Temporal and geographical variation of COVID-19 in-hospital fatality rate in Brazil

BackgroundPrevious studies have shown that COVID-19 In-Hospital Fatality Rate (IHFR) varies between regions and has been diminishing over time. It is believed that the continuous improvement in the treatment of patients, age group of hospitalized, and the availability of hospital resources might be affecting the temporal and regional variation of IHFR. In this study, we explored how the IHFR varied over time and among age groups and federative states in Brazil. In addition, we also assessed the relationship between hospital structure availability and peaks of IHFR. MethodsA retrospective analysis of all COVID-19 hospitalizations with confirmed outcomes in 22 states between March 01 and September 22, 2020 (n=345,281) was done. We fit GLM binomial models with additive and interaction effects between age groups, epidemiological weeks, and states. We also evaluated the association between the modeled peak of IHFR in each state and the variables of hospital structure using the Spearman rank correlation test. ResultsWe found that the temporal variation of the IHFR was heterogeneous among the states, and in general it followed the temporal trends in hospitalizations. In addition, the peak of IHFR was higher in states with a smaller number of doctors and intensivists, and in states in which a higher percentage of people relied on the Public Health System (SUS) for medical care. ConclusionsOur results suggest that the pressure over the healthcare system is affecting the temporal trends of IHFR in Brazil. Key MessagesO_LITemporal variation of age adjusted In-Hospital Fatality Rate (IHFR) was markedly heterogeneous among Brazilian states from March to September of 2020. C_LIO_LIIn several states, the IHFR increased in association with the increase in the number of hospitalizations, which suggests that the overload of the healthcare system might be affecting the temporal trends of IHFR in Brazil. C_LIO_LIThe IHFR remained low in the states with higher rates of hospital resources, even with the high demand for hospitalization. C_LIO_LIThe number of doctors and intensivist physicians per habitant was more strongly correlated with the peak of IHFR in the Brazilian states than the number of ICU beds. C_LI