Suicidal Ideation After Discharge From Psychiatric Hospital: Momentary Assessment Study

High suicide risk is observed after discharge among patients hospitalized with suicidal thoughts and behaviors. Suicidal ideation (SI) is a well-known precursor to suicide attempts (SAs) and suicide, and it is a central indicator of subjective distress.

This study aimed to examine patient characteristics and predischarge symptoms that might (1) distinguish between participants with and without postdischarge SI, (2) predict peak SI frequency, and (3) predict the intensity and longitudinal trajectory of SI.

Before discharge, patients hospitalized due to high suicide risk were screened for diagnosis (MINI 7.0.2 [Mini International Neuropsychiatric Interview] and SCID-5-PD [Structured Clinical Interview for DSM-5 Personality Disorders]), general symptom severity (Outcome Questionnaire-45 [OQ-45]), depression (Patient Health Questionnaire-9 [PHQ-9]), SI (Suicide Status Form-IV [SSF-IV] and Beck Scale for Suicide Ideation), and suicidal behaviors (Suicide Attempt Self-Injury Count). Twenty male and 16 female adult participants installed a mobile app on their smartphones and reported their level of SI on a 5-point Likert scale (1=not at all to 5=very much) 5 times per day for 10 days after discharge. Associations between baseline characteristics and postdischarge SI were analyzed using logistic regression, median regression, and mixed-effects linear regression.

SI was present in 49.2% (528/1073) of all completed surveys. Women reported SI more frequently than men (χ21=63.39; P=.001). No baseline variables differentiated participants who later did (28/36, 77.8%) and did not (8/36, 22.2%) report momentary SI after discharge. Participants hospitalized due to a high risk of suicide (12/36, 33.3%), compared with those hospitalized following an SA (24/36, 66.7%), reported SI more often (χ21=22.64; P=.001) and had a higher frequency of peak SI scores (β=17.74, 95% CI 8.56-26.91; P=.001). Moreover, the variation in the temporal trajectory of postdischarge SI was characterized by large differences between participants (intraclass correlation coefficient [ICC]=0.682), while the intensity of SI did not change over the 10-day period at the group level (β=0.03, 95% CI -0.01 to 0.06; P=.09). Predischarge scores on the SSF-IV chronic subscale, including hopelessness, self-hate, and psychological pain, predicted the intensity of SI after discharge (β=0.13, 95% CI 0.05-0.20; P=.002). No baseline variables predicted the trajectory of SI.

SI after discharge was common, varied substantially between participants, and may be more prevalent than previously reported, particularly in groups with high morbidity and prior SAs. The SSF-IV may be a valuable tool for treatment planning and assessment during the inpatient stay.