Table 1.

Review inclusion and exclusion criteria: what was the age-adjusted association between comorbidities and severe or critical care outcomes in hospital patients with COVID-19 in the early stages of the pandemic?

	Include	Exclude
Population	Adult COVID-19 hospital patients. Studies with 10 or more patients.	Samples nested in clinical trials, samples from cruise ships, familial clusters. Community cases not receiving care in hospitals, including general population estimates of the spread of COVID-19. Studies focusing solely on infants and children (not part of a study including adults).
Outcomes	Relative risk, hazard ratio, odds ratio associated with comorbidity (pre-existing condition, chronic illness) status on admission, of: i. progression to severe disease ii. admission to critical or intensive care unit iii. invasive or non-invasive ventilation iv. death in hospital v. any adverse event (i.e., composite indicators of any of i–iv), for any reported comorbidity.	Other treatments inside and outside critical or intensive care departments, e.g., rates of patients receiving oxygen supplementation.
Comparison	Patients with and without any comorbidity at admission to hospital. Comorbidity was defined as pre-existing health conditions present at admission to hospital with COVID-19, including obesity.	Comparisons within a sample of patients who all have a comorbidity (e.g., studies of cancer patients only). Comparisons between groups of people based on their health-related behaviours (e.g., smoking), ethnicity, or socioeconomic circumstances.
Study design	All primary quantitative empirical observational studies that reported estimates of the independent relative hazard/odds of experiencing a severe outcome according to comorbidity status, adjusted for age only, or age and other plausible confounders of that association.	Any studies in which all estimates of excess risk associated with comorbidity were also adjusted for potential mediators between comorbidity and severe disease outcomes, such as clinically ascertained biomarkers (e.g., inflammatory response or organ function). Causal interpretation of hazard/odds ratios is inappropriate from models not designed to account for confounding of the exposure-outcome association of interest (9, 10), therefore in this review estimates would likely be biased towards the null if adjusted for clinical biomarkers. Qualitative studies. Intervention studies (e.g., clinical trials of new treatments for COVID-19). Projections or estimations of potential outcomes. Non-empirical studies, including editorials, opinions, or discussion pieces. Studies that do not report comorbidity-related risk estimates Review-level evidence
	Include	Exclude
Publication characteristics
Publication stage, type	Pre-prints, peer-reviewed publications, grey literature on empirical evidence (e.g., official statistics).	Not applicable.
Language	English language publications.	Non-English language publications (not available for full text).
Date	Studies published between December 2019 and 14th May 2020.