A comprehensive analysis of features that suggest bipolarity in patients with a major depressive episode: Which is the best combination to predict soft bipolarity diagnosis?

Background udy aimed to identify specific predictors of soft bipolarity (bipolar II disorder or bipolar disorder not otherwise specified) in depressed patients and to evaluate the global predictive performance of combinations of these predictors. s ts included 199 patients with a major depressive episode (MDE) due to soft bipolarity or major depressive disorder. Independent predictors of soft bipolar diagnosis were extracted from 12 previously proposed bipolar features using multiple logistic regression analyses, and the global performance of the combination of these predictors was evaluated using a receiver operating characteristic (ROC) curve. s ent MDEs, family history of bipolar disorders in first-degree relatives, cyclothymic temperament, early age at onset of first MDE, and depressive mixed state were independent predictors of soft bipolarity diagnosis [odds ratio (95% confidence interval): 11.22 (2.19–57.63), 8.82 (1.31–59.15), 7.32 (2.22–24.19), 6.22 (1.58–24.57), and 5.57 (1.91–16.30), respectively]. The area under the ROC curve for the relationship between soft bipolarity diagnosis and the number of these five predictors in each patient was 0.911 (highly accurate). The presence of one or more predictors in each patient resulted in highest sensitivity (92.5%) and good specificity (73.1%), whereas that of two or more predictors resulted in good sensitivity (70.0%) and highest specificity (97.5%) for soft bipolarity diagnosis. tions ured/semistructured interviews were not used. Tools for temperament assessments were different between institutions. sions ination of these predictors was quite helpful for a precise diagnosis of soft bipolarity in patients with depression.