febrile non‐haemolytic transfusion reaction occurrence and ... · correspondence: mikhail menis,...

Vox Sanguinis (2014)

ORIGINAL PAPER © 2014 International Society of Blood TransfusionDOI: 10.1111/vox.12215

Febrile non-haemolytic transfusion reaction occurrence andpotential risk factors among the U.S. elderly transfused inthe inpatient setting, as recorded in Medicare databasesduring 2011–2012M. Menis,1 R. A. Forshee,1 S. A. Anderson,1 S. McKean,2 R. Gondalia,2 R. Warnock,2 C. Johnson,2 P. D. Mintz,1

C. M. Worrall,3 J. A. Kelman3 & H. S. Izurieta11Food and Drug Administration, Silver Spring, MD, USA2Acumen LLC, Burlingame, CA, USA3Centers for Medicare & Medicaid Services, Baltimore, MD, USA

Received: 23 July 2014,revised 25 September 2014,accepted 30 September 2014

Background and Objectives Febrile non-haemolytic transfusion reaction (FNHTR)is an acute transfusion complication resulting in fever, chills and/or rigours.Study’s objective was to assess FNHTR occurrence and potential risk factorsamong inpatient U.S. elderly Medicare beneficiaries, ages 65 and older, during2011–2012.

Materials and Methods Our retrospective claims-based study utilized large Medi-care administrative databases. FNHTR was ascertained via ICD-9-CM diagnosiscode, and transfusions were identified by recorded procedure and revenue centrecodes. The study ascertained FNHTR rates among the inpatient elderly overalland by age, gender, race, blood components and units transfused. Multivariatelogistic regression analyses were used to assess potential risk factors.

Results Among 4 336 338 inpatient transfusion stays for elderly during 2011–2012, 2517 had FNHTR diagnosis recorded, an overall rate of 58�0 per 100 000stays. FNHTR rates (per 100 000 stays) varied by age, gender, number of units andblood components transfused. FNHTR rates were substantially higher for RBCs- andplatelets-containing transfusions as compared to plasma only. Significantly higherodds of FNHTR were identified with greater number of units transfused (P < 0�01),for females vs. males (OR = 1�15, 95% CI 1�04–1�27), and with 1-year histories oftransfusion (OR = 1�25, 95% CI 1�10–1�42), lymphoma (OR = 1�22, 95% CI 1�02–1�46), leukaemia (OR = 1�90, 95% CI 1�56–2�31) and other diseases.

Conclusions Our study shows increased FNHTR occurrence among elderly withgreater number of units and with RBCs- and platelets-containing transfusions,suggesting need to evaluate effectiveness of prestorage leucoreduction in elderly.The study also suggests importance of prior recipient alloimmunization andunderlying health conditions in the development of FNHTR.

Key words: administrative databases, elderly, febrile non-haemolytic transfusionreaction, inpatient, Medicare, potential risk factors.

Introduction

Febrile non-haemolytic transfusion reaction (FNHTR) is

one of the most frequently occurring acute transfusion

complications, which results in fever, chills, rigours,

Correspondence: Mikhail Menis, Office of Biostatistics and Epidemiology,Center for Biologics Evaluation and Research, FDA, 10903 NewHampshire Avenue, White Oak Building #71, Silver Spring, MD 20993-0002, USAE-mail: [email protected]

1

headache, nausea and/or vomiting [1–7]. The two known

mechanisms for FNHTR occurrence include an antileuco-

cyte antibody mediated mechanism and accumulation of

pro-inflammatory cytokines during storage of cellular

components. During the immune-based reaction, pre-

formed leucocyte antibodies in the recipients due to prior

alloimmunization (e.g. transfusion, pregnancy) interact

with leucocytes of transfused blood components resulting

in release of endogenous pyrogens (e.g. interleukins,

tumour necrosis factor) and consequently in FNHTR.

Additionally, the storage of cellular components, espe-

cially platelets, results in accumulation of pro-inflamma-

tory cytokines which may cause FNHTR [1, 2, 5, 6, 8–11].

Although FNHTR is non-life-threatening, it causes patient

discomfort and can potentially exacerbate underlying

health conditions as well as result in interruption of

needed transfusions [1, 5, 7]. Specifically, as fever associ-

ated with FNHTR may also be representative of serious

health conditions (e.g. sepsis, haemolytic transfusion

reactions), the life-saving transfusion(s) are likely to be

stopped until reaction investigation is completed [1, 5, 7].

To our knowledge, there have been no published popu-

lation-based studies assessing FNHTR occurrence and

potential risk factors among the inpatient U.S. elderly. As

published literature suggests a disproportionately high

blood utilization among the elderly and the importance

of prior alloimmunizations (e.g. transfusions, pregnancies)

in FNHTR occurrence, the elderly population may be at

an increased risk for FNHTR [1, 10, 12–16]. The Centers

for Medicare & Medicaid Services (CMS) administers

Medicare, a national health insurance that provides cov-

erage to virtually all U.S. elderly persons ages 65 and

older and maintains large administrative databases [17–

19]. The primary objective of our retrospective popula-

tion-based study was to assess FNHTR occurrence and

potential transfusion and recipient risk factors (e.g. blood

components, number of units transfused, demographics,

health conditions) among the inpatient U.S. elderly Medi-

care beneficiaries during 2011–2012, using large CMS

administrative databases.

Materials and methods

This retrospective claims-based study utilized large Medi-

care administrative databases in co-ordination with the

Centers for Medicare & Medicaid Services and under

CBER’s Safe Blood Program within the SafeRx Project.

The study utilized the 100% Inpatient and Medicare

Enrollment Common Working Files (CWF) for calendar

years 2011 and 2012. The Inpatient Files contain inpa-

tient hospital claims with medical services rendered and

include patient diagnoses and procedures. The Medicare

Enrollment Files contain demographic and enrolment

information, and help in ascertaining coverage eligibility.

Transfusions of blood and blood components [e.g. whole

blood, packed red blood cells (RBCs), platelets, plasma,

autologous blood] were identified by recorded Interna-

tional Classification of Diseases, 9th Revision, Clinical

Modification (ICD-9-CM) procedure codes (99.00–99.05,

99.07) and revenue centre codes (0381–0384, 0390–0392,

0399), while FNHTR was ascertained via ICD-9-CM diag-

nosis code 780.66 titled ‘Febrile non hemolytic transfu-

sion reaction’ [20, 21]. Revenue centre units were used to

quantify blood use [21, 22]. Inpatient transfusion stays

for the elderly, defined as inpatient hospital stays with at

least one blood component transfused, and unique by

beneficiary and hospital admission date, were used as

study’s unit of analysis to assess FNHTR occurrence rates

and conduct multivariate regression analyses among the

U.S. elderly. During inpatient hospital stays, the diagnoses

are generally made and recorded into the medical records

by treating physicians and based on patients’ clinical pre-

sentations and laboratory values. Upon discharge, the

diagnoses are coded by medical coders for billing pur-

poses and captured in the large CMS databases. Inpatient

transfusion stays were grouped into ten mutually exclu-

sive blood component categories: RBCs only; Plasma

only; Platelets only; Plasma and Platelets; RBCs and

Plasma; RBCs and Platelets; RBCs, Plasma and Platelets;

Autologous Transfusion only; Non-specific Blood Compo-

nent (i.e. based on non-specific revenue centre codes

only); and Other Transfusion Combinations. Analyses

using blood quantities (i.e. blood units) transfused

excluded transfusion stays with either zero units or 200

units or more recorded, as these were likely due to either

blood quantity reporting or recording errors.

The study evaluated FNHTR rates (per 100 000 inpa-

tient transfusion stays) among elderly Medicare beneficia-

ries, overall and by age, gender, race, number of units

and blood components transfused. Univariate comparisons

of transfusion stays with FNHTR vs. without FNHTR were

conducted by demographic characteristics, admission

type, mechanical ventilation, inpatient mortality, length

of stay, the use of intensive care unit (ICU) and coronary

care unit (CCU). Multivariate logistic regression analyses

were used to assess the effect of potential transfusion

(e.g. blood components, number of units transfused) and

recipient (e.g. age, gender, race, Charlson Comorbidity

Index [23], health conditions) risk factors on FNHTR

occurrence, while controlling for potential confounders.

The final multivariate regression analyses retained char-

acteristics that are likely biologically and clinically

important for FNHTR occurrence and thus essential in

adjustment for the underlying FNHTR risk, including but

not limited to blood components, number of units trans-

fused, age, gender, race, Charlson Comorbidity Index

© 2014 International Society of Blood TransfusionVox Sanguinis (2014)

2 M. Menis et al.

Score (CCIS) [23] and underlying health conditions. As

transfusion recipient prior alloimmunization plays an

important role in the development of FNHTR, the regres-

sion analyses also adjusted for history of prior transfu-

sion(s). Health conditions and procedures were

ascertained within 1 year prior to inpatient admission (i.e.

1-year histories) to assure their presence before the stud-

ied outcome, while other potential risk factors (e.g. num-

ber of units, blood components) were assessed as recorded

during inpatient transfusion stay. The Elixhauser Comor-

bidities [24] were used to define health conditions when-

ever available. Charlson Comorbidity Index Score is a

comorbidity burden measure based on 17 underlying

health conditions, with a weighted score assigned to each

comorbid condition depending on the relative risk of 1-

year mortality [23, 25]. Both the Elixhauser Comorbidities

[24] and CCIS [23] were based on the enhanced ICD-9-

CM coding algorithm as described by Quan et al. [25].

The multivariate regression analyses were conducted

overall, by gender and age groups (65–79 vs. >79) to help

better assess their effects on FNHTR occurrence. The sen-

sitivity multivariate regression analyses included age and

CCIS as continuous variables to help quantify their effect

on FNHTR occurrence.

The univariate and multivariate analyses utilized SAS

9.2 (SAS Institute, Cary, NC, USA) to assess associations

between potential risk factors (e.g. blood components,

number of units transfused, recipient demographic char-

acteristics, CCIS [23], underlying health conditions) and

FNHTR occurrence. A priori level of significance was

P < 0�05. Specifically, the unadjusted relative risks (i.e.

rate ratios or RRs) and 95% confidence intervals (CI) were

estimated using SAS. Chi-squared tests were used to com-

pare proportions. Trend analyses of FNHTR occurrence by

age and units transfused were assessed using the Coch-

ran–Armitage statistic. Multivariate logistic regression

analyses were used to estimate adjusted odds ratios (ORs)

and 95% CIs for FNHTR occurrence among the elderly

Medicare beneficiaries with quantifiable number of blood

units recorded during transfusion stays, who were also

continuously enrolled in Medicare Parts A (inpatient

health coverage) and B (outpatient health coverage) and

not C (supplemental non-Fee-For-Service coverage) for at

least 1 year prior to their transfusion stays. The regres-

sion model fit and predictive ability were estimated using

the Hosmer–Lemeshow Goodness-of-Fit test and Area

Under the ROC Curve (AUC) [26, 27]. Multicollinearity

was evaluated using Variance Inflation Factors (VIFs) for

each model covariate. Firth’s method of penalized likeli-

hood was applied to multivariate regression analyses to

account for potential separability issues and help correct

for a possible small-sample bias in the multivariate

regression analyses [28, 29]. As part of sensitivity analy-

ses, multivariate regression splines were implemented in

STATA Version 11 (STATA Corp, College Station, TX,

USA) using software package by Royston et al. [30] to

ascertain potentially nonlinear associations between

FNHTR and continuous variables of blood units trans-

fused, age and CCIS, while controlling for other model

covariates. The continuous variables were categorized into

piecewise polynomial functions (i.e. splined functions),

and adjusted odds of FNHTRs and corresponding 95% CIs

were estimated for each of the continuous covariates.

Results

Among 4 336 338 inpatient transfusion stays for elderly

beneficiaries during 2011–2012, 2517 had FNHTR diagno-

sis code recorded, resulting in an overall rate of 58�0 per

100 000 stays. Table 1 compares unadjusted FNHTR rates

(per 100 000 transfusion stays) by blood component

groups, age, gender, race and number of units transfused.

When compared to plasma only stays, the unadjusted rate

comparisons show significantly increased risk of FNHTR

(P < 0�0001) with RBCs only (RR = 3�1, 95% CI 2�2–4�3),platelets only (RR = 2�6, 95% CI 1�6–4�3), RBCs and

plasma (RR = 2�7, 95% CI 1�8–3�9), RBCs and platelets

(RR = 8�9, 95% CI 6�2–12�8), and RBCs, plasma and plate-

lets (RR = 3�9, 95% CI 2�5–6�1). FNHTR rate was also

found to be significantly higher for females vs. males

(RR = 1�1, 95% CI 1�0–1�2). The unadjusted FNHTR rates

increased significantly with increasing number of units

transfused (P < 0�01), and oppositely, decreased with

advancing age (P < 0�01) (Table 1).

Table 2 shows comparison of transfusion stays with

FNHTR vs. without FNHTR by demographic and other

characteristics. Transfusion stays with FNHTR recorded

were significantly more likely to be female (P < 0�02) andyounger elderly ages 65–79 (P < 0�0001) and admitted on

urgent basis (P < 0�01). In contrast, stays with FNHTR

were significantly less likely (P < 0�0001) to be in the

ICU, 27�1% vs. 41�1%; in the CCU, 9�3% vs. 14�3%; as

well as to have continuous mechanical ventilation, 2�8%vs. 9�5%; and a length of stay of more than 10 days,

19�3% vs. 27�0%, as compared to stays without FNHTR.

Inpatient mortality was also significantly lower for

FNHTR stays vs. non-FNHTR stays, 2�6% vs. 6�9%, respec-

tively (Table 2).

Figure 1 shows the overall multivariate logistic regres-

sion analysis, which assesses potential risk factors for

FNHTR occurrence among inpatient elderly Medicare bene-

ficiaries during 2011–2012. When compared to plasma

only group, the adjusted overall regression results show

significantly elevated FNHTR odds ratios (P < 0�05) for

RBCs only (OR = 2�43, 95% CI 1�63–3�64), RBCs and

plasma (OR = 2�01, 95% CI 1�26–3�20), RBCs and platelets


FNHTR occurrence and risk factors in elderly 3

(OR = 3�18, 95% CI 2�04–4�97), RBCs, plasma and platelets

(OR = 2�31, 95% CI 1�31–4�06) and other transfusion com-

binations (OR = 2�51, 95% CI 1�27–4�96). Significantly ele-

vated odds of FNHTR were also identified for stays with 5–

9 units (OR = 1�36, 95% CI 1�15–1�61) and >9 units trans-

fused (OR = 1�39, 95% CI 1�13–1�71) as compared to 1 unit

transfused. In contrast, the results show significantly lower

odds of FNHTR with higher CCIS of 3 or more (P < 0�05).The overall regression analysis also shows significantly

higher odds of FNHTR for females vs. males (OR = 1�15,95% CI 1�04–1�27), as well as for inpatient transfusion

stays with 1-year histories of transfusion (OR = 1�25, 95%CI 1�10–1�42), lymphoma (OR = 1�22, 95% CI 1�02–1�46),leukaemia (OR = 1�90, 95% CI 1�56–2�31), metastatic can-

cer (OR = 1�23, 95% CI 1�04–1�45), blood loss anaemia

(OR = 1�34, 95% CI 1�15–1�56), aplastic anaemia and other

bone marrow failure (OR = 1�47, 95% CI 1�23–1�75), che-motherapy (OR = 1�36, 95% CI 1�13–1�64), and neoplasm

of uncertain behaviour of other lymphatic and hematopoi-

etic tissues (OR = 1�41, 95% CI 1�19–1�68) (Fig. 1)Multivariate regression analyses on FNHTR occurrence

by age (Fig. 2) and gender (Fig. 3) showed generally

Table 1 Comparison of unadjusted febrile non-haemolytic transfusion reaction (FNHTR) rates by blood components, units and demographic characteris-

tics among inpatient transfusion stays for elderly Medicare beneficiaries during 2011–2012

CategoryNumber oftransfusion stays

Number of stayswith FNHTR

FNHTR rate(per 100 000) RRa 95% CIb P-valuec

Overall 4 336 338 2517 58�0Blood Component Groupsd

Plasma only (Reference) 153 279 35 22�8 1�0RBC only 2 193 302 1546 70�5 3�1 2�2–4�3 <0�0001Platelets only 53 245 32 60�1 2�6 1�6–4�3 <0�0001Platelets and Plasma 10 383 4 38�5 1�7 0�6–4�8 0�32RBCs and Plasma 166 771 102 61�2 2�7 1�8–3�9 <0�0001RBCs and Platelets 98 928 201 203�2 8�9 6�2–12�8 <0�0001RBCs, Plasma, and Platelets 46 984 42 89�4 3�9 2�5–6�1 <0�0001Non-specific Blood Componente 1 539 643 525 34�1 1�5 1�1–2�1 <0�0001Autologous Transfusion Only 46 981 8 17�0 0�7 0�4–1�6 0�45Other Transfusion Combinations 26 822 22 82�0 3�6 2�1–6�1 <0�0001

Agef

65–69 (Reference) 799 836 502 62�8 1�070–74 835 434 524 62�7 1�0 0�9–1�1 0�9975–79 856 207 529 61�8 1�0 0�9–1�1 0�8080–84 833 696 475 57�0 0�9 0�8–1�0 0�13≥ 85 1 011 165 487 48�2 0�8 0�7–0�9 <0�0001

Sex

Male (Reference) 1 887 221 1038 55�0 1�0Female 2 449 117 1479 60�4 1�1 1�0–1�2 0�02

Race

Non-White (Reference) 739 783 429 58�0 1�0White 3 596 555 2088 58�1 1�0 0�9–1�1 0�98

Quantifiable Transfusionsg 3 982 634 2318 58�2Number of Units Transfusedf

1 (Reference) 742 201 323 43�5 1�02–4 2 326 874 1273 54�7 1�3 1�1–1�4 <0�00015–9 597 866 460 76�9 1�8 1�5–2�0 <0�0001>9 315 693 262 83�0 1�9 1�6–2�3 <0�0001

aUnadjusted relative risks (i.e. rate ratios) comparing FNHTR rates for a specific category vs. the reference category.b95% confidence intervals.cChi-Squared test to compare proportions.dBlood component groups are mutually exclusive.eThe non-specific blood component group is based on inpatient transfusion stays with recorded non-specific revenue centre codes only (0390, 0391,

0392, or 0399).fChi-Squared tests for FNHTR trend analyses: by age (P < 0�01) and by units transfused (P < 0�01).gQuantifiable transfusions group includes inpatient transfusion stays with nonzero units recorded of <200 units per stay.


4 M. Menis et al.

similar results, with some exceptions as displayed. Specif-

ically, Fig. 2 shows elevated odds of FNHTR for platelets

only as compared to plasma only group in younger

elderly ages 65–79 (OR = 2�05, 95% CI 0�98–4�28), and in

contrast, non-elevated FNHTR odds in older elderly ages

80 and older (OR = 0�65, 95% CI 0�20–2�12). Figure 2

also shows a significant reduction in FNHTR odds with

CCIS of 5 and above in elderly ages 65 to 79 (OR = 0�71,95% CI 0�57–0�89) and a non-significant reduction

among older elderly (OR = 0�87, 95% CI 0�67–1�13). Fig-ure 3 shows significantly elevated odds of FNHTR for

platelets only group in females (OR = 2�92, 95% CI 1�37–6�21), and in contrast, reduced FNHTR odds in males

(OR = 0�42, 95% CI 0�13–1�37). Figure 3 also shows a sig-

nificant reduction in FNHTR odds among males with CCIS

of 2 (OR = 0�71, 95% CI 0�54–0�92), 3 to 4 (OR = 0�71,95% CI 0�55–0�91) and 5 and above (OR = 0�61, 95% CI

0�46–0�79), and not among females.

Figures 4–6, respectively, show plots of multivariate

regression splines with odds of FNHTR by age, number of

units and CCIS, treated as continuous variables, and while

controlling for other covariates in the final regression

models. Figure 4 shows a linear reduction in the odds of

FNHTR with advancing age. In contrast, Fig. 5 shows a

nonlinear increase in the odds of FNHTR with increasing

number of units transfused and a nonzero FNHTR risk

starting at one unit transfused. Figure 6 shows a linear

reduction in the odds of FNHTR with increasing CCIS.

Figure 7 shows the overall sensitivity multivariate regres-

sion analysis with age and CCIS treated as continuous

variables. The results show a 1% reduction in the odds of

FNHTR with each additional year of transfusion recipi-

ent’s age (OR = 0�99, 95% CI 0�99–1�00, P = 0�069)among elderly Medicare beneficiaries. Similarly, Fig. 7

shows a 4% reduction in the odds of FNHTR for each

additional unit-increase in CCIS (OR = 0�96, 95% CI

0�94–0�99, P < 0�05), thus further suggesting the impor-

tance of underlying health conditions in FNHTR occur-

rence.

Discussion

Our retrospective claims-based study is the first popula-

tion-based nationwide study that assessed FNHTR occur-

rence and potential risk factors among the inpatient

elderly Medicare beneficiaries. FNHTR occurrence among

elderly in the U.S. varied by age, gender, number of units

and blood components transfused as well as by different

other characteristics (e.g. health conditions). Overall, our

population-based study showed a substantially elevated

FNHTR risk for stays with RBCs- and platelets-containing

transfusions. The study also showed an increased FNHTR

risk with greater number of units transfused, especially

with five or more units, for females vs. males, and for

elderly with histories of transfusion, chemotherapy, lym-

phoma, leukaemia and other health conditions (e.g.

malignancies, anaemias). In contrast, our study among

the U.S. elderly suggests a decline in FNHTR occurrence

with advancing age and with increasing CCIS as well as

suggests the lowest FNHTR risk for plasma only and

autologous only transfusions, which need further investi-

gations.

Table 2 Comparison of transfusion stays with febrile non-haemolytic

transfusion reaction (FNHTR) vs. without FNHTR by demographic and

other characteristics among inpatient elderly Medicare beneficiaries dur-

ing 2011–2012

Characteristics

FNHTRN = 2517(%)

No FNHTRN = 4 333 821(%)

UnadjustedORa P-valueb

Sex

Male 41�2 43�5 0�9 0�02Female 58�8 56�5 1�1 0�02

Race

White 83�0 82�9 1�0 0�98Non-White 17�0 17�1 1�0 0�98

Age (years)

65–79 61�8 57�5 1�2 <0�0001>79 38�2 42�5 0�8 <0�0001

Admission typec

Emergency 55�2 56�9 0�9 0�08Urgent 19�6 17�2 1�2 <0�01Elective 24�4 24�8 1�0 0�66

Continuous mechanical ventilation

Overalld 2�8 9�5 0�3 <0�0001<96 h 1�5 5�0 0�3 <0�0001≥96 h 1�4 4�8 0�3 <0�0001

Intensive care unit (ICU)

Yes 27�1 41�1 0�5 <0�0001Coronary care unit (CCU)

Yes 9�3 14�3 0�6 <0�0001Length of stay (days)

1–2 5�5 4�9 1�1 0�183–4 24�9 21�7 1�2 <0�00015–6 25�4 21�6 1�2 <0�00017–10 24�9 24�7 1�0 0�88> 10 19�3 27�0 0�6 <0�0001

Inpatient mortality

Yes 2�6 6�9 0�4 <0�0001

aUnadjusted Odds Ratios comparing the presence of specific characteris-

tics for stays with FNHTR vs. no FNHTR.bChi-Squared test to compare proportions; a priori level of significance is

P<0�05.cThis category may not add up to 100%.dOverall group includes continuous mechanical ventilation of <96 h,

≥96 h, and of unknown duration, which can be recorded within the same

transfusion stay.



Although currently available literature on FNHTR

occurrence is based on non-population-based studies (e.g.

single-centre studies) and the incidence of FNHTR occur-

rence among the elderly is unknown, the available litera-

ture generally supports our study’s findings of increased

FNHTR risk with RBCs- and platelets-containing transfu-

sions, greater number of units, among persons with prior

alloimmunization (e.g. transfusions, pregnancies), as well

as the importance of underlying health conditions (e.g.

malignancies) [1, 10–12, 31, 32]. Specifically, our study

shows an increased risk of FNHTR with 1-year histories

of blood transfusions, in females as compared to males,

as well as in persons with histories of different oncologi-

cal conditions and chemotherapy. The study’s findings of

significantly elevated odds of FNHTR with RBCs- and

platelets-containing transfusions, greater number of units

and prior alloimmunization suggest the need for further

clinical investigations of prestorage leucoreduction of cel-

lular components as a potential FNHTR prevention strat-

egy among the elderly, which is generally supported by

the literature [32–37]. Additionally, the study’s findings

of decreased FNHTR risk among elderly with advancing

age, with higher CCIS, and with histories of specific

chronic comorbidities (e.g. congestive heart failure, car-

diac arrhythmias, hypertension), suggest potential impor-

tance of underlying immunity among the elderly for

FNHTR occurrence, which needs further clinical evalua-

tions. In support of the literature, our study also suggests

FNHTR to be one of the most frequent transfusion-related

adverse events among the elderly [1, 3–5, 38–41]. As

Fig. 1 The overall multivariate regression analysis to assess potential risk factors for febrile non-haemolytic transfusion reaction occurrence among inpa-

tient elderly during 2011–2012, while controlling for potential confounders.


6 M. Menis et al.

such, physicians should be aware of FNHTR occurrence

among the elderly, which results in patient discomfort

and therapeutic intervention. FNHTR could also poten-

tially exacerbate underlying comorbidities or suggest

occurrence of serious transfusion-related events (e.g. sep-

sis, haemolytic transfusion reaction), and therefore delay

needed transfusions in the elderly and result in costly

transfusion reaction investigations and potentially in poor

patient outcomes (e.g. morbidity, mortality).

Our large population-based retrospective study to

assess FNHTR occurrence and potential risk factors among

the U.S. elderly used administrative claims databases and

thus limitations include potential under- or misrecording

of transfusion procedures, units and FNHTR diagnosis

code, as well as lack of clinical details to validate the

recorded FNHTR diagnosis. As diagnosis of FNHTR is typ-

ically of exclusion of other transfusion-related events that

may result in fever (e.g. bacterial contamination, haemo-

lytic transfusion reaction, TRALI), under- or misdiagnos-

ing is possible. For example, mild septic reactions or

other transfusion reactions that result in fever may

potentially be misclassified as FNHTR. Additionally, tem-

porality could not be established between all potential

risk factors and FNHTR, and as such additional clinical

investigations are needed to confirm our findings. As the

study was based on claims data, it was not able to evalu-

ate contribution of cellular component storage to the

occurrence of FNHTR among the elderly, potential donor

risk factors (e.g. presence of antileucocyte antibodies,

demographic characteristics), as well as lacked clinical

information to confirm FNHTR diagnosis and could not

evaluate effectiveness of prestorage leucoreduction

among the inpatient elderly. Furthermore, some of the

resulting FNHTR rates may be unstable because of very

small number of events recorded, particularly in associa-

tion with specific blood components, and as such may

result in non-significant associations. As rare events may

bias logistic regression results, especially when adjusting

for multiple confounders, our study used Firth’s method

of penalized likelihood to help correct for potential

small-sample bias [28, 29]. Finally, as our study is the

first population-based study on FNHTR occurrence

Fig. 2 Multivariate regression analyses to assess potential risk factors for febrile non-haemolytic transfusion reaction occurrence among inpatient elderly

ages 65–79 vs. 80 and older, during 2011–2012.



among the elderly, some of the results may be hypothe-

sis-generating. Therefore, additional clinical and epidemi-

ological investigations are needed to better understand

potential underlying recipient, transfusion and donor risk

factors for FNHTR occurrence as well as their clinical

significance.

Fig. 3 Multivariate regression analyses to assess potential risk factors for febrile non-haemolytic transfusion reaction occurrence among inpatient elderly

by gender, during 2011–2012.

Fig. 4 The overall adjusted odds of febrile non-haemolytic transfusion

reaction (FNHTR) by age, while controlling for covariates in the final

regression model, among elderly Medicare beneficiaries during 2011–

2012.


reaction (FNHTR) by number of units transfused, while controlling for co-

variates in the final regression model, among elderly Medicare beneficia-

ries during 2011–2012.


8 M. Menis et al.

In summary, our retrospective claims-based cohort

study is the first population-based study that assessed

FNHTR occurrence and potential risk factors among the

inpatient U.S. elderly. The study showed a statistically

significant increase in FNHTR occurrence for stays with

RBCs- and platelets-containing transfusions, with five or

more units transfused, as well as for females vs. males,

and in the elderly with histories of transfusions, malig-

nancies and anaemias. As supported by the literature

[32–37] and by the Canadian experience with universal

leucoreduction [42, 43], there may be a need for further

investigations and consideration of universal prestorage

leucoreduction of cellular components for FNHTR

prevention in the elderly, and especially for persons with

potential prior alloimmunizations (e.g. pregnancies, trans-

fusions) who are likely to be at a higher risk for FNHTR

occurrence. The study also suggests potential importance

of underlying immunity among the elderly for FNHTR

occurrence, which needs further clinical evaluations. As

FNHTR is one of the most frequently occurring transfu-

sion-related complications which may delay life-saving


reaction (FNHTR) by Charlson Comorbidity Index Score, while controlling

for covariates in the final regression model, among elderly Medicare ben-

eficiaries during 2011–2012.

Fig. 7 The overall sensitivity multivariate regression analysis with age and Charlson Comorbidity Index Score treated as continuous variables to assess

their effects on febrile non-haemolytic transfusion reaction occurrence among inpatient elderly during 2011–2012.



transfusions, our study suggests the importance of physi-

cian awareness of FNHTR occurrence, potential risk fac-

tors and specific prevention strategies among the elderly.

Finally, the study shows utility of large administrative

databases as an additional epidemiological tool in assess-

ment of FNHTR occurrence and potential risk factors and

thus will serve to help reduce the occurrence of FNHTR

and improve public health.

Acknowledgements

All authors made substantial contributions to research

design, or the acquisition, analysis or interpretation of

data and were involved in drafting of the paper or revis-

ing it critically. The manuscript has been approved by all

authors. Authors thank Renata Moldavskaya for her con-

tribution in editing manuscript drafts.

Funding

This study was funded by the U.S. Food and Drug Admin-

istration, Center for Biologics Evaluation and Research.

Conflict of interest

The authors declare that they have no conflict of interests

relevant to the manuscript submitted to Vox Sanguinis.

References1 Heddle NM: Febrile Nonhemolytic

transfusion reactions; in: Popovsky

MA (ed.): Transfusion Reactions, 3rd

edn. Bethesda, MD, AABB Press, 2007:

57–103

2 Pomper GJ: Chapter 53: febrile, aller-

gic, and nonimmune transfusion reac-

tions; in: Simon TL, Snyder EL,

Solheim BG, Stowell CP, Strauss RG,

Petrides M (eds): Rossi’s Principles of

Transfusion Medicine, 4th edn.

Bethesda, MD, AABB Press, 2009:

826–846

3 Public Health Agency of Canada.

Transfusion transmitted injuries.

Noninfectious Risks of Transfusion.

Available from: http://www.

phac-aspc.gc.ca/hcai-iamss/tti-it/risk-

s-eng.php [Last accessed 3 November

2014]

4 Serious Hazards of Transfusion Steer-

ing Group. Annual Serious Hazards of

Transfusion (SHOT) report 2011.

Available from: http://www.shotuk.

org/wp-content/uploads/2012/07/

SHOTANNUAL-REPORT_FinalWeb

VersionBookmarked_2012_06_22.pdf

[Last accessed 10 March 2014]

5 Torres R, Kenney B, Tormey CA:

Diagnosis, treatment, and reporting of

adverse effects of transfusion. Lab

Medicine 2012; 43:217–231

6 Perrotta PL, Snyder EL: Non-infectious

complication of transfusion therapy.

Blood Rev 2001; 15:69–83

7 Eder AF, Chambers LA: Noninfectious

complications of blood transfusion.

Arch Pathol Lab Med 2007; 131:708–

718

8 Heddle NM: Pathophysiology of feb-

rile nonhemolytic transfusion reac-

tions. Curr Opin Hematol 1999;

6:420–426

9 Heddle NM: Febrile nonhemolytic

transfusion reactions to platelets. Curr

Opin Hematol 1995; 2:478–483

10 Heddle NM, Klama L, Meyer R, et al.: A

randomized controlled trial comparing

plasma removal with white cell reduc-

tion to prevent reactions to platelets.

Transfusion 1999; 39:231–238

11 Heddle NM, Klama LN, Griffith L,

et al.: A prospective study to identify

the risk factors associated with acute

reactions to platelet and red cell trans-

fusions. Transfusion 1993; 33:794–797

12 Kleinman S, Chan P, Robillard P: Risks

associated with transfusion of cellular

blood components in Canada. Transfus

Med Rev 2003; 17:120–162

13 Kleinman S, Marshall D, AuBuchon J,

et al.: Survival after transfusion as

assessed in a large multistate US

cohort. Transfusion 2004; 44:386–390

14 Cobain TJ, Vamvakas EC, Wells A,

et al.: A survey of the demographics

of blood use. Transfus Med 2007;

17:1–15

15 Vamvakas EC, Taswell HF: Epidemiol-

ogy of blood transfusion. Transfusion

1994; 34:464–470

16 Wells AW, Mounter PJ, Chapman CE,

et al.: Where does blood go? Prospec-

tive observational study of red cell

transfusion in north England. BMJ

2002; 325:803

17 Virnig BA, McBean M: Administrative

data for public health surveillance and

planning. Annu Rev Public Health

2001; 22:213–230

18 Medicare Program - General Informa-

tion [monograph on the Internet].

Baltimore: CMS; [updated 2013 Aug

27]. Available from: http://www.cms.

gov/Medicare/Medicare-General-Infor

mation/MedicareGenInfo/index.html

[Last accessed 10 February 2014]

19 Files for Order - General Information

[monograph on the Internet]. Baltimore:

CMS; [updated 2013 Apr 23]. Available

from: http://www.cms.gov/Research-Sta

tistics-Data-and-Systems/Files-for-Order/

FilesForOrderGenInfo/index.html [Last

accessed 10 February 2014]

20 Hart AC, Stegman MS, Ford B (eds):

2011 Expert ICD-9-CM for hospitals-

volumes 1, 2, & 3: International

Classification of Diseases, 9th Revi-

sion, Clinical Modification. 6th edn.

Ingenix, 2010.

21 AABB billing guide for transfusion &

cellular therapy services [monograph

on the Internet]. Bethesda, MD, Ameri-

can Association of Blood Banks, 2007

October. Available from: http://

www.aabb.org/advocacy/reimburseme

ntinitiatives/Documents/reimbguidev0710

17.pdf [Last accessed 12 June 2014.]

22 Inpatient RIF: Data Documentation

[database on the Internet]. Minneapo-

lis, MN, Research Data Assistance Cen-

ter (ResDAC); c2013. Available from:

http://www.resdac.org/cms-data/files/ip-

rif/data-documentation [Last accessed

10 February 2014]

23 Charlson ME, Pompei P, Ales KL,

et al.: A new method of classifying


10 M. Menis et al.

prognostic comorbidity in longitudinal

studies: development and validation. J

Chron Dis 1987; 40:373–383

24 Elixhauser A, Steiner C, Harris DR,

et al.: Comorbidity measures for use

with administrative data. Med Care

1998; 36:8–27

25 Quan H, Sundararajan V, Halfon P,

et al.: Coding algorithms for defining

comorbidities in ICD-9-CM and ICD-10

administrative data. Med Care 2005;

43:1130–1139

26 Hosmer DW, Lemeshow S: Applied

Logistic Regression, 2nd edn. New

York, John Wiley & Sons, Inc., 2000

27 Paul P, Pennell ML, Lemeshow S: Stan-

dardizing the power of the Hosmer-

Lemeshow goodness of fit test in large

data sets. Stat Med 2013; 32:67–80

28 Firth D: Bias reduction of maximum

likelihood estimates. Biometrika 1993;

80:27–38

29 King G, Zeng L: Logistic regression in

rare events data. Polit Anal 2001;

9:137–163

30 Royston P, Sauerbrei W: Multivariable

modeling with cubic regression

splines: a principled approach. Stata J

2007; 7:45–70

31 Anderson NA, Gray S, Copplestone JA,

et al.: A prospective randomized study

of three types of platelet concentrates

in patients with haematological malig-

nancy: corrected platelet count incre-

ments and frequency of nonhaemolytic

febrile transfusion reactions. Transfus

Med 1996; 6:33–39

32 Da Ponte A, Bidoli E, Talamini R,

et al.: Pre-storage leucocyte depletion

and transfusion reaction rates in can-

cer patients. Transfus Med 2005;

15:37–43

33 Yazer MH, Podlosky L, Clarke G, et al.:

The effect of prestorage WBC reduc-

tion on the rates of febrile nonhemo-

lytic transfusion reactions to platelet

concentrates and RBC. Transfusion

2004; 44:10–15

34 Pruss A, Kalus U, Radtke H, et al.: Uni-

versal leukodepletion of blood compo-

nents results in a significant reduction

of febrile non-hemolytic but not aller-

gic transfusion reactions. Transfus

Apheresis Sci 2004; 30:41–46

35 Paglino JC, Pomper GJ, Fisch GS,

et al.: Reduction of febrile but not

allergic reactions to RBCs and platelets

after conversion to universal prestor-

age leukoreduction. Transfusion 2004;

44:16–24

36 King KE, Shirey RS, Thoman SK,

et al.: Universal leukoreduction

decreases the incidence of febrile non-

hemolytic transfusion reactions to

RBCs. Transfusion 2004; 44:25–29

37 Federowicz I, Barrett BB, Andersen JW,

et al.: Characterization of reactions after

transfusion of cellular blood compo-

nents that are white cell reduced before

storage. Transfusion 1996; 36:21–28

38 Menis M, Anderson SA, Forshee RA,

et al.: Transfusion-Related Acute Lung

Injury (TRALI) and potential risk fac-

tors among the inpatient U.S. elderly

as recorded in Medicare claims data,

during 2007-2011 Period. Transfusion

2014; 54:2182–2193

39 Menis M, Anderson SA, Forshee RA,

et al.: Transfusion-Associated Circula-

tory Overload (TACO) and potential

risk factors among the inpatient U.S.

elderly as recorded in Medicare admin-

istrative databases during 2011. Vox

Sang 2014; 106:144–152

40 Menis M, Izurieta HS, Anderson SA,

et al.: Outpatient transfusions and

occurrence of serious noninfectious

transfusion-related complications

among U.S. elderly, 2007-2008: utility

of large administrative databases in

blood safety research. Transfusion

2012; 52:1968–1976

41 Menis M, Forshee RA, Anderson SA,

et al.: Post-Transfusion Purpura (PTP)

occurrence and potential risk factors

among the inpatient U.S. elderly

as recorded in large Medicare data-

bases during 2011 through 2012.

Transfusion 2014; doi: 10.1111/trf.

12782

42 Fergusson D, H�ebert PC, Lee SK, et al.:

Clinical outcomes following institution

of universal leukoreduction of blood

transfusions for premature infants.

JAMA 2003; 289:1950–1956

43 H�ebert PC, Fergusson D, Blajchman

MA, et al.: Clinical outcomes follow-

ing institution of the Canadian univer-

sal leukoreduction program for red

blood cell transfusions. JAMA 2003;

289:1941–1949

