Two hundred and sixty-four individuals (40% of the baseline cohort; 60% of those still alive and locally resident) had complete information collected for the 18-year clinical examination, as shown in . Overall, only 35 individuals (12%) reported currently taking antidepressant medications, where BDI scores ranged from 0 to 40 and CES-D scores from 0 to 46. Twenty-six percent of those with high BDI scores were taking antidepressant medications, while 16% of those with high CES-D scores reported taking antidepressant medications at the current time. Of those with extremely high PAID scores, 12.5% reported taking antidepressant medications, 64% had BDI scores ≥ 16 and 80% had CES-D scores ≥ 16.
Depression and diabetes-related distress scores (mean, median, sd) of the study population at 18-year follow-up of the EDC study
Neither current age nor age at onset of diabetes were significantly correlated with the BDI, CES-D or PAID. Duration of diabetes was not significantly correlated with the PAID, but was correlated with the BDI (r = 0.17; P < 0.01) and the CES-D (r = 0.23; P < 0.01). A significantly greater proportion of women had high CES-D scores compared with men (25% vs. 11%; P < 0.01). Excluding those on antidepressant medications did not alter these results.
Overall a strong correlation was observed between the CES-D and the PAID (), and also between the BDI and the PAID (r = 0.49; P < 0.01). A similar pattern was observed when men and women were analysed separately (data not shown) and when the data were analysed excluding those on antidepressant medications.
Correlation between depression (CES-D scores) and diabetes-related distress (PAID) scores.
All four physical activity (energy expenditure) variables were significantly and negatively correlated with the BDI (r between −0.20 and −0.27; P < 0.01), the CES-D (r between −0.16 and −0.33; P < 0.01) and the PAID (r between −0.14, P < 0.05, and −0.23, P < 0.01). Overall, HbA1c and PAID scores were significantly correlated (r = 0.16; P < 0.01), but this was not the case when adjusted for sex. Depression and PAID scores did not differ significantly according to smoking status or alcohol intake.
Those who had CAD at 18-year follow-up had significantly higher depression and poorer diabetes-related distress scores compared with those who had not developed CAD (CES-D: 11.6 ± 10.8 vs. 8.3 ± 8.3, P = 0.009; BDI: 10.9 ± 9.3 vs. 6.9 ± 6.4, P = 0.000; PAID: 21.8 ± 20.2 vs. 15.3 ± 14.7, P < 0.01). Mean CES-D and BDI, but not PAID scores, were significantly lower in those without any major complications at 18-year follow-up compared with those with one and those with two μ three major complications (CES-D: 8.3 ± 8.6 vs. 9.8 ± 10.2 vs. 13.4 ± 10.5, P = 0.024; BDI: 6.9 ± 6.7 vs. 9.2 ± 9.2 vs. 11.1 ± 7.5, P = 0.007).
Linear regression analyses demonstrated that higher PAID scores (P < 0.001), longer duration of diabetes (P < 0.001) and lower household income (P = 0.028) were all significant independent correlates of depression (CES-D) (r2 = 0.43). Substituting BDI scores for CES-D as the dependent variable did not change the model. Further linear regression models were calculated in order to identify the significant independent correlates of self-care. Lower HbA1c (P < 0.001), CAD (P = 0.004) and ON (P = 0.005) were significantly independently correlated with SMBG (r2 = 0.14). Significant independent correlates of physical activity, with energy expenditure during the past week (KCAL/WK) used as the dependent variable, were CES-D (P = 0.002) and smoking (P = 0.017) (r2 = 0.10). If energy expenditure in the past year was used as the dependent variable, HbA1c (P = 0.003) and duration of diabetes (P = 0.025) were significant along with CES-D (P < 0.001). PAID scores did not enter the model except when CES-D scores were removed or when BDI scores were substituted for CES-D scores. Substituting number of major complications for specific complications did not alter any of these models.