(PS11-A8) Resilience in the Aftermath of Losing a Loved One: Examining Predictors of PTSD and PGD Trajectories in a Community Sample

            Bereavement can be an excruciating experience that can lead to prolonged distress reactions. Maladaptive adjustment after loss is associated with considerable public health cost (e.g., Maercker et al. 2013). Understanding the development and predictors of grief-related psychopathology (e.g., Post-Traumatic Stress Disorder, PTSD; Prolonged Grief Disorder, PGD) is therefore pressing and of great significance to prevention and intervention efforts.

            This study recruited bereaved participants (N = 305) younger than 65 years of age who had recently lost a spouse. Recruitment was conducted by sending invitation letters based on public death listings obituaries and support group referrals, as well as internet, fliers, and newspaper advertisements. Bereavement was verified by death certificates. Participants completed structured clinical interviews at 3-, 14-, and 25- months after their loss to assess symptoms of PTSD and PGD. In addition, participants completed surveys assessing flexibility, satisfaction with their marital relationship, distress tolerance, experience of continuing bonds, instrumental support, social network, ego resiliency, experience in close relationship, loneliness, and nature of death (i.e., was the death sudden).

First, latent growth mixture modeling (LGMM) was run to identify trajectories of PTSD and PGD in the aftermath of loss (Muthén & Muthén, 2019). LGMM does not rely on the assumption that individuals can be meaningfully described by a homogeneous mean response, instead it allows to tease out subpopulations (or classes) characterized by discrete growth patterns (i.e., heterogeneous reactions to the loss). For PGD, five classes were identified (i.e., moderate stable, acute recovery, moderate improve, prolonged worsening, and resilient). For PTSD, the 2-class solution was identified as the best fit (i.e., chronic and resilient).

Nex, the least absolute shrinkage and selection operator (LASSO) logistic regression, a form of supervised machine learning suited for high-dimensional data (i.e., where sample size does not substantially exceed the number of predictors), was built to predict resilience against all other trajectories. To select the optimal shrinkage parameters for the LASSO models and obtain mean cross-validation estimates of model performance, 10-fold cross-validation with three repetitions was performed when building each model. The model performances for both PTSD and PGD were fair. Both outcomes exhibited some similarly important predictors, such as loneliness. However, there were also some differences. For PTSD, the best predictor for resilience following loneliness was the size of social network, suggesting the important role of social support in the context of loss. For PGD, the most important predictor was internal continuing bond, followed by loneliness and negative absorption.

Together, these findings suggest that 1) reactions to loss are heterogenous, and 2) result from a combination of different predictors with small effects, 3) targeting perceived loneliness, social support, and distress tolerance could be fruitful approaches for psychotherapy.