Risk factors for excess all-cause mortality during the first wave of the COVID-19 pandemic in England: A retrospective cohort study of primary care data.

Background: The COVID-19 pandemic's first wave in England during spring 2020 resulted in an approximate 50% increase in all-cause mortality. Previously, risk factors such as age and ethnicity, were identified by studying COVID-related deaths only, but these were under-recorded during this period.

Objective: To use a large electronic primary care database to estimate the impact of risk factors (RFs) on excess mortality in England during the first wave, compared with the impact on total mortality during 2015-19.

Methods: Medical history, ethnicity, area-based deprivation and vital status data were extracted for an average of 4.8 million patients aged 30-104 years, for each year between 18-March and 19-May over a 6-year period (2015-2020). We used Poisson regression to model total mortality adjusting for age and sex, with interactions between each RF and period (pandemic vs. 2015-19). Total mortality during the pandemic was partitioned into "usual" and "excess" components, assuming 2015-19 rates represented "usual" mortality. The association of each RF with the 2020 "excess" component was derived as the excess mortality ratio (EMR), and compared with the usual mortality ratio (UMR).

Results: RFs where excess mortality was greatest and notably higher than usual were age >80, non-white ethnicity (e.g., black vs. white EMR = 2.50, 95%CI 1.97-3.18; compared to UMR = 0.92, 95%CI 0.85-1.00), BMI>40, dementia, learning disability, severe mental illness, place of residence (London, care-home, most deprived). By contrast, EMRs were comparable to UMRs for sex. Although some co-morbidities such as cancer produced EMRs significantly below their UMRs, the EMRs were still >1. In contrast current smoking has an EMR below 1 (EMR = 0.80, 95%CI 0.65-0.98) compared to its UMR = 1.64.

Conclusions: Studying risk factors for excess mortality during the pandemic highlighted differences from studying cause-specific mortality. Our approach illustrates a novel methodology for evaluating a pandemic's impact by individual risk factor without requiring cause-specific mortality data.