Arthritis pain · Arthritis Rheumatol · 2019
Summary: 17% on the drug, 10% on placebo had health problems. The paper’s table: 62%, 55%.
Share of patients with any health problem, however minor, during 16 weeks of treatment
The paper says
A trial compared an arthritis-pain drug with a placebo, a look-alike with no drug. The paper’s summary says 17% on the drug and 10% on placebo had a health problem. The term it uses covers any new or worsening problem, however minor, whatever the cause.
Its own data show
The paper’s table counts 209 of 337 on the drug (four dose strengths combined) and 45 of 82 on placebo during treatment: 62% and 55%. Nothing in the paper gives 17% and 10% together: serious problems alone were 1% and 2%, and in the 20 weeks after treatment, 48% and 38%. Their source is unknown.
A placebo group shows how often problems turn up anyway: the drug’s effect is the extra on top. The summary mainly makes problems look uncommon: 1 in 6 on the drug, not 3 in 5. In proportion the drug also looks worse, 1.7 times placebo’s rate, not 1.13 times, though the extra is about 7 in 100 either way.