REVIEW Platelet transfusion: a systematic review of the clinical evidence Ambuj Kumar,1,2 Rahul Mhaskar,1* Brenda J. Grossman,3 Richard M. Kaufman,4 Aaron A.R. Tobian,5 Steven Kleinman,6 Terry Gernsheimer,7 Alan T. Tinmouth,8 and Benjamin Djulbegovic,1,2 on behalf of AABB Platelet Transfusion Guidelines Panel
BACKGROUND: Platelet (PLT) transfusion is indicated either prophylactically or therapeutically to reduce the risk of bleeding or to control active bleeding. Significant uncertainty exists regarding the appropriate use of PLT transfusion and the optimal threshold for transfusion in various settings. We formulated 12 key questions to assess the role of PLT transfusion. STUDY DESIGN AND METHODS: We performed a systematic review (SR) of randomized controlled trials (RCTs) and observational studies. A comprehensive search of PubMed, Web of Science, and Cochrane registry of controlled trials was performed. Methodologic quality of included studies was assessed and a metaanalysis was performed if more than two studies with similar designs were identified for a specific question. RESULTS: Seventeen RCTs and 55 observational studies were included in the final SR. Results from RCTs showed a beneficial effect of prophylactic compared with therapeutic transfusion for the prevention of significant bleeding in patients with hematologic disorders undergoing chemotherapy or stem cell transplantation. We found no difference in significant bleeding events related to the PLT count threshold for transfusion or the dose of PLTs transfused. Overall methodologic quality of RCTs was moderate. Results from observational studies showed no evidence that PLT transfusion prevented significant bleeding in patients undergoing central venous catheter insertions, lumbar puncture, or other surgical procedures. The methodologic quality of observational studies was very low. CONCLUSION: We provide a comprehensive assessment of evidence on the use of PLT transfusions in a variety of clinical settings. Our report summarizes current knowledge and identifies gaps to be addressed in future research.
he number of patients receiving platelet (PLT) transfusions has increased over the past several years.1 PLT transfusions are used either prophylactically to reduce the risk of bleeding or therapeutically to control active bleeding.2 AABB previously estimated that more than 70% of PLT transfusions are performed prophylactically.3 Important questions for PLT transfusion therapy include the appropriate PLT count threshold for prophylactic transfusion, the dosage of PLTs that should be used, and the clinical conditions that should prompt transfusion. To address the existing uncertainties related to the appropriate indications for PLT ABBREVIATIONS: DIC = disseminated intravascular coagulation; GRADE = Grading of Recommendations Assessment, Development and Evaluation; HUS = hemolytic uremic syndrome; LP(s) = lumbar puncture(s); SR = systematic review; TBI = traumatic brain injury; TTP = thrombotic thrombocytopenic purpura. From the 1Division of Evidence-Based Medicine, Department of Internal Medicine, Morsani College of Medicine, University of South Florida, and the 2Moffitt Cancer Center, Tampa, Florida; the 3Washington University School of Medicine, St Louis, Missouri; 4Brigham and Women’s Hospital, Boston, Massachusetts; 5Johns Hopkins Hospital, Baltimore, Maryland; the 6Center for Blood Research, The University of British Columbia, Vancouver, British Columbia, Canada; the 7 University of Washington Medical Center and Seattle Cancer Care Alliance, Seattle, Washington; and the 8Ottawa Hospital and Research Institute, Ottawa, Ontario, Canada. Address reprint requests to: Ambuj Kumar, MD, MPH, Division of Evidence-Based Medicine, Department of Internal Medicine, University of South Florida & H. Lee Moffitt Cancer Center & Research Institute, 3515 East Fletcher Avenue, MDT 1201, MDC 27, Tampa, FL 33612; e-mail: [email protected]
*Equal contribution. Received for publication June 17, 2014; revision received October 9, 2014, and accepted October 9, 2014. doi: 10.1111/trf.12943 © 2014 AABB TRANSFUSION **;**:**-**. Volume **, ** **
KUMAR ET AL.
transfusions, we conducted a systematic review (SR) with the aim of synthesizing the current evidence for many of the common situations in which PLT transfusions are considered.
MATERIALS AND METHODS Development of clinical questions Clinical questions were developed by an AABB Guidelines Panel consisting of 21 members with expertise in transfusion medicine, hematology, surgery, anesthesiology, intensive care, and guidelines development method. The panel was charged with providing direction for development of this SR, which in turn was used to inform the development of clinical practice guidelines (reported in a separate document). This SR was performed according to the standards of The Cochrane Collaboration,4 using a standardized protocol and reported as per the PRISMA guidelines.5
Search methods for identification of studies We searched the Cochrane CENTRAL, PubMed, and Web of Science databases since inception until September 5, 2014. There was no restriction on eligibility criteria in terms of when the studies were performed or language. We first searched for existing guidelines/SR using the search filter developed by Haynes and colleagues.7 We manually scanned references of obtained articles from identified guidelines and SRs to identify other potentially relevant articles for inclusion in the review. We also canvassed members of the AABB Guidelines Panel to identify potentially eligible unpublished or ongoing studies (a detailed search strategy is shown in Appendix S1, available as supporting information in the online version of this paper). When deemed necessary, study authors were contacted for information not available or not well described in the original papers.
Data collection and analysis Criteria for considering studies for this review
Selection of studies
Types of studies
All titles, abstracts, and full-text reports were reviewed by two authors of this SR (AK, RM); further review was conducted by a third author (BD or RK) when there was disagreement about eligibility of the study for inclusion in this SR.
Randomized controlled trials (RCTs) and observational studies (prospective or retrospective cohort studies, casecontrol studies, and those with no control arm) were eligible for inclusion. Although all observational studies meeting the inclusion criteria were reviewed, data from these studies were not used when more than two RCTs addressed a particular question. The use of evidence summaries based only on RCTs was favored due to confounding-by-indication bias in observational transfusion studies: that is, PLT transfusion is often administered based on disease severity.6
Data extraction Data extraction forms were piloted. The final forms were used to extract data in duplicate, and a further review for accuracy was conducted by an independent member of the panel.
Assessment of risk of bias in included studies Types of participants Patients undergoing PLT transfusions in RCTs or observational studies were eligible for inclusion.
Types of interventions Apheresis or whole blood–derived (buffy coat or PLT-rich plasma) PLT concentrates whether used as therapeutic or prophylactic interventions for any clinical condition were eligible for inclusion. The comparison group (if any) was a medical management strategy without PLT transfusion or, in the case of the PLT dosing or PLT count threshold question, PLT transfusion strategy using a different dose or threshold.
Assessment of the risk of bias in the included RCTs was done according to The Cochrane Collaboration’s tool for assessing the risk of bias.4 The risk of bias in observational studies was assessed using the Newcastle Ottawa Scale,8 supplemented by the guidance paper by Wells and coworkers.9 Overall quality of the studies was categorized in four categories according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) method (details on GRADE method are provided in Appendix S2, available as supporting information in the online version of this paper).10 A formal assessment of publication bias was not performed, since a small number of studies (