Sir Winston Churchill was an inveterate enemy to all physical exertion that went by the name of exercise. He attributed his productivity in life to his physical indolence and once gave the advice that you should never stand when you can sit and never sit when you can lie. He did much of his work in bed.

Modern medicine is decisively against him in his opposition to exercise. Reading the introduction to a paper in a recent edition of the British Medical Journal, you might be forgiven for concluding that the panacea has at long last been found, and that it is exercise. People who are physically active live longer and suffer less from heart disease, strokes, cancer, and diabetes than do the sedentary.  They do better in the hospital; and physical inactivity has been estimated to be the fifth most serious contributor to the disease burden of Europe.

The authors of this paper attempted to find out whether exercise is as effective as drugs in reducing mortality in a variety of conditions such as diabetes, stroke, coronary artery disease, and heart failure. They did no actual trial themselves, but rather performed a meta-analysis of the meta-analyses of all the trials that have been published and are relevant to this question: in other words their paper is what might be called a meta-meta-analysis.

Exercise has rarely been compared directly with drug treatments; the authors had therefore mainly to compare the published statistical effect of exercise (compared with no exercise) with that of medication of various sorts (compared with placebo or other medicines). They analyzed the results of 16 meta-analyses which were based upon a total of 254 trials, 54 of them on the effects of exercise and 248 of them on the effects of drugs, involving 339,274 patients in all.

In short they found that the only statistically significant differences in mortality were in stroke and in heart failure.  Drugs (diuretics) were superior to exercise in the latter, and exercise to drug treatment in the former.  Otherwise there was no difference between drug treatment and exercise.