Time trends in family risks and their impact Stephan Collishaw & Barbara Maughan MRC SGDP Centre Institute of Psychiatry Kings College London
Time trends in family risks Recent decades Marked changes in family demographicsMarked changes in family demographics –age at marriage / cohabitation –age at birth of first child –family size –partnership stability / breakdown –complexity of family forms Increases in psychosocial disorders among young peopleIncreases in psychosocial disorders among young people
Time trends in adolescent conduct problems Collishaw et al, 2004
Questions do changing family patterns explain trends in child outcomes?do changing family patterns explain trends in child outcomes? do the same risk factors contribute todo the same risk factors contribute to –individual differences in behaviour problems? –overall levels of behaviour problems? does the impact of family risks change when their prevalence changes?does the impact of family risks change when their prevalence changes?
ONS Population Trends 102, 2000 & Social Trends 20, 1990 Divorce rate per 1,000 married population (England & Wales)
Hypothesis increasing prevalence of divorce associated with reduction in impactincreasing prevalence of divorce associated with reduction in impact –pre-divorce selection effects less troubled familiesless troubled families less parental discordless parental discord –post-divorce consequences less social stigmaless social stigma increased awareness of impact for childrenincreased awareness of impact for children
Parental divorce and child outcomes meta analysis (Amato, 2001) Decade of publication
Parental divorce and child outcomes meta analysis (Amato, 2001) Decade of publication
UK Cohorts Ely et al, 1999; Sigle-Rushton et al, 2005 Cohort
Possible explanations change in nature of marital dissolutionchange in nature of marital dissolution –low discord divorce especially distressing for children increasing gap in economic well-beingincreasing gap in economic well-being –single-parent families not benefited from economic expansion
Aims update picture to include more recent cohortupdate picture to include more recent cohort examine 3 family indicatorsexamine 3 family indicators –family type –family income –family size explore changes in prevalence and impactexplore changes in prevalence and impact illustrate changes in correlatesillustrate changes in correlates test how far changing family risks account for rising levels of conduct problemstest how far changing family risks account for rising levels of conduct problems
Samples Age 16
Family-based correlates – Family type (single vs. step vs. intact) – Family income (<60% median vs. remainder) – Family size (4+ children vs. 1-3 children) Adolescent conduct problems – fighting – bullying – stealing – lying – disobedience Measures
ONS Population Trends 102, 2000 & Social Trends 20, 1990 Cohort 1 Cohort 2 Cohort 3 Divorce rate per 1,000 married population (England & Wales)
Teenagers living with both birth parents
Teenagers living in a step-family
Teenagers living in single-parent households
Low income by cohort: intact families
Proportion of families in poverty: Single parent vs. intact families Single Intact
Increasing income-inequality: single parent vs. intact families OR = 8.5 Single Intact
Increasing income-inequality: single parent vs. intact families OR = 8.5OR = 10.1 Single Intact
Increasing income-inequality: single parent vs. intact families OR = 8.5OR = 10.1OR = 19.4 Single Intact
Family type and conduct problems by cohort Family type OR = 1.9
Family type and conduct problems by cohort Family type OR = 1.9 OR = 2.1
Family type and conduct problems by cohort Family type OR = 1.9 OR = 2.1 OR = 1.8
Family type and conduct problems by cohort Family type OR = 2.7
Family type and conduct problems by cohort Family type OR = 2.7 OR = 2.2
Family type and conduct problems by cohort Family type OR = 2.7 OR = 2.2 OR = 1.6
Summary: family type Risk factorExposureImpact Single parent familyUpNo change Step family UpDown?
Social Trends, 2002 Absolute household disposable income in the UK ( )
Social Trends, 2006 Relative poverty in the UK ( )
Relative poverty by study cohort
Relative poverty and conduct problems by cohort Low income OR = 1.4
Relative poverty and conduct problems by cohort Low income OR = 1.4 OR = 2.1
Relative poverty and conduct problems by cohort Low income OR = 1.4 OR = 2.1 OR = 1.7
Family size by cohort: % four or more children
Family size and conduct problems by cohort N children OR = 3.0
Family size and conduct problems by cohort N children OR = 3.0 OR = 2.3
Family size and conduct problems by cohort N children OR = 3.0 OR = 2.3 OR = 1.8
Summary of findings up to now Risk factorExposureImpact Single parent familyUpNo change Step family UpDown? Relative povertyNo changeUp Large family sizeDownDown
Conduct problems: high scores Total OR = 1.56 per cohort
What contributes to time trends in conduct problems?
What contributes to time trends in conduct problems?
What contributes to time trends in conduct problems?
What contributes to time trends in conduct problems?
Some conclusions 1. Parallel trends in risks and outcomes do not imply a causal link –Increase in divorce rate over the past 30 years –Increase in conduct problems over the same time period –However, trends in family type appear largely independent of trends in conduct problems
Some conclusions 2. Correlates of risk factors may change over time –E.g. Amato –Only had limited data with which to look at this –Socio-economic disadvantage even more strongly associated with single parenthood in more recent cohorts
Some conclusions 3. Implication: changes in impact of a risk factor as important as changes in exposure –Focus on changes in exposure insufficient –Also need to consider possible change in association between risk and outcome –E.g. 1: family type and conduct problems; impact the same or reduced over time –E.g. 2: relative poverty and conduct problems; impact gone up over time
Some conclusions. 4. Different explanatory models needed for understanding individual differences and level differences –Family type, income and size all associated with CP at individual level –But, trends in these aspects of family life only made modest contribution to understanding of level differences –Different factors may be relevant for the two
Time trends in family risks and their impact Stephan Collishaw & Barbara Maughan MRC SGDP Centre Institute of Psychiatry Kings College London
Prevalence and odds of low income by family type and cohort OR =