In response to the interesting article by Barnett et al. (1), we would like to offer the following comments. Diabetes control has been shown to improve with diet and exercise regimens (2,3). The degree of study participants’ compliance with diet and exercise regimens may have confounded the change in A1C reported in the study (1). Also, the independent effect of BMI on both diabetes control and response to therapy has been studied extensively (4). The effect of modification of baseline BMI on diabetes control among various strata of BMI in both study groups needs clarification.
The open-blinded design of the study (1), especially since it involves diabetes education and self monitoring, can significantly impact internal validity due to both performance bias of the subject with respect to compliance with lifestyle modifications as well as detection bias of the health care providers in ascertaining adverse outcomes (5). In addition, the noninferiority design offers no protection against a predetermined idea of equivalence by the investigator, who could allocate similar scores to responses and events of all study subjects (6).
