A Proof of Concept Study on Individual Trends in Suicidal Ideation: An Ecological Momentary Assessment Study of 5 Patients Over Three Months

J Pers Oriented Res. 2023 Jun 17;9(1):42-50. doi: 10.17505/jpor.2023.25265. eCollection 2023.

Abstract

Background: Suicidal ideation (SI) is a significant and long-lasting mental health problem, with a third of individuals still experiencing SI after two years. To date, most Ecological Momentary Assessment (EMA) studies of SI have assessed its day-to-day course over one to four consecutive weeks and found no consistent trends in average SI severity over time.

Aim: The current proof of concept study assessed daily fluctuations of SI over a time span of 3 to 6 months to explore whether individual trends in SI severity could be detected, and if so, if the trajectory of changes were gradual or sudden. The secondary aim was to explore whether changes in SI severity could be detected at an early stage.

Method: Five adult outpatients with depression and SI used an EMA app on their smartphone in addition to their regular treatment for 3 to 6 months, where SI was assessed 3 times a day. To detect trends in SI for each patient, three models were tested: a null model, a gradual change model and a sudden change model. To detect changes in SI before a new plateau was reached, Early Warning Signals and Exponentially Weighted Moving Average control charts were used.

Results: In each patient, average SI severity had a unique trajectory of sudden and/or gradual changes. Additionally, in some patients, increases in both sudden and gradual SI could be detected at an early stage.

Conclusions: The study presents a first indication of unique individual trends in SI severity over a 3 to 6 months period. Though replication in a larger sample is needed to test how well results generalize, a first proof-of-concept is provided that both sudden and gradual changes in SI severity may be detectable at an early stage using the dynamics of time-series data.

Keywords: Suicide; complex systems; early warning signals; ecological momentary assessment; experience sampling methods; time-series analyses.