Sample size calculation for double blind placebo controlled trial I want to do a double- blind sham controlled clinical trial of a laser therapy for a skin disease. As the primary endpoint is change in inflammation between pre and post treatment/sham biopsies I expect the sham arm to have very little to no effect. I expect that the treatment will have moderate-high improvement- lets say 60%. 
Because of cost issues, I would prefer the active treatment arm to be 20 patients. Can you tell me how many patients I need in the control arm? Can I get away with 10?
 A: Doing pre- and post-treatment biopsies in a sham-controlled group whom you expect (by your own admission) to enjoy no benefit seems grossly unethical. You would expose the sham subjects to pain and other potential harms from the biopsies, without any benefit. Could you not employ photographic and symptomatic assessments of this dermatologic condition?
Also, if this is a skin disease with multiple lesions, can you not use subjects as their own controls, by randomly exposing (or masking) some lesions on each subject? Thoughtful design could surely blind subjects to which among their lesions received active laser treatment.
Furthermore, would you not do better to investigate a dose-response curve? Does the proposed laser treatment have any potential harms? Can you select two doses (e.g., laser intensities or durations of treatment sessions) about which you and your colleagues have equipoise regarding the balance between harms (or burdens, due e.g. to lengthy or frequent treatment sessions) and efficacy?
Would the disease by any chance be hidradenitis suppurativa (HS)? (I am guessing this based on your tag 'obstetrics' and your interest in an inflammatory skin condition thought to be treatable by laser.) It would improve your question to identify whatever disease you are dealing with, providing a link to a lay explanation of it. Please also clarify whether the laser treatments are repeated, such that you have longitudinal assessments in this trial. In that case, ethics might demand adaptive design.
In closing, I would like to emphasize that you cannot consider your trial ethical unless you would willingly allow your mother or sister to be enrolled and randomly assigned to any of its arms.

Addendum: In follow-up to the brief exchange with Frank Harrell below, I offer the following open-access references (neither comprehensive nor even representative) that may prove useful to someone dipping a toe into clinical trial ethics.


*

*Sackett DL. Why randomized controlled trials fail but needn’t: 1. Failure to gain “coal-face” commitment and to use the uncertainty principle. CMAJ. 2000;162(9):1311-1314. Free full text

*Shapiro SH, Glass KC. Why Sackett’s analysis of randomized controlled trials fails, but needn’t. CMAJ. 2000;163(7):834-835. Free full text

*Sackett DL. Equipoise, a term whose time (if it ever came) has surely gone. CMAJ. 2000;163(7):835-836. Free full text

*Berry DA. Bayesian Statistics and the Efficiency and Ethics of Clinical Trials. Statist Sci. 2004;19(1):175-187. doi:10.1214/088342304000000044. Free full text
