In Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology
We aimed to develop and validate classification models able to identify individuals at high risk for transition from a diagnosis of depressive disorder to one of bipolar disorder. This retrospective health records cohort study applied outpatient clinical data from psychiatry and nonpsychiatry practice networks affiliated with two large academic medical centers between March 2008 and December 2017. Participants included 67,807 individuals with a diagnosis of major depressive disorder or depressive disorder not otherwise specified and no prior diagnosis of bipolar disorder, who received at least one of the nine antidepressant medications. The main outcome was at least one diagnostic code reflective of a bipolar disorder diagnosis within 3 months of index antidepressant prescription. Logistic regression and random forests using diagnostic and procedure codes as well as sociodemographic features were used to predict this outcome, with discrimination and calibration assessed in a held-out test set and then a second academic medical center. Among 67,807 individuals who received at least one antidepressant medication, 925 (1.36%) subsequently received a diagnosis of bipolar disorder within 3 months. Models incorporating coded diagnoses and procedures yielded a mean area under the receiver operating characteristic curve of 0.76 (ranging from 0.73 to 0.80). Standard supervised machine learning methods enabled development of discriminative and transferable models to predict transition to bipolar disorder. With further validation, these scores may enable physicians to more precisely calibrate follow-up intensity for high-risk patients after antidepressant initiation. Fig. 1BIPOLAR RATES AMONG ALL INDEX PRESCRIPTIONS BETWEEN 2008 AND 2017 FOR DIFFERENT ANTIDEPRESSANT CATEGORIES.: SNRIs serotonin and norepinephrine reuptake inhibitors, SSRIs selective serotonin reuptake inhibitors, MDD major depressive disorder, BP bipolar disorder.Fig. 2AREA UNDER THE CURVE (AUC) IN TEST SET FOR THE LOGISTIC REGRESSION CLASSIFIER (LEFT) AND RANDOM FOREST CLASSIFIER (RIGHT) FOR SITE A AND SITE B.: Input data include sociodemographic features, specifically age, gender, and race (dem), date of prescription (date), type of insurance (insurance), type of provider (provider), and diagnostic/procedure codes (codes). Confidence intervals computed using 500 bootstraps across 50 different splits of the data in train/test/validation sets.Fig. 3LIFT HISTOGRAM FOR THE RANDOM FOREST CLASSIFIER (1ST ROW) AND LOGISTIC REGRESSION CLASSIFIER (2ND ROW) IN SITE A (1ST COLUMN) AND SITE B (2ND COLUMN) FOR A SINGLE SPLIT.: Prescriptions are sorted according to their predicted probability of transition to bipolar disorder. The dashed line corresponds to the average BPD rate at each site respectively: bars above the dashed line correspond to patients high higher-than-average BPD risk.Fig. 4POSITIVE PREDICTIVE VALUES (PPV) VERSUS NEGATIVE PREDICTIVE VALUES (NPV) FOR THE RANDOM FOREST CLASSIFIER (1ST ROW) AND LOGISTIC REGRESSION CLASSIFIER (2ND ROW) IN SITE A (1ST COLUMN) AND SITE B (2ND COLUMN) FOR A SINGLE SPLIT.: Each blue point corresponds to a different operating point (threshold) of the classifier.
Pradier Melanie F, Hughes Michael C, McCoy Thomas H, Barroilhet Sergio A, Doshi-Velez Finale, Perlis Roy H