Citation

Dolsen EA, Nishimi K, LeWinn KZ, Byers AL, Tripp P, Woodward E, Khan AJ, Marx BP, Borsari B, Jiha A, Neylan TC, O'Donovan A. J. Psychiatr. Res. 2022; 156: 186-193.

Copyright

(Copyright © 2022, Elsevier Publishing)

DOI

10.1016/j.jpsychires.2022.10.009

PMID

36252348

Abstract

The coronavirus disease 2019 (COVID-19) pandemic has created a global health crisis, with disproportionate effects on vulnerable sociodemographic groups. Although the pandemic is showing potential to increase suicide ideation (SI), we know little about which sociodemographic characteristics or COVID-19 experiences are associated with SI. Our United States-based sample (n = 837 adults [mean age = 37.1 years]) completed an online survey during August-September 2020. The study utilized an online convenience sample from a prior study, which was enriched for exposure to trauma and experiences of posttraumatic stress symptoms. We assessed SI using the Beck Depression Inventory-II. Traditional (i.e., logistic regression) and machine learning (i.e., LASSO, random forest) methods evaluated associations of 148 self-reported COVID-19 factors and sociodemographic characteristics with current SI. 234 participants (28.0%) reported SI. Twenty items were significantly associated with SI from logistic regression. Of these 20 items, LASSO identified seven sociodemographic characteristics (younger age, lower income, single relationship status, sexual orientation other than heterosexual as well as specifically identifying as bisexual, non-full-time employment, and living in a town) and six COVID-19 factors (not engaging in protective COVID-19 behaviors, receiving mental health treatment (medication and/or psychotherapy) due to the COVID-19 pandemic, socializing during the pandemic, losing one's job due to COVID-19, having a friend with COVID-19, and having an acquaintance with COVID-19) associated with SI. Random forest findings were largely consistent with LASSO. These findings may inform multidisciplinary research and intervention work focused on understanding and preventing adverse mental health outcomes such as SI during and in the aftermath of the pandemic.

Language: en

Keywords

Suicide; Depression; Machine learning; Covid-19; Posttraumatic stress disorder