI, while controlling for underlying
differences in risk. In order to accomplish this, we performed a nested case-control study, the
results of which are presented in this chapter. Critically ill children who developed AKI
during their ICU admission were matched to patients who did not develop AKI. Exposure to
nephrotoxic medications was compared between the two patient groups. The analysis was
adjusted for underlying differences in risk factors of AKI based on the cohort analysis
previously presented (Chapter 4).
5.2 Introduction
The incidence of acute kidney injury (AKI) in hospitalized children ranges from 9% to 83%
(Hassinger et al., 2012; P. Mehta et al., 2012; Morgan et al., 2013; M. B. Slater et al., 2012;
Toth et al., 2012) and critically ill children are at particular risk of developing AKI (Kidney
104
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012).
Children who develop AKI have poor outcomes, including increased morbidity and mortality
(D. J. Askenazi et al., 2006; D. J. Askenazi et al., 2009; Hui-Stickle et al., 2005). A number
of factors increase a patient’s risk for developing AKI, such as critical illness, sepsis,
circulatory shock and trauma (Kidney Disease: Improving Global Outcomes (KDIGO) Acute
Kidney Injury Working Group, 2012). While some factors, such as age and gender (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012),
are non-modifiable, others, including exposure to nephrotoxic medications, are under the
control of the prescriber and present a potential opportunity to decrease the risk of AKI.
Exposure to nephrotoxic medications is the second most common cause of AKI, estimated to
account for 16% of pediatric AKI (Hui-Stickle et al., 2005). However, there is lack of
pediatric data available regarding the actual incidence of nephrotoxicity for many
medications known to have deleterious effects on the kidneys. Moffet and Goldstein recently
assessed the association between nephrotoxic medications and the risk of developing AKI in
hospitalized, non-critically ill pediatric patients (Moffett & Goldstein, 2011); however, the
risk of nephrotoxic medication-induced AKI in critically ill children remains unknown.
The objective of this study was to determine the risk of developing AKI after the initiation of
nephrotoxic medication within the intensive care unit (ICU) setting among critically ill
children.
105
5.3 Methods
We performed a nested case-control study in a cohort of critically ill children at The Hospital
for Sick Children in Toronto, Ontario, Canada. The cohort that was used in the previous
study regarding risk factors of AKI (Chapter 4) was the basis for the nested case-control
study. As stated in the previous chapter, children who were admitted to the ICU between
January 2006 and June 2009 were eligible for the cohort and were followed for the duration
of their ICU stay. Patients with an admission diagnosis of renal disease (known primary
renal failure, post-renal transplant, or chronic renal failure), drug overdose, tumor lysis
syndrome, haemolytic uraemic syndrome (conditions with known renal complications) or
methylmalonic acidemia or urea cycle defect (metabolic conditions treated with dialysis)
were excluded. In addition, we excluded patients who were less than two weeks of age to
minimize the impact of maternal creatinine on measured creatinine levels in the child
(Guignard & Drukker, 1999; Miall et al., 1999) and those who were admitted to the ICU for
less than six hours as these patients tend to be admitted for observation or minor procedures
only. Patients with no creatinine or urine output measurements during their ICU admission
were also excluded as they were missing necessary measurements to assess renal function.
Only the first ICU admission for each patient was considered; repeat admissions during the
study period were not included.
5.3.1 Case Definition
Cases were classified as having AKI according to the RIFLE (Bellomo et al., 2004) criteria: a
RIFLE severity of ‘Risk’ or higher (a 1.5 increase in serum creatinine from baseline or a
urine output of <0.5ml/kg/hour for 6 hours) defined a patient as having AKI. To assess AKI
106
occurring due to exposures specifically within the ICU environment, the baseline creatinine
measure was slightly modified from the RIFLE and defined as the first serum creatinine
measurement within 24 hours of ICU admission. If this creatinine measure was unavailable,
consistent with the RIFLE (Bellomo et al., 2004) criteria the most recent creatinine
measurement within 3 months prior to ICU admission was used; otherwise the mean age- and
gender-adjusted norm (Schwartz, Haycock, & Spitzer, 1976) was used. The date and time of
the first diagnosis of AKI was defined as the index date for each case.
5.3.2 Identification of Controls
Incidence density sampling (Szklo, 2006) was used to identify control patients for each case
patient. Using this sampling methodology, controls were selected from a random sample of
the patients remaining in the cohort at the time each case occurs (the index date), equivalent
to matching cases and controls on follow-up time (Szklo, 2006) (Figure 5.1). As we were
concerned about controlling for exposure to nephrotoxic medications prior to ICU admission,
controls were matched 1:1 to cases on the basis of exposure to nephrotoxic medications
during the hospitalization directly prior to admission to the ICU. Prior nephrotoxic drug
administration was classified according to three categories: yes, no or unknown. The
‘unknown’ classification indicated that the patient was admitted directly to the ICU from
another hospital. This reflected the fact that we were unable to assess medications
administered in other hospitals or at home and thus were only able to assess nephrotoxic drug
exposure during the same hospital admission.
107
Figure 5.1: Incidence density sampling At the time a patient is diagnosed with AKI and becomes a ‘case’, they are matched to a control. Controls are randomly selected from the patients still in the cohort at this time. Thus, a control can later become a case. This process essentially matches cases and controls on follow-up time. Modified from Szklo & Nieto. Epidemiology: Beyond the Basics. 2nd ed. p28 (Szklo, 2006) !
5.3.3 Nephrotoxic Drug Exposure
The main exposure of interest for this study is the administration of nephrotoxic medications
during the patient’s ICU stay. As previously discussed in Chapter 4, a list of potential
nephrotoxic medications was compiled according to adverse event information contained in
the drug monographs of the hospital formulary, the Compendium of Pharmaceutical and
Specialties (Compendium of Pharmaceuticals and Specialties 2013, 2012) and the Pediatric
Dosage Handbook (Taketomo et al., 2012). Using a Delphi process involving clinical
C
O
H
O
R
T
CASES
CONTROLS
Time = 0
108
experts, including a paediatric intensivist, paediatric pharmacologist and paediatric
pharmacists, the following 29 medications were considered to have relevant nephrotoxic
effects: acyclovir, amikacin, amphotericin, captopril, carboplatin, cidofovir, cisplatinum,
cyclophosphamide, cyclosporine, enalapril, enalaprilat, ethacyrnic acid, flucytosine,
foscarnet, furosemide, ganciclovir, gentamicin, hydrochlorothiazide, ibuprofen, ifosfamide,
indomethacin, ketorolac, methotrexate, penicillin, ramipril, sirolimus, tacrolimus,
tobramycin, and vancomycin. Further details of this process are outlined in Section 4.3.1.
During the patient’s ICU stay, the administration of each drug was captured as a binary
variable (yes/no). All data were evaluated prior to the index date of the first diagnosis of
AKI.
While the main interest of this study was exposure to nephrotoxic medications during the
ICU stay, we wanted to control for previous exposures to nephrotoxic drugs. As such,
exposure to any of the 29 nephrotoxic medications listed above during the hospitalization
immediately before the ICU admission was captured for each patient in the cohort. Patients
who were admitted directly to the ICU from another hospital were classified as having
unknown prior nephrotoxic medication exposure. As previously stated, cases and controls
were matched on their prior exposure. Analyses were stratified according to the
administration of nephrotoxic drugs during the hospitalization prior to ICU admission as a
means to control for differences in past exposures.
5.3.4 Covariates
Based on previous research (Chapter 4), the following known risk factors for AKI were
abstracted: age; unplanned ICU admission; admission type (surgical, cardiac; surgical, non-
109
cardiac; medical); International Classification of Disease (ICD-10) (World Health
Organization, 2010) admission diagnoses of circulatory or respiratory disorders, Pediatric
Risk of Mortality (PRISM) score (Pollack et al., 1988); in-ICU neurological or respiratory
dysfunction, defined as a Pediatric Logistic Organ Dysfunction (PELOD) (Leteurtre et al.,
1999) score of one or greater for each system; administration of intravenous radiologic
contrast; and the use of extracorporeal membrane oxygenation (ECMO). All data were
evaluated prior to date of the first diagnosis of AKI (the index date).
5.3.5 Data Analysis
Descriptive statistics were used to characterize the demographic and clinical composition of
the cases and controls. Conditional logistic regression was used to estimate the association
between individual nephrotoxic medications and AKI. Medications with a bivariate
significance level of 0.2 or less were included in a multivariable conditional logistic
regression model. Drugs with no independent association at the bivariate level were grouped
into a single binary variable (‘other nephrotoxic medications’) in order to control for the
potential additive effect of drug exposure. To adjust for underlying differences in AKI risk,
this model included risk factors of AKI found to be statistically significant (p<0.05) in
bivariate analyses. Finally, stratified multivariable models based on prior nephrotoxic
medication exposure were also conducted. All analysis was performed using SAS software,
version 9.3 (SAS Institute Inc., Cary, NC). Small sample sizes (<5) were suppressed for
publication purposes.
This study received approval from The Hospital for Sick Children’s Research Ethics Board.
110
5.4 Results
A total of 6 326 admissions to the ICU were identified during the study period. Of these, 3
865 patients were eligible for the cohort and 915 (23.7%) were categorized as having AKI.
Nine hundred and fourteen case-control pairs were identified and these patients constitute the
study sample that was evaluated.
Forty-one percent of the study sample was over 6 years of age and 43% were female (Table
5.1). Cases and controls were matched 1:1 according to previous exposure to nephrotoxic
drugs: 16.2% (n=148) of case-control pairs were exposed to a nephrotoxic medication during
their hospitalization prior to the ICU admission; the previous exposure status of 24.3%
(n=222) was unknown as they were transferred directly to the ICU from another hospital.
This process identified 914 case-control pairs, with one case unable to be matched a control.
The median observation time, from time of ICU admission to the onset of AKI (the index
date), of the case-control pairs was 28 hours (IQR=43 hours).
111
Table 5.1: Demographics and clinical composition of case and control patients; n(%)
unless specified otherwise
Characteristic Overall (n=1828)
Cases (n=914)
Controls (n=914)
Age group Term neonate (≤27 days) 51 (2.8) 28 (3.1) 23 (2.5) Infant (28 days – 12 months) 670 (36.7) 330 (36.1) 340 (37.2) Toddler (13 months – 2 years) 156 (8.5) 71 (7.8) 85 (9.3) Early childhood (2 – 5 years) 210 (11.5) 100 (10.9) 110 (12.0) Middle childhood (6 – 11 years) 307 (16.8) 141 (15.4) 166 (18.2) Early adolescence (12 – 18
years) 434 (23.7) 244 (26.7) 190 (20.8)
Sex [Female] 778 (42.6) 380 (41.6) 398 (43.5) Nephrotoxic medication exposure prior to ICU admissiona,b
Yes 296 (16.2) 148 (16.2) 148 (16.2) No 1088 (59.5) 544 (59.5) 544 (59.5) Unknown 444 (24.3) 222 (24.3) 222 (24.3) Unplanned admission 1131 (61.8) 552 (60.4) 579 (63.4) Admission type Surgical, cardiac 536 (29.3) 318 (34.8) 218 (23.9) Surgical, non-cardiac 259 (14.2) 101 (11.0) 158 (17.3) Medical 1033 (56.5) 495 (54.2) 538 (58.9) Circulatory admission diagnosis 170 (9.3) 96 (10.7) 72 (7.9) Respiratory admission diagnosis 228 (12.5) 122 (13.4) 106 (11.6) PRISM score (median, IQR) 5.0, 7.0 5.0, 7.0 5.0, 7.0 In-ICU neurologic organ dysfunction
1367 (74.9) 656 (71.8) 711 (77.8)
In-ICU respiratory organ dysfunction
1365 (74.7) 731 (80.0) 634 (69.4)
Intravenous radiologic contrast 337 (18.4) 131 (14.3) 206 (22.5) ECMO 36 (2.0) 20 (2.2) 16 (1.8) In-ICU administration of nephrotoxic medicationb
1474 (80.6) 801 (87.6) 673 (73.6)
ECMO: extracorporeal membrane oxygenation; ICU: intensive care unit; IQR: interquartile range; PRISM: Pediatric Risk of Mortality a Equal distribution between cases and controls due to matching strategy b Nephrotoxic medication defined as administration of any of the following: acyclovir, amikacin, amphotericin, captopril, carboplatin, cidofovir, cisplatinum, cyclophosphamide, cyclosporine, enalapril, enalaprilat, ethacyrnic acid, flucytosine, foscarnet, furosemide, ganciclovir, gentamicin, hydrochlorothiazide, ibuprofen, ifosfamide, indomethacin, ketorolac, methotrexate, penicillin, ramipril, sirolimus, tacrolimus, tobramycin, and vancomycin!
112
When comparing known risk factors for AKI, cases were more likely to be admitted to the
ICU post cardiac surgery (OR=1.73, 95% CI: 1.34-2.22), admitted with an ICD-10 admission
diagnosis of circulatory system disease (OR=1.39, 95% CI: 1.02-1.91) and develop
respiratory dysfunction while in the ICU (OR=1.81, 95% CI: 1.45-2.26) than their matched
controls. Cases were also less likely to develop in-ICU neurological dysfunction (OR=0.72,
95% CI: 0.58-0.89) or receive intravenous radiologic contrast (OR=0.56, 95% CI: 0.43-0.72).
Eighty percent of the study sample was exposed to one or more nephrotoxic medication
during the observation period, with a median of 2 nephrotoxic medications (minimum=0,
maximum=8). Furosemide (administered to 67.8% of patients), vancomycin (28.7%) and
gentamicin (21.4%) were the most frequently administered nephrotoxic medications while
cidofovir (0.2%), enalaprilat (0.2%), foscarnet (0.2%), carboplatin (0.1%), indomethacin
(0.1%), methotrexate (0.1%), ramipril (0.1%) and sirolimus (0.1%) were rarely used.
Cisplatinum, fluctosine and ifosfamide were not administered to any patient during the
observation period.
Cases were significantly more likely than controls to have been exposed to at least one
nephrotoxic medication (OR=2.60, 95% confidence interval (CI): 2.01-3.37). Exposure to
additional nephrotoxic medications increased the risk of AKI by 30% for each additional
drug (OR=1.30, 95% CI: 1.21-1.40). Compared to their matched control, cases were more
likely to have been administered one or more of the following nephrotoxic medications prior
to developing AKI (Table 5.2): the immunosuppressants, cyclosporine (OR=3.0, 95% CI:
1.28-7.06) and tacrolimus (OR=2.06, 95% CI: 1.17-3.61); antibiotics, gentamicin (OR=1.84,
95% CI: 1.44-2.35) and vancomycin (OR=1.29, 95% CI: 1.04-1.59); captopril, an
angiotensin-converting enzyme (ACE) inhibitor (OR=1.74, 95% CI: 1.15-2.64); furosemide,
113
a diuretic (OR=2.23. 95% CI: 1.80 -2.76); ganciclovir, an antiviral (OR=2.93, 95% CI: 1.63-
5.27); and ketorolac, a non-steroidal anti-inflammatory (NSAID) (OR=1.62, 95% CI: 1.18-
2.25).
Table 5.2: Medication exposure during ICU treatment for case and control patients for
all identified nephrotoxic drugs
Medication Cases (n=914) Controls (n=914) ORa (95% CI) Acyclovir 69 (7.6) 63 (6.9) 1.10 (0.77 – 1.57) Amikacin 10 (1.1) 9 (1.0) 1.11 (0.45 – 2.73) Amphotericin 30 (3.3) 23 (2.5) 1.33 (0.76 – 2.35) Captopril 65 (7.1) 39 (4.3) 1.74 (1.15 – 2.64) Carboplatin 0 (0.0) <5 (<0.5) n/a Cidofovir <5 (<0.5) <5 (<0.5) 1.00 (0.14 – 7.10) Cisplatinum 0 (0.0) 0 (0.0) n/a Cyclophosphamide 6 (0.7) <5 (<0.5) 2.00 (0.5 – 8.0) Cyclosporine 21 (2.3) 7 (0.8) 3.00 (1.28 – 7.06) Enalapril 13 (1.4) 8 (0.9) 1.71 (0.68 – 4.35) Enalaprilat <5 (<0.5) <5 (<0.5) 0.50 (0.05 – 5.51) Ethacyrnic acid 24 (2.6) 17 (1.9) 1.47 (0.76 – 2.83) Flucytosine 0 (0.0) 0 (0.0) n/a Foscarnet 0 (0.0) <5 (<0.5) n/a Furosemide 695 (76.1) 544 (59.5) 2.23 (1.80 – 2.76) Ganciclovir 46 (5.0) 17 (1.9) 2.93 (1.63 – 5.27) Gentamicin 238 (26.0) 154 (16.9) 1.84 (1.44 – 2.35) Hydrochlorothiazide 10 (1.1) 5 (0.6) 2.00 (0.68 – 5.85) Ibuprofen 137 (15.0) 114 (12.5) 1.25 (0.95 – 1.64) Ifosfamide 0 (0.0) 0 (0.0) n/a Indomethacin <5 (<0.5) <5 (<0.5) 1.00 (0.06 – 16.00) Ketorolac 109 (11.9) 72 (7.9) 1.62 (1.18 – 2.25) Methotrexate <5 (<0.5) <5 (<0.5) 1.00 (0.06 – 16.00) Penicillin 15 (1.6) 19 (2.1) 0.78 (0.39 – 1.56) Ramipril <5 (<0.5) 0 (0.0) n/a Sirolimus <5 (<0.5) 0 (0.0) n/a Tacrolimus 37 (4.1) 18 (2.0) 2.06 (1.17 – 3.61) Tobramycin 13 (1.4) 10 (1.1) 1.33 (0.56 – 3.16) Vancomycin 284 (31.1) 240 (26.3) 1.29 (1.04 – 1.59) a Adjusted for sampling and matching strategy
114
Nine nephrotoxic medications were included in a multivariable conditional logistic regression
(Table 5.3). The remaining 20 medications, while having no independent association at the
bivariate level, were grouped together to control for the potential additive effect of
medication exposure. Administration of ganciclovir, furosemide, ketorolac or gentamicin
increased the odds of developing AKI. Controlling for underlying differences in risk factors
for AKI, only ganciclovir (adjusted OR=4.23, 95% CI: 1.54-11.58), furosemide (adjusted
OR=1.88, 95% CI: 1.45-2.43), and gentamicin (adjusted OR=1.73, 95% CI: 1.32-2.29)
remained statistically significant.
When the models were stratified according to previous nephrotoxic medication exposure, few
differences were seen in the model results between patients who did and did not receive
nephrotoxic medications during their hospitalization prior to ICU admission (Table 5.4).
115
Table 5.3: Conditional logistic regression model including all nephrotoxic medications
(OR, 95% CI)
Medication Adjusted modela Model further adjusted for risk factorsb
Ganciclovir 2.80 (1.13 – 6.95) 4.23 (1.54 – 11.58) Furosemide 1.99 (1.59 – 2.49) 1.88 (1.45 – 2.43) Gentamicin 1.67 (1.29 – 2.18) 1.73 (1.32 – 2.29) Ketorolac 1.72 (1.22 – 2.42) 1.19 (0.79 – 1.78) Captopril 1.48 (0.96 – 2.29) 1.36 (0.86 – 2.14) Cyclosporine 2.00 (0.78 – 5.14) 1.49 (0.55 – 4.03) Ibuprofen 0.95 (0.70 – 1.28) 0.86 (0.63 – 1.17) Tacrolimus 0.95 (0.40 – 2.27) 0.70 (0.27 – 1.80) Vancomycin 1.14 (0.89 – 1.46) 1.18 (0.90 – 1.54) Other nephrotoxic medicationsc
1.11 (0.84 – 1.47) 1.17 (0.87 – 1.58)
a Adjusted for sampling and matching strategy b Adjusted for: admission type, circulatory admission diagnosis, in-ICU neurological dysfunction, in-ICU respiratory dysfunction, intravenous radiologic contrast c Includes: acyclovir, amikacin, amphotericin, carboplatin, cidofovir, cisplatinum, cyclosporine, enalapril, enalaprilat, ethacyrnic acid, flucytosine, foscarnet, hydrochlorothiazide, ifosfamide, indomethacin, methotrexate, penicillin, ramipril, sirolimus, tobramycin Note: Adjusting for all risk factors (significant and non-significant) did not have an effect on the magnitude of the estimates
116
Table 5.4: Multivariable conditional logistic regression model, stratified by nephrotoxic
medication prior to ICU admission, adjusted for matching strategy and
underlying risk factors for AKIa (OR, 95% CI)
Medication Nephrotoxic medication during hospital admission prior to ICU admission (n=296)
No nephrotoxic medication during hospital admission prior to ICU admission (n=1088)
Ganciclovir 4.14 (0.97 – 17.69) 7.66 (1.39 – 42.16) Furosemide 3.35 (1.34 – 8.38) 1.96 (1.39 – 2.78) Gentamicin 2.03 (1.04 – 3.96) 1.71 (1.18 – 2.49) Ketorolac 1.28 (0.20 – 8.21) 1.13 (0.73 – 1.74) Captopril 1.06 (0.39 – 2.92) 1.53 (0.79 – 2.99) Cyclosporine 1.39 (0.39 – 4.92) 0.49 (0.02 – 15.45) Ibuprofen 0.93 (0.38 – 2.28) 0.91 (0.60 – 1.37) Tacrolimus 0.58 (0.14 – 2.44) 0.50 (0.12 – 2.13) Vancomycin 1.28 (0.68 – 2.41) 0.95 (0.64 – 1.40) Other nephrotoxic medicationsb
1.37 (0.71 – 2.67) 1.30 (0.81 – 2.08)
a Adjusted for: admission type, circulatory admission diagnosis, in-ICU neurological dysfunction, in-ICU respiratory dysfunction, intravenous radiologic contrast b Includes: acyclovir, amikacin, amphotericin, carboplatin, cidofovir, cisplatinum, cyclosporine, enalapril, enalaprilat, ethacyrnic acid, flucytosine, foscarnet, hydrochlorothiazide, ifosfamide, indomethacin, methotrexate, penicillin, ramipril, sirolimus, tobramycin
5.5 Discussion
We conducted a nested case-control study to assess the association between nephrotoxic
medication exposure and the development of AKI in critically ill children. To the best of our
knowledge, this is the first study to do so in this patient population. We found a high
nephrotoxic medication exposure rate with 80% of the study sample being exposed to one or
more nephrotoxic medications during the observation period. A similar rate of nephrotoxic
medication use was reported in a study of non-critically ill children (Moffett & Goldstein,
2011), suggesting that use of nephrotoxic drugs is common among both critically ill and non-
critically ill hospitalized children.
117
We found that patients who developed AKI had an increased odds of exposure to at least one
nephrotoxic medication compared to patients without AKI. However, few individual
medications had a statistically significant association with the development of AKI. Many of
the medications defined as nephrotoxic for the purposes of this study have a low reported
incidence of nephrotoxic side effects but were included due to the lack of paediatric data.
Despite this, our study has captured significant nephrotoxic effects of medications with a
variety of therapeutic uses and administration frequencies in our critically ill patient
population. Those effects were still significant after adjusting for differences in underlying
risk factors of AKI. Interestingly, after adjustment the effect of ganciclovir increased by a
factor of 1.5 (unadjusted OR=2.8, adjusted OR=4.2); in the case of ketorolac, the association
became non-significant after adjustment.
Ganciclovir is known to have nephrotoxic effects; it is insoluble in human urine and
precipitates in the renal tubules causing tubular obstructions (Naughton, 2008). We found
ganciclovir to have the largest nephrotoxic effect (adjusted OR=4.2), though it was
infrequently administered. This relative infrequency of use is likely due to institutional
practice; the use of ganciclovir is restricted at our institution and there are alternative
therapies available. Gentamicin, administered to approximately 21% of the study patients,
was associated with the development of AKI (adjusted OR=1.73). While aminoglycoside
antibacterial agents such as gentamicin are known to have nephrotoxic side effects (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012),
they are used as a cost-effective, first-line treatment due to their efficacy against Gram-
negative bacteria (Rea & Capitano, 2007). Recent guidelines suggest that aminoglycosides
should only be used in cases where no suitable therapeutic alternatives are available (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012).
118
Drugs with similar efficacy, such as amikacin, are thought to also have nephrotoxic effects
(Compendium of Pharmaceuticals and Specialties 2013, 2012; Taketomo et al., 2012);
unfortunately, non-nephrotoxic alternatives are less effective, broad-spectrum antibiotics.
Furosemide was also found to have significant nephrotoxic effects (adjusted OR=1.9). While
its use as a diuretic is important to note as a treatment of AKI, the administration of
furosemide (and all other drugs) in this study occurred prior to the first sign of AKI, which
was conservatively defined as a RIFLE severity of ‘Risk’. As such, it is unlikely that effect
of furosemide in this study was linked to AKI treatment. Recent guidelines have suggested
that diuretics should not be used to prevent or treat AKI, except for the management of fluid
overload (Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury
Working Group, 2012).
5.5.1 Strengths and Limitations
This study utilized a nested case-control design which combines the advantages of both
cohort and case-control studies (Szklo, 2006) and the use of incidence sampling in this study
design is equivalent to matching cases and controls on the duration of follow-up (Szklo,
2006). However, there are some limitations to our study. First, our case definition was based
on the RIFLE (Bellomo et al., 2004) criteria rather than the suggested pediatric modification,
pRIFLE (Akcan-Arikan et al., 2007), which requires the calculation of estimated creatinine
clearance (Schwartz, Brion, & Spitzer, 1987). We were unable to calculate estimated
creatinine clearance as the data sources we used did not capture height. In addition, we were
unable to reconcile medications from other institutions, nor did we have access to medical
records from other institutions. As such, we were unable to assess the exposure to
nephrotoxic medications for patients who were not admitted to our institution prior to their
119
ICU admission; we expect that the subgroup of patients deemed to have ‘unknown’ exposure
to be a combination of patients both with and without previous nephrotoxic drug exposure.
Accordingly, this group was excluded from the stratified analysis. In addition, the use of a
binary exposure for medication exposure may simplify the association between drug
exposure and nephrotoxicity; however, an evaluation of the drug-dose response for each
medication was beyond the scope of this work. Further studies should incorporate a measure
of dosage, either in total dose per kilogram or days of medication exposure. Due to small
sample sizes, we did not evaluate interactions or drug synergies between medications, which
have been reported to be statistically significant in other studies (Moffett & Goldstein, 2011).
It may be that for some drugs, exposure to the individual medication is not as harmful as is
combination with other nephrotoxic drug exposures. It should be noted that a non-significant
nephrotoxic effect in our study does not necessarily indicate that the drug has no nephrotoxic
effect. The infrequent use of some of the medications may have affected the power of our
analyses to detect a significant effect. Furthermore, the study period was limited to the ICU
setting and, as such, we may not have had adequate follow-up time for the effects of certain
medications to be observed, especially those that may require longer use before AKI
symptoms are evident.
5.6 Conclusions
This is the first study to assess the association between nephrotoxic medication exposure and
the development of AKI in critically ill children. We found significant nephrotoxic effects of
drugs in our critically ill patient population; these effects were still significant after adjusting
120
for differences in underlying risk factors of AKI. Nephrotoxic medication exposure was
common among critically ill children, suggesting that alternative therapies should be
considered to reduce the burden of nephrotoxicity in this high-risk population.
121
Chapter 6 General Discussion
122
6.1 Overview
The aim of this dissertation was to contribute new insights to the area of acute kidney injury
and nephrotoxic medication exposure in critically ill children. Specifically, the goal of this
thesis project was to evaluate the contribution of drug therapy to the development of acute
kidney injury (AKI), a common adverse event in critically ill children. In order to
accomplish this goal, we first needed to understand how AKI is defined and operationalized
in the literature, and chose to focus on a specific definition of AKI known as the RIFLE. We
then needed to quantify characteristics and clinical factors that place patients at higher risk
for developing AKI. Finally, combining these results, we evaluated the association between
specific medications and the development of AKI, controlling for underlying differences in
risk. The specific objectives of this thesis work were as follows:
1. Systematically evaluate the published literature describing the use of the RIFLE
definition of acute kidney injury in children
2. Determine the clinical risk factors of acute kidney injury in critically ill children
3. Evaluate the contribution of drug therapy to the development of acute kidney injury.
This chapter will present a summary of the findings for each of the above objectives. In
addition, the strengths and limitations of the dissertation will be presented and the
implications and future directions of the work will also be discussed.
123
6.2 Summary of Major Findings
6.2.1 Summary by Thesis Objectives
A summary of the major findings of this work are presented below, according to the thesis
objectives.
6.2.1.1 Systematic Review of the Use of RIFLE Definition of AKI
In Chapter 3, we systematically evaluated the published literature describing the use of the
RIFLE definition of acute kidney injury in children. We found that the methods used to
diagnose and assess the severity of AKI were incompletely described, with few studies
providing sufficient detail to confirm that the application of the RIFLE was consistent with
the cited methodology. The application of the individual criteria of the RIFLE varied
significantly between studies, including exclusion of the urine output criteria and differences
in the operationalization of the GFR criteria, baseline GFR definitions, and methods for
estimating missing baseline values. Our findings suggest that paediatric populations may be
vulnerable to the same measurement-induced variability that has been reported in adult
populations (Ricci et al., 2008). Furthermore, unlike adult studies (Ricci et al., 2008), our
systematic review demonstrated inconsistent relationships between the RIFLE classification
and measures of mortality and morbidity in children. We found discordant relationships
between increasing RIFLE severity and mortality, length of stay, illness severity and other
measures of kidney function.
Overall, the results of this systematic review suggest that the use of the RIFLE to assess AKI
in children supports the concept of a standardized measure of acute kidney injury in children.
124
However, inconsistencies in the use of the RIFLE have not facilitated the comparison of rates
and outcomes of AKI in this patient population. The measurement and operational issues
raised by this work are relevant to newer definitions of AKI, including the AKIN (R. L.
Mehta et al., 2007) and KDIGO (Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012) definitions of AKI.
6.2.1.2 Determining the Clinical Risk Factors of AKI
In the second study of this thesis work (Chapter 4), we evaluated a large cohort of critically
ill children to determine the clinical risk factors of acute kidney injury in this patient
population. We found that AKI was a common event in the paediatric ICU, occurring in
nearly one-quarter of critically ill children. Our results reinforced the clinical relevance of
AKI in this patient population as AKI was associated with an increased ICU length of stay
and mortality rate.
Twelve independent risk factors of AKI were identified: seven were evident prior to ICU
admission, including age, unplanned admission to the ICU, admission type, admission
diagnoses of circulatory or respiratory disease, illness severity, and administration of a
nephrotoxic medication prior to ICU admission. The five risk factors manifesting during
ICU admission were markers of disease severity (neurologic or respiratory organ
dysfunction) or therapeutic intervention (intravenous contrast, ECMO, nephrotoxic
medication). The single greatest risk factor of ICU-acquired AKI was the in-ICU
administration of nephrotoxic medication(s), underscoring the importance of furthering the
limited information available regarding drug safety in paediatric populations. We found that
administration of radiologic contrast was associated with lower rates of AKI, which adds to
the growing evidence supporting the lack of association between AKI and contrast use
125
(Brasch, 2008; Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury
Working Group, 2012; McDonald et al., 2013; Seeliger et al., 2012; Waybill & Waybill,
2001).
6.2.1.3 Evaluation of the Contribution of Drug Therapy to Acute Kidney Injury
In the third study of this dissertation (Chapter 5), we evaluated the contribution of
nephrotoxic medications to the development of acute kidney injury in critically ill children
using a nested case-control study design. Our results showed that nephrotoxic medication
exposure was common, with over 80% of the study population being exposed to one or more
nephrotoxic medications during the study period. Patients who developed AKI had an
increased odds of exposure to at least one nephrotoxic medication as compared to those
patients without AKI.
We found significant nephrotoxic effects of medications with a variety of therapeutic uses
and administration frequencies in the population of critically ill children studied. These
effects remained significant after adjusting for differences in underlying risk factors of AKI.
The administration of ganciclovir, furosemide and gentamicin was significantly associated
with an increased risk of AKI. Other medications, including ketorolac, captopril,
cyclosporine, and vancomycin were also associated with an increased risk of AKI, though the
results were not statistically significant.
6.2.2 Five Main Findings
Overall, we identified wide variation in the application of a consensus definition of AKI, the
RIFLE. We found a relatively high incidence of AKI (23.7%) among our critically ill
paediatric patient population. After identifying paediatric patients at increased risk of AKI,
126
we found that nephrotoxic medication administration during a patient’s ICU admission was
the single greatest risk factor of AKI. In addition, we found a high incidence of exposure to
nephrotoxic mediations with over 80% of patients exposed to one or more nephrotoxic
medications. This suggests that nephrotoxic medication use should be targeted as a means to
reduce the burden of AKI in this patient population.
There are five major findings from the dissertation work as a whole. These findings will be
used to frame the discussion regarding implications and future directions presented later in
this chapter. The main findings are:
1. Pharmacoepidemiologic methods are feasible to study adverse drug events in children
2. Acute kidney injury is common in critically ill children
3. Definitions of acute kidney injury are used inconsistently
4. Critically ill children who are at high risk for acute kidney injury can be identified
5. Nephrotoxicity is an important contributor to acute kidney injury in critically ill
children.
127
6.3 Strengths and Limitations
6.3.1 Strengths
This dissertation has several strengths, namely the large cohort and comprehensive data
available for analysis, our methodologic approach, and the definition of AKI used as the
outcome measure.
6.3.1.1 Data Sources
Our quantitative analyses were based on a large cohort of critically ill children admitted to
the ICU at The Hospital for Sick Children over a 42-month period. To the best of our
knowledge, this is the largest cohort of critically ill paediatric patients evaluated for AKI to
date. Evaluation of this large patient cohort available for study was feasible due to our use of
electronic patient data. Clinical data for patients admitted to the intensive care unit at The
Hospital for Sick Children are documented electronically, including demographics, clinical
history, recorded vital signs and laboratory results that are documented during the patient’s
stay. Using the patient’s unique medical record number (MRN), we were able to link these
data to data from other clinical databases of radiology procedures, microbiology results, ICD-
coded admission diagnoses, and creatinine values and medication administrations prior to
ICU admission.
6.3.1.2 Methodologic Approach
Due to the large sample available, we were able to assess a large number of potential risk
factors of AKI, some evident prior to ICU admission and others occurring during the ICU
stay. As a number of clinical risk factors of interest were correlated, we addressed the issues
128
of collinearity within our data, excluding variables from the multivariable model and
strengthening our analysis. The use of multivariable modeling (Chapters 4 and 5) allowed us
to assess clinical factors affecting AKI as a whole rather than independently; as such we were
able to model clinical reality more accurately as disease processes and treatments do not
occur in isolation.
Our use of a nested case-control study design (Chapter 5) allowed us to combine the
advantages of both cohort and case-control studies (Szklo, 2006). In addition, the use of
incidence sampling in this study design is equivalent to matching cases and controls on the
duration of follow-up (Szklo, 2006). We studied the nephrotoxic effects of a number of
medications on the risk of AKI, rather than focusing on just one medication. Finally, we
adjusted our estimates of risk to account for differences in underlying risk factors of AKI.
As safety data describing adverse events is limited for paediatric populations (Gilman & Gal,
1992; Uppal et al., 2008), we combined information in drug monographs with clinical expert
opinions. We included a variety of specialties in the expert panel, which included a
paediatric intensivist, a paediatric pharmacologist and paediatric pharmacists. Studies have
shown that panel composition influences ratings, varying across specialties and between
mixed and single-specialty panels (Campbell, Hann, Roland, Quayle, & Shekelle, 1999;
Coulter, Adams, & Shekelle, 1995; Kahan et al., 1996; Leape, Park, Kahan, & Brook, 1992).
It is suggested that expert panels should reflect the full range of stakeholders who have an
interest in the study results (Boulkedid, Abdoul, Loustau, Sibony, & Alberti, 2011).
Heterogeneity amongst panel members also enriches the Delphi process, as different panel
members bring varying points of view regarding the quality of patient care (Hong et al.,
2010).
129
We also evaluated the clinical use of medications rather than looking at dosage, error or
appropriate dosing or prescribing. Defining an ‘appropriate’ dose is complex, especially in
critically ill patients given the complexity of their disease and varying degrees of organ
dysfunction. The use of a binary measure of medication administration was used as a
pragmatic representation of therapeutic intent.
6.3.1.3 Definition of AKI
We used an accepted and widely used measure, the RIFLE, for our outcome of acute kidney
injury, allowing for comparison of our results to other studies and patient populations. At the
time of the thesis work, the RIFLE had been widely used in both adult (Ricci et al., 2008) and
paediatric (M. B. Slater et al., 2012) studies. In addition, we conducted a systematic review of
the existing literature regarding the use of the RIFLE to define AKI in children, allowing us
to apply the lessons learned from our systematic review to our outcome definition and study
methodology for our quantitative analyses. We also demonstrated the feasibility of using
electronic patient data to assess AKI using the RIFLE criteria. We suggest that these results
can be generalized to other, more recently proposed, measures of AKI such as the AKIN and
KDIGO definitions given the modest differences between the RIFLE and these newer
definitions.
6.3.2 Limitations
While this thesis work has a number of strengths, there are also some limitations that should
be noted. Firstly, we studied critically ill children who were admitted to a paediatric,
academic, tertiary care hospital. As such, our results may not be generalizable to
nonacademic hospitals in which most children receive care. However, acutely ill children are
130
likely to be transferred to an academic centre, thus our results are likely generalizable to
critically ill paediatric patient populations. Other limitations regarding data sources, the
operationalization of nephrotoxic medications, the definition of AKI used for the outcome
measure and issues related to the study period are discussed in detail below.
6.3.2.1 Data Sources
Although this work benefitted from the use of electronic patient data from a large cohort,
there are also some limitations as a result of using these data. Some clinical factors thought
to be associated with AKI, such cardiac arrest and duration of cardiac bypass (Aydin et al.,
2012; Fadel et al., 2012; Li et al., 2011; Rosner & Okusa, 2006), are not collected within the
electronic charting systems and thus were not incorporated in our analyses. In addition, our
data sources did not include medication reconciliation from other institutions. As such, we
were only able to assess nephrotoxic drug exposure during the same hospital admission prior
to ICU admission. Medications administered in other hospitals or at home could not be
captured. It is also important to note that the data sources used for the analytic work also
affected our definition of AKI; the pediatric modification of the RIFLE (Akcan-Arikan et al.,
2007) is based on the calculation of estimated creatinine clearance, which requires patient
height, a field which is not routinely captured in our electronic data sources.
Some risk factors had similar definitions and thus the data were highly correlated (e.g.,
systemic inflammatory response syndrome (SIRS) and sepsis, shock and cardiovascular
dysfunction, mechanical ventilation and respiratory dysfunction). As a result, some variables
were excluded from further analysis on the basis of multicollinearity. The exclusion of
unmeasured and collinear variables from multivariable analyses may limit the comparability
of our results to other studies. For example, while we chose to include respiratory
131
dysfunction over mechanical ventilation in our multivariable modeling, others may not have
made the same decision.
6.3.2.2 Medications
We compiled a list of potentially nephrotoxic medications for our analysis. The list was
compiled based on adverse reactions noted in drug monographs and in consultation with
clinical experts; however, we may have excluded potentially relevant medications. As drug
safety data is limited in children, it is possible that there are medications with unknown
nephrotoxic effects in this patient population that were excluded from analysis. We should
also note that some medications that were identified as being nephrotoxic, either through
information in drug monographs or based on clinical expertise, were not found to have a
significant nephrotoxic effect. However, it should be noted that a non-significant effect (i.e.
an odds ratio with a confidence interval that includes 1.0, the null hypothesis) is very
different from a result indicating that a drug is not nephrotoxic (or ‘safe’).
The use of a binary measure of medication exposure may have oversimplified the association
between drug exposure and nephrotoxicity. However, an evaluation of the drug-dose
response for each medication of interest was beyond the scope of this dissertation as we were
interested in the clinical use of medications, a so-called “intent-to-treat” analysis.
Interactions or drug synergies between medications, which have been found to be significant
in other studies (Moffett & Goldstein, 2011), were not assessed. In some drugs, exposure to
the individual medication may not be as harmful as is combination with other nephrotoxic
drug exposures. In addition, the infrequent use of some medications may have affected the
power of our analysis to detect a significant effect.
132
6.3.2.3 Definition of AKI
As previously discussed (Chapter 2), the RIFLE definition of AKI was chosen to define AKI
for this thesis. While other definitions of AKI are available, the RIFLE was chosen for a
number of reasons. First, both the KDIGO definition and the concept of renal angina were
published after the majority of the thesis work was completed. At the time of the initial
planning of the study methodology, the RIFLE was widely used in both adult (Ricci et al.,
2008) and paediatric studies (M. B. Slater et al., 2012). While the AKIN definition (R. L.
Mehta et al., 2007) had recently been introduced, studies showed that both definitions were
concordant when applied to critically ill patients (Bagshaw et al., 2008; Lopes et al., 2008)
and that the AKIN did not improve mortality prediction over the RIFLE (Bagshaw et al.,
2008; Chang et al., 2010). A modification of the RIFLE for use in paediatric populations
(known as the pRIFLE (Akcan-Arikan et al., 2007)) had been introduced however limited
justification was provided for the changes made from the original RIFLE (Akcan-Arikan et
al., 2007). For example, modification of the urine output assessment from a 6-hour to an 8-
hour interval may be illustrative of pragmatic customization of the RIFLE due to institutional
practice or difficulty with measurement accuracy. As such, we felt the use of the RIFLE over
the pRIFLE was justified.
The RIFLE and other definitions of AKI can be used to define a binary measure of acute
kidney injury, as used in this thesis, as well as a measure of the severity of the injury (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012).
While understanding factors that affect changes, either improvement or decline, in kidney
function, we felt it prudent to attempt the proposed pharmacoepidemiologic analysis with a
simpler, dichotomous outcome measure before exploring more complex analyses (i.e.
133
longitudinal analyses with a continuous and/or dynamic outcome measure). The cut-off to
define AKI was the lowest severity level of the RIFLE (‘Risk’). While some may argue that
this results in a more liberal estimate of AKI incidence, research has shown that even small
changes in serum creatinine are associated with adverse clinical outcomes (Coca et al., 2007;
Eric A. J. Hoste et al., 2006; Lassnigg et al., 2004; Nash et al., 2002; Ostermann & Chang,
2007; Shigehiko Uchino et al., 2006) and thus any level of harm to the kidney is important to
capture. The use of this cut-off coincides with the new KDIGO definition of AKI whereby a
patient reaching any severity level or stage of AKI is considered to have AKI (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012).
6.3.2.3.1 Incidence of AKI Using the KDIGO Definition
In 2012, the Kidney Disease: Improving Global Outcomes (KDIGO) group proposed a new
definition and severity classification for AKI. It incorporates previous classifications
(RIFLE, pRIFLE and AKIN) and is the first consensus definition where all elements are
applicable to both paediatric and adult populations (Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012). The major change from
previous definitions is the incorporation of a serum creatinine increase of 0.3 mg/dl within 48
hours to the definition of AKI. While definition was published after the thesis work was
completed, we were interested in understanding how this change in AKI definition affected
the estimated incidence of AKI in our patient population.
As previously reported (Chapter 4), according to the RIFLE definition of AKI, 915 patients
(23.7%) developed AKI during their ICU admission. Based on the KDIGO definition, the
estimated incidence of AKI is 24.0%, with a total of 929 patients classified as developing
AKI in the ICU. Overall, a total of 117 patients met the new criteria, having a serum
134
creatinine increase of 0.3 mg/dl. However, the majority of these patients (88%) also met the
existing criteria of a relative increase in creatinine. Only 14 patients captured by the KDIGO
definition were excluded in the RIFLE definition. Thus, it appears that the addition of a
serum creatinine increase of 0.3 mg/dl does not significantly change the estimated incidence
of AKI.
Little evidence exists in the literature surrounding the benefit of the KDIGO criteria over
other consensus definitions. A study of various AKI definitions in patients admitted with
heart failure reported an AKI incidence of 36.7% based on KDIGO versus 25.6% and 27.9%
according to the RIFLE and AKIN, respectively (Roy et al., 2013). During the first 7 days of
hospitalization for adult patients with acute myocardial infarction (AMI), the incidence of
AKI was 14.8% and 36.6% according to the RIFLE and KDIGO definitions (Rodrigues et al.,
2013). Patients undergoing cardiac surgery had post-surgical AKI rates of 25.9% according
to either the AKIN or KDIGO and 24.9% based on the RIFLE (Bastin et al., 2013). All
studies were performed in adult patient populations and used only the serum creatinine
criteria of the definitions.
6.3.2.4 Study Period
The study period for all quantitative analyses in this thesis was the patient’s ICU admission.
One concern with the study period being limited to a patient’s ICU stay is that this approach
overlooks AKI that presents after ICU discharge. We initially considered following patients
from the time they were admitted to the ICU until they were discharged from the hospital.
However, clinical practice varies between the ICU and the hospital wards. For example, vital
signs are collected more frequently in the ICU. Creatinine is not part of routine clinical
assessment in the ward and thus is only measured or monitored in patients who are judged to
135
be at risk of kidney injury. In addition, urine output is not consistently documented in
patients on the ward. Given the discrepancies in data availability between the ICU and
hospital ward, the study period was limited to the ICU setting and, as such, we may not have
had significant follow-up time for the effects of certain medications to be observed,
especially those that may require longer use before AKI symptoms are evident. The issue of
timing may also explain our findings related to contrast use: serum creatinine levels have
been shown to peak three days after contrast administration (Marenzi et al., 2004; Solomon,
1998) and, as such, the effects of contrast use may not be apparent until after ICU discharge
(i.e. after the period of ‘critical illness’ associated with ICU admission).
6.4 Implications and Future Directions
6.4.1 Implications
This dissertation has added to the literature regarding the incidence of acute kidney injury in
critically ill children and helps to shed light on the importance and scope of AKI in this
patient population. The intent for this section is to focus on the clinical and research
implications of the main findings of the thesis work.
6.4.1.1 Pharmacoepidemiologic Methods Are Feasible
Overall, our work illustrates that a pharmacoepidemiologic approach can be utilized to
quantify and study adverse drug events in paediatric patients, a generally understudied patient
population. Given the small sample sizes available for studies of paediatric patients and the
limited availability of administrative pharmacological data sources for paediatric populations
136
(i.e. for adults over the age of 65, all prescriptions are available through the Ontario Drug
Benefit claims database), using hospital data systems as sources of drug safety may address a
significant gap in drug safety research. This approach can be utilized as a method to study
other adverse drug events in this population or other small patient populations.
6.4.1.2 AKI Is Common In Critically Ill Children
In our large cohort of critically ill children, we found that AKI was a relatively common
occurrence, with roughly one-quarter of critically ill children developing some degree of
renal injury during their ICU admission. This is as common as other types of organ failure
(Devarajan, 2013) and clinicians should be aware of the risk of AKI in critically ill children.
Despite our findings, AKI appears to be underappreciated by clinicians and researchers and is
not recognized by the public (Kellum, Bellomo, & Ronco, 2012). Placing AKI in the
perspective of other well-publicized diseases, AKI has an estimated incidence similar to that
of acute myocardial infarction, a condition widely recognized by clinicians, researchers and
the public (Ali et al., 2007). Strategies are needed to increase awareness of AKI and its
implications amongst all groups.
6.4.1.3 AKI Definitions Are Used Inconsistently
The field of AKI continues to evolve. With the publication of the RIFLE in 2004, four
consensus definitions of AKI have been released over an 8-year period. The most recent
consensus definition, the KDIGO, which incorporates earlier AKI classifications (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012),
appears to be the new, accepted definition of AKI. While an accurate definition of AKI is
important, ever-changing guidelines make it difficult for clinicians and researchers to ensure
137
they are using the most recent definition. Our post-hoc evaluation of the newly proposed
KDIGO criteria illustrated minimal impact of the revised criteria on patients identified as
having AKI, questioning the value of this new refinement. Little evidence exists in the
literature surrounding the benefit of the KDIGO criteria over other consensus definitions and
available data are conflicting (Bastin et al., 2013; Rodrigues et al., 2013; Roy et al., 2013).
Given the inconsistent application of explicitly defined criteria such as the RIFLE, work
should be focused on appropriate usage of existing definitions and improving consistency in
reporting to improve transparency and comparability of study results. Consistent application
of any consensus definition is crucial to provide valid estimates of AKI (Siew et al., 2010;
Zavada et al., 2010) and allow meaningful comparisons across study populations to evaluate
treatments and outcomes of AKI. Any modifications to a standardized outcome definition
such as the RIFLE (Bellomo et al., 2004) or the new KDIGO definition (Kidney Disease:
Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012) reduces
the utility of a consensus definition of AKI.
The importance of this issue becomes more relevant when we consider that these standard
definitions are being used as outcome measures in studies evaluating the effectiveness of
treatments and other interventions and are being used as the “gold standard” comparison for
the development of new biomarkers to improve the timeliness of AKI diagnosis. In order for
clinicians and researchers to make appropriate evidence-based decisions, we must ensure that
the data being analyzed are valid and consistent. Recognizing that pragmatic decisions are
being made when these definitions are applied in epidemiologic studies, consistency and
transparency are important. A number of studies will mention the use of the RIFLE or other
consensus definition in their methodology but will make modifications to the definition,
138
potentially excluding either of the GFR or urine output criteria. Consensus panels should
provide clear directions for how to appropriately use and reference these methodologies
including outlining what modifications or exclusions, if any, are permissible. This will help
to improve the transparency and consistency of the definition of AKI in the literature similar
to the reporting of randomized trials via the Consolidated Standards of Reporting Trials
(CONSORT) statement (Moher, Schulz, & Altman, 2001) or the PRISMA statement for
reporting systematic reviews and meta-analyses (Liberati et al., 2009).
6.4.1.4 Critically Ill Children at Risk for AKI Can Be Identified
Understanding the factors that place children at higher risk of developing AKI will allow
clinicians to prospectively profile patients and adjust therapeutic decisions based on the
patient’s risk of kidney injury. Clinicians need to be able to prospectively disentangle drug
and disease effects on the kidneys in individual patients to appropriately target therapeutic
interventions and avoid high-risk treatments in patients who are more susceptible to AKI. In
certain circumstances, it may be impossible to avoid certain treatments or factors that may
damage kidneys (i.e. no ‘safer’ alternatives exist for a needed therapy) but co-interventions
can be selected to mitigate the negative effects of these treatments.
Our analysis of the risk factors of AKI revealed two important findings: trauma and
radiologic contrast were not associated with subsequent development of AKI. Trauma has
been previously reported as a risk factor of AKI in both adults (Kidney Disease: Improving
Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012; Shigehiko Uchino et
al., 2006) and children (Wohlauer et al., 2012); however, Bailey et al (Bailey et al., 2007) and
Plotz et al (F. B. Plotz et al., 2005) found the association between AKI and an admission
diagnosis of trauma to be statistically non-significant in critically ill children. These studies
139
involved smaller sample sizes and may have been underpowered to detect a statistically
significant difference. More importantly, we found that radiologic contrast administration
was associated with lower rates of AKI. Previous literature has reported contrast
administration as a risk factor of AKI (Brasch, 2008; Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012; Seeliger et al., 2012;
Waybill & Waybill, 2001); however, the association between contrast administration and
AKI has recently been questioned (McDonald et al., 2013). While more work is needed to
elucidate the causal relationships between trauma and contrast administration and the
development of kidney injury, it may be that the factors that were thought to predispose
patients to AKI may have changed (for example, trauma management, patient selection for
contrast, and contrast agents themselves) and there may be a corresponding need to educate
clinicians as to the changing landscape of AKI.
6.4.1.5 Nephrotoxicity Is an Important Contributor to AKI in Critically Ill Children
We found that the single greatest risk factor of ICU-acquired AKI was the administration of
nephrotoxic medications, indicating that scrutiny of the drugs used in the paediatric ICU
setting is warranted. Our results reinforce the high level of nephrotoxicity associated with
furosemide, ganciclovir and gentamicin. While other drugs also showed non-significant
associations with the development of AKI, including captopril, cyclosporine, ketorolac and
vancomycin, it should be noted that a non-significant nephrotoxic effect in our study results
does not necessarily indicate that the drug has no nephrotoxic effect. A lack of association
between a potentially nephrotoxic drug and AKI in our study does not guarantee that the
medication is safe or has no nephrotoxic effects given some of the limitations mentioned
earlier within this chapter. Our work has illustrated the need for more work to understand the
140
nephrotoxic potential of medications in paediatric populations as safety data from adult
populations may not be applicable.
Our findings have implications on prescribing practices: lessons from contrast guidelines
(Benko et al., 2007; Solomon & Deray, 2006; Stacul et al., 2011) and recent
recommendations for aminoglycoside and amphotericin-induced AKI prevention (Kidney
Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012)
should be applied to all potential nephrotoxic drugs: nephrotoxic medications should be used
only where clinically sensible alternatives do not exist for all patients, especially those
identified to be at high-risk.
6.4.2 Future Directions
This section describes the potential future directions for this body of research. Here the most
important new questions that arose from this dissertation are described and means to address
them are proposed.
6.4.2.1 Expand Pharmacoepidemiologic Approach
This work has demonstrated the feasibility of a pharmacoepidemiologic approach to evaluate
adverse drug events in critically ill children. The specific approach used to study acute
kidney injury should be expanded in a number of ways. Firstly, our analysis dealt with a
simple, binary outcome of acute kidney injury. Further work should expand this outcome
definition to incorporate disease severity; there may be certain clinical factors or medications
that are associated with milder or more severe forms of renal injury.
141
Second, future work could incorporate the reversibility of kidney injury. Roughly 80% of
children with hospital-acquired AKI either fully recover or have improved renal function by
hospital discharge (Hui-Stickle et al., 2005) and a recent study of vancomycin-induced AKI
in critically ill children found that the renal injury was resolved in 87% of patients (Totapally
et al., 2013). Incorporating the reversibility of AKI would allow us to not only understand
factors associated with the development of AKI but also factors that affect renal recovery.
This analysis may help develop potential therapies for AKI and assist clinicians in providing
better care and improve outcomes for patients.
Third, the focus of this work was limited to the ICU environment; future studies should look
at post-ICU kidney injury and long-term renal function. This will also allow an expansion of
the definition of AKI to include the full spectrum of severity according to the RIFLE criteria;
this dissertation did not include severity levels of ‘Loss’ and ‘End-stage’ which are defined
by long-term use of renal replacement therapy (>4 weeks) and not applicable given that most
patients are discharged from the ICU within 2 days. Interestingly, newer definitions of AKI,
including the AKIN and KDIGO, do not include renal replacement therapy in their
definitions.
Fourth, enhanced description of the exposure of interest may allow for a more detailed
understanding of nephrotoxicity as a dose-dependent phenomenon. In this work, we
considered drug administration as a binary measure; future work should incorporate drug
dosage or length of drug exposure (Moffett & Goldstein, 2011). For drugs monitored for
therapeutic efficacy (i.e. vancomycin), peak and trough concentrations may be studied. For
example, while a recent study of vancomycin-induced AKI in critically ill children did not
find a significant relationship between the development of AKI and total vancomycin dose
142
and peak or trough levels (Totapally et al., 2013), high vancomycin trough concentrations
have been shown to be associated with an increased risk of nephrotoxicity (Elting et al.,
1998; Hermsen et al., 2010; Hidayat, Hsu, Quist, Shriner, & Wong-Beringer, 2006; Lodise,
Patel, Lomaestro, Rodvold, & Drusano, 2009; McKamy et al., 2011; Pritchard et al., 2010).
We considered 29 medications to have potential nephrotoxic effects and included these drugs
in our analyses. However, given the limited knowledge of drug safety in children, it is
possible that we may have excluded relevant medications. Future work should broaden the
scope of medications used in these analyses or limit a priori assumptions of toxicity of
medications. The role of drug-drug and drug-disease interactions should also be assessed in
future studies. For some drugs, exposure to the individual medication may not be as harmful
as when these therapies are combined with other nephrotoxic exposures; drug interactions
have been reported as having significant associations with nephrotoxicity in other studies
(Moffett & Goldstein, 2011).
The recent changes to the consensus definition have incorporated a rise in creatinine of 0.3
mg/dl SCr within 48 hours, which is applicable to pediatric patients (Kidney Disease:
Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012). We
have shown that this new definition does not significantly alter the estimated incidence of
AKI, however, the quantitative analyses presented here should be rerun to confirm that this
change in definition does not alter the associations described in this dissertation. Finally, in
addition to expanding the scope of the evaluation of AKI, the approach taken with this
adverse drug event could be utilized to study hepatotoxicity, electrolyte or metabolic
disturbances or other adverse events.
143
6.4.2.2 Improve Consistency in the Use of AKI Definitions
Any modifications to a standardized outcome definition such as the RIFLE (Bellomo et al.,
2004) or the new KDIGO definition (Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012) reduces the effectiveness of a consensus
definition of AKI. Further work is needed to understand the impact of the KDIGO criteria on
the estimated incidence of AKI relative to existing definitions; currently little work has been
done in this area and the results are conflicting (Bastin et al., 2013; Rodrigues et al., 2013;
Roy et al., 2013).
Given the inconsistencies in the use of consensus definitions of AKI, work should be focused
on appropriate usage of existing definitions and improving consistency in reporting to
improve transparency and comparability of study results. One solution is the development of
a framework, such as the Consolidated Standards of Reporting Trials (CONSORT) statement
(Moher et al., 2001) or the PRISMA statement for reporting systematic reviews and meta-
analyses (Liberati et al., 2009). Future studies should also look to the clinical relevance of
changes to consensus guidelines. Do clinicians feel that the minor changes made to the
existing definitions (i.e. RIFLE and AKIN) to develop the KDIGO are clinically relevant? A
survey of clinicians may help expert future panels to identify clinically relevant questions and
considerations during the development of consensus definitions. In addition, an analysis
comparing various definitions of AKI and their relationships with clinical outcomes such as
morbidity and mortality is also necessary to determine if one definition is more predictive of
patient outcome than others.
Despite the changes in guidelines, the lessons learned from this thesis regarding the
implementation of RIFLE are applicable to this and other outcome or disease definitions. We
144
found a large variation in how a consensus definition of AKI was applied. It is widely agreed
that urine output has a high sensitivity and specificity to diagnose AKI (Eric A. J. Hoste &
Kellum, 2006), however studies have excluded the urine output criteria from their AKI
definition. If few institutions capture urine output in a usable way, is it appropriate or useful
for a consensus definition to include it? With current and future research focused on the
development of biomarkers a substitute measure for urine output may be found and
incorporated in the next consensus definition. Perhaps rather than continually perfecting a
definition that uses traditional measures of kidney function (serum creatinine and urine
output), future consensus definitions should only be changed when more time-sensitive
biomarkers are developed and readily available for day-to-day clinical use.
While many studies claim that the RIFLE is a validated measure of AKI in both adults (Ricci
et al., 2008) and children (Akcan-Arikan et al., 2007), our work found inconsistent
relationships between the RIFLE classification and measures of mortality and morbidity in
children. Further work is needed to clarify these associations. However, while
demonstration of correlations between the RIFLE and measures of mortality and morbidity
are useful, they are not demonstrative of the RIFLE’s ability to accurately reflect acute
kidney injury. Correlation with mortality is only one form of validation and, on its own,
insufficient. In other words, the RIFLE may be a good measure of illness severity, but may
not truly capture the presence of AKI. Future work should appropriately validate the RIFLE,
or KDIGO definitions of AKI, as a true measure of kidney injury. Ideally, validation would
occur with a comparison of the RIFLE with a ‘gold standard’ measure of acute kidney injury.
However, unlike chronic kidney disease, no ‘gold standard’ diagnostic test exists for the
diagnosis and severity classification of acute kidney injury; it is based on a combination of
patient symptoms, clinical tests, and clinical judgment. Substitute measures of kidney
145
function have been assessed, such as the relationship between the RIFLE and renal
replacement therapy (Duzova et al., 2010; Frans B. Plotz et al., 2008; Zappitelli et al., 2009).
However, the value of these assessments is limited as the use of renal replacement therapy is
captured within the RIFLE definition and non-renal indications for dialysis also exist. For
example, renal replacement therapy can be used to treat patients with tumor lysis syndrome
or drug toxicity (Fleming et al., 2012). Thus, surrogate measures of kidney function that are
not captured in the RIFLE definition should be assessed. For example, one major function of
the kidneys is regulation of electrolytes, abnormally low or high levels of sodium or
potassium would be indicative of kidney dysfunction. Thus, one method of validating the
RIFLE would be to compare RIFLE severity levels against electrolyte levels. Again, it is
useful to point out that issues related to validation also are relevant to other, newer definitions
of AKI, including the KDIGO (Kidney Disease: Improving Global Outcomes (KDIGO)
Acute Kidney Injury Working Group, 2012) consensus definitions.
6.4.2.3 Improve Ability to Identify At Risk Populations
We identified a number of risk factors of AKI in critically ill children. The next step for this
work is to develop and validate a prospective risk assessment or prediction model to help
clinicians identify high-risk patients. This model may take the form of a risk score, such as
the Pediatric Risk of Mortality (PRISM) score (Pollack et al., 1988) or the Paediatric Index of
Mortality (PIM2) score (A. Slater, Shann, & Pearson, 2003), currently used in paediatric
medicine, or may follow a decision tree. The goal is to develop an easy set of rules to help
clinicians identify and target patients at risk for AKI. Once this work is complete, it will
need to be evaluated for its effectiveness: is this information valuable to clinicians? Does it
provide clinicians with a way to easily identify and target at-risk children, improve their
146
ability to provide appropriate and effective care and ultimately improve patient outcomes?
Ultimately, this work could be incorporated into an electronic medical record (EMR) to
provide real-time decision support to clinicians.
This work has also been hypothesis-generating and has highlighted areas where more
research is needed. For example, our work found radiologic contrast administration, long
thought to be a risk factor for AKI (Brasch, 2008; Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012; Seeliger et al., 2012;
Waybill & Waybill, 2001), was associated with lower rates of AKI in our cohort of critically
ill children. A recent analysis of observational studies showed that contrast administration
was not associated with AKI, death or dialysis (McDonald et al., 2013). There are a number
of reasons that may explain these findings (for instance, clinicians may avoid using contrast
in critically ill children who they perceive are at risk for renal injury); as such, further work is
needed to elucidate the causal association between contrast administration and the
development of acute kidney injury in critically ill children. While there may be a number of
explanations for the effect we found, if contrast is indeed nephrotoxic, clinicians may be
approaching patients given contrast in a manner as to dampen the potential injury to the renal
system. It is important to understand this complex process as the lessons from contrast
guidelines (Benko et al., 2007; Solomon & Deray, 2006; Stacul et al., 2011) may be applied
to other nephrotoxic medications to reduce their burden on the renal system. Prospective
studies of clinical practice and clinical decision-making surrounding the treatment of patients
who are administered contrast are needed to elucidate this complex process.
147
6.4.2.4 Improve Safety Profiles and Increase Alternative Treatments to
Nephrotoxic Drugs
Given that our work has shown that nephrotoxic drugs are the primary risk factor of ICU-
acquired kidney injury, investigations into the use of potential alternatives with less
nephrotoxic profiles is necessary. For example, we found that gentamicin was significantly
associated with the development of AKI. While aminoglycoside antibacterial agents such as
gentamicin are known to have nephrotoxic side effects (Kidney Disease: Improving Global
Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012), they are used as a cost-
effective, first-line treatment due to their efficacy against Gram-negative bacteria (Rea &
Capitano, 2007). Recent guidelines suggest that aminoglycosides should only be used in
cases where no suitable therapeutic alternatives are available (Kidney Disease: Improving
Global Outcomes (KDIGO) Acute Kidney Injury Working Group, 2012). Drugs that may
have similar efficacy, such as amikacin and vancomycin, are thought to also have
nephrotoxic effects (Compendium of Pharmaceuticals and Specialties 2013, 2012; Taketomo
et al., 2012); unfortunately, non-nephrotoxic alternatives may be less effective. This points to
an obvious need to find suitable, effective and safe alternatives to medications with
nephrotoxic side effects or safer methods of administration (i.e. once daily aminoglycoside
dosing, continuous infusions of amphotericin B).
Our work has also highlighted a number of drugs for which there is limited empiric
information regarding their nephrotoxic effects. A number of drugs, including captopril,
cyclosporine, ketorolac and vancomycin, showed non-significant associations with the
development of AKI in this work. In the case of vancomycin, studies have shown differing
results: while high vancomycin trough concentrations have been shown to be associated with
148
an increased risk of nephrotoxicity (Elting et al., 1998; Hermsen et al., 2010; Hidayat et al.,
2006; Lodise et al., 2009; McKamy et al., 2011; Pritchard et al., 2010), a recent study of
vancomycin-induced AKI in critically ill children did not find a significant relationship
between the development of AKI and total vancomycin dose and peak or trough levels
(Totapally et al., 2013). Recent adult data shows vancomycin to be minimally nephrotoxic,
even among high-risk, critically ill patients (Davies, Guidry, Petroze, Hranjec, & Sawyer,
2013). This highlights the need for more bench and bedside research to clearly understand
the mechanisms of drug-induced kidney injury.
In addition, our work demonstrates the need for better articulation of what constitutes a
nephrotoxic medication. When refining the list of potentially nephrotoxic medications for
inclusion in the analytic work, our expert panel demonstrated resolvable disagreement which
points to some ambiguity in the field. Our definition of nephrotoxic medications was
supported by ‘overall’ assessment in our cohort study (i.e. administration of a nephrotoxic
medication was associated with the development of acute kidney injury). However, when the
association between individual medications and the development of AKI was studied, the
assumed nephrotoxicity of many medications was not supported by our results. While these
results may be due to the limitations of our work, it may be that, for some medications, the
label of nephrotoxic may be historic and not well deserved. It may be that a definition of
nephrotoxicity of a medication needs to not only be comprehensive in its assessment of
frequency and severity, but may also need to incorporate the mechanism of injury or be
applied separately to specific age groups.
Future work should include interventional studies evaluating the safety and efficacy of
alternatives to nephrotoxic medications. In addition, we suggest that it would be useful to
149
study the effectiveness of applying the recent recommendations for aminoglycoside and
amphotericin-induced AKI prevention (Kidney Disease: Improving Global Outcomes
(KDIGO) Acute Kidney Injury Working Group, 2012) to all potentially nephrotoxic drugs.
This could either be done in a randomized controlled trial, with certain patients or institutions
randomized to follow the recommendations and others randomized to standard care, or in a
before-after study where the standard of care is followed for a certain amount of time and
then the recommendations are applied. In the latter situation, each institution would act as its
own control. If feasibility prohibits these controlled study designs, a pharmacoepidemiologic
approach, such as was used in this thesis, could be used to compare the effects of differing
treatment approaches on the development of AKI, allowing for risk-adjustment to control for
underlying differences between patients.
6.5 Summary
In summary, the current investigation has confirmed that acute kidney injury is a common
event in critically ill children and that nephrotoxic medications are an important contributor
to this disease. The findings from these three studies have provided important contributions
to, and advanced our knowledge of, our understanding nephrotoxic medications and other
risk factors of AKI. Further research should focus on methods to reduce the occurrence of
acute kidney injury in critically ill children, including improving awareness of the burden of
AKI and improving consistency in the use of consensus definitions of AKI. It is necessary to
improve clinicians’ abilities to identify patients at risk of developing AKI and efforts are
needed to provide therapeutic alternatives to medications with high nephrotoxic profiles.
150
Finally, the pharmacoepidemiologic approach used in this dissertation should be expanded in
the study of AKI in critically ill children and could be applied to the study of other adverse
drug events. The results of this and future work can help to optimize prescribing and aid
clinicians in making optimal treatment choices, reducing therapeutic harm and improving
patient outcomes.
151
References
About the BCPA. (2011, May 13, 2011). Retrieved February 1, 2014, from http://bpca.nichd.nih.gov/about/Pages/Index.aspx
Ajami, G., Derakhshan, A., Amoozgar, H., Mohamadi, M., Borzouee, M., Basiratnia, M., . . . Soltani, M. (2010). Risk of Nephropathy after Consumption of Nonionic Contrast Media by Children Undergoing Cardiac Angiography: A Prospective Study. Pediatric Cardiology, 31(5), 668-673.
Akcan-Arikan, A., Zappitelli, M., Loftis, L. L., Washburn, K. K., Jefferson, L. S., & Goldstein, S. L. (2007). Modified Rifle Criteria in Critically Ill Children with Acute Kidney Injury. Kidney International, 71(10), 1028-1035.
Alexander, B. D., & Wingard, J. R. (2005). Study of Renal Safety in Amphotericin B Lipid Complex-Treated Patients. Clinical Infectious Diseases, 40 Suppl 6, S414-421.
Ali, T., Khan, I., Simpson, W., Prescott, G., Townend, J., Smith, W., & Macleod, A. (2007). Incidence and Outcomes in Acute Kidney Injury: A Comprehensive Population-Based Study. Journal of the American Society of Nephrology, 18(4), 1292-1298.
Alkandari, O., Eddington, K. A., Hyder, A., Gauvin, F., Ducruet, T., Gottesman, R., . . . Zappitelli, M. (2011). Acute Kidney Injury Is an Independent Risk Factor for Pediatric Intensive Care Unit Mortality, Longer Length of Stay and Prolonged Mechanical Ventilation in Critically Ill Children: A Two-Center Retrospective Cohort Study. Critical Care, 15(3), R146.
Amann, K., Wanner, C., & Ritz, E. (2006). Cross-Talk between the Kidney and the Cardiovascular System. Journal of the American Society of Nephrology, 17(8), 2112-2119.
American Academy of Pediatrics. Committee on Drugs. Guidelines for the Ethical Conduct of Studies to Evaluate Drugs in Pediatric Populations. (1977). Pediatrics, 60(1), 91-101.
Anderson, G. D. (2002). Children Versus Adults: Pharmacokinetic and Adverse-Effect Differences. Epilepsia, 43 Suppl 3, 53-59.
Andreoli, S. P. (2009). Acute Kidney Injury in Children. Pediatric Nephrology, 24(2), 253-263.
Arany, I., Kaushal, G. P., Portilla, D., Megyesi, J., Price, P. M., & R.L., S. (2008). Cellular Mechanisms of Nephrotoxicity. In M. E. De Broe, G. A. Porter, W. M. Bennett & G. Deray (Eds.), Clinical Nephrotoxins: Renal Injury from Drugs and Chemicals (3rd ed.). Berlin: Springer.
152
Arthur, J. M., Hill, E. G., Alge, J. L., Lewis, E. C., Neely, B. A., Janech, M. G., . . . Shaw, A. D. (2014). Evaluation of 32 Urine Biomarkers to Predict the Progression of Acute Kidney Injury after Cardiac Surgery. Kidney International, 85(2), 431-438.
Ashp Guidelines on Preventing Medication Errors in Hospitals. (1993). American Journal of Hospital Pharmacy, 50(2), 305-314.
Askenazi, D. (2011). Evaluation and Management of Critically Ill Children with Acute Kidney Injury. Current Opinion in Pediatrics, 23(2), 201-207.
Askenazi, D. J., Feig, D. I., Graham, N. M., Hui-Stickle, S., & Goldstein, S. L. (2006). 3-5 Year Longitudinal Follow-up of Pediatric Patients after Acute Renal Failure. Kidney International, 69(1), 184-189.
Askenazi, D. J., Griffin, R., McGwin, G., Carlo, W., & Ambalavanan, N. (2009). Acute Kidney Injury Is Independently Associated with Mortality in Very Low Birthweight Infants: A Matched Case-Control Analysis. Pediatric Nephrology, 24(5), 991-997.
Awdishu, L., & Bouchard, J. (2011). How to Optimize Drug Delivery in Renal Replacement Therapy. Seminars in Dialysis, 24(2), 176-182.
Aydin, S. I., Seiden, H. S., Blaufox, A. D., Parnell, V. A., Choudhury, T., Punnoose, A., & Schneider, J. (2012). Acute Kidney Injury after Surgery for Congenital Heart Disease. The Annals of Thoracic Surgery, 94(5), 1589-1595.
Bagshaw, S. M., George, C., Bellomo, R., & Committe, A. D. M. (2008). A Comparison of the Rifle and Akin Criteria for Acute Kidney Injury in Critically Ill Patients. Nephrology Dialysis Transplantation, 23(5), 1569-1574.
Bailey, D., Phan, V., Litalien, C., Ducruet, T., Merouani, A., Lacroix, J., & Gauvin, F. (2007). Risk Factors of Acute Renal Failure in Critically Ill Children: A Prospective Descriptive Epidemiological Study. Pediatric Critical Care Medicine, 8(1), 29-35.
Baker, G. R., Norton, P. G., Flintoft, V., Blais, R., Brown, A., Cox, J., . . . Tamblyn, R. (2004). The Canadian Adverse Events Study: The Incidence of Adverse Events among Hospital Patients in Canada. Canadian Medical Association Journal, 170(11), 1678-1686.
Barker, K. N., Flynn, E. A., Pepper, G. A., Bates, D. W., & Mikeal, R. L. (2002). Medication Errors Observed in 36 Health Care Facilities. Archives of Internal Medicine, 162(16), 1897-1903.
Barletta, G. M., & Bunchman, T. E. (2004). Acute Renal Failure in Children and Infants. Current Opinion in Critical Care, 10(6), 499-504.
Bastin, A. J., Ostermann, M., Slack, A. J., Diller, G. P., Finney, S. J., & Evans, T. W. (2013). Acute Kidney Injury after Cardiac Surgery According to Risk/Injury/Failure/Loss/End-Stage, Acute Kidney Injury Network, and Kidney
153
Disease: Improving Global Outcomes Classifications. Journal of Critical Care, 28(4), 389-396.
Basu, R. K., Chawla, L. S., Wheeler, D. S., & Goldstein, S. L. (2012). Renal Angina: An Emerging Paradigm to Identify Children at Risk for Acute Kidney Injury. Pediatric Nephrology, 27(7), 1067-1078.
Basu, R. K., Devarajan, P., Wong, H., & Wheeler, D. S. (2011). An Update and Review of Acute Kidney Injury in Pediatrics. Pediatric Critical Care Medicine, 12(3), 339-347.
Basu, R. K., & Wheeler, D. S. (2013). Kidney-Lung Cross-Talk and Acute Kidney Injury. Pediatric Nephrology.
Basu, R. K., Zappitelli, M., Brunner, L., Wang, Y., Wong, H. R., Chawla, L. S., . . . Goldstein, S. L. (2013). Derivation and Validation of the Renal Angina Index to Improve the Prediction of Acute Kidney Injury in Critically Ill Children. Kidney International.
Bates, D. W., Boyle, D. L., Vander Vliet, M. B., Schneider, J., & Leape, L. (1995). Relationship between Medication Errors and Adverse Drug Events. Journal of General Internal Medicine, 10(4), 199-205.
Bates, D. W., Cullen, D. J., Laird, N., Petersen, L. A., Small, S. D., Servi, D., . . . et al. (1995). Incidence of Adverse Drug Events and Potential Adverse Drug Events. Implications for Prevention. Ade Prevention Study Group. Journal of the American Medical Association, 274(1), 29-34.
Bates, D. W., Leape, L. L., & Petrycki, S. (1993). Incidence and Preventability of Adverse Drug Events in Hospitalized Adults. Journal of General Internal Medicine, 8(6), 289-294.
Bates, D. W., Spell, N., Cullen, D. J., Burdick, E., Laird, N., Petersen, L. A., . . . Leape, L. L. (1997). The Costs of Adverse Drug Events in Hospitalized Patients. Adverse Drug Events Prevention Study Group. Journal of the American Medical Association, 277(4), 307-311.
Bavdekar, S. B., & Gogtay, N. J. (2005). Unlicensed and Off-Label Drug Use in Children. Journal of Postgraduate Medicine, 51(4), 249-252.
Beckmann, U., Bohringer, C., Carless, R., Gillies, D. M., Runciman, W. B., Wu, A. W., & Pronovost, P. (2003). Evaluation of Two Methods for Quality Improvement in Intensive Care: Facilitated Incident Monitoring and Retrospective Medical Chart Review. Critical Care Medicine, 31(4), 1006-1011.
Bellomo, R., Kellum, J., & Ronco, C. (2001). Acute Renal Failure: Time for Consensus. Intensive Care Medicine, 27(11), 1685-1688.
Bellomo, R., Kellum, J. A., & Ronco, C. (2012). Acute Kidney Injury. Lancet, 380(9843), 756-766.
154
Bellomo, R., Ronco, C., Kellum, J. A., Mehta, R. L., & Palevsky, P. (2004). Acute Renal Failure - Definition, Outcome Measures, Animal Models, Fluid Therapy and Information Technology Needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (Adqi) Group. Critical Care, 8(4), R204-212.
Benko, A., Fraser-Hill, M., Magner, P., Capusten, B., Barrett, B., Myers, A., & Owen, R. J. (2007). Canadian Association of Radiologists: Consensus Guidelines for the Prevention of Contrast-Induced Nephropathy. Canadian Association of Radiologists Journal, 58(2), 79-87.
Bennett, M., Dent, C. L., Ma, Q., Dastrala, S., Grenier, F., Workman, R., . . . Devarajan, P. (2008). Urine Ngal Predicts Severity of Acute Kidney Injury after Cardiac Surgery: A Prospective Study. Clinical Journal of the American Society of Nephrology, 3(3), 665-673.
Blinder, J. J., Goldstein, S. L., Lee, V. V., Baycroft, A., Fraser, C. D., Nelson, D., & Jefferies, J. L. (2012). Congenital Heart Surgery in Infants: Effects of Acute Kidney Injury on Outcomes. The Journal of Thoracic and Cardiovascular Surgery, 143(2), 368-374.
Blumer, J. L. (1999). Off-Label Uses of Drugs in Children. Pediatrics, 104(3 Pt 2), 598-602.
Boulkedid, R., Abdoul, H., Loustau, M., Sibony, O., & Alberti, C. (2011). Using and Reporting the Delphi Method for Selecting Healthcare Quality Indicators: A Systematic Review. PloS One, 6(6), e20476.
Brasch, R. C. (2008). Contrast Media Toxicity in Children. Pediatric Radiology, 38 Suppl 2, S281-284.
Brennan, T. A., Leape, L. L., Laird, N. M., Hebert, L., Localio, A. R., Lawthers, A. G., . . . Hiatt, H. H. (1991). Incidence of Adverse Events and Negligence in Hospitalized Patients. Results of the Harvard Medical Practice Study I. The New England Journal of Medicine, 324(6), 370-376.
Breslow, L. H. (2003). The Best Pharmaceuticals for Children Act of 2002: The Rise of the Voluntary Incentive Structure and Congressional Refusal to Require Pediatric Testing. Harvard Journal on Legislation, 40(1), 133-193.
Bresolin, N., Silva, C., Halllal, A., Toporovski, J., Fernandes, V., Goes, J., & Carvalho, F. L. (2009). Prognosis for Children with Acute Kidney Injury in the Intensive Care Unit. Pediatric Nephrology, 24(3), 537-544.
Buckley, M. S., Erstad, B. L., Kopp, B. J., Theodorou, A. A., & Priestley, G. (2007). Direct Observation Approach for Detecting Medication Errors and Adverse Drug Events in a Pediatric Intensive Care Unit. Pediatric Critical Care Medicine, 8(2), 145-152.
Calabrese, A. D., Erstad, B. L., Brandl, K., Barletta, J. F., Kane, S. L., & Sherman, D. S. (2001). Medication Administration Errors in Adult Patients in the Icu. Intensive Care Medicine, 27(10), 1592-1598.
155
Camire, E., Moyen, E., & Stelfox, H. T. (2009). Medication Errors in Critical Care: Risk Factors, Prevention and Disclosure. CMAJ : Canadian Medical Association Journal, 180(9), 936-943.
Campbell, S. M., Hann, M., Roland, M. O., Quayle, J. A., & Shekelle, P. G. (1999). The Effect of Panel Membership and Feedback on Ratings in a Two-Round Delphi Survey: Results of a Randomized Controlled Trial. Medical Care, 37(9), 964-968.
Carleton, B., Poole, R., Smith, M., Leeder, J., Ghannadan, R., Ross, C., . . . Hayden, M. (2009). Adverse Drug Reaction Active Surveillance: Developing a National Network in Canada's Children's Hospitals. Pharmacoepidemiology and Drug Safety, 18(8), 713-721.
Chang, C.-H., Lin, C.-Y., Tian, Y.-C., Jenq, C.-C., Chang, M.-Y., Chen, Y.-C., . . . Yang, C.-W. (2010). Acute Kidney Injury Classification: Comparison of Akin and Rifle Criteria. Shock, 33(3), 247-252.
Chantler, C., Garnett, E. S., Parsons, V., & Veall, N. (1969). Glomerular Filtration Rate Measurement in Man by the Single Injection Methods Using 51cr-Edta. Clinical Science, 37(1), 169-180.
Chertow, G. M., Burdick, E., Honour, M., Bonventre, J. V., & Bates, D. W. (2005). Acute Kidney Injury, Mortality, Length of Stay, and Costs in Hospitalized Patients. Journal of the American Society of Nephrology, 16(11), 3365-3370.
Choonara, I., & Conroy, S. (2002). Unlicensed and Off-Label Drug Use in Children: Implications for Safety. Drug safety : an international journal of medical toxicology and drug experience, 25(1), 1-5.
Ciarimboli, G., Koepsell, H., Iordanova, M., Gorboulev, V., Durner, B., Lang, D., . . . Schlatter, E. (2005). Individual Pkc-Phosphorylation Sites in Organic Cation Transporter 1 Determine Substrate Selectivity and Transport Regulation. Journal of the American Society of Nephrology, 16(6), 1562-1570.
Coca, S. G., Peixoto, A. J., Garg, A. X., Krumholz, H. M., & Parikh, C. R. (2007). The Prognostic Importance of a Small Acute Decrement in Kidney Function in Hospitalized Patients: A Systematic Review and Meta-Analysis. American journal of kidney diseases : the official journal of the National Kidney Foundation, 50(5), 712-720.
Coca, S. G., Singanamala, S., & Parikh, C. R. (2012). Chronic Kidney Disease after Acute Kidney Injury: A Systematic Review and Meta-Analysis. Kidney International, 81(5), 442-448.
Cohen, J. (1968). Weighted Kappa: Nominal Scale Agreement with Provision for Scaled Disagreement or Partial Credit. Psychological Bulletin, 70(4), 213-220.
Compendium of Pharmaceuticals and Specialties 2013. (2012). Toronto: Canadian Pharmaceutical Association.
156
Conroy, S., & McIntyre, J. (2005). The Use of Unlicensed and Off-Label Medicines in the Neonate. Seminars in Fetal & Neonatal Medicine, 10(2), 115-122.
Coulter, I., Adams, A., & Shekelle, P. (1995). Impact of Varying Panel Membership on Ratings of Appropriateness in Consensus Panels: A Comparison of a Multi- and Single Disciplinary Panel. Health Services Research, 30(4), 577-591.
Cullen, D. J., Sweitzer, B. J., Bates, D. W., Burdick, E., Edmondson, A., & Leape, L. L. (1997). Preventable Adverse Drug Events in Hospitalized Patients: A Comparative Study of Intensive Care and General Care Units. Critical Care Medicine, 25(8), 1289-1297.
Cummings, B. S., & Schnellmann, R. G. (2001). Pathophysiology of Nephrotoxic Cellular Injury. In R. W. Schrier (Ed.), Diseases of the Kidney and Urogenital Tract (pp. 1071-1136). Philadelphia, PA: Lippincott Williams & Wilkinson.
Cuzzolin, L., Zaccaron, A., & Fanos, V. (2003). Unlicensed and Off-Label Uses of Drugs in Paediatrics: A Review of the Literature. Fundamental & Clinical Pharmacology, 17(1), 125-131.
Davidman, M., Olson, P., Kohen, J., Leither, T., & Kjellstrand, C. (1991). Iatrogenic Renal Disease. Archives of Internal Medicine, 151(9), 1809-1812.
Davies, S. W., Guidry, C. A., Petroze, R. T., Hranjec, T., & Sawyer, R. G. (2013). Vancomycin and Nephrotoxicity: Just Another Myth? Journal of Trauma and Acute Care Surgery, 75(5), 830-835.
de Mendonca, A., Vincent, J. L., Suter, P. M., Moreno, R., Dearden, N. M., Antonelli, M., . . . Cantraine, F. (2000). Acute Renal Failure in the Icu: Risk Factors and Outcome Evaluated by the Sofa Score. Intensive Care Medicine, 26(7), 915-921.
Dennen, P., Altmann, C., Kaufman, J., Klein, C. L., Andres-Hernando, A., Ahuja, N. H., . . . Faubel, S. (2010). Urine Interleukin-6 Is an Early Biomarker of Acute Kidney Injury in Children Undergoing Cardiac Surgery. Critical Care, 14(5), R181.
Dent, C. L., Ma, Q., Dastrala, S., Bennett, M., Mitsnefes, M. M., Barasch, J., & Devarajan, P. (2007). Plasma Neutrophil Gelatinase-Associated Lipocalin Predicts Acute Kidney Injury, Morbidity and Mortality after Pediatric Cardiac Surgery: A Prospective Uncontrolled Cohort Study. Critical Care, 11(6), R127.
Devarajan, P. (2008). The Future of Pediatric Acute Kidney Injury Management--Biomarkers. Seminars in Nephrology, 28(5), 493-498.
Devarajan, P. (2011). Biomarkers for the Early Detection of Acute Kidney Injury. Current Opinion in Pediatrics, 23(2), 194-200.
Devarajan, P. (2013). Pediatric Acute Kidney Injury: Different from Acute Renal Failure but How and Why. Current Pediatrics Reports, 1(1), 34-40.
157
Dick, A., Keady, S., Mohamed, F., Brayley, S., Thomson, M., Lloyd, B. W., . . . Afzal, N. A. (2003). Use of Unlicensed and Off-Label Medications in Paediatric Gastroenterology with a Review of the Commonly Used Formularies in the Uk. Alimentary Pharmacology & Therapeutics, 17(4), 571-575.
Donchin, Y., Gopher, D., Olin, M., Badihi, Y., Biesky, M., Sprung, C. L., . . . Cotev, S. (1995). A Look into the Nature and Causes of Human Errors in the Intensive Care Unit. Critical Care Medicine, 23(2), 294-300.
Donchin, Y., & Seagull, F. J. (2002). The Hostile Environment of the Intensive Care Unit. Current Opinion in Critical Care, 8(4), 316-320.
Du, Y., Zappitelli, M., Mian, A., Bennett, M., Ma, Q., Devarajan, P., . . . Goldstein, S. L. (2011). Urinary Biomarkers to Detect Acute Kidney Injury in the Pediatric Emergency Center. Pediatric Nephrology, 26(2), 267-274.
Duzova, A., Bakkaloglu, A., Kalyoncu, M., Poyrazoglu, H., Delibas, A., Ozkaya, O., . . . Turkish Society for Pediatric Nephrology Acute Kidney Injury Study, G. (2010). Etiology and Outcome of Acute Kidney Injury in Children. Pediatric Nephrology, 25(8), 1453-1461.
Elting, L. S., Rubenstein, E. B., Kurtin, D., Rolston, K. V., Fangtang, J., Martin, C. G., . . . Bodey, G. P. (1998). Mississippi Mud in the 1990s: Risks and Outcomes of Vancomycin-Associated Toxicity in General Oncology Practice. Cancer, 83(12), 2597-2607.
Ernst, F. R., & Grizzle, A. J. (2001). Drug-Related Morbidity and Mortality: Updating the Cost-of-Illness Model. Journal of the American Pharmaceutical Association, 41(2), 192-199.
Etwel, F. A., Rieder, M. J., Bend, J. R., & Koren, G. (2008). A Surveillance Method for the Early Identification of Idiosyncratic Adverse Drug Reactions. Drug Safety, 31(2), 169-180.
Evenepoel, P. (2004). Acute Toxic Renal Failure. Best Practice & Research. Clinical Anaesthesiology, 18(1), 37-52.
Eyler, R. F., & Mueller, B. A. (2011). Antibiotic Dosing in Critically Ill Patients with Acute Kidney Injury. Nature reviews. Nephrology, 7(4), 226-235.
Fadel, F. I., Abdel Rahman, A. M., Mohamed, M. F., Habib, S. A., Ibrahim, M. H., Sleem, Z. S., . . . Soliman, M. M. (2012). Plasma Neutrophil Gelatinase-Associated Lipocalin as an Early Biomarker for Prediction of Acute Kidney Injury after Cardio-Pulmonary Bypass in Pediatric Cardiac Surgery. Archives of medical science : AMS, 8(2), 250-255.
Fernandez, E., Perez, R., Hernandez, A., Tejada, P., Arteta, M., & Ramos, J. T. (2011). Factors and Mechanisms for Pharmacokinetic Differences between Pediatric Population and Adults. Pharmaceutics, 3, 52-72.
158
Ferranti, J., Horvath, M. M., Cozart, H., Whitehurst, J., & Eckstrand, J. (2008). Reevaluating the Safety Profile of Pediatrics: A Comparison of Computerized Adverse Drug Event Surveillance and Voluntary Reporting in the Pediatric Environment. Pediatrics, 121(5), e1201-1207.
Fink, A., Kosecoff, J., Chassin, M., & Brook, R. H. (1984). Consensus Methods: Characteristics and Guidelines for Use. Am J Public Health, 74(9), 979-983.
Fischer, J. E., Bachmann, L. M., & Jaeschke, R. (2003). A Readers' Guide to the Interpretation of Diagnostic Test Properties: Clinical Example of Sepsis. Intensive Care Medicine, 29(7), 1043-1051.
Fleming, G. M., Walters, S., Goldstein, S. L., Alexander, S. R., Baum, M. A., Blowey, D. L., . . . Prospective Pediatric Continuous Renal Replacement Therapy Registry, G. (2012). Nonrenal Indications for Continuous Renal Replacement Therapy: A Report from the Prospective Pediatric Continuous Renal Replacement Therapy Registry Group. Pediatric Critical Care Medicine, 13(5), e299-304.
Gambaro, G., & Perazella, M. A. (2003). Adverse Renal Effects of Anti-Inflammatory Agents: Evaluation of Selective and Nonselective Cyclooxygenase Inhibitors. J Intern Med, 253(6), 643-652.
Ghaleb, M. A., Barber, N., Franklin, B. D., Yeung, V. W., Khaki, Z. F., & Wong, I. C. (2006). Systematic Review of Medication Errors in Pediatric Patients. The Annals of pharmacotherapy, 40(10), 1766-1776.
Gill, A. M., Leach, H. J., Hughes, J., Barker, C., Nunn, A. J., & Choonara, I. (1995). Adverse Drug Reactions in a Paediatric Intensive Care Unit. Acta Paediatrica, 84(4), 438-441.
Gilman, J. T., & Gal, P. (1992). Pharmacokinetic and Pharmacodynamic Data Collection in Children and Neonates. A Quiet Frontier. Clinical Pharmacokinetics, 23(1), 1-9.
Goldstein, B., Giroir, B., & Randolph, A. (2005). International Pediatric Sepsis Consensus Conference: Definitions for Sepsis and Organ Dysfunction in Pediatrics. Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies, 6(1), 2-8.
Goldstein, S. L., & Chawla, L. S. (2010). Renal Angina. Clinical journal of the American Society of Nephrology : CJASN, 5(5), 943-949.
Gonzalez-Martin, G., Caroca, C. M., & Paris, E. (1998). Adverse Drug Reactions (Adrs) in Hospitalized Pediatric Patients. A Prospective Study. International Journal of Clinical Pharmacology and Therapeutics, 36(10), 530-533.
Gourishankar, S., Courtney, M., Jhangri, G. S., Cembrowski, G., & Pannu, N. (2008). Serum Cystatin C Performs Similarly to Traditional Markers of Kidney Function in the Evaluation of Donor Kidney Function Prior to and Following Unilateral Nephrectomy. Nephrology, dialysis, transplantation : official publication of the
159
European Dialysis and Transplant Association - European Renal Association, 23(9), 3004-3009.
Grams, M. E., & Rabb, H. (2012). The Distant Organ Effects of Acute Kidney Injury. Kidney International, 81(10), 942-948.
Greenhill, L. L., Vitiello, B., Abikoff, H., Levine, J., March, J. S., Riddle, M. A., . . . Zito, J. M. (2003). Developing Methodologies for Monitoring Long-Term Safety of Psychotropic Medications in Children: Report on the Nimh Conference, September 25, 2000. Journal of the American Academy of Child and Adolescent Psychiatry, 42(6), 651-655.
Guidelines for the Ethical Conduct of Studies to Evaluate Drugs in Pediatric Populations. Committee on Drugs, American Academy of Pediatrics. (1995). Pediatrics, 95(2), 286-294.
Guignard, J. P., & Drukker, A. (1999). Why Do Newborn Infants Have a High Plasma Creatinine? Pediatrics, 103(4).
Guo, X., & Nzerue, C. (2002). How to Prevent, Recognize, and Treat Drug-Induced Nephrotoxicity. Cleveland Clinic Journal of Medicine, 69(4), 289-290, 293-284, 296-287 passim.
Haase, M., Bellomo, R., Devarajan, P., Schlattmann, P., & Haase-Fielitz, A. (2009). Accuracy of Neutrophil Gelatinase-Associated Lipocalin (Ngal) in Diagnosis and Prognosis in Acute Kidney Injury: A Systematic Review and Meta-Analysis. American Journal of Kidney Diseases, 54(6), 1012-1024.
Haase, M., Devarajan, P., Haase-Fielitz, A., Bellomo, R., Cruz, D. N., Wagener, G., . . . Mertens, P. R. (2011). The Outcome of Neutrophil Gelatinase-Associated Lipocalin-Positive Subclinical Acute Kidney Injury: A Multicenter Pooled Analysis of Prospective Studies. Journal of the American College of Cardiology, 57(17), 1752-1761.
Hall, I. E., Coca, S. G., Perazella, M. A., Eko, U. U., Luciano, R. L., Peter, P. R., . . . Parikh, C. R. (2011). Risk of Poor Outcomes with Novel and Traditional Biomarkers at Clinical Aki Diagnosis. Clinical Journal of The American Society of Nephrology: CJASN, 6(12), 2740-2749.
Harty, L., Johnson, K., & Power, A. (2006). Race and Ethnicity in the Era of Emerging Pharmacogenomics. Journal of Clinical Pharmacology, 46(4), 405-407.
Hassinger, A. B., Backer, C. L., Lane, J. C., Haymond, S., Wang, D., & Wald, E. L. (2012). Predictive Power of Serum Cystatin C to Detect Acute Kidney Injury and Pediatric-Modified Rifle Class in Children Undergoing Cardiac Surgery. Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies, 13(4), 435-440.
160
Herget-Rosenthal, S., Marggraf, G., Husing, J., Goring, F., Pietruck, F., Janssen, O., . . . Kribben, A. (2004). Early Detection of Acute Renal Failure by Serum Cystatin C. Kidney International, 66(3), 1115-1122.
Hermsen, E. D., Hanson, M., Sankaranarayanan, J., Stoner, J. A., Florescu, M. C., & Rupp, M. E. (2010). Clinical Outcomes and Nephrotoxicity Associated with Vancomycin Trough Concentrations During Treatment of Deep-Seated Infections. Expert Opinion on Drug Safety, 9(1), 9-14.
Hidayat, L. K., Hsu, D. I., Quist, R., Shriner, K. A., & Wong-Beringer, A. (2006). High-Dose Vancomycin Therapy for Methicillin-Resistant Staphylococcus Aureus Infections: Efficacy and Toxicity. Archives of Internal Medicine, 166(19), 2138-2144.
Hoffman, T. M., Wernovsky, G., Atz, A. M., Bailey, J. M., Akbary, A., Kocsis, J. F., . . . Wessel, D. L. (2002). Prophylactic Intravenous Use of Milrinone after Cardiac Operation in Pediatrics (Primacorp) Study. Prophylactic Intravenous Use of Milrinone after Cardiac Operation in Pediatrics. American Heart Journal, 143(1), 15-21.
Hong, C. S., Atlas, S. J., Chang, Y., Subramanian, S. V., Ashburner, J. M., Barry, M. J., & Grant, R. W. (2010). Relationship between Patient Panel Characteristics and Primary Care Physician Clinical Performance Rankings. Journal of the American Medical Association, 304(10), 1107-1113.
Hoste, E. A. J., Clermont, G., Kersten, A., Venkataraman, R., Angus, D. C., De Bacquer, D., & Kellum, J. A. (2006). Rifle Criteria for Acute Kidney Injury Are Associated with Hospital Mortality in Critically Ill Patients: A Cohort Analysis. Critical Care, 10(3), R73.
Hoste, E. A. J., & De Corte, W. (2011). Clinical Consequences of Acute Kidney Injury. In J. A. Kellum, C. Ronco & J. L. Vincent (Eds.), Controversies in Acute Kidney Injury (Vol. 174, pp. 56-64). Basel: Karger.
Hoste, E. A. J., & Kellum, J. A. (2006). Acute Kidney Injury: Epidemiology and Diagnostic Criteria. Current Opinion in Critical Care, 12(6), 531-537.
Hou, S. H., Bushinsky, D. A., Wish, J. B., Cohen, J. J., & Harrington, J. T. (1983). Hospital-Acquired Renal Insufficiency: A Prospective Study. The American journal of medicine, 74(2), 243-248.
Hsieh, E. M., Hornik, C. P., Clark, R. H., Laughon, M. M., Benjamin, D. K., Jr., Smith, P. B., & on Behalf of the Best Pharmaceuticals for Children Act-Pediatric Trials, N. (2013). Medication Use in the Neonatal Intensive Care Unit. American Journal of Perinatology.
Hui-Stickle, S., Brewer, E. D., & Goldstein, S. L. (2005). Pediatric Arf Epidemiology at a Tertiary Care Center from 1999 to 2001. American journal of kidney diseases : the official journal of the National Kidney Foundation, 45(1), 96-101.
161
Hussain, E., & Kao, E. (2005). Medication Safety and Transfusion Errors in the Icu and Beyond. Critical Care Clinics, 21(1), 91-110, ix.
Impicciatore, P., Choonara, I., Clarkson, A., Provasi, D., Pandolfini, C., & Bonati, M. (2001). Incidence of Adverse Drug Reactions in Paediatric in/out-Patients: A Systematic Review and Meta-Analysis of Prospective Studies. British Journal of Clinical Pharmacology, 52(1), 77-83.
International Drug Monitoring. The Role of the Hospital. (1969). World Health Organization Technical Report Series, 425, 5-24.
Izzedine, H., Launay-Vacher, V., & Deray, G. (2005). Antiviral Drug-Induced Nephrotoxicity. American Journal of Kidney Diseases, 45(5), 804-817.
K/DOQI Clinical Practice Guidelines for Chronic Kidney Disease: Evaluation, Classification, and Stratification. (2002). American Journal of Kidney Diseases, 39(2 Suppl 1), S1-266.
Kahan, J. P., Park, R. E., Leape, L. L., Bernstein, S. J., Hilborne, L. H., Parker, L., . . . Brook, R. H. (1996). Variations by Specialty in Physician Ratings of the Appropriateness and Necessity of Indications for Procedures. Medical Care, 34(6), 512-523.
Kane-Gill, S., & Weber, R. J. (2006). Principles and Practices of Medication Safety in the Icu. Critical Care Clinics, 22(2), 273-290, vi.
Kaushal, R., Bates, D. W., Landrigan, C., McKenna, K. J., Clapp, M. D., Federico, F., & Goldmann, D. A. (2001). Medication Errors and Adverse Drug Events in Pediatric Inpatients. JAMA : the journal of the American Medical Association, 285(16), 2114-2120.
Kavaz, A., Ozcakar, Z. B., Kendirli, T., Ozturk, B. B., Ekim, M., & Yalcinkaya, F. (2012). Acute Kidney Injury in a Paediatric Intensive Care Unit: Comparison of the Prifle and Akin Criteria. Acta Paediatrica, 101(3), e126-129.
Kearns, G. L. (2000). Impact of Developmental Pharmacology on Pediatric Study Design: Overcoming the Challenges. The Journal of allergy and clinical immunology, 106(3 Suppl), S128-138.
Kearns, G. L., Abdel-Rahman, S. M., Alander, S. W., Blowey, D. L., Leeder, J. S., & Kauffman, R. E. (2003). Developmental Pharmacology--Drug Disposition, Action, and Therapy in Infants and Children. The New England journal of medicine, 349(12), 1157-1167.
Kellum, J. A., Bellomo, R., & Ronco, C. (2012). Kidney Attack. JAMA : the journal of the American Medical Association, 307(21), 2265-2266.
Kelly, W. N. (2008). How Can I Recognize an Adverse Drug Event. http://www.medscape.org/viewarticle/569794
162
Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Working Group. (2012). Kdigo Clinical Practice Guideline for Acute Kidney Injury. Kidney International, Supplement 2, 1-138.
Kilbridge, P. M., Noirot, L. A., Reichley, R. M., Berchelmann, K. M., Schneider, C., Heard, K. M., . . . Bailey, T. C. (2009). Computerized Surveillance for Adverse Drug Events in a Pediatric Hospital. Journal of the American Medical Informatics Association : JAMIA, 16(5), 607-612.
Kimland, E., Bergman, U., Lindemalm, S., & Bottiger, Y. (2007). Drug Related Problems and Off-Label Drug Treatment in Children as Seen at a Drug Information Centre. European Journal of Pediatrics, 166(6), 527-532.
Kinsey, V. E., & Zacharias, L. (1949). Retrolental Fibroplasia; Incidence in Different Localities in Recent Years and a Correlation of the Incidence with Treatment Given the Infants. Journal of the American Medical Association, 139(9), 572-578.
Kohli, H. S., Bhaskaran, M. C., Muthukumar, T., Thennarasu, K., Sud, K., Jha, V., . . . Sakhuja, V. (2000). Treatment-Related Acute Renal Failure in the Elderly: A Hospital-Based Prospective Study. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association, 15(2), 212-217.
Koren, G., & Haslam, R. H. (1994). Pediatric Medication Errors: Predicting and Preventing Tenfold Disasters. Journal of Clinical Pharmacology, 34(11), 1043-1045.
Koyner, J. L., Garg, A. X., Coca, S. G., Sint, K., Thiessen-Philbrook, H., Patel, U. D., . . . Consortium, T.-A. (2012). Biomarkers Predict Progression of Acute Kidney Injury after Cardiac Surgery. Journal of the American Society of Nephrology, 23(5), 905-914.
Koyner, J. L., Vaidya, V. S., Bennett, M. R., Ma, Q., Worcester, E., Akhter, S. A., . . . Murray, P. T. (2010). Urinary Biomarkers in the Clinical Prognosis and Early Detection of Acute Kidney Injury. Clinical Journal of The American Society of Nephrology: CJASN, 5(12), 2154-2165.
Krahenbuhl-Melcher, A., Schlienger, R., Lampert, M., Haschke, M., Drewe, J., & Krahenbuhl, S. (2007). Drug-Related Problems in Hospitals: A Review of the Recent Literature. Drug safety : an international journal of medical toxicology and drug experience, 30(5), 379-407.
Krawczeski, C. D., Goldstein, S. L., Woo, J. G., Wang, Y., Piyaphanee, N., Ma, Q., . . . Devarajan, P. (2011). Temporal Relationship and Predictive Value of Urinary Acute Kidney Injury Biomarkers after Pediatric Cardiopulmonary Bypass. Journal of the American College of Cardiology, 58(22), 2301-2309.
Krawczeski, C. D., Woo, J. G., Wang, Y., Bennett, M. R., Ma, Q., & Devarajan, P. (2011). Neutrophil Gelatinase-Associated Lipocalin Concentrations Predict Development of
163
Acute Kidney Injury in Neonates and Children after Cardiopulmonary Bypass. The Journal of pediatrics, 158(6), 1009-1015 e1001.
Lane, K., Dixon, J. J., Macphee, I. A., & Philips, B. J. (2013). Renohepatic Crosstalk: Does Acute Kidney Injury Cause Liver Dysfunction? Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association, 28(7), 1634-1647.
Lassnigg, A., Schmidlin, D., Mouhieddine, M., Bachmann, L. M., Druml, W., Bauer, P., & Hiesmayr, M. (2004). Minimal Changes of Serum Creatinine Predict Prognosis in Patients after Cardiothoracic Surgery: A Prospective Cohort Study. Journal of the American Society of Nephrology : JASN, 15(6), 1597-1605.
Lazarou, J., Pomeranz, B. H., & Corey, P. N. (1998). Incidence of Adverse Drug Reactions in Hospitalized Patients: A Meta-Analysis of Prospective Studies. JAMA : the journal of the American Medical Association, 279(15), 1200-1205.
Le, J., Nguyen, T., Law, A. V., & Hodding, J. (2006). Adverse Drug Reactions among Children over a 10-Year Period. Pediatrics, 118(2), 555-562.
Leape, L. L., Brennan, T. A., Laird, N., Lawthers, A. G., Localio, A. R., Barnes, B. A., . . . Hiatt, H. (1991). The Nature of Adverse Events in Hospitalized Patients. Results of the Harvard Medical Practice Study Ii. The New England journal of medicine, 324(6), 377-384.
Leape, L. L., Park, R. E., Kahan, J. P., & Brook, R. H. (1992). Group Judgments of Appropriateness: The Effect of Panel Composition. Quality Assurance in Health Care, 4(2), 151-159.
Leeder, J. S. (2003). Developmental and Pediatric Pharmacogenomics. Pharmacogenomics, 4(3), 331-341.
Leteurtre, S., Martinot, A., Duhamel, A., Gauvin, F., Grandbastien, B., Nam, T. V., . . . Leclerc, F. (1999). Development of a Pediatric Multiple Organ Dysfunction Score: Use of Two Strategies. Medical decision making : an international journal of the Society for Medical Decision Making, 19(4), 399-410.
Leteurtre, S., Martinot, A., Duhamel, A., Proulx, F., Grandbastien, B., Cotting, J., . . . Leclerc, F. (2003). Validation of the Paediatric Logistic Organ Dysfunction (Pelod) Score: Prospective, Observational, Multicentre Study. Lancet, 362(9379), 192-197.
Lex, D. J., Toth, R., Cserep, Z., Alexander, S. I., Breuer, T., Sapi, E., . . . Szekely, A. (2014). A Comparison of the Systems for the Identification of Postoperative Acute Kidney Injury in Pediatric Cardiac Patients. Annals of Thoracic Surgery, 97(1), 202-210.
Li, S., Krawczeski, C. D., Zappitelli, M., Devarajan, P., Thiessen-Philbrook, H., Coca, S. G., . . . Parikh, C. R. (2011). Incidence, Risk Factors, and Outcomes of Acute Kidney Injury after Pediatric Cardiac Surgery: A Prospective Multicenter Study. Critical Care Medicine, 39(6), 1493-1499.
164
Liano, F., Junco, E., Pascual, J., Madero, R., & Verde, E. (1998). The Spectrum of Acute Renal Failure in the Intensive Care Unit Compared with That Seen in Other Settings. The Madrid Acute Renal Failure Study Group. Kidney international. Supplement, 66, S16-24.
Liberati, A., Altman, D. G., Tetzlaff, J., Mulrow, C., Gotzsche, P. C., Ioannidis, J. P., . . . Moher, D. (2009). The Prisma Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration. PLoS Medicine, 6(7), e1000100.
Lodise, T. P., Patel, N., Lomaestro, B. M., Rodvold, K. A., & Drusano, G. L. (2009). Relationship between Initial Vancomycin Concentration-Time Profile and Nephrotoxicity among Hospitalized Patients. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America, 49(4), 507-514.
Lopes, J., x00E, Ant, x00F, nio, Fernandes, P., . . . Prata, M. M. (2008). Acute Kidney Injury in Intensive Care Unit Patients: A Comparison between the Rifle and the Acute Kidney Injury Network Classifications. Critical Care, 12(4), R110.
Luo, X., Doherty, J., Cappelleri, J. C., & Frush, K. (2007). Role of Pharmacoepidemiology in Evaluating Prescription Drug Safety in Pediatrics. Current Medical Research and Opinion, 23(11), 2607-2615.
Manrique, A., Jooste, E. H., Kuch, B. A., Lichtenstein, S. E., Morell, V., Munoz, R., . . . Davis, P. J. (2009). The Association of Renal Dysfunction and the Use of Aprotinin in Patients Undergoing Congenital Cardiac Surgery Requiring Cardiopulmonary Bypass. Anesthesia & Analgesia, 109(1), 45-52.
Marenzi, G., Lauri, G., Assanelli, E., Campodonico, J., De Metrio, M., Marana, I., . . . Bartorelli, A. L. (2004). Contrast-Induced Nephropathy in Patients Undergoing Primary Angioplasty for Acute Myocardial Infarction. Journal of the American College of Cardiology, 44(9), 1780-1785.
Martinez-Mir, I., Garcia-Lopez, M., Palop, V., Ferrer, J. M., Rubio, E., & Morales-Olivas, F. J. (1999). A Prospective Study of Adverse Drug Reactions in Hospitalized Children. British Journal of Clinical Pharmacology, 47(6), 681-688.
Matlow, A. G., Baker, G. R., Flintoft, V., Cochrane, D., Coffey, M., Cohen, E., . . . Nijssen-Jordan, C. (2012). Adverse Events among Children in Canadian Hospitals: The Canadian Paediatric Adverse Events Study. Canadian Medical Association Journal, 184(13), E709-718.
May, F. E., Stewart, R. B., & Cluff, L. E. (1977). Drug Interactions and Multiple Drug Administration. Clinical Pharmacology and Therapeutics, 22(3), 322-328.
McDonald, J. S., McDonald, R. J., Comin, J., Williamson, E. E., Katzberg, R. W., Murad, M. H., & Kallmes, D. F. (2013). Frequency of Acute Kidney Injury Following Intravenous Contrast Medium Administration: A Systematic Review and Meta-Analysis. Radiology, 267(1), 119-128.
165
McKamy, S., Hernandez, E., Jahng, M., Moriwaki, T., Deveikis, A., & Le, J. (2011). Incidence and Risk Factors Influencing the Development of Vancomycin Nephrotoxicity in Children. The Journal of pediatrics, 158(3), 422-426.
McMaul, M., & Van Hollen, C. (2011). Bpca and Prea Reauthorization. Retrieved February 1, 2014, from http://childhoodcancer-mccaul.house.gov/issue/bpca-and-prea-reauthorization
Mehta, P., Sinha, A., Sami, A., Hari, P., Kalaivani, M., Gulati, A., . . . Bagga, A. (2012). Incidence of Acute Kidney Injury in Hospitalized Children. Indian Pediatrics, 49(7), 537-542.
Mehta, R. L., & Chertow, G. M. (2003). Acute Renal Failure Definitions and Classification: Time for Change? Journal of the American Society of Nephrology : JASN, 14(8), 2178-2187.
Mehta, R. L., Kellum, J. A., Shah, S. V., Molitoris, B. A., Ronco, C., Warnock, D. G., & Levin, A. (2007). Acute Kidney Injury Network: Report of an Initiative to Improve Outcomes in Acute Kidney Injury. Critical Care, 11(2), R31.
Mehta, R. L., Pascual, M. T., Soroko, S., Savage, B. R., Himmelfarb, J., Ikizler, T. A., . . . Chertow, G. M. (2004). Spectrum of Acute Renal Failure in the Intensive Care Unit: The Picard Experience. Kidney International, 66(4), 1613-1621.
Miall, L. S., Henderson, M. J., Turner, A. J., Brownlee, K. G., Brocklebank, J. T., Newell, S. J., & Allgar, V. L. (1999). Plasma Creatinine Rises Dramatically in the First 48 Hours of Life in Preterm Infants. Pediatrics, 104(6), e76.
Mishra, J., Dent, C., Tarabishi, R., Mitsnefes, M. M., Ma, Q., Kelly, C., . . . Devarajan, P. (2005). Neutrophil Gelatinase-Associated Lipocalin (Ngal) as a Biomarker for Acute Renal Injury after Cardiac Surgery. Lancet, 365(9466), 1231-1238.
Moffett, B. S., & Goldstein, S. L. (2011). Acute Kidney Injury and Increasing Nephrotoxic-Medication Exposure in Noncritically-Ill Children. Clinical journal of the American Society of Nephrology : CJASN, 6(4), 856-863.
Moher, D., Schulz, K. F., & Altman, D. G. (2001). The Consort Statement: Revised Recommendations for Improving the Quality of Reports of Parallel-Group Randomised Trials. Lancet, 357(9263), 1191-1194.
Morgan, C. J., Zappitelli, M., Robertson, C. M., Alton, G. Y., Sauve, R. S., Joffe, A. R., . . . Rebeyka, I. M. (2013). Risk Factors for and Outcomes of Acute Kidney Injury in Neonates Undergoing Complex Cardiac Surgery. The Journal of pediatrics, 162(1), 120-127 e121.
Moyen, E., Camire, E., & Stelfox, H. T. (2008). Clinical Review: Medication Errors in Critical Care. Critical Care, 12(2), 208.
166
Nash, K., Hafeez, A., & Hou, S. (2002). Hospital-Acquired Renal Insufficiency. American journal of kidney diseases : the official journal of the National Kidney Foundation, 39(5), 930-936.
Naughton, C. A. (2008). Drug-Induced Nephrotoxicity. American Family Physician, 78(6), 743-750.
Niederhauser, V. P. (1997). Prescribing for Children: Issues in Pediatric Pharmacology. The Nurse practitioner, 22(3), 16-18, 23, 26-18 passim.
O'Brien, R. M. (2007). A Caution Regarding Rules of Thumb for Variance Inflation Factors. Quality & Quantity, 41(5), 673-690.
Ostermann, M., & Chang, R. (2007). Acute Kidney Injury in the Intensive Care Unit According to Rifle. Critical Care Medicine, 35(8), 1837-1843; quiz 1852.
Packer, M., Carver, J. R., Rodeheffer, R. J., Ivanhoe, R. J., DiBianco, R., Zeldis, S. M., . . . et al. (1991). Effect of Oral Milrinone on Mortality in Severe Chronic Heart Failure. The Promise Study Research Group. The New England journal of medicine, 325(21), 1468-1475.
Palmieri, T., Lavrentieva, A., & Greenhalgh, D. (2009). An Assessment of Acute Kidney Injury with Modified Rifle Criteria in Pediatric Patients with Severe Burns. Intensive Care Medicine, 35(12), 2125-2129.
Pandolfini, C., Campi, R., Clavenna, A., Cazzato, T., & Bonati, M. (2005). Italian Paediatricians and Off-Label Prescriptions: Loyal to Regulatory or Guideline Standards? Acta Paediatrica, 94(6), 753-757.
Pandolfini, C., Impicciatore, P., Provasi, D., Rocchi, F., Campi, R., & Bonati, M. (2002). Off-Label Use of Drugs in Italy: A Prospective, Observational and Multicentre Study. Acta Paediatrica, 91(3), 339-347.
Pannu, N., & Nadim, M. K. (2008). An Overview of Drug-Induced Acute Kidney Injury. Critical Care Medicine, 36(4 Suppl), S216-223.
Parikh, C. R., Devarajan, P., Zappitelli, M., Sint, K., Thiessen-Philbrook, H., Li, S., . . . Krawczeski, C. D. (2011). Postoperative Biomarkers Predict Acute Kidney Injury and Poor Outcomes after Pediatric Cardiac Surgery. Journal of the American Society of Nephrology : JASN, 22(9), 1737-1747.
Parikh, C. R., Edelstein, C. L., Devarajan, P., & Cantley, L. (2007). Biomarkers of Acute Kidney Injury: Early Diagnosis, Pathogenesis, and Recovery. Journal of investigative medicine : the official publication of the American Federation for Clinical Research, 55(7), 333-340.
Patz, A. (1957). The Role of Oxygen in Retrolental Fibroplasia. Pediatrics, 19(3), 504-524.
167
Patzer, L. (2008). Nephrotoxicity as a Cause of Acute Kidney Injury in Children. Pediatric Nephrology, 23(12), 2159-2173.
Perazella, M. A. (2005). Drug-Induced Nephropathy: An Update. Expert Opinion on Drug Safety, 4(4), 689-706.
Perazella, M. A. (2009). Renal Vulnerability to Drug Toxicity. Clinical Journal of The American Society of Nephrology: CJASN, 4(7), 1275-1283.
Perazella, M. A. (2012). Drug Use and Nephrotoxicity in the Intensive Care Unit. Kidney International, 81(12), 1172-1178.
Plotz, F. B., Bouma, A. B., van Wijk, J. A. E., Kneyber, M. C. J., & Kenkamp, A. (2008). Pediatric Acute Kidney Injury in the Icu: An Independent Evaluation of Prifle Criteria. Intensive Care Medicine, 34(9), 1713-1717.
Plotz, F. B., Hulst, H. E., Twisk, J. W., Bokenkamp, A., Markhorst, D. G., & van Wijk, J. A. (2005). Effect of Acute Renal Failure on Outcome in Children with Severe Septic Shock. Pediatric Nephrology, 20(8), 1177-1181.
Pollack, M. M., Ruttimann, U. E., & Getson, P. R. (1988). Pediatric Risk of Mortality (Prism) Score. Critical Care Medicine, 16(11), 1110-1116.
Pritchard, L., Baker, C., Leggett, J., Sehdev, P., Brown, A., & Bayley, K. B. (2010). Increasing Vancomycin Serum Trough Concentrations and Incidence of Nephrotoxicity. The American journal of medicine, 123(12), 1143-1149.
Prodhan, P., McCage, L. S., Stroud, M. H., Gossett, J., Garcia, X., Bhutta, A. T., . . . Blaszak, R. T. (2012). Acute Kidney Injury Is Associated with Increased in-Hospital Mortality in Mechanically Ventilated Children with Trauma. The journal of trauma and acute care surgery, 73(4), 832-837.
Pronovost, P., Weast, B., Schwarz, M., Wyskiel, R. M., Prow, D., Milanovich, S. N., . . . Lipsett, P. (2003). Medication Reconciliation: A Practical Tool to Reduce the Risk of Medication Errors. Journal of Critical Care, 18(4), 201-205.
Proulx, N. L., Akbari, A., Garg, A. X., Rostom, A., Jaffey, J., & Clark, H. D. (2005). Measured Creatinine Clearance from Timed Urine Collections Substantially Overestimates Glomerular Filtration Rate in Patients with Liver Cirrhosis: A Systematic Review and Individual Patient Meta-Analysis. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association, 20(8), 1617-1622.
Rea, R. S., & Capitano, B. (2007). Optimizing Use of Aminoglycosides in the Critically Ill. Seminars in Respiratory and Critical Care Medicine, 28(6), 596-603.
Regulations Requiring Manufacturers to Assess the Safety and Effectiveness of New Drugs and Biological Products in Pediatric Patients. (1997). Washington, D.C.: Department of Health and Human Services.
168
Renal Failure. (2006). In M. H. Beers, R. S. Porter, T. V. Jones, J. L. Kaplan & M. Berkwits (Eds.), The Merck Manual of Diagnosis and Therapy (18th ed., pp. 1926-2058). Whitehouse Station: Merck Research Laboratories.
Ricci, Z., Cruz, D., & Ronco, C. (2008). The Rifle Criteria and Mortality in Acute Kidney Injury: A Systematic Review. Kidney International, 73(5), 538-546.
Roberts, R., Rodriguez, W., Murphy, D., & Crescenzi, T. (2003). Pediatric Drug Labeling: Improving the Safety and Efficacy of Pediatric Therapies. Journal of the American Medical Association, 290(7), 905-911.
Rodrigues, F. B., Bruetto, R. G., Torres, U. S., Otaviano, A. P., Zanetta, D. M., & Burdmann, E. A. (2013). Incidence and Mortality of Acute Kidney Injury after Myocardial Infarction: A Comparison between Kdigo and Rifle Criteria. PloS One, 8(7), e69998.
Rosner, M. H., & Okusa, M. D. (2006). Acute Kidney Injury Associated with Cardiac Surgery. Clinical Journal of The American Society of Nephrology: CJASN, 1(1), 19-32.
Rothschild, J. M., Landrigan, C. P., Cronin, J. W., Kaushal, R., Lockley, S. W., Burdick, E., . . . Bates, D. W. (2005). The Critical Care Safety Study: The Incidence and Nature of Adverse Events and Serious Medical Errors in Intensive Care. Critical Care Medicine, 33(8), 1694-1700.
Rougier, F., Ducher, M., Maurin, M., Corvaisier, S., Claude, D., Jelliffe, R., & Maire, P. (2003). Aminoglycoside Dosages and Nephrotoxicity: Quantitative Relationships. Clinical Pharmacokinetics, 42(5), 493-500.
Roy, A. K., Mc Gorrian, C., Treacy, C., Kavanaugh, E., Brennan, A., Mahon, N. G., & Murray, P. T. (2013). A Comparison of Traditional and Novel Definitions (Rifle, Akin, and Kdigo) of Acute Kidney Injury for the Prediction of Outcomes in Acute Decompensated Heart Failure. Cardiorenal Medicine, 3(1), 26-37.
Safer, D. J., Zito, J. M., & Gardner, J. E. (2001). Pemoline Hepatotoxicity and Postmarketing Surveillance. Journal of the American Academy of Child and Adolescent Psychiatry, 40(6), 622-629.
Schetz, M., Dasta, J., Goldstein, S., & Golper, T. (2005). Drug-Induced Acute Kidney Injury. Current Opinion in Critical Care, 11(6), 555-565.
Schneider, J., Khemani, R., Grushkin, C., & Bart, R. (2010). Serum Creatinine as Stratified in the Rifle Score for Acute Kidney Injury Is Associated with Mortality and Length of Stay for Children in the Pediatric Intensive Care Unit. Critical Care Medicine, 38(3), 933-939.
Schwartz, G. J., Brion, L. P., & Spitzer, A. (1987). The Use of Plasma Creatinine Concentration for Estimating Glomerular Filtration Rate in Infants, Children, and Adolescents. Pediatric Clinics of North America, 34(3), 571-590.
169
Schwartz, G. J., Haycock, G. B., Edelmann, C. M., Jr., & Spitzer, A. (1976). A Simple Estimate of Glomerular Filtration Rate in Children Derived from Body Length and Plasma Creatinine. Pediatrics, 58(2), 259-263.
Schwartz, G. J., Haycock, G. B., & Spitzer, A. (1976). Plasma Creatinine and Urea Concentration in Children: Normal Values for Age and Sex. The Journal of pediatrics, 88(5), 828-830.
Seeliger, E., Sendeski, M., Rihal, C. S., & Persson, P. B. (2012). Contrast-Induced Kidney Injury: Mechanisms, Risk Factors, and Prevention. European Heart Journal, 33(16), 2007-2015.
Siew, E. D., Matheny, M. E., Ikizler, T. A., Lewis, J. B., Miller, R. A., Waitman, L. R., . . . Peterson, J. F. (2010). Commonly Used Surrogates for Baseline Renal Function Affect the Classification and Prognosis of Acute Kidney Injury. Kidney International, 77(6), 536-542.
Silva, D. C., Araujo, O. R., Arduini, R. G., Alonso, C. F., Shibata, A. R., & Troster, E. J. (2013). Adverse Drug Events in a Paediatric Intensive Care Unit: A Prospective Cohort. BMJ Open, 3(2).
Singbartl, K., Bishop, J. V., Wen, X., Murugan, R., Chandra, S., Filippi, M. D., & Kellum, J. A. (2011). Differential Effects of Kidney-Lung Cross-Talk During Acute Kidney Injury and Bacterial Pneumonia. Kidney International, 80(6), 633-644.
Singh, N. P., Ganguli, A., & Prakash, A. (2003). Drug-Induced Kidney Diseases. The Journal of the Association of Physicians of India, 51, 970-979.
Slater, A., Shann, F., & Pearson, G. (2003). Pim2: A Revised Version of the Paediatric Index of Mortality. Intensive Care Medicine, 29(2), 278-285.
Slater, M. B., Anand, V., Uleryk, E. M., & Parshuram, C. S. (2012). A Systematic Review of Rifle Criteria in Children, and Its Application and Association with Measures of Mortality and Morbidity. Kidney International, 81(8), 791-798.
Smyth, R. M., Gargon, E., Kirkham, J., Cresswell, L., Golder, S., Smyth, R., & Williamson, P. (2012). Adverse Drug Reactions in Children--a Systematic Review. PloS One, 7(3), e24061.
Solomon, R. (1998). Contrast-Medium-Induced Acute Renal Failure. Kidney International, 53(1), 230-242.
Solomon, R., & Deray, G. (2006). How to Prevent Contrast-Induced Nephropathy and Manage Risk Patients: Practical Recommendations. Kidney international. Supplement(100), S51-53.
Srisawat, N., Hoste, E. E. A., & Kellum, J. A. (2010). Modern Classification of Acute Kidney Injury. Blood Purification, 29(3), 300-307.
170
Stacul, F., van der Molen, A. J., Reimer, P., Webb, J. A., Thomsen, H. S., Morcos, S. K., . . . Heinz-Peer, G. (2011). Contrast Induced Nephropathy: Updated Esur Contrast Media Safety Committee Guidelines. European Radiology, 21(12), 2527-2541.
Stevens, L. A., & Levey, A. S. (2009). Measured Gfr as a Confirmatory Test for Estimated Gfr. Journal of the American Society of Nephrology, 20(11), 2305-2313.
Strom, B. L. (2012). What Is Pharmacoepidemiology? In B. L. Strom, S. E. Kimmel & S. Hennessy (Eds.), Pharmacoepidemiology (5th ed.). Oxford: Wiley-Blackwell.
Sural, S., Sharma, R. K., Singhal, M., Sharma, A. P., Kher, V., Arora, P., . . . Gulati, S. (2000). Etiology, Prognosis, and Outcome of Post-Operative Acute Renal Failure. Renal Failure, 22(1), 87-97.
Sutherland, S. M., Zappitelli, M., Alexander, S. R., Chua, A. N., Brophy, P. D., Bunchman, T. E., . . . Goldstein, S. L. (2010). Fluid Overload and Mortality in Children Receiving Continuous Renal Replacement Therapy: The Prospective Pediatric Continuous Renal Replacement Therapy Registry. American journal of kidney diseases : the official journal of the National Kidney Foundation, 55(2), 316-325.
Symons, J. M., Chua, A. N., Somers, M. J., Baum, M. A., Bunchman, T. E., Benfield, M. R., . . . Goldstein, S. L. (2007). Demographic Characteristics of Pediatric Continuous Renal Replacement Therapy: A Report of the Prospective Pediatric Continuous Renal Replacement Therapy Registry. Clinical journal of the American Society of Nephrology : CJASN, 2(4), 732-738.
Szklo, M. N., J. (2006). Epidemiology: Beyond the Basics (2nd ed.). Burlington: Jones and Bartlett Learning.
t Jong, G. W., Vulto, A. G., de Hoog, M., Schimmel, K. J., Tibboel, D., & van den Anker, J. N. (2001). A Survey of the Use of Off-Label and Unlicensed Drugs in a Dutch Children's Hospital. Pediatrics, 108(5), 1089-1093.
Taber, S. S., & Mueller, B. A. (2006). Drug-Associated Renal Dysfunction. Critical Care Clinics, 22(2), 357-374, viii.
Taketomo, C. K., Hodding, J. H., & Kraus, D. M. (2012). Pediatric and Neonatal Dosage Handbook (19th ed.). Hudson: Lexi-Comp.
Tallgren, M., Niemi, T., Poyhia, R., Raininko, E., Railo, M., Salmenpera, M., . . . Hynninen, M. (2007). Acute Renal Injury and Dysfunction Following Elective Abdominal Aortic Surgery. European Journal of Vascular & Endovascular Surgery, 33(5), 550-555.
Thaul, S. (2012). Fda's Authority to Ensure That Drugs Prescribed to Children Are Safe and Effective. Washington, DC: Congressional Research Service Report for Congress.
Totapally, B. R., Machado, J., Lee, H., Paredes, A., & Raszynski, A. (2013). Acute Kidney Injury During Vancomycin Therapy in Critically Ill Children. Pharmacotherapy, 33(6), 598-602.
171
Toth, R., Breuer, T., Cserep, Z., Lex, D., Fazekas, L., Sapi, E., . . . Szekely, A. (2012). Acute Kidney Injury Is Associated with Higher Morbidity and Resource Utilization in Pediatric Patients Undergoing Heart Surgery. The Annals of thoracic surgery, 93(6), 1984-1990.
Toto, R. D. (1995). Conventional Measurement of Renal Function Utilizing Serum Creatinine, Creatinine Clearance, Inulin and Para-Aminohippuric Acid Clearance. Current Opinion in Nephrology and Hypertension, 4(6), 505-509; discussion 503-504.
Traynor, J., Mactier, R., Geddes, C. C., & Fox, J. G. (2006). How to Measure Renal Function in Clinical Practice. British Medical Journal, 333(7571), 733-737.
Uchino, S., Bellomo, R., Goldsmith, D., Bates, S., & Ronco, C. (2006). An Assessment of the Rifle Criteria for Acute Renal Failure in Hospitalized Patients. Critical Care Medicine, 34(7), 1913-1917.
Uchino, S., Kellum, J. A., Bellomo, R., Doig, G. S., Morimatsu, H., Morgera, S., . . . Ronco, C. (2005). Acute Renal Failure in Critically Ill Patients: A Multinational, Multicenter Study. Journal of the American Medical Association, 294(7), 813-818.
Ulrich, C. M., Bigler, J., & Potter, J. D. (2006). Non-Steroidal Anti-Inflammatory Drugs for Cancer Prevention: Promise, Perils and Pharmacogenetics. Nature Reviews Cancer, 6(2), 130-140.
United States Government Accountability Office Report to Congressional Committees. (2007). Pediatric Drug Research: Studies Conducted under Best Pharmaceuticals for Children Act (Vol. GAO-07-557). Washington, D.C.: U.S. Government Accountability Office.
Uppal, N. K., Dupuis, L. L., & Parshuram, C. S. (2008). Documentation of Pediatric Drug Safety in Manufacturers' Product Monographs: A Cross-Sectional Evaluation of the Canadian Compendium of Pharmaceuticals and Specialities. Paediatric Drugs, 10(3), 193-197.
Urinary System. (2002). In J. L. Stockslager, R. Cheli & K. Haworth (Eds.), Lippincott Professional Guides: Anatomy & Physiology (2nd ed.). Philadelphia: Lippincott Williams & Wilkins.
Vachvanichsanong, P., Dissaneewate, P., Lim, A., & McNeil, E. (2006). Childhood Acute Renal Failure: 22-Year Experience in a University Hospital in Southern Thailand. Pediatrics, 118(3), e786-791.
Valentin, A., Capuzzo, M., Guidet, B., Moreno, R. P., Dolanski, L., Bauer, P., & Metnitz, P. G. (2006). Patient Safety in Intensive Care: Results from the Multinational Sentinel Events Evaluation (See) Study. Intensive Care Medicine, 32(10), 1591-1598.
172
Viswanathan, S., Manyam, B., Azhibekov, T., & Mhanna, M. J. (2012). Risk Factors Associated with Acute Kidney Injury in Extremely Low Birth Weight (Elbw) Infants. Pediatric Nephrology, 27(2), 303-311.
Washburn, K. K., Zappitelli, M., Arikan, A. A., Loftis, L., Yalavarthy, R., Parikh, C. R., . . . Goldstein, S. L. (2008). Urinary Interleukin-18 Is an Acute Kidney Injury Biomarker in Critically Ill Children. Nephrology Dialysis Transplantation, 23(2), 566-572.
Waybill, M. M., & Waybill, P. N. (2001). Contrast Media-Induced Nephrotoxicity: Identification of Patients at Risk and Algorithms for Prevention. Journal of vascular and interventional radiology : JVIR, 12(1), 3-9.
Wheeler, D. S., Devarajan, P., Ma, Q., Harmon, K., Monaco, M., Cvijanovich, N., & Wong, H. R. (2008). Serum Neutrophil Gelatinase-Associated Lipocalin (Ngal) as a Marker of Acute Kidney Injury in Critically Ill Children with Septic Shock. Critical Care Medicine, 36(4), 1297-1303.
White, T. J., Arakelian, A., & Rho, J. P. (1999). Counting the Costs of Drug-Related Adverse Events. Pharmacoeconomics, 15(5), 445-458.
Willy, M. E., Manda, B., Shatin, D., Drinkard, C. R., & Graham, D. J. (2002). A Study of Compliance with Fda Recommendations for Pemoline (Cylert). Journal of the American Academy of Child and Adolescent Psychiatry, 41(7), 785-790.
Wohlauer, M. V., Sauaia, A., Moore, E. E., Burlew, C. C., Banerjee, A., & Johnson, J. (2012). Acute Kidney Injury and Posttrauma Multiple Organ Failure: The Canary in the Coal Mine. The journal of trauma and acute care surgery, 72(2), 373-378; discussion 379-380.
World Health Organization. (2010). International Statistical Classification of Diseases and Related Health Problems 10th Revision. Retrieved February 12, 2013, 2013, from apps.who.int/classifications/icd10/browse/2010/en
Wyatt, C. M., Arons, R. R., Klotman, P. E., & Klotman, M. E. (2006). Acute Renal Failure in Hospitalized Patients with Hiv: Risk Factors and Impact on in-Hospital Mortality. AIDS, 20(4), 561-565.
Zappitelli, M., Bernier, P.-L., Saczkowski, R. S., Tchervenkov, C. I., Gottesman, R., Dancea, A., . . . Alkandari, O. (2009). A Small Post-Operative Rise in Serum Creatinine Predicts Acute Kidney Injury in Children Undergoing Cardiac Surgery. Kidney International, 76(8), 885-892.
Zappitelli, M., & Goldstein, S. L. (2009). Acute Kidney Injury: General Aspects. In S. G. Kiessling, J. Goebel & M. J. Somers (Eds.), Pediatric Nephrology in the Icu. Berlin: Springer.
Zappitelli, M., Moffett, B. S., Hyder, A., & Goldstein, S. L. (2011). Acute Kidney Injury in Non-Critically Ill Children Treated with Aminoglycoside Antibiotics in a Tertiary
173
Healthcare Centre: A Retrospective Cohort Study. Nephrology Dialysis Transplantation, 26 (1), 144-150.
Zappitelli, M., Parikh, C. R., Akcan-Arikan, A., Washburn, K. K., Moffett, B. S., & Goldstein, S. L. (2008). Ascertainment and Epidemiology of Acute Kidney Injury Varies with Definition Interpretation. Clinical Journal of The American Society of Nephrology: CJASN, 3(4), 948-954.
Zappitelli, M., Washburn, K. K., Arikan, A. A., Loftis, L., Ma, Q., Devarajan, P., . . . Goldstein, S. L. (2007). Urine Neutrophil Gelatinase-Associated Lipocalin Is an Early Marker of Acute Kidney Injury in Critically Ill Children: A Prospective Cohort Study. Critical Care, 11(4), R84.
Zavada, J., Hoste, E., Cartin-Ceba, R., Calzavacca, P., Gajic, O., Clermont, G., . . . Kellum, J. A. (2010). A Comparison of Three Methods to Estimate Baseline Creatinine for Rifle Classification. Nephrology Dialysis Transplantation, 25(12), 3911-3918.
Zito, J. M., Derivan, A. T., Kratochvil, C. J., Safer, D. J., Fegert, J. M., & Greenhill, L. L. (2008). Off-Label Psychopharmacologic Prescribing for Children: History Supports Close Clinical Monitoring. Child and Adolescent Psychiatry and Mental Health, 2(1), 24.