In the current study, we used data from two waves of medical assessments of the older subsample of BASE-II participants, which represent up to 10.4 years of follow-up (mean follow up at 7.4 ±1.5 years). Data were available for 1,671 (mean age 68.8 ±3.7 years, 51.6% women) and 1,083 (mean age 75.6 ±3.8 years, 52.0% women) participants of baseline and follow-up assessments, respectively. Detailed characteristics are shown in table 1.
Two hundred and nine participants were diagnosed with T2D at baseline out of 1,625 for whom T2D data were available (12.9%, 68.7 ±3.7 years, 37.3% women), 52 of them were newly diagnosed (24.9%). One hundred eighty-five participants (out of 1,083) had this diagnosis at the time of follow-up (17.1%, 75.6 ± 4.2 years, 41.1% female), including 111 prevalent and 74 incident cases (Figure 1). The corresponding incidence rate is 10.1 new T2D diagnoses per 1000 person-years. Of the 185 T2D cases at follow-up, 41 participants (22.2%, females N=21) were not aware of the disease, which resembles the proportion observed at baseline. At baseline, men had a T2D prevalence of 16.2% and women of 9.0% which increased at follow-up to 21.0% and 13.5%, respectively. Baseline and follow-up characteristics of participants with T2D are displayed in supplementary table 2. Details of the analytical sample of 111 participants diagnosed with T2D at baseline and follow-up, the prevalent cases, are displayed in table 2.
At baseline 38.9% of the participants had prediabetes. When focussing on the 74 incident T2D cases, for which 64 full laboratory datasets were available, 61 of 64 (95.3%) had prediabetes at baseline, and as expected they had significantly higher fasting blood glucose and HbA1c values at baseline when compared to the participants who had not developed T2D at the time of follow-up (both p<0.05, Welch’s t-test).
We next evaluated baseline HbA1c and fasting blood glucose in the 41 participants with incident T2D who were diagnosed at follow-up for the first time. This revealed that the mean baseline HbA1c was within the prediabetic range for both, men and women (both 5.8% ±0.3), whereas this was the case for fasting glucose only in men (101.9 mg/dl ±8.7) and not in women (97.7 mg/dl ±9.3).
Diabetes diagnostic criteria and antidiabetic medication at follow-up
We next focused on the diagnostic criteria and their combinations leading to the T2D diagnosis in the 185 participants with T2D at follow-up (figure 2). A total of 143 participants were diagnosed based on at least one laboratory parameter, fasting glucose, 2h-glucose (OGTT) or HbA1c, and 46 participants fulfilled the maximum of four diagnostic criteria (OGTT was only performed when T2D was not known): anamnestic information, antidiabetic medication, fasting glucose and HbA1c. In 42 participants the T2D diagnosis was based solely on anamnestic information on an existing T2D diagnosis and/or antidiabetic medication use without any of the laboratory parameters reaching the diagnostic cut-off. With 24 out of the 41 incident T2D cases at follow-up more than half of the newly diagnosed participants were diagnosed solely by the 2h-OGTT. Interestingly, the T2D diagnosis based on impaired glucose tolerance (OGTT) as the only criterion among the incident cases was found more frequently in women (N=16) than in men (N=8), a difference which was statistically significant (p=0.028, Fisher’s exact test, supplementary figure 2 and 3).
We next evaluated the capacity of the three laboratory parameters, fasting glucose, HbA1c and 2h-glucose (OGTT), as assessed at baseline to predict incident T2D at follow-up (7.4 ±1.4 years later, a dataset of N=860 with all three parameters available was used for this analysis). Figure 3A shows the ROC curves from this analysis, which revealed that the AUCs for fasting glucose, HbA1c and 2h-glucose were comparable with overlapping 95% confidence intervals (CI) and can be classified as within the acceptable range: fasting glucose, AUC: 0.820, 95% CI 0.767-0.874; HbA1c, AUC: 0.792, 95% CI 0.735-0.850 and 2h-glucose (OGTT), AUC: 0.718, 95% CI 0.648-0.788. Stratification of this analysis by sex revealed that all three parameters of the glucose status tested are equally able to predict incident T2D with AUCs comparable to the values yielded from the not stratified analysis, with the exception of the 2h-glucose (OGTT) in men which predicted incident T2D less accurate (AUC: 0.671, 95% CI 0.570-0.771, Figure 3B and C).
Evaluating the antidiabetic medication of our sample in the follow-up dataset revealed that 100 out of the 185 participants diagnosed with T2D were treated with antidiabetic drugs. The majority (N=85) used metformin, 47 of them as the only antidiabetic medication and 38 in combination with another oral antidiabetic drug or insulin; the most frequent combination being metformin and a dipeptidyl peptidase 4 (DPP-4) inhibitor (N=15). There was no significant difference between women and men with respect to the antidiabetic medication, when considering each medication separately. For details see figure 4.
Diabetes complications severity index (DCSI) at baseline and follow-up for prevalent cases
We computed the DCSI for both waves of assessments as a measure of T2D severity in prevalent cases and determined its change between the two assessments to evaluate T2D progression. The DCSI significantly increased in the 7.3 ±1.5 years between baseline and follow-up (mean DCSI 1.1 ±1.2 vs. 2.0 ±1.8; range 0-5 vs. 0-6). The mean DCSI in women (N=40) increased from 0.7 ±0.8 to 1.8 ±1.7 and in men (N=71) from 1.3 ±1.4 to 2.1 ±1.8 (all p<0.01, Wilcoxon signed rank test). Thus, while the DCSI was higher in men at both time points, the increase of T2D severity as assessed with the DCSI was higher in women, but not statistically significant. The DCSI change per year was 0.12 in men and 0.14 in women. Results are displayed in figure 5 and 6.
In a next step we compared the different DCSI constituting categories (see supplementary table 3) with respect to their impact among the participants with prevalent T2D at both time points of assessment. This revealed that cardiovascular complications was the complication with the highest impact of 43.2% at baseline, and a steep increase to 67.6% at follow-up. This was followed by the DCSI categories nephropathy (21.6% and 61.3%) and peripheral vascular disease (27.9% and 28.8%). When looking at sex-differences, at baseline men were significantly more likely affected from cardiovascular diseases than women, but at follow-up this difference was not significant anymore (p<0.05 and p=0.214, Mann-Whitney-U test). At follow-up nephropathy had the highest impact (60.0%) in women, followed by cardiovasculardiseases (52.5%), whereas for men it was the other way around (76.1% and 62.0% respectively). When comparing the progression of each category, there was no significant difference between the sexes (p>0.05, Mann-Whitney U test).