Introduction
Methods
Data sources and searches
Study selection
Data extraction
Study quality assessment
Data synthesis and analysis
Ethical approval
Results
Study selection
Study | Study designation | Inclusion population | Exclusion population | Adjustment for confounding |
---|---|---|---|---|
Vincent et al. [2] | Prospective, multiple center, observational study in western Europe | All ICU patients | Not reported | Patients’ admission variables |
Corwin et al. [3] | Prospective, multiple center, observational cohort study in the United States | Not reported | Admission to a pediatric, cardiothoracic, cardiac, neurologic, or burn ICU; renal failure on dialysis | Propensity to receive a transfusion |
Robinson et al. [32] | Retrospective, observational study | Not reported | Patients with blunt hepatic, splenic, or both injuries | Shock indices and associated injury severity |
Croce et al. [31] | Retrospective, observational study | Patients with blunt injury and ISS <25, survival of at least 48 hours, and no blood transfusion within the first 48 hours from admission | Patients who received any transfusion within the first 48 hours from admission, ISS > =25 | We use original numbers to analyze RR, because the author reported a distracting result (OR2.46,95%CI, 3.17 to 11.56) |
Taylor et al. [19] | Prospective, observational, cohort study | All ICU admissions | Not reported | Unadjusted |
Netzer et al. [33] | Prospective, cohort study | Patients with ALI/ARDS | Patients were excluded if they had current or prior congestive heart failure, respiratory disease, or conditions that mimicked ALI/ARDS, including vasculitis with diffuse alveolar hemorrhage; were burned 30% of total body area; or were lung or bone marrow recipients. | Age, gender, APACHE III score, and precipitating event |
Ruttinger et al. [34] | Retrospective, observational cohort study | All consecutive cases admitted immediately or delayed after a surgical procedure | Patients who had not undergone surgery during their present hospital stay and who had been admitted only for medical reasons, and patients with a rapidly fatal clinical course or with minimal disease severity | Admission variables, maximum APACHE II score, maximum number of failing organs, duration of invasive ventilation, duration of catecholamine therapy, and duration of renal replacement therapy |
Vincent et al. [35] | Prospective, multicenter, observational study | All ICU patients | Not reported | Sex and age, type of admission, main medical history, fluid balance, SAPS II, and severity of organ dysfunction on admission as SOFA score |
Bochicchio et al. [36] | Prospective | Trauma patients admitted >48 hours to the ICU | Not reported | Age, sex, race, and ISS |
Bursi et al. [37] | Retrospective observational study | Stable patients after elective major vascular surgery | Patients who had hemorrhagic hypovolemic shock requiring emergency RBC transfusion, severely anemic | Baseline characteristics, surgical risk, bleeding, presence of anemia, and propensity to receive transfusion |
Engoren et al. [38] | Retrospective study | All ICU patients | Cardiac surgical patients | APACHE II scores and propensity to receive a transfusion |
Sakr et al. [21] | Retrospective study | All surgical ICU patients | Not reported | Patients’ propensity to receive a transfusion |
Parsons et al. [39] | A secondary analysis | Patients with new-onset ALI, sepsis and shock | Patients with trauma or multiple transfusion | Age, sex, race, randomization arm and APACHE III score |
Sheth et al. [40] | Retrospective, observational cohort study | Patients with intracerebral hemorrhage | Patients younger than 18 years of age or with ICH secondary to antecedent head trauma, acute ischemic stroke with hemorrhagic transformation, brain tumor, vascular malformation, venous thrombosis, vasculitis of the central nervous system, hematological malignances, blood dyscrasia, or coagulopathy | Anemia, warfarin use, admission GCS score, hematoma volume, hematoma location, and DNR/CMO status |
Park et al. [4] | Prospective, multicenter observational study | Patients with severe sepsis or septic shock | Not reported | Propensity to receive a transfusion |
Brophy et al. [41] | A cross-sectional retrospective study | Anemia and renal dysfunction | Patients with anemia of neoplastic diseases or those receiving chemotherapy | Age, race, sex, ICU LOS, ESA use, transfusion status, mechanical ventilation or CPAP status, vasopressor use, severity of, illness, and presence of, following comorbid conditions, GI bleed, sepsis, and neurologic injury. |
Silva et al. [42] | Prospective observational cohort study | All ICU admissions | Acute coronary syndrome, ischemic stroke, acute hemorrhage, prior transfusion, pregnant women and Jehovah’s Witnesses | Sex, origin, previous disease, ventilation mode |
Sekhon et al. [43] | Retrospective cohort study | Severe TBI patients | Nontraumatic etiology, consciousness, concomitant traumatic quadriparesis | Age, admission GCS score, insertion of EVD, mean 7-day hemoglobin |
Risk of bias assessment
Synthesis of results
Number of studies excluded | Number of studies included | Effect size (95%CI) | Homogeneity index, Q | Between-study variability, I 2 | |
---|---|---|---|---|---|
Excluded studies did not give confounder-adjusted estimates | 16 | 1.211 (0.975 to 1.505) | 114.881, P <0.01 | 86.943 | |
Excluded studies got a `moderate’ or `low’ risk of bias | 15 | 1.178 (0.937 to 1.481) | 144.866, P <0.01 | 90.336 | |
Excluded studies did not reported RR or HR as effect size measurement | 6 | 0.901 (0.622 to 1.305) | 31.039, P <0.01 | 91.391 |
Number of studies included | Effect size (95%CI) | Homogeneity index, Q | Between-study variability, I 2 | |
---|---|---|---|---|
Type of patient | ||||
All ICU | 1.513 (1.123 to 2.039) | 39.822, P <0.01 | 84.933 | |
Sepsis and shock | 0.831 (0.203 to 3.413) | 8.041, P = 0.005 | 87.563 | |
Surgical | 1.174 (0.682 to 2.023) | 6.793, P = 0.033 | 70.556 | |
Trauma | 2.705 (0.815 to 8.973) | 30.695, P <0.01 | 93.484 | |
Other | 0.940 (0.417 to 2.116) | 15.505, P <0.01 | 87.101 | |
Outcome measurement | ||||
Count data | 3.969 (2.023 to 7.788) | 28.936, P <0.01 | 96.544 | |
OR | 1.465 (1.049 to 2.045) | 30.191, P <0.01 | 70.190 | |
RR and HR | 0.901 (0.622 to 1.305) | 31.039, P <0.01 | 91.391 | |
Adjustment | ||||
Unadjusted | 3.933 (2.107 to 7.343) | 28.936, P <0.01 | 96.544 | |
No propensity matched but multiple adjusted | 1.358 (0.965 to 1.910) | 25.626, P <0.01 | 68.782 | |
Propensity score matched | 1.089 (0.767 to 1.546) | 60.523, P <0.01 | 90.086 | |
Leukoreduced usage | ||||
Not reported | 1.851 (1.229 to 2.786) | 86.863, P <0.01 | 88.488 | |
<50% | 1.155 (0.583 to 2.290) | 52.156, P <0.01 | 96.165 | |
~76% | 0.966 (0.412 to 2.267) | 14.812, P <0.01 | 93.249 | |
~100% | 1.031 (0.452 to 2.352) | 1.261, P = 0.26 | 20.696 |
Number of studies | Coefficient (95% CI) | P value | |
---|---|---|---|
Age of patient | -0.0417 (-0.0680 to -0.0154) | 0.002 | |
Publication year | -0.1056 (-0.2103 to -0.0009) | 0.048 | |
Admission Hb | -0.1848 (-0.4343 to 0.0648) | 0.147 | |
Units transfused | -0.1676 (-0.3856 to 0.0504) | 0.132 | |
Pretransfusion Hb | -0.2157 (-0.8588 to 0.4275) | 0.511 | |
APACHE II | 0.0202 (-0.0569 to 0.0973) | 0.607 |
Publication bias
Discussion
Limitations
Conclusions
Key messages
-
There was no consistent evidence to link RBC transfusion with mortality in ICU patients.
-
In studies that had adjusted for confounders, RBC transfusion was not associated with in-hospital mortality in ICU patients.
-
The discrepancies on risk of mortality and RBC transfusion between previous studies may be due to different types of patients, information on leukoreduction, study design and statistical methods, age of patients, and publication year.