Two of today's findings turn on the comparator rather than the intervention.
The biofeedback trial used a waitlist, whose migraine days changed by exactly 0.0. A waitlist controls for the passage of time and for regression to the mean, and for nothing else — not expectation, not attention, not the discipline of doing something every day. Against a sham the difference would very likely be smaller.
The dementia meta-analysis turns on a different distinction: whether the activity was chosen. Leisure and household activity are volitional and protective; occupational and commuting activity are imposed and were associated with harm. Whatever the mechanism, an exposure that people select carries everything else about the people who select it, which is why observational activity research so often disagrees with trials.
Both reduce to one habit: before reading an effect size, ask what the comparison group actually experienced, and whether the exposure was something participants chose.
- A waitlist control measures time and regression, not expectation or attention
- A control arm that changes by exactly zero suggests no non-specific effect was captured
- Chosen exposures carry the characteristics of the people who choose them
- Same activity, different domain, opposite direction — that is a confounding signature
- Ask what the comparator did before you read what the intervention did
The statistics, in plain English
Confounding by indication and healthy-user effects both arise from the same root: people who take up an exposure differ from those who do not, in ways that also affect the outcome. Randomisation removes it; statistical adjustment removes only the part you thought to measure, which is why adjusted observational estimates and trial results so often diverge.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for neurology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free