Prediction of diabetes mellitus (NIDDM).

This study was made to analyze predictive NIDDM markers using a long-term GTT follow-up observation period of 1-30 years. Subjects of this study were 5446 cases (3994 males, 1452 females). Results are as follows: (1) NIDDM development rate increased gradually with increasing 2 h-PG levels at GTT, but for groups with 2 h-PG > or = 170 mg/dl, the rate rose rapidly. (2) PG at GTT was higher in the NIDDM development group than in the control group. Mean 1 h-PG reached > or = 200 mg/dl for 4 years before onset in the NIDDM group. Frequency of 1 h-PG > or = 200 mg/dl was 54% 4 years before and 67% 1 year before onset. (3) The highest NIDDM prediction accuracy was in 1 h-PG levels of 200 mg/dl or more and/or 2 h-PG levels of 170 mg/dl or more. Sensitivity was 75.2%, and specificity, 63.4% within 3 years before onset. (4) With addition of delta IRI/delta PG, sensitivity increased, but specificity decreased. (5) The highest relationship with NIDDM development was for high PG levels (1 h-PG > or = 200 mg/dl and/or 2 h-PG > or = 170 mg/dl), the odds ratio being 5.65. The odds ratio of delta IRI/delta PG was lower than ratio of high PG levels. (6) NIDDM development rate increased about 50% in the under-60-years age group and in the group of BMI > or = 25.