School of Geography FACULTY OF ENVIRONMENT

School of GeographyFACULTY OF ENVIRONMENT

ESRC Research Award RES-165-25-0032

What happens when international migrants settle? Ethnic group population trends and projections for UK local areas

Estimates of ethnic mortality in the UK

Phil Rees and Pia Wohland

Presentation at the ESRC Research Methods Festival, St. Catherine’s College, Oxford, 1st July 2008

Session: Research Methods for Understanding Population Trends and Processes using

secondary data

School of GeographyFACULTY OF ENVIRONMENT

Introduction-a

Outline of the Presentation

1. INTRODUCTION

• Context and Aim

• Review of previous work

2. METHODS

• Life tables for local areas (plus SMRs, plus survivorship probabilities)

• Estimating the Standardised Illness Ratios (SIRs) for ethnic groups

• Predicting ethnic group SMRs from SIRs

• Adjusting ethnic group SMRs to agree with local mortality

3. RESULTS

• SMRs for ethnic groups for local areas in the UK

• Life tables for ethnic groups for local areas in the UK

4. CONCLUSIONS

•Robustness: check against geographically weighted method

•Robustness: check against new infant mortality data

1. Introduction

2. Methods

3. Results

4. Conclusions

School of GeographyFACULTY OF ENVIRONMENT

ESRC Project: What happens when international migrants settle? Ethnic group population trends and projections for UK local areas

Context and aim1. Introduction

2. Methods

3. Results

4. Conclusions

Mortality component: survivorship probabilities

Fertility component: age specific fertility rates

International migration component:

Internal migration component: census migration updated using Patient Registration Data

New Migrant

databankNew Migrant

databank

Projection model:

Aim: to develop projections of ethnic populations for all local areas in the UK

School of GeographyFACULTY OF ENVIRONMENT

Introduction-a

Review of work on ethnic mortality

1. Introduction

2. Methods

3. Results

4. Conclusions

•United States: Routine to project the population by race (three groups, White, Black, Other). Mortality statistics collected by race.•New Zealand: Interesting model that project groups with inclusion of mixed origin persons included in both parental groups.•European countries: Use nationality (foreign, native) as the classification variable (reviewed in Coleman 2006, PDR paper)•United Kingdom: self-identified ethnic groups used in estimation or projection models but no ethnic mortality differences recognised

• Coleman & Scherbov 2005: all group mortality rates only• Rees and Parsons 2006: all group survivorship probabilities only• Large and Ghosh 2006a, 2006b: all group mortality rates only• Some cause of death work uses country of birth data• In June 2008 ONS has begun publishing new statistics on infant mortality by ethnicity in England and Wales

School of GeographyFACULTY OF ENVIRONMENT

1.Introduction

2.Methods

3.Results

4.Conclusions

Introduction-a

To project populations by ethnicity for local authorities we need to estimate the mortality rates and survivorship probabilities for ethnic groups

Previous UK projections have only used all group mortality and survivorship probabilities

But we know that mortality varies greatly between persons classified by country of birth and from work with the Longitudinal Study

We use an indirect method to estimate ethnic group mortality by using SIRs to estimate SMRs for ethnic groups

Methods: overview (1)

School of GeographyFACULTY OF ENVIRONMENT

1.Introduction

2.Methods

3.Results

4.Conclusions

Methods

We use spreadsheets in a new way to compute life tables efficiently for a large number of local authorities by single year of age and sex. The LTs yield mortality rates … survivorship probabilities and so SMRs.

We estimate SIRs for LAs using the Limiting Long Term Illness variable from the 2001 Census.

We then explore use of alternative sets of LAs for fitting regression models that predict SMR from SIR:•All UK LAs•LAs in England, Wales, Scotland and Northern Ireland separately•LAs in England with & without high BME populations

Methods: overview (2)

School of GeographyFACULTY OF ENVIRONMENT

1. Introduction

2. Methods

3. Results

4. Conclusions

Method 1Using illness data from the census

Stage A compute life tables for all UK LAs: mortality rates

School of GeographyFACULTY OF ENVIRONMENT

1. Introduction

2. Methods

3. Results

4. Conclusions

Method 1Using illness data from the census

Stage A compute life tables for all UK LAs: lx survivors

School of GeographyFACULTY OF ENVIRONMENT

1. Introduction

2. Methods

3. Results

4. Conclusions

Method 1Using illness data from the census

Stage A compute life tables for all UK LAs: survivorship probabilities

School of GeographyFACULTY OF ENVIRONMENT

Stage B Estimate standardized illness ratios (SIRs) by ethnicity for local areas (LAs)

Stage B.1 Estimate SIRs by ethnic group and local areawhere there are enough ill people and total population

1. Introduction

2. Methods

3. Results

4. Conclusions

Method: Using illness data from the census

Map: SIRs for Indians

Indian SIR

150 to 200 (1)

102 to 150 (102)

98 to 102 (21)

75 to 98 (168)

30 to 75 (62)

School of GeographyFACULTY OF ENVIRONMENT

Stage B.2 Estimation of SIRs when numbers are small We used the modelSIR(i,e) = SIR(e) × SIR(i)/100which assumes independence of the national and local effects

1. Introduction

2. Methods

3. Results

4. Conclusions

Method: Using illness data from the census

Bangladeshi SIR

200 to 400 (4)

150 to 200 (63)

102 to 150 (227)

98 to 102 (19)

75 to 98 (40)

30 to 75 (1)

50 100 150 200

50

10

01

50

20

0

Black Caribbean modelled data

Bla

ck C

ari

bb

ea

n o

rig

ina

l da

ta

School of GeographyFACULTY OF ENVIRONMENT

Stage C The relationship between SIR and SMR across local authorities

Illness is linked to mortality at the individual and area scales.

We measured the relationship for various sets of local areas, choosing separate equations for each country

1. Introduction

2. Methods

3. Results

4. Conclusions

Method: Using illness data from the census

Gender Nation r2 Intercept Slope

  England 0.51 52.1 0.48

Females Wales 0.78 43.9 0.37

  Scotland 0.69 60.5 0.64

  Northern Irland 0.16 71.2 0.26

  England 0.63 47.3 0.52

Males Wales 0.56 54.9 0.39

  Scotland 0.75 28.3 0.82

  Northern Irland 0.40 59.9 0.36

School of GeographyFACULTY OF ENVIRONMENT

Stage D Estimate mortality rates by age and sex for ethnic groups from estimated SMRs:

Miexg = Mi

xg × SMRi eg / 100

We assume each ethnic group has the same mortality pattern by age

Stage EAdjust the estimated mortality rates so that they agree with the all group mortality rates

Miexg[2] = Mi

exg[1] × Dixg/Σe Mi

exg[1] Piexg

1. Introduction

2. Methods

3. Results

4. Conclusions

Method: Using illness data from the census

SMR Bangladeshi Males

150 to < 175 (4)

130 to < 150 (41)

115 to < 130 (98)

101 to < 115 (146)

99 to < 101 (22)

85 to < 99 (43)

62 to < 85 (0)

Introduction-a

SMR White British Males

150 + (0)

130 to < 150 (1)

115 to < 130 (18)

101 to < 115 (75)

99 to < 101 (25)

85 to < 99 (202)

62 to <85 (33)

SMR Indian Males

150 + (0)

130 to <150 (1)

115 to <130 (19)

101 to <115 (79)

99 to <101 (17)

85 to <99 (187)

62 to <85 (51)

SMR Chinese Males

150 + (0)

130 to < 150 (0)

115 to < 130 (0)

101 to < 115 (1)

99 to < 101 (0)

85 to < 99 (58)

55 to < 85 (295)

SMRs for White British Males

SMRs for Indian Males

SMRs for Bangladeshi Males

SMRs for Chinese males

Results: SMRs

SMRS for all groups

Introduction-a

All people

White British

Bangladeshi

Indian

Chinese

75,000 to 79,500 (0)

70,000 to 75,000 (29)

65,000 to 70,000 (131)

55,000 to 65,000 (168)

42,500 to 55,000 (26)

Survival at age 75 for males

School of GeographyFACULTY OF ENVIRONMENT

1.Introduction

2.Methods

3.Results

4.Conclusions

Introduction-a

Conclusions: what has been done

•We have produced the first estimates of ethnic group mortality specific to local areas in England

We have produced the first set of life tables for ethnic groups by local area in England

School of GeographyFACULTY OF ENVIRONMENT

1.Introduction

2.Methods

3.Results

4.Conclusions

Introduction-a

Conclusions: checks need doing

•We will develop some estimates of the confidence intervals around selected key life table statistics

•We will compare the method presented here with an alternative based on geographical weighting:

•Estimate the national mortality rates for an ethnic group by averaging the mortality rates for local areas weighted by the population of that group•Feed those estimates into the adjustment step described above

School of GeographyFACULTY OF ENVIRONMENT

1.Introduction

2.Methods

3.Results

4.Conclusions

Introduction-a

Conclusions: what is to be done

•Extend analysis to Wales, Scotland and Northern Ireland

•Develop better models of the SIR-SMR relationship

•Develop life tables for 2002-2006

•Analyse trends and develop projections, variants and scenarios