A public health approach to prevent psychosis

Yung et al. (Reference Yung, Wood, Malla, Nelson, McGorry and Shah2019) are correct to chide us for not considering ARMS services to be part of a public health approach. We accept that ARMS services can be regarded as selective prevention. However, even if ARMS services for psychosis were able to prevent patients from developing a psychotic disorder, the distribution of risk in society would remain largely unaffected by their intervention (McKinlay, Reference McKinlay1993). This is because ARMS clinics do not influence the exposures across the population which increase the risk of psychosis. Therefore, we argue that a better approach would be public health interventions with the primary focus on reducing exposure to the risk factors. This approach was famously employed by John Snow in 1854 when he brought a cholera epidemic to an end by taking the handle off the Broad Street water pump. In medicine greater change has often been produced by universal primary prevention which targets the whole population rather than intensive treatment of the individual; for example, minimizing or eradicating infectious diseases by vaccination (e.g. smallpox) or improved hygiene (e.g. COVID19 in the absence of a vaccine) or remedying maternal nutritional deficits, such as folic acid deficiency to avoid neural tube defects. Another approach sets out to achieve lifestyle change. A recent successful example of the latter has been national campaigns against tobacco smoking which have resulted in large reductions in smoking rates and smoking-related diseases.

Moving on to population-based preventive approaches

A number of authors have raised the possibility of applying the principles of primary prevention to psychiatry in general (Arango et al., Reference Arango, Diaz-Caneja, McGorry, Rappaport, Sommer, Vorstman and Carpenter2018) and to depression (Hoare, Callali, & Berk, Reference Hoare, Callali and Berk2020), but only a few studies have focussed on psychosis. Regarding the latter, both McGrath (Reference McGrath2000) and Warner (Reference Warner2001) advocated improving pre-and perinatal care including nutrition, particularly for mothers who had schizophrenia, in the expectation that this would result in fewer babies with early neurodevelopmental damage and consequently predisposed to schizophrenia. In a similar vein, Brown and McGrath (Reference Brown and McGrath2010) considered the targeting of prenatal infections and nutritional deficiencies. Sommer et al. (Reference Sommer, Bearden, van Dellen, Breetvelt, Duijff, Maijer and Vorstman2016) took a different tack and pointed to the social and cognitive impairments which typically develop years before the ARMS stage; they recommended that prevention should focus on those children at risk for psychosis by virtue of increased familial/genetic risk or minor psychotic symptoms. They suggested that interventions could aim to improve stress resilience, optimize brain maturation, and prevent or alleviate adverse environmental circumstances. Recently, Anglin, Galea, and Bachman (Reference Anglin, Galea and Bachman2020) made a passionate plea for a recalibration of the priorities of US funding institutions towards research into reduced exposure to putative social risk factors for psychosis such as poverty and racism.

We consider that interventions aimed at the whole population would be most fruitful. In medicine, such strategies generally depend on identifying exposures that increase the risk of the disorder in question and then publicly advocating avoidance of that risk factor. What do we know about exposures relevant for psychosis? Studies demonstrating that psychosis varies in incidence might provide some clues as to the exposures of interest. The recent EU-GEI study of 16 sites reported that the incidence of psychosis was more than five times higher in certain of the Northern European cities studied than that in parts of Italy and Spain (Jongsma et al., Reference Jongsma, Gayer-Anderson, Lasalvia, Quattrone, Mulè and Szöke2018).

It is implausible to think that the differences in the EU-GEI study are due to genetic differences between Northern and Southern Europeans. Differential exposure to environmental risk factors is much more likely. However, there are a considerable number of candidate factors (Stilo & Murray, Reference Stilo and Murray2019). Brown and McGrath (Reference Brown and McGrath2010) suggested the use of the population attributable fraction (PAF) as a measure which could help to decide which factors were important enough to be selected for preventative effort. The PAF is the estimated proportion of all cases of a disorder which could be prevented if one could completely remove the risk factor under study. In interpreting PAF estimates, it is always worth recalling that because psychosis is multifactorial, adding up PAFs for different component causes can come to over 100%.

Kirkbride et al. (Reference Kirkbride, Morgan, Fearon, Dazzan, Yang, Lloyd and Jones2010) estimated the PAF for urbanicity in England to be 19.3% and that for ethnic minority populations 21.6%; the latter means that if all the factors associated with an increased risk of psychotic illness in ethnic minority populations could be successfully identified and removed from the whole population, up to 21.6% of all cases of could be prevented. Bebbington et al., estimated the PAF for child sexual abuse in England as 14%. Obviously, the PAF will vary geographically depending on the prevalence of the risk factor in that particular area. For example, the PAF for use of high potency cannabis varied in the EUGEI study from less than 2% in Bologna, Italy to 50% in Amsterdam, Holland.

We will now consider the prospects for primary prevention focussing on what we regard as the leading candidates: urbanicity, migration, and cannabis use.

The example of urbanicity

Although being born or brought up in a city rather than a rural setting is a highly replicated risk factor (Stilo & Murray, Reference Stilo and Murray2019), urbanicity must be a proxy for one or more specific component causes. High population density, greater exposure to stress, drug abuse, crime (Newbury et al., Reference Newbury, Arseneault, Caspi, Moffitt, Odgers and Fisher2018) have all been suggested. More recently, novel hypotheses focussing on air pollution (Newbury et al., Reference Newbury, Arseneault, Beevers, Kitwiroon, Roberts, Pariante and Fisher2019), lack of green space (Engemann et al., Reference Engemann, Pedersen, Bøcker, Larse, Constantinos, Mortenson and Jens-Christain2018), and lack of social capital (Rotenberg, Anderson, & McKenzie, Reference Rotenberg, Anderson and McKenzie2020) have all been investigated.

Is it possible that cities could be designed so that their poor areas are less toxic? Certainly, planning cities to have more green space or lower pollution could benefit not only those at risk of schizophrenia but children from poor families more widely (Centre for urban design & mental health, 2020). However, at present, there is insufficient evidence concerning the exact pathogenic mechanism for us to advocate particular changes in town planning which would diminish the psychotogenicity of cities. Nevertheless, there is an urgent need for careful evaluation of the effects of urban planning on mental health.

The example of migration/ethnic minority status

One factor which contributed significantly to the difference between the incidence of psychosis in Northern and Southern European sites in the EU-GEI study was the substantial population of people from ethnic minorities in Northern cities, mostly migrants and their children. When migrants/ethnic minorities were removed from the analysis, the incidence of psychosis in South London decreased from 61/100 000 per annum to 45/100 000 per annum (Jongsma et al., Reference Jongsma, Gayer-Anderson, Lasalvia, Quattrone, Mulè and Szöke2018). If the increased risk is connected to racism and/or discrimination, which is a plausible hypothesis, then we have yet another reason (if one were needed) to back political change to eradicate them (Morgan & Hutchinson, Reference Morgan and Hutchinson2010). Unfortunately, we do not think that as yet we are sure enough of the mechanism underlying the increased risk in migrants and their children (Boydell et al., Reference Boydell, van Os, McKenzie, Allardyce, Goel, McCreadie and Murray2001; Jongsma et al., Reference Jongsma, Gayer-Anderson, Lasalvia, Quattrone, Mulè and Szöke2018; Sharpley, Hutchinson, Murray, & McKenzie, Reference Sharpley, Hutchinson, Murray and McKenzie2001) to launch an initiative on this basis.

The example of cannabis use

Use of cannabis is associated with an increased risk of psychosis. As with the initial reports 60 years ago of a causal association between cigarette smoking and lung cancer, there was initial scepticism about the idea that cannabis use increases the risk of psychosis. Critics suggested that the association is not causal but it is due to concomitant use of other drugs, pre-existing deviant personality, or self-medication for preceding psychiatric or prodromal symptoms. One by one these assertions have been rebutted. The risk increases with the frequency and length of use, and the potency of the cannabis used (Di Forti et al., Reference Di Forti, Quattrone, Freeman, Tripoli, Gayer-Anderson, Quigley and Murray2019; Marconi, Di Forti, Lewis, Murray, & Vassos, Reference Marconi, Di Forti, Lewis, Murray and Vassos2016).

The long-term adverse effects of tobacco and alcohol track the extent of their use. Lung cancer reached epidemic proportions after cigarette smoking spread, and the frequency of alcoholic liver disease waxes and wanes in proportion to changes in population alcohol consumption. Growing evidence suggests that this is also the case for cannabis and psychosis. Boydell et al. (Reference Boydell, van Os, Caspi, Kennedy, Giouroukou, Fearon and Murray2006) showed that the incidence of schizophrenia doubled in London, UK, between 1965 and 1999, and attributed this in large part to the increased use of cannabis. Hjorthøj, Larsen, Starzer, and Nordentoft (Reference Hjorthøj, Larsen, Starzer and Nordentoft2019) demonstrated that the incidence of cannabis-induced psychosis more than doubled in Denmark between 2006 and 2016. Gonçalves-Pinho, Bragança, and Freitas (Reference Gonçalves-Pinho, Bragança and Freitas2020) who studied hospitalizations in Portugal with a primary diagnosis of psychotic disorder or schizophrenia in the 15 years following the decriminalisation of cannabis in 2001, found that those who received a secondary diagnosis of cannabis use rose from 0.87% in 2000 to 10.60% in 2015. Di Forti et al. (Reference Di Forti, Quattrone, Freeman, Tripoli, Gayer-Anderson, Quigley and Murray2019) found that the incidence rate of psychosis in 11 areas across five countries in Europe was positively correlated highly (r = 0.8) with the prevalence of daily cannabis use in the general population in each site. The population attributable fractions (PAFs) calculated indicated that if high-potency cannabis were no longer available, some 12% of cases of first-episode psychosis could be prevented across the 11 sites, rising to 30% in London and 50% in Amsterdam.

Is the evidence sufficiently persuasive to merit preventative action? Gage, Hickman, and Zammit (Reference Gage, Hickman and Zammit2016) who exhaustively scrutinized the epidemiological literature for possible confounding, bias, misclassification, reverse causation and other explanations for the association between cannabis use and later psychosis, concluded that ‘epidemiologic studies provide strong enough evidence to warrant a public health message that cannabis use can increase the risk of psychotic disorders’. We concur. Given the lack of an animal model for psychosis and of the equivalent of painting tobacco tar on mice to demonstrate its carcinogenicity, it is not sensible to wait for absolute proof that cannabis is a component cause of psychosis.