Leonardo Tondo1, Caterina Visioli2, Antonio Preti2, Ross J Baldessarini3. 1. International Consortium for Bipolar Disorder Research, Mailman Research Center, McLean Division of Massachusetts General Hospital, Boston, MA, USA; Centro Lucio Bini Mood Disorders Center, Cagliari and Rome, Italy. Electronic address: ltondo@mclean.harvard.edu. 2. Centro Lucio Bini Mood Disorders Center, Cagliari and Rome, Italy. 3. International Consortium for Bipolar Disorder Research, Mailman Research Center, McLean Division of Massachusetts General Hospital, Boston, MA, USA; Department of Psychiatry, Harvard Medical School, Boston, MA, USA.
Abstract
OBJECTIVE: Most first lifetime episodes among persons eventually diagnosed with bipolar disorder are depressive, often with years of delay to a final differentiation from unipolar major depression. To support early differentiation, we tested several predictive factors for association with later diagnoses of bipolar disorder. METHOD: With data from mood-disorder patients with first-lifetime episodes of major depression, we used multivariate, logistic modeling and Bayesian methods including Receiver Operating Characteristic curves to evaluate ability of one or more selected factors to differentiate patients who later met DSM-IV-TR diagnostic criteria for bipolar disorder and not unipolar major depressive disorder. RESULTS: We analyzed data from 2146 patients (642 bipolar, 1504 unipolar) at risk for 13 years following initial depressive episodes. In multivariate modeling for 812 subjects with information on all clinical factors considered, seven significantly and independently differentiated bipolar from unipolar disorders, ranking (by significance): (a) ≥4 previous depressive episodes, (b) suicidal acts, (c) cyclothymic temperament, (d) family history of bipolar disorder, (e) substance-abuse, (f) younger-at-onset, or onset-age <25, and (g) male sex; four of these (c, d, f, g) can be identified at illness-onset. Bayesian analysis indicated optimal sensitivity and specificity at 2-4 factors/person and correct classification of 64-67% of cases, and ROC analysis of factors/person yielded a significant area-under-the-curve of 0.72 [CI: 0.68-0.75]. CONCLUSIONS: In multivariate modeling, 7 factors were significantly and independently associated with bipolar disorder diagnosed up to 13 years after initial depression.
OBJECTIVE: Most first lifetime episodes among persons eventually diagnosed with bipolar disorder are depressive, often with years of delay to a final differentiation from unipolar major depression. To support early differentiation, we tested several predictive factors for association with later diagnoses of bipolar disorder. METHOD: With data from mood-disorderpatients with first-lifetime episodes of major depression, we used multivariate, logistic modeling and Bayesian methods including Receiver Operating Characteristic curves to evaluate ability of one or more selected factors to differentiate patients who later met DSM-IV-TR diagnostic criteria for bipolar disorder and not unipolar major depressive disorder. RESULTS: We analyzed data from 2146 patients (642 bipolar, 1504 unipolar) at risk for 13 years following initial depressive episodes. In multivariate modeling for 812 subjects with information on all clinical factors considered, seven significantly and independently differentiated bipolar from unipolar disorders, ranking (by significance): (a) ≥4 previous depressive episodes, (b) suicidal acts, (c) cyclothymic temperament, (d) family history of bipolar disorder, (e) substance-abuse, (f) younger-at-onset, or onset-age <25, and (g) male sex; four of these (c, d, f, g) can be identified at illness-onset. Bayesian analysis indicated optimal sensitivity and specificity at 2-4 factors/person and correct classification of 64-67% of cases, and ROC analysis of factors/person yielded a significant area-under-the-curve of 0.72 [CI: 0.68-0.75]. CONCLUSIONS: In multivariate modeling, 7 factors were significantly and independently associated with bipolar disorder diagnosed up to 13 years after initial depression.
Authors: A K Leonpacher; D Liebers; M Pirooznia; D Jancic; D F MacKinnon; F M Mondimore; B Schweizer; J B Potash; P P Zandi; F S Goes Journal: Psychol Med Date: 2015-04-08 Impact factor: 7.723