Social connectedness as a predictor of suicidal outcomes in psychiatric outpatients: a 12-month prospective EMA study

Introduction

Suicide is a leading global public health problem, accounting for over 700,000 deaths each year [1]. Individuals with a history of self-harm exhibit higher rates of premature mortality from both external and natural causes [2]. This underlines the need to advance both suicide prevention strategies and our understanding of the factors that influence suicide risk.Social connectedness has been identified as a key protective factor for suicide, yet most studies rely on static, cross-sectional assessments. Ecological Momentary Assessment (EMA) allows for the capture of real-time fluctuations in risk factors and the protective nature of social connectedness. We conducted a 12-month prospective EMA study involving 550 psychiatric outpatients from four Spanish hospitals who had experienced recent suicidal thoughts or attempts.

Methods

Participants received a daily survey containing 2–4 randomly selected items from a pool of 34. These items included measures of social connectedness, active and passive suicidal ideation, and non-suicidal self-injury (NSSI). Information on suicide attempts was extracted from electronic medical records. Multilevel mixed-effects models were used to distinguish between-person (average levels) and within-person (daily deviations) associations, adjusting for age, sex, time, and missingness.

Results

There was a total of 585 participants in the original EMA sample; 35 participants were removed for having 3 or less EMA responses, leaving a final sample of 550. The average age of the sample was 44.1 years of age (SD =14.4), and reported sex was 64.4% female ,34.9% male, and 0.7% other. A total of 81,023 unique EMAs were answered in the study during the 365-day period (M=147.32 items completed per person, SD= 170.06).

The trajectories of all suicidal ideation outcomes with time as the sole predictor were non-significant (p > .05). Similarly, the slope of social connectedness across time was non-significant (p < 0.05), indicating no systematic changes during follow-up.

On days when participants reported higher-than-usual social connectedness relative to their own mean, they also reported lower active suicidal ideation (within-person effect: b = −0.18, 95% CI [−0.33, −0.04], p < .001).

Higher average connectedness was also inversely associated with NSSI (b = −0.28, p = .003). Within-person increases in connectedness also predicted reductions in active suicidal ideation (b = −0.12, p < .001), but not NSSI or passive ideation. Lastly, the only significant predictor of suicide attempts was the missingness of active suicidal ideation EMA in the 7 days prior.

Conclusions

This study shows social connectedness is a key predictor of active suicidal thoughts and, to a lesser extent, NSSI in high-risk outpatients. Both stable interpersonal vulnerabilities and moment-to-moment fluctuations in belongingness appear to be relevant to suicide risk. In contrast, social connectedness did not appear as a statistically significant predictor of suicide attempts. Instead, disengagement from EMA was the strongest behavioral signal preceding attempts. These findings highlight the need to monitor both connectedness, patterns of engagement, and to integrate passive EMA signals with self-report to better capture moments of acute suicide risk.

References

1. World Health Organization (WHO). Suicide. Retrieved from https://www.who.int/news-room/fact-sheets/detail/suicide; 2024.
2. Bergen H, Hawton K, Waters K, et al. Premature death after self-harm: A multicentre cohort study. Lancet 2012;380(9853):1568–1574; doi.org/10.1016/S0140-6736(12)61141-6