Childhood cancer · PLoS Med · 2026
Halved cancer-diagnosis waits: chance 1 in 1,000, says the paper’s table. Same row: 1 in 6.
How often chance alone gives this drop; convincing if rarer than 1 in 20
The paper says
After Northwest Ethiopian health workers were trained, children’s median wait for a cancer diagnosis (half waited less) fell 54%, from 56.5 to 25.8 days. The paper’s table says chance alone would give a drop this big 1 time in 1,000, if waits hadn’t really changed.
Its own data show
The same row prints the score behind that figure: 1.41, the drop measured against chance’s typical swing. Scores convert into chance figures by one fixed rule, like Celsius into Fahrenheit: 1.41 means about 1 time in 6; 1 in 1,000 needs 3.29. The paper’s 11 other scores obey it; only this table’s three waits don’t.
A result usually counts as convincing only if chance alone would produce it under 1 time in 20. The 1.41 and the 1 in 1,000 can’t both be right; the paper doesn’t publish each child’s wait, so readers can’t tell which is. If the 1.41 is right, the paper’s conclusion linking training to “reduced diagnostic delays” has no convincing test.