OBJECTIVES: Our goal was to examine the association between poverty, in the first and fourth years of life and cumulatively in the first and fourth years of life, and the health of children in the fourth year of life in the UK Millennium Cohort Study and in the Quebec Longitudinal Study of Child Development (QLSCD).
METHODS: Data from the UK Millennium Cohort Study of 14 556 children and from the QLSCD of 1950 children were analyzed. Comparable measures of poverty were households in receipt of the safety-net benefit: income support in the United Kingdom and social welfare in Quebec. Three parent-reported health indicators were examined: asthma attack, long-standing illness, and limiting long-standing illness by the fourth year of life. Associations were explored with logistic regression modeling controlling for child characteristics and maternal education.
RESULTS: Poverty only in the first year of life significantly increased the risk of asthma attacks and limiting long-standing illness in the fourth year of life among UK children; trends were in the expected direction in the QLSCD but did not reach statistical significance. Poverty in the fourth year of life only significantly increased the risk of all 3 outcomes for UK children but not for Quebec children. For children experiencing poverty in both the first and fourth years of life, the risks for all 3 outcomes also increased in the United Kingdom, whereas only the risk of limiting long-standing illness increased in Quebec. Adjustment for confounding had little effect on the increased risks associated with poverty.
CONCLUSIONS: These findings suggest that experience of poverty at various times in early childhood increases the risk of asthma attacks and chronic illness in the fourth year of life; however, they also indicate that poverty at different stages of the early childhood life course may have different effects on chronic illness in different country settings.
WHAT'S KNOWN ON THIS SUBJECT:
Poverty in early childhood is detrimental to children's health, but the mechanisms are poorly understood. There is limited evidence on the impact of cumulative poverty in early childhood on chronic illness in childhood.
WHAT THIS STUDY ADDS:
First-year-of-life and cumulative poverty have similar effects on chronic illness in the United Kingdom and Quebec. Poverty in the fourth year of life only is associated with chronic illness in the United Kingdom but not Quebec.
Although poverty is widely accepted as a major determinant of child health,1,–,5 the mechanisms by which it impacts child health remain unclear especially in rich countries with relatively well developed social protection systems. Poverty in childhood has been shown to have an adverse effect on adult health,6,–,8 and there is a new interest in how early “adverse experiences associated with poverty can lead to a lifetime of illness and diminished capacities.”9 There are limited data on the impact of duration of poverty on health in early childhood, but authors suggest that chronic poverty is more detrimental to health in early childhood than transient poverty.2,10,–,12 Early poverty seems to have an impact on child health at 10 to 11 years,12 and growing up in poverty has been reported to carry a higher risk of chronic illness and asthma in childhood.13,–,15 However, few authors take into account the dynamics of poverty when studying the health of children, and we know little about the impact of diverse social policies on the health of children.16
The multiple correlates of poverty, such as increased levels of maternal smoking and low maternal education, pose significant methodologic challenges in studying the mechanisms linking poverty to child health. However, a clearer understanding of these mechanisms is key to informing social policies aimed at addressing the social determinants of child health.
International comparisons, although methodologically problematic, offer the possibility of comparing the impact of poverty on child health in different countries and social policy settings. Important differences have been reported in child wellbeing when comparing income inequalities in industrialized countries at an ecological level.17 Few authors have compared the impact of long-term poverty during early childhood on child health by using individual data from different countries.
On the basis of secondary analysis of longitudinal data from ongoing cohort studies in the United Kingdom and Quebec, we explored the relationship between duration of poverty by using comparable measures collected in the first and fourth years of life, and chronic illness in 2 countries. In addition, we tested whether maternal education, maternal smoking, migrant status, violence since birth, and lone parenthood mediated the association of poverty and chronic illness and whether the association was explained by confounding by child and household characteristics.
Data from the first (9 months) and second (36 months) sweeps of the UK Millennium Cohort Study (UKMCS) and from the first (5 months) and fourth (41 months) cycles of the Quebec Longitudinal Study of Child Development (QLSCD) were analyzed. The UKMCS, coordinated by the Centre for Longitudinal Studies, Institute of Education, University of London, is specifically designed to follow “the new-century infants” and their families.18 Live births were drawn from the child benefit register after a multistage cluster sampling strategy that selected a random sample of 398 electoral wards, disproportionately stratified to ensure adequate representation of all 4 UK countries, deprived areas, and those with high concentrations of black and ethnic minority families.19 All infants who were alive and living in the selected electoral wards were eligible if they were born between September 1, 2000, and August 31, 2001, in England and Wales, and between November 22, 2000, and January 11, 2002, in Scotland and Northern Ireland, and in receipt of child benefit.18 Of the 20 646 targeted children, data were obtained for 18 819 living infants born to 18 553 participating families at the first sweep. Participating families were resurveyed when their infants were 36 and 60 months old, and will be followed-up every 2 years.
The QLSCD is a birth cohort, coordinated by the Direction Santé Québec of the Institut de la Statistique du Québec since 1998. The sample was drawn from the Quebec live birth registry. Sampling followed a 3-step strategy.20 The resulting sample was representative of singleton live births registered in the Quebec live births registry in 1997–1998 with the exception of those on the Cri and Inuit territories, on Indian reservations, or in the northern region of Quebec (2.1% of live births). Infants born before 24 or after 42 weeks' gestation (0.1%) and those with unknown gestational age (1.3%) were excluded. A random sample of 2940 singleton live-born infants was initially selected. Parents of 2675 children were reachable and 83.1% consented to participate.20 The baseline data collected at 5 months old included 2223 children of whom 2120 children were since resurveyed annually.
Our analysis included singleton births only, involving 14 556 children with complete data of 15 596 children followed at 9 months and 36 months by the UKMCS (participation rate of 80% for singletons) and 1893 of 1950 children followed at 5 months and 41 months by the QLSCD (participation rate of 92%). The authors of both studies used structured questionnaires to collect data on children's health and development, parental characteristics, and on household and community level living conditions. Trained interviewers conducted computer-assisted face-to-face interviews with the person most knowledgeable about the child: the mother in 99% of cases in the UKMCS and in 98% of cases in the QLSCD. Interviews were administrated in English in the UKMCS and in English or French in the QLSCD, at respondents' homes. Respondents also answered paper-based self-administrated questionnaires. Data on birth weight were collected during the interview regarding 9-month-old infants for the UKMCS. Children's neonatal health data were collected from hospital records in the QLSCD.
Poverty was defined as households in receipt of the safety net state benefits: income support (IS) in the United Kingdom and social welfare (SW) in Quebec (see Appendixes 1 and 2 for details of monetary benefits according to family type). These measures were chosen because they were the most comparable across the 2 cohorts. IS in the United Kingdom is offered to families with savings of less than £16 000 and low income in which no member works >16 hours/week. The amount received depends on family size and is below 60% of median income for households with children. In Quebec, families are eligible to receive SW when they have no source of income. For a lone parent with 1 child and a couple with 2 children, the allocated amounts were, 60% and 54%, respectively, of the low income cutoff defined by Statistic Canada.21 Households in receipt of IS when the infant was 9 months old in the United Kingdom and SW when the infant was 5 months old in Quebec were defined as poor in the infant's first year of life. Those households in receipt of IS or SW when the child was 3 years old in the United Kingdom or 3.5 years old in Quebec were defined as poor in the infant's fourth year of life. Poverty status was categorized as follows: never poor; poor only in the first year of life; poor only in the fourth year of life; or poor in both the first and fourth year of life (designated as cumulative poverty).
Three health indicators, for which comparable data were available in both cohorts, were used to examine chronic health problems in the first 4 years of life: the occurrence of asthma attacks at any time up to the child's fourth year of life; the occurrence of any long-standing illness; and the presence of limiting long-standing illness. In the UKMCS, asthma attack was not specifically defined in the question, “… has the child ever had asthma?” Chronic illness was defined as any long-term condition that had been diagnosed by a health professional and had lasted for >3 months or was expected to continue for at least 3 months. These conditions include, among others, vision and hearing problems, asthma, and convulsions. Limiting chronic illness was defined as any long-term condition that limited the child at play or from joining in any activity normal for a child of his or her age.
In the QLSCD, the question about asthma attack was, “In the past 12 months did [child's name] have an asthma attack?” Combining mothers' reports for the 4 first waves, asthma attack was operationalized as the occurrence of any asthma attack since the child was born. The presence of a chronic illness was established when a mother reported that during the previous 12-month period her child was given a diagnosis by a health professional of allergy, heart disease, bronchitis, kidney disease, mental disability, epilepsy, cerebral palsy, or any other health problem that lasted or might last for 6 months or more. Mothers specified whether the illness limited their child at play or from joining in any activity normal for a child of his or her age. A list of the questions used by the UKMCS and the QLSCD to collect data on the dependent variables is included in Appendix 3.
Descriptive analyses were first conducted by using sample weights to account for the complex sampling design in both cohorts. We explored the associations between poverty and chronic illnesses with logistic regression modeling introducing potential mediators known to be important variables in the pathway from poverty to ill health in childhood. These included maternal smoking, maternal education, lone parenthood, marital violence, and immigration status. Maternal smoking was defined as those smoking daily and subdivided into those who smoked during the child's first year of life, during the child's fourth year of life, or during both the first and fourth years of life. In the UKMCS, maternal education represents educational qualification obtained from the age of 16 years (no academic qualification, academic qualification at 16, academic qualification at 18, degree or equivalent) as reported by the mother at the first sweep. In the QLSCD, maternal education represents the highest educational level attained (partial or completed college or university studies; high school, vocational, or trade school diploma; or no high school diploma) at the baseline data collection. Marital violence was established when a mother reported having been physically hurt or threatened by her partner since her child's birth. A mother was classified as an immigrant if she migrated from a non-European country. Mediation was assessed following the 4 relational criteria established by Baron and Kenny.22 None of the potential mediators fulfilled all 4 conditions. We controlled for potential confounding including child characteristics (gender, age, birth order, and birth weight) and the number of household members at baseline. Because children in both countries have free-at-time-of-use access to all levels of health care, health care access was not relevant as a mediator or confounder. Because maternal smoking and education were not shown to act as mediators, they were treated as potential confounders.
We conducted separate but identical analysis for each health outcome and each cohort by using SPSS 13.00 (SPSS Inc, Chicago, IL).
Descriptive results are summarized in Table 1. In the UKMCS, asthma was reported in 1688 children (11.6%), long-standing illness in 2300 children (15.8%), and limiting long-standing illness in 422 children (2.9%). In the QLSCD, at least 1 asthma attack since birth was reported in 258 children (13.6%), long-standing illness in 341 children (18.1%), and limiting long-standing illness in 31 children (1.6%).
In the UKMCS, 13.5% of children participating at 3 years old were living in families receiving IS during their first year of life, whereas in the QLSCD, 11.0% of children were living in households receiving SW at that age. However, 4.4% in the UKMCS and 5.9% in the QLSCD were living in poor families only during their first year of life, and 3.7% and 2.4%, respectively, only in the fourth year. For cumulative poverty, there were 9.1% of children in the UKMCS and 5.2% of children in the QLSCD who experienced poverty both during their first and fourth years of life (Table 1). In both countries, univariate associations suggest that children living in poverty have a higher risk of having 1 of the 3 chronic conditions: asthma attacks; long-standing illness; and limiting long-standing illness (Table 2).
Odds ratios (ORs) and 95% confidence intervals (CIs) estimated at various steps of the modeling process are presented in Table 3. Model 1 provided unadjusted estimates of the association between poverty indicators and the outcome variables. Model 2 adjusted for maternal smoking, and model 3 additionally adjusted for the child's and household characteristics. The full model 4 controlled for all covariates by adding maternal education.
In the UKMCS, unadjusted estimates show that experience of poverty only in the first year of life significantly increased the risk of asthma attacks (OR: 2.04 [CI: 1.70–2.44]) and limiting long-standing illness (OR: 1.49 [95% CI: 1.02–2.17]) in the fourth year of life. Adjustment for maternal smoking status, the child's characteristics, and maternal education changed very little the strength of the association between asthma and first-year poverty, which remained statistically significant at the 5% level. In the QLSCD, none of the 3 health indicators were significantly associated with poverty in the child's first year of life, although trends were in the expected direction. Table 3 also shows that in the United Kingdom but not in Quebec, exposure to poverty in the fourth year of life only is associated with higher risk for all 3 indicators of ill health before and after adjustment.
The adjusted odds of asthma attacks, long-standing illness, and limiting long-standing illness remained significantly higher for UK children in cumulative poverty compared with the reference group. Only limiting long-standing illness was significantly associated with cumulative poverty for children in the QLSCD cohort.
These analyses show that, in both countries, relative poverty during early childhood has health consequences for the child later on. In both countries, a comparable percentage of children in their first and fourth years of life were living in families receiving IS or SW. However, there was a higher percentage of cumulatively poor in the UKMCS than in the QLSCD. These differences remained even after applying the sampling weights that account for oversampling of low-income families in the UK sample.
Our results demonstrate that in the UK poverty only in the first year of life was associated with an increased risk of asthma attacks and limiting long-standing illness in children in the fourth year of life. Trends for asthma and limiting long-standing illness were in the expected direction for Quebec children but failed to reach statistical significance as a result of small numbers. Poverty in the fourth year of life had very different effects in the 2 cohorts: in the United Kingdom, the odds for all 3 outcomes were higher than those for poverty in the first year only, but, in Quebec, there was no association between fourth-year-only poverty and any of the outcomes. Cumulative poverty was associated with higher risks for all 3 health problems in the UKMCS but only for limiting long-standing illness in the QLSCD, although the ORs for asthma and long-standing illness were in the expected direction. In the UKMCS, the higher risks remained significant for all 3 health outcomes after adjustment for confounding as did limiting long-standing illness in the QLSCD cohort. In addition, the findings indicate that poverty at different stages of the early childhood life course may have different effects on chronic illness in different country settings. The increased risk of chronic illness associated with poverty only in the fourth year among UK children but not among Quebec children may result from the known tendency for UK households with disabled children to become increasingly disadvantaged as the children get older.23
In both cohorts, ORs for limiting long-standing illness were higher and more consistently significant than for the other 2 outcomes. Children whose long-standing illness is activity-limiting are more likely to be more severely affected by their illness than those with no activity limitation, suggesting that poverty may be causally related to increased severity of long-standing illness or activity limitation may be differently reported by poor compared with nonpoor parents.
Comparison With Published Literature
We are not aware of authors of published articles who compared individual level data on the relationship of early and cumulative poverty to chronic illness in early childhood. Social disparities in chronic illness in childhood have been reported from authors of various countries24,–,26; however, the specific role of early and cumulative poverty has received less attention.2,3 Our findings, particularly the UKMCS data, suggest that poor children are at higher risk of experiencing asthma attacks. Some authors suggest that disparities are related more to difference in management leading to more frequent and more severe attacks.2,27,–,33
Our findings that early and cumulative poverty are related to chronic health problems in the fourth year of life support Chen et al's12 conclusions that the effect of poverty on child health is cumulative rather than latent. In addition to their adverse effects on children's health in the fourth year of life, early and cumulative poverty are likely to exert negative influences on health across the life course.4,16
Our study has a number of limitations that need to be taken into account when interpreting the results. The measures of poverty used are likely to have limited our capacity to demonstrate health differences between poor children and children classified as nonpoor in this study. Using safety net benefits to classify poverty excludes children of working poor families. For example, in the QLSCD, 50% of those children living below Statistics Canada's low income cutoff are in working households that are not reliant on SW.15 Therefore, our results might underestimate the relationship between poverty and child health. The outcomes measured in the 2 cohorts are not identical and no information on the severity of asthma attacks or long-standing illnesses was available. The small numbers in the QLSCD resulted in unstable findings with wide CIs in some of the regression models. However, the direction of the effects was similar to that of the UK data lending support to the overall conclusions.
The main strength of the study is the comparison between longitudinal cohort studies in the United Kingdom and Quebec at about the same period of time with high participation rates and with comparable data.
Social Policy Implications
Despite differences, for example in maternity benefits and child care, both countries have protective social policies for families and children. However, poor children are still more at risk of ill health. We suggest that policies aimed at more effectively reducing child poverty, particularly in early childhood, may reduce the burden of chronic illness in childhood.
The UKMCS was supported by the Economic and Social Research Council, the Office of National Statistics, and various government departments. The study was led by the Centre for Longitudinal Studies at the Institute of Education of the University of London. We thank the Economic and Social Data Service and the United Kingdom Data Archive for permission to access the study data. The QLSCD was supported by Canadian Institutes of Health Research grants 200309MOP-123079 and 200609MOP-165867 and by the Institut de la Statistique du Québec, Direction Santé Québec, which was responsible for data collection and validation.
- Accepted September 28, 2009.
- Address correspondence to Nick Spencer, FRCP, FRCPCH, MPhil, University of Warwick, School of Health and Social Studies, Coventry CV4 7AL, United Kingdom. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- UKMCS =
- UK Millennium Cohort Study •
- QLSCD =
- Quebec Longitudinal Study of Child Development •
- IS =
- income support •
- SW =
- social welfare •
- OR =
- odds ratio •
- CI =
- confidence interval
- Phipps S
- Séguin L,
- Nikiéma B,
- Gauvin L,
- Zunzunegui MV,
- Xu Q
- Séguin L,
- Xu Q,
- Gauvin L,
- Zunzunegui MV,
- Potvin L,
- Frohlich KL
- Spencer N
- Shonkoff JP
- Chen E,
- Martin AD,
- Matthews KA
- Newacheck PW,
- Rising JP
- Rona RJ
- Séguin L,
- Xu Q,
- Potvin L,
- Dumas C,
- Frohlich KL
- Pickett KE,
- Wilkinson RG
- Dex S,
- Joshi H
- Plewis I,
- Calderwood L,
- Hawkes D,
- Hughes G,
- Joshi H
- Jetté M,
- Desgroseillers L
- 21.↵National Council of Welfare. National Council of Welfare: welfare incomes 2005. Report volume 125. Ottawa, Ontario, Canada; Minister of Public Works and Government Services, 2006
- Burchardt T
- Gordon D,
- Parker R,
- Loughran F,
- Heslop P
- Berntsson L,
- Kohler L
- Chen E
- Chen E,
- Fisher E,
- Bacharier L,
- Strunk R
- Copyright © 2010 by the American Academy of Pediatrics