Outcomes of Protocol-Driven Care of Critically Ill Severely Anemic Patients for Whom Blood Transfusion Is Not an Option* Aryeh Shander, MD; Mazyar Javidroozi, MD, PhD; Carmine Gianatiempo, MD; Nisha Gandhi, MD; John Lui, MD; Frank Califano, MD; Margit Kaufman, MD; Sajjad Naqvi, MD; Faraz Syed, MD; Oshuare Aregbeyen, MD
Objective: To compare the outcomes of severely anemic critically ill patients for whom transfusion is not an option (“bloodless” patients) with transfused patients. Design: Cohort study with propensity score matching. Setting: ICU of a referral center. Patients: One hundred seventy-eight bloodless and 441 transfused consecutive severely anemic, critically ill patients, admitted between May 1996 and April 2011, and having at least one hemoglobin level less than or equal to 8 g/dL within 24 hours of ICU admission. Patients with diagnosis of brain injury, acute myocardial infarction, or status postcardiac surgery were excluded. Interventions: Allogeneic RBC transfusion during ICU stay. *See also p. 1238. All authors: Department of Anesthesiology and Critical Care Medicine, Englewood Hospital and Medical Center, Englewood, NJ. This work was performed at Englewood Hospital and Medical Center. All authors contributed equally to the design of the study. Drs. Shander, Gianatiempo, Gandhi, Lui, Califano, Kaufman, and Syed were responsible for management of the ICU patients during the study period. Drs. Naqvi and Aregbeyen were primarily responsible for data collection and verification. All authors had unrestricted access to the study data. Statistical analysis was performed by Dr. Javidroozi. All authors contributed to preparing the article and reviewed it prior to submission. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ ccmjournal). Dr. Shander received funding from Bayer, Luitpold, and Masimo; has been a consultant or speaker with honorarium for or received research support from Bayer, Luitpold, Masimo, Novartis, Novo Nordisk, OrthoBiotech, Pfizer, Masimo, and Zymogenetics; and is a founding member of the Society for the Advancement of Blood Management (SABM). Dr. Javidroozi been a consultant and contractor for and received funding from SABM and Gauss Surgical. Dr. Naqvi disclosed work at Englewood Hospital and Medical Center as a TeamHealth employee. Dr. Aregbeyen is employed at Englewood Hospital and Medical Center, New Jersey. The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: 10.1097/CCM.0000000000001599
Critical Care Medicine
Measurements and Main Results: Primary outcome was in-hospital mortality. Other outcomes were ICU mortality, readmission to ICU, new electrocardiographic or cardiac enzyme changes suggestive of cardiac ischemia or injury, and new positive blood culture result. Transfused patients were older, had higher hemoglobin level at admission, and had higher Acute Physiology and Chronic Health Evaluation II score. Hospital mortality rates were 24.7% in bloodless and 24.5% in transfused patients (odds ratio, 1.01; 95% CI, 0.68–1.52; p = 0.95). Adjusted odds ratio of hospital mortality was 1.52 (95% CI, 0.95–2.43; p = 0.08). No significant difference in ICU readmission or positive blood culture results was observed. Analysis of propensity score–matched cohorts provided similar results. Conclusions: Overall risk of mortality in severely anemic critically ill bloodless patients appeared to be comparable with transfused patients, albeit the latter group had older age and higher Acute Physiology and Chronic Health Evaluation II score. Use of a protocol to manage anemia in these patients in a center with established patient blood management and bloodless medicine and surgery programs is feasible and likely to contribute to improved outcome, whereas more studies are needed to better delineate the impact of such programs. (Crit Care Med 2016; 44:1109–1115) Key Words: anemia; bloodless medical and surgical procedures; blood transfusion; critical illness; Jehovah’s Witnesses
A
llogeneic RBC transfusion is commonly used in anemia to avoid tissue hypoxia and its consequences. Despite significant risks, costs, and challenges to determine reliably the patients in whom its benefits are likely to outweigh the risks (1), millions of units of allogeneic blood are transfused annually (2). Certain patient populations, namely the critically ill, are particularly at risk (3). Caring for anemic patients for whom blood transfusions are not an option (called “bloodless” here) has been a challenge. Refusal of transfusion does not equate refusal of medical care, and these patients can still benefit from treatments to www.ccmjournal.org
1109
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
augment their intrinsic hematopoietic capabilities, minimize blood loss, and improve physiologic response to anemia (4). These patients may also provide an opportunity to study the association between transfusion and outcomes since the decision not to transfuse is not dictated by their clinical status. We compared the outcomes of severely anemic critically ill “bloodless” patients with transfused patients at a referral center with programs in place to care for these patients. We hypothesized that the outcomes of bloodless patients would be worse than their transfused counterparts.
MATERIALS AND METHODS The study was approved by the local hospital institutional review board, and individual patient consent was waived. Consecutive patients admitted to a medical/surgical ICU between May 1996 and April 2011, in a hospital with established patient blood management (PBM) and bloodless medicine and surgery (BMS) were screened retrospectively. The main eligibility criterion was severe anemia at ICU admission defined as at least one hemoglobin level less than or equal to 8 g/dL from a blood sample taken within 24 hours of ICU admission. Exclusion criteria were age less than 18 years old; death within 24 hours of admission; treatment with artificial hemoglobin-based oxygen carriers; admission with diagnosis of traumatic brain injury or acute myocardial infarction; and status postcardiac surgery and pregnancy. For patients with multiple hospital admissions, only the first admission leading to ICU admission was considered. Patients were grouped into two cohorts: bloodless patients who did not receive any allogeneic blood transfusions; and transfused patients, who received allogeneic RBC transfusions during ICU stay. Patients who are unwilling to accept blood transfusions are managed under our BMS program. An advanced directive is obtained as early as permissible to identify what blood components, treatments, and procedures are acceptable to the patients. Management of these patients (including those in this study) is guided by our bloodless protocol, which has been in place at our center since 1994 (Appendix, Supplemental Digital Content 1, http://links.lww.com/CCM/B666). The protocol has been periodically updated, but no significant changes were made during the study period. The primary study outcome was in-hospital mortality that was further stratified into ICU mortality and in-hospital mortality after ICU discharge. Other outcomes were readmission to ICU, new electrocardiographic changes suggestive of cardiac ischemia or injury, new-onset rise in cardiac enzymes, and new positive blood culture result. As a retrospective study, there were no systematic criteria to order any of these tests, and data recorded in the patient charts were used as the sole resource. Lack of performance of any specific test was considered to indicate that the test was not deemed to be indicated by the treating clinicians, and hence the results were assumed to be negative. Acute Physiology and Chronic Health Evaluation (APACHE) II score of the patients was calculated as previously described (5). Additionally, a modified APACHE II score (with 1110
www.ccmjournal.org
age and hematocrit subscores removed) was calculated and used in multivariate analyses when age and hematocrit were entered individually. Statistical Analysis Chi-square test, Student t test, and Mann-Whitney U test were performed as appropriate, and values are accordingly provided in forms of count followed by percentage (%), mean ± sd, and median/interquartile range (IQR). No missing data were encounter for key study variables. Logistic regression analysis was performed with various endpoints as dependent variables to adjust the odds ratio (OR) for other confounders. The 95% CI for ORs are provided. For all studied endpoints, OR values above 1 are indicative of favorable outcomes for transfused versus bloodless patients. Statistical analysis was performed using SPSS (SPSS for Windows, Version 13.0, Chicago, IL). Using the patients’ age, gender, postsurgical status, lowest hemoglobin level within first 24 hours of ICU admission, and APACHE II score, the propensity of being transfused was calculated. Bloodless patients were matched to transfused patients in 1:2 ratio based on propensity score using the nearest neighbor matching within a caliper distance of 0.05 without replacement (6). During the period that the patients were admitted to our ICU, a number of transfusion guidelines were published with a trend toward lowering the hemoglobin threshold for transfusion (7). To account for the potential impact of lower hemoglobin transfusion criteria, we performed two sets of subgroup analysis, excluding patients with nadir ICU hemoglobin levels above 7 and above 6 g/dL. Availability of severely anemic bloodless patients was a limiting factor in our study, and following initial sample size estimations, we resorted to a convenience sampling of 119 bloodless patients, matched 1:2 to 238 transfused patients for the matched analysis. For comparison and assuming a mortality rate of 15%, if there was truly no difference between the groups, 238 patients were required to be 80% sure that a two-sided 95% CI would exclude a mortality rate difference in favor of the transfused group of more than 13%.
RESULTS Study population consisted of 619 severely anemic patients admitted to the ICU, 28.8% of which were bloodless. Mean age was 65.05 ± 16.91 years old and 59% were female. Leading ICU admitting diagnoses were gastrointestinal conditions (42.2%), pulmonary conditions (12.8%), sepsis/septic shock (6.6%), postvascular surgery (5.8%), and gynecologic conditions (4.7%). Approximately 36% of the patients were admitted to ICU following a surgical procedure during the same hospital admission (status “postsurgical”). Transfused patients (n = 441) received median/IQR of 2/2 units of RBC. The last measured hemoglobin level prior to first RBC transfusion in ICU was 6.76 ± 1.11 g/dL. Among the transfused, 51, 28, and 9 patients also received plasma (2/2 units), platelet (1/1 units), and cryoprecipitate (2/3 units), respectively. Table 1 provides the key characteristics June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations
Table 1.
Baseline Characteristics and Outcome Data of Bloodless and Transfused Patients Bloodless (n = 178)
Transfused (n = 441)
Univariate p
60.07 ± 16.57
67.06 ± 16.65
< 0.001
113 (63.5)
252 (57.1)
0.147
80 (44.9)
158 (35.8)
0.035
6.12 ± 1.85
7.07 ± 1.45
< 0.001
20.56 ± 7.71
24.31 ± 7.47
< 0.001
5.30 ± 1.65
6.44 ± 0.97
< 0.001
ICU stay until lowest hemoglobin level (d)
1/1
1/2
< 0.001
Length of ICU stay in survived patients (d)
4/5
3/5
0.171
6.20 ± 1.86
8.44 ± 1.42
< 0.001
ICU
28 (15.7)
72 (16.3)
0.855
Hospital post-ICU
16 (9.0)
36 (8.2)
0.738
Overall (ICU + hospital post-ICU)
44 (24.7)
108 (24.5)
0.952
Total (overall + subsequent admissions)
49 (27.5)
115 (26.1)
0.711
ICU readmission (%)
19 (10.7)
49 (11.1)
0.875
New electrocardiographic changes suggestive of ischemia or injury (%)
22 (12.4)
126 (28.6)
< 0.001
Cardiac enzyme rise (%)
18 (10.1)
136 (30.8)
< 0.001
New positive blood culture result (%)
16 (9.0)
56 (12.7)
0.193
Variables
Age (yr) Female gender (%) Postsurgical cases (%) Lowest hemoglobin level within first 24 hr in ICU (g/dL) Acute Physiology and Chronic Health Evaluation II score Lowest hemoglobin level in ICU (g/dL)
ICU discharge hemoglobin level (g/dL) Mortality (%)
Data are provided as mean ± sd, median/interquartile range, or n (%).
and outcome data. Transfused patients were older, had higher hemoglobin level at admission, and had higher APACHE II score (with significantly higher scores in temperature, heart rate, respiratory rate, oxygenation status, blood pH, WBC count and age components, and significantly lower scores in hematocrit and Glasgow Coma Scale components). One hundred sixty-four patients (26.5%) expired: 100 during ICU stay (16.2%); 52 during hospital stay after ICU discharge (8.4%); and the remaining during subsequent hospital admissions (Table 1). Among the 28 bloodless patients who died in ICU, the last hemoglobin level prior to death was 5.57 ± 1.38 g/dL, and it was measured within 1 or 2 days prior to death. Anemia was classified as the primary cause of death or a main contributing factor in 36.4% of these patients. Other leading causes of death included multiple comorbidities (36.4%), sepsis/septic shock (27.3%), advanced malignancies (13.6%), multiple organ failure (9.1%), respiratory failure (9.1%), and hemorrhagic shock (9.1%). Seven patients had been transferred from other hospitals. Six patients had “do-not-resuscitate” (DNR) order on file, including two who were DNR at admission to hospital. Unadjusted mortality rates according to the nadir hemoglobin levels during ICU stay for bloodless patients are presented in Figure 1. Unadjusted ORs of ICU and overall mortality were 0.957 (95% CI, 0.594–1.540; p = 0.855) and 1.012 (95% CI, 0.676–1.516; p = 0.952), respectively. Following adjustment for Critical Care Medicine
Figure 1. Mortality rates in bloodless patients according to the lowest hemoglobin (Hb) level during ICU stay. In each column, the lower (light gray) part represents mortality during ICU stay and the upper (dark gray) part represents mortality occurring during hospital stay out of ICU. Numbers in parentheses are the total number of cases in each category.
age, gender, postsurgical status, lowest hemoglobin level during first 24-hour of ICU admission, and modified APACHE II score, ORs of death in ICU and during overall hospital stay were 1.299 (95% CI, 0.756–2.234; p = 0.344) and 1.522 (95% CI, 0.955–2.426; p = 0.078), respectively. Older age, lower hemoglobin level, and higher modified APACHE II score were the significant independent predictors of mortality. www.ccmjournal.org
1111
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
Table 2. Comparison of Factors Used in Propensity Score Matching Between Bloodless and Transfused Cohorts Factors
Bloodless (n = 119)
Transfused (n = 238)
Univariate p
Age (yr)
63.54 ± 14.98
64.58 ± 17.36
0.556
Female gender (%)
73 (61.3)
144 (60.5)
0.878
Postsurgical cases (%)
49 (41.2)
99 (41.6)
0.939
6.87 ± 1.44
6.83 ± 1.56
0.785
Acute Physiology and Chronic Health Evaluation II score
22.31 ± 7.55
22.60 ± 7.42
0.733
Propensity score
0.298 ± 0.157
0.298 ± 0.158
0.997
Lowest hemoglobin level within first 24 hr in ICU (g/dL)
Data are provided as mean ± sd or n (%).
Unadjusted OR of readmission to ICU was 0.956 (95% CI, 0.546–1.675; p = 0.875). Results were similar when expired patients were excluded and when adjusted for other parameters mentioned earlier (data not shown). Unadjusted OR of new electrocardiographic changes suggestive of ischemia or injury was 0.353 (95% CI, 0.216–0.577; p < 0.001). Following adjustment for other parameters, OR of new electrocardiographic changes was 0.303 (OR, 0.177–0.520; p < 0.001). Other significant independent predictors were older age, male gender, and lower hemoglobin level. The independent association remained significant following exclusion of all 19 cases admitted to the ICU with cardiac diagnoses (OR, 0.293; 95% CI, 0.169–0.510; p < 0.001). For new-onset rise in cardiac enzymes, unadjusted and adjusted OR were 0.252 (95% CI, 0.149– 0.428; p < 0.001) and 0.324 (95% CI, 0.186–0.564; p < 0.001). Older age and higher modified APACHE II score were other significant independent predictors of new-onset raised cardiac enzyme. The independent association remained significant following exclusion of 19 cases admitted to the ICU with cardiac diagnoses (OR, 0.324; 95% CI, 0.183–0.573; p < 0.001).
Unadjusted OR of new positive blood culture result was 0.679 (95% CI, 0.378–1.219; p = 0.193). Following adjustment for other factors, OR of new positive blood culture result was 0.739 (95% CI, 0.397–1.374; p = 0.339). The factors used in propensity score matching are summarized in Table 2. Comparison of outcomes between matched bloodless and transfused cohorts are provided in Table 3. Figure 2 depicts the mortality rates in matched bloodless and transfused patients according to hemoglobin levels. Following exclusion of patients with nadir hemoglobin level of 7 g/dL or more, 26 of 142 bloodless (18.3%) and 54 of 297 transfused patients (18.2%) died in ICU (OR, 1.009; 95% CI, 0.601–1.693; p = 0.974). OR of ICU mortality following adjustment for other confounders was 1.652 (95% CI, 0.899– 3.036; p = 0.106). Following 1:1 propensity score matching (n = 98 per cohort), 19.4% of bloodless patients and 15.3% of transfused patients died in ICU (OR, 1.331; 95% CI, 0.633–2.800; p = 0.451). When patients with nadir hemoglobin level of 6 g/dL or more were excluded, 22 of 116 bloodless (19.0%) and 20 of
Table 3. Outcomes of Subsets of Bloodless and Transfused Patients Matched Based on the Propensity Score Bloodless (n = 119)
Transfused (n = 238)
OR (95% CI)
p
3/4
3/4
NA
0.353
ICU
17 (14.3)
33 (13.9)
1.035 (0.551–1.947)
0.914
Hospital post-ICU
12 (10.1)
17 (7.1)
1.458 (0.672–3.162)
0.338
Overall (ICU + hospital post-ICU)
29 (24.4)
50 (21.0)
1.212 (0.719–2.042)
0.471
Total (overall + subsequent admissions)
33 (27.7)
54 (22.7)
1.307 (0.791–2.162)
0.296
ICU readmission (%)
17 (13.2)
30 (12.6)
1.156 (0.609–2.192)
0.658
New electrocardiographic changes suggestive of ischemia or injury (%)
13 (10.9)
61 (25.6)
0.356 (0.187–0.678)
0.001
Cardiac enzyme rise (%)
14 (11.8)
65 (27.3)
0.355 (0.190–0.664)
0.001
New positive blood culture result (%)
10 (8.4)
22 (9.2)
0.901 (0.412–1.969)
0.793
Outcomes
Length of ICU stay in survived patients (d) Mortality (%)
OR = odds ratio, NA = not applicable. Data are provided as mean ± sd, median/interquartile range, or n (%).
1112
www.ccmjournal.org
June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations
Figure 2. Mortality rates in propensity score–matched transfused and bloodless patients according to the lowest hemoglobin (Hb) level within first 24 hr of ICU admission (A) and lowest Hb level during ICU stay (B). In each column, the lower (light gray) part represents mortality during ICU stay and the upper (dark gray) part represents mortality occurring during hospital stay out of ICU. Numbers in parentheses are the total number of cases in each category. Star symbol indicates p < 0.5 comparing total mortality rate between the bloodless and transfused patients.
111 transfused patients (18.0%) died in ICU (OR, 1.065; 95% CI, 0.545–2.082; p = 0.854). OR of ICU mortality following adjustment for other confounders was 1.769 (95% CI, 0.814– 3.844; p = 0.150). Following 1:1 propensity score matching (n = 75 per cohort), 21.3% of bloodless patients and 16.0% of transfused patients died in ICU (OR, 1.424; 95% CI, 0.622– 3.260; p = 0.402).
DISCUSSION Current guidelines recommend RBC transfusion in critically ill patients at hemoglobin level less than 6 g/dL in general, hemoglobin level less than 7 g/dL in ventilated or trauma patients and those with stable cardiac disease, and hemoglobin level less than 8 g/dL in those with acute coronary syndrome (8). Clinical, ethical, and legal challenges may arise when transfusion criteria are met in patients for whom blood transfusion is not an option. With limited published literature on critically ill bloodless patients available, here we report the largest study on anemic bloodless patients in critical care setting (9–11). We used transfused patients with hemoglobin level less than or equal to 8 g/dL within first 24 hours of admission to ICU as our comparator. Despite presence of severe anemia and presence of indication(s) for transfusion in many patients, the Critical Care Medicine
mortality rate of bloodless patients managed under our BMS protocol was relatively comparable with transfused patients. Anemia was determined to be the primary or one of the leading causes of death in about one third of bloodless patients who had expired in ICU. Transfused patients were older (by 7 yr on average) and had higher APACHE II scores (by 3.75 points on average, corresponding to ~10% increased risk of mortality) (5), while having higher hemoglobin levels at ICU admission. Nonetheless, the difference in mortality rates did not reach statistical significance (albeit with a trend toward higher mortality in bloodless patients) once the cohorts were matched. A growing number of studies including clinical trials have addressed the impact of transfusion on outcomes, but their relevancy to our population is questionable. In-hospital mortality rates in critically ill adult patients who were randomized to a liberal versus restrictive transfusion strategies were 28.1% versus 22.3% in a landmark trial, and the restrictive transfusion strategy was particularly beneficial in patients with lower APACHE II score and younger age (12). Post hoc analyses of the data further supported this notion in critically ill patients with cardiovascular disease (13), with moderateto-severe head injury (14), and in those undergoing mechanical ventilation in the ICU (15). A meta-analysis of 19 trials involving over 6,000 patients indicated that restrictive transfusion strategies (hemoglobin trigger of 7–8 g/dL) reduced in-hospital mortality (relative risk, 0.77; 95% CI, 0.62–0.95) compared with liberal transfusion strategies (hemoglobin trigger of 9–10 g/dL) (16). A more recent meta-analysis of 31 trials involving 9,813 patients concluded that while restrictive transfusion strategies were associated with reduced transfusion utilization, there was no significant reduction in mortality (relative risk, 0.86; 95% CI, 0.74–1.01) or morbidity rates compared with liberally transfused patients (17). However, a number of recent trials have been suggestive of increased risk of mortality in patients randomized to restrictive transfusion strategy. In one study, 30-day mortality rates in cardiac surgery patients randomized to a restrictive postoperative transfusion threshold (hemoglobin level < 7.5 g/dL) versus liberal transfusion threshold (hemoglobin level < 9 g/dL) were 2.6% versus 1.9%, whereas 90-day mortality rates were 4.2% versus 2.6% (hazard ratio, 1.64; 95% CI, 1.00–2.67) (18). In a pilot trial of 110 patients undergoing cardiac catheterization, 30-day mortality was higher in restrictive (hemoglobin trigger < 8 g/dL, 1.8%) than liberal transfusion arm (hemoglobin trigger < 10 g/dL, 13%) (19). Finally, in a randomized trial of 198 surgical oncology patients, a restrictive transfusion strategy (hemoglobin level < 7 g/dL) was associated with increased 30- and 60-day mortality (23% vs 8% and 24% vs 11%, respectively) compared with liberal transfusion strategy (hemoglobin level < 9 g/dL) (20), although the sharp contrast of the results of this study with previous trials (particularly the magnitude of the reported effect on mortality) are grounds for further studies in this population to corroborate these results. What these trials do not provide is a direct comparison of the outcomes of transfused patients versus those who are not, www.ccmjournal.org
1113
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
or those who are treated with other modalities in severe anemia. Notably, depending on the study’s hemoglobin threshold for transfusion, many to all of our bloodless patients would have been transfused even if they have been randomized to the restrictive transfusion arm of these trials, undermining the relevance of these trials to our study. Several observational studies have also addressed the impact of transfusion on outcomes— often suggesting a negative impact—but these studies are also predominantly based on patients with higher hemoglobin levels, and they suffer from often-present caveat of larger baseline risk and burden of illness in transfused patients (1, 21). It is debated whether the worsening of the outcomes in transfused cohorts is resulted from transfusions or underlying pathologies, which concurrently made transfusions necessary during the course of treatment (21). Studying bloodless patients can also mitigate this issue since not being transfused was not a decision made by the clinicians. Two independent studies on total of 593 Jehovah’s Witness patients with postoperative hemoglobin level less than or equal to 8 g/dL reported that the risk of mortality sharply increased as hemoglobin level dropped below 5–6 g/dL (9, 10). In another study, only 13 of 294 bloodless patients had a nadir hemoglobin level less than 5 g/dL, and all of them survived. There were only two deaths in the bloodless group, and they occurred in patients with nadir hemoglobin levels 5–7 g/dL (11). Another study reported an ICU mortality rate of 19% for critically ill Jehovah’s Witness patients (vs 8.8% in their general ICU population), but it was based on a much smaller sample size (22). In our study, ICU mortality rate in bloodless patients was higher than transfused patients in subgroups with nadir hemoglobin level less than or equal to 6 g/dL within first 24 hours of ICU admission or less than or equal to 5 g/dL during ICU stay (Fig. 2). In the higher hemoglobin categories, the ICU mortality rates were similar in bloodless patients although post-ICU in-hospital mortality rates increased in bloodless patients with higher hemoglobin levels (nadir hemoglobin level > 7 g/dL during the first 24 hr of ICU admission or > 6 g/dL during ICU stay) (Figs. 1 and 2). The increased late mortality is consistent with the findings of a recent trial (18). It can be speculated that some anemic bloodless patients survive ICU stay, but they may still suffer the sequelae of hypoxia or other comorbidities. We did not observe any significant differences in the risk of ICU readmission or emergence of new positive blood culture result between the study cohorts. The latter finding is at odds with the reports indicative that allogeneic blood transfusions may increase the risk of infections (23, 24). Given the limited power of the study and lack of data on clinical infectious complications, our study remains inconclusive on this aspect. Similarly, although our observation of increased risk of electrocardiographic changes and rise of biomarkers suggestive of cardiac ischemia in transfused patients is consistent with reports linking cardiac complications with transfusion (25–28), unavailability of detailed data on baseline cardiac risk of the patients prevents us from drawing conclusions. Although studying bloodless patients can potentially reduce the risk of bias, our study shares many of the limitations of 1114
www.ccmjournal.org
observational studies. Some baseline characteristics of our bloodless patients were more favorable compared with the transfused patients, which were remedied with our matching approach. However, presence of other significant imbalances and measured/unmeasured biases cannot be ruled out. Our baseline data on risk factors of important outcomes such as cardiac or renal complications were limited, which could undermine the matching approach. The transfused patients could have been sicker than bloodless patients in terms of other variables not included in our study, and hence, they could not have been matched accordingly, leaving some bias behind. Nonetheless, we included APACHE II score, which is the most widely used and accepted predictor of mortality in critically ill patients, and accordingly we focused on mortality as the primary outcome in our study. Severely anemic, critically ill bloodless patients represent a rare population, limiting our sample size and the power of the study. Our bloodless patients were treated with structured strategies under our BMS program (Appendix, Supplemental Digital Content 1, http:// links.lww.com/CCM/B666) that could have affected their outcome (4). Although transfused patients were managed under our hospital-wide PBM program, receiving comparable standard of care (4, 10, 29), we cannot rule out the possibility of more aggressive treatment in bloodless patients. Last, although the key tenets of our BMS protocol have remained unchanged throughout the study, it has undergone periodic updates. The trend toward lower hemoglobin thresholds for transfusion (currently set at around 6–7 g/dL in patients with no major comorbidity) (7), which could have made transfusion less likely to be indicated in some of our bloodless patients with higher hemoglobin levels, but we did not see statistically significant differences in mortality rates when analysis was limited to patients with lower nadir hemoglobin levels. Nonetheless, the absolute mortality rates were higher in bloodless patients, and the likelihood of study being underpowered to detect meaningful differences cannot be ruled out.
CONCLUSIONS Our study suggests that within the limitations of controlling for a finite number of risk factors in a rare patient population, the overall risk of mortality in severely anemic critically ill bloodless patients can approach the risk level of those who are managed with transfusion, in a center with BMS and PBM programs. When stratified based on the nadir hemoglobin levels, survival is negatively affected in lower hemoglobin levels. Use of a protocol to manage anemia in these patients is feasible and likely to contribute to improved outcome, whereas more studies are needed to better delineate the impact of such programs.
REFERENCES
1. Shander A, Fink A, Javidroozi M, et al; International Consensus Conference on Transfusion Outcomes Group: Appropriateness of allogeneic red blood cell transfusion: The international consensus conference on transfusion outcomes. Transfus Med Rev 2011; 25:232.e53–246.e53 June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations 2. Wier LM, Pfuntner A, Maeda J et al: HCUP Facts and Figures: Statistics on Hospital-Based Care in the United States, 2009. Rockville, MD, Agency for Healthcare Research and Quality, 2011 3. Athar MK, Puri N, Gerber DR: Anemia and blood transfusions in critically ill patients. J Blood Transfus 2012; 2012:629204 4. Shander A, Javidroozi M, Perelman S, et al: From bloodless surgery to patient blood management. Mt Sinai J Med 2012; 79:56–65 5. Knaus WA, Draper EA, Wagner DP, et al: APACHE II: A severity of disease classification system. Crit Care Med 1985; 13:818–829 6. Austin PC: An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46:399–424 7. Shander A, Gross I, Hill S, et al; College of American Pathologists; American Society of Anesthesiologists; Society of Thoracic Surgeons and Society of Cardiovascular Anesthesiologists; Society of Critical Care Medicine; Italian Society of Transfusion Medicine and Immunohaematology; American Association of Blood Banks: A new perspective on best transfusion practices. Blood Transfus 2013; 11:193–202 8. Napolitano LM, Kurek S, Luchette FA, et al; American College of Critical Care Medicine of the Society of Critical Care Medicine; Eastern Association for the Surgery of Trauma Practice Management Workgroup: Clinical practice guideline: Red blood cell transfusion in adult trauma and critical care. Crit Care Med 2009; 37:3124–3157 9. Carson JL, Noveck H, Berlin JA, et al: Mortality and morbidity in patients with very low postoperative Hb levels who decline blood transfusion. Transfusion 2002; 42:812–818 10. Shander A, Javidroozi M, Naqvi S et al: An update on mortality and morbidity in patients with very low postoperative hemoglobin levels who decline blood transfusion. Transfusion 2014; 54(10 Pt 2):2688–2695 11. Frank SM, Wick EC, Dezern AE, et al: Risk-adjusted clinical outcomes in patients enrolled in a bloodless program. Transfusion 2014; 54:2668–2677 12. Hébert PC, Wells G, Blajchman MA, et al: A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999; 340:409–417 13. Hébert PC, Yetisir E, Martin C, et al; Transfusion Requirements in Critical Care Investigators for the Canadian Critical Care Trials Group: Is a low transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001; 29:227–234 14. McIntyre LA, Fergusson DA, Hutchison JS, et al: Effect of a liberal versus restrictive transfusion strategy on mortality in patients with moderate to severe head injury. Neurocrit Care 2006; 5:4–9 15. Hébert PC, Blajchman MA, Cook DJ, et al; Transfusion Requirements in Critical Care Investigators for the Canadian Critical Care Trials
Critical Care Medicine
Group: Do blood transfusions improve outcomes related to mechanical ventilation? Chest 2001; 119:1850–1857 16. 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 17. Holst LB, Petersen MW, Haase N, et al: Restrictive versus liberal transfusion strategy for red blood cell transfusion: Systematic review of randomised trials with meta-analysis and trial sequential analysis. BMJ 2015; 350:h1354 18. Murphy GJ, Pike K, Rogers CA, et al; TITRe2 Investigators: Liberal or restrictive transfusion after cardiac surgery. N Engl J Med 2015; 372:997–1008 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.e1–971.e1 20. de Almeida JP, Vincent JL, Galas FR, et al: Transfusion requirements in surgical oncology patients: A prospective, randomized controlled trial. Anesthesiology 2015; 122:29–38 21. Shander A, Javidroozi M, Ozawa S, et al: What is really dangerous: Anaemia or transfusion? Br J Anaesth 2011; 107(Suppl 1):i41–i59 22. MacLaren G, Anderson M: Bloodless intensive care: A case series and review of Jehovah’s Witnesses in ICU. Anaesth Intensive Care 2004; 32:798–803 23. Hendrickson JE, Hillyer CD: Noninfectious serious hazards of transfusion. Anesth Analg 2009; 108:759–769 24. Rohde JM, Dimcheff DE, Blumberg N, et al: Health care-associated infection after red blood cell transfusion: A systematic review and meta-analysis. JAMA 2014; 311:1317–1326 25. Möhnle P, Snyder-Ramos SA, Miao Y, et al; Multicenter Study of Perioperative Ischemia (McSPI) Research Group: Postoperative red blood cell transfusion and morbid outcome in uncomplicated cardiac surgery patients. Intensive Care Med 2011; 37:97–109 26. Vivacqua A, Koch CG, Yousuf AM, et al: Morbidity of bleeding after cardiac surgery: Is it blood transfusion, reoperation for bleeding, or both? Ann Thorac Surg 2011; 91:1780–1790 27. Whitson BA, Huddleston SJ, Savik K, et al: Risk of adverse outcomes associated with blood transfusion after cardiac surgery depends on the amount of transfusion. J Surg Res 2010; 158:20–27 28. Sherwood MW, Wang Y, Curtis JP, et al: Patterns and outcomes of red blood cell transfusion in patients undergoing percutaneous coronary intervention. JAMA 2014; 311:836–843 29. McEvoy MT, Shander A: Anemia, bleeding, and blood transfusion in the intensive care unit: Causes, risks, costs, and new strategies. Am J Crit Care 2013; 22:eS1–eS13; quiz eS14
www.ccmjournal.org
1115
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Objective: To compare the outcomes of severely anemic critically ill patients for whom transfusion is not an option (“bloodless” patients) with transfused patients. Design: Cohort study with propensity score matching. Setting: ICU of a referral center. Patients: One hundred seventy-eight bloodless and 441 transfused consecutive severely anemic, critically ill patients, admitted between May 1996 and April 2011, and having at least one hemoglobin level less than or equal to 8 g/dL within 24 hours of ICU admission. Patients with diagnosis of brain injury, acute myocardial infarction, or status postcardiac surgery were excluded. Interventions: Allogeneic RBC transfusion during ICU stay. *See also p. 1238. All authors: Department of Anesthesiology and Critical Care Medicine, Englewood Hospital and Medical Center, Englewood, NJ. This work was performed at Englewood Hospital and Medical Center. All authors contributed equally to the design of the study. Drs. Shander, Gianatiempo, Gandhi, Lui, Califano, Kaufman, and Syed were responsible for management of the ICU patients during the study period. Drs. Naqvi and Aregbeyen were primarily responsible for data collection and verification. All authors had unrestricted access to the study data. Statistical analysis was performed by Dr. Javidroozi. All authors contributed to preparing the article and reviewed it prior to submission. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ ccmjournal). Dr. Shander received funding from Bayer, Luitpold, and Masimo; has been a consultant or speaker with honorarium for or received research support from Bayer, Luitpold, Masimo, Novartis, Novo Nordisk, OrthoBiotech, Pfizer, Masimo, and Zymogenetics; and is a founding member of the Society for the Advancement of Blood Management (SABM). Dr. Javidroozi been a consultant and contractor for and received funding from SABM and Gauss Surgical. Dr. Naqvi disclosed work at Englewood Hospital and Medical Center as a TeamHealth employee. Dr. Aregbeyen is employed at Englewood Hospital and Medical Center, New Jersey. The remaining authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: 10.1097/CCM.0000000000001599
Critical Care Medicine
Measurements and Main Results: Primary outcome was in-hospital mortality. Other outcomes were ICU mortality, readmission to ICU, new electrocardiographic or cardiac enzyme changes suggestive of cardiac ischemia or injury, and new positive blood culture result. Transfused patients were older, had higher hemoglobin level at admission, and had higher Acute Physiology and Chronic Health Evaluation II score. Hospital mortality rates were 24.7% in bloodless and 24.5% in transfused patients (odds ratio, 1.01; 95% CI, 0.68–1.52; p = 0.95). Adjusted odds ratio of hospital mortality was 1.52 (95% CI, 0.95–2.43; p = 0.08). No significant difference in ICU readmission or positive blood culture results was observed. Analysis of propensity score–matched cohorts provided similar results. Conclusions: Overall risk of mortality in severely anemic critically ill bloodless patients appeared to be comparable with transfused patients, albeit the latter group had older age and higher Acute Physiology and Chronic Health Evaluation II score. Use of a protocol to manage anemia in these patients in a center with established patient blood management and bloodless medicine and surgery programs is feasible and likely to contribute to improved outcome, whereas more studies are needed to better delineate the impact of such programs. (Crit Care Med 2016; 44:1109–1115) Key Words: anemia; bloodless medical and surgical procedures; blood transfusion; critical illness; Jehovah’s Witnesses
A
llogeneic RBC transfusion is commonly used in anemia to avoid tissue hypoxia and its consequences. Despite significant risks, costs, and challenges to determine reliably the patients in whom its benefits are likely to outweigh the risks (1), millions of units of allogeneic blood are transfused annually (2). Certain patient populations, namely the critically ill, are particularly at risk (3). Caring for anemic patients for whom blood transfusions are not an option (called “bloodless” here) has been a challenge. Refusal of transfusion does not equate refusal of medical care, and these patients can still benefit from treatments to www.ccmjournal.org
1109
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
augment their intrinsic hematopoietic capabilities, minimize blood loss, and improve physiologic response to anemia (4). These patients may also provide an opportunity to study the association between transfusion and outcomes since the decision not to transfuse is not dictated by their clinical status. We compared the outcomes of severely anemic critically ill “bloodless” patients with transfused patients at a referral center with programs in place to care for these patients. We hypothesized that the outcomes of bloodless patients would be worse than their transfused counterparts.
MATERIALS AND METHODS The study was approved by the local hospital institutional review board, and individual patient consent was waived. Consecutive patients admitted to a medical/surgical ICU between May 1996 and April 2011, in a hospital with established patient blood management (PBM) and bloodless medicine and surgery (BMS) were screened retrospectively. The main eligibility criterion was severe anemia at ICU admission defined as at least one hemoglobin level less than or equal to 8 g/dL from a blood sample taken within 24 hours of ICU admission. Exclusion criteria were age less than 18 years old; death within 24 hours of admission; treatment with artificial hemoglobin-based oxygen carriers; admission with diagnosis of traumatic brain injury or acute myocardial infarction; and status postcardiac surgery and pregnancy. For patients with multiple hospital admissions, only the first admission leading to ICU admission was considered. Patients were grouped into two cohorts: bloodless patients who did not receive any allogeneic blood transfusions; and transfused patients, who received allogeneic RBC transfusions during ICU stay. Patients who are unwilling to accept blood transfusions are managed under our BMS program. An advanced directive is obtained as early as permissible to identify what blood components, treatments, and procedures are acceptable to the patients. Management of these patients (including those in this study) is guided by our bloodless protocol, which has been in place at our center since 1994 (Appendix, Supplemental Digital Content 1, http://links.lww.com/CCM/B666). The protocol has been periodically updated, but no significant changes were made during the study period. The primary study outcome was in-hospital mortality that was further stratified into ICU mortality and in-hospital mortality after ICU discharge. Other outcomes were readmission to ICU, new electrocardiographic changes suggestive of cardiac ischemia or injury, new-onset rise in cardiac enzymes, and new positive blood culture result. As a retrospective study, there were no systematic criteria to order any of these tests, and data recorded in the patient charts were used as the sole resource. Lack of performance of any specific test was considered to indicate that the test was not deemed to be indicated by the treating clinicians, and hence the results were assumed to be negative. Acute Physiology and Chronic Health Evaluation (APACHE) II score of the patients was calculated as previously described (5). Additionally, a modified APACHE II score (with 1110
www.ccmjournal.org
age and hematocrit subscores removed) was calculated and used in multivariate analyses when age and hematocrit were entered individually. Statistical Analysis Chi-square test, Student t test, and Mann-Whitney U test were performed as appropriate, and values are accordingly provided in forms of count followed by percentage (%), mean ± sd, and median/interquartile range (IQR). No missing data were encounter for key study variables. Logistic regression analysis was performed with various endpoints as dependent variables to adjust the odds ratio (OR) for other confounders. The 95% CI for ORs are provided. For all studied endpoints, OR values above 1 are indicative of favorable outcomes for transfused versus bloodless patients. Statistical analysis was performed using SPSS (SPSS for Windows, Version 13.0, Chicago, IL). Using the patients’ age, gender, postsurgical status, lowest hemoglobin level within first 24 hours of ICU admission, and APACHE II score, the propensity of being transfused was calculated. Bloodless patients were matched to transfused patients in 1:2 ratio based on propensity score using the nearest neighbor matching within a caliper distance of 0.05 without replacement (6). During the period that the patients were admitted to our ICU, a number of transfusion guidelines were published with a trend toward lowering the hemoglobin threshold for transfusion (7). To account for the potential impact of lower hemoglobin transfusion criteria, we performed two sets of subgroup analysis, excluding patients with nadir ICU hemoglobin levels above 7 and above 6 g/dL. Availability of severely anemic bloodless patients was a limiting factor in our study, and following initial sample size estimations, we resorted to a convenience sampling of 119 bloodless patients, matched 1:2 to 238 transfused patients for the matched analysis. For comparison and assuming a mortality rate of 15%, if there was truly no difference between the groups, 238 patients were required to be 80% sure that a two-sided 95% CI would exclude a mortality rate difference in favor of the transfused group of more than 13%.
RESULTS Study population consisted of 619 severely anemic patients admitted to the ICU, 28.8% of which were bloodless. Mean age was 65.05 ± 16.91 years old and 59% were female. Leading ICU admitting diagnoses were gastrointestinal conditions (42.2%), pulmonary conditions (12.8%), sepsis/septic shock (6.6%), postvascular surgery (5.8%), and gynecologic conditions (4.7%). Approximately 36% of the patients were admitted to ICU following a surgical procedure during the same hospital admission (status “postsurgical”). Transfused patients (n = 441) received median/IQR of 2/2 units of RBC. The last measured hemoglobin level prior to first RBC transfusion in ICU was 6.76 ± 1.11 g/dL. Among the transfused, 51, 28, and 9 patients also received plasma (2/2 units), platelet (1/1 units), and cryoprecipitate (2/3 units), respectively. Table 1 provides the key characteristics June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations
Table 1.
Baseline Characteristics and Outcome Data of Bloodless and Transfused Patients Bloodless (n = 178)
Transfused (n = 441)
Univariate p
60.07 ± 16.57
67.06 ± 16.65
< 0.001
113 (63.5)
252 (57.1)
0.147
80 (44.9)
158 (35.8)
0.035
6.12 ± 1.85
7.07 ± 1.45
< 0.001
20.56 ± 7.71
24.31 ± 7.47
< 0.001
5.30 ± 1.65
6.44 ± 0.97
< 0.001
ICU stay until lowest hemoglobin level (d)
1/1
1/2
< 0.001
Length of ICU stay in survived patients (d)
4/5
3/5
0.171
6.20 ± 1.86
8.44 ± 1.42
< 0.001
ICU
28 (15.7)
72 (16.3)
0.855
Hospital post-ICU
16 (9.0)
36 (8.2)
0.738
Overall (ICU + hospital post-ICU)
44 (24.7)
108 (24.5)
0.952
Total (overall + subsequent admissions)
49 (27.5)
115 (26.1)
0.711
ICU readmission (%)
19 (10.7)
49 (11.1)
0.875
New electrocardiographic changes suggestive of ischemia or injury (%)
22 (12.4)
126 (28.6)
< 0.001
Cardiac enzyme rise (%)
18 (10.1)
136 (30.8)
< 0.001
New positive blood culture result (%)
16 (9.0)
56 (12.7)
0.193
Variables
Age (yr) Female gender (%) Postsurgical cases (%) Lowest hemoglobin level within first 24 hr in ICU (g/dL) Acute Physiology and Chronic Health Evaluation II score Lowest hemoglobin level in ICU (g/dL)
ICU discharge hemoglobin level (g/dL) Mortality (%)
Data are provided as mean ± sd, median/interquartile range, or n (%).
and outcome data. Transfused patients were older, had higher hemoglobin level at admission, and had higher APACHE II score (with significantly higher scores in temperature, heart rate, respiratory rate, oxygenation status, blood pH, WBC count and age components, and significantly lower scores in hematocrit and Glasgow Coma Scale components). One hundred sixty-four patients (26.5%) expired: 100 during ICU stay (16.2%); 52 during hospital stay after ICU discharge (8.4%); and the remaining during subsequent hospital admissions (Table 1). Among the 28 bloodless patients who died in ICU, the last hemoglobin level prior to death was 5.57 ± 1.38 g/dL, and it was measured within 1 or 2 days prior to death. Anemia was classified as the primary cause of death or a main contributing factor in 36.4% of these patients. Other leading causes of death included multiple comorbidities (36.4%), sepsis/septic shock (27.3%), advanced malignancies (13.6%), multiple organ failure (9.1%), respiratory failure (9.1%), and hemorrhagic shock (9.1%). Seven patients had been transferred from other hospitals. Six patients had “do-not-resuscitate” (DNR) order on file, including two who were DNR at admission to hospital. Unadjusted mortality rates according to the nadir hemoglobin levels during ICU stay for bloodless patients are presented in Figure 1. Unadjusted ORs of ICU and overall mortality were 0.957 (95% CI, 0.594–1.540; p = 0.855) and 1.012 (95% CI, 0.676–1.516; p = 0.952), respectively. Following adjustment for Critical Care Medicine
Figure 1. Mortality rates in bloodless patients according to the lowest hemoglobin (Hb) level during ICU stay. In each column, the lower (light gray) part represents mortality during ICU stay and the upper (dark gray) part represents mortality occurring during hospital stay out of ICU. Numbers in parentheses are the total number of cases in each category.
age, gender, postsurgical status, lowest hemoglobin level during first 24-hour of ICU admission, and modified APACHE II score, ORs of death in ICU and during overall hospital stay were 1.299 (95% CI, 0.756–2.234; p = 0.344) and 1.522 (95% CI, 0.955–2.426; p = 0.078), respectively. Older age, lower hemoglobin level, and higher modified APACHE II score were the significant independent predictors of mortality. www.ccmjournal.org
1111
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
Table 2. Comparison of Factors Used in Propensity Score Matching Between Bloodless and Transfused Cohorts Factors
Bloodless (n = 119)
Transfused (n = 238)
Univariate p
Age (yr)
63.54 ± 14.98
64.58 ± 17.36
0.556
Female gender (%)
73 (61.3)
144 (60.5)
0.878
Postsurgical cases (%)
49 (41.2)
99 (41.6)
0.939
6.87 ± 1.44
6.83 ± 1.56
0.785
Acute Physiology and Chronic Health Evaluation II score
22.31 ± 7.55
22.60 ± 7.42
0.733
Propensity score
0.298 ± 0.157
0.298 ± 0.158
0.997
Lowest hemoglobin level within first 24 hr in ICU (g/dL)
Data are provided as mean ± sd or n (%).
Unadjusted OR of readmission to ICU was 0.956 (95% CI, 0.546–1.675; p = 0.875). Results were similar when expired patients were excluded and when adjusted for other parameters mentioned earlier (data not shown). Unadjusted OR of new electrocardiographic changes suggestive of ischemia or injury was 0.353 (95% CI, 0.216–0.577; p < 0.001). Following adjustment for other parameters, OR of new electrocardiographic changes was 0.303 (OR, 0.177–0.520; p < 0.001). Other significant independent predictors were older age, male gender, and lower hemoglobin level. The independent association remained significant following exclusion of all 19 cases admitted to the ICU with cardiac diagnoses (OR, 0.293; 95% CI, 0.169–0.510; p < 0.001). For new-onset rise in cardiac enzymes, unadjusted and adjusted OR were 0.252 (95% CI, 0.149– 0.428; p < 0.001) and 0.324 (95% CI, 0.186–0.564; p < 0.001). Older age and higher modified APACHE II score were other significant independent predictors of new-onset raised cardiac enzyme. The independent association remained significant following exclusion of 19 cases admitted to the ICU with cardiac diagnoses (OR, 0.324; 95% CI, 0.183–0.573; p < 0.001).
Unadjusted OR of new positive blood culture result was 0.679 (95% CI, 0.378–1.219; p = 0.193). Following adjustment for other factors, OR of new positive blood culture result was 0.739 (95% CI, 0.397–1.374; p = 0.339). The factors used in propensity score matching are summarized in Table 2. Comparison of outcomes between matched bloodless and transfused cohorts are provided in Table 3. Figure 2 depicts the mortality rates in matched bloodless and transfused patients according to hemoglobin levels. Following exclusion of patients with nadir hemoglobin level of 7 g/dL or more, 26 of 142 bloodless (18.3%) and 54 of 297 transfused patients (18.2%) died in ICU (OR, 1.009; 95% CI, 0.601–1.693; p = 0.974). OR of ICU mortality following adjustment for other confounders was 1.652 (95% CI, 0.899– 3.036; p = 0.106). Following 1:1 propensity score matching (n = 98 per cohort), 19.4% of bloodless patients and 15.3% of transfused patients died in ICU (OR, 1.331; 95% CI, 0.633–2.800; p = 0.451). When patients with nadir hemoglobin level of 6 g/dL or more were excluded, 22 of 116 bloodless (19.0%) and 20 of
Table 3. Outcomes of Subsets of Bloodless and Transfused Patients Matched Based on the Propensity Score Bloodless (n = 119)
Transfused (n = 238)
OR (95% CI)
p
3/4
3/4
NA
0.353
ICU
17 (14.3)
33 (13.9)
1.035 (0.551–1.947)
0.914
Hospital post-ICU
12 (10.1)
17 (7.1)
1.458 (0.672–3.162)
0.338
Overall (ICU + hospital post-ICU)
29 (24.4)
50 (21.0)
1.212 (0.719–2.042)
0.471
Total (overall + subsequent admissions)
33 (27.7)
54 (22.7)
1.307 (0.791–2.162)
0.296
ICU readmission (%)
17 (13.2)
30 (12.6)
1.156 (0.609–2.192)
0.658
New electrocardiographic changes suggestive of ischemia or injury (%)
13 (10.9)
61 (25.6)
0.356 (0.187–0.678)
0.001
Cardiac enzyme rise (%)
14 (11.8)
65 (27.3)
0.355 (0.190–0.664)
0.001
New positive blood culture result (%)
10 (8.4)
22 (9.2)
0.901 (0.412–1.969)
0.793
Outcomes
Length of ICU stay in survived patients (d) Mortality (%)
OR = odds ratio, NA = not applicable. Data are provided as mean ± sd, median/interquartile range, or n (%).
1112
www.ccmjournal.org
June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations
Figure 2. Mortality rates in propensity score–matched transfused and bloodless patients according to the lowest hemoglobin (Hb) level within first 24 hr of ICU admission (A) and lowest Hb level during ICU stay (B). In each column, the lower (light gray) part represents mortality during ICU stay and the upper (dark gray) part represents mortality occurring during hospital stay out of ICU. Numbers in parentheses are the total number of cases in each category. Star symbol indicates p < 0.5 comparing total mortality rate between the bloodless and transfused patients.
111 transfused patients (18.0%) died in ICU (OR, 1.065; 95% CI, 0.545–2.082; p = 0.854). OR of ICU mortality following adjustment for other confounders was 1.769 (95% CI, 0.814– 3.844; p = 0.150). Following 1:1 propensity score matching (n = 75 per cohort), 21.3% of bloodless patients and 16.0% of transfused patients died in ICU (OR, 1.424; 95% CI, 0.622– 3.260; p = 0.402).
DISCUSSION Current guidelines recommend RBC transfusion in critically ill patients at hemoglobin level less than 6 g/dL in general, hemoglobin level less than 7 g/dL in ventilated or trauma patients and those with stable cardiac disease, and hemoglobin level less than 8 g/dL in those with acute coronary syndrome (8). Clinical, ethical, and legal challenges may arise when transfusion criteria are met in patients for whom blood transfusion is not an option. With limited published literature on critically ill bloodless patients available, here we report the largest study on anemic bloodless patients in critical care setting (9–11). We used transfused patients with hemoglobin level less than or equal to 8 g/dL within first 24 hours of admission to ICU as our comparator. Despite presence of severe anemia and presence of indication(s) for transfusion in many patients, the Critical Care Medicine
mortality rate of bloodless patients managed under our BMS protocol was relatively comparable with transfused patients. Anemia was determined to be the primary or one of the leading causes of death in about one third of bloodless patients who had expired in ICU. Transfused patients were older (by 7 yr on average) and had higher APACHE II scores (by 3.75 points on average, corresponding to ~10% increased risk of mortality) (5), while having higher hemoglobin levels at ICU admission. Nonetheless, the difference in mortality rates did not reach statistical significance (albeit with a trend toward higher mortality in bloodless patients) once the cohorts were matched. A growing number of studies including clinical trials have addressed the impact of transfusion on outcomes, but their relevancy to our population is questionable. In-hospital mortality rates in critically ill adult patients who were randomized to a liberal versus restrictive transfusion strategies were 28.1% versus 22.3% in a landmark trial, and the restrictive transfusion strategy was particularly beneficial in patients with lower APACHE II score and younger age (12). Post hoc analyses of the data further supported this notion in critically ill patients with cardiovascular disease (13), with moderateto-severe head injury (14), and in those undergoing mechanical ventilation in the ICU (15). A meta-analysis of 19 trials involving over 6,000 patients indicated that restrictive transfusion strategies (hemoglobin trigger of 7–8 g/dL) reduced in-hospital mortality (relative risk, 0.77; 95% CI, 0.62–0.95) compared with liberal transfusion strategies (hemoglobin trigger of 9–10 g/dL) (16). A more recent meta-analysis of 31 trials involving 9,813 patients concluded that while restrictive transfusion strategies were associated with reduced transfusion utilization, there was no significant reduction in mortality (relative risk, 0.86; 95% CI, 0.74–1.01) or morbidity rates compared with liberally transfused patients (17). However, a number of recent trials have been suggestive of increased risk of mortality in patients randomized to restrictive transfusion strategy. In one study, 30-day mortality rates in cardiac surgery patients randomized to a restrictive postoperative transfusion threshold (hemoglobin level < 7.5 g/dL) versus liberal transfusion threshold (hemoglobin level < 9 g/dL) were 2.6% versus 1.9%, whereas 90-day mortality rates were 4.2% versus 2.6% (hazard ratio, 1.64; 95% CI, 1.00–2.67) (18). In a pilot trial of 110 patients undergoing cardiac catheterization, 30-day mortality was higher in restrictive (hemoglobin trigger < 8 g/dL, 1.8%) than liberal transfusion arm (hemoglobin trigger < 10 g/dL, 13%) (19). Finally, in a randomized trial of 198 surgical oncology patients, a restrictive transfusion strategy (hemoglobin level < 7 g/dL) was associated with increased 30- and 60-day mortality (23% vs 8% and 24% vs 11%, respectively) compared with liberal transfusion strategy (hemoglobin level < 9 g/dL) (20), although the sharp contrast of the results of this study with previous trials (particularly the magnitude of the reported effect on mortality) are grounds for further studies in this population to corroborate these results. What these trials do not provide is a direct comparison of the outcomes of transfused patients versus those who are not, www.ccmjournal.org
1113
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Shander et al
or those who are treated with other modalities in severe anemia. Notably, depending on the study’s hemoglobin threshold for transfusion, many to all of our bloodless patients would have been transfused even if they have been randomized to the restrictive transfusion arm of these trials, undermining the relevance of these trials to our study. Several observational studies have also addressed the impact of transfusion on outcomes— often suggesting a negative impact—but these studies are also predominantly based on patients with higher hemoglobin levels, and they suffer from often-present caveat of larger baseline risk and burden of illness in transfused patients (1, 21). It is debated whether the worsening of the outcomes in transfused cohorts is resulted from transfusions or underlying pathologies, which concurrently made transfusions necessary during the course of treatment (21). Studying bloodless patients can also mitigate this issue since not being transfused was not a decision made by the clinicians. Two independent studies on total of 593 Jehovah’s Witness patients with postoperative hemoglobin level less than or equal to 8 g/dL reported that the risk of mortality sharply increased as hemoglobin level dropped below 5–6 g/dL (9, 10). In another study, only 13 of 294 bloodless patients had a nadir hemoglobin level less than 5 g/dL, and all of them survived. There were only two deaths in the bloodless group, and they occurred in patients with nadir hemoglobin levels 5–7 g/dL (11). Another study reported an ICU mortality rate of 19% for critically ill Jehovah’s Witness patients (vs 8.8% in their general ICU population), but it was based on a much smaller sample size (22). In our study, ICU mortality rate in bloodless patients was higher than transfused patients in subgroups with nadir hemoglobin level less than or equal to 6 g/dL within first 24 hours of ICU admission or less than or equal to 5 g/dL during ICU stay (Fig. 2). In the higher hemoglobin categories, the ICU mortality rates were similar in bloodless patients although post-ICU in-hospital mortality rates increased in bloodless patients with higher hemoglobin levels (nadir hemoglobin level > 7 g/dL during the first 24 hr of ICU admission or > 6 g/dL during ICU stay) (Figs. 1 and 2). The increased late mortality is consistent with the findings of a recent trial (18). It can be speculated that some anemic bloodless patients survive ICU stay, but they may still suffer the sequelae of hypoxia or other comorbidities. We did not observe any significant differences in the risk of ICU readmission or emergence of new positive blood culture result between the study cohorts. The latter finding is at odds with the reports indicative that allogeneic blood transfusions may increase the risk of infections (23, 24). Given the limited power of the study and lack of data on clinical infectious complications, our study remains inconclusive on this aspect. Similarly, although our observation of increased risk of electrocardiographic changes and rise of biomarkers suggestive of cardiac ischemia in transfused patients is consistent with reports linking cardiac complications with transfusion (25–28), unavailability of detailed data on baseline cardiac risk of the patients prevents us from drawing conclusions. Although studying bloodless patients can potentially reduce the risk of bias, our study shares many of the limitations of 1114
www.ccmjournal.org
observational studies. Some baseline characteristics of our bloodless patients were more favorable compared with the transfused patients, which were remedied with our matching approach. However, presence of other significant imbalances and measured/unmeasured biases cannot be ruled out. Our baseline data on risk factors of important outcomes such as cardiac or renal complications were limited, which could undermine the matching approach. The transfused patients could have been sicker than bloodless patients in terms of other variables not included in our study, and hence, they could not have been matched accordingly, leaving some bias behind. Nonetheless, we included APACHE II score, which is the most widely used and accepted predictor of mortality in critically ill patients, and accordingly we focused on mortality as the primary outcome in our study. Severely anemic, critically ill bloodless patients represent a rare population, limiting our sample size and the power of the study. Our bloodless patients were treated with structured strategies under our BMS program (Appendix, Supplemental Digital Content 1, http:// links.lww.com/CCM/B666) that could have affected their outcome (4). Although transfused patients were managed under our hospital-wide PBM program, receiving comparable standard of care (4, 10, 29), we cannot rule out the possibility of more aggressive treatment in bloodless patients. Last, although the key tenets of our BMS protocol have remained unchanged throughout the study, it has undergone periodic updates. The trend toward lower hemoglobin thresholds for transfusion (currently set at around 6–7 g/dL in patients with no major comorbidity) (7), which could have made transfusion less likely to be indicated in some of our bloodless patients with higher hemoglobin levels, but we did not see statistically significant differences in mortality rates when analysis was limited to patients with lower nadir hemoglobin levels. Nonetheless, the absolute mortality rates were higher in bloodless patients, and the likelihood of study being underpowered to detect meaningful differences cannot be ruled out.
CONCLUSIONS Our study suggests that within the limitations of controlling for a finite number of risk factors in a rare patient population, the overall risk of mortality in severely anemic critically ill bloodless patients can approach the risk level of those who are managed with transfusion, in a center with BMS and PBM programs. When stratified based on the nadir hemoglobin levels, survival is negatively affected in lower hemoglobin levels. Use of a protocol to manage anemia in these patients is feasible and likely to contribute to improved outcome, whereas more studies are needed to better delineate the impact of such programs.
REFERENCES
1. Shander A, Fink A, Javidroozi M, et al; International Consensus Conference on Transfusion Outcomes Group: Appropriateness of allogeneic red blood cell transfusion: The international consensus conference on transfusion outcomes. Transfus Med Rev 2011; 25:232.e53–246.e53 June 2016 • Volume 44 • Number 6
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
Clinical Investigations 2. Wier LM, Pfuntner A, Maeda J et al: HCUP Facts and Figures: Statistics on Hospital-Based Care in the United States, 2009. Rockville, MD, Agency for Healthcare Research and Quality, 2011 3. Athar MK, Puri N, Gerber DR: Anemia and blood transfusions in critically ill patients. J Blood Transfus 2012; 2012:629204 4. Shander A, Javidroozi M, Perelman S, et al: From bloodless surgery to patient blood management. Mt Sinai J Med 2012; 79:56–65 5. Knaus WA, Draper EA, Wagner DP, et al: APACHE II: A severity of disease classification system. Crit Care Med 1985; 13:818–829 6. Austin PC: An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46:399–424 7. Shander A, Gross I, Hill S, et al; College of American Pathologists; American Society of Anesthesiologists; Society of Thoracic Surgeons and Society of Cardiovascular Anesthesiologists; Society of Critical Care Medicine; Italian Society of Transfusion Medicine and Immunohaematology; American Association of Blood Banks: A new perspective on best transfusion practices. Blood Transfus 2013; 11:193–202 8. Napolitano LM, Kurek S, Luchette FA, et al; American College of Critical Care Medicine of the Society of Critical Care Medicine; Eastern Association for the Surgery of Trauma Practice Management Workgroup: Clinical practice guideline: Red blood cell transfusion in adult trauma and critical care. Crit Care Med 2009; 37:3124–3157 9. Carson JL, Noveck H, Berlin JA, et al: Mortality and morbidity in patients with very low postoperative Hb levels who decline blood transfusion. Transfusion 2002; 42:812–818 10. Shander A, Javidroozi M, Naqvi S et al: An update on mortality and morbidity in patients with very low postoperative hemoglobin levels who decline blood transfusion. Transfusion 2014; 54(10 Pt 2):2688–2695 11. Frank SM, Wick EC, Dezern AE, et al: Risk-adjusted clinical outcomes in patients enrolled in a bloodless program. Transfusion 2014; 54:2668–2677 12. Hébert PC, Wells G, Blajchman MA, et al: A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med 1999; 340:409–417 13. Hébert PC, Yetisir E, Martin C, et al; Transfusion Requirements in Critical Care Investigators for the Canadian Critical Care Trials Group: Is a low transfusion threshold safe in critically ill patients with cardiovascular diseases? Crit Care Med 2001; 29:227–234 14. McIntyre LA, Fergusson DA, Hutchison JS, et al: Effect of a liberal versus restrictive transfusion strategy on mortality in patients with moderate to severe head injury. Neurocrit Care 2006; 5:4–9 15. Hébert PC, Blajchman MA, Cook DJ, et al; Transfusion Requirements in Critical Care Investigators for the Canadian Critical Care Trials
Critical Care Medicine
Group: Do blood transfusions improve outcomes related to mechanical ventilation? Chest 2001; 119:1850–1857 16. 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 17. Holst LB, Petersen MW, Haase N, et al: Restrictive versus liberal transfusion strategy for red blood cell transfusion: Systematic review of randomised trials with meta-analysis and trial sequential analysis. BMJ 2015; 350:h1354 18. Murphy GJ, Pike K, Rogers CA, et al; TITRe2 Investigators: Liberal or restrictive transfusion after cardiac surgery. N Engl J Med 2015; 372:997–1008 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.e1–971.e1 20. de Almeida JP, Vincent JL, Galas FR, et al: Transfusion requirements in surgical oncology patients: A prospective, randomized controlled trial. Anesthesiology 2015; 122:29–38 21. Shander A, Javidroozi M, Ozawa S, et al: What is really dangerous: Anaemia or transfusion? Br J Anaesth 2011; 107(Suppl 1):i41–i59 22. MacLaren G, Anderson M: Bloodless intensive care: A case series and review of Jehovah’s Witnesses in ICU. Anaesth Intensive Care 2004; 32:798–803 23. Hendrickson JE, Hillyer CD: Noninfectious serious hazards of transfusion. Anesth Analg 2009; 108:759–769 24. Rohde JM, Dimcheff DE, Blumberg N, et al: Health care-associated infection after red blood cell transfusion: A systematic review and meta-analysis. JAMA 2014; 311:1317–1326 25. Möhnle P, Snyder-Ramos SA, Miao Y, et al; Multicenter Study of Perioperative Ischemia (McSPI) Research Group: Postoperative red blood cell transfusion and morbid outcome in uncomplicated cardiac surgery patients. Intensive Care Med 2011; 37:97–109 26. Vivacqua A, Koch CG, Yousuf AM, et al: Morbidity of bleeding after cardiac surgery: Is it blood transfusion, reoperation for bleeding, or both? Ann Thorac Surg 2011; 91:1780–1790 27. Whitson BA, Huddleston SJ, Savik K, et al: Risk of adverse outcomes associated with blood transfusion after cardiac surgery depends on the amount of transfusion. J Surg Res 2010; 158:20–27 28. Sherwood MW, Wang Y, Curtis JP, et al: Patterns and outcomes of red blood cell transfusion in patients undergoing percutaneous coronary intervention. JAMA 2014; 311:836–843 29. McEvoy MT, Shander A: Anemia, bleeding, and blood transfusion in the intensive care unit: Causes, risks, costs, and new strategies. Am J Crit Care 2013; 22:eS1–eS13; quiz eS14
www.ccmjournal.org
1115
Copyright © 2016 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.
