View article

Mocanu and colleagues importantly draw attention to the continued varied reporting of weight-related outcomes after bariatric surgery and the barrier this presents to comparative analyses. They equate reporting body mass index (BMI) with weight, while of course BMI is a derived mathematical term (weight in kg/height in m) 2, a proxy for obesity. BMI changes should therefore reflect both a re-measured weight and remeasured height. Not unique to reports on surgical interventions, height is rarely re-measured at the time of postintervention assessment: the preoperative height is assumed to be constant. In these circumstances, any BMI change is merely a weight change modified by the constant value of height2. Statistical significance testing of BMI changes will give the spuriously identical p value and confidence intervals to comparisons of weight changes. While the assumption of height constancy may be valid for …