Results: Background: To test the hypothesis that in adults with diabetes increasing severity of depression is associated with worse glycemic control.
Methods: Study subjects were 14,144 adults with diabetes mellitus. We classified depression status based on ICD-9 depression codes (296.2, 296.20-26, 296.3, 296.30-36, 296.82, 298.0, 300.4, 311), and use of anti-depression medications as follows: (a) No evidence of depression, (b) mild depression_one or more depression codes, (c) moderate depression_one or more depression codes plus one or more depression drugs, (d) severe depression_one or more hospital admissions with principal discharge diagnosis of major depression. Multivariate linear regression models were used to adjust covariates including age, gender, Charlson comorbidity score, number primary care visits and likelihood of A1c test.
Results: As depression severity increased from category (a) to (d) above, mean age decreased (p<.0001), proportion female increased (p <.0001), medical comorbidity increased (p<.0001), and number of primary care visits increased (p <.0001). A1c test rates were, from (a) to (d) 71.3%, 64.6%, 67.9%, and 62.8% (p <.0001). Unadjusted mean A1c was 7.7% in those with no depression (N=11,802), 7.7% in mild depression (N=1,475), 7.6% in moderate depression (N=754), and 7.5% in severe depression (N=113; p = .22). In a multivariate model that controlled for patient age, gender, comorbidity, primary care visits and test likelihood, depression severity was not related to A1c level (p= .12).
Conclusion: As depression status worsened, number of medical visits and medical comorbidity increased, but we found no evidence that worsening depression status is related to worse glycemic control.