Patterns of stress generation differ depending on internalizing symptoms, alcohol use, and personality traits in early adulthood: a five year longitudinal study

ABSTRACT

Background

Depression is thought to generate stressful life events. However, other internalizing symptoms such as anxiety or post-traumatic stress and individual difference variables such as personality traits and alcohol use may contribute to stressful life events. Whether stress generation is specific to depression or generalized to these other variables is unclear. Therefore, we tested whether stress generation was depression specific or generalizable to anxiety, PTSD, alcohol use, neuroticism, and extraversion.

Design

Two-wave longitudinal study with a five-year follow-up.

Methods

917 young adults completed measures of internalizing symptoms, alcohol use, neuroticism, and extraversion during college and five years later along with an interview-based measure of life events.

Results

Symptoms of depression, anxiety, PTSD, and neuroticism exhibited bivariate predictive effects on interpersonal-dependent events. When considering internalizing symptoms in the aggregate, stress generation was specific to symptoms rather than neuroticism. Furthermore, interpersonal-dependent life events mediated Time 1 internalizing symptoms predicting Time 2 symptoms.

Conclusion

Our results indicate that stress generation applies to internalizing symptoms broadly rather than specifically to depression. Moreover, neuroticism was no longer a significant predictor of life events when examined with internalizing symptoms simultaneously. These results support the value of integrative models that test numerous factors predicting stressful life events.

KEYWORDS:

• Internalizing symptoms
• alcohol use
• neuroticism
• extraversion
• life events
• stress generation

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 All effects remained significant in our mediation model when ethnic/minority status was used as a covariate.

2 We conducted additional models to examine how the results may have changed if we used all 1142 participants that were invited, but did not participate at follow-up. Using either a saturated FIML model or multiple imputation, we observed the same pattern of results as obtained from the sample of 917 who participated at follow-up.

Additional information

Funding

This research was supported by grant P60-AA003510 from NIAAA. Data available in https://osf.io/93gu5/?view_only=3f1b855d3e6b4b4da3ee9c435163f978.