Introduction Household Structure and Childhood Mortality in Ghana Winfred Avogo Victor Agadjanian Department of Sociology/ Center for Population Dynamics, Arizona State University Child mortality public health priority in Africa Analysis show mortality decline has stalled Focus on socio-economic & biological factors Little known about influence of living arrangements of children on child mortality Examine household structure within cultural and socio-economic context children live Setting Objectives 2 Ghana; population of 20 million in West Africa Estimate effect of household structure on child survival Economy stable; estimated poverty rate at 35% of population Examine variations of household structure and child survival across rural and urban areas 25% Urban and 75% rural Young population with 46.9% 0-14 Chart trends in childhood mortality from 1993- 2003 3 Overall life expectancy of 59.6 for females and 55.4 for males 4 Hypotheses Theoretical framework Three major theoretical perspectives: ‘New household economic models’ Becker, 1981; Schultz, 1974 ¾ Critically examine household as a single entity ¾ Resource pooling within the family ¾ Household boundaries and resource availability Literature on socioeconomic influences on child survival Biological proximate determinants of child survival 5 Children in nuclear households have health advantage if constraints to resource pooling exist Effects of nuclear households depend on mother’s education and socio-economic status Effects mediated by bio-demographics of mother and utilization of maternal services Effects of household structure stronger in rural than in urban areas 6 1 Data and Methods Ghana Demographic Health Surveys (GDHS), crosssectional; 1993, 1998 & 2003 Sample size: 6351 last born children •Outcome Variable: ¾Childhood mortality: probability of dying from birth to age 5 •Main predictor: Household structure: two broad forms: ¾(1) Elementary: nuclear & single-parent Households ¾(2) Extended:three generational & lateral households Statistical Model Discrete-time logistics models ¾ 5 Age intervals specified ¾ Each child contributes 1-5 observations depending on age at death or censoring 7 •Control variables: person months lived, socioeconomic, bio-demographic, maternal health factors 8 Figure 1: Household Structure Analytical Strategies ¾ Model 1: Household structure, 2: socio- Lateral extended (20.36%) economic, 3: biological and maternal utilization ¾ Urban and Rural Areas: Same model specification Nuclear (49.28%) Three generational (12.41%) Single parent (17.95%) 10 9 Household structure by selected characteristics 160 ild ir th de 108 111 100 80 77 66 60 57 64 40 20 0 1983-1987 1989-1993 1994-1998 1999-2003 Ch ea 119 120 Approximate GDHS period Infant mortality (1q0) an ag at h n ba tb Ur S r al Ru S w Lo m Me diu SE S SE uc SE gh Hi uc Ed No uc Ed er a ry im Pr ig h 155 140 Nuclear Single parent Three generational Lateral extended Me Se c /H 180 Percent Ed 90 80 70 60 50 40 30 20 10 0 Figure 3: Trends in Infant and Under-five Mortality Rates Ghana 1988-2003 Deaths per 1000 Live Data and Methods Under-five mortality (5q0) Selected Characteristics 11 12 2 Figure 4: Child Survival Estimates by Household Structure Multivariate Results Table : Odds ratio of the relationship between family structure and child death All Rural Urban Family Structure Single-parent 1.06 0.87 1.08 Three-generational 1.59 + 1.89 * 1.11 Laterally extended 1.51 * 1.67 * 1.45 (Nuclear) 1 1 1 Month ('0) 1 1 1.00 '1-5 0.10 ** 0.08 0.14 * '6-11 0.09 ** 0.08 * 0.08 * '12-23 0.27 ** 0.31 0.16 '24-59 0.58 0.78 * 0.27 Mother's eductaion Secondary or Higher 0.63 + 0.82 0.78 ** Primary 0.71 0.60 * 0.32 (No education) 1 1 1 Standard of living index High 0.27 0.81 0.27 ** Medium 0.51 * 1.09 0.61 ( Low) 1 1 1 Prenatal care and delivery assistance Received either Received neither (Received both) 13 0.82 0.80 1 1.241 1.58 1 14 Conclusion and Implications Extended households seem to have negative impact on child mortality Household structure important for rural areas does not appear so in urban areas Cost of health care, social inequalities may explain effects of household structure in rural and urban areas 1 *= <.05, ** = <.01 + = <.1 Summary of findings 0.81 0.72 Education and standard of living more important in urban areas 15 Community level health interventions need to adjust to specific needs of household forms More research is needed to identify vulnerable children in changing household situations Comprehensive sociological models needed to account for household effects 16 THANK YOU 17 3