Re: Effect of a ketogenic diet versus Mediterranean diet on glycated hemoglobin in individuals with prediabetes and type 2 diabetes mellitus: the interventional Keto-Med randomized crossover trial.

Dear Editor:

We applaud Gardner et al. (1) for their study, especially during a pandemic. Lowering dietary carbohydrate is known to improve glycemic control (2), but this study has several methodological limitations, appearing as if there were a bias against the keto diet [Well Formulated Ketogenic Diet (WFKD)]. Bias #1: medication reduction was different for the 2 periods: during the WFKD period, subjects were instructed to stop the medication (which affects the main outcome of HbA1c), yet in the Mediterranean-Plus (Med-Plus) diet period, the dose was reduced by only 50% [the article states “excluding the four participants (who discontinued the medications), we observed a significant difference for both HbA1c and HDL-C between diets, favoring WFKD”]. Bias #2: changes in glycemic control “were adjusted for weight loss,” diminishing one of the advantages of the WFKD as compared with other diets: superior weight loss. Most concerning, though, are the basic design flaws that this crossover study did not have a washout period (to ensure a return to baseline state), nor a power calculation. Because a return to baseline state did not occur, most methodologists would suggest to treat the study as a parallel-group design and confine analysis to the first period alone (3). So, for the primary outcome, HbA1c, we would like to see a statistical analysis confined to the first period alone for the WFKD compared with Med-Plus diets (mean change: 0.66 compared with 0.31; n = 16; Supplemental Table 6A). If nonsignificant, we would like to see a post hoc power calculation to assess whether sufficient subjects were included in the study to exclude a clinically meaningful difference between the 2 diets.