Screening identified 640 eligible patients and 506 completed T1 assessment (79.0%). There were no significant differences between eligible patients who participated at T1 and those who refused in age, sex, ethnicity, years with diabetes, number of comorbidities DDS and CES-D scores, or any CIDI condition. At T1, age was 57.8 ± 9.86 years (mean ± sd), 57.0% were female, and the average number of comorbidities was 3.8 ± 2.5 ().
Of the 506 T1 patients, 411 (81.2%) completed all three study waves, 21 (4.2%) missed T2 only, 40 (7.9%) missed T3 only and 34 (6.7%) missed both T2 and T3. Thus, overall attrition was 18.8% over 18 months. We compared 28 diabetes status and demographic variables in those who completed all three waves with those who missed one or two waves. Two low but statistically significant associations were found, which, given the number of analyses run, could have resulted by chance: those who missed a wave more often spoke Spanish than English (r = 0.09, P = 0.04) and experienced more years with diabetes (r = 0.12, P = 0.01). Those with CIDI conditions did not drop out more frequently than those without.
Prevalence of disorder
There were no significant differences in prevalence of each condition across the three waves, so shows the prevalence of each condition at T1. Also included are data from the NCS-R, the prevalence of each condition over 18 months (any one or more waves), and the persistence of each condition over time (any single wave, any two waves, all three waves). Four findings are noteworthy. First, patients with diabetes had a significantly higher rate of MDD, GAD, DYS and PANIC at each wave than prevalence rates reported by the NCS-R: 60% higher for MDD, 122% higher for GAD, 85% higher for PANIC and 6% higher for DYS. Second, the prevalences of diabetes distress and depressive affect were each between 60 and 110% higher than MDD across all study waves. Third, about 30% of those with MDD and about 50% of those with GAD reached criteria for a dual MDD/GAD diagnosis at each study wave; anxiety and depression frequently co-occurred among a substantial number of patients. Fourth, the percentage of individual patients who reached criteria for a condition any time over 18 months was substantially higher than the percentage at any one wave. For example, the mean prevalence of MDD and GAD across the three waves was 10.1 and 8.0%, respectively; but the percentage of individual patients receiving a diagnosis of MDD or GAD at any of the three waves was 19.8 and 17.0%, respectively. These were approximately double the rate for that condition assessed at one point in time; and the percentages for high depressive affect and diabetes distress were about 60% higher. Expanding the time frame from one point in time, as in cross-sectional observation, to 18 months highlights the sharply increased prevalence of affective, anxiety and distress conditions among individual patients with diabetes.
Prevalence of each condition across waves
Persistence of disorder
For those 411 patients on whom we had data at all three time points, shows the number of waves at which individual patients reached criteria for a diagnosed condition. On average, the vast majority of patients with affective and anxiety disorders received a diagnosis only at one wave, whereas more patients with high depressive affect or diabetes distress reached criteria at multiple study waves. For example, 14.9, 4.2 and 1.7% of patients recorded a diagnosis of MDD at one, two and three study waves, respectively. Patients with high depressive affect and no affective and anxiety disorder, however, reported significantly greater persistence of their condition across study waves: 15.5, 11.8 and 9.0% (Bowker's test of symmetry, P < 0.001 for each contrast). Stated differently, 76% of those with ≥ 16 CES-D scores at one time point scored similarly at a second wave, and 77% of these patients scored similarly at a third wave. For the DDS, the percentages were 50 and 86%, respectively. In contrast, only 28% of those who reached criteria for MDD at one wave did so at a second, and only 14% of these did so at a third wave. The comparable percentages for GAD were only 27 and 19%. These findings suggest a greater persistence of depressive affect and distress over time, compared with the more episodic presentation of affective and anxiety disorders.
We used multiple regression to determine which factors were associated with the persistence of each condition across the three waves (). Education, duration of diabetes and non-HDL-cholesterol failed to display a significant, independent relationship with the frequency of presentation over time for any of the six conditions. However, younger age, female gender and high comorbidities were consistently and independently related to greater persistence of most conditions over time. High HbA1c at T1 was also associated with persistence of depressive affect over time. Significance was maintained in all analyses when only patients who reached a criterion for a condition at one, two or three waves were included. Therefore, the findings were not due only to comparisons between those with and without a condition across waves.
Regression weights (β) of demographic and diabetes variables predicting frequency of occurrence of each condition (0, 1, 2 or 3 waves; N = 411)
Linkages with HbA1c
Mean HbA1c did not differ significantly across the three study waves (F = 0.77, P = 0.46). CES-D scores and DDS scores were each significantly and positively associated with HbA1c at each wave: CES-D, r = 0.11–0.17 (P < 0.01); DDS, r = 0.11–0.13 (P < 0.01). In contrast, no significant correlation was found between any affective or anxiety disorder and HbA1c at any study wave. Thus, only high depressive affect and diabetes distress were consistently linked to high HbA1c over time.