Among patients with certain chronic diseases, those who are overweight survive longer than those of normal weight, although excess weight raises the risk of developing the disease.
Reported in dialysis and heart failure patients in the late 1990s and named in cardiology soon after. The explanations divide: selection, since the comparison begins only after diagnosis; reverse causation, since illness itself causes weight loss; confounding by smoking, which lowers weight and raises mortality; and the crudeness of body mass index as a measure of fat. Studies that exclude smokers and patients with recent weight loss tend to weaken or remove the effect.
In practice
- A registry reports better survival for heavier heart failure patients, and the advantage disappears once those who lost weight in the previous year are excluded.
- A clinician advises a patient to lose weight while the registry on his desk seems to say otherwise, because the registry counts only people who already have the disease.
- Possible explanations
- Candidate explanations are selection, since the comparison begins after diagnosis; reverse causation, since illness itself causes weight loss; confounding by smoking, which lowers weight and raises mortality; and the crudeness of body mass index as a measure of fat.
- In its favour
- The pattern in the data is not in doubt, and it has been reported repeatedly across heart failure, dialysis and other chronic conditions.
- Against it
- Studies that exclude smokers and patients with recent weight loss weaken or remove it, which is what an artefact of conditioning on diagnosis would predict.
- Standing
- Contested. The pattern in the data is not in doubt; whether it is a fact about bodies or an artefact of conditioning on diagnosis is. Compare the birth-weight paradox, its structural twin.
- Subject
- Science & Epistemology
- Essay
- Paradoxes in the data, and sayings about knowing