Variation in self-identified most stressful life event by outcome of previous pregnancy in a population-based sample interviewed 6–36 months following delivery

The majority of health research uses a deductive approach to measure stressful life events, despite evidence that perception of what is stressful varies. The goal of this project was to 1) describe the distribution of self-identified most stressful life events in a cohort of women who experienced a perinatal loss (stillbirth or neonatal death) or live birth in the previous three years and 2) test how childhood adversity influences participant selection of their most stressful life event. We used data from 987 women (282 with stillbirth, 657 without loss, and 48 with a neonatal death in the first 28 days) in the Stillbirth Collaborative Research Network – OASIS (Outcomes after Study Index Stillbirth) follow-up study, a population-based sample set in five U.S. states in 2009. We applied an inductive coding process to open-ended responses to a question about the most stressful event or major crisis that participants had ever experienced, resulting in a set of 15 categories. We compare psychologic wellbeing across self-identified most stressful life event, accounting for sampling and loss-to-follow-up weights. Overall, stillbirth was most commonly identified as the most stressful event (18.3% [95% CI: 15.6, 21.5]), followed by loss by death of someone other than a child (17.25% [95% CI: 13.9, 20.3]). For participants who experienced a perinatal loss, we fit multivariable logistic regression models to quantify the association between report of childhood maltreatment and identifying the perinatal loss as the most stressful life event, calculating risk ratios (RRs). Reporting any moderate or severe childhood maltreatment was associated with 24% lower risk of identifying the perinatal loss as the most stressful life event (adjusted RR: 0.76 [95% CI: 0.58, 1.01]), after adjusting for race/ethnicity, age, and education. These results demonstrate the value of combining standardized measures with open-ended, inductive approaches to measuring stress in large, population-based studies.