How often do people taking placebo in clinical trials report a given side effect? Look up the background rate by condition and follow-up, then compare a drug's reported rate against it. Built from placebo-controlled trials and placebo participants on ClinicalTrials.gov.
Data: every placebo-, sham- or vehicle-controlled trial with posted results on ClinicalTrials.gov (API v2, retrieved ). Only placebo groups not combined with another treatment are used; vaccine trials are excluded.
Trials list non-serious side effects only when they reach a reporting threshold (often 5%) in some group, so a side effect that is not listed is known only to be below that threshold. The estimates treat it that way (a censored random-effects model) instead of counting it as zero or ignoring it; both of those shortcuts distort the rate.
The 95% prediction interval is the range expected for the placebo arm of a new comparable trial, and it is the number to compare an observed rate with; the mean is the expected rate in such an arm. These are trial-level background rates, not an individual's risk, and the comparison is a rough guide, not a test of whether a drug causes an effect. Not medical advice.