WHO’s Ebola death rate has climbed to 46.8 per cent in a month. Its own recovery counts show what is moving

The World Health Organization’s latest Disease Outbreak News on the Bundibugyo virus outbreak in the Democratic Republic of the Congo, issued 14 August and current to 12 August, reports 4,665 confirmed cases, 2,184 deaths and a crude case fatality ratio of 46.8 per cent. Two notices earlier, in mid-July, the same ratio was 39 per cent. A rate rising nine points in a month reads as a disease becoming more lethal.

WHO publishes the figures that show it is mostly something else. Its notices also carry recoveries — "at least 986 patients have recovered" in the latest, 965 in the count as of 12 August — and dividing deaths by the cases whose outcome is actually known gives 68.0 per cent in mid-July, 70.9 at the end of the month and 69.4 now. That ratio has been flat. What has moved is how much of the caseload has a recorded outcome at all: 57.3 per cent, then 62.1, then 67.5. The crude ratio is the product of the two, and it is rising because the backlog is clearing.

Neither number is a good estimate of how deadly the infection is, and the direction of the error is knowable. A death is ascertained within days; a discharge requires two negative tests and comes weeks later, so the resolved cohort is weighted toward deaths at any moment. WHO labels its own figure "crude" for this reason. The honest bracket from its data runs from 46.8 per cent, which assumes every unresolved case survives, to 79.3 per cent, which assumes none does — and the true figure will sit inside that and is not yet determinable.

The geography is where the outbreak stays uneven, and the province detail is not WHO’s. Its notice puts Ituri, which carries about 85 per cent of cases, at 43.4 per cent; the DRC health ministry’s daily situation reports and a European Commission emergency-mapping product built from them put North Kivu at about 67 per cent. Once Ituri is printed below the national average, everywhere else is arithmetically forced above it, so the gap is not in itself a finding — and differences in surveillance depth and access to treatment centres shape it as much as lethality does.