Retrospective diagnostic co-factors among Covid-19 cases sourced from Medicare, 1999-2021: an all-cause mortality review

Abstract

Abstract Introduction: SARS-CoV-2 infections co-occurred with other diverse pre-existing clinical conditions in mortality cases. We use encounter level health data to evaluate the impact of non-Covid-19 diagnostic events on all-cause mortality observed among Covid-19 positive cases billing Medicare. We further investigate prior diagnostic codes which occur in pre-pandemic study years among cases presenting to Medicare clinically with Covid-19 and cases with Covid-19 who experience all-cause mortality to inform patient population management. Methods: We aggregated encounter level records sourced from all Medicare beneficiaries from 1999-2021. Odds ratios were constructed using diagnostic history, age decile, study year and survival status. We used Generalized Linear Model (GLM) to predict the Decedent Observation Odds Ratio (DOOR) from study year, case observation odds ratio, age decile, non-covid conditions within counts of distinct covid-ever cases and their decedents. Odds ratios are relative to covid-never cases, or cases who didnot present with Covid-19 clinically. Results: High explanatory DOOR measures are observed for diagnostic codes commonly associated with inpatient Covid-19 mortality. High DOOR measures are also observed for individuals living with specific kinds of cancers, experiencing cardiac arrest or acute tubular necrosis. Conclusion: Covid-ever mortality is influenced by primary infection itself and exacerbations of pre-existing conditions. Consequences of primary infection are observable in GLM, as well as meaningful prior clinical risk factors such as cancer, diabetes, cardiac and respiratory disease. Long-covid conditions require surviving Covid-19 clinical presentation and are predictable from GLM models.