The selected topic is the prevention of type 2 diabetes. The research question is “Does a low-fat diet prevent impaired glucose tolerance (IGT), fasting blood glucose > 120, and / or high HemoglobinA1C (AIC) level in high-risk and healthy individuals?” The study to be considered is a 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study, published in The Lancet, 2009 (see citation below).
The Diabetes Prevention Program Outcomes Study was the follow-up to a USA multicenter randomized clinical trial that examined the utility of intensive lifestyle intervention or the drug metformin to prevent or delay incidence of type 2 diabetes. Participants were considered at high risk of diabetes due to raised fasting glucose, impaired glucose tolerance, or obesity. The sample included 3234 participants who were enrolled between 1996 and 1999. Random assignment placed participants in one of three groups: intensive lifestyle intervention, metformin, or placebo.
Qualitative variables, such as gender and ethnicity, were analyzed using the Χ2 test, while quantitative variables, such as fasting glucose level and glucose tolerance test results, were analyzed with t-tests or ANOVA. The exception to this was triglyceride levels, which had highly skewed results and required a special inferential test. Treatment outcomes over time were analyzed using longitudinal regression techniques. Alpha levels were 0.05, but because the study had three groups, this was adjusted to 0.017 for each group. The alpha level was divided to reduce Type 1 errors — that is, failure to reject a true null hypothesis.
After 2 years the placebo group was given metformin, since it had been shown to reduce incidence of diabetes by 31%. A new group was then enrolled, totaling 2766 individuals, for the follow-up study. Lifestyle intervention group classes were offered to all participants to encourage weight loss and physical activity, since lifestyle changes had reduced incidence of diabetes by 58% in the original study. The previous metformin group continued to take metformin unless they developed diabetes with HbA1c of more than 7% or experienced serious side effects. Primary outcome was development of diabetes — fasting blood glucose 7.0 mmol/L or higher, or 2-h blood glucose 11.1 mmol/L or higher after 75 g oral glucose load.
Weight gain during the follow-up period was common — the mean weight for the original metformin and lifestyle groups decreased for the first two years, then increased (especially for the lifestyle group), but both groups stayed under their baseline weight. Development of diabetes during the study showed similar curves for all three groups, but the lifestyle intervention group had lower incidences throughout the 10 year follow-up period. This effect was most pronounced for those aged 60 and older. Mean fasting glucose levels showed a different pattern, with the metformin group lower than the lifestyle group. Beginning at approximately year 5, the lifestyle group surpassed the original metformin group with regard to use of anti-diabetic drugs.
The purpose of this study was to determine what treatments prevent or delay the development of type 2 diabetes in individuals with high risk factors or prediabetes. Incidence of diabetes has been growing for many years, with a high economic cost due to the widespread effects of high blood sugar on the organs of the body. Therefore, gathering evidence on the best way to prevent diabetes is crucial to public health. The DPP study was a large, randomized, multicenter study designed to maximize the statistical foundation of the results. The follow-up study described here was a longitudinal expansion of the original study intended to determine the effects over time of the two interventions. The follow-up concluded that the interventions in the original study did reduce incidence of diabetes over time, and that overall, the lifestyle intervention was the most effective.
Since this study was designed to be a rigorous examination of intervention effects that would potentially guide physicians in their treatment of prediabetes, the inferential statistics chosen (Χ2, t-tests, ANOVA, and longitudinal regression) were among the strongest available that could be used for the various types of data. Without these statistics as well as the alteration in the alpha level, the study would lack a firm foundation for its conclusions, in spite of its design.
- Diabetes Prevention Program Research Group. (2009). 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. The Lancet, 374(9702), 1677-1686.
