REVIEW OF: Ross EL et al, JAMA Psychiatry2024;81(2):135–143

STUDY TYPE: Analysis of past data using machine learning

When a patient presents with active suicidality, we usually recommend psychiatric hospitalization to keep them safe. But does this actually reduce future suicidality? This study looked at whether hospitalization is effective in preventing future suicidality and aimed to create a personalized approach using advanced data analysis.

Investigators analyzed data from 196,610 visits to EDs and urgent care clinics by veterans who had either suicidal ideation (SI) or suicide attempts (SAs) between 2010 and 2015. Patients were grouped based on their psychiatric diagnosis and on the nature and timing of their suicidality (SI only, SA in the past week, or SA the day before). The primary outcome was whether patients had another SA within the next year.

• About 71.5% of these patients were hospitalized, and the rest were discharged home. The overall risk of another SA within a year was 11.9% for those who were hospitalized and 12.0% for those who weren’t, a negligible difference.
• Hospitalization significantly reduced SA risk in patients with an SA within the past day, with reductions ranging from 6.9% to 9.6% across diagnoses.
• Overall, no significant risk reduction was observed for patients with SI only or an SA in the past two to seven days. The exceptions to this trend were patients with depression who had an SA two to seven days prior to the visit—hospitalization did lower future risk of SA for this group. 
• The machine learning model identified that hospitalization reduced SA risk in 28.1% of patients but increased it in 24.0%.

CARLAT TAKE

Assessment of suicidality should be done case by case. Hospitalization for suicidality should not be a reflex decision. Those most likely to benefit with respect to future suicidality are patients with SAs the day before, or, in the case of depression, within the past week. However, for those with SI or more remote SAs, hospitalization does not clearly demonstrate benefits and may even pose risks.