Paperonium

Infectious disease · BMJ Glob Health · 2020

“No significant changes” in Ebola’s death rate. The figure it cites shows a clear change.

The summary: “nosignificant changes”would sit at 1The number it citesrules out 1: a clear changeIts table: 71% died, then 62%a clear change, but a fall1.6×1.51.80.560.790.5×0.7×1×1.5×2×no changea falllikelier
The summary:“no significantchanges”would sit at 1The numberit citesrules out 1: aclear changeIts table:71% died,then 62%a clear change,but a fall1.6×1.5 to 1.80.560.790.5×1×2×no changea falllikelier

Deaths per survivor, 2019 against 1976

The paper says

The paper’s summary says the share of Ebola patients who died showed “no significant changes” over four decades. Its evidence: deaths per survivor were 1.6 times as high in 2019 as in 1976 (a range of 1.5 to 1.8), where 1 would mean no change.

Its own data show

Two problems. That range, 1.5 to 1.8, leaves out 1, so the cited figure shows a clear change. And 1.6 says dying got likelier, yet its own table shows a smaller share died: 71% in 1976–86 (about 5 deaths per 2 survivors), 62% in 2009–19 (about 5 per 3). That is 0.67 times, a fall.

The range is where the true figure plausibly lies. If it leaves out 1, chance alone would rarely produce a change this big, which is what “significant” means. The summary uses “no change” to argue Africa still struggles to control Ebola. Allowing for how much outbreaks vary, the paper’s counts leave room for no change; it reports no such analysis.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

Forty-two years of responding to Ebola virus outbreaks in Sub-Saharan Africa: a review · BMJ Glob Health (2020) · doi:10.1136/bmjgh-2019-001955