Prospective Cohort, Retrospective Cohort
Study population and setting
This study included 1044 adult patients (median age: 70 years, 57% male), who had active cancer and a positive SARS-CoV-2 PCR test between March 18 and May 8, 2020 at one of 61 participating centers of the UK Coronavirus Cancer Monitoring Project (UKCCMP). The analysis explored the association between cancer subtype and inpatient, all cause, case-fatality. Tumor types were classified according to ICD-10 codes. The UKCCMP population was also compared to all cancer patients recorded in ONS (Office of National Statistics) cancer census.
Summary of Main Findings
Descriptive, unadjusted comparisons of UKCCMP patients to the ONS cohort suggest that cancer patients with COVID-19 were somewhat more likely to be male (56.9% vs. 51.3%) but had similar age distributions. Hematological malignancies, such as leukemia, myeloma, and lymphoma, were slightly more common in UKCCMP patients than the ONS cohort. Overall, there were 319 deaths (92% due to COVID-19) among the 1044 UKCCMP cohort. After adjusting for age and sex, the only cancer subtype associated with increased risk of death was leukemia (OR 2.25, 95% CI: 1.13-4.57).
This was a relatively large study of cancer patients in the UK with prospective follow-up for outcomes.
The adjusted analyses only accounted for confounding due to age and sex, but confounding due to many other factors, including COVID-19 disease severity and other comorbidities, is likely and, thus, results may be biased. The primary outcome of interest was all-cause mortality, rather than COVID-19 specific mortality, which may influence observed results of specific cancer subtypes are more likely to have increased rates of death due to other causes (i.e. cancer itself). Missing data were excluded, which leads to bias when the population who have missing data are different than those without missing data. The comparison to the ONS cohort was unadjusted and selection bias, among other limitations, is very likely; that is, patients who are exposed to COVID-19, get tested, and test positive, are likely different than individuals who do not get tested or do not have exposures to COVID-19, which may explain some of the differences between these two cohorts. The length of follow-up was relatively short, and all outcomes due to COVID-19 might not have been captured.
This was one of the first and largest analyses to explore the association between cancer subtype and COVID-19 infection and outcomes.
This review was posted on: 19 December 2020