Download - Di Gessa G, Glaser K and Tinker A
The health of grandparents caring for their grandchildren: The role of early and mid-life conditions
Di Gessa G, Glaser K and Tinker AInstitute of Gerontology, Department of Social Science, Health & Medicine,
King’s College London
United Kingdom
Outline
• Background• Aim and objectives• Data and Methods• Results• Conclusion
Background
• Grandparents play crucial role in family life• Evidence of the impact of childcare on
grandparents’ health is mixed: Custodial/Primary grandchild carers
experience poorer health; Higher quality of life, lower depression and
loneliness among grandparents providing grandchild care (vs no care).
Background /2
• Most studies are cross-sectional and samples consist mostly of US grandparents;
• Focus on primary and custodial care;• Few studies have studied the link between
grandchild care and grandparents’ health using a cumulative advantage/disadvantage framework.
Aim and objectives
Examine the effects of caring for grandchildren on health among European grandparents
i) Using longitudinal data
ii) Using life history data, and controlling for cumulative experiences across the life course (paid work histories; health and socio-economic position in childhood).
Data
3 waves of multidisciplinary comparable surveys, representative of individuals 50+– Survey of Health, Ageing and Retirement in
Europe (SHARE) (N~27,000);France, Austria, Germany, Sweden, Denmark, Switzerland,The Netherlands, Italy, Spain, Greece, Belgium
– Household response rate: 62%, with individual response rates higher than 85%;
– First wave collected in 2004/05.
Focus on grandparents
Data /2
Waves 1, 2 provide information on grandparents, including demographic and socio-economic characteristics, health, and household characteristics.
Wave 3 collects retrospective life history information about childhood conditions, and life events in adulthood.
Data /3
«During the last 12 months, have you looked after your grandchild[ren]
without the presence of the parents?» If so i) «how often?» [daily, weekly, monthly, less often]
ii) «about how many hours?»
Intensive grandparental childcare if grandchildren were looked after by grandparents on
a daily basis or at least 15 hours per week
Overview of Analysis
Latent Health w2Baseline Characteristics (w1)
Age; Gender; Education;Household type, Country;Wealth quintiles;Number & Age of grandchildren;Childcare;Paid work and social engagement;
Latent Health;Health behaviour (BMI, smoking);Depression; Cognitive function;
Latent childhood disadvantage
classes
Number of unions;In paid work >75%;Has never worked;Has suffered i. Hunger; ii. ‘Adverse’ event; iii. Long periods of ill health
Measures
We used Latent Class Analysis in order to classify respondents into childhood conditions advantage/ disadvantage subgroups;
[Dysfunctional parent; One parent dead; Occupation of breadwinner; Books in HH; toilet; Hot water; Bath; Heating; Poor health; In hospital or bed for one month or more; With severe illness at age of 10 ]
We used a Latent Variable to represent the concept of ‘somatic’ health;
[Self-rated health, Self report of conditions (cancer, lung, heart, stroke, diabeter), Self report of limiting disability, Activities of Daily Living, Instrumental activities of daily living]
Sample and Methods
Sample:• ~16,000 grandparents aged 50+ at baseline;• ~ 9,700 grandparents at 24-month follow-up;• ~ 7,200 with history data.• ~ 6,500 complete cases (~41%)
AnalysisLinear regression of the latent variable health at follow-up, controlling for both baseline and history socio-economic and demographic characteristics.
Results – descriptive /1
Grandparental childcare Wave 1 Wave 2
Not looking after 50.2 50.2
Not intensive 36.1 36.8
Intensive 13.7 13.0
Total 15,887 9,644
Distribution of grandparent childcare, by wave
Source: SHARE 2004/05, 2006Countries: France, Austria, Germany, Sweden, Denmark, Switzerland, The Netherlands, Italy, Spain, Greece, Belgium
Results – descriptive /2
Not looking after
Not intensive
Intensive
SRH fair/poor 46.9 30.5 36.7ADL limitations 16.9 6.9 7.4Depressed 30.5 20.7 27.2
In couple >80% 71.0 78.9 83.2Never worked (W) 27.9 14.4 29.1Suffered hunger 13.6 8.9 9.5
Childhood advantage 19.2 33.5 17.3Poor 73.5 58.7 75.8
Distribution of selected grandparent’s characteristics, by childcare
Results – linear regression /1Beta coefficients from models of ‘good’ health at wave 2
• Younger grandparents in higher education, and in the higher wealth quintiles at baseline are less likely to report poor health at wave 2;
• No gender differences;• No household composition differences; age and
number of grandchildren not significant;• Social engagement at baseline not significant.
Results – linear regression /2Latent health 0.558 < 0.001In lowest cognitive quintile – 0.049 0.005Depressed – 0.094 < 0.001Obese – 0.077 < 0.001Smoking – 0.009 0.543
2 or more marital unions – 0.018 0.352In paid work for < 75% of working life – 0.022 0.114Has never worked – 0.046 0.019Has suffered long periods of ill health – 0.154 <0.001Has suffered hunger – 0.022 0.228Has suffered any ‘adverse’ event – 0.019 0.298Poor but in good health at 10 0.001 0.932Poor and in poor health at age 10 – 0.039 0.054
Not intensive 0.033 0.010Intensive 0.033 0.019
Conclusions
Using waves 1, 2 and history data
i) Provision of childcare –both intensive and non-intensive –positively associated with good health over time;
ii) Such relationship remains even when accounting for childhood and adulthood disadvantage;
iii) Expected relationships between baseline health, socio-economic and demographic characteristics and health at follow-up.
Limitations & Future work
• Attrition can bias results, especially in the older population where the most ‘disadvantaged’ have a higher probability of dropping out of the study;
Multiple Imputations, Sensitivity analysis
• “Selection effect” of grandparents who look after grandchildren. Unmeasured factor?
Thanks for your attention!
Questions, comments and feedback are welcome.
Childhood
3 classes
• Class proportion: 68%; 24%; 8%• Classification accuracy: 0.84• Average Latent Class probability
1 2 3
1 0.94 0.04 0.02
2 0.02 0.96 0.02
3 0.09 0.06 0.85
Childhood /2
Dysfuncti
onal pare
nt
One parent d
ead
Occup LO
W
No books
No toile
t
No hot wate
r
No bath
No heating
SRH = poor o
r bad
In Hospita
l or b
edSic
k0
0.2
0.4
0.6
0.8
1
Class 1Class 2Class 3
Figure 1. Conditional Response Probabilities
Somatic Health
We used:• Self rated health• Self report of long-term health problems• Self report of heart failure, chronic lung
disease, stroke, diabetes, and cancer• Activities of Daily Living• Instrumental activities of daily living
CFI TLI RMSEA
Unidimensional model 0.977 0.969 0.037