Health and mortality strand abstracts

Predictors of overweight and obesity in children and adolescents: Monday 9 September, 1.30pm

Understanding ethnic differences in BMI in the US and UK
Afshin Zilanawala, Yvonne Kelly, Amanda Sacker, University College London
Health disparities are an important indicator of inequality in society. In the UK and US, racial and ethnic health disparities have been documented. Research has focused on adults and less is known about disparities among children. Early childhood health is an important predictor of adult health and wellbeing. Previous work suggests observed disparities are explained by socioeconomic, cultural, and behavioural factors. Little research examines racial and ethnic disparities in childhood health from a comparative perspective. Comparative analyses offer the opportunity to take into account cultural and societal factors in explaining ethnic/racial differences. Our study investigates ethnic inequalities in adiposity, as measured by BMI and obesity/overweight, among 5 year olds. We examine socioeconomic position, markers of cultural tradition, and family routine characteristics to understand ethnic inequalities in adiposity. Analyses use two comparable nationally representative birth cohort studies from the US and UK. We use a detailed ethnic classification and a rich set of explanatory factors. In the UK sample, we find Indian and Pakistani children have lower BMI and Black Caribbean children have higher BMI compared to White children. Bangladeshi and Black African children were more likely to be obese (as compared to normal weight), but these disadvantages were mostly due to cultural and socioeconomic factors. Unexplained ethnic differences were seen among Black Caribbeans. In the US sample, Mexican children’s higher BMI is explained by cultural factors, and the American Indian disadvantage for BMI was not explained by explanatory factors.

Family structure, maternal nativity, and childhood obesity: Evidence for the United Kingdom
Wendy Sigle-Rushton, Alice Goisis, London School of Economics

Using data from the British Millennium Cohort Study (MCS), we examine the risk of overweight at age 5, exploring whether and how it varies by family structure and maternal nativity. Because the resources available to different family types, and the form and function of the family, are likely to vary across different nativity groups, a closer examination of heterogeneous effects may shed light on the underlying processes that drive gaps in well-being by family structure and parental marital status. Although within all maternal nativity groups children of married mothers have better health outcomes, our findings suggest that predominant theoretical explanations which link marriage to greater access to resources and better health behaviours may be less relevant for the children of foreign-born and to some extent second-generation mothers.

Community socio-economic status influences on dietary intake in South African adolescents living in an urban area
R. Pradeilles1 ; P. Griffiths1,2 ; E. Rousham1; S. Norris2  ; A. Feele2; 1Centre for Global Health and Human Development, SSEHS, Loughborough University; MRC/Wits Developmental Pathways for Health Research Unit, Department of Paediatrics, School of Clinical Medicine, The University of the Witwatersrand

Background and objectives
The influence of the community environment on adolescent dietary intakes has not yet been investigated in low- and middle-income countries. This work investigates associations between community and household socio-economic status (SES), gender, ethnicity and diet.
Analyses of data at 17-19 years from the 1990 born Johannesburg-Soweto Birth to Twenty cohort (n=631) were conducted. Associations between gender, ethnicity, caregiver’s education, community SES indices, and dietary intake were examined using linear regression whilst controlling for potentially confounding factors. Community SES indices were created using principal components analysis applied to proxy indicators of the community environment. Dietary intake was assessed using a locally developed Food Frequency Questionnaire.
Adjusted linear regression results showed there was a significant relationship between gender and fat intake (+2.3% of total energy for girls vs. boys; p<0.0001). The difference in mean energy intake by ethnicity was significant in girls (-0.5 z-scores for mixed ancestry vs. blacks; p<0.0001).
Mean energy intake was higher in girls living with a caregiver who had lower education (+0.39 z-scores for the non-educated or primary school category vs. higher education; p=0.023 and +0.41 z-scores for the secondary school category vs. higher education; p=0.002).
Mean carbohydrates intake in girls was higher among those reporting a high need for neighbourhood services (+2.17%; p=0.012).
Mean energy intake in girls was higher among those reporting higher problems in their neighbourhood (+0.3 z-scores; p=0.002).
Findings of this analysis show that gender, ethnicity, caregiver’s education and community SES factors have an influence on diet in this cohort of adolescents in Johannesburg.

Socio-demographic predictors of health and mortality: Tuesday 10 September, 4.45pm

Do short birth intervals have long-term implications for parental health? An investigation using Norwegian complete cohort register data
Oystein Kravdal, University of Oslo Emily Grundy, University of Cambridge

Short inter pregnancy intervals are known to have adverse effects on perinatal incomes and child health, especially in poorer countries. Inter pregnancy or birth intervals have also been associated with maternal outcomes in both poorer and richer societies, although results have been disputed. Apart from possible maternal depletion, having two closely spaced children may lead to stresses on both parents which are manifested in later poorer health. We investigate associations between birth intervals and later use of prescribed medication and mortality risks using discrete time hazard modelling of complete cohort Norwegian register data. Results show significant adverse effects on both outcomes of birth intervals of less than 18 months, although effects are relatively small.

Intra-migration and health in England and Wales: Does where you move to impact upon your health?
Mark A Green, University of Sheffield

A small sub-set of geographers have been interested in the role that migration plays in influencing health other than for international movements. Despite the success of the ‘Moving to Opportunity’ scheme in the USA, very few studies have focused on how neighbourhood effects and conditions independently affect this relationship. Of the research conducted in Britain, most focuses on an aspatial approach. However this is problematic, since they assume that this experienced effect (i.e. migration) is consistent. Furthermore, given that movers and non-movers vary in their characteristics, selection bias is usually unaccounted for when tested such effects. As a result, analyses violate their model assumptions meaning findings may be false, a product of model conditions. This presentation seeks to evaluate whether there is an additional observed impact on an individual’s health associated with migration within England and Wales. A new and innovative methodology is applied to this theme to allow us to gain a more accurate analysis of patterns. This surrounds a recently developed matching method (Coarsened Exact Matching), which manipulates a dataset to allow for more accurate parametric testing. Matching methods are an under-utilised research method, despite their ability to transform analytical design towards a pseudo-experimental approach. Results from a larger study showing the importance of accounting for origin and location in movements are presented to further advocate the implementation of this research methodology/design as a means of improving spatial and analytical research in Geography.

The relationship between age, period and cohort trends in mortality and self-assessed health in Britain from 1974 to 2006
Michael Murphy, London School of Economics

Over the past 30 years, mortality has improved substantially in Britain. However, a range of health indicators, especially self-reported measures, suggested that health improvements did not match those of mortality or even deteriorated over this period. Trends in mortality and morbidity might be expected to be causally related to both current (period) and earlier (cohort) experiences, but effects such as selection and scarring might be expected to work in opposite directions. Thus the relationship between age, period, and cohort morbidity and mortality at the population level remains an empirical question and this paper assesses these relationships in Britain over the past 30 years, using a range of different self-reported morbidity indicators in order to elucidate possible joint future patterns of mortality and morbidity. The morbidity data used are from the General Household Survey (GHS) time series file, interviewing about 20 thousand adults aged 16 and over per annum, The total sample size is 671,773 including information on: 1. long-standing (chronic) illness 2. limiting long-standing illness 3. general health 4. current illness 5. acute illness 6. smoking The mortality data covering the same period are taken from the Human Mortality Database (HMD). I fit a series of age, period and cohort (APC) epidemiologically-informed models using Poisson & log-binomial models to compare period & cohort trends. I discuss the results and methodological issues including robustness of models and the implications of these findings for future patterns of morbidity.

Identifying housing transitions in relation to subsequent mortality at older ages
Maria Evandrou, Jane Falkingham, James Robards, Athina Vlachantoni, EPSRC Care Life Cycle, University of Southampton

While transitions into residential care and informal caring arrangements within the household have been researched in numerous cases, there has been less consideration of such transitions in the context of subsequent mortality. Analyses have generally either considered longer-term mortality transitions over ten years or more using longitudinal data or short-term transitions in relation to palliative care. Neither approach allows for the estimation of the relationship between housing pathways and risk of mortality depending on the housing pathway. Therefore, the present study uses waves 1-18 of the British Household Panel Survey (BHPS) to identify housing transitions over a period of three years and their association with subsequent mortality. BHPS members aged 65 years and over and resident at two consecutive waves were selected and followed through their housing pathways. Mortality after the three waves of observation was estimated according to the housing pathways observed and also controlling for a range of covariates which include age, sex, marital status, health status and socioeconomic measures. Results show that transitions to residential housing within 1 year of the final wave of observation before dying are the strongest predictor of subsequent mortality. Excess male mortality across all housing transitions was evident, as was higher mortality among the never married group and among those reporting poor health at the baseline. A finer-grained set of mortality transition estimates are presented than possible using other longitudinal data which has considered longer time intervals between transitions.

Health and mortality in low and middle-income countries: Wednesday 11 September, 9.00am

Intra-partum care among the urban poor in India: analysis of temporal dynamics in decision-making
Eleri Jones, London School of Economics
Aim: To understand and explain the temporal process and dynamics of decision-making on intra-partum care for first births among the urban poor in India.
Methods: The study uses a qualitative longitudinal research design. Two stages of semi-structured in-depth interviews were conducted in January-July 2012 with primiparous women, as well as their husbands, mothers-in-law and/or mothers (depending on household composition around the time of the birth). The first interview was conducted in the third trimester of pregnancy and the second within the first 8 weeks postpartum. 16 households were sampled purposively from 12 slums in the city of Indore, Madhya Pradesh. 75 interviews, background information and field notes were coded and analysed thematically and longitudinally.
Results: Households identify and register at a facility as a priority, but make financial and transport arrangements in the final stages. Where intentions for home delivery are influenced by strong norms and/or lack of experience with the health system in general, planning and preparation is limited. Even where planning for intrapartum care is explicit and begins early in pregnancy, uncertainty is a feature. Anxiety; discontent following antenatal visits; and mixed messages from members of the social network and health providers all contribute to this uncertainty regarding health care provider. Intentions may change during pregnancy and outcomes do not always match intentions, regardless of preparedness, due to multiple influences at the individual, household and health system levels.
Conclusion: There is a need to reconceptualise decision-making and birth preparedness as temporal processes with non-linear courses of action.

Regional and socioeconomic differentials of mortality risk at young ages in Brazil. Mario F. G. Monteiro, Alba Zaluar, NUPEVI, Instituto de Medicina Social, UERJ, Rio de Janeiro

Probability of premature death between the ages of 15 and 30 (15q15) is particularly important in studies on violence because external causes account for 72% of ther death in this age group. The objective of this paper is to study the ethnic and regional risk differences of mortality among young people, estimating the probability of an adolescent that is 15 years old to die before reaching 30 years old (15q15). This information can be useful for implementing public policies that aim a reduction of mortality risk at young ages. The estimates have been done for five selected States of Brazil: Rio de Janeiro, São Paulo and Minas Gerais, in the Southeast Region, and Pernambuco and Bahia in the poorer Northeast Region. Using the MortPak, program, developed by the UN Population Bureau, with information from the Demographic Censuses of 2000 and 2010 on children born alive and children surviving according to the mother’s age (Method of Brass), were estimated the probability of death between the ages of 15 and 30 years old (15q15). The results showed higher risks for the Black population than for the White, mainly in Pernambuco in 2000 (15q15 = 48/1000, meaning that 4,8% of young Black, aged 15 years old, will die before completing 30 years) and Bahia (15q15 = 34/1000, in 2000). Nevertheless, significant reductions of 15q15 from 2000 to 2010 were observed for Black and White in five States. In 2010, for the Black population in Pernambuco, the estimate of 15q15 was reduced to 26/1000.

An investigation of seasonal variation of child under-nutrition in Malawi: is seasonal food availability an important factor?
Lana Chikhungu, University of Southampton

Recent estimates indicate that the levels of child stunting and underweight in Malawi are 47% and 13% respectively (NSO-Malawi and MEASURE DHS, 2011). These levels are extremely high considering it is expected that in a well-nourished population only 2% of the population should be stunted or underweight (ORC-Macro, 2006). This study uses the 2004 Malawi Integrated Household Survey which was collected across the year to study the variation of child stunting and child underweight across the year. The multivariate analysis findings show that stunting levels are significantly higher in the months of March to August (the harvest period) compared to September to February (the cropping season) similar to the findings by Ferro-Luzzi et al. (2001). The observed seasonal pattern of stunting is consistent with the seasonal pattern of childhood illnesses such as pneumonia and fever in line with previous studies (Panter-Brick, 1997, Simondon et al., 2001, Rowland et al., 1977). On the other hand, anecdotal evidence suggests reduced child care with mothers spending more time tending to the harvests during the months of March to August may also play a part in the increased stunting rates during this period. Reduced child care during busy agricultural activities has been reported in other settings (Chen et al., 1979, Huffman et al., 1980). It is therefore important that nutrition interventions should take into consideration this seasonal variation of child stunting and reduced child care children might receive during the busy harvesting time when implementing their programmes.

Geographic and socio-economic inequalities in women and children’s nutritional status in Pakistan
Mariachiara Di Cesare, Imperial College London; Zaid Bhatti, The Aga Khan University, Pakistan; Lea Fortunato, Imperial College London;  Zulfiqar Bhutta, The Aga Khan University, Pakistan

Background: Under-nutrition represents an important health risk factor among children and women in low-income countries. Despite the general global improvement in nutritional status observed in the last decade still in 2011 more than 300 million children in the world were undernourished. In particular, inequalities in the nutritional status of children and women suggest that the burden of under-nutrition is not equally shared within a country and across the population. Pakistan is one of the countries in the world with the highest burden of child under-nutrition and the lowest women’s BMI.
Methods: Using data from the National Nutritional Survey of Pakistan 2011 we analyze geographical inequalities in nutrition among children and women as well as individual and household determinants. We use a Bayesian spatial model for the geographical analysis and multilevel regression models for the individual level analysis.
Results: In the best-off 10% of the sample the prevalence of stunting was 24% while it was 56% in the poorest. Similarly women BMI ranged from 20.6 kg/m2 to 26.5 kg/m2 across deciles of wealth. The geographical analysis showed a high level of inequalities with areas characterized by a stunting prevalence from 22% to 75%. In multivariate analysis, after adjustment, household assets and mother’s anthropometric status were associated with higher height, while food insecurity was associated with lower weight-for-age z-score and women’s BMI.
Conclusions: Having identified the main determinants of under-nutrition and at the same time the vulnerable areas in Pakistan is an opportunity for discussing specific local actions for reducing nutritional inequalities.

Inequalities in health in England and Wales: Wednesday 11 September, 11.00am

Gender convergence in human survival and the postponement of death
Les Mayhew & David Smith, Cass Business School, City University

It has been a long accepted demographic maxim that females outlive males. Using data for England and Wales, we show that life expectancy at age 30 is converging and continuation of this long-term trend suggests it could reach parity in 2030. Key among the reasons identified for the narrowing of the gap are differences in smoking prevalence between males and females which have narrowed considerably. Using data from 37 comparator countries gender differences in smoking prevalence are found to explain over 75% of the variance in the life expectancy gap, but other factors such as female emancipation and better health care are also considered. The paper presents a model which considers differences in male and female longevity in greater detail using novel methods for analysing life tables. It considers the ages from which death is being postponed to the ages at which people now die; the relative speed at which these changes are taking place between genders; how the changes observed are affecting survival prospects at different ages up to 2030. It finds that as life expectancy continues to rise there is evidence for convergence in the oldest ages to which either gender will live.

Social inequalities in mortality in English regions and Wales by NS-SEC, 2001 to 2010
Llio Owen, Office for National Statistics

In England and Wales, it is well-established that mortality varies according to socio-economic position. It is usually only possible to publish accurate and precise mortality rates for these social strata around the time of the decennial census, when populations are enumerated by occupation and employment status. However, to improve the timeliness of updates to such statistics, ONS in 2010 conducted a study to investigate the feasibility of using the Labour Force Survey (LFS) to provide inter-censal population denominators for the estimation of mortality rates by the National Statistics Socio-economic Classification (NS-SEC) during the inter-censal period. Further to this, ONS published national results from 2001 to 2010 which showed that there has been a steady decrease in mortality rates for both men and women in most classes over the period. Estimates have now been extended to regional level for three-year rolling periods from 2001-03 to 2008-10, which highlight geographical differences in the social inequality. In this analysis, the LFS was used to provide population denominators by age and NS-SEC for men and women of working age. Numbers of deaths by NS-SEC were obtained from death registrations (numerators) and the populations were derived from the weighted LFS datasets (denominators). Both relative and absolute indicators, Slope Index of Inequality (SII) ad Relative Index of Inequality (RII), are used to examine the trend in the size of the social gradient in mortality over time. The results discussed will be published in a statistical bulletin at the beginning of May 2013.

The changing structure of English society, implications for health: evidence from the Health Survey for England 1996-2010
Fran Darlington, Paul Norman, Dimitris Ballas, University of Leeds

Recent releases from the 2011 UK Census have confirmed that Britain is changing; minority ethnic groups are not the minority they once were, particularly within England. This has implications for those monitoring the health of the English population: inequalities in health are widely documented including variations by ethnic group which cannot be explained away by genetic or cultural differences. Indeed it has been argued that such inequalities are rooted within socio-economic differences. Consequently, exploring how the distribution of key socio-economic variables and health status has changed over time by ethnic group is important to efforts made to understand changing health gradients. Drawing on evidence from the Health Survey for England between 1996 and 2010, this work will show how changes to the social and economic mix of England have differentially influenced the probability of reporting poor health or limiting long-term illness (LLTI) by ethnic group. Time trends illustrating patterns in the distribution of minority ethnic groups by social, class, education, housing tenure, employment status or health status will reveal changes in society’s structure. To assess changing implications for ethnic variations in health over time, binary logistic regression will be used to model the probability of reporting poor health or LLTI each year. These results will be supplemented by calculating SIRs for each ethnic group by, for example, social class, housing tenure and education at each survey year.

The Influence of area disadvantage on LLTI Reporting in England, 2001. Contributing Evidence to the Joint Action European Health and Life Expectancy Information System Programme (Work Package 6: The National Experiences in SES differentials in Health Expectancies)
Eleanor Evans, Office for National Statistics.

The influence of an ecological measure of area deprivation (Index of Multiple Deprivation (IMD), 2004) and a number of household and individual characteristics were used to assess the reporting of limiting long-term illness (LLTI) in England using the Office for National Statistics Longitudinal Study (ONS LS). This research formed part of a larger project to investigate the viability of using the LS as a basis for calculating life and health expectancies by National Statistics Socio-Economic Classification (NS-SEC), with the idea of testing the comparative discrimatory power of individual socioeconomic position and area cluster on a health outcome collected at Census 2001. Binary logistic regression analysis found, amongst others, increasing age, low educational attainment and high area based deprivation to be significant in increasing the odds of reporting limiting long-term illness. A significant interaction between IMD and housing tenure was also present: in the most deprived areas, a protective modifying effect was found for residents of council and other social housing forms of tenure and raised for residents of private rented and other accommodation types of tenure when compared to owner occupiers. This study adds further to the body of evidence linking socio-economic factors with subsequent LLTI reporting and also represents an important step in the study of life and health expectancies using the LS. In addition, the finding of a significant interaction between IMD and housing tenure in LLTI reporting warrants further investigation and may signify differences in housing quality, maintenance and access. ONS LS Clearance Number: 20139.