Migration and child health inequities in Nigeria: a multilevel analysis of contextual- and individual-level factors

Objective To assess the role of rural–urban migration in the risks of under-five death; to identify possible mechanisms through which migration may influence mortality; and to determine individual- and community-level relationships between migration status and under-five death. Method Multilevel Cox regression analysis was used on a nationally representative sample of 6029 children from 2735 mothers aged 15–49 years and nested within 365 communities from the 2003 Nigeria Demographic and Health Survey. Hazard ratios with 95% confidence intervals were used to express the measures of association between the characteristics, and intra-class coefficients were used to express the measures of variation. Results Children of rural non-migrant mothers had significantly lower risks of under-five death than children of rural–urban migrant mothers. The disruption of family and community ties, low socio-economic position and vulnerability, and the difficulties migrants face in adapting into the new urban environment, may predispose the children of rural–urban migrants to higher mortality. Conclusion Our results stress the need for community-level and socio-economic interventions targeted at migrant groups within urban areas to improve their access to health care services, maternal education, as well as the general socio-economic situation of women. Objectif: Evaluer le rôle de la migration rurale-urbaine dans les risques de décès chez les moins de cinq ans, identifier les mécanismes possibles par lesquels la migration peut influer sur la mortalité et déterminer les relations entre l’individu, la communauté et l’état de la migration avec les décès des moins de cinq ans. Méthode: L’analyse de régression de Cox multi-niveau a été utilisée sur un échantillon national représentatif de 6029 enfants de 2735 mères âgées de 15 à 49 ans et nichées au sein de 365 communautés provenant de l’Enquête sur la Démographique et la Santé de 2003 au Nigéria. Des rapports de risque avec des intervalles de confiance de 95% ont été utilisés pour exprimer les mesures d’associations entre les caractéristiques et des coefficients intra-classe ont été utilisés pour exprimer les mesures de variation. Résultats: Les enfants de mères rurales non-migrantes avaient des risques de décès des moins de cinq ans significativement plus faibles que les enfants de mères migrantes rurales-urbaines. La perturbation des liens familiaux et communautaires, la faible position socio-économique et la vulnérabilité ainsi que les difficultés rencontrées par les migrants à s’adapter dans le nouvel environnement urbain peuvent prédisposer les enfants des migrants ruraux-urbains à une mortalité plus élevée. Conclusion: Nos résultats soulignent la nécessité pour des interventions socio-économiques à l’échelle de la communauté et ciblées sur des groupes migrants dans les zones urbaines, pour améliorer leur accès aux services de soins de santé, à l’éducation maternelle ainsi que la situation socio-économique des femmes. Mots-clés: migration, perturbation, sélectivité, adaptation, mortalité nourrissons, enfants et moins de cinq ans, modélisation multi-niveau, Nigeria. Objetivo: Evaluar el papel de la migración rural-urbana en los riesgos de muerte de menores de cinco años; para identificar posibles mecanismos a través de los cuales la migración puede influenciar la mortalidad; y determinar las relaciones a nivel individual y comunitaria entre el estatus de migrante y la mortalidad de menores de cinco año. Método: Se utilizó un análisis de regresión de Cox multinivel sobre una muestra representativa a nivel nacional de 6029 niños de 2735 madres con 15-49 años y anidado en 365 comunidades del censo demográfico y sanitario de Nigeria del 2003. Las tasas de riesgo relativo, con intervalos de confianza del 95% se utilizaron para expresar las medidas de asociación entre las características, y los coeficientes intra-clase se utilizaron para expresar medidas de variación. Resultados: Los niños de madres rurales no inmigrantes tenían un riesgo significativamente menor de morir antes de los cinco años que los niños con madres rurales e inmigrantes. La ruptura de lazos familiares y comunitarios, la baja posición socio-económica y la vulnerabilidad, así como las dificultades enfrentadas por los inmigrantes a la hora de adaptarse al nueve ambiente urbano, podrían predisponer a una mayor mortalidad a los niños de familias inmigrantes de zonas rurales a zonas urbanas. Conclusión: Nuestros resultados subrayan la necesidad de intervenciones a nivel comunitarios y socio-económico especialmente enfocadas hacia grupos inmigrantes dentro de áreas urbanas, que mejoren su acceso a servicios sanitarios, la educación materna, así como en general a la situación socio-económica de la mujer. Palabras clave: Migración, ruptura, selectividad, adaptación, mortalidad infantil, mortalidad menores de cinco años, modelos multinivel, Nigeria. Rural–urban migration still accounts for most of the migratory movements in west Africa (Black et al. 2004). Most of the migrants typically move from hostile areas characterized by drought and low agricultural productivity in search of more fertile lands to farm or for better paying jobs in urban areas. While many of these movements are temporary or seasonal, they provide the rural poor with avenues to supplement meagre farm earnings during unproductive seasons while ensuring that they keep the rights to their land (Findley 1994). Rural–urban migration in sub-Saharan Africa, either temporary or permanent, appears to be the most significant form of movement for long-term spatial redistribution and is therefore regarded by many policy makers and governments as the overriding internal migration pattern in the region, after the exclusion of periodic and seasonal movements (Oucho & Gould 1993; Oucho 1998; Chattopadhyay et al. 2006). Migration is an important strategy for the diversification of livelihoods for many in the world’s poorest nations (Quisumbing & McNiven 2007). Several studies in developing countries have shown rural–urban child mortality differentials in favour of urban areas and indicate that survival chances of children in rural areas could be improved by urban migration (Brockerhoff 1994; Ssengonzi et al. 2002). Explanations for this proposed rural disadvantage in child mortality include the concentration of public health resources in urban areas (UNICEF 1994), low immunization coverage levels and family planning programmes in remote areas (USAID 1991) and the effect of malaria and other infectious diseases in some tropical regions (WHO 1991). Contradicting evidence suggests that urban child survival advantage over rural areas has narrowed considerably over time and that standard of living in some urban areas is even worse than in rural areas as a result of high levels of poverty in several urban populations (Brockerkhoff & Brennan 1998). Most rural–urban migrants initially settle in poor neighbourhoods, which are characterized by lack of adequate sanitation and clean water, poor housing and overcrowding, and lack of access to modern health services (Woldemicael 2000; Todaro 1996; Crompton & Savioli 1993; Aaby 1992). Some previous studies indicate that rural–urban migrants tend to be young, often students between 15 and 29 years of age, predominantly single, educated and seeking better educational opportunities (Brockerhoff 1994; Sastry 1996; Oucho 1998). This is believed to have a negative effect on the development of rural areas (Mabogunje 1988). However, other studies suggest differently stating that migration impacts positively upon sending areas (Hugo 1998), in that household members who migrate can facilitate investments in new activities by providing rural households with liquidity in the form of remittances and income security in the event of an adverse income shock (Adepoju 1974; Wouterse & Taylor 2008). Migration to urban areas results in a dramatic short-term increase in the risk of child mortality. With extended urban residence, survival chances of children tend to improve compared to children of rural non-migrants (Brockerhoff 1995) as a result of behavioural change and economic progress (Brockerhoff 1994), improved access to maternal child health services, improved housing conditions and overall levels of fertility (Tam 1994). With the shift in reliance of the Nigerian economy from agriculture to heavy dependence on crude oil as the major source of foreign revenue, the rural economy has experienced significant deterioration (Ayadi 2005). This decline triggered a steady rural–urban migration in search of emerging job opportunities in the oil and related sectors. Better social conditions and government policies favourable to cities have helped sustain rural–urban migration with resulting increased unemployment and urban poverty (USAID 2002). Existing studies on internal migration processes in Nigeria are few and limited to a few villages or medium-sized towns (Olurode 1995; Adepoju 1986). Most migrants in Nigeria (especially rural–urban) are young persons between 15 and 29 years of age and unemployed (Adepoju 1986). The objectives of this study were to (i) estimate and compare the under-five mortality rates for the different migrant groups; (ii) assess the role of rural–urban migration on under-five mortality; (iii) assess the possible mechanisms through which migration may influence under-five mortality; and to (iv) assess the individual- and community-level relationship between migration status and the risk of under-five death. Three perspectives of migration could be used to explain the differential health outcomes between migrants and non-migrants. First, migrant disruption suggests that the process of migration disrupts the natural progression of demographic events in the lives of the migrants. For instance, migration causes a break in mothers’ network of social support in terms of advice on childcare and treatment during the children’s illness, financial support and even social and cultural practices (Ssengonzi et al. 2002). Shorter birth intervals are associated with reduced child survival, and rural women are reported to have longer birth intervals than urban women (Vitzthum 2001), thus increasing the survival chances of their children (Witwer 1993). The disruption perspective therefore argues that despite availability of better health services in urban areas, children of rural–urban migrants will have lower survival chances than children of non-migrants owing to consequences of the migration itself. The variable migrant status was used to test the migrant disruption perspective in this study. Second, migrant selectivity suggests that rural–urban migration is selective for those with specific demographic and socio-economic characteristics that are favourable to child survival (e.g. education, occupation and wealth). These are also characteristics that increase their propensity to migrate (Ssengonzi et al. 2002; Stephenson et al. 2003; Amankwaa et al. 2003). Demographic and socio-economic characteristics were used to operationalize migrant selectivity perspective. Third, migrant adaptation refers to the extent to which migrants effectively use healthcare facilities and thus adapt to the new urban environment (Tam 1994). Social institutions, such as health services, family members already living in the host area and community groups are important in aiding the adaptation of migrants into the host population and hence child survival (Stephenson et al. 2003). Failure of migrants to adapt to their new urban environment may lead to their continued use of traditional rural medical services and under-utilization of modern health services (Davidson 1983; Uyanga 1983). In addition, migrants tend to be concentrated in low income and informal employment, which often prevents them from fully adapting into urban societies (Stephenson et al. 2003). These factors are believed to result in mortality differentials between migrant and non-migrant populations. Characteristics associated with health care utilization were used to operationalize migrant adaptation. Cross-sectional data from the 2003 Nigeria Demographic and Health Survey (NDHS) were used in this study. This is a nationally representative sample collected using a stratified two-stage and of the data are 2004). such as and of status and age at death the child had were collected for of these were collected by from women aged 15–49 years who a of 6029 children to the This study is on analysis of data with The was by the in the of and the of the to use the data in this study was from The variable is the risk of under-five as a child between birth and the In terms of migrant status was as urban rural non-migrant and rural–urban migrant was as a who between of rural and urban areas in the years to the The 2003 migration to of years in the of as and and of the previous and of and These were used to migration a variable that the migration into rural non-migrants and urban non-migrants. migrant children were in and were from the In this migration status was by a who of an and may those in migration within an were from the who reported previous as rural and as urban was as a rural–urban The non-migrant groups were as rural non-migrant or urban non-migrant upon their reported at the of as factors were into and socio-economic and health care utilization (i) of the with as the (ii) birth and of higher birth with birth as (iii) mothers’ age, years and years or with years as (iv) mothers’ age at birth of years or with years or as the and with as mothers’ education, no and education, with or higher as mothers’ agricultural and and with as of and with as The was using of a of household a through and the resulting were in to a standard with a of and standard of household was a for and the were for were to the of the household in which they and the sample was into and in with as the of of with as and care by with as The community-level were at the of the are areas used as for or et al. are and to be with to population characteristics, economic status and living conditions and are of or more areas which are the for which data are in Nigeria. of a of with a a had households 2004). risk factors were as (i) of residence, of of with as and (ii) mothers’ education, as the of mothers with or higher in the of of with as and (iii) of the variable with as the This variable was as the of children that the in the on by the Health and at of and at and of age, of at birth and at and of age, and of at of (iv) as the of mothers who their child in a in the of of with as low and the of the children and mothers in the sample by migration status and compared the of the different characteristics using The data were using Cox & which data as a variable can that the have a effect on The variable for this study is the time between birth and death of a child years of age or the is into no variable and to the into individual and community and to identify a possible that can be by of under-five mortality within et al. 2005). mothers’ migration status as the demographic characteristics of children and mothers of the birth mothers’ age, mothers’ age at birth of child and migration status and socio-economic education, mothers’ occupation and migration status and health care utilization in of of child and care by community-level of residence, community child community and community mothers’ that were significant in the previous used survival regression to for the of the data of 6029 children were nested within mothers who were in nested within 365 communities The of individual- and community-level characteristics, and results were as and 95% confidence intervals The was using and in & 2005). The of the study characteristics the migrant groups in the study sample is in With the of of the characteristics significantly between the migrant Rural–urban migrants were the most and urban non-migrants were the migrant to the other rural–urban migrants the of mothers years or at the birth of their child with no mothers in the poorest mothers who had during mothers who at and mothers who had care by a the pattern in which factors were in the Migration status was into the as the variable to assess the influence of migration on under-five mortality migration status into in lower risks of for children of urban non-migrant 95% and rural non-migrant 95% children compared to children of rural–urban migrant mothers. This that the process of mothers’ migration significantly the risk of under-five death for the children of rural–urban migrant mothers. for demographic characteristics change the association between the risks of and migration that the effect of migration on under-five death is of demographic The risks of were higher for 95% and children of mothers 95% compared to children of birth and children of socio-economic characteristics with migration with the risks of significant for children of rural non-migrants 95% and for children of urban non-migrants compared to children of rural–urban migrants. This suggests that in the risk of under-five death are by the between socio-economic characteristics of the migrant and non-migrant The risks of were for children of mothers with 95% and no 95% than for children of mothers with or higher characteristics associated with health care utilization and migration with the risks of lower for children of rural non-migrants 95% compared to children of rural–urban migrants and higher for children of mothers who had during 95% compared to children of mothers who that utilization of health care services the mortality between the migrant and non-migrant With the of community-level with migrant status the risks for children of rural non-migrants were lower 95% compared to children of rural–urban that this association could be by community-level characteristics in this study. The risk was lower for children of mothers in communities with high of 95% than for children of mothers in communities with the of at the The significant in previous and significantly lower risks of for children of rural non-migrant mothers 95% than for children of rural–urban migrants. The risk of was higher for mothers with no 95% and 95% compared to mothers with or higher The for between under-five mortality outcomes and factors is shown in The were significantly different communities and mothers within the different thus the use of in this The risk of under-five death was higher for children of rural–urban migrant mothers than for children of urban non-migrant mothers in the Rural–urban migrant children experienced higher risks of under-five death than urban non-migrant This could be to in socio-economic position and health care utilization between migrant groups and is in with from other studies (Brockerhoff 1994; et al. & 2008). mortality was by migrant disruption as by significantly lower risk of under-five death non-migrant children compared to rural–urban migrant children with the of migration status in disruption may have from the migration process associated with disruption of family and community and Demographic and socio-economic characteristics of migrant in and the risks of for rural non-migrant children compared to rural–urban migrant children of mothers and mothers with or no had significantly increased risks of under-five death. These demographic and socio-economic characteristics indicate the of these characteristics in the mortality between rural–urban and rural non-migrants. of the characteristics of the study sample that and more mothers and unemployed mothers who had their child at years or and in the were most to the other rural non-migrants were and more to their they had better child survival outcomes than rural–urban migrants. could be that is the of the migration process and the that rural–urban migrants are a more population that the risk of under-five death for children of rural–urban migrants compared to rural non-migrants. Health care utilization of migrant in the risks for the children of rural the risks still and some of the mortality between children of rural non-migrants and rural–urban thus a significant role in the survival of children of rural–urban migrants. during significantly increased the risks for their children the age of years compared to mothers who during Characteristics of the study sample that rural–urban migrants were compared to rural non-migrants and urban non-migrants in terms of during care by and of their Migration between different with in the availability of health services, conditions and social may be associated with such as the continued use of traditional rural medical This with the socio-economic selectivity may to in mortality between migrant and rural non-migrant of such as population and high population household and poor social in urban areas of countries such as Nigeria may predispose to such mortality The of community-level factors in the risks for children of rural non-migrants who had lower risks of under-five death compared to children of rural–urban migrants. in a community with high of mothers who had was associated with significantly lower risks of under-five death compared to living in a community with and the This that community-level factors were important of under-five mortality. In a general a of with significant on under-five mortality from previous the risk of under-five death was lower for urban non-migrant children than for rural–urban migrant The risks of under-five death were higher for children of mothers and mothers with or no no significant in survival could be between children of rural–urban migrants and urban non-migrants after for the characteristics in The pattern of in this study the of community in under-five mortality and more to the community in under-five mortality are of characteristics of the who in these communities in or of factors that to communities in The at the community in was the of on under-five mortality between This is by the significant and lower risks of under-five death associated with living in communities with a high of thus to of of the study include (i) other factors in the study may also be of under-five mortality migrant and non-migrant groups; (ii) the used as a for in this study may of into of and the of (iii) other community to the risks of under-five death may have or are such as to immunization and of immunization (iv) data have the social of to the extent of mothers’ in the community they and data on household income or which are the used to are collected in the The used is a for household economic status and may results to those from of income and such data are or can be collected & This study had First, the use of to test the perspectives of migration on the risks of under-five death; the that are nationally representative and for the of the results the 2006). are countries and results are therefore countries 2006). Rural–urban migration is an important of under-five mortality. Our suggest that the traditional migration to in under-five mortality between migrant and non-migrant The disruption of family and community from the migration the selectivity and low socio-economic of migrants and in adapting into the new urban environment predispose the children of rural–urban migrants to higher mortality. These results the need to migrant groups within urban areas in the of increasing and health care services, maternal and the general socio-economic situation of women. are to for the data used in this study.