Mapping the gender gradient in posttraumatic stress disorder prevalence: A machine learning approach

Area of Science:

• Psychiatry
• Sociology
• Computational Psychology

Background:

• Posttraumatic stress disorder (PTSD) is diagnosed more frequently in women than men.
• The specific mechanisms driving this gender disparity remain unclear.
• Traditional studies often rely on binary sex comparisons, potentially overlooking nuanced factors.

Purpose of the Study:

• To investigate the impact of gender-linked sociocultural factors on PTSD.
• To move beyond binary sex categories in PTSD research.
• To quantify the association between gendered social circumstances and PTSD.

Main Methods:

• Developed a supervised machine learning model to derive continuous gender scores from predictors of roles, relationships, and institutional facets.
• Assessed the association of these gender scores with past-year PTSD in trauma-exposed individuals (N=23,936).
• Compared the predictive power of gender scores against binary sex classifications.

Main Results:

• A significant dose-response relationship was observed between female-typical social circumstances and past-year PTSD.
• Gender scores demonstrated a stronger association with PTSD (aOR=4.03) than binary sex (aOR=2.69).
• These findings persisted after controlling for trauma exposure and other risk factors.

Conclusions:

• Sociocultural factors associated with gender play a crucial role in the development of PTSD.
• Quantitative approaches that consider continuous gender dimensions offer deeper insights than binary sex comparisons.
• This research underscores the need to examine gendered social contexts in understanding traumatic stress.