Background: Studies of recognition of depression in older (aged 65 or more) medical inpatients show low rates of recognition of depression by attending physicians. However, few studies have compared different measures of recognition of depression.
Objectives: (1) To compare the validity of four indicators of recognition of depression and a global measure of recognition against a diagnosis of depression and (2) to explore the effect of patient characteristics on recognition of depression.
Methods: In a cohort of 264 medical inpatients 65 years and older (115 with major or minor depression, 78 with no depression), sensitivities, specificities, and diagnostic odds ratios (DOR) of 4 indicators of recognition (symptoms, diagnosis, treatment, and referral) and a global measure of recognition (any of the 4 indicators) were calculated. The associations between patient characteristics (age, sex, history of depression, antidepressant use before admission, severity of depression, comorbidity, duration of hospitalization, disability, and hospital of admission) and recognition were explored using multiple logistic regression.
Results: Less than half of the depressed patients were recognized. The indicator with the highest sensitivity was treatment (27.8%, 95% confidence interval [CI] 20.0-37.0), whereas the indicator with the best specificity was diagnosis (96.6%, 95% CI 91.9-98.7). The unadjusted DOR of global recognition was 2.6 (95% CI 1.5, 4.4). Less comorbidity, more severe depression symptoms, a history of depression, longer hospital stay, and antidepressant use before admission were significantly associated with better global recognition.
Conclusion: Recognition of depression in elderly medical inpatients depends upon the indicator of recognition used.