Exploring heterogeneity in reported venous thromboembolism risk in COVID-19 and comparison to other viral pneumonias: a systematic review and meta-regression

BACKGROUND: Sources of heterogeneity in venous thromboembolism (VTE) risk in COVID-19 are unclear and comparisons to other viruses are lacking. OBJECTIVES: To describe VTE risk in patients with COVID-19, explore sources of heterogeneity, and make comparisons with other viral pneumonia. METHODS: PubMed and Embase data were searched on March 14, 2021, for studies on VTE in adults hospitalized with viral pneumonia. VTE risk estimates were pooled in a random effects meta-analysis stratified by virus type. Heterogeneity in COVID-19 was explored in multivariable meta-regression. RESULTS: Seventy studies in COVID-19 (intensive care [ICU] [47] vs ward [23]), 4 studies in seasonal influenza (ICU [3] vs ward [1]), 2 ICU studies in H1N1 and 1 ICU study in SARS-CoV-1 were included. For COVID-19 ICU, pooled VTE risk was 19.6% (95% confidence interval [CI], 16.2%–23.5; I(2) = 92.8%) for nonscreening studies and 30.0% (95% CI, 17.9%–45.7%; I(2) = 81.9%) for screening studies. For COVID-19 ward, pooled VTE risk was 3.4% (95% CI, 2.4%–4.7%; I(2) = 91.3%) and 22.5% (95% CI, 10.2%–42.7%; I(2) = 91.6%) for nonscreening and screening studies, respectively. Higher sample size was associated with lower VTE risk. Pooled VTE risk in seasonal influenza and H1N1 at ICU were 9.0% (95% CI, 5.6%–14.2%; I(2) = 39.7%) and 29.2% (95% CI, 8.7%–64.2%; I(2) = 77.9%), respectively. At ward, VTE risk of seasonal influenza was 2.4% (95% CI, 2.1%–2.7%). In SARS-CoV-1, VTE risk was 47.8% (95% CI, 34.0–62.0). CONCLUSION: Pooled risk estimates in COVID-19 should be interpreted cautiously as a high degree of heterogeneity is present, which hinders comparison to other viral pneumonia. The association of VTE risk in COVID-19 to sample size suggests publication bias.