Prediction of diabetes with body mass index, oral glucose tolerance test and islet cell autoantibodies in a regional population

Abstract. Rolandsson O, Hägg E, Nilsson M, Hallmans G, Mincheva‐Nilsson L, Lernmark Å (Umeå University, Sweden; and University of Washington, Seattle, WA, USA). Prediction of diabetes with body mass index, oral glucose tolerance test and islet cell autoantibodies in a regional population. J Intern Med 2001; 249: 279–288. Objective. Our aim was to test the hypothesis that a combination of markers for Type 1 diabetes (glutamate decarboxylase and IA‐2 autoantibodies) and for Type 2 diabetes oral glucose tolerance test (OGTT) and body mass index (BMI)], would predict clinical diabetes in a regional population. Design. A population‐based follow‐up cohort study. Setting. Participants visited the primary health care centre in Lycksele, Sweden in 1988–92. Participants. A cohort of 2278 subjects (M/F 1149/1129) who were studied at follow‐up in 1998. At base line there were 2314 subjects (M/F 1167/1147) who participated in the Västerbotten Intervention Program on their birthday when turning either 30, 40, 50 or 60 years of age. Main outcome measurements. A clinically diagnosed diabetes at follow‐up when the medical records were reviewed for diagnosis of diabetes. At base line, the participants were subjected to a standard OGTT and their BMI determined along with the autoantibodies. Results. At follow‐up, 42/2278 (1.8%, 95% CI 1.2–2.3) (M/F 23/19) had developed diabetes: 41 subjects were clinically classified with Type 2 and one with Type 1 diabetes. There was no significant relation between autoantibody levels at base line and diabetes at follow‐up. Stepwise multiple logistic regression showed that the odds ratio for developing diabetes was 10.8 (95% CI 6.3–18.9) in subjects in the fourth quartile of BMI (BMI > 27) compared with 7.8 (95% CI 4.8–12.6) in the fourth quartile of 2‐h plasma glucose (>7.5 mmol L^?1) and 7.2 (95% CI 4.8–11.4) in the fourth quartile of the fasting plasma glucose (>5.6 mmol L^?1). Conclusion. Islet cell autoantibodies did not predict diabetes at follow‐up. BMI measured at base line was as effective as 2‐h plasma glucose and fasting plasma glucose to predict diabetes in this adult population.