A new meta-analysis by Chau and colleagues (2019) examines the association between loneliness and psychotic symptoms. As another recent meta-analysis had already found a positive association between the two (Michalska et al., 2018), perhaps one of the most interesting aspects of this paper is how the authors framed their study in terms of wider debates about how we should understand mental health, particularly psychosis.
The first debate concerns whether we should understand mental health in terms of unified disorders (the traditional psychiatric view), correlated dimensions (a view more common amongst psychologists, e.g. Kruger & Eaton, 2015), or as individual symptoms (e.g. network models of mental health; Borsboom, 2017).
The authors start off by assuming a two-dimensional model of psychosis. More specifically, loneliness is examined in relation to positive (e.g. hallucinations, delusions) and negative (e.g. asociality, anhedonia) symptoms. They also examine the association between loneliness and individual symptoms.
The second debate concerns whether clinical, ‘diagnosable’ cases of psychosis should be understood as being on a continuum with psychotic experiences in the general population or as being categorically different. The study authors assume a continuum model and therefore examine the association between loneliness and psychosis in both clinical and non-clinical populations.
So, can the experience of loneliness shed any light on these debates and how we might understand mental health more generally?
The search strategy included a search of five electronic databases (PsychINFO, Medline, Scopus, Web of Science and EMBASE) up to October 2018, a scoping review of Google Scholar and checking reference lists of previous reviews.
There were two inclusion criteria. All studies using a quantitative measure of (1) psychosis, and (2) loneliness, were included.
A number of different analyses were conducted to investigate the association between psychosis and loneliness, including:
- Random-effects meta-analysis of associations (Pearson’s r) with separate analyses for positive and negative symptoms;
- Exploratory analyses for individual symptoms;
- Sub-group analyses comparing ‘clinical’ (diagnosed) and ‘non-clinical’ (no diagnosis reported) samples were conducted.
Studies were quality checked using a tool developed by the Agency for Health Research and Quality.
31 studies met the inclusion criteria with a total of 17,927 participants:
- Clinical = 2,613
- Non-clinical = 314
- Undifferentiated (clinical status unknown/clinical and non-clinical samples combined) = 15,000
All studies met the authors’ quality threshold criteria. In terms of potential bias, there was little evidence of publication bias from funnel plot analyses though studies with smaller sample sizes demonstrated larger effect sizes.
The authors first reported the association between loneliness and positive psychotic symptoms (30 studies, n=17,832). The pooled sample, including all positive symptoms, revealed a medium sized association (r=0.302). However, when looking at these separately, a medium to large association was found for paranoia and loneliness, and a small to medium association for hallucinations and loneliness.
Comparing the clinical and non-clinical samples, there was a small-to-medium association between positive symptoms and loneliness in clinical samples and a medium-to-large association in non-clinical samples (r=0.389). The difference between clinical and non-clinical samples was significant (p=0.010).
The association between loneliness and negative symptoms was evaluated in 15 studies (n=5,567). Across all 15 studies a medium sized association between loneliness and all negative symptoms was found (r=0.347). The association with individual negative symptoms was not examined.
Similarly to the findings for positive symptoms, a small-to-medium association was found for clinical samples (r=0.127) and a medium-to-large association found for non-clinical samples (r=0.476). Again, this difference between samples was significant (p<0.001).
The authors conclude that loneliness has a medium association with both positive and negative psychotic symptoms, though within the positive dimension, loneliness was more strongly associated with paranoia than with hallucinations.
There were also significant differences between clinical and non-clinical samples with loneliness showing a stronger association with symptoms in non-clinical samples.
Strengths and limitations
Should we take a single symptom approach?
One strength of this study is that the authors examined psychosis by operationalising this experience in different ways. That is, loneliness was examined in relation to a two-dimensional model (positive and negative symptoms), as well as individual symptoms. Though overall symptoms were not examined.
Paranoia showed a stronger association with loneliness than did hallucinations. This lends some support to the single symptom approach which suggests that risk and protective factors correspond to individual symptoms of mental health problems rather than unified disorders (Bentall et al., 2012).
This might inform the future development and evaluation of loneliness interventions. For example, interventions might be evaluated for their effectiveness in reducing paranoia, rather than psychotic symptoms more generally.
Difficulties in examining loneliness in relation to single mental health conditions
One limitation is that confounding factors between loneliness and psychosis were not taken into account. A recent systematic review of the association between loneliness and psychosis reported mixed findings, with depression appearing to be a confounding factor (Lim et al., 2018). In the current meta-analysis the authors included a wide range of psychotic disorders in their analysis, including mood disorders with psychotic features. It may have been interesting to examine the association between loneliness and different disorders.
The relationship between loneliness, psychosis and general mental health is complex. It is difficult to determine the strength of the association between loneliness and single disorders such as psychosis without taking other mental health conditions into consideration.
A psychosis continuum?
A second limitation is that the authors started by assuming a continuum model of psychosis and interpreted the results in accordance with this model without considering alternative explanations. However, it is not clear that the results of this study support this model.
Significant differences in the association between loneliness and experiences of psychosis were found between the clinical and non-clinical samples. This appears to contradict the continuum model insofar as you would expect a consistent association across samples (though note it is difficult to assess from the data whether there could be relevant non-linear distributions, such as a dose-response curve).
The authors suggest that the results may be due to a score range restriction in clinical samples whereby participants score the highest possible scores. However, the data from the included studies did not indicate any ceiling effects for the measures used.
A second possibility, consistent with the continuum model, is that there are other variables which moderate the strength of association at different degrees of severity. However, most of the candidate variables, such as stigma, social support and interpersonal competence (Lim et al., 2015), would predict a stronger association with loneliness in clinical samples whilst the results of this paper found a weaker association.
So, finally, we come to the most parsimonious interpretation; that there are in fact relevant differences in the experience of psychosis between clinical and non-clinical populations (i.e. supporting a categorical model of psychosis, not a continuum). It may be that loneliness is a pertinent risk factor for psychosis-like experiences in the general population but is less of a concern for clinical populations. If this were true then it could help to prioritise interventions with different treatment targets for particular groups.
The main message here is probably a reminder for all of us to be mindful of the assumptions that we make in relation to mental health given how heavily these influence our research, practice and the understanding of our own mental health.
Implications for practice
Loneliness has a detrimental influence on a variety of physical and mental health outcomes (Hare-Duke, 2017). The findings of this study suggest that loneliness is moderately associated with psychotic symptoms. Reducing loneliness amongst service users might therefore be one aim for mental health professionals.
Preliminary evaluations of loneliness interventions for people with psychosis have been promising (Lim et al., 2019a; Lim, et al., 2019b) but there is generally a dearth of robust trial evidence in this area (Ma R et al., 2019). Hence, it is unclear at this stage which types of interventions might be effective.
Conflicts of interest
Laurie Hare-Duke reports no conflict of interest in regards to the reviewed paper.
Chau, A. K., Zhu, C., & So, S. H. W. (2019). Loneliness and the psychosis continuum: a meta-analysis on positive psychotic experiences and a meta-analysis on negative psychotic experiences [abstract]. International Review of Psychiatry, 1-20.
Bentall RP, Wickham S, Shevlin M, & Varese F (2012). Do specific early-life adversities lead to specific symptoms of psychosis? A study from the 2007 the Adult Psychiatric Morbidity Survey [pdf]. Schizophrenia bulletin, 38(4), 734-740.
Borsboom, D. (2017), A network theory of mental disorders [pdf]. World Psychiatry, 16: 5-13.
Hare-Duke L (2017). The importance of social ties in mental health [pdf]. Mental health and social inclusion, 21(5), 264-270.
Lim MH, Gleeson JF, Alvarez-Jimenez M, & Penn DL (2018). Loneliness in psychosis: a systematic review [abstract]. Social psychiatry and psychiatric epidemiology, 53(3), 221-238.
Lim MH, Gleeson JF, Rodebaugh TL, Eres R, Long KM, Casey K, & Penn DL (2019a). A pilot digital intervention targeting loneliness in young people with psychosis [abstract]. Social psychiatry and psychiatric epidemiology, 1-13.
Lim MH, Penn DL, Thomas N, & Gleeson JFM (2019b). Is loneliness a feasible treatment target in psychosis? [abstract]. Social psychiatry and psychiatric epidemiology, 1-6.
Ma R, Mann F, Wang J, Lloyd-Evans B, Terhune J, Al-Shihabi A, & Johnson S. (2019). The effectiveness of interventions for reducing subjective and objective social isolation among people with mental health problems: a systematic review [abstract]. Social psychiatry and psychiatric epidemiology, 1-38.
Michalska da Rocha, B., Rhodes, S., Vasilopoulou, E., & Hutton, P. (2018). Loneliness in psychosis: A meta-analytical review [pdf]. Schizophrenia Bulletin, 44(1), 114–125.