(PS2-83) Associations Between Ecological Momentary Assessments and Clinical Interview Ratings of Mood Symptoms in Bipolar Disorders

Abstract Text: Bipolar Disorders (BD), defined by swings in clinical states of (hypo)mania, depression, and euthymia, are linked to impaired functioning and high suicide risk. Tracking daily mood changes and first signs of episodes can effectively prevent BD episode relapse. A promising tool is mHealth, or smartphone/wireless technology that can track mood in real-time and real-world settings and deliver ecological momentary assessments (EMA) of self-reported mood daily. Gold-standard assessments rely on interview ratings of BD symptoms based on one’s retrospective report. Studies show that EMA and clinical interview ratings are equally reliable and valid for assessing BD mood symptoms within short time frames (e.g., 48 hours). Still, there is a need to examine associations between daily EMA ratings and clinical interviews over longer timeframes (weeks) to provide stronger evidence for EMA BD mood-tracking in real-world clinical settings.
The present study investigates associations between concurrent EMA (daily) and clinical interview ratings (weekly) over three months in 44 individuals with BD (25 BD I; 15 BD II; 4 BD Other Specified; mean age = 48; 68% male). Participants responded to daily self-report EMA of BD symptoms using items from the Internal State Scale (ISS) via the VA mPRO smartphone app. The ISS is a self-report instrument assessing manic symptoms (Activation subscale) and depressive symptoms (Depression subscale) for the prior 24 hours. Biweekly clinical interviews retrospectively assessed BD symptoms for each prior study week using the Young Mania Rating Scale (YMRS) and a modified Hamilton Rating Scale for Depression (MHRSD). Separate hierarchical linear models (HLM) for YMRS versus MHRSD scores examined their associations with several summary indices of EMA data (e.g., weekly maximum vs mean EMA subscale score).
Participant mean completion was 65% of EMA and 92% of clinical interview ratings. HLM predicting YMRS scores yielded significant associations between maximum, mean, and median EMA Activation indices (unstandardized b = 0.022 to 0.033). Models with mean and median EMA Activation scores yielded the best fit to data based on AIC and BIC scores and had conditional R2 of 0.39 and 0.41, respectively. HLM predicting MHRSD scores yielded significant associations between maximum, mean, and median EMA Depression indices (unstandardized b = 0.045 to 0.062). Models with mean and maximum EMA Depression scores yielded the best fit to data based on AIC and BIC scores and had conditional R2 of 0.56 and 0.54, respectively. Ongoing analyses are exploring associations based on proximity of EMA ratings to clinical interview assessment time. Figures will show daily changes in EMA for participants predominantly exhibiting different mood states.
This study identifies specific summary variables of daily EMA mood assessments strongly associated with retrospective clinical interview ratings of BD symptoms over weeks or durations that better approximate real-world clinical applications. EMA summary indices differ between BD mood states, thus guiding future studies of large EMA datasets predicting BD episode relapse. More studies are needed to fully test the reliability of these estimates in larger samples and to assess their clinical utility.