Mortality in Cancer Patients With COVID-19 Who Are Admitted to an ICU or Who Have Severe COVID-19: A Systematic Review and Meta-Analysis

PURPOSEThere are scarce data to aid in prognostication of the outcome of critically ill cancer patients with COVID-19. In this systematic review and meta-analysis, we investigated the mortality of critically ill cancer patients with COVID-19.METHODSWe searched online databases and manually searched for studies in English that reported on outcomes of adult cancer patients with COVID-19 admitted to an intensive care unit (ICU) or those with severe COVID-19 between December 2019 and October 2020. Risk of bias was assessed by the Modified Newcastle-Ottawa Scale. The primary outcome was all-cause mortality. We also determined the odds of death for cancer patients versus noncancer patients, as also outcomes by cancer subtypes, presence of recent anticancer therapy, and presence of one or more comorbidities. Random-effects modeling was used.RESULTSIn 28 studies (1,276 patients), pooled mortality in cancer patients with COVID-19 admitted to an ICU was 60.2% (95% CI, 53.6 to 6.7; I2 = 80.27%), with four studies (7,259 patients) showing higher odds of dying in cancer versus noncancer patients (odds ratio 1.924; 95% CI, 1.596 to 2.320). In four studies (106 patients) of patients with cancer and severe COVID-19, pooled mortality was 59.4% (95% CI, –39.4 to 77.5; I2 = 72.28%); in one study, presence of hematologic malignancy was associated with significantly higher mortality compared with nonhematologic cancers (odds ratio 1.878; 95% CI, 1.171 to 3.012). Risk of bias was low.CONCLUSIONMost studies were reported before the results of trials suggesting the benefit of dexamethasone and tocilizumab, potentially overestimating mortality. The observed mortality of 60% in cancer patients with COVID-19 admitted to the ICU is not prohibitively high, and admission to the ICU should be considered for selected patients (registered with PROSPERO, CRD42020207209).

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.