Mladsi DM, Barnett CL, Kaye JA. Identification of potential bias from data sources and study designs used to estimate platelet transfusion adverse event rates. Poster presented at the 2018 ISPOR 21st Annual European Congress; November 12, 2018. Barcelona, Spain.


OBJECTIVES: Optimal use of prophylactic platelet transfusions requires knowledge of adverse event (AE) risks. Rates of several AEs were published in a consensus AABB guideline (Kaufmann et al., 2015); however, these estimates come from studies employing various data sources and study designs. The objective of this review was to help those considering prophylactic use of platelet transfusions better understand the risks summarized in consensus documents and uncertainty surrounding these estimates.

METHODS: We reviewed the publications upon which the AE rates in the AABB guidelines were based and applied the hierarchical system for rating levels of evidence promoted by the Center for Evidence-based Medicine (CEBM). We also validated the rates presented in the guidelines against source documents cited.

RESULTS: The AABB guideline cites 7 sources, 1 per AE, as evidence of AE rates associated with platelet transfusion. Two were randomized controlled trials assessing the frequency of acute AEs after alternative platelet preparation strategies; 2 were published surveillance reports; and 3 were cited as “personal communication,” with no study characteristics provided (but apparently based on analyses of surveillance data from the American Red Cross). We could not consider preponderance of evidence for any AE because each rate estimate was based on a single study. The RCTs on which 2 AEs were based were found to be of high quality (level 1b per CEBM, 2009); however, the risks presented in the guidelines were not calculable from the published trial data. The remaining 5 sources were surveillance data, which are likely to underestimate AE rates.

CONCLUSIONS: AE rates for platelet transfusions reported in the AABB consensus guideline are likely biased due to suboptimal data sources, study designs, and validation. Consensus documents would better support decision making by making more transparent the basis for estimating AE rates, including assessing quality of evidence and validation.

Share on: