Abstracts - health and mortality strand

Strand organisers: Dr. Amos Channon, University of Southampton; Dr. Paula Griffiths, Loughborough University

Abstracts are listed in the order they are scheduled for presentation. Please refer to the programme for timings.

Closing the gap on Indigenous life expectancies: What if we succeed?
Andrew Taylor, The Northern Institute, Charles Darwin University; Tony Barnes, Northern Territory Treasury, Darwin

In 2007-2008 the Council of Australian Governments agreed to six ambitious targets for addressing long-standing disparities between Indigenous and other Australians in health, education and employment outcomes. The 'National Integrated Strategy for Closing the Gap on Indigenous Disadvantage' (colloquially 'Closing the Gap') includes the goal of eliminating life expectancy gaps within a generation. This policy says nothing about what changes in the demographic structure of the population might be expected should there be even partial success towards this ambitious target. Information is required to analyse shifts in age-sex burdens for their effects on health, education and other service provision areas. In this study we use cohort component projections modelling to assess the demographic impacts of closing the gap in a generation for the Northern Territory, the Australian jurisdiction with the highest Indigenous composition and largest life expectancy gaps. Modelling is based on three scenarios: 1) No changes to Indigenous life expectancies from those estimated in 2010; 2) Success in closing the gap within a generation; and 3) A continuation of current expectations about Indigenous life expectancy changes. Although closing the gap would only produce small increases in the size of the next generation's Indigenous population, over and above those increase projected with lower life expectancy scenarios, it will result in substantial changes to the age composition of the population and produce associated shifts in various demographic indices pertinent to planning and the funding of core services.

Email: Andrew Taylor: andrew.taylor@cdu.edu.au

Prevalence of life-limiting and life-threatening illness in children and young people in England: time trends by area type
Paul Norman, Centre for Spatial Analysis & Policy, School of Geography, University of Leeds and Lorna Fraser, Paediatric Epidemiology Group, Division of Epidemiology, University of Leeds

An independent review of children's palliative care services in England highlighted the lack of available data on the number of children and young people with 'life limiting conditions' who received or who would benefit from palliative care services. Recent research determined that the prevalence of life-limiting conditions in children and young people in England, at 32 per 10,000 population, was double the previously reported estimates. In this paper we build on the previous research by analysing 2000-01 to 2009-10 time-trends in the prevalence of life limiting conditions by small area deprivation and by geodemographic area types. We find that the more deprived locations and areas classified as 'Multicultural City Life' and 'Disadvantaged Urban Communities' have significantly higher prevalence rates of children with life limiting conditions than other area types. Whilst rates are highest for children age less than one year of age and rates then reduce with age, the time trends show no increase over time for the youngest age group but significant increases in prevalence during the decade for older age-groups. We conclude that the increase in prevalence rates for children aged between 1 and 19 is because survival is improving rather than due to increases in incidence. This is likely to be due to the quality of care received in Children's hospices but some of the geographical inequalities in prevalence may be due to differences in accessibility.

Email: Dr. Paul Norman: p.d.norman@leeds.ac.uk

Inequalites in disability-free life expectancy and healthy life expectancy between ethnic groups in England and Wales
Pia Wohland, Institute for Ageing & Health, University of Newcastle; Carol Jagger, Institute for Ageing & Health, University of Newcastle Phil Rees, School of Geography, University of Leeds;

Health expectancy measures are essential to quantify and monitor a population's health and wellbeing. The proportion of persons from different ethnic backgrounds have increased considerably over the last two decades in England and Wales (E&W), but to date little is known about health expectancies of these groups. Mortality estimates, an indirect health indicator, show variations across ethnic groups in E&W with a gap in life expectancy (LE) at birth of about 4 years for men as well as women across ethnic groups. In this study we estimate two health measures, disability free life expectancy (DFLE) and healthy life expectancy (HLE), for ethnic groups in E&W for 2001 with the Sullivan method combining ethnic group mortality estimates (Rees et al, 2009) and self reported limiting long term illness and health information from the 2001 Census. The focus of this study is to investigate how DFLE and HLE vary across ethnic groups at different ages in E&W and to see to what extent the observed variations resemble those observed in mortality. First results show, that the gap in DFLE is more pronounced than the one found in LE. DFLE at birth varies by about 10 years for females across ethnic groups. The absolute differences decrease but a gap persists in all ages. On the other hand, ranks of ethnic group DFLE and LE are similar, for females, Pakistani and Bangladeshi women have the lowest DFLE and Chinese and the Other White women the highest.

Email: Dr. Pia Wohland: pia.wohland@ncl.ac.uk

The Socio-economic factors affecting over-nutrition of Thai adults
Chayada Bhadrakom, Bhavani Shankar, C S Srinivasan, Nipa Rojroongwasinkul, University of Reading

Changes in socio-economic structure and dietary pattern have conduced to prevalence of overnutrition in Thailand. Since overnutrition is directly related to obesity, determining the factors driving this prevalence is necessary for health public policy to promote healthier consumption behaviours. However, the risk of inadequate or excess diet mostly occurs at the tails of nutrient intake distribution rather than at average; thus, this study goes beyond the previous studies conducted in Thailand by estimating a richer quantile regression model that allows the effects of socio-economic factors to vary across the intake distribution. The database from Thai Food Consumption Survey in 2004 with sample of 5,264 Thai adults aged between 18 and 65 years was employed. The quantile regressions present that despite the growth of economic shifting the structure of nutrient diet in Thailand, it is no longer the prior factor creating overnutrition. Lower energy but more fat dense diet associated to females than males, urban than rural and higher socio-economic status than lower socio-economic status implies a need for attention from policy makers. Furthermore, non-impact of smoking on fat density intake at upper tail indicates that smoking cessation would not conflict with the reduction in fat intake policy for over consuming people.

Email: Chayada Bhadrakom: C.Bhadrakom@pgr.reading.ac.uk

Residential mobility, socioeconomic context and body mass index in a cohort of urban South African adolescents
Carren Ginsburg, Paula Griffiths, Linda M. Richter, Shane A. Norris, Loughborough University

Adolescents who are changing residence, as well as their social and economic circumstances are likely to experience lifestyle changes that place them at greater risk of obesity. This paper uses data from Birth to Twenty, a birth cohort of South African urban children, to determine the relationship between residential mobility and body mass index (BMI) amongst Black adolescents aged 15 (n = 1613), and to examine the role of changes in household socioeconomic status (SES). The prevalence of overweight and obesity in the sample was 25% in females and 8% in males. Amongst the females, a strong positive association between residential mobility and BMI was observed for those who also experienced an increase in household SES between birth and 15 years (beta= 0.42, SE = 0.13), while no effect was identified for males. The study shows the potential for environmental change and increased resources to influence the risk for obesity. It also highlights the value in considering the range of social environmental factors and changes across the early life course that might play a part in evolving nutritional patterns in urban transitioning environments.

Email: Dr. Paula Griffiths: p.griffiths@lboro.ac.uk

Being Stunted-Overweight - A Multilevel Analysis of the Determinants of Under-Fives Nutritional Status in Albania 2008-09
Katie Bates, London School of Economics

Background: Albania has been shown to be experiencing one of the highest prevalence of stunted overweight under-fives (children that are concurrently both stunted and overweight) worldwide at 9.89% of all under-fives. To the author's knowledge, the determinants of this nutritional status from 'within a population perspective' have not been addressed. Methods: Cross-sectional data from the Albanian Demographic and Health Survey 2008-09 providing anthropometric, socioeconomic, dietary, maternal and health data was use to explore the determinants of being stunted-overweight from a multilevel perspective. Multinomial modelling for four nutritional outcomes (not malnourished, stunted, overweight and stunted-overweight) incorporating a random intercept for mother, household and community fitted. In addition, context effects were tested utilising random coefficients. Results (Preliminary): At this stage, the results show previous internal migratory trajectories of the mother, which include residency in urban areas, are associated with increased odds of an individual being stunted-overweight compared to all other nutritional statuses. In addition, a socioeconomic gradient in nutritional status has been shown - with children from low socioeconomic background most likely to be stunted, followed by stunted-overweight, obese and not malnourished as socioeconomic conditions improve. Only a random intercept for community has been shown to be significant, with community accounting for 42.15% of the variance. *Random coefficients modelling yet to be conducted.

Email: Katie Bates: k.m.bates@lse.ac.uk

Participation in health services, complementary feeding and children\'s nutritional status in Malawi
Lana Chikhungu, Sabu Padmadas, Nyovani Madise, University of Southampton

Malawi just as many other developing countries, continues to be challenged by the problem of child under-nutrition. Recent estimates indicate that in Malawi, 47% of children under the age of five are stunted whilst 13% are underweight (ORC Macro and NSO, 2011). Both the stunting and underweight rates are high considering that the expected level of under-nutrition in a well-nourished population is about 2% (ORC Macro, 2006). The quantity and quality of nutrients a child takes and whether they are ill or not are directly associated with a child's nutritional status. National level efforts to tackle the problem of child under-nutrition in Malawi include implementation of food security, nutritional and health education programmes but also intensification of immunisation and vitamin A supplementation programmes to improve children\'s immune status for better health. Poor feeding practices such as early introduction of complementary foods, not practising exclusive breastfeeding and child illnesses have been reported as some of the contributing factors to poor child nutritional status in Malawi by previous studies (Madise and Mpoma, 1997). Most of the previous studies in Malawi and other developing countries also report a sharp rise of child under-nutrition during the weaning (age 7 to 18 months) the period when children get introduced to new foods and are more prone to illness as they become more mobile and exploratory. Using the Malawi Demographic and Household Survey of 2004, this study looks at how participation in health services as evidenced by mother's possession of a child health card, child's immunisation and receipt of vitamin A supplements is associated with a child's nutritional status amongst all children as well as children of different age groups but also how different kinds of foods consumed during the weaning time are associated with a child's nutritional status. The multivariate analysis shows that consumption of animal sources of food such as meat, fish, chicken and poultry and vitamin A rich foods such as mangoes and papaya during the weaning time and the mother's possession of a child health card are associated with a lower likelihood of the child being underweight. Children who consume higher quantities of a variety of foods are less likely to be underweight and that there is no significant association between consumption of local grain or legumes and a child's nutritional status. The chi- square bivariate analysis results show that there is a lower likelihood of being underweight amongst children that were fully immunised compared to those that were partially immunised as well as amongst children that received vitamin A supplements compared to those that did not for children aged between 7 to 18 months.

Email: Lana Chikhungu: lcc1g08@soton.ac.uk

Gender differences in anaemia, stunting and thinness among school-aged children and adolescents in 11 low-income countries
Emily Rousham, Loughborough University; Natalie Roschnik, Save the Children (USA); Andrew Hall, Save the Children (UK)

Background: Iron deficiency anaemia is a significant global public health problem, with young children and women of childbearing age being seen as the most at-risk groups. Few studies have examined the association between undernutrition and anaemia in adolescent males in socioeconomically disadvantaged populations. This paper examines the relationship between undernutrition (stunting and thinness) and anaemia among school age boys and girls. Methods: Anthropometric and haemoglobin data from samples of school children (aged 5-17 years) from 10 countries within Save the Children School (USA) Health and Nutrition programmes were analysed (n=19,722). Results: Among all age groups, stunted children and adolescents (height-for-age < -2 z scores of reference median) were significantly more likely to have anaemia than their non-stunted counterparts. The prevalence of anaemia was 42.7% and 27.9% for stunted and not stunted boys respectively. In girls the prevalences were 36.5% versus 27.3% respectively. Stunted adolescent boys had a higher prevalence of anaemia than stunted girls. Despite a high prevalence of thinness (BMI-for-age < -2SD of the WHO reference median) in many countries, there was no association between thinness and anaemia at any age. Summary and conclusions: These data reveal age-related changes in the prevalence of anaemia in stunted boys and girls, with stunted adolescent boys being at greater risk of anaemia than stunted adolescent girls. Factors contributing to this pattern of anaemia and undernutrition will be explored. In these children and adolescents with a high prevalence of undernutrition, stunting is a risk factor for anaemia, but thinness is not.

Email: Dr. Emily Rousham: e.k.rousham@lboro.ac.uk

Socioeconomic and intergenerational factors impacting the health of Maya families in Mexico
Maria Ines Varela-Silva, Hugo Azcorra, Hannah Wilson, Federico Dickinson, Paula Griffiths, Barry Bogin, Loughborough University

The overall aim of our project is to identify factors that impact on the health of the Maya families in Merida, Yucatan, Mexico. The ultimate goal is to use our findings to implement intervention programmes among the Maya communities in order to improve their health and well-being. Our research has shown that the Maya in the Yucatan are poorer, less healthy, less educated, with very short stature (stunting), and with fewer opportunities to improve their lives than non-Maya groups. The Maya are undergoing nutritional and epidemiological transitions. The nutritional transition is leading them to shift from a traditional diet with more fibre, less fat and less calories to a globalised diet, rich in calories and fat content and poor in fibre. Consequently they are becoming very overweight. The epidemiological transition means that the Maya are suffering less from infectious diseases but more from non-communicable diseases such as diabetes and heart disease. In turn, the epidemiological transition is aggravated by their overweight status. Short stature associated with overweight is the worst possible health outcome. When the two outcomes (very short stature and overweight) coexist in the same group, same family or same individual we say that a situation of nutritional dual-burden has occurred. Our results show that 81% of the maternal grandmothers and 58,75% of the mothers are dual-burden individuals. More than 30% of the children are overweight or obese. In conclusion, there is an alarming rate of nutritional dual-burden heavily impacting the quality of life of the Mexican Maya families.

Email: Dr. Ines Varela-Silva: m.i.o.varela-silva@lboro.ac.uk

Modal age at death: mortality trends in England and Wales, 1841-2010
Emily Clay, Office for National Statistics

Measuring mortality and life expectancy has been and remains very important for our understanding of longevity, ageing societies and mortality improvements of human populations. Life expectancy, the mean age at death according to a hypothetical population that is subject to a particular mortality experience, has been calculated through life tables as far back as 1841 in England and Wales and remains the leading indicator of population mortality improvements today. For low mortality countries such as England and Wales, which also see large percentages of people survive to older ages, the late modal age at death may be a better measure of life duration. Modal age at death is more sensitive to changes in elderly mortality; variation can also be captured through the standard deviation of ages at death around the mode. This research uses smoothed mortality rates for England and Wales, using the non-parametric method of 2-Dimensional P-splines, to construct life tables in order to obtain modal age at death figures from 1841-2010. This allows the analysis of old age mortality compression and the more recent phenomena of the shift in the survival curve or age pattern of mortality to the right. This presentation will focus on the changes over time of modal age at death and standard deviations. The results of this research will add to the growing body of literature around the modal age at death and it will also help to inform mortality assumptions used in the ONS National Population Projections.

Email: Emily Clay: Emily.clay@ons.gsi.gov.uk

Socio-demographic factors associated with mortality among older adults between waves: an analysis based on SHARE data
Georgia Verropoulou, University of Piraeus

Aims: The study aims at exploring socio-demographic factors related to mortality between the first and the second waves of the Survey of Health, Ageing and Retirement in Europe (SHARE). Data and methods: The data used in the analysis come from waves 1 and 2 of SHARE. Wave 1 of the survey was carried out in 2004 in 11 European countries representing Scandinavia, Central Europe and the Mediterranean; the target population was persons aged 50 or higher. The second wave of the survey was conducted over 2006-2007. The sample used in the analysis comprises 20,547 persons, 19,919 of which were successfully re-interviewed in wave 2 while 628 or 3.1% died between the waves. The importance of factors associated with mortality between the waves was assessed using logistic regression. Results: The findings indicate that older age at baseline (wave 1) is related to higher chances of dying between the waves. Regarding associations with the demographic characteristics of the respondents, female gender has a significant and protective effect (halves the chances of death between the waves) and the same holds for marriage. In fact, single persons run a 50% higher risk of dying compared to married while chances of death for separated, divorced and widowed persons are higher by 25%. With respect to the socio-economic characteristics of the respondents, more years in education have a significant protective effect and the same holds for a higher level of net wealth. Level of income is not significant once net wealth has been taken into account. NS-SEC based on occupation is not significant once education has been included in the predictors. Risky health behaviours such as smoking,and low levels of physical activity have a strong negative effect.

Email: Dr. Georgia Verropoulou: gverrop@unipi.gr

Are health inequalities evident at all ages?
Paul Norman, School of Geography, University of Leeds; Paul Boyle, Longitudinal Studies Centre – Scotland (LSCS), School of Geography & Geosciences, University of St Andrews

The notion that mortality inequalities across area deprivation vary by age is logical since not all causes of death rates increase with age and not all causes of death are related to the gradient of deprivation. Dibben and Popham (2012) investigate these phenomena and find that heightened exposure to the risk of land transport accidents increases levels of mortality in the least deprived areas such that inequalities disappear at this age. In addition to the interaction between the cause–age and cause–deprivation relationships, population migration may redistribute the population such that the health–deprivation relationship varies by age. This proposition is based on
the distinctive age schedule of migration, the types of areas people typically move from and to at different ages and that the migration process itself is health selective. This paper will demonstrate how population movements between differently deprived areas mean that inequalities in health vary by age. In addition, this paper will demonstrate that restricting the age ranges for summary measures such as standardised mortality and illness ratios may enable poor health areas to be more readily highlighted.

Email: Dr. Paul Norman: p.d.norman@leeds.ac.uk 

Reserve capacity, childhood growth, age at reproduction, and physical performance in Bangladeshi women.
B. Bogin, Loughborough University; D. Harper, Loughborough University; J. Merrell, Swansea University; J. Chowdhury, Swansea University; B. Molik, University of Bristol; M. Heinrich, University of London; V. Garaj, Brunel University; P. Meier, University of Sheffield; J.L. Thompson, University of Bristol

Purpose: Using the evolutionary model of Life History Theory we analyse trade-offs between childhood growth, age at reproduction, physical function, and aging in Bangladeshi women living in the UK. Healthy childhood growth allows for the development of adult reserve capacity, which slows the rate of aging and promotes greater longevity. Materials & Methods: This presentation is based on our research with Bangladeshi women (mothers & daughters) in Cardiff, UK and in Bangladesh (see www.Projectmina.org ). As an ethnic minority, many of the Bangladeshi community in the United Kingdom (UK) are socially disadvantaged and suffer from high levels of disability, disease and rapid aging. Little is known about the causes. Results: UK-living women have higher rates of overweight and obesity than the Bangladeshi- living women. All women are of short stature, (X=150.5 cm, sd= 6.4), but daughters are significantly taller than their mothers. UK-born daughters are taller than Bangladeshi-born daughters who migrated to the UK and most of the difference is due to greater knee height - a biomarker of better health in infancy and childhood. A 7-component lower body physical function test finds that Bangladeshi mothers have significantly better total scores than UK mothers.
Discussion: Greater knee height, a greater age at first birth, and a greater time lag between marriage and first birth predict slower rates of physical decline in all women.

Funded by the New Dynamics of Ageing Programme, Economic and Social Research Council.

Email: Professor Barry Bogin: b.a.bogin@lboro.ac.uk

Family Structure and Child Health: (Re-)Framing the Question
Wendy Sigle-Rushton, London School of Economics

It is well established that children who grow up with married biological parents have, on average, better outcomes than children who experience other family structures.  It  is less clear what these associations mean and how they should be interpreted. Does the association between family structure and well-being represents something "real" or is it merely spurious? Although what exactly the causal effects of family structure are, particularly in the production of child health, remains poorly specified, economists have made significant contributions to debates about the importance of unobserved heterogeneity as a potentially important source of bias. Their work has outlined important concerns and emphasized the need for careful interpretation of the data.  Nonetheless, the influence of economists and economic modelling may have diverted attention from other equally relevant questions about how we should understand the crude relationship between family structure and child health.  Borrowing concepts from discursive analysis and the feminist literature on intersectionality, this paper questions whether empirical research on the effects of family structure has been  too fixated on issues of causality and selection. Empirical work that seeks simply to determine whether any statistically significant association can be “written off” as a spurious relationship may be framed in too narrow and too limiting a way.  When we are so concerned with "whether" family structure is important that we stop asking "why", possibilities for change are obscured.

Email: Dr. Wendy Sigle-Rushton: W.Sigle-Rushton@lse.ac.uk

Measuring inequalities in levels of emergency hospital admissions between ethnic groups in England
Allan Baker, Justine Fitzpatrick, Ed Klodawski, London Health Observatory; Paul Fryers, David Jephson, Heather Heard, East Midlands Public Health Observatory

Background: higher percentages of emergency hospital admissions may reflect some patients not receiving the care most suited to managing their conditions. As emergency admissions cannot be predicted, they can also impact on other hospital care and carry a high cost.
Objective: to measure whether there are inequalities in levels of emergency hospital admissions between ethnic groups in England. Participants: patients admitted to hospital between April 2010 and March 2011 using data from Hospital Episode Statistics, the statistical database of hospital patient care provided by the NHS in England.
Main outcome measure: emergency admissions as a percentage of total hospital admissions. Data have been indirectly age standardised to take account of differences in the age structures of populations, between areas and between ethnic groups. Results: results have been calculated for five ethnic groups (White, Mixed, Asian, Black and Chinese) as well as for all ethnic groups combined and those whose ethnicity is not known / not stated. Results have been calculated for ethnic groups within England as a whole and within every English local authority. As these figures will be released as Official Statistics, they are restricted until early July 2012.

Email: Allan Baker: allan.baker@lho.nhs.uk

Towards the demography of ill-health: comparing the geographical distributions of mortality and health in Ghana
Allan G. Hill, University of Southampton and Harvard School of Public Health and John R. Weeks, San Diego State University

It is becoming increasingly clear that the demography of health (or ill-health) is very different from the demography of mortality (or premature death) (Riley 1997; Harris 1999; Harris 1999). We are at the beginning of the explorations of the connections between the two in both contemporary and historical societies (Riley 1999; Riley 1999). One way in which such linkages can be explored is through the study of trends and differentials in mortality and morbidity in sub-populations located in particular geographical areas (Weeks, Hill et al. 2006). In a broad way, the restriction of the analysis to specific geographic locations allows us to control the sets of (probably separate) risk factors which predispose people to illness or to premature death. Here we make use of a nationally representative dataset with precise geographical codes to compare levels of mortality and morbidity in Ghana. The data were collected under the WHO World Health Survey programme in 2003 and have the advantage of including a wide range of health measures for children and adults, both self-reported and objectively assessed, in addition to a full both history and information about recent deaths in the household. First, using the data from the 2000 census, we allocate the survey data points to regions and districts with common socio-economic and other characteristics. For these new geographical units, we then calculate childhood mortality rates from the full birth histories and adult death rates from the survey data on household deaths. Second, we explore the health data using the standardised instruments embedded in the questionnaire, such as SF 36, and briefly examine the association between the scores and other measures of health (Ware and Sherbourne 1992). Next, we examine the spatial distributions of infant and adult mortality by the statistical districts we have created and compare this with the distribution of two domains of self-reported health (physical and mental) derived from the same source and for the same districts. This analysis allows us to comment on the differing distributions of premature death and current morbidity for the whole population and within the statistical districts we have identified. This work at the national level is a precursor to some more detailed geographical analysis at a smaller scale within the Region of Accra.

Email: Professor Allan Hill: ah4e10@soton.ac.uk

How structural determinants influence intermediary determinants of early childhood health in Colombia: exploring the role of communities
Ana Maria Osorio, University of Barcelona; Catalina Bolancé, Universityy of Barcelona; Nyovani Madise, University of Southampton

This study examines how structural determinants influence intermediary factors of child health inequities and how they operate through the communities where children live. In particular, we explore individual, family and community level characteristics associated with a composite indicator that quantitatively measures intermediary determinants of early childhood health in Colombia. We use data from the 2010 Colombian Demographic and Health Survey (DHS). Adopting the conceptual framework of the Commission on Social Determinants of Health (CSDH), three dimensions related to child health are represented in the index: behavioural factors, psychosocial factors and the health system. In order to generate the weight of the variables and take into account the discrete nature of the data, principal component analysis (PCA) using polychoric correlations are employed in the index construction. Weighted multilevel models are used to examine community effects. The results show that the effect of a household's SES is attenuated when community characteristics are included, indicating the importance that the level of community development may have in mediating individual and family characteristics. The findings indicate that there is a significant variance in intermediary determinants of child health between-community, especially for those determinants linked to the health system, even after controlling for individual, family and community characteristics. These results likely reflect that whilst the community context can exert a greater influence on intermediary factors linked directly to health, in the case of psychosocial factors and the parent's behaviours, the family context can be more important. This underlines the importance of distinguishing between community and family intervention programmes.

Email: Ana Maria Osorio: anaosorio@ub.edu

Understanding the link between HIV infection and poverty in Kenya: Evidence from the Demographic and Health Surveys
Monica A. Magadi, Department of Sociology, University of Hull

The link between HIV infection and poverty has attracted considerable research attention in recent years, but the relationship is rather complex and findings from existing studies remain inconclusive. While some argue that poverty increases vulnerability, existing empirical evidence from sub-Saharan Africa largely support the view that wealthier men and women have higher prevalence of HIV. Results of a recent cross-national analysis of poverty and the risk of HIV infection in 20 countries of sub-Saharan Africa revealed that the urban poor in SSA have significantly higher odds of HIV infection than their urban non-poor counterparts, despite poverty being associated with a significantly lower risk among rural residents. Furthermore, the gender disparity in HIV infection (i.e. the disproportionate higher risk among women) was amplified among the urban poor. In this paper, we focus on the association between HIV infection and poverty in Kenya, paying particular attention to gender disparities and differences in risk factors of HIV infection by urban/rural residence. The study is based on secondary analysis of data from the Kenya Demographic and Health Surveys conducted in 2003 and 2008. We apply multilevel logistic regression models, allowing the poverty risk factor to vary across communities (i.e. clusters) in Kenya. A set of factors that may explain the apparent link between poverty and HIV infections, including sexual behaviour factors, are introduced in the model in successive stages. Preliminary analyses suggest a positive gradient between wealth and the risk of HIV infection for both males and females, resident in urban and rural areas. Further analysis will explore the role of a set of factors (including sexual behaviour factors) in explaining the apparent link between poverty and HIV infections in Kenya.

Email: Professor Monica Magadi: m.magadi@hull.ac.uk