Go to ...

• HOME
• ABOUT
• SERVICES
• PROJECTS
• HELP & SUPPORT
• NEWS & EVENTS
• CONTACT

Digimap Collections Menu

• Login
• Description
• Access & Subscription
• Help & Support
• EDINA News & Events

Digimap Case Study:
Neighbourhood variation in the incidence of psychotic disorders in SE London

Authors

James B.Kirkbride, Paul Fearon, Craig Morgan, Paola Dazzan, Kevin Morgan, Robin M. Murray, Peter B. Jones

Title

Neighbourhood variation in the incidence of psychotic disorders in SE London

Date

30 April 2007

Application Area

Spatial Epidemiology

Application to other subject areas

Medical Geography

Project type

PhD

Summary

Using Bayesian disease mapping approaches to examine the variation in incidence rates of schizophrenia using small area maps to underpin the GIS component.

Abstract: Background Urbanicity is a risk factor for schizophrenia, but it is unclear whether this risk is homogenous across urban areas. Aims To determine whether the incidence of psychotic disorders varied within an urban area, beyond variation attributable to individual-level characteristics. Methods: All incident cases of ICD-10 psychoses from a large, 2-year, epidemiological study of first-episode psychoses in Southeast London were identified. Incidence rates for 33 wards were standardised for age, sex and ethnicity. Bayesian models produced accurate relative risk estimates that were then mapped. Results: 295 cases were identified during 565,000 person-years of followup. We observed significant heterogeneity in relative risks for broad and non-affective psychoses (schizophrenia), but not for affective psychoses. Highest risks were observed in contiguous wards. Conclusions Neighbourhood variation in the incidence of non-affective psychoses could not be explained by individual-level risk, implicating neighbourhood-level socioenvironmental factors in their aetiology. The findings are consistent with classical sociological models of mental disorders.

Datasets Used

• Name: English wards
• Source: Digimap, UKBORDERS

Aims and Objectives

Investigation of the geographical variation of psychotic disorders using contemporary multilevel techniques has been predominantly restricted to common mental disorders. While some studies have investigated the relative roles of individual and neighbourhood-level characteristics in psychoses, to our knowledge, no previous study has explicitly modelled the geographical clustering of psychotic disorders in a multilevel framework using Bayesian techniques. The aim is to test rates of schizophrenia with increasing exposure to urbanicity and to find out if the role of neighbourhood (socio-) environments is etiologically relevant. The relationship between place and the affective psychoses is not well understood, aim to shed light on this. Using data from a large epidemiological study of 33 neighbourhoods in southeast London, we tested whether the incidence of psychotic disorders was heterogeneous between neighbourhoods having accounted for individual-level risk factors which potentially explained this variation. Demonstration of heterogeneity would implicate environmental risk factors at the neighbourhood-level in the aetiology of schizophrenia, adding weight to calls to modify current psychiatric paradigms of disease causation.

Methodology

We analysed data on first-onset cases of psychotic disorders collected from the Southeast London study area of the Aetiology & Ethnicity in Schizophrenia and Other Psychoses [ÆSOP] study.

The use of SIR (standardised incidence ratios) for disease mapping is problematic, particularly for relatively rare disorders such as psychoses, because point estimates are often based on small counts of cases in each ward. Estimates are highly influenced by sampling variability (over-dispersion), which cannot be adequately handled by the Poisson distribution which SIR are normally assumed to follow. In addition, spatial patterning may also exist in the data, because counts of cases in neighbouring wards may be more similar than counts of cases in wards further apart. This dependence is known as spatial autocorrelation

The Bayesian approach ‘‘smoothes’’ SIR (henceforth, Bayesian relative risks [RR]) in each ward by fitting Bayesian models with a random effects term which weights the RR in a given ward, according to RR in neighbouring wards. The nature of the random effects term is dependent on prior assumptions concerning the structure of this variability. We tested four
Bayesian models with different random effect terms for each outcome; unstructured - assumes no spatial autocorrelation is present in the RR (Model 1), structured - assumes RR in neighbouring wards are more similar than in wards further apart (spatial autocorrelation, Model 2), convolution -combines models 1 and 2(Model 3); and mixture - which suggests transition of RR between wards is not necessarily smooth, but discontinuities across the study area may be present. For example, where a rural and urban ward are adjacent, there may be valid reasons for large differences in relative risk (Model 4). Bayesian models were fitted using WinBUGS (version 1.4) and its spatial extension GeoBUGS (version 1.2).

Results

Incidence of the broad psychoses category in Southeast London was not homogeneous. It followed a nonrandom geographical distribution, independent of differences in the age, sex and ethnic composition of neighbourhoods, having accounted for over-dispersion and spatial autocorrelation using Bayesian spatial models. Similar findings pertained for the nonaffective psychoses separately, but we found no evidence of heterogeneity in rates for the affective psychoses. This is in accordance with the sociological models proposed by Faris and Dunham.

The distribution of elevated rates of non-affective psychoses followed a distinct horseshoe pattern encompassing some of the most deprived neighbourhoods in Southwark and Lambeth, including Brixton, Camberwell and Peckham. This heterogeneity was present after adjustment for age, sex and ethnicity, which is consistent with the hypothesis that neighbourhood- level environmental risk factors may be relevant to the aetiology of non-affective psychoses. Such risk factors may include socioeconomic deprivation, ethnic density or social cohesion.

This study has demonstrated considerable heterogeneity in incidence rates for non-affective psychoses but not for affective psychoses in a highly urbanized environment, having controlled for age, sex, and ethnicity. Highest rates were observed in contiguous wards. This supports the hypothesis that potentially important environmental risk factors, that may play a part in the aetiology of non-affective psychoses, vary at the local level within environments, rather than being ubiquitous. This heterogeneity should also be taken into consideration in terms of health services planning.

 

Additional Information

Full Paper published in: Soc Psychiatry Psychiatr Epidemiol (2007) DOI 10.1007/s00127-007-0193-0

	Fig. 1: Mapping posterior parameter estimates for the broad psychoses phenotype in Southeast London (Model 4). (a) Posterior relative risk, (b) Posterior probability of RR greater than 1.0 (where theta represents the relative risk [RR] in the ith ward)
	Fig. 2: Mapping posterior parameter estimates for non-affective psychoses in Southeast London (Model 4). (a) Posterior relative risk, (b) Posterior probability of RR greater than 1.0 (where theta represents the relative risk [RR] in the ith ward)
	Fig. 3: Mapping posterior parameter estimates for affective psychoses in Southeast London (Model 4). (a) Posterior relative risk, (b) Posterior probability of RR greater than 1.0 (where theta represents the relative risk [RR] in the ith ward)
	Fig 4: Ward names and Southeast London study centre location in relation to Greater London

References

1. Allardyce J, Gilmour H, Atkinson J, Rapson T, Bishop J, McCreadie RG (2005) Social fragmentation, deprivation and urbanicity: relation to first-admission rates for psychoses. Br J
Psychiatry 187:401–406
2. Besag J, York J, Mollie A (1991) Bayesian image restoration, with application in spatial statistics with discussion. Ann Inst Stat Math 43:1–59
3. Boydell J, van Os J, McKenzie K, et al. (2001) Incidence of schizophrenia in ethnic minorities in London: ecological study into interactions with environment. Br Med J 323:1336–1338
4. Cannon M, Clarke MC (2005) Risk for schizophrenia – broadening the concepts, pushing back the boundaries. Schizophr Res 79:5–13
5. Crow TJ (2000) Schizophrenia as the price that homo sapiens pays for language: a resolution of the central paradox in the origin of the species. Brain Res Rev 31:118–129
6. Faris REL, Dunham HW (1939) Mental disorders in urban areas. University of Chicago Press, Chicago 7. Harrison G, Brewin J, Cantwell R, et al. (1996) The increased risk of psychosis in African-Caribbean migrants to the UK: a replication. Schizophr Res 18:102
8. Harrison G, Gunnell D, Glazebrook C, Page K, Kwiecinski R (2001) Association between schizophrenia and social inequality at birth: case-control study. Br J Psychiatry 179:346–350
9. Haukka J, Suvisaari J, Varilo T, Lonnqvist J (2001) Regional variation in the incidence of schizophrenia in Finland: a study of birth cohorts born from 1950 to 1969. Psychol Med 31:1045– 1053
10. Jablensky A, Sartorius N, Ernberg G, et al. (1992) Schizophrenia: manifestations, incidence and course in different cultures. A World Health Organization ten-country study. Psychol Med Monogr Suppl 20:1–97
11. Kirkbride JB, Fearon P, Morgan C, et al. (2006) Heterogeneity in incidence rates of schizophrenia and other psychotic syndromes: findings From the 3-center ÆSOP study. Arc Gen Psychiatry 63:250–258
12. Kirkwood BR, Sterne JAC (2003) Essential medical statistics. Blackwell, Malden, MA
13. Lawson AB, Browne W, Vidal Rodeiro CL (2003) Disease mapping with WinBUGS and MLwiN. Wiley & Sons, Chichester
14. Lawson AB, Clark A (2002) Spatial mixture relative risk models applied to disease mapping. Stat Med 21:359–370
15. Lewis G, David A, Andreasson S, Allebeck P (1992) Schizophrenia and city life. Lancet 340:137–140
16. MRC (2004) winBUGS. MRC Biostatistics Unit, Cambridge
17. Martuzzi M, Hills M (1995) Estimating the degree of heterogeneity between event rates using likelihood. Am J Epidemiol 141:369–374
18. McGovern D, Cope RV (1987) First psychiatric admission rates of first and second generation Afro Caribbeans. Soc Psychiatr 22:139–149
19. Mollie A (1996) Bayesian mapping of disease. In: Gilks WR, Richardson S, Spiegelhalter DJ (eds) Markov chain Monte Carlo in practice. Chapman and Hall, London, pp 359–379
20. Mortensen PB, Pedersen CB, Melbye M, Mors O, Ewald H (2003) Individual and familial risk factors for bipolar affective disorders in Denmark. Arch Gen Psychiatry 60:1209–1215
21. Mortensen PB, Pedersen CB, Westergaard T, et al. (1999) Effects of family history and place and season of birth on the risk of schizophrenia. New Engl J Med 340:603–608
22. van Os J, Driessen G, Gunther N, Delespaul P (2000) Neighbourhood variation in incidence of schizophrenia. Evidence for person–environment interaction. Br J Psychiatry 176:243–248
23. Pedersen CB, Mortensen PB (2001) Evidence of a dose–response relationship between urbanicity during upbringing and schizophrenia risk. Arch Gen Psychiatry 58:1039–1046
24. Pedersen CB, Mortensen PB (2006) Urbanicity during upbringing and bipolar affective disorders in Denmark. Bipolar Disord 8:242–247
25. Pereira R (2002) The census coverage survey - the key element of a one number Census. Office for National Statistics, Titchfield
26. Richardson S, Thomson A, Best N, Elliott P (2004) Interpreting posterior relative risk estimates in disease-mapping studies. Environ Health Perspect 112:1016–1025
27. Scully PJ, Owens JM, Kinsella A, Waddington JL (2004) Schizophrenia, schizoaffective and bipolar disorder within an epidemiologically complete, homogeneous population in rural
Ireland: small area variation in rate. Schizophr Res 67:143–155
28. Silver E, Mulvey EP, Swanson JW (2002) Neighborhood structural characteristics and mental disorder: Faris and Dunham revisited. Soc Sci Med 55:1457–1470
29. Spiegelhalter DJ, Best N, Carlin BP, van der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc Ser B Stat Methodol 64:583–640
30. StataCorp (2005) Stata. StataCorp, College Station, Texas
31. Sundquist K, Frank G, Sundquist J (2004) Urbanisation and incidence of psychosis and depression: follow-up study of 4.4 million women and men in Sweden. Br J Psychiatry 184:293– 298
32. Torrey EF, Mortensen PB, Pedersen CB, Wohlfahrt J, Melbye M (2001) Risk factors and confounders in the geographical clustering of schizophrenia. Schizophr Res 49:295–299
33. Tsuang MT, Tohen M, Zahner GEP (1995) Textbook in psychiatric epidemiology. John Wiley & Sons, New York
34. Wakefield J, Best N, Waller L (2001) Bayesian approaches to disease mapping. In: Elliott P, Wakefield J, Best N, Briggs D (eds) Spatial epidemiology: methods and applications. Oxford
University Press, Oxford
35. Wakefield JC, Kelsall JE, Morris SE (2001) Clustering, cluster detection and spatial variation in risk. In: Elliott P, Wakefield JC, Best N, Briggs D (eds) Spatial epidemiology: methods and applications. Oxford University Press, Oxford
36. Weich S, Holt G, Twigg L, Jones K, Lewis G (2003) Geographic variation in the prevalence of common mental disorders in Britain: a multilevel investigation. Am J Epidemiol 157:730–737

Publishing Institution

Dept of Psychiatry-University of Cambridge, Institute of Psychiatry-Kings College London, Dept of Psychology-University of Westminster.