A nine-year follow-up of the UK Biobank cohort (Shaw et al., 2021) found that for men, living alone and living with non-partners each roughly doubled suicide mortality, while loneliness was only modestly associated with suicide death; for women, none of these social-connection measures predicted suicide. Loneliness and lack of emotional support, in contrast, predicted hospital admissions for self-harm in both sexes. The authors concluded that the causal and mechanistic structure linking social connection to lethal versus non-lethal suicidal behaviour — in particular why cohabiting with a partner protects men but not women from suicide death, and whether loneliness is causally involved — remains unresolved. Later work has replicated the living-alone/suicide association prospectively (meta-analytic pooled adjusted hazard ratio 1.27, significantly stronger in men than women), added time-varying causal survival analyses of living arrangements, and provided Mendelian-randomization evidence (a method using genetic variants as quasi-random instruments to test causal effects) consistent with a causal effect of loneliness on suicide attempts; however, no study has tested the proposed mediators (perceived burdensomeness, masculinity-related self-image) of the sex-specific cohabitation effect, no longitudinal study has predicted suicide death from repeatedly measured multi-item loneliness, and no randomized intervention targeting loneliness has assessed self-harm or suicide outcomes. The problem is to establish the causal and mechanistic role of loneliness, emotional support, and household composition in lethal versus non-lethal suicidal behaviour, including sex specificity.

Background

Suicide and hospital-treated self-harm are rare outcomes whose social antecedents are difficult to study. The UK Biobank is a cohort of over 500,000 middle-aged UK residents recruited and assessed between 2006 and 2010 and linked thereafter to hospital-admission and mortality records; Cox regression (a survival model yielding hazard ratios, HRs, for the rate of an event) can then relate baseline characteristics to subsequent suicide or self-harm. Shaw et al. (2021) used this design to study three social-connection exposures: household composition (living alone, living with a partner, living with non-partners), subjective loneliness (a single yes/no item), and perceived emotional support (a single item). For men, both living alone (HR 2.16) and living with non-partners (HR 1.80) predicted death by suicide, independently of loneliness, which had only a modest association (HR 1.43); for women, no social-connection measure predicted suicide death, though the confidence intervals could not exclude modest associations. Loneliness predicted hospital admission for self-harm in both sexes (women HR 1.89, men HR 1.74), and loneliness and emotional support explained much of the association between living alone and self-harm for men. Two prominent frameworks interpret such findings. The Interpersonal Theory of Suicide proposes that suicidal desire arises from thwarted belongingness (loneliness plus absence of reciprocal caring relationships) and perceived burdensomeness (feeling so flawed as to be a liability to others), with progression to death by suicide requiring an acquired capacity for lethal self-injury; emotional dysregulation theory instead emphasizes that some affect-regulation difficulties drive non-lethal self-harm while inhibiting feared lethal acts. Shaw et al.'s pattern — cohabitation strong for suicide death in men only, loneliness strong for self-harm in both sexes — is consistent with different pathways to lethal versus non-lethal behaviour, and the authors proposed that perceived burdensomeness and masculinity-linked self-image might mediate the male-specific protection from cohabiting with a partner; men's reluctance to seek help was raised by the authors as an alternative explanation via residual confounding rather than as a proposed mediator, and one that would not itself explain the increased risk among men living with non-partners. The study's acknowledged limitations frame what any resolution must overcome: single-item, single-timepoint exposure measurement; healthy-volunteer selection bias (only 5.5% invitation response; suicide rates below national averages); no data on personality disorders, stressful life events, or help-seeking; self-harm ascertainment restricted to English hospital admissions; and, fundamentally, observational design with possible reverse causation (poor health may drive both social disconnection and suicidality). Mendelian randomization (MR) — using genetic variants associated with an exposure as quasi-random instruments to test causality — is one design later work has applied to this question. The Interpersonal Theory constructs and the sex-specific cohabitation effect are the mechanistic target; the measurement, selection, and causal-direction issues are the inferential obstacles.

Problem Statement

Establish the causal and mechanistic role of social connection — household composition (living alone, living with a partner, living with non-partners), subjective loneliness, and perceived emotional support — in lethal versus non-lethal suicidal behaviour in the general population, in a way that resolves the specific pattern Shaw et al. (2021) documented: partner cohabitation protects men, but not women, against suicide death, while loneliness predicts hospital-treated self-harm in both sexes. Concretely, the resolution requires determining: (1) whether loneliness and lack of emotional support are causal contributors to self-harm and suicide (as opposed to markers of poor health or confounded associations), on evidence that establishes temporal ordering; (2) which mechanisms mediate the male-specific protective effect of partner cohabitation on suicide death — in particular whether perceived burdensomeness and masculinity-linked self-image and self-worth account for it; and (3) whether the theoretical pathways to lethal and non-lethal suicidal behaviour genuinely differ, as the Interpersonal Theory of Suicide (thwarted belongingness and perceived burdensomeness versus acquired capacity) and emotional-dysregulation accounts would predict. Determinations (1)–(3) are jointly required for a resolution. Where the available literature and theory do not settle a determination, the resolution should identify what further evidence — and, where relevant, what measures or design — would settle it. A further implication suggested by Shaw et al.'s results — the authors suggested that addressing loneliness in the general population may reduce the risk of self-harm, without making intervention evidence a condition for resolving their open causal and mechanistic question — is whether interventions that reduce loneliness (individually or at population level) reduce hospital-treated self-harm or suicide, and whether such effects differ by sex or household composition; this intervention question is a further implication rather than a required determination. A resolution must address, rather than reproduce, the known limitations of this literature: single-item, single-timepoint exposure measures, healthy-volunteer-biased samples, and admitted-self-harm-only outcome ascertainment.

References

  1. Shaw RJ, Cullen B, Graham N, Lyall DM, Mackay D, Okolie C, Pearsall R, Ward J, John A, Smith DJ (2021). Living alone, loneliness and lack of emotional support as predictors of suicide and self-harm: A nine-year follow up of the UK Biobank cohort. Journal of Affective Disorders 279:316-323. DOI 10.1016/j.jad.2020.10.026, https://doi.org/10.1016/j.jad.2020.10.026
  2. Luo Z, Wang J, Chen X, Cheng D, Zhou Y (2024). Assessment of the relationship between living alone and suicidal behaviors based on prospective studies: a systematic review and meta-analysis. Frontiers in Public Health 12:1444820. DOI 10.3389/fpubh.2024.1444820, https://doi.org/10.3389/fpubh.2024.1444820
  3. Narita Z, Shinozaki T, Goto A, Hori H, Kim Y, Wilcox HC, Inoue M, Tsugane S, Sawada N (2024). Time-varying living arrangements and suicide death in the general population sample: 14-year causal survival analysis via pooled logistic regression. Epidemiology and Psychiatric Sciences 33:e30. DOI 10.1017/s2045796024000325, https://doi.org/10.1017/s2045796024000325
  4. Alothman D, Lewis S, Card T, Tyrrell E, Fogarty AW (2024). The association between living alone, loneliness and suicide mortality and effect modification by age: A case:control study. Journal of Affective Disorders 352:278-280. DOI 10.1016/j.jad.2024.02.047, https://doi.org/10.1016/j.jad.2024.02.047
  5. Shoib S, Amanda TW, Saeed F, Ransing R, Bhandari SS, Armiya'u AY, Gürcan A, Chandradasa M (2023). Association Between Loneliness and Suicidal Behaviour: A Scoping Review. Turkish Journal of Psychiatry (Turk Psikiyatri Dergisi) 34(2):125-132. DOI 10.5080/u27080, https://doi.org/10.5080/u27080
  6. Lu A, Chan A, Menon T, Le GH, Wong S, Ho R, Lo HKY, Rhee TG, Lim PK, Guillen-Burgos HF, McIntyre RS (2025). Association between loneliness and suicidality among general populations and persons with depressive and bipolar disorders: A systematic review. Journal of Affective Disorders 380:777-801. DOI 10.1016/j.jad.2025.03.150, https://doi.org/10.1016/j.jad.2025.03.150
  7. Hu W, Liu TS, Tian G, Wang JN, Ma YB, Zheng ZA, Feng TJ, Niu XX, Yan YN, Jia CX, Liu BP (2026). Associations of loneliness and social isolation with the risk of suicide attempts: Insights from an observational and genetic analysis. Journal of Affective Disorders 408:121856. DOI 10.1016/j.jad.2026.121856, https://doi.org/10.1016/j.jad.2026.121856
  8. Birken M, Ikhtabi S, Steare T, Johnson S, Benedetto G, Lin F, Scott HR, Harju-Seppänen J, McCloud T, Shafran R, Pitman A (2025). Investigating the effectiveness of interventions intended to reduce loneliness using psychological strategies and a theory of change: a systematic review of interventional studies and meta-analysis. BMC Psychology 14:131. DOI 10.1186/s40359-025-03639-3, https://doi.org/10.1186/s40359-025-03639-3
  9. Gundersen K, Dzurova D, Csemy L (2026). Late-Life Suicidal Ideation in Nordic and Central-European Countries: Gender Differences and Risk Factors from the SHARE Study. Journal of Cross-Cultural Gerontology 41. DOI 10.1007/s10823-026-09576-x, https://doi.org/10.1007/s10823-026-09576-x
  10. Cranney S, Wilkinson ST (2025). The relationship between marital status and risk for suicidal thoughts, plans, and attempts: Evidence for a growing association. Journal of Affective Disorders 391:119862. DOI 10.1016/j.jad.2025.119862, https://doi.org/10.1016/j.jad.2025.119862
  11. Husky MM, Amieva H, Bergua V, Péres K, Pic O, Beck F, Léon C, Gillaizeau I, Vasiliadis HM (2026). Living alone, major depression and suicidal ideation in older adults in the community: Findings from 2005 to 2021. Journal of Affective Disorders 400:121203. DOI 10.1016/j.jad.2026.121203, https://doi.org/10.1016/j.jad.2026.121203
  12. Wahring IV, Simpson JA, Van Lange PAM (2026). Romantic Relationships Matter More to Men than to Women. Behavioral and Brain Sciences 49 (first published online December 2024). DOI 10.1017/s0140525x24001365, https://doi.org/10.1017/s0140525x24001365
  13. Lubas-Grzyb A, Rode D, Rode M, Marganski AJ (2026). Gender-Specific Correlates of Suicidal Behaviour: Insights from the Interpersonal Theory of Suicide. Journal of Clinical Medicine 15:1335. DOI 10.3390/jcm15041335, https://doi.org/10.3390/jcm15041335