According to a new study in the BMJ, between 46-58% of 679 surveyed US internists and rheumatologists regularly prescribe placebo treatments. Indeed, a large majority of the respondents - 62% - said they think the practice is ethically permissible. Interestingly, though, only a minority of the surveyed physicians prescribe inert substances (like sugar pills), preferring painkillers and vitamins.
While medical ethicists disapprove, arguing prescribing placebos undermines trust and violates the principle of informed consent, I'm not so sure it's a bad thing. If we view medicine as a technology for healing (a perspective persuasively advanced by David Wootton in Bad Medicine), then disallowing anything that works seems odd. Indeed, given how powerful the placebo effect is - often significantly outperforming pharmacological or surgical interventions - not occasionally prescribing placebos seems perverse. Patients, it seems to me, care about getting better, not about how they get better. If so, I agree with Ben Goldacre: prescribing placebos is permissible.
(See also: Goldacre's two-part radio program on the placebo effect and the NYT's piece on the BMJ study).