Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


Social inequality & psychosis risk

Findings from the ELFEP study on psychosis risk across urban environments. Includes translational PsyMaptic presentation

James Kirkbride

on 10 April 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Social inequality & psychosis risk

Social inequality, psychosis risk & the provision of mental health services Why non-affective vs. affective differential? What factors are are associated with psychosis risk in urban areas? Future directions James B. Kirkbride Ph.D Can we translate the
epidemiology of psychosis
into effective
public mental health? Prediction is very difficult, especially if it's about the future. Methods Can we use knowledge
of the epidemiology of
psychotic disorders to
make useful predictions about psychosis incidence? Empirical first episode psychosis data from 2 large epidemiological studies in England " " Niels Bohr Nobel Prize winner for Physics Rich epidemiology of psychotic disorders Caribbean migration to England Eastern European migrants to England Kirkbride et al. 2006, Arch Gen Psych AESOP & East London FEP studies
1029 FEP cases
16-64 years
Sociodemographic & environmental data obtained
Poisson regression to estimate incidence coefficients Risk coefficients applied to population at-risk in new region

Accounts for population structure & environmental characteristics

Expected number of incident FEP cases estimated in new region

Several models tested

95% "prediction intervals" Predicted incidence cases obtained for East Anglia region, over 2.5 years

Denominator: 2009 mid-year Census estimate

Predicted cases aged 16-64 and 16-35 years estimated Observed Predicted Social epidemiology of Psychoses in East Anglia [SEPEA] study (www.sepea.org)

FEP cases incepted to EIS ascertained over same 2.5-year period

6 EIS services in region

Sole referral point for people aged 16-35 years experiencing psychotic symptoms

Clinical diagnosis 6 months after EIS acceptance Results 522 FEP cases observed in East Anglia, 16-35 years Several models tested:
1. Age, sex & ethnicity
2. Model 1 + deprivation
3. Model 1 + pop. density
4. Uniform rate: 51 per 100k p-y Observed vs. Predicted Did the observed N fall within prediction intervals at: EIS level? Local Authority level? Which model performed best across these metric? Model with age X sex, ethnicity & pop. density provides best predictor of FEP, 16-35 years old Further 262.9 FEP cases aged 36-64 yrs predicted by model No external validation possible, but apparent (i.e. internal) validity was satisfactory Summary Potential model for the population-level prediction of FEP Requires precise characterisation of epidemiological risk Requires precise estimates of population at-risk:

We used 2009 mid-year Census estimates

2011 census will update data (when available) Allows local service planning for EIS, based on local need: www.psymaptic.org Limitations Further validation in more regions required Models based on clinically relevant FEP EIS remit wider, including ARMS & false positives Such cases may not have FEP, but require psychiatric triage Incorporate further validation regions Update denominator Validation for
specific demographic groups
specific diagnoses Other regions & disorders "I confess that in 1901, I said to my brother Orville that man would not fly for fifty years . . . Ever since, I have distrusted myself and avoided all predictions. " " Wilbur Wright Co-inventor & creator of first powered human flight 1908 Acknowledgments Peter Jones, University of Cambridge

Robin Murray (East London) & Jeremy Coid (ELFEP)

AESOP & ELFEP studies

EIS in East Anglia

Fredrik Oher (Lund University) James B Kirkbride jbk25@cam.ac.uk @dr_jb_kirkbride www.psymaptic.org Overall regional level? Refine models: i.e. finer-grained geography James B. Kirkbride Ph.D. Is living in an urban environment the incidence of psychotic disorder? associated with If so, what are the factors psychosis risk in urban areas? associated with Geographical variation in incidence of psychotic disorders in Southeast London Incidence data from AESOP study
2 year epidemiology FEP study
Bayesian hierarchical analysis to detect spatial variation Is living in an urban environment the incidence of psychotic disorder? associated with Psychosis Risk, urban birth & upbringing Urban birth & upbringing increase schizophrenia risk
Dose-response gradient with duration & "dose"
Not observed for affective psychoses Mortensen et al. (1999) New Eng J Med.

Adjusted for age-sex interaction, parental age, family history and season of birth

See also Lewis et al (1992)

Pedersen & Mortensen (2006)

- No evidence of dose-response relationship between bipolar disorder & urban birth / upbringing in 5m First contact study, 1996-1998 All service bases followed-up including leakage study DSM-IV diagnosis using SCAN assessment Denominator estimated from 2001 Census Neighbourhood environmental variables estimated using census & other administrative data Analysis: Bayesian hierarchical modelling of incidence rates of non-affective & affective psychoses Models spatial dependency and multilevel structure of the data Variables included in analysis Individual: Age, sex, ethnicity & social class Neighbourhood: Population density People per km2 Socioeconomic deprivation Multiple deprivation: income, education, crime, health, service access, housing Social cohesion Voter turnout at local elections Income inequality Voter turnout at local elections Dispersion of income deprivation in each neighbourhood estimated using Gini coefficient Ethnic density & integration Perfect integration Perfect segregation Estimated for each ethnic group cf all other groups East London First Episode Psychosis study [ELFEP] 484 cases identified in ELFEP study 427 FEP included (NFA excluded) 313 non-affective psychosis cases, 114 affective Epidemiologically representative FEP sample See Kirkbride et al. 2008. BJP & Coid et al. 2008. Arch Gen Psych Bayesian Relative Risk, after adjustment for age, sex, ethnicity & social class Bayesian posterior probability of RR>1 No significant spatial variation in affective psychoses detected Incidence of non-affective psychosis independently associated with neighbourhood: Multiple deprivation
Per standard deviation increase:
RR: 1.28 (95% CI: 1.08-1.51) Two neighbourhood ethnic composition effects observed: Income inequality
Per standard deviation increase:
RR: 1.25 (95% CI: 1.04-1.49) Population density
Per standard deviation increase:
RR: 1.18 (95% CI: 1.00-1.41) Black African pop. Increased own-group
ethnic density for black
African group reduced incidence:
RR: 0.70 (95% CI: 0.48-0.99) Increased own-group
ethnic segregation for black Caribbean group increased incidence:
RR: 1.54 (95% CI: 1.12-2.03) Black Caribbean pop. Kirkbride et al. 2013. Schizophrenia Bulletin Summary Not explained by age, sex, ethnicity or social class differences Geographical variation associated with deprivation, inequality & population density Some ethnic composition effects observed - may vary by ethnic group Some limitations
Cross-sectional study, possible reverse causation i.e. social drift
Ecological fallacy
Power to detect effects in affective sample
Control for other individual level confounders Possible mechanisms? Mechanisms? Positive symptoms of psychosis influenced by dopaminergic dysfunction May be under stress control, including social stressors encountered in more urban environments Multilevel linear & ordinal logistic regression models Inspected neighbourhood differences in positive symptoms, including paranoia SCAN item group checklist; factor analysis to identify symptom dimensions Look at geographical variation in symptom dimension severity in FEP cases in AESOP 469 FEP cases included In London vs. Nottingham, increase in
Positive symptoms overall (Effect size: 0.17; 95% CI: 0.07, 0.27)
Paranoia (OR: 1.85; 95% CI: 1.31, 2.61) ~5% variance in positive symptoms attributable to neighbourhood level Controlled for age, sex, ethnicity, mode of onset, drug use, family hx Premorbid cognitive impairment in people with non-affective psychosis No premorbid cognitive impairment in people with affective psychosis Prenatal malnutrition, OCs, infections associated with NA No association with perinatal events Early neuro-developmental delays Increased lability to social environmental stressors Increased exposure to social environmental stressors (drift) No increased lability / exposure to deleterious social environments Childhood trauma Substance abuse (?) Migration & ethnicity Significant geographical variation in incidence of non-affective psychoses, but not affective psychoses Social stress associated with urban living, in more densely populated neighbourhoods Compounded / exacerbated by both absolute (i.e. deprivation) and relative (i.e. inequality, ethnic composition) issues What may explain the non-affective / affective differential? No early neuro-developmental delays Oher et al. Submitted Kirkbride et al. 2006. Arch Gen Psych; Kirkbride et al. 2007 SPPE Kirkbride et al. 2013. Schizophrenia Bulletin Kirkbride et al. 2013. Schizophrenia Bulletin Kirkbride et al. 2013. Schizophrenia Bulletin
Full transcript