Background: Otitis media (OM), a prevalent middle ear inflammation, is a leading cause of medical consultations and antibiotic prescriptions among children globally, imposing significant financial burdens on healthcare systems. Untreated acute OM can progress to chronic suppurative OM, a primary cause of pediatric hearing loss and potential cognitive delays, affecting speech acquisition. Environmental factors such as air pollution and socioeconomic conditions contribute to OM's multifactorial etiology, yet research on these associations, especially in low- and middle-income countries (LMICs), remains limited. The impact of preventive measures such as pneumococcal conjugate vaccines (PCVs), known to reduce AOM incidence and subsequent hearing loss, on cognitive development requires further investigation. This dissertation addresses these gaps through an aggregated analysis integrating spatial data, examining OM prevalence, hearing loss, and cognitive development among children in Bohol, Philippines, while identifying demographic, environmental, and socioeconomic correlates.
Methods: Utilizing data from a follow-up assessment of a randomized controlled trial (ISRCTN 62323832) evaluating an 11-valent PCV in Bohol, this dissertation integrated demographic, socioeconomic, and health data collected via parental questionnaires, ear assessments, and cognitive tests. This study redefined neighborhoods using a square grid methodology to standardize geographic units. Then I aggregated participant data at the neighborhood level and created maps to visualize the spatial distribution of each outcome. I assessed global spatial autocorrelation using Moran’s I index, supplemented by Local Indicators of Spatial Association (LISA) maps to pinpoint local clusters. The three specific aims were: (1) to identify factors associated with OM prevalence using multivariable ordinary least squares (OLS) and spatial regression models; and to assess spatial heterogeneity in associations using geographically weighted regression (GWR); (2) to examine characteristics associated with hearing loss incidence through Poisson regression and evaluate spatial non-stationarity of the associations with GWR; and (3) to assess factors associated with cognitive development and regional variations in the associations using OLS regression and GWR.
Results: Areal OM prevalence averaged 41% and exhibited significant spatial clustering (Moran’s I=0.31, p<0.001). OM prevalence was associated with household size, maternal insurance coverage, breastfeeding practices, and exposure to traffic-related air pollution. Neighborhood-level hearing loss prevalence averaged 8.5% without significant global spatial autocorrelation (Moran’s I=0.048, p=0.26), with associations found for OM prevalence, distance from hospitals, household smoking exposure, and household wealth. Cognitive development showed significant global spatial autocorrelation (Moran’s I=0.17, p<0.001), and was associated with OM prevalence, household wealth, and parental education levels. GWR identified regional variations in these associations, emphasizing the spatial heterogeneity. PCV coverage was not statistically significantly associated with any outcomes on neighborhood level.
Discussion: The dissertation explores critical health and developmental outcomes among a cohort of children in Bohol, Philippines, within the context of an LMIC. Aim 1 reveals the role of reducing respiratory pathogen exposure, enhancing breastfeeding rates, and promoting health insurance coverage in mitigating OM prevalence. Surprisingly, proximity to healthcare facilities did not significantly correlate with OM prevalence, highlighting multifaceted influences of healthcare accessibility. Aim 2 delves into pediatric hearing loss, emphasizing its associations with poor healthcare access and household smoking, with socioeconomic disparities showing the protective effect of economic advantage. Aim 3 demonstrates that higher household wealth and parental education correlate with improved cognitive outcomes. The spatial approach, integrating GWR, illuminates localized variations in each outcome, suggesting tailored public health strategies are crucial. While the cross-sectional nature of this study limits causal inference, the findings advocate for comprehensive interventions addressing health and socioeconomic disparities. These insights, although specific to the study population, provide valuable context to inform public health strategies in other LMICs. The methodologies employed offer a framework guiding similar public health interventions across diverse LMIC settings, facilitating targeted and effective approaches to enhance child health and development outcomes. This dissertation advocates for nuanced, community-specific policies and practices to mitigate health disparities, thereby contributing to the broader effort to improve global child health.