Objective: To determine whether pediatric office measures (waist circumference, body mass index [BMI], systolic [SBP] and diastolic [DBP] blood pressure, and parental diabetes) and laboratory measures (glucose, triglyceride, high-density lipoprotein cholesterol, and insulin) predict risk of type 2 diabetes mellitus (T2DM) at ages 19 and 39 years.
Design: Nine- and 26-year prospective follow-ups of schoolchildren.
Setting: Urban and suburban schools.
Participants: One thousand sixty-seven girls starting at age 10 years in the National Growth and Health Study and 822 schoolchildren aged 6 to 18 years at entry from the Princeton Follow-up Study. Outcome Measure Development of T2DM.
Results: In the Princeton Follow-up Study, childhood SBP and BMI in the top fifth percentile and black race predicted T2DM at age 39 years (area under the receiver-operator curve [AUC] = 0.698). Adding a childhood glucose level of 100 mg/dL or higher, and high-density lipoprotein cholesterol in the bottom fifth percentile and triglyceride concentration in the top fifth percentile as explanatory variables increased AUC to 0.717 and 0.709, respectively. If childhood BMI, SBP, and DBP were all lower than the 75th percentile, likelihood of T2DM at age 39 years was 2%; the likelihood was 1% if the parents had no DM. In the National Growth and Heath Study, SBP in the top fifth percentile and parental diabetes predicted T2DM at age 19 years (AUC = 0.699). Adding insulin in the top fifth percentile increased AUC to 0.764, with insulin being a significant variable. If childhood BMI, SBP, and DBP were all lower than the 75th percentile, the likelihood of T2DM at age 19 years was 0.2%, 0.2% if the parents were also free of DM, and 0.3% if childhood insulin was also less than the 75th percentile.
Conclusions: Office-based childhood measures predict the presence and absence of future T2DM 9 and 26 years after baseline. Childhood insulin measurement improves prediction, facilitating approaches to primary prevention of T2DM.