Estimating the current size of the 2026 DRC Bundibugyo virus outbreak

WHO collaboratory call, 10 June 2026

Sam Abbott, Kath Sherratt, Samuel Brand and Sebastian Funk

https://epiforecasts.io

Daily SitReps

https://insp.cd/category/actualites/

What can we learn from the SitReps about transmission dynamics?

An external view aiming to support interpretation of the data provided by INSP and DRC authorities.

Model schematic

Generative process from import through cryptic growth and renewal to the observed streams

Initial model inspired by McCabe et al., 2026.

Delay estimates from 2012 Isiro outbreak

The data

stream value used as of
exported cases (Uganda) 3 (dated 11, 16, 23 May) 23 May
export deaths 1 (dated 14 May) 14 May
suspected cases (DRC) 1 077 26 May
suspected deaths (DRC) 246 26 May
confirmed cases (DRC) 550 7 June
confirmed deaths (DRC) 101 7 June
specimens analysed 755 28 May
genetic TMRCA 25 March bound on age

Counts come from INSP situation reports (DRC). Suspected cases and deaths include non-BVD, which the model aims to account for.

Fit to the data

Posterior predictive check: modelled cumulative counts against observed, by stream

Transmission dynamics

Reproduction number over the established outbreak

Doubling time ~16 d, but very uncertain (90% 5–53) — Rₜ ≈ 1.7 (90% 1.2–2.9)

Current size

Cumulative infections, onsets and deaths over time with cut-off densities

Estimated true burden ≈ 3,500 infections (90% 2,100 – 9,000)

Consistency with McCabe et al. 

Comparison with McCabe et al., matched in time

Comparison to earlier versions

Comparison with earlier versions of our report

Reconciling different data streams

Comparison of estimates from different data streams vs. joint

Limitations

  • Based on a few weeks of data for an outbreak that started
  • Several priors (delays, incubation, CFR, ascertainment) are weakly informed or imported from other outbreaks/contexts (e.g. 2012 Isiro)
  • Aggregated model for the national level
  • An external view of the SitRep data
  • Work in progress; model and analysis drafted with an LLM under human oversight