Key to the scientific understanding of psychotic disorders, and the design of effective, efficient and equitable services, is a clear picture of how often they occur, and where. This is important for many reasons: psychosis is considered a “rare” disorder, and this idea of rarity has shaped ideas about its causation, natural history, economic burden, and the impact on people’s lives. The idea of the rarity of psychosis shapes how much public resource is devoted to services for its management. Therefore, understanding variation in the occurrence of psychotic disorders is vital to arguing for the right level of service provision. On the other hand there are various problems introduced by incorrect information on occurrence of psychotic disorders: inconsistently staffed services (including for early intervention), and inefficient studies. Good epidemiological studies on the occurrence of psychosis are therefore necessary. It is also important to understand changes in the occurrence of psychosis over time, and between places. The paper that is the subject of this blog addresses this question, that of spatio-temporal variation in the incidence of psychosis (Jongsma et al., 2018).
An important basic question is what we mean by variation. Essentially this involves calculating averages for different places and different times, and assessing whether differences exist. But what level of variation is important variation? This question has proven a significant challenge for researchers. There were disagreements about the interpretation of the World Health Organization’s International Pilot Study of Schizophrenia incidence studies of schizophrenia, whose results implied to some that the incidence of schizophrenia was uniform across countries, and to others demonstrated significant variation between regions (Sartorius et al., 1974). One answer is that important variation is that variation which represents differences in causal factors; variation between those who smoke cannabis and those who do not smoke cannabis would have relevant implications for psychosis incidence of limiting accessibility to cannabis in future public health strategies. Another could be that important variation indicates the need for variation in the way in which services are planned and organised; for example, excess of psychotic disorders in urban environments, compared to rural environments, might indicate specific strategies across regions.
In this large study of the occurrence of new cases of schizophrenia in a defined period, located in 6 countries with 17 settings, occurrence was examined in relation to a range of factors, including age, gender, and population density of where they lived. The different centres were staffed by a research team that was trained in the study design, and the principles of clinical assessment for psychosis. Patients were included if they met criteria for a first psychotic episode, if they had not previously made contact with health services for psychosis, no evidence of psychosis due to an identified physiological process in the brain (such as HIV), no evidence of psychosis due to drug intoxication, and were between the ages of 18-64. Patients meeting these criteria underwent diagnostic ratings to get more specific diagnoses and a range of neuropsychiatric assessments.
Crude rates of psychosis were calculated by taking the number of new cases occurring within each catchment area (and in each group of interest within the study: men vs women, young vs old, etc), and dividing by the number of people at risk for the outcome. This denominator was based on best estimates for the total number of people living in each catchment area who would have been included in the cases had they developed psychosis (“population at risk”). These rates across catchment areas were adjusted to account for differences in age and gender distributions between areas, by a process called “direct standardization”. Statistical modelling was then used to assess relationships between incidence and a range of risk factors.
- The investigators found that 2,774 people presented to services with first episode psychosis, corresponding to around 21 cases in 100,000 people, per year
- Men comprised 58%
- People from ethnic minorities comprised 40%
- Age at first contact was earlier in men than women, but did not vary by ethnic group
- Age patterns differed slightly between men and women
- Incidence was commonest in both men and women in the youngest age group,
- but in men became rapidly less common with increasing age,
- whereas in women this drop in occurrence was less pronounced – by the age of 30-34,
- incidence differences between men and women were small, except for another peak in women occurring in the 45-49 group, which has been identified previously.
- Differences in incidence between regions were pronounced, with the incidence of schizophrenia in south-east London identified as 10 times higher than that found in Santiago, Spain
- Area-level factors were consistently related to incidence: higher rates of psychosis across catchment areas were associated with:
- lower rates of people owning their own homes,
- higher rates of unemployed people in the area,
- and higher rates of single people.
The authors summarise their results by emphasising geographic variation in incidence of psychotic disorders, higher rates in ethnic minority groups, men, and age distributions of incidence which were in line with previous research. Rates were similar to those identified in a previous WHO study of psychotic disorders incidence.
What are the implications of this research for practitioners, policy-makers, researchers, service-users and the general public?
- Consistent geographic variation in the incidence of psychotic disorders implies a role for environmental factors in the causation of psychosis.
- This variation also suggests that health services need to pay attention to which areas have particularly high rates of psychosis for the planning and staffing of services.
- On the other hand, this study does not identify causes of psychosis, and does not aim to evaluate the natural history of psychosis in detail.
- Considering the spread of incidence rates across settings in this study, it is possible that very low rates, for example in Santiago, could be a reflection not of differences in causal factors for psychosis (which would be relevant for prevention or treatment), but of differences in the structure of mental health services, and differences in the way in which these services link together with other health services.
- This study forms part of an important update of the scientific information on who gets diagnosed with psychosis, when, and where.
- However, this study should be put in context of the structure of health services in each particular region, and local understanding of which individuals seek help for psychosis, and how.
Jongsma HE, Gayer-Anderson C, Lasalvia A, et al. (2018) Treated incidence of psychotic disorders in the multinational EU-GEI study [published online December 6, 2017]. JAMA Psychiatry. doi:10.1001 /jamapsychiatry.2017.3554
Sartorius, N., Shapiro, R. & Jablensky, A. (1974). The international pilot study of schizophrenia (PDF). Schizophrenia Bulletin 1, 21.