Response

A recent article reported the findings from two systematic reviews examining the association between personality disorders and antisocial behaviour.^{Reference Chow, Yu, Geddes and Fazel1} In the first review, 21 studies reported the association between personality disorders and antisocial behaviour. The overall association for any personality disorder was strong, with an odds ratio of 4.5, indicating over four times the odds of antisocial behaviour among individuals with any personality disorder compared to those without. Looking at specific personality disorders, the effect was strongest for Antisocial Personality Disorder (ASPD), with an odds ratio of 7.6, but the overall effect was still significant – though much smaller – excluding ASPD. In the second review, 39 studies reported the association between personality disorders and criminal recidivism. The effect was significant but not as large as in the first review, with any personality disorder and recidivism having an odds ratio of 2.3, indicating justice-involved individuals with any personality disorder had more than two times the odds of recidivism compared to justice-involved individuals without personality disorder. The odds ratio specifically for ASPD was 2.8, by far the most common personality disorder diagnosis in criminal justice settings, even though it is less common than other personality disorders in the general community.^{Reference Fazel and Danesh2,Reference Winsper, Bilgin, Thompson, Marwaha, Chanen and Singh3}

As the authors acknowledge in their discussion of these two reviews, correlation is not causation. Though longitudinal studies such as recidivism studies demonstrate temporal precedence, the second causal condition after covariation, neither cross-sectional nor longitudinal studies can definitively rule out the possibility of third factor explanations. For example, both antisocial behaviour and personality disorder may be ‘caused’ by adverse childhood experiences, and this creates a correlation between the two constructs. In particular, a parsimonious explanation for at least part of the association between any personality disorder and antisocial behaviour is prior antisocial behaviour. Prior antisocial behaviour is directly relevant to some of the diagnostic criteria of ASPD and Borderline Personality Disorder (BPD) in the DSM,⁴ indirectly relevant to other personality disorder criteria (e.g. interpersonally exploitative and lacking empathy for Narcissistic personality disorder), and likely considered less directly in ICD-11 assessment of personality disorder,⁵ which is salient for the first review. Indeed, ASPD is unlike other personality disorders in the DSM in requiring a development precursor in the form of a Conduct Disorder diagnosis, meaning someone must have engaged in a pattern of antisocial behaviour in order to qualify for ASPD. For the second review, it is an empirically supported truism that past (antisocial) behaviour is a strong predictor of future antisocial behaviour.

An important question then is how much of the association between personality disorder and antisocial behaviour is something distinctive about the personality disorder versus content overlap with antisocial behaviour, where a diagnosis of ASPD requires prior antisocial behaviour, and antisocial behaviour can also be relevant to some BPD criteria (less so for other personality disorders). Relevant here is research on psychopathy, an antisocial personality syndrome characterised by interpersonal/affective features such as grandiosity, callousness, and shallow emotions and by antisocial behaviour, including juvenile delinquency and criminal versatility. A two-factor structure of psychopathy involving interpersonal/affective features (Factor 1) and antisocial behaviour (Factor 2) features are repeatedly found in psychometric analyses in community and criminal justice samples across cultures.^{Reference Hare, Harpur, Hakstian, Forth, Hart and Newman6} Psychopathy is not ASPD, though there is a strong but asymmetric overlap, where most individuals high in psychopathy will meet criteria for ASPD, yet many individuals with ASPD would not meet thresholds for psychopathy. But interestingly, Factor 2 is a stronger predictor of recidivism than Factor 1.^{Reference Walters7} The psychopathy literature suggests the relationship between ASPD and recidivism would be much smaller if the relevance of previous antisocial behaviour was statistically or methodologically controlled. For example, would ASPD still be correlated with recidivism if the analysis controlled for previous criminal history?

Another interesting finding in both reviews is that most studies of personality disorder and antisocial behaviour or recidivism have been conducted in high-income countries, where personality disorder is more common according to a review of community samples published in this journal.^{Reference Winsper, Bilgin, Thompson, Marwaha, Chanen and Singh3} This greater prevalence could reflect differential detection (the cost of assessment and diagnosis), an effect of sociocultural factors on expression of personality disorder or differences in the biological predispositions towards personality disorder. This may also be true for antisocial behaviour, where all but two studies in Chow et al relied on national registries to identify recidivism, and the availability and quality of these registries vary greatly.^{Reference Chow, Yu, Geddes and Fazel1} The association between personality disorder and recidivism may be stronger in higher-income countries with higher quality registries, unlike other countries where new criminal offences are not reported, prosecuted or registered. I agree that more research is needed in not only low and middle income countries, but more broadly, in countries that are not WEIRD (Westernised, Educated, Industrialised, Rich and Democratic).^{Reference Henrich, Heine and Norenzayan8} In addition to comparing countries, it would be fruitful to consider sociocultural factors more broadly. For example, men who are high in psychopathy are more likely to be diagnosed with ASPD than BPD, whereas women who are high in psychopathy may be more likely to be diagnosed with BPD than ASPD, suggesting similar phenotypes might nonetheless be viewed as having different personality disorders due to gendered stereotypes, roles or practices.^{Reference Sprague, Javdani, Sadeh, Newman and Verona9}