Methods: This cross-sectional descriptive study analyzed the 2007 Ghana Maternal Health Survey (GMHS), the first and the only maternal health survey conducted in Ghana (Ghana Statistical Service, 2009). The primary data were collected in two-phases of fieldwork. In phase I, a nationally representative sample of 240,000 households was selected from the 10 administrative regions of Ghana across rural and urban areas, out of which 226,209 completed the household questionnaire. We analyzed the data using STATA version 13 (College Station, TX). Descriptive inferential statistics were used to summarize the means of continuous variables and proportions of categorical variables and to cross-tabulate each variable, including the household resources, by maternal report of child death in the full sample and in the sample stratified by rurality of residence. Means were compared statistically with t-tests and proportions with chi-square tests. We examined the household resources and other variables for their intra-class correlations and assessed whether variable reduction strategies (i.e. factor analysis) were required. Logistic regression analysis was used to examine the association between household resources and child mortality, controlling for the covariates. In all of the analyses, survey weights were applied to account for the complex survey design.
Results: Women who resided in the urban area were more likely to report access to all the household resources measured than those in rural area, except kerosene lantern for which more rural women reported possession. In unadjusted comparisons, a significantly lower proportion of women who reported possession of each of the household resources reported child mortality with one exception. There was no significant difference in the proportion of women reporting child mortality by the reported number of sleeping rooms. Women who had primary, middle school and secondary or tertiary education had lower odds of reporting child mortality as compared to those who had no education. The odds ratios ranged from 0.78 [95% CI: 0.65, 0.94] for primary school to 0.55 [95% CI: 0.38, 0.78] for those who had secondary or tertiary education. In the model restricted to women living in rural areas, none of the household resources were associated with child mortality. In contrast, in the model restricted to women living in urban areas and towns, having refrigerator at the time of interview was associated with lower odds of reporting child mortality 0.63 [95% CI: 0.48, 0.83]; whereas, having a kerosene lantern [AOR=1.40; 95% CI: 1.06, 1.85] or flush toilet [AOR=1.84; [95% CI: 1.23, 2.75] was associated with higher odds of reporting child mortality.
Conclusion: Possession of a refrigerator may play a role in child mortality, particularly in urban areas. This finding may reflect unmeasured SES or the importance of access to refrigeration in preventing diarrheal disease or other proximal causes of child mortality in sub-Saharan Africa. Further studies should focus on understanding the association of household resources and child mortality in rural areas and the contribution of community socio-economic attributes to child mortality.
See more of: Research Sessions: Oral Paper & Posters