mardiana_rct2011
TRANSCRIPT
-
8/3/2019 Mardiana_RCT2011
1/9
Permissive underfeeding and intensive insulin therapy in critically ill
patients: a randomized controlled trial13
Yaseen M Arabi, Hani M Tamim, Gousia S Dhar, Abdulaziz Al-Dawood, Muhammad Al-Sultan, Maram H Sakkijha,Salim H Kahoul, and Riette Brits
ABSTRACT
Background: Nutritional support has been recognized as an essen-
tial part of intensive care unit management. However, the appropri-
ate caloric intake for critically ill patients remains ill defined.
Objective: We examined the effect of permissive underfeeding
compared with that of target feeding and of intensive insulin therapy
(IIT) compared with that of conventional insulin therapy (CIT) on
the outcomes of critically ill patients.Design: This study had a 2 2 factorial, randomized, controlled
design. Eligible patients were randomly assigned to permissive un-
derfeeding or target feeding groups (caloric goal: 6070% com-
pared with 90100% of calculated requirement, respectively) with
either IIT or CIT (target blood glucose: 4.46.1 compared with
1011.1 mmol/L, respectively).
Results: Twenty-eight-day all-cause mortality was 18.3% in the per-
missive underfeeding group compared with 23.3% in the target feed-
ing group (relative risk: 0.79; 95% CI: 0.48, 1.29; P = 0.34). Hospital
mortality was lower in the permissive underfeeding group than in the
target group (30.0% compared with 42.5%; relative risk: 0.71; 95%
CI: 0.50, 0.99; P = 0.04). No significant differences in outcomes were
observed between the IIT and CIT groups.
Conclusion: In critically ill patients, permissive underfeeding may be
associated with lower mortality rates than target feeding. This trial
was registered at controlled-trials.com as ISRCTN96294863. Am J
Clin Nutr 2011;93:56977.
INTRODUCTION
Nutritional support has been recognized as an essential part of
intensive care unit (ICU) management (1). However, the ap-
propriate caloric dose for critically ill patients remains ill defined.
The perceived benefit of achieving the caloric target is to at-
tenuate malnutritiona common complication during critical
illness (2) that is associated with increased morbidity and
mortality (2, 3). In fact, several studies have shown worse out-
comes in patients receiving a low caloric intake.
On the other hand, some evidence supports caloric restriction.
Studies have shown that caloric restriction prolongs the life span
in several species (4, 5), promotes mammalian cell survival (6),
and improves longevity biomarkers in humans (7). These effects
have been attributed to several mechanisms, including a re-
duction in the metabolic rate and oxidative stress (8), a reduction
in mitochondrial free radical generation (9), an up-regulation of
the plasma membrane redox system (10), an improvement in
insulin sensitivity, modification of cardiovascular disease risk
(11), an improvement in myocardial ischemic tolerance (12), and
changes in neuroendocrine and sympathetic nervous system
function (5). Although the applicability of these findings to
critically ill patients is unknown, physiologically stressed criti-
cally ill patients are likely to be in a hypercatabolic state (13) and
to have augmented oxidative stress (14), insulin resistance (15),
and altered neuroendocrine and sympathetic nervous system
function (5). In fact, some clinical studies have shown thata lower caloric intake in critically ill patients is associated with
better outcomes (1619).
Because of this controversy, it remains unclear what con-
stitutes an appropriate caloric dose for critically ill patients (20,
21). Although clinical practice guidelines recommended initi-
ating nutritional support early in the course of critical illness (20,
22), the evidence for achieving the caloric target was insufficient
to make recommendation (20).
The purpose of our study was to examine the effect of per-
missive underfeeding compared with that of a targeted caloric
intake and of intensive insulin therapy compared with conven-
tional insulin therapy on the outcomes of critically ill patients.
SUBJECTS AND METHODS
Setting
The study was conducted in the 21-bed medical-surgical ICU
of a tertiary care academic hospital accredited by Joint Com-
mission International. The ICU is run as a closed unit by 24 h in-
house critical care board-certified intensivists, as described
elsewhere (23), with an approximate nurse-to-patient ratio of 1 to
1.2.
1From the Departments of Intensive Care Medicine (YMA, GSD, AA-D,
MA-S, MHS, SHK, and RB) and Epidemiology and Biostatistics (HMT),
King Abdulaziz Medical City and King Saud Bin Abdulaziz University for
Health Sciences, Riyadh, Saudi Arabia.2
Supported by the King Abdulaziz City for Science and Technology
(LG 10-30).3 Address correspondence to YM Arabi, Intensive Care Department, Col-
lege of Medicine, King Saud Bin Abdulaziz University, King Abdulaziz
Medical City, PO Box 22490, Intensive Care Department, MC 1425, Riyadh,
11426, Saudi Arabia. E-mail: [email protected] or [email protected].
Received June 7, 2010. Accepted for publication December 9, 2010.
First published online January 26, 2011; doi: 10.3945/ajcn.110.005074.
Am J Clin Nutr2011;93:56977. Printed in USA. 2011 American Society for Nutrition 569
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
2/9
Patients
Patients aged !18 y with a blood glucose concentration of
.6.1 mmol/L (110 mg/dL), receiving enteral feeding, and ex-
pected to stay for !48 h were eligible for the study. Exclusion
criteria included type 1 diabetes, diabetic ketoacidosis, hypo-
glycemia, brain death, do-not-resuscitate status, terminal illness,
postcardiac arrest, seizures within the past 6 mo, pregnancy,
liver transplant, burn, readmission to the ICU within the samehospitalization, enrollment in a competing trial, oral feeding,
and total parenteral nutrition. Informed consent was obtained
from the patient or his or her next of kin within 48 h of the
randomization window.
Study design
The study was a 2 2 factorial-design, randomized controlled
trial (RCT). On the basis of computer-generated random per-
muted blocks, the enrolled patients were randomly assigned by
using concealed envelops to 1 of the 4 study groups: permissive
underfeeding with intensive insulin therapy (IIT), permissive
underfeeding with conventional insulin therapy (CIT), targetfeeding with IIT, or target feeding with CIT. Stratified ran-
domization was performed for diabetic and nondiabetic patients.
Similar to other IIT trials, the assignment to IIT or CIT was not
blinded because of the need for dose titration. The feeding
strategy was also not blinded because of the need for titration
according to tolerance and gastric residuals. The study was
approved by the Institutional Review Board and was conducted
between April 2006 and January 2008.
Interventions
Permissive underfeeding compared with target feeding
The standard caloric requirement was estimated by the di-
etitian using the Harris-Benedict equations and adjusting for
stress factors (24). The goal of caloric intake in the target feeding
group was 90100% of the standard caloric requirement, and in
the permissive underfeeding group it was 6070% of the standard
caloric requirement. The selection of formula was left to dis-
cretion of the attending physician as long as it satisfied the total
caloric intake criteria and was not enriched with immunonu-
trients. Enteral feeding was administered with an enteral feeding
protocol published elsewhere from our ICU (25). The dietitianassessed the caloric intake for the previous day and compensated
FIGURE 1. Patients in the study. All randomly assigned patients were included in the analysis as per intention-to-treat principle. DNR, do not resuscitate;NPO, nothing by mouth; IIT, intensive insulin therapy; CIT, conventional insulin therapy.
570 ARABI ET AL
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
3/9
for variation from the target on a daily basis by increasing or
decreasing the intake accordingly. Calculation of caloric intake
took into account intravenous dextrose and propofol infusions.
The protein requirement was calculated as 0.81.5 g/kg on the
basis of patient condition and underlying diseases (24). To avoid
protein malnutrition in the permissive underfeeding group,
additional protein (Resource Beneprotein; Nestle Healthcare
Nutrition Inc, Minneapolis, MN) was added to maintain the
full protein requirement without affecting the assigned caloric
intake.
Intensive compared with conventional insulin therapy
We used the previously published IIT and CIT protocols fromour center (26). Insulin infusion was adjusted to maintain a target
blood glucose concentration of 4.46.1 mmol/L (80110mg/dL)
in the IIT group and of 1011.1mmol/L (180200 mg/dL) in
the CIT group. The frequency of blood glucose monitoring in-
creased to every 20 min when blood glucose concentrations
decreased to 3.2 mmol/L (58 mg/dL) and reduced to every 24 h
when measurements were stable. The protocols included sev-
eral safeguards to reduce the incidence of hypoglycemia, which
included reducing or holding insulin infusion and/or adding
intravenous dextrose when glucose concentrations dropped
abruptly and during discontinuation or intolerance of feeding
(26).
The patients were followed until discharge from the ICU, except
if the patient tolerated oral feeding, had a do-not-resuscitate order
written (after enrollment), or became brain dead (after enroll-
ment). In the latter situations, the intervention was stopped butthe
outcome data were collected.
Data collection
Data were collected by the study coordinator using pre-
established definitions. At baseline, the following data were
recorded: patient demographic characteristics, Acute Physiologic
and Chronic Health Evaluation (APACHE) II scores (27), Se-
quential Organ Failure Assessment (SOFA) scores (28), admission
category (postoperative compared with nonoperative), diabeteshistory, inclusion blood glucose, ICU admission diagnosis and
presence of chronic illnesses on the basis of APACHE II defi-
nitions (27), vasopressor therapy (defined as use of any vaso-
pressor infusion except dopamine ,5 lg kg21 min21),
mechanical ventilation, serum creatinine, platelet count, biliru-
bin, international normalization ratio, partial pressure of oxygen
to fraction of inspired oxygen ratio (PaO2:FiO2), and Glasgow
Coma Scale. We calculated daily mean blood glucose concen-
trations and total daily insulin doses. We also documented the
daily total caloric intake, including what was received from the
enteral feeding formula and from propofol and dextrose. Daily
protein intake was recorded.
TABLE 1
Baseline characteristics of patients in the permissive underfeeding, target feeding, intensive insulin therapy, and conventional insulin therapy groups 1
Variable
Caloric intake Insulin therapy
Permissive underfeeding
(n = 120)
Target feeding
(n = 120) P value2
Intensive
(n = 120)
Conventional
(n = 120) P value2
Age (y) 50.3 6 21.33
51.9 6 22.1 0.56 53.0 6 21.3 49.3 6 22.0 0.19
Female sex [n (%)] 34 (28.3) 42 (35) 0.27 40 (33.3) 36 (30) 0.58
Height (cm) 164.0 6 10.0 164.0 6 12.0 0.53 164.0 6 12.0 165.0 6 11.0 0.47
Weight (kg) 77.0 6 17.8 76.3 6 21.1 0.80 75.8 6 18.9 77.5 6 20.0 0.52
BMI (kg/m2) 28.5 6 7.4 28.5 6 8.4 0.96 28.4 6 7.7 28.6 6 8.1 0.85
Diabetes [n (%)] 47 (39.2) 48 (40) 0.90 48 (40) 47 (39.2) 0.90
Inclusion blood glucose (mmol/L)4
12.2 6 4.7 11.6 6 4.2 0.26 12.3 6 4.7 11.5 6 4.2 0.15
Admission category [n (%)]
Nonoperative 95 (79.2) 103 (85.8) 0.17 99 (82.5) 99 (82.5) 1.0
Postoperative 25 (20.8) 17 (14.2) 21 (17.5) 21 (17.5)
Traumatic brain injury [n (%)] 35 (29.2) 31 (25.8) 0.56 34 (28.3) 32 (26.7) 0.77
APACHE II score 25.2 6 7.5 25.3 6 8.2 0.89 25.3 6 7.5 25.2 6 8.2 0.99
SOFA score, day 1 10.2 6 3.3 10.3 6 3.3 0.78 9.9 6 3.0 10.6 6 3.5 0.10
Mechanical ventilation [n (%)] 119 (99.2) 119 (99.2) 1.0 120 (100) 118 (98.3) 0.16
Vasopressor [n (%)] 77 (64.2) 78 (65) 0.89 75 (62.5) 80 (66.7) 0.50
Sepsis [n (%)] 35 (29.2) 37 (30.8) 0.78 33 (27.5) 39 (32.5) 0.40
Creatinine (lmol/L)
4
160.56
169.0 169.26
167.9 0.69 161.06
162.7 168.86
174.1 0.72Bilirubin (lmol/L)
443.8 6 82.1 61.3 6 128.5 0.38 40.4 6 76.5 64.7 6 131.7 0.22
Platelets (109 /L) 205.06 131.0 219.0 6 141.0 0.42 217.0 6 142.0 207.0 6 130.4 0.59
INR 1.4 6 0.7 1.5 6 0.7 0.28 1.4 6 0.7 1.5 6 0.7 0.68
PaO2:FiO2 201.8 6 105.5 207.8 6 97.3 0.65 209.1 6 97.5 200.5 6 105.3 0.52
GCS 7.2 6 3.5 7.4 6 3.5 0.66 7.1 6 3.3 7.4 6 3.6 0.44
Time to randomization (h) 18.0 6 13.5 19.9 6 13.8 0.27 19.3 6 14.1 18.6 6 13.2 0.68
1 APACHE II, Acute Physiology and Chronic Health Evaluation II; SOFA, Sequential Organ Failure Assessment; INR, international normalized ratio;
PaO2:FIO2, the ratio of partial pressure of oxygen to the fraction of inspired oxygen; GCS, Glasgow Coma Scale.2
A t test was used for continuous variables, and a chi-square test was used for categorical variables. No significant interaction was found between the 2
interventions.3 Mean 6 SD (all such values).4 To convert to conventional units in mg/dL, divide by 0.0555 for glucose, 88.4 for creatinine, and 17.1 for bilirubin.
PERMISSIVE UNDERFEEDING IN CRITICALLY ILL PATIENTS 571
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
4/9
FIGURE 2. Daily caloric intake, protein intake, average glucose concentrations, and insulin doses from study days 17 expressed as box plots. P values forthe comparison of the 2 groups on each day are shown (by t test). AD: Permissive underfeeding group (n = 120) compared with the target feeding group (n =120). Compared with the target feeding group, patients in the permissive underfeeding group consumed fewer calories but had similar protein intakes, glucoseconcentrations, and insulin doses. EH: Intensive insulin therapy group (IIT; n = 120) group compared with the conventional insulin therapy group (CIT; n =120). Compared with the CIT group, patients in the IIT group had higher insulin doses and lower glucose concentrations but had similar calorie and proteinintakes.
572 ARABI ET AL
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
5/9
Outcomes
The primary endpoint was 28-d all-cause mortality. Sec-
ondary endpoints included ICU, hospital, and 180-d mortality;
ICU and hospital length of stay (LOS); and mechanical ven-
tilation duration (MVD). We documented the occurrence of
health careassociated infections up to 48 h after ICU discharge,
which included bacteremia, catheter-related bloodstream in-
fection, urinary tract infection, ventilator-associated pneumo-
nia, and skin and soft tissue infections using the National
Nosocomial Infection Surveillance (NNIS) System (29). Wealso documented the need for renal replacement therapy and
packed red blood cell transfusion. We monitored the occurrence
of hypoglycemia (defined as a blood glucose concentration
2.2mmol/L or 40 mg/dL) and hypokalemia (defined as a po-
tassium concentration ,2.8 mmol/L).
Sample size
No RCT study has examined the effect of permissive un-
derfeeding on mortality. On the basis of a recently published
cohort study from our center (30), we observed a relative dif-
ference of 50% in ICU mortality between patients receiving
.90% of caloric requirements and those receiving 6070% of
caloric requirements (28% compared with 14%) (30). On the
basis of an estimated 28-d mortality rate of 25%, a power of 0.8,
and an a of 0.05, the number of subjects needed to show a re-
duction in mortality was 120 in each group.
Statistical analysis
The analysis was designed on an intention-to-treat principle.
No stopping rules or interim analysis were planned. Statistical
analyses were performed with the Statistical Analysis Software
(SAS; release 8, 1999; SAS Institute Inc, Cary, NC). Baseline
characteristics and outcome variables were compared by using a
ttest for continuous variables and a chi-square test for categorical
variables. For mortality data, we calculated the relative risk
(RR) and 95% CIs. Kaplan-Meier survival curves were con-
structed and compared by using a log-rank test. For outcomes
presented as rates, such as hypoglycemia, we used Z approxi-
mation. Statistical significance was defined as a P value 0.05.
Given the 2 2 factorial design, we tested for interactions
between the 2 interventions by multivariate logistic regression
modeling. The P value for this interaction was 0.067 for 28-d
TABLE 2
Energy and protein intakes and insulin and glucose data in the permissive underfeeding, target feeding, intensive insulin therapy, and conventional insulin
therapy groups
Variable
Caloric intake Insulin therapy
Permissive underfeeding
(n = 120)
Target feeding
(n = 120) P value1
Intensive
(n = 120)
Conventional
(n = 120) P value1
Calculated energy requirement (kcal/d) 1833.0 6 335.82
1767.6 6 311.3 0.12 1769.6 6 296.5 1830.9 6 349.3 0.14
Study energy intake target (kcal/d) 1336.7 6 282.2 1767.6 6 311.3 ,0.0001 1508.1 6 341.6 1571.9 6 401.4 0.19
Achieved energy intake (kcal/d) 1066.6 6 306.1 1251.7 6 432.5 0.0002 1151.4 6 344.0 1166.9 6 423.7 0.76
Percentage energy intake/requirement (kcal) 59.0 6 16.1 71.4 6 22.8 ,0.0001 65.9 6 19.5 64.5 6 21.8 0.61
Calculated protei n requirement (g/d) 74.1 6 17.8 69.1 6 14.5 0.02 71.0 6 16.2 72.2 6 16.6 0.56
Achieved protein intake (g/d) 47.5 6 21.2 43.6 6 18.9 0.14 45.4 6 19.1 45.7 6 21.2 0.91
Percentage protein intake/requirement (g) 65.2 6 25.7 63.7 6 25.0 0.63 64.8 6 23.7 64.1 6 26.8 0.84
Average energy intake, enteral (kcal) 915.96 346.6 1102.8 6 451.0 0.0004 996.6 6 377.1 1022.1 6 445.7 0.63
Average propofol intake (kcal) 33.9 6 65.7 34.8 6 80.9 0.93 31.8 6 66.5 36.9 6 80.2 0.59
Average dextrose intake (kcal) 117.1 6 105.6 114.1 6 101.4 0.82 123.2 6 108.2 107.96 98.0 0.25
Received insulin [n (%)] 92 (76.7) 97 (80.8) 0.43 119 (99.2) 70 (58.3) ,0.0001
Average insulin daily dose (units) 43.1 6 41.6 42.8 6 42.2 0.96 62.8 6 39.9 23.0 6 33.4 ,0.0001
Average glucose concentration (mmol/L)3
7.4 6 1.9 7.5 6 2.0 0.67 6.2 6 0.7 8.6 6 2.0 ,0.0001
1A t test was used for continuous variables, and a chi-square test was used for categorical variables. No significant interaction was found between the
2 interventions.
2 Mean 6 SD (all such values).3 To convert to conventional units in mg/dL, divide by 0.0555.
TABLE 3
Mortality rate in the permissive underfeeding, target feeding, intensive insulin therapy, and conventional insulin therapy groups 1
Caloric intake Insulin therapy
P value2
Permissive underfeeding
(n = 120)
Target feeding
(n = 120) RR (95% CI) P value2
Intensive
(n = 120)
Conventional
(n = 120) RR (95% CI)
28-d Mortality [n (%)] 22/120 (18.3) 28/120 (23.3) 0.79 (0.48, 1.29) 0.34 23/120 (19.2) 27/120 (22.5) 0.85 (0.52, 1.40) 0.52
180-d Mortality [n (%)] 38/116 (32.8) 52/117 (44.4) 0.74 (0.53, 1.03) 0.07 45/118 (38.1) 45/115 (39.1) 0.97 (0.71, 1.35) 0.88
ICU mortality [n (%)] 21/120 (17.5) 26/120 (21.7) 0.81 (0.48,1.35) 0.42 21/120 (17.5) 26/120 (21.7) 0.81 (0.48, 1.35) 0.42
Hospital mortality [n (%)] 36/120 (30) 51/120 (42.5) 0.71 (0.50, 0.99) 0.04 42/120 (35) 45/120 (37.5) 0.93 (0.67, 1.31) 0.69
1 ICU, intensive care unit; RR, relative risk.2
A chi-square test was used for categorical variables. No significant interaction was found between the 2 interventions.
PERMISSIVE UNDERFEEDING IN CRITICALLY ILL PATIENTS 573
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
6/9
mortality, 0.45 for ICU mortality, 0.29 for hospital mortality, and
0.26 for 180-d mortality, which indicated no statistically sig-
nificant interaction. Accordingly, we conducted pooled analyses
comparing patients in the permissive underfeeding group with
those in the target feeding group and the IIT group compared
with the CIT group (31). Moreover, for the comparison of one
intervention, we carried out unadjusted and adjusted analyses for
the second intervention and found similar results and presented
the unadjusted results.
RESULTS
Patient characteristics
Of 1587 patients assessed for eligibility, 240 met the inclusion
criteria and were recruited for the study (Figure 1). The baseline
characteristics of the study groups, stratified by the inter-
ventions, are presented in Table 1. No significant differences in
baseline characteristics were observed between patients in the
permissive underfeeding and target feeding groups or between
patients in the IIT and CIT groups.
Interventions
Caloric intake was consistently lower in the permissive un-
derfeeding group than in the target feeding group (Figure 2, A).
The average percentage caloric intake/requirement throughout
the ICU stay was 59.0 6 16.1% in the permissive underfeeding
group compared with 71.4 6 22.8% in the target feeding group
(P , 0.0001) (Table 2). Protein intake, glucose concentrations,
and insulin doses were similar in the permissive underfeeding
and target feeding groups (Figure 2, BD; Table 2). Regarding
the IIT and CIT groups, the average daily insulin dose
throughout the study period was 62.8 6 39.9 units in the IIT
group and 23.0 6 33.4 in the CIT group (P , 0.0001), with
corresponding average glucose concentrations of 6.2 6 0.7 and
8.66 2.0 mmol/L (P, 0.0001), respectively, whereas the calorie
and protein intakes were similar (Table 2; Figure 2, EH).
Outcomes
Permissive underfeeding compared with target feeding
Mortality. The primary endpoint of 28-d all-cause mortality was
18.3% in the permissive underfeeding group compared with
23.3% in the target feeding group (RR: 0.79; 95% CI: 0.48,
1.29; P = 0.34) (Table 3). Hospital mortality was lower in the
permissive underfeeding group than in the target feeding group
(30.0% compared with 42.5%; RR: 0.71; 95% CI: 0.50, 0.99;
P = 0.04). Mortality at 180 d was 32.8% in the permissive
underfeeding compared with 44.4% in the target feeding group
(RR: 0.74; 95% CI: 0.53, 1.03; P = 0.07). Kaplan-Meier sur-
vival estimates are shown in Figure 3, which indicate separa-
tion in the probability of survival between the 2 groups,
although not statistically significant (log-rank test, P = 0.16).
Other endpoints. ICU LOS and MVD were 11.7 6 8.1 compared
with 14.5 6 15.5 (P = 0.09) and 10.6 6 7.6 compared with
13.2 6 15.2 (P = 0.10) in the permissive underfeeding and
target feeding groups, respectively (Table 4). No significant
TABLE 4
Secondary endpoints in the permissive underfeeding, target feeding, intensive insulin therapy, and conventional insulin therapy groups1
Variable
Caloric intake Insulin therapy
Permissive underfeeding
(n = 120)
Target feeding
(n = 120) P value2Intensive
(n = 120)
Conventional
(n = 120) P value2
Cause of death [n (%)]
Multiorgan failure 22 (84.6) 29 (93.6) 0.55 24 (88.9) 27 (90) 0.71
Cardiac 2 (7.7) 1 (3.2) 1 (3.7) 2 (6.7)
Other 2 (7.7) 1 (3.2) 2 (7.4) 1 (3.3)
ICU LOS (d) 11.7 6 8.13 14.5 6 15.5 0.09 13.1 6 9.8 13.1 6 14.7 0.95
Hospital LOS (d) 70.2 6 106.9 67.2 6 93.6 0.81 70.7 6 106.3 66.7 6 94.3 0.76
Mechanical ventilation duration (d) 10.6 6 7.6 13.2 6 15.2 0.10 11.6 6 8.6 12.1 6 14.8 0.74
Hypoglycemic episodes/100 treatment days [n (%)] 37/1408 (2.6) 47/1736 (2.7) 0.89 67/1577 (4.2) 17/1566 (1.1) ,0.0001
Hypoglycemia [n (%)] 25 (20.8) 21 (17.5) 0.51 38 (31.7) 8 (6.7) ,0.0001
PRBC transfusion (units/d) 0.07 6 0.16 0.12 6 0.24 0.03 0.07 6 0.16 0.11 6 0.24 0.13
Renal replacement therapy [n (%)] 15 (12.5) 23 (19.2) 0.16 20 (16.7) 18 (15) 0.72
Hypokalemic episodes [n (%)] 9 (7.5) 23 (19.2) 0.008 17 (14.2) 15 (12.5) 0.70
ICU-acquired infections (n)
Urinary tract infection/1000 Foley catheter days 2 5.4 0.09 5.3 2.4 0.18
Catheter-related infection/1000 central line days 5.9 10 0.23 7.6 8.7 0.75
Ventilator associated pneumonia/1000 ventilator days 14 10 0.34 14.8 8.9 0.15
Tracheobronchitis/1000 ventilator days 25.7 24.5 0.84 23.9 26.1 0.71
Any ICU-acquired infection/1000 ICU days 54.7 53.6 0.89 56.4 51.7 0.56
ICU-acquired sepsis [n (%)]
All sepsis episodes 53 (44.2) 56 (46.7) 0.70 59 (49.2) 50 (41.7) 0.24
Severe sepsis/septic shock 53 (44.2) 55 (45.8) 0.80 58 (48.3) 50 (41.7) 0.30
1ICU, intensive care unit; LOS, length of stay; PRBC, packed red blood cell.
2 A t test was used for continuous variables, a chi-square test was used for categorical variables, and Z approximation was used for rates. No significant
interaction was found between the 2 interventions.3
Mean 6 SD (all such values).
574 ARABI ET AL
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
7/9
difference was observed in hypoglycemia rates, ICU-acquired
infection rates, and the need for renal replacement therapy.
Patients in the permissive underfeeding group required less
packed red cell transfusions than did the target group (0.07 6
0.16 compared with 0.12 6 0.24 units/d; P = 0.03) and had
less hypokalemic episodes (7.5% of patients compared with
19.2% of patients; P = 0.008).
IIT compared with CITMortality. No differences in 28-d all-cause mortality or in the
ICU, hospital, and 180-d mortality rates were observed be-
tween the IIT and CIT groups (Table 3)a finding that
reflected in Kaplan-Meier survival estimates (log-rank test,
P = 0.71; Figure 3).
Other endpoints. No significant differences in any of the sec-
ondary endpoints were observed between the IIT and CIT
groups (Table 4), except for hypoglycemia, which occurred
more frequently in the IIT group than in the CIT group
(31.7% compared with 6.7%; P , 0.0001).
DISCUSSIONThe main finding in our study was that permissive un-
derfeeding was associated with lower mortality and morbidity
than was target feeding in critically ill patients. IIT was not
associated with an improvement in outcomes relative to CIT.
Studies examining the optimal caloric dose in critically ill
patients have yielded different results (16, 18, 19, 3236), pos-
sibly because of their heterogeneity with regard to study design,
patient population, route of delivery, timing, and dose of nutri-
tion. In a cohort study of 48 critically ill patients, Villet et al (32)
found that a cumulative energy deficit was associated with longer
ICU LOS, increased MVD, and more complications. Rubinsonet al (33) found, in a study of 138 medical ICU patients, that
patients receiving ,25% of prescribed energy requirements had
a higher risk of bloodstream infections than did all other pa-
tients. In an RCT of 82 severe traumatic brain injury patients,
Taylor et al (34) found that patients in the enhanced nutrition
group had better neurologic outcomes after 3 mo of the injury
than did patients in the standard group. However, no differences
in neurologic outcome at 6 mo or in mortality were observed
(34). In a multicenter cluster RCT that examined the effect of
the implementation of evidence-based feeding algorithms (35),
the investigators showed that the intervention led to a statisti-
cally nonsignificant increase in caloric delivery (1264 compared
with 998 kcal; P = 0.31) and a significant reduction in hospitalLOS. However, it remains unclear whether the large changes in
clinical outcomes are related to the small increase in dose of
nutrition (21). Other studies showed no effect on patient out-
comes with increased caloric intake, as seen in a cluster RCT in
27 hospitals in Australia and New Zealand (36). Finally, other
studies showed improved outcomes with reduced caloric intakes.
Krishnan et al (16) found that a moderate caloric intake (ie, 33
65% of the recommended targets) was associated with lower
MVD, ICU LOS, and hospital mortality than was a higher ca-
loric intake. We documented similar finding in a nested cohort
study of 523 patients (17). In a study of 40 critically ill obese
patients, Dickerson et al (18) found that lower calorie intakes
were associated with a lower ICU LOS and duration of antibi-otic therapy. Ibrahim et al (19) compared early aggressive and
late feeding protocols in 150 patients and found that early
feeding was associated with higher incidences of ventilator-
associated pneumonia and Clostridium difficileassociated di-
arrhea and longer ICU and hospital LOS.
Our study was the first RCT to compare permissive enteral
underfeeding with target enteral feeding in critically ill medical
surgical patients. A statistically significant treatment effect in the
primary endpoint (ie, 28-d all-cause mortality) was not found, but
a reduction in the hospital and 180-d mortality rates was observed
in the permissive underfeeding group. The lack of a statistically
significant difference in the 28-d mortality rate might be related to
several factors. First, the achieved calorie intake, especially in thetarget feeding group, was below that which was planned (ach-
ieved: 71%; planned: 90100%), which resulted in a smaller
difference in interventions (59% compared with 71%). The
difficulty of achieving nutritional targets in the ICU is well
documented in the literature (37), including in interventional
studies aimed at augmenting caloric intake (3436). Such dif-
ficulty is due to intrinsic patient factors, including impaired
gastric emptying as well as practice-related factors such as delays
in initiating enteral feeding, slow advancement of feeding, and
waiting for active bowel sounds. To minimize the effect of the
latter, we followed a standardized validated protocol (25) and
used a compensation strategy. Nevertheless, our study showed
FIGURE 3. Kaplan-Meier survival curves for the permissive under-feeding and target feeding groups (A) and for the intensive insulin therapyand conventional insulin therapy groups (B).
PERMISSIVE UNDERFEEDING IN CRITICALLY ILL PATIENTS 575
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
8/9
that a target caloric intake of.90% may be achievable in some
patients, but it is difficult to achieve in a group of critically ill
patients in an intervention study. This high expectation has
probably overestimated the treatment effect used in the sample
size calculation and undermined the power of the study. Second,
the permissive underfeeding target of 6070% might not have
been hypocaloric enough and that the benefit is expected with
lower caloric intakes, as shown in several observational studies
(16). Nevertheless, our trial compared the best achievable target(target group) in a reduced-calorie group (permissive un-
derfeeding) as reflected by the clear separation in caloric intake
between the 2 groups. Third, the sample size might have been
underestimated because it was based on anticipated treatment
effect derived from observational studies; which often show
larger treatment effects than in randomized studies (38). Fourth,
although commonly used in critical care trials, the endpoint of
28-d mortality might not detect the effect of the intervention,
and long-term outcome measures might be better endpoints. In
fact, our study showed larger treatment effects on long-term
mortality rates, such as hospital and 180-d mortality rates, than
on short-term mortality rates, including ICU and 28-d mortality
rates. Fifth, the lack of a statistically significant difference mightreflect the true effect; ie, no difference in outcomes between the
permissive underfeeding and target feeding groups. However,
we observed a statistically significant difference in hospital
mortality. In addition, the point estimates of 28-d, ICU, and
180-d mortality rates were consistently better in the permissive
underfeeding group, which suggested a potential improvement
in outcome with reduced calorie intakes.
Our study showed a lack of benefit with IIT and a significant
increase in the risk of hypoglycemia. IIT has gained great interest
after van den Berghe et al (39) reported that IITwas associated with
a reduction in mortality and morbidity in surgical ICU patients,
which led to calls to adopt this therapy as a standard of care for ICU
patients (4042). Several subsequent RCTs did not show a clearclinical outcome benefit, including a trial from our center; from the
multicenter Glucontrol, VISEP, and NICE-SUGAR trials (26, 39,
4345); and from 2 subsequent meta-analyses (46, 47). Of note,
our insulin protocols had the same glucose concentration targets as
did both van den Berghe trials, and we achieved comparable
glucose concentrations. Mean glucose concentrations were 6.2 6
0.7 compared with 8.6 6 2.0. mmol/L in our IIT and CIT groups,
5.76 1.1 compared with 8.5 6 1.8 mmol/L in the trial in surgical
patients, and 6.2 6 1.6 compared with 8.5 6 1.7 mmol/L in the
trial in medical patients. The lack of benefit from IIT raises the
question of what the optimal blood glucose concentration is in
critically ill patients and whether other factors, such as glucose
variability, are as important (48).Our study should be interpreted in light of its strengths and
weaknesses. The strengths included the study design (ie, RCT)
and the concealment resulting in well-balanced groups. The use
of supplemental protein allowed the maintenance of similar
protein intakes in the 2 groups, despite significant differences in
calorie intake. The limitations included being monocenter, the
unblinded nature of the study, and the lower than anticipated
calorie intake in the target group.
In conclusion, the study suggests that permissive underfeeding
may be associated with lower mortality than target feeding in
critically ill patients and that a larger multicenter study be con-
ducted to confirm these findings.
We acknowledge Abdullah Al Shimemeri, Asgar H Rishu, Emma
P Querubin, Molly M Mdletshe, Brintha Naidu, Monica Pillay, and Azzam
Mohammed, who are all from King Abdulaziz Medical City, Riyadh, Saudi
Arabia.
The authors responsibilities were as followsYMA (Principal Investiga-
tor): had full access to all of the data, takes full responsibility for the integrity
of the data and the accuracy of the data analysis, and was responsible for the
study concept and design, overall supervision, statistical expertise, and draft-
ing of the manuscript; HMT: responsible for the study concept and design,
analysis and interpretation of the data, statistical expertise, and drafting of
the manuscript; GSD: responsible for the study concept and design, data col-
lection, drafting of the manuscript, and critical revision of the manuscript;
AAD and MAS: responsible for the study concept and design, study super-
vision, anddrafting of themanuscript;MHS: responsible forthe study concept
and design, data collection, dietary protocol implementation, and regular in-
services to the staff; SHK: responsible for the study concept and design, data
collection, protocol implementation, and training of the staff; and RB: re-
sponsible for the study concept and design, data collection, protocol imple-
mentation, and training of the staff. None of the authors had any conflicts of
interest.
REFERENCES1. Marik PE, Zaloga GP. Early enteral nutrition in acutely ill patients:
a systematic review. Crit Care Med 2001;29:226470.2. Norman K, Pichard C, Lochs H, Pirlich M. Prognostic impact of
disease-related malnutrition. Clin Nutr 2008;27:515.3. Goiburu ME, Goiburu MM, Bianco H, et al. The impact of malnutrition
on morbidity, mortality and length of hospital stay in trauma patients.
Nutr Hosp 2006;21:60410.4. Lin SJ, Kaeberlein M, Andalis AA, et al. Calorie restriction extends
Saccharomyces cerevisiae lifespan by increasing respiration. Nature
2002;418:3448.5. Heilbronn LK, Ravussin E. Calorie restriction and aging: review of the
literature and implications for studies in humans. Am J Clin Nutr 2003;
78:3619.6. Cohen HY, Miller C, Bitterman KJ, et al. Calorie restriction promotes
mammalian cell survival by inducing the SIRT1 deacetylase. Science
2004;305:3902.
7. Heilbronn LK, de Jonge L, Frisard MI, et al. Effect of 6-month calorierestriction on biomarkers of longevity, metabolic adaptation, and oxi-
dative stress in overweight individuals: a randomized controlled trial.
JAMA 2006;295:153948.8. Dandona P, Mohanty P, Ghanim H, et al. The suppressive effect of
dietary restriction and weight loss in the obese on the generation of
reactive oxygen species by leukocytes, lipid peroxidation, and protein
carbonylation. J Clin Endocrinol Metab 2001;86:35562.9. Gredilla R, Sanz A, Lopez-Torres M, Barja G. Caloric restriction de-
creases mitochondrial free radical generation at complex I and lowers
oxidative damage to mitochondrial DNA in the rat heart. FASEB J
2001;15:158991.10. Hyun DH, Emerson SS, Jo DG, Mattson MP, de Cabo R. Calorie re-
striction up-regulates the plasma membrane redox system in brain cells
and suppresses oxidative stress during aging. Proc Natl Acad Sci USA
2006;103:1990812.
11. Shinmura K, Tamaki K, Saito K, Nakano Y, Tobe T, Bolli R. Car-dioprotective effects of short-term caloric restriction are mediated by
adiponectin via activation of AMP-activated protein kinase. Circulation
2007;116:280917.12. Shinmura K, Tamaki K, Bolli R. Impact of 6-mo caloric restriction on
myocardial ischemic tolerance: possible involvement of nitric oxide-
dependent increase in nuclear Sirt1. Am J Physiol Heart Circ Physiol
2008;295:H234855.13. Bouffard YH, Delafosse BX, Annat GJ, Viale JP, Bertrand OM, Motin
JP. Energy expenditure during severe acute pancreatitis. JPEN J Pa-
renter Enteral Nutr 1989;13:269.14. Abiles J, de la Cruz AP, Castano J, et al. Oxidative stress is increased
in critically ill patients according to antioxidant vitamins intake, in-
dependent of severity: a cohort study. Crit Care 2006;10:R146.15. Zauner A, Nimmerrichter P, Anderwald C, et al. Severity of insulin
resistance in critically ill medical patients. Metabolism 2007;56:15.
576 ARABI ET AL
atUniversidadeFederaldaBahiaonMay4,2011
www.ajcn.org
Downloadedfrom
http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/http://www.ajcn.org/ -
8/3/2019 Mardiana_RCT2011
9/9