- Email: [email protected]
Migration and health impact assessment
Public Health (2003) 117, 305–311
Migration and health impact assessment S.J. Lewis* Department of Public Health and Epidemiology, University of Birmingham, Birmingham, UK Received 24 April 2002; received in revised form 29 January 2003; accepted 18 February 2003
KEYWORDS Health impact assessment; Migration; Regeneration; Residential mobility
Summary Government policies, programmes and projects can have a significant impact on health. Health impact assessments (HIAs) seek to estimate this impact, but they often do so by measuring intermediate or proxy indicators and factors that act to determine health. These measures frequently assume a static population. However, regeneration policies can work hard for several years to no apparent effect. One explanation could be migration. Families who have benefited move from the area and other, perhaps more deprived, families move in. Conversely, healthy, prosperous families may move into an improved area, giving the impression that the health of the population has changed, when in fact it is the actual population that has changed. Census data in England and Wales show that a positive correlation exists between migration within wards and deprivation scores. This paper explores the possible implications of migration for HIA. The census, NHS central register, electoral register, labour force survey, central index of the Department of Social Security, council tax database and other data sources are examined to identify what migration data are available at a local level. Factors that determine rates of migration at a local level have been reviewed, with special reference to the differences between population subgroups. The paper concludes with recommendations to take account of residential mobility and changes in migration patterns when carrying out HIAs. Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction There is a growing emphasis placed by the European Union on the need to assess new major policies, projects and programmes for their impact on health.1 This is particularly important in addressing inequalities in health.2 There are many definitions of what constitutes a health impact assessment *Address: Department of Epidemiology and Public Health, University College London, 1– 19 Torrington Place, London WC1E 6BT, UK. E-mail address: [email protected]
(HIA), although the one provided by the European Centre for Health Policy encompasses the main objectives of these assessments which are: ‘to improve knowledge about the potential impact of a policy or programme, inform decision makers and affected people, and facilitate adjustment of the proposed policy in order to mitigate the negative and maximize the positive impacts’.3 The methods used to carry out HIAs are relatively new and not particularly well defined.4 An initial scoping exercise is usually performed that identifies health impacts by literature review, drawing on experience elsewhere and through consultations with key
0033-3506/03/$ - see front matter Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S0033-3506(03)00099-4
306
informants and the community. Measurement of the change in health status of a community following an intervention may be carried out by collecting quantitative measures of health from routine data sources, and also by interviewing the community to collect qualitative information on health. Health impacts are often measured via known determinants of health, which may be physical, social, environmental and lifestyle factors.3 One potential indicator of health within a given community is the percentage of the population that is unemployed. There is a danger that those using or interpreting this data to look at change may not appreciate the dynamic nature of the population in question. Simply presenting the proportion of unemployed individuals within a population may give a limited view of what is actually happening within that group of individuals. We know that the 1991 census recorded 4.6 million residents in England living at a different usual address compared to 12 months earlier.5 This figure is believed to be an underestimate and does not account for multiple moves within the year. Furthermore, the 1991 census covered a period of historically low residential mobility. Therefore, the actual volume of migration in England is probably much greater than suggested above. In fact, turnovers for tenants of local authority stock of 20% per annum are not unusual.6 The aim of this paper is to increase our understanding of the challenges presented by migration when carrying out an HIA. Specifically, four objectives were identified: † to describe factors that determine migration at a local level † to identify the main sources of small-area migration data and assess their value in monitoring migration † to identify methodological problems which may be caused by migration when carrying out HIAs † to make recommendations for handling migration when carrying out HIAs.
Factors that determine migration at a local level
S.J. Lewis
expected to result in a net change in health status, will ultimately result in non-random shifts in migration patterns since the two appear to be interlinked.7 This means, for example, that policies intended to regenerate a community may appear to have no effect when evaluated, because families which have benefited have moved away and other, already deprived, families have moved in. Alternatively, investment within an area may lead to an inward movement of the advantaged. Improved housing, schools and other facilities within an area is likely to attract more prosperous and ‘healthy’ families, which will appear as a net improvement in the health of the population, although in actual fact, the health of the target population may have stayed the same. Using census data, Bentham7 showed a correlation between health and migration, which differed according to age and distance migrated. This suggests that migration is likely to be selective in relation to health and, as such, may fall into one of the following categories:7 † the migration of sick people away from areas with health hazards † migration of sick people in order to be better placed to receive care † migration for housing or labour markets reasons of healthy people. A recent study showed that selective migration over the whole life course at local level does appear to have significantly altered the geographical pattern of mortality seen in Britain today.8 This is important considering that between-district mortality ratios vary by as much as a factor of two, and by electoral wards there may be as much as a fourfold difference. Young migrants, especially those moving longer distances, have been shown to be healthy relative to non-migrants of the same age.7 Since it is young adults who make up the majority of migrants, areas of net outmigration of this age group are likely to consist of a less healthy population, with the reverse being true for areas of net inmigration. Migration amongst elderly people tends to be those in poor health moving to avoid environmental hazards or to be better placed to receive care.7 This could result in an increase in morbidity and mortality in areas with good medical care and favourable environmental conditions.9
Health status and migration Age and stage of life Very little has been published on the impact of regeneration policies on migration. However, it is conceivable that policies, programmes and projects that are aimed at improving health, or are at least
Studies have shown that factors such as housing tenure, socio-economic position and educational status, which may be used as surrogate measures
Migration and health impact assessment
of health in HIA, also affect distance and levels of migration.8 The influence that these factors have on migration differs systematically between population subgroups. For this reason, population composition is a key correlate of migration. Migration has been shown to differ with age. Rates for young children (and their parents) decline gradually until age 15. From 16 years, migration rates rise again to reach a peak at 22 years. This is followed by a decline to a minimum in the late 50s. There is a slight increase in migration propensity at age 60 – 65 years, about the age when people retire. This then falls, followed by a further increase in migration when people are in their 70 and 80s.5,10 Up until age 16, a child’s migration pattern commonly follows that of their parents, but from age 16 onwards, people are usually responsible for their own movements. The first move as an adult tends to be leaving a parent’s home. This may be a local move or movement into a city to begin higher education or a job. At around 20 –25 years of age, people may move again to take up residency with a partner. Other moves for people in their 20s might be to take up a career position or movement into a larger house because of the first child. Mid-career promotion or inheritance may cause movement to a larger house for people in their 30 – 50s. Divorce and second cohabitations or marriages are also common reasons for migration among the middle aged. Retirement often results in a move of house. This may be followed by elderly people moving due to a bereavement, frailty or chronic illness.8
Cultural context and ethnicity When age and recent immigration are taken into account, black and minority ethnic groups move house less frequently than whites.11 However, the migratory rates of these ethnic groups differ considerably, with Chinese and Black Africans being twice as likely to migrate as Indians, Black Caribbeans and Pakistanis. There are also large differences in migration rates within ethnic groups across the country. In general, turnover in small communities in peripheral areas is much higher than in established minority ethnic group concentrations in London, the Midlands and West Yorkshire.5,12 Cultural and religious factors may also influence population movement and migration, independent of ethnicity.12
Housing tenure The majority of short-distance moves are for improved housing reasons.5 However, moves over longer distances face constraints such as variations
307
in house prices and rents, and, in particular, administrative controls restricting potential population movement into council housing over long distances.13 This means that although those moving into council housing are more likely to move than those moving into owner-occupied housing, they are less likely to move long distances.13 For certain relatively mobile subgroups, such as young professionals, privately rented housing offers the most flexible choice of tenure. However, only a small proportion of property in the UK falls into this category.13
Occupation and educational status The relationship between socio-economic status and mobility depends on the distance of migration. Inter-regional migration is dominated by professionals and managers, and hence, those educated to degree level.14 People employed in manual occupations have relatively low inter-regional migration rates. The majority of inter-regional migrants are employed at the start and end of the year of migration. However, migrants moving out of employment exceed those moving into employment. One possible explanation for this could be the return migration of those who have not been successful in the employment market.5 A study of low-skilled unemployed people in Liverpool showed that secure local housing and support of family and friends carried stronger weight than uncertain job prospects elsewhere.15 Migration over short distances is a different story. Within regions, it is the low socio-economic groups who are more likely to migrate.11 In support of this, a positive correlation exists between migration and Townsend deprivation score for within-ward migration, but no such correlation exists for other types of migration.
Environmental factors Apart from 16 to 29 year olds, who migrate towards the cities either for higher education or jobs, migration often tends to be away from highpopulation density areas.5 The advantages of living in a physically attractive environment and the desire to seek an alternative lifestyle and community to that in the city are the main driving factors. Other considerations influencing where people choose to live might be: accessibility to friends and family, level of social ‘buzz’, degree of antisocial behaviour, congestion, noise, proximity to retail facilities, entertainment and leisure facilities, medical facilities and nursing homes.5
308
Migration data sources An ideal source of migration data for use when carrying out HIAs would: † provide total coverage of the study population to reduce the possibility of bias † be updated at least yearly to allow the monitoring of migration trends around the time of the intervention † be disaggregated to a small-area level so that migration trends can be monitored among local communities † provide information on socio-economic characteristics, age and sex prior to the move to allow analysis of migration for different subgroups of the population. A review of the main sources of migration data was carried out in 199116 and the situation has changed very little since then.5 None of the data sources identified fulfill all the above criteria; however, the strengths and weaknesses of each have been identified (Table 1).
Other sources of migration data Television licence, driving licence, utilities (water, gas and electricity), and bank and credit card records may all provide some information on migration. They are less comprehensive than those reviewed in Table 1, either because they offer incomplete coverage or individuals are under no obligation to report changes of address. In the case of television licensing and utilities, only the head of the household is recorded;16 this concept will be meaningless in group housing, for example students sharing. Even within traditional households, the person taking responsibility for the bills, and therefore given ‘head of household’ status, may not be male and the main earner. However, these data sources may be of use in monitoring changes in mobility within small areas. If the HIA being carried out is concerned with a confined community, such as a housing estate, then additions and deductions to local general practitioner lists, school registers and housing authority records may be of some use in determining migration patterns. The quality of the data and the information available from these sources will be variable, and the validity of the data source will depend on the HIA in question. For instance, if a project is evaluating the impact of providing free winter fuel for the elderly within a community, general practitioner lists might be a more relevant source of migration data than school registers.
S.J. Lewis
The problem of migration in health impact assessments Migration is an extremely important influence on the population demographics of an area and vice versa.5 Migration dynamics in the UK differ widely particularly between small areas within a region. This may be due to population demographics, as well as environmental and socio-economic factors. The magnitude and direction of migration has been shown to be volatile in the short term and this volatility again is greater within regions than between regions.5 This makes levels of migration extremely difficult to predict.5 Migration is a selective process, and selectivity is related to health.7 Migrants often tend to be healthier; therefore, areas with high levels of inmigration are characterized by a healthy population. They also tend to move to areas that are economically successful. Other factors known to be associated with migration, i.e. socio-economic status, educational level and housing tenure, are typically measured as determinants of the health of a population. This means that health indicators within a given area will be affected by a sudden change in the pattern of migration. Therefore, migration may be a confounder in HIAs if the intervention being evaluated improves health and increases migration. At present, there is no comprehensive source of migration data at a small-area level relating to number of migrants, let alone socio-demographic characteristics and motivations for moving. The census is the only possibility for studying socioeconomic characteristics of migrants. However, this is of limited use for HIAs since it is only updated every 10 years and only provides a snapshot of moves in this period. The council tax database, Department of Social Security and telephone directories all offer some information on migration at a small-area level, but are open to bias due to non-registration to avoid payment and lack of a telephone, etc. Other local sources of data such as general practitioner lists, school registers and housing authority turnovers need to be investigated locally to determine the level of information available. Recommendations for dealing with migration in HIAs: † As migration is associated with health and determinants of health, HIAs should at least consider whether the intervention in question is likely to result in changes in the migratory potential of the community being studied.
Migration and health impact assessment
309
Table 1 Migration data sources. Data source
Collection frequency
Area covered
Accessibility
Information
Weaknesses
Census
Every 10 years
Total UK population broken down into ward level
Accessible 18 months after collection
Current address and address 12 months prior to census, sex, age, ethnic group, current health status, economic position and tenure
May be up to 11 years 6 months out of date. Only provides a snapshot of what is happening 1 year in every 10. Does not provide data on multiple moves within the year. Does provide data on socio-economic status prior to move.
NHS central register
Updated continuously
Movement between health authorities for all individuals with an NHS number
Published in: key population vital statistics, regional trends, social trends, population trends (ONS). Data may be accessed directly via health authority
Age, sex, health authority of previous registration
Does not provide data on moves at small-area level. Does not provide data on the socio-economic characteristics of migrants. Likely to be a delay between people moving and registering with a new CP. Some moves are not recorded as individuals do not always register with a new before the next move.
Electoral register
Updated yearly
All British Common wealth Citizens over 18 years of age who are resident on a qualifying date
Widely accessible
Name and address (may be possible to make some assumptions about socio-economic status from postcode)
Wide variation in frequency and thoroughness with which electoral registration offices remove people from the register. Provides no information on age. Excludes children and aliens. Likely to be some dual registration as it is not illegal to register in two places.
Labour force survey
Carried out quarterly
A sample of 60,000 households in the UK. Each is included in the survey five times before being replaced
Accessible via ONS
Age, sex, address, address 1 year prior to survey, country of birth, current socio-economic position and socioeconomic position 1 year prior to survey
Only a sample of the population is represented. Does not necessarily collect data on all moves. Limited value when looking at small areas due to large standard errors.
Council tax database
Updated continuously
All households in the UK are registered with their local authority
Data on deductions and additions to the list can be obtained by contacting local authorities
Number of people within a household and address (may be possible to make some assumptions about socioeconomic status from postcode)
Does not provide data on age or sex. Information on forward and previous address not collected. There are problems with non-registration that are biased towards frequent migrants. (continued on next page)
310
S.J. Lewis
Table 1 (continued) Data source
Collection frequency
Area covered
Accessibility
Information
Weaknesses
Department of social security (DSS)
Updated continuously
Every person in the UK with a national insurance number is included
It may be possible to access information on changes of address between given time intervals
Age, sex and benefit claims
Database only stores most recent address and so cannot be used to detect all moves. Moves by individuals not working or claiming benefits will not be detected. Database is biased towards those claiming benefits, who are more likely to report changes in address.
Telephone directories
Updated continuously
All households with a telephone are included
It may be possible to access information on changes of address between given time intervals
Name of head of household and address (may be possible to make some assumptions about socio-economic status from postcode)
Incomplete coverage as only one name per household is recorded. No information on age or sex. Data is biased towards those with a land-line telephone. Telephone owners can opt to be excluded from directories.
Consideration should also be given as to whether migration is likely to be an outcome variable or a confounder of the HIA. † In order to determine any changes in migration resulting from the intervention in question, it will be necessary to determine migration patterns for a considerable period prior to and following the intervention. If possible, age, sex and socio-economic characteristics of migrants should also be collected to determine any changes in who is actually migrating. This could be carried out by monitoring one or more of the routine data sources outlined previously or by interviews with the community. † Often it is not feasible to carry out a longitudinal HIA that monitors changes within a population following an intervention. Most HIAs to date have only documented likely changes prior to an event occurring. In this scenario, it is still important to build up a picture of local demographics and assess the fluidity of the population in order to determine how this may be affected, and whether this may confound any changes in the health of the population. This could be carried out by looking for recent migration patterns to see how many individuals or families are moving into and out of the area. One example of a very crude way of estimating current migration patterns at the family level would be to compare
a snapshot of telephone directories from the 2 consecutive years prior to the intervention, although this will be opened to bias as described in Table 1. † The data source used to determine migration levels should be relevant to the intervention in question. For example, in the case of some HIAs, it may be sufficient to determine migration at health authority or district level, whereas for others, migration at a much smaller level is needed. The data source selected should also be tailored to the population, it would be futile to use a telephone directory to determine levels of migration in a community of high deprivation in which only a small proportion of homes own a telephone. † It may be necessary to determine migration levels via a number of different sources in order to overcome potential biases. Where an appropriate data source does not exist, questions on migration may be incorporated into surveys conducted as part of the HIA process.
Acknowledgements I would like to acknowledge Drs Jayne Parry and Richard Wilson, Patrick Saunders, Duncan Cooper
Migration and health impact assessment
and Professor Andrew Stevens of the Department of Public Health and Epidemiology, University of Birmingham for their help with this work.
References 1. Lehto J, Ritsatakis A. Health impact assessment as a tool for intersectoral health policy. Proceedings of the Health Impact Assessment: From Theory to Practice Conference, Oct 28—31; Gothenburg, Sweden. World Health Organisation, Regional Office for Europe, 1999 2. Department of Health, Our healthier nation: a contract for health. Cm 3852. London: The Stationery Office; 1998. 3. European Centre for Health Policy, World Health Organization Regional Office for Europe, Health impact assessment: main concepts and suggested approach. Brussels: WHO; 1999. 4. Parry J, Stevens A. Prospective health impact assessment: pitfalls, problems and possible ways forward. BMJ 2001;323: 1177—82. 5. Champion T, Fotheringham S, Rees P, Boyle P, Stillwell J. The determinants of migration flows in England: a review of existing data and evidence. A report prepared for the Department of the Environment, Newcastle: The University of Newcastle upon Tyne; 1998. 6. Rhodes J, Tyler P. Health impact assessment—methodology. Department of Health. Health impact assessment: report of a methodological seminar. London: The Stationery Office; 1999.
311
7. Bentham G. Migration and morbidity: implications for geographical studies of disease. Soc Sci Med 1988;26:49—54. 8. Brimblecombe N, Dorling D, Shaw M. Mortality and migration in Britain, first results from the British Household Panel Survey. Soc Sci Med 1999;49:981—8. 9. Boyle PJ, Gatrell AC, Duke-Williams O. The effect on morbidity of variability in deprivation and population stability in England and Wales: an investigation at smallarea level. Soc Sci Med 1999;49:791—9. 10. Stillwell JCH. Monitoring intercensal migration in the United Kingdom. Environ Plan 1994;26:1711—30. 11. Owen D. Migration and employment. In: Stillwell J, Rees P, Boden P, editors. Migration processes and patterns. Volume 2: population redistribution in the United Kingdom. London: Belhaven; 1992. 12. Robinson V. Move on up: the mobility of Britain’s AfroCaribbean and Asian populations. In: Stillwell J, Rees P, Boden P, editors. Migration processes and patterns. Volume 2: population redistribution in the United Kingdom. London: Belhaven; 1992. 13. Boyle P. Migration and housing tenure in South East England. Environ Plan 1992;30:855—6. 14. Owen D, Green A. Migration patterns and trends. In: Champion A, Fielding A, editors. Migration processes and patterns. Volume 1: research progress and prospects. London: Belhaven; 1992. 15. Kitching R. Migration behaviour among the unemployed and low-skilled. In: Johnson JH, Salt J, editors. Labour migration: the internal geographic mobility of labour in the developed world. London: David Fulton; 1990. 16. Bulusu L. A review of migration data. OPCS occasional paper 39. London: Office of Population Censuses and Surveys; 1991.
Migration and health impact assessment S.J. Lewis* Department of Public Health and Epidemiology, University of Birmingham, Birmingham, UK Received 24 April 2002; received in revised form 29 January 2003; accepted 18 February 2003
KEYWORDS Health impact assessment; Migration; Regeneration; Residential mobility
Summary Government policies, programmes and projects can have a significant impact on health. Health impact assessments (HIAs) seek to estimate this impact, but they often do so by measuring intermediate or proxy indicators and factors that act to determine health. These measures frequently assume a static population. However, regeneration policies can work hard for several years to no apparent effect. One explanation could be migration. Families who have benefited move from the area and other, perhaps more deprived, families move in. Conversely, healthy, prosperous families may move into an improved area, giving the impression that the health of the population has changed, when in fact it is the actual population that has changed. Census data in England and Wales show that a positive correlation exists between migration within wards and deprivation scores. This paper explores the possible implications of migration for HIA. The census, NHS central register, electoral register, labour force survey, central index of the Department of Social Security, council tax database and other data sources are examined to identify what migration data are available at a local level. Factors that determine rates of migration at a local level have been reviewed, with special reference to the differences between population subgroups. The paper concludes with recommendations to take account of residential mobility and changes in migration patterns when carrying out HIAs. Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction There is a growing emphasis placed by the European Union on the need to assess new major policies, projects and programmes for their impact on health.1 This is particularly important in addressing inequalities in health.2 There are many definitions of what constitutes a health impact assessment *Address: Department of Epidemiology and Public Health, University College London, 1– 19 Torrington Place, London WC1E 6BT, UK. E-mail address: [email protected]
(HIA), although the one provided by the European Centre for Health Policy encompasses the main objectives of these assessments which are: ‘to improve knowledge about the potential impact of a policy or programme, inform decision makers and affected people, and facilitate adjustment of the proposed policy in order to mitigate the negative and maximize the positive impacts’.3 The methods used to carry out HIAs are relatively new and not particularly well defined.4 An initial scoping exercise is usually performed that identifies health impacts by literature review, drawing on experience elsewhere and through consultations with key
0033-3506/03/$ - see front matter Q 2003 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/S0033-3506(03)00099-4
306
informants and the community. Measurement of the change in health status of a community following an intervention may be carried out by collecting quantitative measures of health from routine data sources, and also by interviewing the community to collect qualitative information on health. Health impacts are often measured via known determinants of health, which may be physical, social, environmental and lifestyle factors.3 One potential indicator of health within a given community is the percentage of the population that is unemployed. There is a danger that those using or interpreting this data to look at change may not appreciate the dynamic nature of the population in question. Simply presenting the proportion of unemployed individuals within a population may give a limited view of what is actually happening within that group of individuals. We know that the 1991 census recorded 4.6 million residents in England living at a different usual address compared to 12 months earlier.5 This figure is believed to be an underestimate and does not account for multiple moves within the year. Furthermore, the 1991 census covered a period of historically low residential mobility. Therefore, the actual volume of migration in England is probably much greater than suggested above. In fact, turnovers for tenants of local authority stock of 20% per annum are not unusual.6 The aim of this paper is to increase our understanding of the challenges presented by migration when carrying out an HIA. Specifically, four objectives were identified: † to describe factors that determine migration at a local level † to identify the main sources of small-area migration data and assess their value in monitoring migration † to identify methodological problems which may be caused by migration when carrying out HIAs † to make recommendations for handling migration when carrying out HIAs.
Factors that determine migration at a local level
S.J. Lewis
expected to result in a net change in health status, will ultimately result in non-random shifts in migration patterns since the two appear to be interlinked.7 This means, for example, that policies intended to regenerate a community may appear to have no effect when evaluated, because families which have benefited have moved away and other, already deprived, families have moved in. Alternatively, investment within an area may lead to an inward movement of the advantaged. Improved housing, schools and other facilities within an area is likely to attract more prosperous and ‘healthy’ families, which will appear as a net improvement in the health of the population, although in actual fact, the health of the target population may have stayed the same. Using census data, Bentham7 showed a correlation between health and migration, which differed according to age and distance migrated. This suggests that migration is likely to be selective in relation to health and, as such, may fall into one of the following categories:7 † the migration of sick people away from areas with health hazards † migration of sick people in order to be better placed to receive care † migration for housing or labour markets reasons of healthy people. A recent study showed that selective migration over the whole life course at local level does appear to have significantly altered the geographical pattern of mortality seen in Britain today.8 This is important considering that between-district mortality ratios vary by as much as a factor of two, and by electoral wards there may be as much as a fourfold difference. Young migrants, especially those moving longer distances, have been shown to be healthy relative to non-migrants of the same age.7 Since it is young adults who make up the majority of migrants, areas of net outmigration of this age group are likely to consist of a less healthy population, with the reverse being true for areas of net inmigration. Migration amongst elderly people tends to be those in poor health moving to avoid environmental hazards or to be better placed to receive care.7 This could result in an increase in morbidity and mortality in areas with good medical care and favourable environmental conditions.9
Health status and migration Age and stage of life Very little has been published on the impact of regeneration policies on migration. However, it is conceivable that policies, programmes and projects that are aimed at improving health, or are at least
Studies have shown that factors such as housing tenure, socio-economic position and educational status, which may be used as surrogate measures
Migration and health impact assessment
of health in HIA, also affect distance and levels of migration.8 The influence that these factors have on migration differs systematically between population subgroups. For this reason, population composition is a key correlate of migration. Migration has been shown to differ with age. Rates for young children (and their parents) decline gradually until age 15. From 16 years, migration rates rise again to reach a peak at 22 years. This is followed by a decline to a minimum in the late 50s. There is a slight increase in migration propensity at age 60 – 65 years, about the age when people retire. This then falls, followed by a further increase in migration when people are in their 70 and 80s.5,10 Up until age 16, a child’s migration pattern commonly follows that of their parents, but from age 16 onwards, people are usually responsible for their own movements. The first move as an adult tends to be leaving a parent’s home. This may be a local move or movement into a city to begin higher education or a job. At around 20 –25 years of age, people may move again to take up residency with a partner. Other moves for people in their 20s might be to take up a career position or movement into a larger house because of the first child. Mid-career promotion or inheritance may cause movement to a larger house for people in their 30 – 50s. Divorce and second cohabitations or marriages are also common reasons for migration among the middle aged. Retirement often results in a move of house. This may be followed by elderly people moving due to a bereavement, frailty or chronic illness.8
Cultural context and ethnicity When age and recent immigration are taken into account, black and minority ethnic groups move house less frequently than whites.11 However, the migratory rates of these ethnic groups differ considerably, with Chinese and Black Africans being twice as likely to migrate as Indians, Black Caribbeans and Pakistanis. There are also large differences in migration rates within ethnic groups across the country. In general, turnover in small communities in peripheral areas is much higher than in established minority ethnic group concentrations in London, the Midlands and West Yorkshire.5,12 Cultural and religious factors may also influence population movement and migration, independent of ethnicity.12
Housing tenure The majority of short-distance moves are for improved housing reasons.5 However, moves over longer distances face constraints such as variations
307
in house prices and rents, and, in particular, administrative controls restricting potential population movement into council housing over long distances.13 This means that although those moving into council housing are more likely to move than those moving into owner-occupied housing, they are less likely to move long distances.13 For certain relatively mobile subgroups, such as young professionals, privately rented housing offers the most flexible choice of tenure. However, only a small proportion of property in the UK falls into this category.13
Occupation and educational status The relationship between socio-economic status and mobility depends on the distance of migration. Inter-regional migration is dominated by professionals and managers, and hence, those educated to degree level.14 People employed in manual occupations have relatively low inter-regional migration rates. The majority of inter-regional migrants are employed at the start and end of the year of migration. However, migrants moving out of employment exceed those moving into employment. One possible explanation for this could be the return migration of those who have not been successful in the employment market.5 A study of low-skilled unemployed people in Liverpool showed that secure local housing and support of family and friends carried stronger weight than uncertain job prospects elsewhere.15 Migration over short distances is a different story. Within regions, it is the low socio-economic groups who are more likely to migrate.11 In support of this, a positive correlation exists between migration and Townsend deprivation score for within-ward migration, but no such correlation exists for other types of migration.
Environmental factors Apart from 16 to 29 year olds, who migrate towards the cities either for higher education or jobs, migration often tends to be away from highpopulation density areas.5 The advantages of living in a physically attractive environment and the desire to seek an alternative lifestyle and community to that in the city are the main driving factors. Other considerations influencing where people choose to live might be: accessibility to friends and family, level of social ‘buzz’, degree of antisocial behaviour, congestion, noise, proximity to retail facilities, entertainment and leisure facilities, medical facilities and nursing homes.5
308
Migration data sources An ideal source of migration data for use when carrying out HIAs would: † provide total coverage of the study population to reduce the possibility of bias † be updated at least yearly to allow the monitoring of migration trends around the time of the intervention † be disaggregated to a small-area level so that migration trends can be monitored among local communities † provide information on socio-economic characteristics, age and sex prior to the move to allow analysis of migration for different subgroups of the population. A review of the main sources of migration data was carried out in 199116 and the situation has changed very little since then.5 None of the data sources identified fulfill all the above criteria; however, the strengths and weaknesses of each have been identified (Table 1).
Other sources of migration data Television licence, driving licence, utilities (water, gas and electricity), and bank and credit card records may all provide some information on migration. They are less comprehensive than those reviewed in Table 1, either because they offer incomplete coverage or individuals are under no obligation to report changes of address. In the case of television licensing and utilities, only the head of the household is recorded;16 this concept will be meaningless in group housing, for example students sharing. Even within traditional households, the person taking responsibility for the bills, and therefore given ‘head of household’ status, may not be male and the main earner. However, these data sources may be of use in monitoring changes in mobility within small areas. If the HIA being carried out is concerned with a confined community, such as a housing estate, then additions and deductions to local general practitioner lists, school registers and housing authority records may be of some use in determining migration patterns. The quality of the data and the information available from these sources will be variable, and the validity of the data source will depend on the HIA in question. For instance, if a project is evaluating the impact of providing free winter fuel for the elderly within a community, general practitioner lists might be a more relevant source of migration data than school registers.
S.J. Lewis
The problem of migration in health impact assessments Migration is an extremely important influence on the population demographics of an area and vice versa.5 Migration dynamics in the UK differ widely particularly between small areas within a region. This may be due to population demographics, as well as environmental and socio-economic factors. The magnitude and direction of migration has been shown to be volatile in the short term and this volatility again is greater within regions than between regions.5 This makes levels of migration extremely difficult to predict.5 Migration is a selective process, and selectivity is related to health.7 Migrants often tend to be healthier; therefore, areas with high levels of inmigration are characterized by a healthy population. They also tend to move to areas that are economically successful. Other factors known to be associated with migration, i.e. socio-economic status, educational level and housing tenure, are typically measured as determinants of the health of a population. This means that health indicators within a given area will be affected by a sudden change in the pattern of migration. Therefore, migration may be a confounder in HIAs if the intervention being evaluated improves health and increases migration. At present, there is no comprehensive source of migration data at a small-area level relating to number of migrants, let alone socio-demographic characteristics and motivations for moving. The census is the only possibility for studying socioeconomic characteristics of migrants. However, this is of limited use for HIAs since it is only updated every 10 years and only provides a snapshot of moves in this period. The council tax database, Department of Social Security and telephone directories all offer some information on migration at a small-area level, but are open to bias due to non-registration to avoid payment and lack of a telephone, etc. Other local sources of data such as general practitioner lists, school registers and housing authority turnovers need to be investigated locally to determine the level of information available. Recommendations for dealing with migration in HIAs: † As migration is associated with health and determinants of health, HIAs should at least consider whether the intervention in question is likely to result in changes in the migratory potential of the community being studied.
Migration and health impact assessment
309
Table 1 Migration data sources. Data source
Collection frequency
Area covered
Accessibility
Information
Weaknesses
Census
Every 10 years
Total UK population broken down into ward level
Accessible 18 months after collection
Current address and address 12 months prior to census, sex, age, ethnic group, current health status, economic position and tenure
May be up to 11 years 6 months out of date. Only provides a snapshot of what is happening 1 year in every 10. Does not provide data on multiple moves within the year. Does provide data on socio-economic status prior to move.
NHS central register
Updated continuously
Movement between health authorities for all individuals with an NHS number
Published in: key population vital statistics, regional trends, social trends, population trends (ONS). Data may be accessed directly via health authority
Age, sex, health authority of previous registration
Does not provide data on moves at small-area level. Does not provide data on the socio-economic characteristics of migrants. Likely to be a delay between people moving and registering with a new CP. Some moves are not recorded as individuals do not always register with a new before the next move.
Electoral register
Updated yearly
All British Common wealth Citizens over 18 years of age who are resident on a qualifying date
Widely accessible
Name and address (may be possible to make some assumptions about socio-economic status from postcode)
Wide variation in frequency and thoroughness with which electoral registration offices remove people from the register. Provides no information on age. Excludes children and aliens. Likely to be some dual registration as it is not illegal to register in two places.
Labour force survey
Carried out quarterly
A sample of 60,000 households in the UK. Each is included in the survey five times before being replaced
Accessible via ONS
Age, sex, address, address 1 year prior to survey, country of birth, current socio-economic position and socioeconomic position 1 year prior to survey
Only a sample of the population is represented. Does not necessarily collect data on all moves. Limited value when looking at small areas due to large standard errors.
Council tax database
Updated continuously
All households in the UK are registered with their local authority
Data on deductions and additions to the list can be obtained by contacting local authorities
Number of people within a household and address (may be possible to make some assumptions about socioeconomic status from postcode)
Does not provide data on age or sex. Information on forward and previous address not collected. There are problems with non-registration that are biased towards frequent migrants. (continued on next page)
310
S.J. Lewis
Table 1 (continued) Data source
Collection frequency
Area covered
Accessibility
Information
Weaknesses
Department of social security (DSS)
Updated continuously
Every person in the UK with a national insurance number is included
It may be possible to access information on changes of address between given time intervals
Age, sex and benefit claims
Database only stores most recent address and so cannot be used to detect all moves. Moves by individuals not working or claiming benefits will not be detected. Database is biased towards those claiming benefits, who are more likely to report changes in address.
Telephone directories
Updated continuously
All households with a telephone are included
It may be possible to access information on changes of address between given time intervals
Name of head of household and address (may be possible to make some assumptions about socio-economic status from postcode)
Incomplete coverage as only one name per household is recorded. No information on age or sex. Data is biased towards those with a land-line telephone. Telephone owners can opt to be excluded from directories.
Consideration should also be given as to whether migration is likely to be an outcome variable or a confounder of the HIA. † In order to determine any changes in migration resulting from the intervention in question, it will be necessary to determine migration patterns for a considerable period prior to and following the intervention. If possible, age, sex and socio-economic characteristics of migrants should also be collected to determine any changes in who is actually migrating. This could be carried out by monitoring one or more of the routine data sources outlined previously or by interviews with the community. † Often it is not feasible to carry out a longitudinal HIA that monitors changes within a population following an intervention. Most HIAs to date have only documented likely changes prior to an event occurring. In this scenario, it is still important to build up a picture of local demographics and assess the fluidity of the population in order to determine how this may be affected, and whether this may confound any changes in the health of the population. This could be carried out by looking for recent migration patterns to see how many individuals or families are moving into and out of the area. One example of a very crude way of estimating current migration patterns at the family level would be to compare
a snapshot of telephone directories from the 2 consecutive years prior to the intervention, although this will be opened to bias as described in Table 1. † The data source used to determine migration levels should be relevant to the intervention in question. For example, in the case of some HIAs, it may be sufficient to determine migration at health authority or district level, whereas for others, migration at a much smaller level is needed. The data source selected should also be tailored to the population, it would be futile to use a telephone directory to determine levels of migration in a community of high deprivation in which only a small proportion of homes own a telephone. † It may be necessary to determine migration levels via a number of different sources in order to overcome potential biases. Where an appropriate data source does not exist, questions on migration may be incorporated into surveys conducted as part of the HIA process.
Acknowledgements I would like to acknowledge Drs Jayne Parry and Richard Wilson, Patrick Saunders, Duncan Cooper
Migration and health impact assessment
and Professor Andrew Stevens of the Department of Public Health and Epidemiology, University of Birmingham for their help with this work.
References 1. Lehto J, Ritsatakis A. Health impact assessment as a tool for intersectoral health policy. Proceedings of the Health Impact Assessment: From Theory to Practice Conference, Oct 28—31; Gothenburg, Sweden. World Health Organisation, Regional Office for Europe, 1999 2. Department of Health, Our healthier nation: a contract for health. Cm 3852. London: The Stationery Office; 1998. 3. European Centre for Health Policy, World Health Organization Regional Office for Europe, Health impact assessment: main concepts and suggested approach. Brussels: WHO; 1999. 4. Parry J, Stevens A. Prospective health impact assessment: pitfalls, problems and possible ways forward. BMJ 2001;323: 1177—82. 5. Champion T, Fotheringham S, Rees P, Boyle P, Stillwell J. The determinants of migration flows in England: a review of existing data and evidence. A report prepared for the Department of the Environment, Newcastle: The University of Newcastle upon Tyne; 1998. 6. Rhodes J, Tyler P. Health impact assessment—methodology. Department of Health. Health impact assessment: report of a methodological seminar. London: The Stationery Office; 1999.
311
7. Bentham G. Migration and morbidity: implications for geographical studies of disease. Soc Sci Med 1988;26:49—54. 8. Brimblecombe N, Dorling D, Shaw M. Mortality and migration in Britain, first results from the British Household Panel Survey. Soc Sci Med 1999;49:981—8. 9. Boyle PJ, Gatrell AC, Duke-Williams O. The effect on morbidity of variability in deprivation and population stability in England and Wales: an investigation at smallarea level. Soc Sci Med 1999;49:791—9. 10. Stillwell JCH. Monitoring intercensal migration in the United Kingdom. Environ Plan 1994;26:1711—30. 11. Owen D. Migration and employment. In: Stillwell J, Rees P, Boden P, editors. Migration processes and patterns. Volume 2: population redistribution in the United Kingdom. London: Belhaven; 1992. 12. Robinson V. Move on up: the mobility of Britain’s AfroCaribbean and Asian populations. In: Stillwell J, Rees P, Boden P, editors. Migration processes and patterns. Volume 2: population redistribution in the United Kingdom. London: Belhaven; 1992. 13. Boyle P. Migration and housing tenure in South East England. Environ Plan 1992;30:855—6. 14. Owen D, Green A. Migration patterns and trends. In: Champion A, Fielding A, editors. Migration processes and patterns. Volume 1: research progress and prospects. London: Belhaven; 1992. 15. Kitching R. Migration behaviour among the unemployed and low-skilled. In: Johnson JH, Salt J, editors. Labour migration: the internal geographic mobility of labour in the developed world. London: David Fulton; 1990. 16. Bulusu L. A review of migration data. OPCS occasional paper 39. London: Office of Population Censuses and Surveys; 1991.