Addiction treatment · JAMA Intern Med · 2014
The paper says its headline measure was chosen beforehand. The trial’s public plan never listed it.
Main measures as the paper describes them, and in the trial’s 2009 public plan
The paper says
Opioid-addicted hospital patients were randomly either started on a substitute drug and booked into the hospital’s addiction clinic, or weaned off opioids. The paper says two main measures were chosen beforehand: reaching that clinic (72% vs 12%) and days on illegal opioids (40% fewer).
Its own data show
In 2009, years before any results, the trial posted its plan on the public website ClinicalTrials.gov. Its two main measures: opioid use (matching the 40%) and HIV-risk behaviour. Reaching the clinic is not listed at all, and ten later updates, three after the paper, never added it. The paper reports no HIV result.
Trials post their main measures publicly before results exist, so readers can check that the results reported are the ones planned. The paper’s summary leads with the 72%, a measure the plan never had, and one that follows closely from the treatment: that group was booked into that very clinic. The plan’s central question, HIV-risk behaviour, goes unreported.