Thoughts of self‐harm in late adolescence as a risk indicator for mental disorders in early adulthood

Early intervention for youth mental disorders has received increasing attention in recent decades. For psychosis, this is exemplified by the clinical high‐risk (CHR) paradigm, which has been highly successful in defining a subpopulation at enhanced risk. However, the subpopulation captured by CHR services represents a small proportion of all psychosis cases ¹ , highlighting the need for additional approaches to early detection of at‐risk individuals.

Thoughts of self‐harm are common in youth populations and are associated with several psychiatric outcomes. A recent Finnish registry study found that 18% of young people in Finland who presented to hospital with self‐harm were diagnosed with a psychotic disorder by age 28 ² , suggesting that hospital presentation with self‐harm may be a system‐based risk marker for psychosis. However, most individuals with self‐injurious thoughts or behaviours do not present to hospital, and only a small proportion (4%) of future psychosis cases were captured in that study.

Expanding on this approach, we examined whether having thoughts of self‐harm in late adolescence (irrespective of hospital presentation) was a risk indicator for development of psychotic disorder, as well as depressive disorder and generalized anxiety disorder (GAD), in early adulthood. In exploratory secondary analyses, we also examined whether telling a general practitioner (GP) about thoughts of self‐harm was a risk marker for these disorders.

The sample was drawn from the Avon Longitudinal Study of Parents and Children (ALSPAC) ³ , ⁴ , ⁵ . Pregnant women in Avon, UK with expected delivery dates between April 1, 1991 and December 31, 1992 were invited to participate. 14,541 pregnancies were enrolled (13,988 children alive at 1 year of age). When the oldest children were approximately 7 years of age, an attempt was made to bolster the initial sample with eligible cases who did not join originally. The total sample size for analyses using data collected after age 7 is 15,447 pregnancies (14,901 children alive at 1 year of age). Data were collected and managed using REDCap ⁶ , ⁷ . Ethical approval was obtained from ALSPAC Ethics and Law Committee and local research ethics committees. Informed consent for use of questionnaire and clinic data was obtained following recommendations of the above‐mentioned Committee.

At age 17, participants completed the Clinical Interview Schedule Revised (CIS‐R) ⁸ . This included a question asking whether the participant had thoughts of self‐harm in the week prior to assessment. This was coded as a binary exposure variable (yes/no).

At age 24, participants completed the semi‐structured Psychosis‐Like Symptoms Interview (PLIKSi) to assess for psychotic experiences ⁹ . Psychotic disorder was defined as having at least one definite psychotic experience (not attributable to sleep or fever) which recurred at least once per month over the previous six months, and was associated with severe distress, or marked impairment of the participant's social or occupational functioning, or led him/her to seek professional help. We also examined outcomes of moderate/severe depressive disorder and GAD, defined according to the ICD‐10, based on responses to the CIS‐R completed at age 24.

At age 17, where participants reported thoughts of self‐harm, they were also asked if they had spoken to their GP about their thoughts. This variable was coded with four categories: no thoughts of self‐harm; told no‐one; told someone other than their GP; told their GP.

Primary analyses used logistic regression to evaluate associations between thoughts of self‐harm at age 17 and psychotic disorder, depressive disorder and GAD at age 24. Secondary analyses used logistic regression to evaluate associations between telling someone about thoughts of self‐harm at age 17 and the same outcomes at age 24. For all analyses, “no thoughts of self‐harm” was the reference category. For each analysis, participants who already met criteria for the relevant outcome at age 17 were excluded. In keeping with the predictive nature of this study, models were not adjusted for potential confounders. Analyses were performed using Stata 17 (StataCorp).

Participants assessed at age 17 and having data available on thoughts of self‐harm were 4,563. Following exclusion of subjects who met outcomes criteria at age 17, the numbers of participants in each analytical sample were 2,591 for psychotic disorder; 2,622 for depressive disorder; and 2,628 for GAD. The numbers of participants who reported thoughts of self‐harm at age 17 in each analytical sample were 267 (10.3%), 234 (8.9%), and 247 (9.4%), respectively (see also supplementary information).

Of the 18 participants who met criteria for psychotic disorder at age 24, 8 (44.4%) had reported thoughts of self‐harm at age 17. The corresponding numbers were 34 of 157 (21.7%) among those with depressive disorder and 50 of 205 (24.4%) among those with GAD at age 24. On the other hand, the absolute risk of psychotic disorder by age 24 among those with thoughts of self‐harm at age 17 was 3.0% (odds ratio, OR: 7.15, 95% CI: 2.80‐18.27), while it was 14.5% for depressive disorder (OR: 3.19, 95% CI: 2.12‐4.78); and 20.2% for GAD (OR: 3.64, 95% CI: 2.57‐5.17).

Secondary analyses provided evidence of associations between telling a GP about thoughts of self‐harm at age 17 and psychotic disorder (OR: 19.34, 95% CI: 5.11‐73.24), depressive disorder (OR: 14.42, 95% CI: 6.20‐33.53) and GAD (OR: 5.00, 95% CI: 2.20‐11.35) at age 24 (see also supplementary information).

These results suggest that a large proportion of those who develop psychotic disorder (44.4%) may be captured through screening for thoughts of self‐harm in late adolescence. On the other hand, of all those endorsing thoughts of self‐harm at age 17, only 3% developed a psychotic disorder at age 24; 14.5% developed depressive disorder; and 20.2% developed GAD. The simplicity of this approach is that it is based on a single reported symptom. However, in isolation, its utility for defining an at‐risk subgroup is limited, due to low positive predictive values. Nonetheless, the findings underscore the importance of appropriate long‐term follow‐up for young people with thoughts of self‐harm in relation to distal mental health outcomes.

Secondary analyses indicated that presenting to a GP with thoughts of self‐harm may be a particular indicator of risk for psychotic disorder in early adulthood, as well as for depressive disorder and GAD. This suggests a possible system‐based approach for early detection in primary care. However, these results should be viewed as preliminary and interpreted with caution, given the small numbers of participants in the exposure category.

It is notable that effect estimates were highest for psychotic disorder compared to depressive disorder or GAD. However, confidence intervals overlapped, in keeping with the view that thoughts of self‐harm in late adolescence may be a transdiagnostic risk marker.

One possible explanation of our findings is that endorsement of thoughts of self‐harm in late adolescence captures young people exposed to known transdiagnostic risk factors for future mental disorders, such as bullying and other forms of childhood adversity, socio‐economic disadvantage and substance use problems. However, the aims of this study were predictive rather than explanatory, and causal inferences cannot be drawn. If confirmed in further populations, these findings suggest novel opportunities for early detection of young people at risk of mental disorders in early adulthood.