Letter to the Editor

Abstract

I have long had an interest in the possibility of metformin acting as a galactogogue, especially with galegin in Goat’s Rue being a precursor of metformin. I was certainly disappointed in the outcome and even more so in the conclusions reached in “Feasibility and Acceptability of Metformin to Augment Low Milk Supply: A Pilot Randomized Controlled Trial” as published in the May 2019 issue of Journal of Human Lactation (35[2]). The authors stated, “Thus, even though we did not achieve our sample size goal, we did enroll enough mothers to detect a clinically important difference, 73–80% of the time, if one existed.” If the reader implies from this statement (and I am sure many did) that metformin did not act as a galactogogue, this is unfortunate, as I believe this was not a proven conclusion. First of all, this conclusion needs to be parsed more closely due to terminology that the authors use, “We did not achieve our sample size.” Doing so invalidates the hypothesis and any conclusions. Researchers must either achieve their stated sample size and report results, or if the sample size was not achieved, the study is no longer a randomized controlled trial, and only observations can be reported. The authors state that they were able “to detect a clinically important difference . . . if one existed,” but the authors must make sure the reader understands it is not a statistical difference. Even that conclusion is qualified by the statement “if one existed.” The “final” sample size was actually 7 metformin users and 4 placebo users. It is quite apparent that the authors made no provisions to not allow the number of subjects to become so low. A sample size of 12 per group is the rule of thumb for a pilot study. In addition, that is the rule for a pilot study involving a three-period crossover design, which the current study was not. To obtain 80% (not 73% to 80%) power to detect change, the study needed 20 cases in each treatment group, and one of the treatment groups must reach the primary end point, which neither did. My point is not to discourage this type of research, and I am encouraged by it, especially when my bias is to believe that metformin has great theoretical promise to be an important galactogogue, while at the same time there is no good research out there on the subject. Only rigorous research will settle the issue.