The effect of 0ethnic density on psychosis incidence

Lead Research Organisation: King's College London
Department Name: Health and Social Care Research


Members of some ethnic groups have a higher than usual risk of developing a severe mental illness such as schizophrenia or bipolar disorder. Some studies have linked this to how isolated people from some ethnic groups are, so that, for example, black people living in mostly white areas may be more at risk. This has been termed the ethnic density effect. Because these are relatively uncommon diseases it has been difficult to look at any more than one, very crudely defined, ethnic group at a time. Using a very large collection psychiatric records I will be able to compare the experiences of a wide range of ethnic groups and look in more detail at what might be causing this ethnic density effect. It has also been suggested that this effect may be much more relevant to some types of severe mental illness, such as schizophrenia, than others. By accessing such a large collection of health records I will be able to address this question in more detail than has been possible before.

The study will look at how living in a particular type of area can effect one's mental health. This is particularly difficult to research as it is hard to identify the relevant neighbourhood when looking at patient records. Typically studies have relied on the address where someone was living when they were first diagnosed. However, people with a mental illness often change address as they become ill, making it difficult to establish which area is relevant. A number of researchers looking at this problem have concluded that, for research to progress, better methods are needed to assess neighbourhoods effects. The psychiatric records that I will use in this study include information about the areas where patients have lived at different times in their life. This, along with further detailed information about each patient's life history, will allow me for the first time to look much more carefully at how living in a particular neighbourhood may affect someone's mental health.

To find out more about this from the patient's perspective I will also interview people who have suffered a severe mental illness, chosen from ethnic groups and neighbourhoods that most clearly show this ethnic density effect. The results of these interviews will then be used to guide the analysis of patient records that will make up the main part of the study. It has also been argued that, as this is an international phenomena, it is important that studies compare results in more than one country. Therefore I will also do a similar study using health records for patients in the Utrecht region in Holland to see how a different context changes the study results.

The study is intended to find out more about what is causing this increased risk in some ethnic groups. The results of this research will therefore have the potential to help prevent this occurring in the future. By finding out more about what is happening in these groups the study will add to what is generally known about the causes of these diseases and ultimately help inform future treatments and preventative measures.

Technical Summary

to investigate the effect of ethnic density on the risk of psychosis among black and minority ethnic (BME) groups
to assess:
1) whether greater ethnic density, at neighbourhood area level, is inversely related to psychosis incidence for members of BME groups, after prior address history is accounted for
2) whether there are significant differences in this effect between ethnic groups,
3) whether this effect is different for sub-groups of psychotic illness (affective / non-affective)
4) whether this differs at different neighbourhood levels
5) how this might differ in different urban areas, both nationally and internationally
6) possible underlying mechanisms behind these effects reflecting, for example, aspects of social cohesion, social support and social alienation.
I will use a mixed methods design:
Study 1 - Initial analysis using detailed psychiatric linked to stratified area census records (covering 180,000 mental health service users in SE London, including detailed free text allowing address history to be tracked prior to illness onset) - meeting objectives 1 - 4.
Study 2 - Using primary care records (covering 900,000 patients in 1 SE London and 3 East London boroughs) to extend the analysis (validating against the above linked psychiatric records) - obj. 1 - 5.
Study 3 - Qualitative study of life history accounts of neighbourhood factors (comparing accounts in high and low ethnic density areas to generate hypotheses as to mechanisms behind ethnic density effect) - obj. 6
Study 4 - Further analysis of linked records (to investigate possible mechanisms behind the ethnic density effect informed by study 1-3) - obj. 6
Study 5 - International comparison study (apply analyses to similar data collected in Netherlands)- obj. 5.
This has the potential to illuminate mechanisms behind ethnic differences in psychosis that would inform treatment and prevention.

Planned Impact

The results of the proposed study will add to what is known about the causes of psychotic illness and therefore be relevant to those concerned with prevention and treatment measures and mental health in general, including: clinicians, counsellors and psychologists, mental health service users and carers and relevant voluntary sector organisations. Ultimately the question "why do some people develop a severe mental illness and not others" is of interest to everyone, not just mental health professionals. The study also has a wider relevance in that it has the potential to tell us something about the nature of social isolation as it is experienced by particular ethnic groups and how this could be pathogenic. The results could therefore inform much wider debates around ethnicity, assimilation and integration that are of interest to policy makers, the social care and voluntary sector and the general public. This is still a relatively new field of research and, so far, the ethnic density effect has been studied in very few ethic groups. Mine will be the first to look at the experience of clinical psychosis among the Indian, Pakistani and Bangladeshi communities in this way and therefore will also be of direct, and immediate, interest to these communities. Through an extensive programme of dissemination activity the study could therefore have an impact within the duration of the fellowship.

Although the study is not directly concerned with an intervention the results of the study could inform future interventions and therefore have an impact in the long term. One area, for example, is that of treatment measures designed to address the social isolation of service users. Although research in this area is still at an early stage, this study intends to reveal factors behind the elevated rates of psychosis relating to the social isolation of particular ethnic groups. What we learn about these factors could then inform relevant interventions . For example, a recent initiative to develop a social capital intervention addressing the social isolation of mental health service users in London (see: would directly benefit from the results of this study. This is designed to find out how health and social care workers can assist individuals recovering from psychosis to access social capital. On the basis that initiatives such as this could potentially be put in place within 2 to 3 years the results of my study could inform patient care within the next decade.

The study could also identify communities and areas where service users from particular ethnic groups would benefit from mental health services specifically tailored to their ethnic group or, alternatively, measures to help them access ethnic specific services in other areas.

There are other potential areas where the study could influence patient care. For example, it may be that language fluency plays a role in psychosis risk for some of these groups and therefore attention could be paid to language services for mental health service users. It may be that people from particular religious faiths are more at risk than others, in which case resources could be more effectively targeted at these faith communities. Also public health information designed to improve people's awareness of mental illness and sources of support could be targeted at particular groups identified in the study as most isolated and at risk.

The study could also inform wider community services - where particular groups are highlighted as being vulnerable this study could inform housing policies, travel links, town planning measures and the provision of local ethnic specific resources and communications. By successfully disseminating the results via local health trusts, local authority policy makers and other agencies this could potentially have a long term effect.


10 25 50

publication icon
Gilkes A (2016) Does COPD risk vary by ethnicity? A retrospective cross-sectional study. in International journal of chronic obstructive pulmonary disease

Description A Cross-National Study of the Role of Neighbourhood Context in the Mental Health of Refugees
Amount £410,838 (GBP)
Funding ID MR/S025510/1 
Organisation Medical Research Council (MRC) 
Sector Public
Country United Kingdom
Start 06/2019 
End 06/2022
Title Application of AZtool to Danish Parishes to Optimise Area Units for Research 
Description Neighbourhood units based on Danish parishes vary considerably in size hindering model convergence for spatial analysis. For small parishes I therefore combined adjacent units to arrive at an optimum size, using AZtool, the algorithm devised to create UK census area units (Cockings et al., 2011; Martin, 2003). I set the algorithm to aim for an optimum parish size of 3000 inhabitants with no units <200, collapsing 2114 parishes into 1135 units. I also split very large parishes (over 6500) into two, randomly assigning parish members into either unit, giving a final total of 1167 parish units (median size 3564). These were then used to determine the neighbourhood social context based on all residents in the parish in any one year. 
Type Of Material Improvements to research infrastructure 
Year Produced 2017 
Provided To Others? Yes  
Impact This method is widely applicable to other studies using similar area level data and I have disseminated details of the method to academic colleagues in Denmark and Sweden. 
Title Pooled GP data for lambeth and East London 
Description For the paper on antidepressant use / depression diagnosis I arranged for relevant patient level GP data to be extracted at the same time point (October 2013) for all patients in Lambeth, SE London, and Newham, Hackney and Tower Hamlets (East London). 
Type Of Material Database/Collection of data 
Year Produced 2013 
Provided To Others? Yes  
Impact Published paper 
Title Prospective cohort based on historical Danish population register data 
Description Prospective cohort dataset established to investigate the role of neighbourhood social characteristics in the aetiology of psychotic illness. Comprises of 2,224,175 individuals born after 1st January 1965 and living in Denmark on their 15th birthday - followed from 1980 to 2013 resulting in 37,919,670 person years of data. 
Type Of Material Database/Collection of data 
Provided To Others? No  
Impact So far only myself and the collaborating team have access to this database as it is held on a secure server with highly restricted access. 
Title Re-defined Danish area units 
Description Denmark is divided up into parishes that have intrinsic meaning as area units and, in many cases, are consistent over time. However, these vary considerably in size and are often very small and therefore unsuitable for a population level analysis of a rare outcome such as psychotic illness. In order to make these more consistent in size and remove small areas I combined approximately 75% of parishes using AZtool, the algorithm originally devised to create UK census area units (Cockings, Harfoot, Martin, & Hornby, 2011; Martin, 2003). This combines smaller neighbouring units based on pre-set criteria, in this case we set an optimum size of 3000 people. This resulted in new area units with a median size of 3564, IRQ 3654, and no units with a population less than 200. 
Type Of Material Database/Collection of data 
Provided To Others? No  
Impact This has allowed me to analyse neighbourhood effects on a rare outcome (psychotic illness) using area units based on administrative data. 
Description Aarhus University - population register 
Organisation Aarhus University
Country Denmark 
Sector Academic/University 
PI Contribution I contributed a detailed research design protocol for an analysis of Danish population data collected over a 30 year period to assess the role of neighbourhood social composition on later risk of psychotic illness. I am currently carrying out the data analysis and writing a series of journal articles based on the results. In order to access Danish population data all foreign researchers are required to spend 4 weeks in Denmark to gain an understanding of the data and data security processes. I have so far spent 3 weeks in Denmark over 4 separate visits. I have also provided the geographical area units needed to conduct the analysis by re-configuring existing administrative boundary data
Collaborator Contribution Aarhus University have provided access to a statistics Denmark secure server so that i can analyse the data - along with a secure VPN connection to analyse data remotely. On each of my visits they have provided office space, a computer and IT / database support. They have organised the approval process for the study to access statistics Denmark data, requested the relevant data (from the specification I provided) and created the linked datasets that i am using for the analysis. I have discussed the dataset and analysis in detail with Esben Agerbo and Carsten Pedersen with whom i have had regular meetings and both have shared their many years of expertise with these data resources.
Impact I have delivered a presentation at the Centre for Integrated Register-based Research on the proposed study and my work to date. Also I have created newly defined geographical area units covering the whole of Denmark in order to conduct a spatial analysis.
Start Year 2015
Description East London primary care data 
Organisation Queen Mary University of London
Country United Kingdom 
Sector Academic/University 
PI Contribution I have combined GP records for Lambeth (SE London) with data collected by the collaborating team for East London. I standardised the format of the respective datasets and carried out a pooled analysis to investigate the effect of neighbourhood ethnic density on depression diagnosis and antidepressant use. I conducted the data analysis and wrote a recently published paper on this study. I am now advising one of the collaborating team on an analysis they are conducting using similar methods but looking at a different research question,
Collaborator Contribution They provided access to the data via a secure link along with advice about the dataset and the clinical context in which it was created - through a series of meetings over a 12 month period.
Impact Recent journal article: Schofield, P., Das-Munshi, J., Mathur, R., Congdon, P., & Hull, S. (2016). Does depression diagnosis and antidepressant prescribing vary by location? Analysis of ethnic density associations using a large primary-care dataset. Psychological Medicine, 46(06), 1-9. Conference presentation: "The ethnic density effect and antidepressant use" Health Inequalities Research Network (HERON) Conference - London 2014
Start Year 2014
Description Lambeth & Southwark MIND 
Organisation MIND (Mental Health Charity)
Country United Kingdom 
Sector Charity/Non Profit 
PI Contribution Organised project advisory group providing a forum for discussion about ethnicity and mental illness
Collaborator Contribution MIND provided premises, recruited participants and helped facilitate the discussion
Impact Currently planning a series of focus groups investigating the views of mental health service users about ethnic differences in rates of severe mental illness
Start Year 2014
Description Neighbourhood and mental health for migrant groups 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Public/other audiences
Results and Impact This was a presentation and discussion I was invited to by a local Somali mental health charity (Xiishod) as part of a seminar they organised to, in their words: "bring awareness and light to mental health in order for the community to gain a wider understanding of the issues surrounding mental health"
Year(s) Of Engagement Activity 2017
Description Project Advisory Group 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? Yes
Geographic Reach Local
Primary Audience Participants in your research and patient groups
Results and Impact This meeting was very well attended (19 BME service users participated) resulting in a lengthy and very fruitful discussion about why ethnic differences occur in the first place.

It was felt afterwards, by those involved, that further discussions would justify a research study in itself. I am therefore applying for ethical approval to conduct a series of focus groups to discuss ethnic differences in mental health
Year(s) Of Engagement Activity 2014
Description Talk at Aarhus University 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Presented a talk outlining my proposed research study using Danish population register data
Year(s) Of Engagement Activity 2015