Occupational performance characteristics in patients with bipolar disorders: discriminant analysis with major...

This study aimed to examine whether occupational performance characteristics observed during occupational therapy could facilitate the differential diagnosis of individuals with bipolar disorder (BD) and major depressive disorder (MDD).

A retrospective chart review was conducted on patients admitted to a university hospital for a one-week diagnostic evaluation. Eligible participants were aged 20-70 years, diagnosed with BD or MDD, and presented with at least mild depressive symptoms (Montgomery-Åsberg Depression Rating Scale ≥7). Behavioral data were systematically recorded during a structured occupational therapy session involving artistic activities. Occupational performance characteristics were identified via qualitative content analysis and coded as binary variables. Group differences were examined using nonparametric tests, followed by simultaneous discriminant analysis and receiver operating characteristic (ROC) analysis to evaluate diagnostic utility. In addition, leave-one-out cross-validation (LOOCV) was performed as an internal validation procedure to assess the stability of the discriminant model.

The study sample comprised 21 patients with BD and 35 patients with MDD. Among 66 identified occupational performance characteristics, 3 specific behaviors, namely "Messy work," "Psychomotor dysregulation," and "Work on your own without using a textbook," were significantly more prevalent in the BD group. The discriminant model incorporating these characteristics achieved an overall classification accuracy of 80.4%, which was also observed in LOOCV. ROC analysis indicated moderate discriminative power (AUC = 0.765), characterized by relatively high specificity (94.3%) and moderate sensitivity (57.1%).

Distinct occupational performance characteristics observed during occupational therapy provide clinically meaningful insights for differentiating BD from MDD during depressive episodes. Systematic behavioral observation in ecologically valid settings may serve as a valuable adjunct to conventional symptom-based diagnostic frameworks.