A comparison of treatment response between college students with and without suicide risk at university counseling centers

ABSTRACT

This study aims to explore the effect of University Counseling Center (UCC) treatment in a nationally representative sample of 101,354 college students with suicide risk variables (i.e. recent suicidal ideation (SI), recent SI and history of suicide attempts (SA), history of SA) as compared to those without suicide risk seeking services from 160 UCCs. Time Varying Effects Modeling was used to assess changes in distress level and SI over the course of up to 20 sessions of treatment among students with the aforementioned risk variables compared to those without. Reductions in distress and SI were observed in all students with the most significant change seen in the first five sessions. Students with recent SI and no history of attempts seem to achieve comparable amount of improvement to students without suicide risks. Presence of suicide attempt history appears to limit the prognosis. These results entail that students with a history of SA whose levels of risk and chronicity of symptoms may require more treatment and be better managed with alternative forms of treatment.

KEYWORDS:

• Suicidal ideation
• suicide attempt
• suicide risk
• university counseling centers
• treatment

Acknowledgements

This work was supported by the Center for Collegiate Mental Health.

Disclosure statement

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

Data Availability

The data reported in this manuscript were obtained from Center for Collegiate Mental Health. http://ccmh.psu.edu/available-data

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/28324765.2023.2169583

Notes

1. We used the R computing software to perform the analyses in this paper, in particular the pffr package. The framework on which this package is based theoretically enables multiple partially crossed random effects (Scheipl et al., 2015), however the current implementation of the package does not allow for this with longitudinal data (i.e. sparse functional data, in functional data parlance). A fairly new tvem package in R also implements TVEM but does not currently allow for any random effects or hierarchical structure.