Theoretical pharmacokinetic advantages and methodological flaws: glargine is not superior to NPH insulin in children with type 1 diabetes mellitus.

Theoretical pharmacokinetic advantages and methodological flaws: glargine is not superior to NPH insulin in children with type 1 diabetes mellitus To the Editor, We have read with great interest the recent article by Rollin and collaborators (1) concerning the use of glargine and NPH insulin in children (0-8 years old) with type 1 diabetes mellitus (T1DM). We would like to comment on the context which it emerges from, on aspects about the study design, and on how data is presented and interpreted. We believe that taking our comments into account would substantially change the interpretation drawn from this study. Glargine is an insulin analog that, in theory, has pharmacokinetic advantages when compared to NPH insulin (2). Unfortunately, these advantages have not been translated into better glycemic control (3) or into enhanced prevention of chronic complications in children with diabetes. The only benefit demonstrated in children studied in clinical trials was a small reduction in the incidence of nocturnal hypoglycemia (4). Despite of this, glargine is one of the most prescribed drugs for the treatment of diabetes (5). In Brazil, access to health services and medication is a right assured by law. Furthermore, in this country glargine costs over four times more than NPH insulin (R$ 109.65 against R$ 23.97 a the 3 mL ampoule) (6). The study by Rollin and collaborators emerges from this background and is as a source of local (Brazilian) data. Rollin and collaborators aimed to compare the efficacy and safety profile of glargine (compared to NPH insulin) in children with T1DM younger than eight years old. The endpoints chosen to measure these effects were glycosylated hemoglobin (HbA 1c) and mean hypoglycemic events per patient-period, for efficacy and safety, respectively. To access these endpoints, the Authors used a non-randomized design and compared the patients before and after the use of glargine (1). This type of design (quasi-experimental with historical self-control) is prone to many biases. A randomized design would be methodologically more adequate to answer the question raised in the study. If the small number of patients in the institution was a limiting issue, this could have been overcome by means of adopting a cross-over design or, more appropriately, by means of a multicenter study effort. Furthermore, patients, physicians providing care, and investigators were not blinded, which could contribute to measurement bias. Also, this could lead to an intervention bias: better glycemic control secondary to closer monitoring …