[Skip to Navigation]
April 2000

Statistical Analysis to Settle Ethical Issues?

Arch Gen Psychiatry. 2000;57(4):327-328. doi:10.1001/archpsyc.57.4.327

ETHICAL ISSUES cannot be settled by statistical analysis. While I congratulate Khan et al1 on their efforts to "promote informed discussions about the use of placebo in antidepressant clinical trials," I question whether this or any study could be adequately designed "to assess the fate of depressed patients assigned to placebo in antidepressant clinical trials."

My stance on placebo controls changed when, many years ago, while in labor, I was recruited for a placebo-controlled randomized controlled trial (RCT) on the efficacy of an obstetric anesthetic agent. I forbear quoting my exceedingly rude refusal. I, who had advocated and participated in the design, execution, and analyses of many placebo-controlled RCTs, found this request bordering on the medically unethical. The risk of any long-term adverse effects of treatment with placebo was negligible. However, for the duration of the labor, I had a 50% chance of being deprived of the best usually available medical care and suffering the consequences, however temporary. Such deprivation, to my mind, is not in accord with the injunction to "do no harm."

Add or change institution