This review assessed the efficacy and effectiveness of red blood cell leukoreduction. The authors concluded that leukoreduced transfusion may reduce post-operative infections and a larger sample size might have shown a reduction in mortality. The review methods were not fully described and statistical differences between the studies were not assessed. These issues cast doubt on the reliability of the authors' conclusions.

To assess the efficacy and effectiveness of red blood cell (RBC) leukoreduction in reducing post-operative infection, mortality and cancer recurrence.

MEDLINE was searched from 1966 to May 2003; the search terms were stated. The reference lists of selected reports and reviews were examined.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) were eligible for inclusion.

Specific interventions included in the review

Studies that compared leukoreduced allogeneic RBC transfusions with non-leukoreduced allogeneic RBC transfusions were eligible for inclusion. Studies in which either treatment group received only autologous blood were excluded. The included studies used various types of filters, including bedside and pre- and post-storage. The control interventions were whole blood, buffy core-depleted and standard transfusions.

Participants included in the review

Studies of adults undergoing surgery were eligible for inclusion. The participants included patients undergoing colorectal, cardiac and gastrointestinal surgery. One study of patients undergoing surgery for burns was excluded post hoc.

Outcomes assessed in the review

The review assessed post-operative infection, mortality and the recurrence of cancer. Studies evaluating non-clinical outcomes were excluded. The included studies used various time periods in their definitions of post-operative mortality.

How were decisions on the relevance of primary studies made?

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

The review was restricted to RCTs, with information on blinding, randomisation, the proportion of patients randomised but not transfused, and withdrawals extracted. Three reviewers extracted data relating to validity criteria, and these were discussed in the narrative.

Three reviewers extracted the data using a standardised form. Any discrepancies were resolved by consensus. Outcome data were extracted separately for all patients randomised and for only patients who were actually transfused. The relative risk (RR) and 95% confidence interval (CI) of developing an adverse event were calculated for each study.

How were the studies combined?

The pooled RR and 95% CI were calculated using the random-effects model of DerSimonian and Laird. The data were pooled separately for all patients randomised and for only patients who were transfused.

How were differences between studies investigated?

Subgroup analyses were used to examine the effects on the results of type of filter and type of surgery. Heterogeneity did not appear to have been assessed statistically.

Ten RCTs (n not reported) were included.

The proportion of patients randomised but not transfused ranged from 2 to 73% across the 8 studies reporting this information. Four studies analysed patients according to the allocated treatment group, one study analysed patients according to the product they received, and one study excluded patients who had received the wrong product from analysis. The other 4 studies did not report protocol deviations.

Post-operative infection.

The reduction in post-operative infection with leukoreduced transfusion compared with non-leukoreduced transfusion was not statistically significant for all patients randomised (n=3,073; RR 0.76, 95% CI: 0.54, 1.08), but was statistically significant for transfused patients (n=2,358; RR 0.60, 95% CI: 0.38, 0.93).

Post-operative infection was statistically significantly reduced with a bedside filter for all patients randomised (RR 0.38, 95% CI: 0.26, 0.54), for transfused patients (RR 0.19, 95% CI: 0.06, 0.58), for all cardiac patients randomised (RR 0.82, 95% CI: 0.63, 1.00) and for transfused cardiac patients (RR 0.77, 95% CI: 0.61, 0.97). There was no statistically significant difference between leukoreduction using pre- or post-storage filters and non-leukoreduced transfusion.

Mortality.

Mortality was not statistically significantly reduced with leukoreduced transfusion for all patients randomised (n=5,597; RR 0.71, 95% CI: 0.45, 1.13) or for transfused patients (RR 0.61, 95% CI: 0.36, 1.04).

Cancer recurrence (1 study, n=not reported). Based on 1 study, there was no statistically significant difference between leukoreduced transfusion and non-leukoreduced transfusion in cancer recurrence for all patients randomised (27.9% versus 27.8%, P>0.05) or for transfused patients (28.4% versus 31.2%, P>0.05).

The transfusion of leukoreduced RBCs may reduce post-operative infections. If studies had recruited a larger sample size, they might have shown a reduction in mortality.

The review question was clear in terms of the study design, intervention, participants and outcomes. Only one database was searched and no attempt was made to locate unpublished studies, thus raising the possibility of missing studies and publication bias, which the authors did not investigate. It was also unclear whether language limitations had been applied. The methods used to select the studies were not described, so it is not known whether any efforts were made to reduce errors and bias at this stage. However, methods were used to minimise bias in the extraction of data. Only RCTs were included and some information regarding quality criteria extracted, but the quality of the included studies was not formally assessed with a recognised quality tool.

The meta-analyses were performed without assessing statistical heterogeneity. There was insufficient information on the included studies to assess the extent of clinical heterogeneity among studies; in particular, no information was given about how post-operative infection was defined. It was also difficult to judge whether statistical heterogeneity was present because relevant meta-analysis graphs were not provided. Some potential sources of heterogeneity among the studies were examined. The authors did not estimate the minimum sample size required to detect a difference between treatments in mortality, so it was not possible to assess whether the lack of difference in mortality was due to an inadequate sample size. In light of the methodological flaws in the review, the authors' conclusions should not be viewed as robust.

Practice: The authors did not state any implications for practice.

Research: The authors stated that further research, in a variety of patient populations, is required to ensure the results of the review are accurate and generalisable. More studies using cancer recurrence as an outcome are also needed. The authors stated that large double-blinded trials with randomisation at the point of transfusion would best compare leukoreduced transfusions with standard non-filtered blood. However, they also stated that in countries that have already adopted a universal policy of leucoreduction, large before-and-after studies may be the best option of assessing effectiveness.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.