Objectives: To determine whether childhood elevated fasting plasma glucose (FPG) levels within the normoglycemic range predict diabetes in adulthood.
Design: Retrospective cohort study.
Setting: Community of Bogalusa, Louisiana.
Participants: Normoglycemic (n = 1723), prediabetic (n = 79), and type 2 diabetic (n = 47) adults aged 19 to 44 years followed up serially for an average of 21 years since childhood. Main Exposures Association of elevated baseline childhood FPG levels with the prediabetic or diabetic status at the last survey in adulthood.
Main outcome measures: Receiver operating characteristic analysis and longitudinal logistic regression odds ratios.
Results: The prevalent rate of adult diabetes status by quartiles of baseline childhood FPG levels showed an adverse trend for prediabetes (P < .001) and diabetes (P = .03), with an apparent threshold occurring at or above the 50th percentile (86 mg/dL). Regarding the predictive value of the above threshold, the area under the receiver operating curve analysis yielded a C value of 0.855 for prediabetes and 0.789 for diabetes models, with sensitivity and specificity, respectively, of 76.9% and 85.2% for prediabetes and 75.0% and 76.0% for diabetes. In a multivariate analysis that included anthropometric, hemodynamic, and metabolic variables from childhood to adulthood and baseline childhood FPG status (> or = vs < 50th percentile), individuals with elevated childhood FPG levels were 3.40 times more likely to develop prediabetes (P < .001) and 2.06 times more likely to develop diabetes (P = .05) as adults.
Conclusion: The fact that elevated FPG level in childhood, even within the normoglycemic range, is a predictor of type 2 diabetes in younger adulthood has implications for health care policy.