Objective--To analyse anthropometric and metabolic characteristics as risk factors for development of non-insulin dependent diabetes mellitus in middle aged normoglycaemic men.
Design--Prospective population study based on data collected in a health survey and follow up 10 years later.
Setting--Uppsala, a middle sized city in Sweden.
Subjects--2322 men aged 47-53, of whom 1860 attended the follow up 7-14 years later, at which time they were aged 56-64.
Main outcome measures--Incidence of non-insulin dependent diabetes.
Results--In a multivariate logistic regression analysis, variations of 1 SD from the mean of the group that remained euglycaemic were used to calculate odds ratios and 95% confidence intervals. Blood glucose concentration 60 minutes after the start of an intravenous glucose tolerance test (odds ratio=5.93, 95% confidence interval 3.05 to 11.5), fasting serum insulin concentration (2.12, 1.54 to 2.93), acute insulin increment at an intravenous glucose tolerance test (1.71, 1.21 to 2.43), body mass index (1.41, 1.01 to 1.97), and systolic blood pressure (1.23, 0.97 to 1.56) were independent predictors of diabetes. In addition, the use of antihypertensive drugs at follow up (selective or unselective [Beta] blocking agents, thiazides, or hydralazine) was an independent risk factor (1.70, 1.11 to 2.60).
Conclusions--Metabolic and anthropometric characteristics associated with or reflecting insulin resistance as well as a poor acute insulin response to glucose challenge were important predictors of future diabetes in middle aged men. Antihypertensive drugs were found to constitute a further, iatrogenic risk factor.
That obesity is one of the predisposing factors for non-insulin dependent diabetes mellitus has been shown in several prospective population studies. [1-5] Hereditary factors as expressed by a history of diabetes in close relative also seem to be important. [6-9]
During the past decade criteria for impaired glucose tolerance and non-insulin dependent diabetes have been set using the oral glucose tolerance test. In the early 1970s the intravenous glucose tolerance test seemed a likely tool for diagnosing diabetes, but no criteria for using it to diagnose non-insulin dependent diabetes have been proposed.
A low insulin concentration soon after glucose injection is found in normoglycaemic relatives of patients with non-insulin dependent diabetes and also in patients with impaired glucose tolerance or non-insulin dependent diabetes. [10 11] It has been suggested that people with a low early insulin response are at risk of developing non-insulin dependent diabetes,  but this has not been confirmed in prospective studies. Impaired insulin action (insulin resistance) is a precursor of non-insulin dependent diabetes in Pima Indians and Mexican-Americans [13-16] and white European women. 
During 1970-3 the intravenous glucose tolerance test was used in a large population based survey in Uppsala, Sweden, to characteristic glucose metabolism in middle aged men. The present study aims to investigate the incidence of non-insulin dependent diabetes during an observation period of about 10 years in these men and to find the risk factors for non-insulin dependent diabetes by using metabolic features, anthropometric measures, history of diabetes among relatives, and, especially, insulin concentrations at intravenous glucose tolerance test. Observations that diabetes is overrepresented among treated hypertensive men  and women  prompted us to evaluate hypertensive drugs as a risk factor for non-insulin dependent diabetes.
Between 1970 and 1973 the health of 2841 men born 1920-4 and resident in Uppsala was surveyed. In all, 2322 (81.7%) attended. The survey aimed to identify risk factors for cardiovascular disease in middle aged men and to select individuals at high risk for treatment, [20 21] and it used an intravenous glucose tolerance test to characterise the participants with regard to glucose metabolism and insulin values. [10 22]
We started to re-examine the eligible part of this cohort in January 1980. At that time 2130 men were still alive and registered as inhabitants of Uppsala. Sixty seven died before they had been investigated; of the remaining 2063, 1860 attended the second screening, giving an attendance rate of 90% (fig 1). Of the 270 initial participants who did not participate in the 1980 survey (but who were still living in Uppsala), 135 had regularly been checked by physicians. Requests for information about their …