Trends in COVID-19 hospitalisations, outcomes and lengths of stay in England during 2020.

Objectives: Robust outcome information for patients hospitalised for COVID-19 in England is lacking. We estimate monthly trends, accounting for baseline characteristics, for a cohort of hospitalised patients during 2020. Design: Retrospective cohort study using parametric multi-state mixture models. Setting: Hospital and ICU admissions at 31 National Health Service trusts in England collected as part of Public Health England's surveillance of Severe Acute Respiratory Infections (SARI-Watch) and linked to data on deaths. Participants: 20,785 adults (aged 15+) admitted with COVID-19 between 15 March and 31 December 2020. Outcome measures: Estimated monthly trends in hospitalised case-fatality risk (HFR), probability of ICU admission, and lengths of stay, by sex, age, region of residence, ethnicity and number of comorbidities, accounting for missing data and censored outcomes. Results: Between March and June/July/August 2020, the estimated HFR reduced from 31.9% (95% confidence interval 30.3-33.5%) to 10.9% (9.4-12.7%). HFR subsequently increased to 21.6% (18.4-25.5%) in September, rising steadily to 25.7% (23.0-29.2%) in December, with the steepest increases estimated for older patients and those with multiple comorbidities. Probability of ICU admission reduced from 13.9% (12.8-15.2%) in March to 6.2% (5.3-7.1%) in May, rose to 14.2% (11.1-17.2%) in September and fell to 9.6% (8.6-10.9%) in December. Among non-ICU patients, estimated median stay from hospital admission to death increased during the study period (from 6.6 to 12.3 days), as did median stay until discharge (from 6.1 to 9.3 days). Conclusions: Initial improvements in patient outcomes reflect improvements in clinical practice, availability of new therapies, and increased ward capacity. Significantly poorer outcomes were associated with older age, multimorbidity and admission at a time of increased hospital burden.