Agreement between survey- and register-based measures of depression in Denmark.

Background: Register-based studies of major depressive disorder (MDD) do not capture all prevalent cases, as untreated cases and diagnoses made by general practitioners are not recorded in the registers. We examined the prevalence and agreement of survey- and register-based measures of depression, and explored sociodemographic and health-related factors that may have influenced this agreement.

Methods: All 32,407 participants in the 2017 Central Denmark Region How are you? survey were linked to hospital and prescription records. A checklist for depressive symptoms within the last 14 days (Major Depression Inventory; MDI) from the survey was compared with register-based assessment of hospital-diagnosed MDD and/or prescriptions for antidepressants. We estimated agreement between survey-based and register-based measures for depression and used logistic regression models to explore selected associated factors.

Results: In total, 5.9% of How are you? survey participants screened positive for current depression on the MDI. Of these, 51.3% (95% confidence interval (CI): 49.0-53.6) filled a prescription for an antidepressant medication during the 10 years prior or 2 years following the administration of the survey, and 14.5% (95% CI: 12.9-16.2) were treated for MDD in a psychiatric hospital-based setting. When using a higher threshold of the MDI indicating more severe current depression, 22.8% (95% CI: 19.6-26.1) of those who screened positive also received an MDD diagnosis and 63.4% (95% CI: 59.7-67.2) were prescribed antidepressants during this 12-year period. Among those with current depression, female sex, older age, chronic diseases, hospital-treated self-harm, and being permanently outside the workforce were associated with having a register-based MDD diagnosis or antidepressant prescription. Among those with a register-based depression record, female sex, younger age, hospital-treated self-harm, stress, and severe loneliness were associated with current depression.

Conclusion: We found that as few as 15% of individuals with current depression in the general Danish population were captured by the psychiatric hospital register, while 51% of these individuals were identifiable in the prescription register. These findings demonstrate that register-based measures significantly underestimate the true prevalence of depression by identifying only the cases that are most severe.