This article by Dhruv Khullar was published in the NYTimes.

Jan. 10, 2018 –

The author writes: “Behavior contributes to nearly half of cancer deaths in the United States, and up to 40 percent of all deaths.”

The first citation refers to a popular news magazine with no links to the actual study. The second refers to a NEJM article that draws the 40% statistic from yet another article, this one appearing to be a summary in JAMA, associating numbers of deaths with specific behaviors, with virtually no info dabout how that number was calculated.

How do we assign causes of death? I knew an 80yo woman who smoked a pack a day, with no symptoms of heart or lung disease. Got checked regularly. When she dies in her 90s, will her death be attributed to cigarette smoking?

Thin, fit people get diabetes. Non-smokers get lung cancer. Donald Trump avoids exercise, eats junk and remains free of heart disease and diabetes.

From that NEJM article: “Smoking is increasingly concentrated in the lower socioeconomic classes and among those with mental illness or problems with substance abuse.”

Missing from the discussion are (a) medical error: a heart surgeon’s fatal mistake gets recorded as death from heart disease; (b) assuming longevity as desirable outcome, when an early death might be kinder than a long period of painful illness

Perhaps we should encourage a new form of health behavior: learning to interpret statistics in a medical report.