The association between platelet transfusions and mortality in patients with critical illness

Abstract

BACKGROUND: Platelet (PLT) transfusions are frequently administered in the setting of critical illness but their clinical impacts remain unknown. This study examined the association between PLT transfusions and death in a large intensive care unit (ICU) patient population. STUDY DESIGN AND METHODS: Using a transfusionregistry spanning 2008 to 2015, this study assessed effect of in-ICU PLT transfusions on ICU and in-hospital mortality using a stratified, time-dependent Cox proportional hazards model adjusted for illness severity, thrombocytopenia, and other confounders. Patients with known malignancy were excluded. Exposure to PLT transfusions were analyzed dichotomously (ever or never transfused) and continuously (number of transfusions). Medical, general surgery, and cardiac surgery subgroups were analyzed separately. RESULTS:Overall 32,842 adult patients were admitted to ICU, and 4927 patients received PLT transfusion(s). Crude in-ICU and in-hospital mortality were higher for PLT-transfused patients compared to nontransfused patients (9.2% vs. 6.7% and 12.3% vs. 9.3%, respectively). After confounders were adjusted for, PLT transfusions (ever vs. never) were not associated with increased mortality in ICU (hazard ratio [HR], 0.78; 95% confidence interval [CI], 0.60-1.02; p = 0.06) or in hospital (HR, 0.89; 95% CI, 0.68-1.09; p = 0.41). Continuous exposure analysis also showed no association between PLT transfusions and death. PLT transfusions have a protective effect on in-hospital mortality in the subgroup of general surgery patients (HR, 0.71; 95% CI, 0.51-0.99; p = 0.04; ever or never analysis). CONCLUSION:Platelet transfusions were not associated with increased risk of death in critically ill patients. Further studies are required to identify subgroups for which PLT transfusions may be beneficial.ABBREVIATIONS: HR = hazard ratio; ICU = intensive care unit; IQR(s) = interquartile range(s); MODS(s) = multiorgan dysfunction score(s). From the RESULTS Patient demographicsBetween 2008 and 2015, a total of 32,842 nononcology adult patients were admitted to one of three participating ICUs (Fig. 1). Median (IQR) age at admission was 66 (54-76) years and 11,944 (36.5%) of patients were female. Median (IQR) duration of ICU stay was 2 (1-4) days. Thrombocytopenia Volume 59, June 2019 TRANSFUSION 1963 PLT TRANSFUSION AND DEATH IN THE ICU * Data are reported as number (%) or median (IQR). † Cardiac surgery with cardiopulmonary bypass. ‡ Bleeding disorders include qualitative PLT defects, von Willebrand's disease, acquired factor deficiencies. § Mechanical ventilation may either be invasive or noninvasive. k MODS = multiple organ dysfunction score at baseline (range, 0-20). MODS were available for 29,550 (90.0%) of the overall patient population.

Thrombocytopenia in intensive care unit patients: A scoping review

“…Identified studies eligible for inclusion were published in the period from 1983 to 2019. Six of the included studies were case‐control studies, 5,19‐23 one was an RCT, 24 five were post hoc analyses of RCTs, 7,25‐28 one study was a protocol for an RCT, 29 and 57 studies were cohort studies 3,6,8‐14,30‐77 …”

“…Of the 57 cohort studies, 27 (47%) were retrospective, 6,8‐11,37,49,50,52‐59,61‐64,68,71‐76 25 (44%) were prospective, 3,12‐14,30‐36,38‐48,65,67,70 while 5 (9%) had other or unstated design 51,60,66,69,77 . Nine (36%) of the prospective, 14,31,34,35,38,43,45,46,65 and 8 (31%) of the retrospective cohort studies 9,50,58,68,72‐74,76 were multi‐center studies.…”

The epidemiology of platelet transfusions: an analysis of platelet use at 12 US hospitals

“…The REDS-III Recipient and Component databases represent one of the largest data sets available addressing both component and patient clinical characteristics, and we estimate it represents roughly 3-5% of all platelets transfused annually in the United States 2,3 . Other large international databases are available that have been used to address a number of donor and patient transfusion issues, but may lack the granularity and depth of the REDS-III database [11][12][13][14][15]19 . Our data confirm that apheresis platelets are the most common platelet transfusion product, though our data shows less apheresis use than national surveys due to one of the four HUBS manufacturing large numbers of WBD platelets.…”

“…They found that median pre-transfusion platelet count was 55,000/µL for medical ICU patients and 95,000/µL for general surgery patients. They emphasized the need for further prospective studies to address the benefits and harms of platelet transfusions in critically ill patients 11 .…”

“…However, analyses of large databases that contain granular information on platelet use has been lacking. We are aware of only a few including the Society of Thoracic Surgery database on blood use in cardiac surgery that have provided substantial data on platelet use [11][12][13][14][15] .…”