What does a health-zone case-fatality ratio measure during an active outbreak? Reported mortality, mapped health-facility context, and case-death reporting heterogeneity in the 2026 Bundibugyo virus disease epidemic in DR Congo
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
During an active outbreak, deaths divided by confirmed cases can be mistaken for biological severity or quality of care even though outcomes remain unresolved and reporting and ascertainment differ across places. We examined what health-zone case-fatality ratios measured during the 2026 Bundibugyo virus disease epidemic in the Democratic Republic of the Congo.
Methodology/Principal Findings
We analysed national and health-zone cumulative confirmed cases and deaths through 20 July 2026. We calculated reported crude case-fatality ratios, examined case-death reporting synchronisation, and fitted beta-binomial partial-pooling models with four-chain Markov chain Monte Carlo. Exploratory models added mapped clinical-facility availability or distance to the nearest mapped hospital. Reporting-date delay, under-ascertainment, and mortality forecasting were evaluated as diagnostics, scenarios, or developmental analyses rather than patient-level fatality estimation. The national ratio reached 40.4% (999/2,473). Among 14 zones with at least 20 cases, crude ratios ranged from 28.3% to 68.5%. In the primary model, North Kivu had higher posterior odds than Ituri, but with substantial uncertainty (OR 1.80, 95% credible interval 0.92-3.37); epidemic maturity was positively associated (1.49, 1.04-2.19). In exploratory models, greater mapped clinical-facility availability was associated with lower reported fatality (0.67, 0.48-0.96), while greater distance to a mapped hospital was associated with higher reported fatality (1.46, 1.06-2.00). Same-day case-death co-reporting was common, and materially different reporting-delay assumptions fitted similarly.
Conclusions/Significance
Health-zone ratios revealed meaningful surveillance heterogeneity but did not identify biological fatality risk or causal effects of facilities, access, care, or conflict. Epidemic maturity, selective ascertainment, referral, and administrative reporting plausibly shaped the numerator and denominator. These ratios should guide investigation rather than rank health-zone performance; linked patient records are required for clinical fatality and competing-risks analyses.
Author Summary
We studied why reported deaths divided by confirmed cases differed so widely between health zones during the 2026 Bundibugyo virus disease outbreak in the Democratic Republic of the Congo. We found that these ratios were strongly influenced by when an outbreak began locally, how cases and deaths entered situation reports, and which cases were detected. The ratio increased nationally as the outbreak matured, and cases and deaths were often added on the same reporting date. Mapped health-facility indicators provided useful clues after statistical adjustment, but they did not measure whether services were open, staffed, equipped, reachable, or capable of treating this disease. The difference between North Kivu and Ituri may partly reflect chronic isolation, referral disruption, distrust, case selection, or reporting, but the available data cannot assign a causal explanation. We conclude that a high health-zone ratio should prompt investigation, not be treated as a ranking of care quality. To determine the actual probability of death and the role of treatment or access, researchers need linked patient records containing clinical dates, outcomes, referral histories, and treatments.
