Dose-Response Association Between Psychological Distress and Risk of Completed Suicide in the General Population

Elevated suicide rates in people with clinical depression, as indexed by hospitalizations or use of psychiatric outpatient services, are well documented.¹ However, the association between depression across the full range of severity and subsequent suicide risk is unknown. With single-cohort studies insufficiently powered to examine this relation, to our knowledge, we provide the first pooling of individual-level data from a series of large general population–based cohort studies.

Described in detail elsewhere,² 16 independent, geographically representative surveys conducted between 1994 and 2008 of individuals living in private households were used in the present analyses: the Health Survey for England (N = 13) and the Scottish Health Surveys (N = 3). Combining these studies resulted in a total of 193 873 participants, 166 606 (86%) of whom had data on age, sex, and psychological distress. Study members were linked to the UK National Health Service register for primary and contributing causes of death. Ethical approval was given by the London Research Ethics Council, and informed consent was obtained from all participants.

Psychological distress was measured using the 12-item General Health Questionnaire, which contains items principally concerned with symptoms of depression and anxiety. The sensitivity (0.70) and specificity (0.80) against standardized psychiatric interviews are acceptably high.³ In the present analyses, participants were classified according to standard thresholds: asymptomatic (score of 0), subclinically symptomatic (1-3), symptomatic (4-6), and highly symptomatic (7-12).² We used any mention of suicide on the death certificate as our outcome (associations were very similar in analyses using suicide as the underlying cause).

We computed 2 sets of analyses. First, having used Schoenfeld residuals to ascertain that the proportional hazards assumption had not been violated, we calculated hazard ratios (HRs) and accompanying 95% CIs for the association between the categories of psychological distress and suicide using Cox proportional hazard models. Having found no evidence of effect modification by sex (P for interaction = .81), data for men and women were combined. We adjusted HRs for several covariates, the selection of which was based on empirical evidence—in the present data set, the existing literature,⁴ or both—that they are associated with both suicide and psychological distress: socioeconomic position, marital status, frequency of alcohol consumption, smoking status, and presence of a somatic long-standing illness. We accounted for between-study variation using a shared frailty parameter.⁵ Additionally, to allow us to explore inflections in the distress-suicide relation, we used fractional polynomials to estimate the best-fitting dose-response curve for the full distress scale and suicide.

There were 108 deaths ascribed to suicide during a mean duration of 9.5 years of follow-up (1 581 805 person-years). Compared with the asymptomatic group, adjusted HRs (aHRs) for suicide were raised for participants in the symptomatic (aHR, 1.83; 95% CI, 0.99-3.39) and highly symptomatic (aHR, 2.43; 95% CI, 1.38-4.27) groups (P for trend <.001; Table). A 1-SD increase in psychological distress was associated with a 1.29-fold elevation (95% CI, 1.12-1.48) in the risk of suicide.

The Figure shows the best-fitting dose-response curve for psychological distress and suicide. Scores of 2 or more on the distress scale were associated with a stepwise elevation in the risk of suicide.

We observed a dose-response relationship between a single administration measure of psychological distress symptom severity and suicide risk in the general population up to 17 years later. That this association was apparent after adjustment for a range of confounding factors, including long-standing illness, suggests other mechanisms underlie the association. One possibility is stressful life experiences combine with a predispositional vulnerability involving multiple neurobiological pathways, such as serotonergic and noradrenergic systems and the ventromedial prefrontal cortex (diathesis-stress model).⁶ Our findings raise the question whether health care professionals should pay attention to suicide risk at distress levels lower than current recommendations suggest.

Corresponding Author: Steven Bell, PhD, University College London, Research Department of Epidemiology and Public Health, 1-19 Torrington Pl, London WC1E 6BT, England (steven.bell@ucl.ac.uk).

Published Online: November 11, 2015. doi:10.1001/jamapsychiatry.2015.2107.

Author Contributions: Dr Bell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors reviewed and agreed on the final manuscript as well as the decision to submit the manuscript for publication.

Study concept and design: Batty.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Bell, Batty.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Bell.

Administrative, technical, or material support: Russ, Stamatakis.

Study supervision: Batty.

Conflict of Interest Disclosures: None reported.

Funding/Support: Dr Bell is supported by grants from the European Research Council and UK Medical Research Council/Alcohol Research UK. Drs Russ and Batty are members of the University of Edinburgh Centre for Cognitive Ageing and Cognitive Epidemiology, part of the cross council Lifelong Health and Wellbeing Initiative, for which funding from the Biotechnology and Biological Sciences Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, and UK Medical Research Council is gratefully acknowledged. Prof Kivimäki is supported by the UK Medical Research Council, the US National Institutes of Health, NordForsk, the Nordic Council of Ministers, and a professorial fellowship from the UK Economic and Social Research Council.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.