An indirect estimate of the incidence of non-insulin-dependent diabetes mellitus

Abstract

Our goal was to estimate non-insulin-dependent diabetes mellitus incidence in the Netherlands in the absence of equivocal empirical data. Incidence can be expressed as a function of age, sex, prevalence, and mortality. We obtained prevalence data from a study that pooled existing prevalence estimates. We calculated diabetes-related mortality using relative risks on all-cause mortality. Sensitivity for the rate of excess mortality was determined using the 95% confidence intervals (95% CI) of the relative risks. The estimated incidence increases exponentially with age, with a doubling time of 10 years for men and 9 years for women. The rate increases from 8.1 per 10,000 (95% CI = 7.7-8.8) for men ages 40-44 years and 7.0 (95% CI = 6.8-8.0) for women to 79.7 per 10,000 (95% CI = 69.5-90.9) for men ages 75-79 years and 85.8 (95% CI = 80.6-91.0) for women. When empirical estimates of incidence are largely lacking, the methodology described offers a useful alternative, in particular for the assessment of potential intervention effects.