Hemoglobin Level and Hospital Mortality Among ICU Patients With Cardiac Disease Who Received Transfusions

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY

VOL. 66, NO. 22, 2015

ª 2015 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

ISSN 0735-1097/$36.00

PUBLISHED BY ELSEVIER INC.

http://dx.doi.org/10.1016/j.jacc.2015.09.057

Hemoglobin Level and Hospital Mortality Among ICU Patients With Cardiac Disease Who Received Transfusions Yew Y. Ding, MBBS, MPH,*y Boris Kader, PHD,zx Cindy L. Christiansen, PHD,zx Dan R. Berlowitz, MD, MPHzx

ABSTRACT BACKGROUND There is a paucity of randomized clinical trial data on the use of red blood cell (RBC) transfusion in critically ill patients, specifically in the setting of cardiac disease. OBJECTIVES This study examined how hemoglobin (Hgb) level and cardiac disease modify the relationship of RBC transfusion with hospital mortality. The aim was to estimate the Hgb level threshold below which transfusion would be associated with reduced hospital mortality. METHODS We performed secondary data analyses of Veterans Affairs intensive care unit (ICU) episodes across 5 years. Logistic regression quantified the effect of transfusion on hospital mortality while adjusting for nadir Hgb level, demographic characteristics, admission information, comorbid conditions, and ICU admission diagnoses. RESULTS Among 258,826 ICU episodes, 12.4% involved transfusions. Hospital death occurred in 11.6%. Without comorbid heart disease, transfusion was associated with decreased adjusted hospital mortality when Hgb was approximately <7.7 g/dl, but transfusion increased mortality above this Hgb level. Corresponding Hgb level thresholds were approximately 8.7 g/dl when comorbid heart disease was present and approximately 10 g/dl when the ICU admission diagnosis was acute myocardial infarction (AMI). Sensitivity analysis using additional adjustment for selected blood tests in a subgroup of 182,792 ICU episodes lowered these thresholds by approximately 1 g/dl. CONCLUSIONS Transfusion of critically ill patients was associated with reduced hospital mortality when Hgb level was <8 to 9 g/dl in the presence of comorbid heart disease. This Hgb level threshold for transfusion was 9 to 10 g/dl when AMI was the ICU admission diagnosis. (J Am Coll Cardiol 2015;66:2510–8) © 2015 by the American College of Cardiology Foundation.

H Listen to this manuscript’s

emoglobin (Hgb) level thresholds drive

restrictive strategy of receiving transfusions only

red blood cell (RBC) transfusion practices

when Hgb levels fell to <7 g/dl had significantly

in critically ill patients, with additional in-

lower hospital mortality compared with a liberal

fluence exerted by other patient factors, such as

strategy of transfusing when Hgb was <10 g/dl (2).

increased age, severity of illness, gastrointestinal

Subsequent observational studies of intensive care

hemorrhage, comorbid heart disease, and acute

unit (ICU) patients yielded mixed results, with some

myocardial infarction (AMI) (1–7). Over the past 2 de-

demonstrating increased mortality with transfusion,

cades, research has led to better understanding of the

whereas others showed the opposite (4,5,8–12). Sub-

extent to which these transfusion thresholds influ-

groups that derived greater benefit from transfusions

ence mortality.

included those with cardiac disease, such as those

In the landmark TRICC (Transfusion Requirements

with severe ischemic heart disease (13) and AMI (14).

in Critical Care) trial, patients randomly assigned to a

These findings were by no means consistent across

audio summary by JACC Editor-in-Chief Dr. Valentin Fuster.

From the *Department of Geriatric Medicine and Institute of Geriatrics and Active Ageing, Tan Tock Seng Hospital, Singapore; yHealth Services and Outcomes Research, National Healthcare Group, Singapore; zCenter for Healthcare Organization and Implementation Research (CHOIR), Bedford, Massachusetts; and the xBoston University School of Public Health, Boston, Massachusetts. Dr. Ding was funded by the National Medical Research Council of Singapore for his research fellowship in Boston. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Manuscript received May 4, 2015; revised manuscript received September 10, 2015, accepted September 15, 2015.

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

studies, with some showing benefit only when Hgb

METHODS

ABBREVIATIONS AND ACRONYMS

was <8 g/dl (12,15–18). In contrast, a recent, pilot randomized

controlled

2511

Transfusion of ICU Patients With Cardiac Disease

documented

We performed secondary analyses of Ve-

reduced mortality with a liberal transfusion strategy

trial

(RCT)

terans Affairs (VA) electronic databases at

in the setting of acute coronary syndrome (ACS) or

the Center for Healthcare Organization and

stable angina (19). For congestive heart failure (CHF),

Implementation Research. Ethical approval

infarction

the impact of transfusion is even more uncertain (20).

was obtained from the VA Bedford Institu-

CHF = congestive heart failure

Notably, the American Association of Blood Banks

tional Review Board. Individual-level data on

Hgb = hemoglobin

clinical practice guidelines recommended a restric-

VA ICU admissions during fiscal years 2001

ICU = intensive care unit

tive transfusion strategy for hospitalized patients

through 2005 were extracted from national-

with pre-existing cardiovascular disease for an Hgb

level VA databases. For hospitals with sepa-

level #8 g/dl, but these guidelines did not recom-

rate surgical and neurological ICUs, we only

mend a restrictive or liberal transfusion strategy for

selected medical ICU episodes. We excluded

hemodynamically stable patients with ACS. No spe-

operative

cases

with

surgical

ACS = acute coronary syndrome

AMI = acute myocardial

RBC = red blood cell RCT = randomized controlled trial

VA = Veterans Affairs

admitting

cific mention of a suggested strategy for patients with

diagnoses. Only the first ICU episode of each year

CHF was made (21).

for unique patients was included. SEE PAGE 2519

Hospital death was the outcome of interest, whereas RBC transfusion during the first 30 days

More recently, an elegant observational study of

of ICU admission was the treatment of interest.

approximately 35,000 AMI hospitalizations from 57

Transfusion receipt was defined as having an Inter-

centers used propensity score analysis to identify a

national Classification of Diseases-9th Revision-

subset of patients who were well-matched on 45

Clinical Modification (ICD-9-CM) procedure code of

variables. In this subset, transfusion was associated

99.04 (transfusion of packed cells) or 99.03 (trans-

with a 25% reduction in the odds of hospital mortality

fusion of whole blood) in the ICU file. ICU admis-

(22). However, the accompanying editorial cautioned,

sions with at least 1 transfusion documented during

that despite several observational studies on trans-

the first 30 days were collectively assigned as the

fusion in patients with significant cardiac disease,

“transfusion” group; all others were included in the

there is still uncertainty on when to transfuse these

“no transfusion” group. To exclude hospitals that

patients (23). Nonetheless, there is arguably accu-

were possible outliers in terms of transfusion prac-

mulating evidence that suggests that a restrictive

tices, we arbitrarily removed data from 8 of 120

transfusion strategy may not be optimal in this

hospitals where transfusion rates were >2 SD from

setting (13,14,19,22). Although another recent edito-

the mean.

rial echoed the sentiment that evidence on appro-

To specify the nadir Hgb level, we selected the

priate transfusion thresholds for patients with ACS is

lowest value during ICU admission before blood

weak, the possibility was entertained that such pa-

transfusion. If not transfused, we selected the lowest

tients may yet benefit from transfusion at higher Hgb

value during the first 30 days of ICU admission. If

levels (e.g., 9 to 10 g/dl) (24).

none was available, we used the lowest value during

Due to this ongoing debate, we conducted addi-

hospital admission before ICU admission.

tional analyses of a previously developed dataset to

Other explanatory variables included demographic

determine how Hgb level modifies the treatment ef-

characteristics, admission-related information, co-

fect of RBC transfusion during ICU admissions with

morbid

respect to hospital mortality in the context of cardiac

measures [25], Acute Physiology and Chronic Health

disease. Specifically, we sought to estimate the Hgb

Evaluation III chronic health parameters [26], and

level below which transfusion is associated with

others [27,28]), ICU admission diagnoses categories

reduced hospital mortality (or transfusion threshold)

(adapted from the Healthcare Cost and Utilization

across key cardiac conditions, such as comorbid heart

Project) (29), and selected blood test values. Comor-

disease, AMI, unstable angina, and CHF. We hypo-

bid conditions were assigned if their codes were

thesized that transfusion thresholds with these

stated at least once in administrative records during

conditions are higher than those advocated in the

the previous 2 years, but not where there was an

restrictive transfusion strategy. Ultimately, our goal

overlapping admission diagnosis (27). ICU admission

was to assemble additional evidence to assist physi-

diagnoses were defined by the first ICD-9-CM code

conditions

(e.g.,

Elixhauser

comorbidity

cians in deciding when to transfuse critically ill pa-

from ICU files. Chronic kidney disease was defined as

tients with cardiac disease while we await the results

a glomerular filtration rate <60 ml/min/1.73 m 2 for at

of definitive clinical trials.

least 3 months (30).

Ding et al.

2512

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

STATISTICAL ANALYSES. We used principal com-

T A B L E 1 Patient Characteristics

ponent analyses to create a smaller number of Died in Hospital

comorbidity groupings. This is a data reduction method that addresses correlation between a set of

All (N ¼ 258,826)

Yes (n ¼ 30,086)

No (n ¼ 228,740)

11.4  2.6

9.8  2.4

11.7  2.6

66.1  12.5

70.8  11.7

65.5  12.5

251,596 (97.2)

29,552 (98.2)

222,044 (97.1)

White

162,262 (62.7)

18,644 (62.0)

143,618 (62.8)

regression was then performed to build explanatory

Black

43,356 (16.8)

5,205 (17.3)

38,151 (16.7)

models for hospital mortality. Blood transfusion

Others and unknown

53,208 (20.5)

6,237 (20.7)

46,971 (20.5)

Explanatory Variables

Mean hemoglobin level, g/dl Demographic features Mean age, yrs Male

artificial variables or “principal components” that can be used in further analyses. Details of the principal components and their corresponding comorbid conditions are provided in Online Table 1. Logistic

Race

ICU admission-related features Direct admission

observed variables, and it develops a shorter list of

197,336 (76.2)

16,171 (53.7)

4.5  6.8

7.8  13.8

4.1  5.0

57,845 (22.3)

8,864 (29.5)

48,981 (21.4)

56,555 (21.9)

10,090 (33.5)

46,465 (20.3)

on the relationship between transfusion and hospital

90,568 (35.0)

12,553 (41.7)

78,015 (34.1)

mortality, we included corresponding interaction

Neurological disease

15,844 (6.1)

2,657 (8.8)

13,187 (5.8)

variables. To address clustering of admissions within

Respiratory disease

47,607 (18.4)

7,744 (25.7)

39,863 (17.4)

Liver disease

10,876 (4.2)

2,058 (6.8)

8,816 (3.9)

fiscal years, we employed fixed effects for year as

Psychiatric disease

45,386 (17.5)

5,541 (18.4)

Mean ICU length of stay, days Chronic kidney disease

181,165 (79.2)

during the first 30 days of ICU stay was the treatment variable. Other aforementioned explanatory vari-

Heart disease risk factors

effect modification of Hgb, comorbid heart disease, and ICU admission for AMI, unstable angina, and CHF

Comorbidity groupings Heart disease

ables, including Hgb, were included. To examine the

Coagulopathy

8,389 (3.2)

Renal disease

39,845 (17.4)

dummy variables. To understand how absolute hospital mortality rates with and without transfusion

2,083 (6.9)

6,306 (2.8)

15,666 (6.1)

3,023 (10.0)

12,643 (5.5)

varied with Hgb, we computed model-predicted

Anemia

34,261 (13.2)

5,998 (19.9)

28,263 (12.4)

mortality rates at different Hgb values from 6 to 11

Cancer

24,294 (9.4)

5,193 (17.3)

19,101 (8.4)

g/dl. Finally, we conducted sensitivity analyses by

Peptic ulcer disease

5,272 (2.0)

763 (2.5)

4,509 (2.0)

including selected blood tests, namely blood urea

Endocrine disease

9,134 (3.5)

1,443 (4.8)

7,691 (3.4)

nitrogen, blood total white blood cell, serum albumin,

Rheumatological disease

2,984 (1.2)

505 (1.7)

2,479 (1.1)

Angina/unstable angina

7,555 (2.9)

81 (0.3)

7,474 (3.3)

Congestive heart failure

14,105 (5.4)

1,533 (5.1)

12,572 (5.5)

these investigations. SAS version 9.3 (SAS Institute, Cary, North Carolina) was used for all data analyses.

ICU admission diagnoses

Arrhythmia

528 (1.8)

9,465 (4.1)

Acute myocardial infarction

27,770 (10.7)

2,460 (8.2)

25,310 (11.1)

Other cardiovascular diseases

37,733 (14.6)

1,291 (4.3)

36,442 (15.9)

Peripheral vascular disease Valvular heart disease Gastrointestinal bleeding Other gastrointestinal disease

9,993 (3.9)

997 (0.4)

152 (0.5)

and serum creatinine as explanatory variables for a subset of ICU episodes that had complete data on

RESULTS

845 (0.4)

971 (0.4)

91 (0.3)

880 (0.4)

14,069 (5.4)

997 (3.3)

13,072 (5.7)

8,064 (3.1)

1,283 (4.3)

6,781 (3.0)

Among 258,826 ICU episodes, hospital death occurred in 30,086 patients (11.6%). Transfusion was docu-

Gastrointestinal neoplasm

1,514 (0.6)

422 (1.4)

1,092 (0.5)

mented in 32,097 (12.4%) episodes during the first

Serious neurological disease

8,722 (3.4)

1,650 (5.5)

7,072 (3.1)

30 days of ICU admission. Patient characteristics for

568 (0.2)

these episodes are shown in Table 1. For 182,792 ICU

Minor neurological disease

598 (0.2)

30 (0.1)

72,821 (28.1)

6,884 (22.9)

65,937 (28.8)

episodes with complete information on selected

Orthopedic, nonsurgically treated

2,387 (0.9)

320 (1.1)

2,067 (0.9)

blood tests, hospital death occurred in 13%, with

Renal disease

6,070 (2.3)

1,083 (3.6)

4,987 (2.2)

Lung neoplasm

1,940 (0.7)

819 (2.7)

1,121 (0.5)

transfusion received by 14% during the first 30 days

Respiratory arrest/failure

9,881 (3.8)

3,650 (12.1)

6,231 (2.7)

10,702 (4.1)

2,359 (7.8)

8,343 (3.6)

8,685 (3.4)

925 (3.1)

7,760 (3.4)

pected, patients who died in hospital were older, had

Other medical diagnoses

Pneumonia Chronic obstructive pulmonary disease

of ICU stay. There were no major differences from the study population as a whole (Online Table 2). As ex-

Other respiratory disease

4,342 (1.7)

670 (2.2)

3,672 (1.6)

lower Hgb, were more likely to have heart disease and

Infection

9,907 (3.8)

2,858 (9.5)

7,049 (3.1)

most other comorbidities, and stayed longer in the

32,097 (12.4)

6,526 (21.7)

25,571 (11.2)

ICU. Notably, those who died were almost twice as

Blood transfusion during first 30 days of ICU admission Values are mean  SD or n (%). ICU ¼ intensive care unit.

likely to have received transfusion during the first 30 days of ICU admission. Details of logistic regression analyses for hospital mortality are provided in Online Table 3. The most important results are those that show the interactions of blood transfusion with Hgb and cardiac disease.

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

For the whole group, the odds ratio (OR) for the

was the critical point for reversal of the risk–benefit

interaction between blood transfusion and Hgb was

balance.

1.22 (95% confidence interval [CI]: 1.20 to 1.25). This

Beyond obtaining point estimates of the transfusion

indicates that for each 1 g/dl increase in Hgb, trans-

threshold, having a sense of the uncertainty of these

fusion was associated with a 22% increase in odds of

estimates would be helpful. The broken line repre-

hospital mortality. Lower hospital mortality was

senting the lower limit of the 95% CI for the transfusion

associated with transfusion at lower Hgb levels. Cor-

group (blue line) intersects with that of the corre-

responding ORs for interactions with comorbid heart

sponding upper limit for the no transfusion group

disease and ICU admission diagnosis of AMI were 0.82

(orange lines) at Hgb levels of approximately 0.7 to

(95% CI: 0.76 to 0.88) and 0.78 (95% CI: 0.69 to 0.88),

0.9 g/dl above the transition point across all graphs.

respectively. The interpretation was that transfusion

Correspondingly, the broken line representing the

was associated with lower hospital mortality when

upper limit of the 95% CI for the transfusion group

these 2 conditions were present. In other words, for

(blue lines) and that of the corresponding lower limit

any given Hgb, the benefit associated with trans-

for the no transfusion group (orange lines) intersect at

fusion was greater when patients had comorbid heart

Hgb levels approximately 0.7 to 0.9 g/dl below the

disease or were admitted to the ICU for AMI. The in-

transition point. These values represent worst-case

teractions for transfusion with ICU admission for

scenarios with respect to uncertainty around our

unstable angina and CHF were not significant.

estimates of the transition point. In other words, the

This suggests no differences in benefit with trans-

true transition point is very likely within the range of

fusion whether or not these 2 conditions were pre-

0.7 to 0.9 g/dl below and above our point estimates.

sent. The model c-statistic of 0.81 indicates good model discrimination.

Sensitivity analyses using additional blood test information on a smaller subset of patients yielded

For model-predicted hospital mortality rates with

interaction effects in the same direction. However,

and without transfusion at different Hgb levels from 6

the interaction of transfusion with ICU admission for

to 11 g/dl for 3 key permutations of comorbidity and

AMI was not significant (third column of Online

ICU admitting diagnosis, Figures 1A to 1C (unbroken

Table 3). Most importantly, the corresponding trans-

blue and orange lines) represent point estimates of

fusion thresholds were approximately 1 g/dl lower

predicted probability of hospital mortality with and

than those for the whole group, as seen in Online

without transfusion, respectively. The corresponding

Figure 1. We also conducted sensitivity analyses

broken lines represent the 95% CIs for those point

with robust standard error estimation to account for

estimates. Confidence bands around these point es-

clustering by facility. However, this made no differ-

timates for mortality range from approximately 0.02

ence in the results of estimated effects or statistical

to 0.06. As expected, mortality decreased with higher

significance. Finally, we repeated the regression

Hgb for both transfused and nontransfused patients.

models, but with Hgb specified as 1 g/dl categories

However, this decrease was predicted to be steeper

rather than as a continuous variable for both the main

for those who were not transfused. In patients

effect and its interaction with transfusion receipt.

without comorbid heart disease and who were

The parameter estimates on the logit of hospital

admitted to the ICU for a set of reference noncardiac

mortality varied in an approximately linear fashion

diagnoses, mortality was lower with transfusion

across Hgb categories, thereby providing justification

compared with no transfusion below an Hgb of

for specifying Hgb as a continuous variable. Detailed

approximately 7.7 g/dl, but mortality was higher

results are provided in Online Tables 4 and 5, as well

above that cutpoint. We can infer that the critical

as Online Figure 2.

point at which the risk–benefit balance of transfusion reversed was at an Hgb in the region of 7.7 g/dl.

DISCUSSION

However, for those with comorbid heart disease admitted to the ICU for these noncardiac diagnoses,

This is the largest observational study of critically ill

mortality was lower with transfusions below an Hgb

patients to examine the association of blood trans-

of approximately 8.7 g/dl, but mortality was higher

fusions with mortality and effect modification by Hgb

above this level. This indicates that the correspond-

level and cardiac disease. Our premise is that Hgb

ing critical point was in this region of Hgb level.

levels exist below which the transfusion risk–benefit

Finally, for those with comorbid heart disease who

balance reverses (31). Combining our primary results

were admitted to the ICU for AMI, mortality was

and those of sensitivity analyses, we infer that the

lower with transfusion below an Hgb of approxi-

Hgb level below which transfusion was associated

mately 10 g/dl, but was higher above this level, which

with

reduced

hospital

mortality

or

transfusion

2513

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

F I G U R E 1 Model-Predicted Probability of Hospital Mortality

B

0.3 0.2

Probability of Hospital Death

0.0

0.1

0.3 0.2 0.1

Probability of Hospital Death

0.4

0.4

A

0.0 6

7

10

8 9 Hgb (g/dL)

6

11

7

9 8 Hgb (g/dL)

10

11

0.3 0.2 0.1

Probability of Hospital Death

0.4

C

0.0

2514

6

7

9 8 Hgb (g/dL)

10

11

Hospital mortality with and without transfusion across nadir hemoglobin (Hgb) levels for the whole study population of Veterans Affairs intensive care unit (ICU) episodes (n ¼ 258,826) was analyzed for a 60-year-old man transferred to the ICU and who stayed for 30 days in fiscal year 2005 in the setting of (A) no comorbid heart disease admitted to ICU for reference medical diagnoses, (B) comorbid heart disease admitted to ICU for reference medical diagnoses, or (C) comorbid heart disease admitted to ICU for acute myocardial infarction. Orange lines, no transfusion; blue lines, transfusion; unbroken lines, point estimates; broken lines, 95% confidence limits.

threshold is approximately 7 to 8 g/dl across a het-

findings confers added confidence with respect to

erogeneous population of VA ICU patients. Interest-

our extended analyses of subgroups with cardiac

ingly, these results approximate those of TRICC and

disease.

are consistent with a Cochrane Review based on a

It was recognized quite early that critically ill pa-

meta-analysis of 19 RCTs that confirmed the associa-

tients with cardiovascular disease may require a

tion of restrictive transfusion strategies with reduced

different transfusion strategy (33–35). Subsequent

hospital mortality (32). When we consider the results

analyses of these patients in TRICC revealed that

of TRICC as offering true estimates of the treatment

among those with severe ischemic heart disease,

effect of transfusions, then the convergence of

the restrictive transfusion group had nonsignificantly

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

CENTRAL I LLU ST RAT ION Varying Red Cell Transfusion Thresholds: Depending on Cardiac Disease for Veterans Affairs’ Intensive Care Unit Episodes

Ding, Y.Y. et al. J Am Coll Cardiol. 2015; 66(22):2510–8.

Nadir hemoglobin level below which red cell transfusion is associated with lower mortality when (A) comorbid heart disease (8 to 9 g/dl) and (B) acute myocardial infarction (9 to 10 g/dl) are present.

higher mortality than the liberal transfusion group

participants may not be truly representative of the

(13). In a retrospective study of elderly Medicare pa-

population of interest. Inadequate statistical hand-

tients with AMI, transfusion was associated with

ling may occur, too, where transfusion operates as a

significantly lower 30-day mortality if the hematocrit

determinant of predictors of outcomes when not

at admission was #33% (Hgb 11 g/dl) (14). In a recent

randomized (38).

pilot RCT of patients with ACS or stable angina who

In our study, having comorbid heart disease raised

underwent cardiac catheterization, 30-day mortality

the transfusion threshold by approximately 1 g/dl to

was lower for the liberal transfusion group compared

approximately 8 to 9 g/dl (Central Illustration). In

with

(19,36).

other words, the mortality benefit associated with

Together, these findings suggest that patients with

transfusion was accrued at slightly higher Hgb levels

cardiac disease could benefit from a more liberal

in the presence of heart disease. Here, our findings

transfusion strategy.

are consistent with the recommendations of the

the

restrictive

transfusion

groups

In contrast, other studies suggest that a restrictive

AABB

(American

Association

of

Blood

Banks).

transfusion strategy might be beneficial. These

Beyond that, ICU admission for AMI raises this

include a higher predicted 30-day mortality with

threshold by a further 1 g/dl to approximately 9 to

transfusion when the nadir Hgb level was >8 g/dl or

10 g/dl (Central Illustration). Together, our results

there was an equivalent hematocrit value in the

provide additional evidence, albeit observational in

setting of ACS (15,16,37), and higher hospital mor-

nature, to support transfusing critically ill patients

tality with transfusion when Hgb was >7 g/dl in

with these cardiac conditions at higher Hgb levels. It

the presence of cardiovascular comorbidities (12).

is plausible that increased sensitivity to the adverse

The reasons for contrasting effects on mortality

consequences of reduced oxygen delivery with ane-

in different studies are probably methodological,

mia exists in the presence of cardiac disease (23).

including different trial designs and selection issues

Anemia may exacerbate myocardial ischemia and

with

activate the sympathetic nervous system (39,40).

observational

studies

and

RCTs

where

2515

2516

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

Furthermore, the ability to increase cardiac output

specificity for transfusion identification from blood

to compensate for reduced oxygen delivery with

bank records using relevant ICD-9-CM procedure

anemia may be reduced with cardiac disease (41).

codes (49). Third, only approximately 3% of our pa-

The impact of these effects may vary across the

tients were women. Although there were >7,000 fe-

spectrum of cardiac disease, with AMI being at its

male ICU admissions, and sex adjustment was made

most severe end. This may explain why transfusions

in regression analyses, we remained cautious on

are associated with reduced mortality at higher Hgb

extrapolating our results to female patients. As such,

levels as we traverse across the continuum from no

our findings relate largely to critically ill male pa-

cardiac disease to comorbid heart disease, and finally

tients who have cardiac disease. Finally, data cur-

to AMI. Then again, our findings do not indicate that

rency is a potential concern because we used data

these levels should be modified for unstable angina

from 2001 to 2005. However, it is debatable as to

or CHF. However, a national heart failure survey

whether transfusion and ICU practice have changed

found a nonsignificant trend toward lower 30-day

enough over the past decade to expect important

mortality with transfusion among propensity score-

differences in our results if we had used more recent

matched patients (20).

VA data. Although possible, we are of the opinion

Bedside decisions on transfusion attempt to bal-

that such differences are less likely to be operant.

ance achieving greater oxygen carriage for myocardial

Despite these limitations, we believe that this study

support and minimizing risk of transfusion-related

contributes to the ongoing debate on when critically

adverse

injury

ill patients with anemia and cardiac disease should

(42–47) and secondary bacterial infection (48). It is

be transfused. Building on previous work such as the

plausible that specific comorbid conditions and acute

recent study on transfusion in the setting of AMI by

illnesses could shift this balance toward transfusing

Salisbury et al. (22), our findings suggest that the

more liberally. In this respect, we have demonstrated

restrictive transfusion strategy may not be optimal

the varying effect of transfusion on hospital mortality

for patients with selected cardiac conditions. Rather,

across a range of Hgb levels and the effect modifica-

they

tion by cardiac conditions. The situation becomes

Hgb levels 1 to 2 g/dl higher when comorbid heart

more complicated with multiple concurrent illnesses

disease or AMI occur, but not when unstable angina

as commonly encountered in critically ill patients.

or CHF do.

outcomes,

including

acute

lung

Suggested transfusion thresholds, at best, serve as a general guide, and physicians should look beyond them for more complex patients with advanced age, multimorbidities, or frailty in addition to their cardiac illnesses. For them, transfusion decisions are best guided by considered clinical judgment at the bedside. STUDY LIMITATIONS. First, as emphasized, this was

an observational study. Selection bias from unaccounted factors influencing transfusion receipt may exist despite our best efforts to include a wide range of covariates in our risk-adjustment model. Particularly, we did not have data on some aspects of illness

support

setting

transfusion

thresholds

at

CONCLUSIONS RBC transfusion during ICU admission is associated with reduced hospital mortality when pre-transfusion Hgb levels were <8 to 9 g/dl in the setting of comorbid heart disease. The transfusion threshold is raised to 9 to 10 g/dl if the ICU admission is for AMI, but not for patients with unstable angina or CHF. Although we are waiting for more definitive evidence from randomized clinical trials, these findings offer possible interim guidance for physicians on transfusing ICU patients with cardiac disease.

severity, functional status, and do-not-resuscitate

ACKNOWLEDGMENT The Department of Veterans

orders. As shown, sensitivity analyses using addi-

Affairs provided space and computer support.

tional test information suggest that the Hgb level at which the risk–benefit associated with transfusion

REPRINT REQUESTS AND CORRESPONDENCE: Dr.

reverses could be approximately 1 g/dl lower. With

Dan R. Berlowitz, VA Bedford Hospital, Center for

this information, it is quite likely that the true

Healthcare Organization and Implementation Re-

transfusion threshold lies between these 2 sets of

search (CHOIR), Edith Nourse Rogers Memorial Hos-

estimates. In line with this possibility, we adjusted

pital, (152), Building 70, 200 Springs Road, Bedford,

our study conclusions accordingly. Second, we did

Massachusetts 01730. E-mail: [email protected].

not have VA data on the accuracy of ICD-9-CM codes

OR Dr. Yew Y. Ding, Department of Geriatric Medicine &

for transfusion. External evidence on validity of the

Institute of Geriatrics and Active Ageing, Tan Tock

billing for blood transfusion in a non-VA tertiary care

Seng Hospital, 11 Jalan Tan Tock Seng, Singapore 308433.

setting hospital indicated favorable sensitivity and

E-mail: [email protected].

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

PERSPECTIVES COMPETENCY IN PATIENT CARE AND PROCE-

TRANSLATIONAL OUTLOOK: Randomized trials

DURAL SKILLS: Although transfusion decisions should

are necessary to determine the Hgb thresholds for

be individualized based on clinical assessment, critically ill

transfusion associated with survival benefit in criti-

patients with comorbid heart disease may benefit from

cally ill patients with heart disease, including those

Hgb levels >8 to 9 g/dl, and patients with ACS may need

with acute coronary syndromes or decompensated

levels >9 to 10 g/dl.

heart failure.

REFERENCES 1. Reiles E, Van der Linden P. Transfusion trigger in critically ill patients: has the puzzle been completed? Crit Care 2007;11:142.

anemia: is there a benefit? Intensive Care Med 2013;39:445–53.

2. Hébert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial

transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001; 29:227–34.

of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999;340:409–17. 3. Hébert PC, Wells G, Martin C, et al. Variation in red cell transfusion practice in the intensive care unit: a multicentre cohort study. Crit Care 1999;3: 57–63. 4. Vincent JL, Baron JF, Reinhart K, et al. Anemia and blood transfusion in critically ill patients. JAMA 2002;288:1499–507. 5. Corwin HL, Gettinger A, Pearl RG, et al. The CRIT study: anemia and blood transfusion in the critically ill–current clinical practice in the United States. Crit Care Med 2004;32:39–52. 6. Cohen J, Kagan I, Hershcovici R, et al. Red blood cell transfusions–are we narrowing the evidence-practice gap? An observational study in 5 Israeli intensive care units. J Crit Care 2011;26: 106.e1–6. 7. Ding YY, Kader B, Christiansen CL, Berlowitz DR. Patient factors associated with transfusion practices in Veterans Affairs intensive care units: implications for further research. J Crit Care 2011;26: 431.e1–9. 8. Vincent JL, Sakr Y, Sprung C, et al., for the Investigators SOiAIPS. Are blood transfusions associated with greater mortality rates? Results of the Sepsis Occurrence in Acutely Ill Patients study. Anesthesiology 2008;108:31–9.

13. Hébert PC, Yetisir E, Martin C, et al. Is a low

14. Wu WC, Rathore SS, Wang Y, et al. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med 2001;345:1230–6. 15. Aronson D, Dann EJ, Bonstein L, et al. Impact of red blood cell transfusion on clinical outcomes in patients with acute myocardial infarction. Am J Cardiol 2008;102:115–9. 16. Alexander KP, Chen AY, Wang TY, et al. Transfusion practice and outcomes in non–STsegment elevation acute coronary syndromes. Am Heart J 2010;155:1047–53. 17. Shishehbor MH, Madhwal S, Rajagopal V, et al. Impact of blood transfusion on short-term and long-term mortality in patients with ST-segment elevation myocardial infarction. J Am Coll Cardiol Intv 2009;2:46–53. 18. Cooper HA, Rao SV, Greenberg MD, et al. Conservative versus liberal red cell transfusion in acute myocardial infarction (the CRIT Randomized Pilot Study). Am J Cardiol 2011;108:1108–11. 19. Carson JL, Brooks MM, Abbott JD, et al. Liberal versus restrictive transfusion thresholds for patients with symptomatic coronary artery disease. Am Heart J 2013;165:964–71.e1. 20. Garty M, Cohen E, Zuchenko A, et al. Blood transfusion for acute decompensated heart failure–friend or foe? Am Heart J 2009;1584: 653–8.

9. Engoren M, Arslanian-Engoren C. Long-term survival in the intensive care unit after erythrocyte blood transfusion. Am J Crit Care 2009;18:124–31; quiz 32.

21. Carson JL, Grossman BJ, Kleinman S, et al. Red blood cell transfusion: a clinical practice guideline from the AABB. Ann Intern Med 2012;157:49–58.

10. Sakr Y, Lobo S, Knuepfer S, et al. Anemia and blood transfusion in a surgical intensive care unit. Crit Care 2010;14:R92.

transfusion during acute myocardial infarction: association with mortality and variability across hospitals. J Am Coll Cardiol 2014;64:811–9.

11. Sheth KN, Gilson AJ, Chang Y, et al. Packed red blood cell transfusion and decreased mortality in intracerebral hemorrhage. Neurosurgery 2011;68: 1286–92.

23. Yeh RW, Wimmer NJ. Blood transfusion in myocardial infarction: opening old wounds for comparative-effectiveness research. J Am Coll Cardiol 2014;64:820–2.

12. Leal-Noval SR, Muñoz-Gómez M, JiménezSánchez M, et al. Red blood cell transfusion in non-bleeding critically ill patients with moderate

24. Hébert PC CJ. Transfusion threshold of 7 g per deciliter—the new normal. N Engl J Med 2014;371: 1459–61.

22. Salisbury AC, Reid KJ, Marso SP, et al. Blood

25. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27. 26. Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system. Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:1619–36. 27. Johnston JA, Wagner DP, Timmons S, et al. Impact of different measures of comorbid disease on predicted mortality of intensive care unit patients. Med Care 2002;40:929–40. 28. Render ML, Kim HM, Welsh DE, et al. Automated intensive care unit risk adjustment: results from a National Veterans Affairs study. Crit Care Med 2003;31:1638–46. 29. Healthcare Cost and Utilization Project (HCUP) Clinical Classifications Software (CCS) for ICD-9CM. Rockville, MD: Agency for Healthcare Quality and Research. Available at: https://www.hcup-us. ahrq.gov/toolssoftware/ccs/ccs.jsp. Accessed May 2, 2015. 30. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39 2 Suppl 1:S1–266. 31. Goodnough LT, Levy JH, Murphy MF. Concepts of blood transfusion in adults. Lancet 2013;381: 1845–54. 32. Carson JL, Carless PA, Hebert PC. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 2012;4:CD002042. 33. Carson JL, Hill S, Carless P, Hébert P, Henry D. Transfusion triggers: a systematic review of the literature. Transfus Med Rev 2002;16:187–99. 34. Hill SR, Carless PA, Henry DA, et al. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev 2002:CD002042. 35. Fakhry SM, Fata P. How low is too low? Cardiac risks with anemia. Crit Care 2004;8 Suppl 2:S11–4. 36. Sardar P, Nairooz R, Dutu L, et al. Liberal versus restrictive transfusion strategy for patients with coronary artery disease. Am Heart J 2013;166:e25. 37. Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA 2004;292:1555–62.

2517

2518

Ding et al.

JACC VOL. 66, NO. 22, 2015 DECEMBER 8, 2015:2510–8

Transfusion of ICU Patients With Cardiac Disease

38. Alfonso F. Blood transfusion after myocardial infarction: friend, foe, or double-edged sword? J Am Coll Cardiol Intv 2009;2:633–5. 39. Graham LN, Smith PA, Stoker JB, et al. Sympathetic neural hyperactivity and its normalization following unstable angina and acute myocardial infarction. Clin Sci (Lond) 2004;106:605–11. 40. Nikolsky E, Aymong ED, Halkin A, et al. Impact of anemia in patients with acute myocardial infarction undergoing primary percutaneous coronary intervention: analysis from the Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) Trial. J Am Coll Cardiol 2004;44:547–53.

prospective nested case-control study. Am J Respir Crit Care Med 2007;176:886–91.

with emphasis on the critically ill. Br J Haematol 2009;147:431–43.

43. Marik PE, Corwin HL. Efficacy of red blood cell transfusion in the critically ill: a systematic

48. Juffermans NP, Prins DJ, Vlaar AP, et al. Transfusion-related risk of secondary bacterial infections in sepsis patients: a retrospective cohort study. Shock 2011;35:355–9.

review of the literature. Crit Care Med 2008;36: 2667–74. 44. Triulzi DJ. Transfusion-related acute lung injury: current concepts for the clinician. Anesth Analg 2009;108:770–6. 45. van Stein D, Beckers EA, Sintnicolaas K, et al. Transfusion-related acute lung injury reports in the Netherlands: an observational study. Transfusion 2010;50:213–20.

ill cardiac patients. Ann Intensive Care 2014;4:16.

46. Vlaar AP, Schultz MJ, Juffermans NP. Transfusion-related acute lung injury: a change of perspective. Neth J Med 2009;67:320–6.

42. Gajic O, Rana R, Winters JL, et al. Transfusionrelated acute lung injury in the critically ill:

47. Benson AB, Moss M, Silliman CC. Transfusionrelated acute lung injury (TRALI): a clinical review

41. Du Pont-Thibodeau G, Harrington K, Lacroix J. Anemia and red blood cell transfusion in critically

49. Segal JB, Ness PM, Powe NR. Validating billing data for RBC transfusions: a brief report. Transfusion 2001;41:530–3.

KEY WORDS acute coronary syndrome, acute myocardial infarction, heart failure, risk adjustment

A PPE NDI X For supplemental tables and figures, please see the online version of this article.