Background: Loneliness co-occurs alongside many mental health problems and is associated with poorer treatment outcomes. It could therefore be a phenomenon of interest to clinicians as an indicator of generalised risk for psychopathology. The present study tested whether a short measure of loneliness can accurately classify individuals who are at increased risk of common mental health problems.
Methods: Data were drawn from two nationally representative cohorts: the age-18 wave of the UK-based Environmental Risk (E-Risk) Longitudinal Twin Study and the age-38 wave of the New Zealand-based Dunedin Multidisciplinary Health and Development Study. In both cohorts, loneliness was assessed using the three-item UCLA Loneliness Scale, plus two stand-alone items about feeling alone and feeling lonely. Outcome measures consisted of diagnoses of depression and anxiety and self-reports of self-harm/suicide attempts, assessed via a structured interview.
Results: ROC curve analysis showed that the Loneliness Scale had fair accuracy in classifying individuals meeting criteria for all three outcomes, with a cut-off score of 5 (on a scale from 3 to 9) having the strongest empirical support. Both of the stand-alone items showed modest sensitivity and specificity but were more limited in their flexibility. The findings were replicated across the two cohorts, indicating that they are applicable both to younger and older adults. In addition, the accuracy of the loneliness scale in detecting mental health problems was comparable to a measure of poor sleep quality, a phenomenon which is often included in screening tools for depression and anxiety.
Conclusions: These findings indicate that a loneliness measure could have utility in mental health screening contexts, as well as in research.
Keywords: ROC curve; loneliness; mental health; screening.