Should we believe in transfusion-associated enterocolitis? Applying a GRADE to the literature

Numerous observational studies appear to demonstrate an association between packed red blood cell (pRBC) transfusions and necrotizing enterocolitis (NEC). However, the limited numbers of randomized controlled trials (RCTs) do not support a causal relationship between pRBC transfusion and NEC. We sought to determine the quality of the evidence behind transfusion-associated necrotizing enterocolitis (TANEC), and to formulate a GRADE-based recommendation regarding transfusion practices to reduce the risk of TANEC. A systematic search including MEDLINE, Embase, CINAHL, the Cochrane Central Register of Controlled Trials and clinical trials registries was performed for studies assessing the association between transfusion and NEC. Teams of two paired reviewers independently screened studies for eligibility, assessed risk of bias using the GRADE framework, and collected data from each eligible study. We examined studies for two time points following transfusion: within 48h if this was available, and otherwise at any time after transfusion. In total, 23 observational studies and three RCTs met inclusion criteria. The average rating for the quality of evidence of individual studies was between "very low" and "low." On pooling studies for GRADE review, we observed an inconsistency of results. This led to a final overall quality of "very low" for the evidence for an association between transfusions and necrotizing enterocolitis. The pooled outcome of NEC for observational/case control studies was an odds ratio of 1.13 (95% CI: 0.99-1.29) when TANEC was defined as occurring within 48 hours of transfusion. For NEC occurring at any time post-transfusion, the pooled OR was 1.95 (1.60-2.38). Conversely, the pooled outcome of NEC for the RCT data had an odds ratio of 0.6 (0.3, 1.21) with NEC being less frequent in the liberal transfusion group compared to the restrictive transfusion group. The overall quality of the evidence for TANEC is "very low," suggesting very little confidence in the effect estimate. RCT data tended toward apparent protection against NEC. The available evidence is not sufficient to support a practice recommendation around pRBC transfusions in the context of preventing the development of NEC.