Factors Associated with In-Facility and Out-of-Facility COVID-19-associated Deaths in the Free State Province, South Africa from March 2020 to March 2022

Abstract

Background The COVID-19 pandemic caused a global health crisis and has been extensively researched. The gap be- tween reported COVID-19 deaths and excess deaths is less well characterized. The majority of research uses hospital death as an endpoint - a limitation in light of the gap in COVID-19 death reporting. Un- derstanding the true extent and “who”, “what” and “where” of COVID-19-related deaths is crucial. Our study investigates factors associated with in-facility (hospital) and out-of-facility (community) deaths using COVID-19 outbreak data from the Free State province which benefitted from improved com- munity death surveillance. Information generated from this analysis is used to recommend potential surveillance system improvements. Methods This study analyzed electronic medical records from the go.data surveillance system in the Free State, South Africa, during the COVID-19 response. It assessed factors associated with community death and determined factors associated with death using subsets of community cases and hospital cases drawn from a single surveillance system. Results There were 201 078 COVID-19 cases in Free State, of which 179 916 were community cases and 21 162 were hospitalised cases. The 30-39 year age group made up 21% of COVID-19 cases and was the most common age group. There were 8 589 COVID-19-associated deaths of which 3 159 were in the community and 5 430 were in the hospital. The fourth wave was associated with higher odds of being a community death compared to the third wave - this was demonstrated in both univariate and multivariable analysis (4th wave OR 1.88, 95% CI 1.57-2.26; aOR 1.67, 95% CI 1.37-2.03). Among iv hospitalised cases, the odds of death were lower for all waves compared to the third wave. Having an immune condition demonstrated a lower aOR of death in the hospitalised model (aOR 0.69, 95% CI 0.57-0.83) but was associated with a higher aOR of death in the community model (aOR 1.82, 95% CI 1.49-2.2). Conclusions The results demonstrate a difference in factors associated with death in the community versus hospi- talised COVID-19 cases, under similar conditions such as variant exposure. This is suggestive of a heterogeneous population and provides insight into the “who”, “what”, and “where” of otherwise unre- ported COVID-19 deaths, suggested by excess death reports. These results preclude national cause of death reports and highlight the heterogeneity of in-facility and out-of-facility deaths. It is recommended that surveillance systems, such as the civil registration and vital statistics system, be reinforced to dis- tinguish community and hospital deaths in near real-time to assist with evidence for action in future pandemics.