Enhanced comparisons of COVID-19 mortality across populations

BackgroundThe reported crude case fatality rate (CFR) for COVID-19 varies considerably across countries. Crude CFRs could by biased by larger proportions of older COVID-19 cases in population data, who are also at increased mortality risk. Such distorted age case structures are a common feature of selective COVID 19 testing strategies in many countries, and they potentially mask underlying differences arising from other important factors such as health system burden. MethodsWe used the method of direct case-age standardisation to evaluate the effects of age variations on CFRs. Data on cases and death by age from Italy, Spain, China, Australia and South Korea were analysed to derive standardised CFRs. Findings were compared across different case age distribution references as standards. ResultsUsing the South Korean case age distribution as a standard, the fivefold higher crude CFR for Italy is reduced to less than two-fold after adjustment, while the crude CFR difference for Spain is virtually eliminated. The adjusted CFR for Australia is the lowest among all countries. DiscussionMortality differences based on crude CFRs are exaggerated by age structures, which are effectively controlled by case age standardization. Residual CFR differences could be attributed to health and health system factors. The South Korean case age distribution is an appropriate reference standard, given its robust case detection and contact tracing program. Till reliable population level indicators of incidence and mortality are available, the age-standardized CFR could be a viable option for international comparison of the impact of the COVID 19 epidemic. SummaryO_ST_ABSThe knownC_ST_ABSThere are intense debates around the magnitude of and reasons for wide variations in observed case fatality rates (CFRs) from COVID 19 across countries. Age is commonly speculated as a reason, but this has not been technically quantified or explained. The newThe technique of direct standardization using reference distributions of case age structures eliminates the effects of age on CFR, thus enhancing the comparability as well as understanding of differentials The implicationsResidual differences between adjusted CFRs can be used to infer health and health system factors that influence mortality in COVID 19 cases in different populations