Introduction

There are a large number of randomized clinical trials comparing short-term mortality between liberal and restrictive transfusion strategies. However, transfusion is thought to have long-term consequences related to alteration of immune function. These effects have been hypothesized to increase the risk of subsequent infections and recurrence of cancer. Thus, it is possible that transfusion will increase the risk of long-term mortality by increasing the deaths due to infections and recurrent cancer. Alternatively, more liberal transfusion might reduce cardiac complications by increasing oxygen to vulnerable myocardium and reduce deaths from cardiovascular disease. We hypothesized that liberal transfusion would affect cause specific mortality and might affect all-cause mortality, depending on the predominant impact of transfusion.

Methods

We performed a secondary analysis of the FOCUS trial that randomly allocated patients undergoing hip fracture repair with postoperative hemoglobin concentrations below 10 g/dL within 3 days of surgery. Patients were eligible for the trial if they had underlying cardiovascular disease (coronary artery disease, congestive heart failure, stroke, peripheral vascular disease) or cardiovascular risk factors (diabetes mellitus, hypertension, hyperlipidemia, tobacco use, or renal insufficiency). We randomly allocated patients to liberal transfusion where patients received blood transfusion to maintain hemoglobin 10 g/dL or greater, or restrictive transfusion where patients received blood transfusion when hemoglobin level was less than 8 g/dL or for symptoms. Long-term mortality was determined by linking the study subjects to national death registries in US and Canada. We compared survival time between the two transfusion treatment strategies using the unadjusted log-rank test and Cox proportional hazard models. The underlying cause of death was identified by the national death registries and grouped into seven categories: cardiovascular disease, cancer, infection, stroke, dementia, pulmonary, and other.

Results

There were 1007 subjects randomly allocated to the liberal transfusion strategy and 1009 to the restrictive transfusion strategy. The baseline clinical status was similar between the two groups; mean age of the study population was 81.6 years (range, 51 to 103). The liberal transfusion group were transfused a total of 1866 units and restrictive transfusion group 652 units. We established long-term survival for 2002 (99.3%) of the study population. The median follow-up was 3.1 years (interquartile range, 2.4 to 4.1 years) and there were 841 (42.0%) deaths. There was no difference in the long-term mortality between the liberal transfusion strategy (N=432 deaths) and restrictive transfusion strategy (N= 409 deaths); hazard ratio =1.09; 95% confidence interval 0.95 to 1.25 (Figure). The results were consistent across all subgroups including demographics and multiple co-morbidities. There was also no difference in the underlying cause of death between the transfusion strategies (p=0.99) (Table); Cardiovascular disease (liberal-32.6%, restrictive-33.5%), Cancer (liberal-12.5%, restrictive-12.0%), Infection (liberal-9.5%, restrictive 9.0%).

Conclusions

Liberal transfusion of 10 g/dL did not reduce or increase long-term mortality compared to restrictive transfusion strategy using an 8 g/dL threshold or symptoms in a high risk group of elderly patients with underlying cardiovascular disease or risk factors. The underlying causes of death were similar in both arms of the trial and liberal transfusion did not appear to increase risk of death from infection or cancer or reduce the risk of death from cardiovascular disease. These results do not support the hypotheses that transfusion leads to long-term immunosuppression that is severe enough to influence mortality or cause of death. Our findings suggest that clinicians should primarily consider short-term effects of transfusion when deciding whom to transfuse.

Table
      
  Total N (%)
 
Liberal
N (%)
 
Restrictive
N (%)
 
 
 Cardiovascular Disease
 
278 (33.1)
 
141 (32.6)
 
137 (33.5)
 
 
 Cancer
 
103 (12.2)
 
54 (12.5)
 
49 (12.0)
 
 
 Infection
 
78 (9.3)
 
41 (9.5)
 
37 (9.0)
 
 
 Stroke
 
57 (6.8)
 
27 (6.3)
 
30 (7.3)
 
 
 Dementia
 
108 (12.8)
 
56 (13.0)
 
52 (12.7)
 
 
 Pulmonary
 
58 (6.9)
 
29 (6.7)
 
29 (7.1)
 
 
 Other
 
147 (17.5)
 
79 (18.3)
 
68 (16.6)
 
 
 Unknown
 
12 (1.4)
 
5 (1.2)
 
7 (1.7)
 
 
 Totals
 
841
 
432
 
409
 
 
      
  Total N (%)
 
Liberal
N (%)
 
Restrictive
N (%)
 
 
 Cardiovascular Disease
 
278 (33.1)
 
141 (32.6)
 
137 (33.5)
 
 
 Cancer
 
103 (12.2)
 
54 (12.5)
 
49 (12.0)
 
 
 Infection
 
78 (9.3)
 
41 (9.5)
 
37 (9.0)
 
 
 Stroke
 
57 (6.8)
 
27 (6.3)
 
30 (7.3)
 
 
 Dementia
 
108 (12.8)
 
56 (13.0)
 
52 (12.7)
 
 
 Pulmonary
 
58 (6.9)
 
29 (6.7)
 
29 (7.1)
 
 
 Other
 
147 (17.5)
 
79 (18.3)
 
68 (16.6)
 
 
 Unknown
 
12 (1.4)
 
5 (1.2)
 
7 (1.7)
 
 
 Totals
 
841
 
432
 
409
 
 

Disclosures

Magaziner:Ammonett : Consultancy; Sanofi: Consultancy; Regeneron: Consultancy; Novartis: Consultancy; Eli Lilly: Consultancy; American Orthopedic Association: Consultancy.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution