Gabriela Mejia-Pailles 1,2 , Vicky Hosegood 1,2,3 Kathy Ford 4 , Ann - - PowerPoint PPT Presentation

Centre for Population Change

International Population Conference, Cape Town, South Africa 31st October 2017

Dynamics of children’s living arrangement and caregiver churn in rural communities with high HIV prevalence in South Africa

Gabriela Mejia-Pailles1,2 , Vicky Hosegood1,2,3 Kathy Ford4 , Ann Berrington1,3

1 Centre for Population Change, University of Southampton, 2 Africa Health Research

Institute, University of KwaZulu-Natal, 3Department of Social Statistics and Demography, University of Southampton, 4School of Public Health, University of Michigan

Background

• The care that children receive are associated with many aspects of

their development, health and wellbeing

• For the majority of children, biological parents will be centrally

involved in providing care

• In South Africa, three features provide particular motivation to the

documentation and understanding of children’s care arrangements

• Severe HIV epidemic
• High levels of adult and child migration
• Marriage, union instability and separation levels and patterns

• Family-related health and welfare policy in South Africa where

identifying who is providing care to children is important

• Empirical findings about caregiving arrangements in South

African communities are based on cross-sectional sources of data Aim:

• To describe the dynamics of children’s living arrangement and

the frequency and pattern of changes or ‘churn’ in the people identified as being primary caregivers.

Background

Research Questions:

• RQ1. What are children’s living arrangements in a high HIV

prevalence area in rural South Africa?

• RQ2. How often do orphans and non-orphans experience a

change in main caregiver?

• RQ3. How does age at orphaning relate to the churn in

caregivers?

• RQ4. How soon do orphans and non-orphans experience a first

change in caregiver ?

• RQ5. What are the main pathways in the types of caregivers

experienced by orphans and non-orphans?

Data & Methods

Africa Health Research Institute (AHRI) Demographic Surveillance System

• Ongoing Demographic Surveillance System (DSS) collecting

longitudinal data since 2000, in northern KwaZulu-Natal, South Africa

• n 10,000 households & approx. 90,000 household members; 30% of

household member are not resident in the community

• Prospective cohort of over 10,000 non-orphaned children aged 0-10

years on 1st Jan 2005 who were a member of a study household throughout 1st Jan 2005 and 31st Dec 2012

• Main caregiver: person in charge for the child’s care on a daily basis

 Analytical approach

• Survival analysis and Sequence Analysis

• RQ1. Living arrangements of all resident children <18

years in the DSS by orphaning status, 2005-2012

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Both parents Mother

• nly

Father

• nly

Neither parent Both parents Mother

• nly

Father

• nly

Neither parent Both parents Mother

• nly

Father

• nly

Neither parent Non-orphans Maternal orphans Paternal orphans

2005 2007 2010 2012

RQ2 & RQ3. Mean number of changes in caregiver and mean number of different household members acting as caregivers for children in the prospective cohort, 2005-2012

2 4 6 8 10 Total 15-17 10-14 5-9 0-4 Total 15-17 10-14 5-9 0-4 Total 15-17 10-14 5-9 0-4 Non-orphans Double

• rphans

Paternal

• rphans

Maternal

• rphans

. Changes in caregiving Different household members acting as caregivers

• RQ4. Time to first change in caregiver by maternal
• rphaning status

0.00 0.25 0.50 0.75 1.00 2 4 6 8 years non-maternal orphan <1yr maternal orphan >1yr maternal orphan

Kaplan-Meier failure estimates

• RQ5. Types of caregiving trajectories considering

maternal co-residency and maternal survival status

Clu luster Medoid trajectory % ch chil ildren in in clu cluster 1.

• 1. Mos
• stly

resi esident mothers (mother, resident, 11,) – (female relative, mother alive resident, 1) – (father, mother alive resident, 1) – (other, mother alive resident, 1) – (mother, resident, 3) 40% 2.

• 2. Slo

Slower ch changes (mother, alive resident, 1) – (female relative, mother alive resident, 1) – (mother resident, 8) – (self, mother alive non-resident, 2) – (other, mother alive non resident, 1) – (self, mother alive non-resident, 1) - (other, mother alive non- resident, 1) - (grandmother, mother alive non-resident, 2) 23% 3.

• 3. Fas

ast ch changers (grandfather, mother alive non-resident, 1) – (mother, non-resident, 3) - (grandfather, mother alive non-resident, 1) – (father, mother alive non-resident, 3) – (grandfather, mother, alive non-resident, 2) – (mother, non-resident, 1) – (female relative, mother alive non resident, 1) – (mother, non-resident, 2)- (female relative, mother alive non-resident, 1) – (mother, non-resident, 2) 17% 4.

• 4. Mos
• stly no

non resi esidents s mother (grandmother, mother alive non-resident, 1) – (mother, non-resident, 3) – (other, mother alive non-resident, 1) -– (mother, non-resident, 9) – (grandfather, mother alive non-resident, 1) -– (mother, non-resident, 1) – (grandfather, mother alive non- resident, 1) 8% 5.

• 5. Mos
• stly

gr grandmothers (grandmother, mother alive non-res, 10) – (mother, alive and resident, 3) – (grandmother, mother alive and resident, 1) – (father, mother alive resident, 1) – (mother, alive non-resident, 2) 11%

• RQ5. Medoid trajectory for “Mostly Resident

Mothers” cluster

Cluster Med edoid id tr traje ajectory % % ch chil ildren in n clu cluster 1. . Mos

• stly

ly resi esident mot

• thers

(m (moth ther, res resid ident, 11) 11) – (female relative, mother alive resident, 1) – (father, mother alive resident, 1) – (other, mother alive resident, 1) – (m (mother, re resid ident, 3) 40%

• RQ5. Medoid trajectory for “Slow changers”

cluster

Clu luster Med edoid id tr traje ajectory % % ch chil ildren in n clu cluster 2. . Slow ch changers (mother, alive resident, 1) – (female relative, mother alive resident, 1) – (mother resident, 8) – (self, mother alive non-resident, 2) – (other, mother alive non resident, 1) – (self, mother alive non-resident, 1)

• (other, mother alive non-resident, 1) -

(grandmother, mother alive non-resident, 2) 23%

• RQ5. Medoid trajectory for “Fast changers”

cluster

Cluster Med edoid id tr traje ajectory % % ch chil ildren in in clu cluster 3. . Fast t ch changers (grandfather, mother alive non-resident, 1) – (mother, non-resident, 3) - (grandfather, mother alive non-resident, 1) – (father, mother alive non- resident, 3) – (grandfather, mother, alive non- resident, 2) – (mother, non-resident, 1) – (female relative, mother alive non resident, 1) – (mother, non-resident, 2)- (female relative, mother alive non- resident, 1) – (mother, non-resident, 2) 17%

• RQ5. Medoid trajectory for the “Mostly non-

resident mothers” cluster

Cluster Med edoid id tr traje ajectory % % ch chil ildren in n clu cluster 4. . Mos

• stly

ly non non resi esidents mother (grandmother, mother alive non-resident, 1) – (m (moth ther, non-resid ident, 3) – (other, mother alive non- resident, 1) -– (m (mother, non-resident, 9) – (grandfather, mother alive non-resident, 1) -– (m (moth ther, non-resid ident, 1) – (grandfather, mother alive non-resident, 1) 8%

• RQ5. Medoid trajectory for the “Mostly

grandmothers” cluster

Cluster Med edoid id tr traje ajectory % % ch chil ildren in n clu cluster 5. . Mos

• stly

ly gran and- mot

• thers

(gra randmother, mother alive non-res, 10) – (mother, alive and resident, 3) – (gra randmother, mother aliv live and re resid ident, 1) – (father, mother alive resident, 1) – (mother, alive non-resident, 2) 11%

• RQ5. Clusters characteristics considering maternal

co-residency and maternal survival status

Res esid ident mo mothers Slo Slow w changers Fas ast changers Non Non resid ident mo mothers Gr Gran andmot

• the

rs rs Orp Orphanin ing 2005-2012 (%) (%) Ne Never orph

• rphaned duri

during g 2005-12 12 46 46 21 21 15 15 9 9 Be Became do double le duri during 2005-12 12 4 29 29 31 31 6 30 30 Be Became ma maternal l dur durin ing g 2005-12 12 4 32 32 36 36 6 22 22 Be Became pa paternal dur durin ing 2005-12 12 39 39 22 22 16 16 8 14 14 Nu Number of

• f changes of
• f caregiv

ivers (me (mean) 5 7 7 6 7 Nu Number of

• f di

different car aregi givers (me (mean) 2 3 4 3 3 Se Sex x Mal Male 40 40 23 23 17 17 8 11 11 Fem emale les 41 41 22 22 18 18 8 12 12 Ag Age e at t the the begi beginnin ing of

• f ob
• bserv

rvation per perio iod (%) (%) <= <=5 yr yrs 41 41 27 27 13 13 9 10 10 6-10 10 yr yrs 40 40 19 19 22 22 7 13 13

Discussion

• The high levels of migration, parental mortality and low rates of

marriages have resulted in disperse living arrangements for many children and their parents in rural communities in South Africa.

• Our findings showed that children who became orphan and children

whose parents survived during the period of observation in this community both experienced a similarly high mean number of changes in their primary caregiver

• Caregiving responsibilities felt to a small number of household

members, who alternate on the caregiver role

• We found no evidence of a self-care trajectory

Limitations

• Selectivity of the children in the prospective cohort
• Children needed to have survived until 1st January 2005,

particularly older children

• Children needed to be alive and present in the study area for

their inclusion between 2005 and 2012

• Children needed to be non-orphans for their inclusion in the

analysis

Acknowledgments

• All members of the community
• All staff at the Welcome Trust Africa Health Research

Institute (field teams, quality controllers, data capturers, data centre, database team)

• ESRC-UK for funding the project “Improving Measures of the

Family Environment in Longitudinal Population Studies of Child Health in sub-Saharan Africa”

Centre for Population Change

Find our more and contact us

Web: www.cpc.ac.uk Email: cpc@soton.ac.uk Tel: +44 (0)23 80 592 579 Twitter: @CPCpopulation Facebook: CPCpopulation Mendeley: CPC Population