Effectiveness of Bariatric Surgery in the Remission of Type 2 Diabetes and Vascular Complications
We would like to thank Chen et al1 for their study reporting a positive outcome regarding the effectiveness of bariatric surgery in long-term remission and improvement of type 2 diabetes mellitus and associated vascular complications. However, the study fails to address its numerous underlying caveats.
First, the retrospective nature of this study poses several limitations, of which the most significant is the relatively large margin for selection bias. Despite the authors’ best efforts to account for this, it is apparent that the two cohorts were not matched for age or body mass index (BMI). In fact, the average initial BMI was substantially higher in the control group (48.8 kg/m2) compared with the surgical group (45.8 kg/m2) (P = 0.0006) and participants in the control group were also significantly older (P = 0.0219). The combination of these two important confounders, both of which strongly influence the course of diabetes, cast doubt on the validity of the results of this study. Relatedly, the study limits itself to moderate-risk groups, and there is no justification provided for this. Given the retrospective nature of the study, the authors are in the best position to analyze the effectiveness in patients with different extents of obesity. The classification of obesity and subsequently, its effect on the inclusion criteria for surgery has been extensively discussed, and we urge caution over the generalizability of the results, given the neglect of the other groups who form the majority of the patient population.
Furthermore, the two cohorts were matched at baseline for a range of factors including the duration of diabetes and insulin medication. However, it is evident that these factors do not provide an accurate reflection of diabetic control, and it is questionable why HbA1c level, which was the final outcome measure, was not matched at baseline. This is because HbA1c is the best marker of diabetic control and would therefore provide a more comprehensive evaluation of the impact of the interventions.2 Moreover, this study neglected to appraise comorbidities present in the patient population at baseline. In addition to diabetes, metabolic syndrome comprises of hypertension and hyperlipidemia, both of which may be highly prevalent in the moderate-risk category of patients used in this study. Alleviating these two disorders plays a huge role in diabetes control and pharmacological interventions are likely to have a more direct overall impact on the metabolic compared with surgery alone. Thus, the failure to account for this, as would be the in other studies,3 highlights the limitations of the conclusions drawn from this study.
There is still some paucity in the data provided. The only outcome measure for weight is the excess weight loss (EWL), which is highly dependent on the ideal body weight. This is an ambiguous concept heavily influenced by the body frame of patients, which is based on the outdated Metropolitan Life Tables from the 1940s.4 EWL has high reliability when used in conjunction with measurements of changes in BMI, waist circumference, fasting plasma glucose, fasting insulin levels, and HOMA-IR, but no further analysis of any of this is provided for either surgical or control groups. Accordingly, even the few outcome measures given were only analyzed once, instead of at distinct time points where possible, as is conventionally done.5 This will not only provide an indication of the immediate effects of surgery but it also enables the correlation of these effects to the remission of diabetes and related vascular complications, and account for the impact of attrition rate on the data.