The association between platelet transfusion and mortality in patients with sepsis-induced thrombocytopenia: a propensity score-matching analysis

Background: Thrombocytopenia is common among sepsis patients. Platelet transfusion is frequently administered to increase platelet counts but its clinical impacts remains unclear in sepsis-induced thrombocytopenia. Therefore, we intended to explore the association between platelet transfusion and mortality in patients with sepsis-induced thrombocytopenia. Methods: The study was based on the Medical Information Mart for Intensive Care (MIMIC) III database. Sepsis patients were divided into two groups: platelet transfusion group (PT group) and no platelet transfusion group (NPT group). The primary outcome was 28-day mortality, and secondary outcomes were the length of ICU stay (LOS-ICU) and survival days at 28 days. Propensity-score matching was used to reduce the imbalance. Results: 1733 patients were included: 296 patients were in the PT group and 1437 in the NPT group. Overall, 655 patients died by day 28. After propensity score matching, 294 paired patients constituted each group. 28-day mortality did not decrease in PT patients comparing with NPT patients (120(40.54%) vs. 535(37.23%) deaths; p=0.29). LOS-ICU was similar between the PT group and the NPT group (5.84(2.68-11.78) vs. 4.94(2.18-12.72); p=0.44). After confounders were adjusted, it showed no difference between the PT group and the NPT group in 28-day mortality (hazard ratio [HR], 1.28;95% condence interval [CI], 0.96 - 1.17; p=0.09) or LOS-ICU (odds ratio [OR], 0.09; 95%CI, -1.45 - 1.62; p=0.91). Survival days at 28 days showed no difference between the groups according to Kaplan-Meier survival estimates. After propensity score matching, platelet transfusion was also not associated with 28-mortality or LOS-ICU. Conclusions: In the propensity score-matched analysis, it showed that platelet transfusion maybe not associated with 28-day mortality and the length of ICU


Background
Sepsis is the most common disease in the intensive care unit (ICU), de ned as a life-threatening syndrome of organ dysfunction which was caused by the dysregulated host response to severe infection 1 , and it has been considered a major cause of health loss worldwide. According to a recent scienti c publication, there were about 48.9 million cases and 11 million sepsis-related deaths worldwide in 2017, which accounted for almost 20% of all global deaths 2 .
Clinically, platelet count decrease is common among patients with sepsis who are admitted to the ICU.
The incidence of sepsis-induced thrombocytopenia is around 25% on ICU admission 3 and approximately 55% during the hospital stay 4 . Studies have con rmed that platelets play a crucial role in in ammatory balance, immune responses, tissue repair, and regeneration, beyond their important role in hemostasis and thrombosis [5][6][7] . Based on recent studies, thrombocytopenia is closely associated with multiple organ dysfunction syndromes, prolonged ICU stay, and high mortality in sepsis patients 8 . Besides, thrombocytopenia is an early prognostic marker for ICU patients in the rst 24h of septic shock onset 9 . And non-resolution of thrombocytopenia is associated with increased 28-day mortality in this population 3 .
In general, after controlling the infection and improving the patient's conditions, sepsis-induced thrombocytopenia is gradually returning to be normal. Recently, several studies have showed that recombinant human thrombopoietin (rhTPO) can lead to the quick recovery of the platelet count and improve the prognosis of patients with sepsis-induced severe thrombocytopenia [10][11][12] . And platelet transfusion is the most common clinical therapy to increase platelet counts. In theory, one standard unit dose of platelet transfusion can elevate platelet counts by 2010 9 /L. However, platelet transfusion is limited in clinical practice and has a rigorous indication in sepsis patients 13,14 because of resource scarcity, transfusion-related immune, and infectious complications. Thus, large prospective clinical trials are restricted in the exploration of the platelet transfusion impact on sepsis-induced thrombocytopenia.
According to a recent large registry study, it showed that platelet transfusion was not associated with the increased risk of death in critically ill patients 15 . Nonetheless, there is no large study based on platelet transfusion in severe sepsis-induced thrombocytopenia investigating whether platelet administration could improve the prognosis of sepsis patients. In this study, we aimed to determine the potential association between platelet transfusion and clinical outcomes including 28-day mortality and the length of ICU stay.

Database
This was a retrospective study based on an online international database, Medical Information Mart for Intensive Care III (MIMIC III), compromising the information of 46,520 critically ill patients who have been admitted to the Beth Israel Deaconess Medical Center from 2001 to 2012 16 . All the patients in the database were de-identi ed, and the need for informed consent was waived. One author (AZ) obtained access to this database (certi cation number 35752875) and was responsible for data extraction.

Study population and de nitions
Septic patients with a platelet count level ≤ 5010 9 /L were eligible for inclusion in our study. And sepsis was de ned according to the third sepsis de nition 1 , which was extracted as suspected infection and an acute change in cases that total Sequential Organ Failure Assessment (SOFA) score ≥ 2 points. For patients who were readmitted to the ICU, only the rst ICU admissions were included. For patients younger than 18 years or older than 89 years were excluded. The primary outcome was 28-day mortality. The secondary outcome was the length of ICU stay (LOS-ICU).
Propensity score matching PSM was used to minimize the imbalance of the confounding factor between the PT and the NPT groups. A one-to-one nearest neighbor matching algorithm was applied with a caliper width of 0.02 in our study. The following variables were selected to generate the propensity score: age, gender, SOFA, Simpli ed Acute Physiology score II (SAPS II), platelet count, platelet transfusion, diabetes mellitus, hypertension, chronic pulmonary, congestive heart failure, cancer.
The management of missing data Variables with missing data are common in the MIMIC III database. For C-reactive protein, serum lactate, albumin, and procalcitonin values, more than 20% were missing and were removed from this analysis.
For other continuous variables with missing values less than 5%, the missing values were replaced by the mean or median values.

Statistical analysis
Continuous variables were depicted as means with standard deviations or medians with interquartile ranges (IQRs), as appropriate. Student's t test, analysis of variance, Wilcoxon rank-sum test, or Kruskal-Wallis test were used, as appropriate. Categorical data were shown as frequencies and proportions, and they were compared using the X2 test. The association between platelet transfusion and 28-day mortality was determined by logistic regression including the baseline as the covariate and the group as a xed factor. The survival outcomes comparisons between the groups were analyzed by the log-rank test. PSM was used to minimize the imbalance between groups. A two-tailed test was performed, and p<0.05 was considered to indicate statistical signi cance. All statistical analyses were performed using the R package (version 3.6.3).

Baseline characteristics
The data of 1733 patients were included. The ow chart of patient selection is presented in Fig. 1. The overall 28-day mortality rate was 38.9%. The comparisons of the baseline characteristics are listed in Table 1. Patients in the PT group were younger than those in the NPT group (58.17±14.56 vs. 61.00±15.14, p=0.001). The platelet count was signi cantly lower in patients with platelet transfusion Association between platelet transfusion and patient outcomes On univariable analysis of risk of 28-day mortality was similar between the PT group and the NPT group (HR, 1.15; 95%CI, 0.89 -1.48; p=0.29). After confounders(platelet count and SOFA score) were adjusted, it revealed that platelet transfusion was also not associated with the overall 28-day mortality (HR, 1.28; 95%CI, 0.96 -1.71; p=0.09) or LOS-ICU(OR, 0.08; 95%CI, -1.45 -1.61; p=0.92) ( Table 2). Also, there was no signi cant difference on the survival days at 28 days between the groups, according to Kaplan-Meier survival estimates (Figure 2) Outcomes after propensity score matching After PSM, 294 cases from each group were matched by a 1:1 matching algorithm (Table 3). To assess the overall quality of the matched sample, the standardized difference of the means and the ratio of the variances between the propensity scores of both groups were compared, and the propensity scores between the groups was also inspected. There was no signi cant difference between the two matched groups with regards to all eleven covariates(age, gender, SOFA, SAPSII, platelet count, diabetes mellitus, hypertension, chronic pulmonary, congestive heart failure, cancer) (

Discussion
The present study demonstrates that platelet transfusion is not associated with 28-day mortality or LOS-ICU in sepsis patients with severe thrombocytopenia. This result was robust in the PSM analysis after adjustment for covariates including platelet counts and SOFA scores. According to our ndings, it seems that platelet transfusion is not a reasonable choice to rescue sepsis-induced thrombocytopenia for improving the prognosis of sepsis patients with severe thrombocytopenia.
A low platelet count, thrombocytopenia, commonly occurs in sepsis patients. According to previous research, nearly 35% to 59% of patients with sepsis develop thrombocytopenia 17,18 , which has been recognized as an independent risk factor for mortality and a marker for disease severity 19 . Moreover, it is also an important index to evaluate the prognosis of patients 20 . Sepsis patients with a low platelet count or dynamic thrombocytopenia show a poor prognosis and increased mortality 21 . The mechanisms of sepsis-induced thrombocytopenia are complex and probably involve various factors. For instance, endothelial dysfunction is a major consequence of sepsis and plays a crucial role in platelet activation and consumption 22 . This activation which results in aggregation is increased locally by cytokine production 23 . Besides, altered thrombopoiesis and/or hemophagocytosis is the major causes of thrombocytopenia, which would be potentiated by sepsis mediators 24 . In addition, uid resuscitation and surgical operation may have an in uence on platelet count.
Sepsis patients with platelet counts less than 5010 9 /L are considered to have sepsis-induced thrombocytopenia 25 , which has a high mortality and poor prognosis. It has been reported that non-resolution of thrombocytopenia was associated with increased 28-day mortality, instead of thrombocytopenia itself 3 . Currently, there is no effective treatment for this condition. Infection control, organ support therapy and immune response regulation remain the mainstream treatments. In recent years, recombinant human thrombopoietin (rhTPO) was reported improving platelet count and reducing platelet transfusion possibility among patients with severe sepsis and thrombocytopenia in a prospective study 10 . In another research, it suggested that the rescue therapy with rhTPO could rapidly lead to a recovery of the platelet count, increase survival days and reduce the 28-day mortality in sepsis patients with severe thrombocytopenia 12 14 . In our study, we found nearly the same result among sepsis patients with thrombocytopenia, that platelet transfusion was not associated with increased risk of 28-day mortality or LOS-ICU. Regarding to our negative nding, patients with sepsisinduced thrombocytopenia may not bene t from platelet transfusion, which may also put patients into a potential risk of infection 28 . Therefore, it may be reasonable to avoid unnecessary transfusions.
In present study, there are still several limitations. Firstly, as a retrospective design, the adjustment of relevant but missing data was not allowed. Although we did perform propensity score matching to reduce the imbalance, estimation of the propensity score could only be based on the acquirable data. Secondly, bacteria species and sources were not recorded in our data and the purpose of platelet transfusion was also unknown. Thus, these two aspects could not be included in the analysis. Thirdly, patients who have been administrated with β-lactam or sulfa antibiotics, which may result in antibiotic-induced thrombocytopenia, did not excluded in this study. Lastly, well-organized prospective randomized clinical trials are required to analyze the role of platelet transfusion in sepsis-induced thrombocytopenia and identify patients most likely to bene t from platelet transfusion.

Conclusions
In our study, we found that platelet transfusion maybe not associated with 28-day mortality and the length of ICU stay in patients with sepsis-induced thrombocytopenia. And further prospective studies will be needed in the future to address the bene ts and harms of platelet transfusion in sepsis patients with thrombocytopenia, to distinguish different subtypes of sepsis for which platelet transfusions would be of bene t, and to develop optimal platelet transfusion thresholds to better guide clinical treatment.  Tables   Due to technical limitations, table 1 to 3 is only available as a download in the Supplemental Files section. Figure 1 Flow chart of patient selection from the MIMIC III database