Poster #32 - Positive Affect Improves a Transdiagnostic Model of Perinatal Depression Symptoms

• In Event: Poster Session 04

Abstract

Depression is difficult to measure and conceptualize in any population and particularly in perinatal samples. Depression may not be a unidimensional construct (Wang & Gorenstein, 2013; Lee King, 2012) and it is frequently comorbid with other psychiatric disorders (e.g., Kessler et al., 2003; Le Strat et al., 2011; Steffen et al., 2020). Additional challenges specific to measuring and conceptualizing depression in pregnant and postpartum (perinatal) people include the overlap between symptoms of depression and normative perinatal experience (e.g., changes in appetite or sleep; Matthey & Ross-Hamid, 2011) and changes in sensitivity/specificity of depression measures across the perinatal period (Ji et al., 2011). Despite these challenges, accurate measurement of depression during the perinatal period is vital, since perinatal depression is common and associated with long-term adverse outcomes for children (Stein et al., 2014; Woody et al., 2017). To address these concerns, in a recent publication we sought to characterize symptoms of depression using transdiagnostic factors across the perinatal period (*redacted for submission*, 2020). However, our original factor model did not test whether the addition of a factor comprised of items measuring affective states with a positive valence (positive affect; PA) improved our model fit. The aims of the current study were to replicate and refine that transdiagnostic model of depression symptoms by 1) testing whether a factor that measured PA bettered our model fit and 2) replicating our factor model in a second, more socio-demographically diverse sample. We conducted secondary analyses of prospective, longitudinal data from two samples of women at risk of perinatal depression due to a history of neuropsychiatric illness. Sample 1 (n = 657) was 90% White, 87% married, and middle socioeconomic status based on Juhn et al.’s (2011) Hollingshead socioeconomic quartiles. Sample 2 (n = 142) was 74% White, 44% married, younger, and in the lowest socioeconomic quartile. Data were item scores from seven commonly used measures, repeatedly measured over pregnancy and the first year postpartum. We compared fit indices (AIC, BIC, CFI, TLI, and RMSEA) from our original factor model, which included one general and six specific factors derived from the Research Domain Criteria (Loss, Potential Threat, Frustrative Nonreward, and Sleep-Wakefulness) and depression literatures (Somatic and Coping) to our novel factor model, which included the original factors plus our proposed PA factor. In both samples, the addition of a PA factor improved model fit for almost all fit indices in both samples. We found at least partial metric invariance between perinatal periods, except for trimester 3 – postpartum period 1. In two independently collected samples, a PA factor improved a transdiagnostic model of perinatal depression symptoms. Important future directions include understanding how levels of PA increase resilience of vulnerability to other depression symptoms in the perinatal period and how these relate to child outcomes.

Authors

• Blaire Caroline Pingeton, Emory University

• Amy Cochran, University of Wisconsin-Madison

• Sherryl Hope Goodman, Emory University

• Heidemarie Laurent, Pennsylvania State University, University Park

• Marissa D. Sbrilli, University of Illinois at Urbana-Champaign

• Bettina Knight

• D. Jeffrey Newport, UT Health Austin

• Zachary Stowe

• Katherine Anne Lee, Emory University
  Presenting Author