Distress and cancer · BMJ · 2017
The paper’s summary: the most distressed were 32% likelier to die of cancer, even allowing for smoking. Its own table: 26%.
Extra risk of dying of cancer if most distressed, with the likely range
The paper says
The paper’s summary says the most distressed people (scoring 7 to 12 out of 12 on a distress questionnaire) were 32% more likely to die of cancer than the rest, even allowing for smoking, drinking, weight and education.
Its own data show
The paper’s Table 5 has two lines for this. The 32% line allows only for age and sex. The line that also allows for smoking, drinking, weight and education says 26% (146,008 people with all four recorded, against 163,363). The summary’s six other figures, such as prostate and pancreas cancer, match that second line.
“Allowing for” smoking, drinking, weight and education means comparing distressed people with others similar on those four, as recorded, so the comparison takes those recorded differences out. Done that way, the paper gets 26%, not 32%: still raised, but about a fifth smaller. The summary, reprinted in research databases, doesn’t show that drop. Its figures for particular cancers are right.