Symptom‐based case definitions for COVID‐19: Time and geographical variations for detection at hospital admission among 260,000 patients

AbstractIntroductionCase definitions are used to guide clinical practice, surveillance and research protocols. However, how they identify COVID‐19‐hospitalised patients is not fully understood. We analysed the proportion of hospitalised patients with laboratory‐confirmed COVID‐19, in the ISARIC prospective cohort study database, meeting widely used case definitions.MethodsPatients were assessed using the Centers for Disease Control (CDC), European Centre for Disease Prevention and Control (ECDC), World Health Organization (WHO) and UK Health Security Agency (UKHSA) case definitions by age, region and time. Case fatality ratios (CFRs) and symptoms of those who did and who did not meet the case definitions were evaluated. Patients with incomplete data and non‐laboratory‐confirmed test result were excluded.ResultsA total of 263,218 of the patients (42%) in the ISARIC database were included. Most patients (90.4%) were from Europe and Central Asia. The proportions of patients meeting the case definitions were 56.8% (WHO), 74.4% (UKHSA), 81.6% (ECDC) and 82.3% (CDC). For each case definition, patients at the extremes of age distribution met the criteria less frequently than those aged 30 to 70 years; geographical and time variations were also observed. Estimated CFRs were similar for the patients who met the case definitions. However, when more patients did not meet the case definition, the CFR increased.ConclusionsThe performance of case definitions might be different in different regions and may change over time. Similarly concerning is the fact that older patients often did not meet case definitions, risking delayed medical care. While epidemiologists must balance their analytics with field applicability, ongoing revision of case definitions is necessary to improve patient care through early diagnosis and limit potential nosocomial spread.