As we previously discussed, regression to the mean was an important bias of our study, since it was not controlled (1,2). A subgroup analysis may or may not contribute to this bias. Pennant et al. (3) performed statistical simulations of baseline results and found that a reduction from 6.1 to 4.7 (95% CI ±0.2) hypoglycemic events per patient per year may be due to regression to the mean. However, there would still remain a reduction from 4.7 to 1.4 hypoglycemic events per patient per year, which was of clinical importance in this high-risk population.
The Diabetes Treatment and Teaching Program (DTTP) for type 1 diabetes has been extensively studied in controlled trials including patients similar to those studied in the DCCT, with follow-up periods between 6 months and 2 years (4–7). We refer interested readers to these previous publications, which addressed metabolic control, quality of life, and incidence of severe hypoglycemia or ketoacidosis.
We do not agree with the argument that there was important selection bias during recruitment, since the study was performed as a prospective implementation study in German general hospitals within routine care conditions. Participants were consecutively included as they were referred by their family physicians for participation in the DTTP (1,2). Reasons for referral were initiation of flexible, intensive insulin therapy, option for free diet, and improvement of metabolic control or any other problem related to metabolic control, including frequent acute complications such as severe hypoglycemia and ketoacidosis. There were no explicit inclusion or exclusion criteria for participation in the DTTP.
It remains speculative that there was much interviewer bias. Severe hypoglycemia was defined as hypoglycemia treated by intravenous glucose or glucagon injection. Severe ketoacidosis was defined as diabetic ketoacidosis requiring hospital admission (1,2). Therefore, severe hypoglycemia and ketoacidosis were infrequent events with a serious impact on the personal lives of participants, which should reduce individual recall bias. At follow-up, participants should not have underreported severe hypoglycemia and ketoacidosis, since there were no positive or negative consequences for participants and study staff.
Pennant et al. suppose that the younger mean age (27 years) was a main reason for the threefold increase of severe hypoglycemia in the intensive therapy group of the DCCT (8). As discussed in a previous article (2), the DTTP improves self-management skills, which are essential to successful treatment of type 1 diabetes and reduce the risk of severe hypoglycemia. The DTTP gives patients the opportunity to define individual treatment goals and to balance favorable A1C levels and an unacceptable risk of severe hypoglycemia, whereas the DCCT primarily aims at normalization of A1C values.
The study by Tsalikian et al. (9) does not add to solving the problem of severe hypoglycemia in younger people. Only 49 children (aged 8–17 years) were studied, and the observational period was only 2 days. Furthermore, there is strong evidence from the Epidemiology of Diabetes Interventions and Complications Trial (10) that early intensive treatment results in superior long-term results compared with conventional therapy. We agree with Pennant et al. that more studies should evaluate free diet and lifestyle in young adult or adolescent patients with type 1 diabetes.
Our main message from the current and previous studies is that flexible, intensive insulin therapy and free diabetes diet should not be withheld from patients at risk for severe hypoglycemia or ketoacidosis.
