A comment on ‘the risk of coronary heart disease in type 2 diabetic patients exposed to thiazolidinediones compared to metformin and sulfonylurea therapy’

We read with interest the recently published article by Johannes et al. evaluating whether the risk of coronary heart disease differs among type 2 diabetic patients treated with thiazolidinediones (TZDs) compared to those treated with combined oral metformin and sulfonylurea (Mþ S) therapy. This report represents an important contribution to the field, but we would like to comment on some of the authors’ methods. Our primary concern with this study is the definition of study population. Although, according to the title and the introduction, the targeted study population was type 2 diabetic patients, it is not clear that Johannes et al. excluded type 1 diabetic patients and specifically selected type 2 diabetic patients. Although type 1 diabetes mellitus most commonly develops before the age of 30, in contrast to previous studies targeting type 2 diabetic patients, the authors did not include information regarding how patients with type 1 diabetes were excluded which is especially important in the case of patients under the age of 35. Moreover, a previous study observed that the majority of cases of death in diabetic participants was attributable to acute coronary events, and that the 5-year mortality rate for type 2 diabetes (18.9%) was higher than that for type 1 (5.5%). Likewise, type 1 and type 2 diabetes may differ in several aspects, such as their complications. Therefore, the authors should give details of the criteria of the study population. Another concern is that among 12570 TZD initiators, in the ‘as balanced’ analyses, 56% began TZD monotherapy, 16% added a TZD to metformin, 22% added a TZD to sulfonylurea and 6% added a TZD to combination Mþ S therapy; only 38% remained on TZD monotherapy throughout the study period. This leaves us wondering how different the results would have been if the TZD initiators included in the ‘as balanced’ analyses would have been restricted to those who began TZD monotherapy, especially to those who remained on TZD monotherapy throughout the study period. Our third concern with this study is that the authors did not clearly define the non-user group. If non-use was a substitute for ‘no exposure’, which the authors defined, it would be helpful if a different definition were to be used because no exposure, which the authors defined, might result in a different outcome than that seen for patients who have never used TZD. This leaves us wondering whether the results would have been different for as treated analyses if the reference group would have been defined as those who have never used TZD. Finally, the reason that the authors chose as the comparison group patients receiving Mþ S therapy to evaluate ‘the effects of TZD therapy as prescribed in routine clinical practice on coronary heart disease in comparison with other oral antidiabetic therapy’ remains unclear, although authors explained that patients started on TZDs might be more advanced in their disease progression than diabetics started on other oral antidiabetic therapy. Recent studies have shown that the combination of sulfonylureas and metformin might result in a higher risk of adverse cardiovascular outcomes than treatment with metformin alone. In addition, different sulfonylurea derivatives could lead to different outcomes, in particular as regards cardiovascular outcome. Among sulfonylurea derivatives included in final analysis, it would be helpful to include information on what kinds of sulfonylurea derivatives were included.