The proportion of patients with thrombocytopenia in three human‐susceptible coronavirus infections: a systematic review and meta‐analysis

Abstract

At the end of 2019, a novel coronavirus (2019-nCoV or COVID-19) infection broke out in Wuhan and then rapidly swept over China. It was the third instance of a coronavirus epidemic in the last two decades, preceded by the severe acute respiratory syndrome coronavirus (SARS-CoV) in 2003 and the Middle East respiratory syndrome coronavirus (MERS-CoV) in 2012. The above three viral infections have caused >4000 deaths to date, with 2019-nCoV accounting for over half. Haematological abnormalities have been found in almost every patient with coronavirus infection, which may contribute to mortality. According to some retrospective studies, the proportion of patients with thrombocytopenia significantly rises in critically ill patients.1-3 However, the reported proportion of coronavirus-infected patients with thrombocytopenia varies from 0·054 to 0·55.3 As the platelet number might be an indicator for the prognosis of coronavirus infection, it is a prerequisite to determine an exact value for the proportion of coronavirus-infected patients with platelet reduction during infection. It is necessary to further draw the attention of clinicians to platelets, besides typical clinical features and radiographic examinations. As infections caused by the three types of coronavirus share much in common, ranging from transmission patterns, clinical manifestations and diagnostic methods to the lack of effective therapies, a systematic review and meta-analysis was conducted to analyse the proportion of patients with thrombocytopenia in SARS-CoV, MERS-CoV and 2019-nCoV infections. PubMed and the Excerpta Medica dataBASE (EMBASE) databases were searched on 26 February 2020, with key terms relating to coronavirus and thrombocytopenia. Observational studies were selected to calculate the proportion of coronavirus-infected patients with thrombocytopenia, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A total of 159 articles were retrieved from the initial database search, 19 of which were finally included based on pre-defined selection criteria (Data S1), with a sample size of 2130 after removing duplicates, inspecting abstracts and full-text reviews (details in Table I1, 3-20). Then data were extracted, and the Newcastle–Ottawa Quality Assessment Scale (NOS) was used to evaluate the quality of the included studies. All the raw proportions of patients with thrombocytopenia extracted from the 19 studies were transformed with the Freeman–Tukey double arcsine method, and the pooled proportion of patients with thrombocytopenia was subsequently calculated as 0·30 (95% confidence interval [CI] of 0·24–0·37), with considerable heterogeneity (χ2 = 174·59, I2 = 90%, P < 0·01). Using the random-effect model (Forest plot in Fig 1), the results of the Egger's (t = −0·70, P = 0·50) and Begg's (z = −1·44, P = 0·151) tests indicated that there was no publication bias in our present analysis. To explore the heterogeneity, sensitivity analysis and regression diagnosis of linear and generalised linear models were implemented, which indicated that the Wong et al.3 and Xu et al.4 studies contributed to the heterogeneity (plots in Data S1). Meta-regression and subgroup analysis were used to further identify the sources contributing to the heterogeneity. It suggested that virus type (SARS-CoV, MERS-CoV or 2019-nCoV) and publication year (before 2009, before 2019 or after 2020) may be the sources of heterogeneity rather than the sample size (≤100 or >100), age (≤45 or >45 years), region (China or outside China), gender ratio (male: female ≤1 or >1), or NOS score (<5 or ≥5). The P values of the virus type and publication year were P = 0·01 for multiple-moderator and P = 0·02 for single-moderator meta-regression. In subgroup analyses, the heterogeneity was critically mitigated by the pieces of evidence of the between-group P value (P < 0·0001) and the significant decline of I2 in both the MERS-CoV and 2019-nCoV subgroups (Data S1. Moreover, the I2 decreased from 90% to <75% when grouping was based on normal platelet range. In conclusion, our present study found that the estimated proportion of patients with thrombocytopenia and infected with SARS-CoV, MERS-CoV or 2019-nCoV was 0·30 (95% CI 0·24–0·37). The heterogeneity of the present study is ascribed to the different types of viruses and the publication year of the reports. The diverse values for the lower limits of normal platelet range and two specific studies also potentially accounted for the heterogeneity. Our present investigation showed that nearly 30% of coronavirus-infected patients had thrombocytopenia. However, there was critical heterogeneity amongst these studies. Based on sensitivity analysis, we found two suspicious studies (Wong et al.3 and Xu et al.4), which might have contributed to the heterogeneity. The severity of the illness was not clearly distinguished in either study; in the Wong et al. study3 38 patients were in intensive care units, while the number in the Xu et al. study4 was only one. According to the meta-regression and subgroup analysis, the results strongly indicated that the type of virus and the publication year of the studies were the main causes of heterogeneity. The infections caused by the three coronaviruses contain the same transmission pattern and the same series of symptoms. There are also no specific anti-coronavirus drugs that have been proven effective in vivo. However, these three types of virus have different gene sequences. The 2019-nCoV has 88% nucleotide similarity with SARS-CoV, but it only shares <50% homology with MERS-CoV.21 In the present study, the quality of published reports on coronaviruses appeared to improve over time from the subgroup analysis. Different criteria for thrombocytopenia also lead to heterogeneity according to the meta-analysis. Six of the 19 studies Table 1 had limits lower than 150 × 109/l for the normal platelet range, which excluded patients who should have been included in our present study. The phenomenon was demonstrated extremely in the Xu et al.4 study, with a proportion of patients with thrombocytopenia of 0·05, which was the lowest amongst all the studies. Another reason for the heterogeneity might lie in the incubation period, which means the time of the blood test could possibly represent diverse stages of the illness in individuals. The mechanism of coronavirus-related thrombocytopenia may be partly explained by the inflammation microenvironment created by coronaviruses and the cytokine storm afterwards, impairment of platelet biogenesis in lungs, as well as sequestration of platelets in capillary-rich organs, such as the lungs and liver.22-24 This work was supported by the National Natural Science Foundation of China (81873432 and 81670132), grants from the Jiangsu Province of China (BE2016665, SBE2016740635 and ZDRCA2016047), The Natural Science Foundation of the Jiangsu Higher Education Institution of China (18KJA320006), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). We declare no competing interests. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Topics & Concepts