Menu
Menu
Menu
Menu
Globally, approximately 5.9 million children under 5 years of age died in 2015, a reduction of over 50% since 1990. Millennium Development Goal 4 established the goal of reducing child mortality by two-thirds by 2015. Multiple countries have surpassed this goal; however, regional and within-country inequities exist. We sought to study determinants of health-care utilization among children 6-59 months of age with fever, diarrhea, and respiratory symptoms in Zambézia Province, Mozambique. We conducted a population-based cross-sectional survey of female heads of household between April and May 2014. Mobile teams conducted interviews in 262 enumeration areas, with three distinct districts being oversampled for improved precision. Descriptive statistics and logistic regression using Stata 13.1 and R 3.2.2 were used to examine factors associated with health-care utilization. A total of 2,317 children were evaluated in this study. Mothers’ median age was 26 years, whereas child median age was 24 months. The proportion of children reporting fever, diarrhea, or respiratory illness in the prior 30 days was 44%, 22%, and 22%, respectively. Health-care utilization varied with 65% seeking health care for fever, compared with 57% for diarrhea and 25% for respiratory illness. In multivariable logistic regression, the characteristics most associated with health-care utilization across illnesses were delivery of last child at a facility, higher maternal education, and household ownership of a radio. The decision or ability to use health care is a multifaceted behavior swayed by societal norms, values, socioeconomics, and perceived need. Recognizing the predictors of a particular population may offer useful information to increase uptake in health-care services.
We used cross-sectional data from a household survey that was conducted as part of the Ogumaniha project (meaning “united for a common purpose” in the local Echuabo language), funded under the U.S. Agency for International Development (USAID) Strengthening Communities through Integrated Programming award. Ogumaniha’s goal was to reduce poverty and improve the health of people living in Zambézia Province through cohesive community-based programming.32 A strong monitoring system and project evaluation based on performance indicators were agreed upon with USAID and the provincial government and were central to Ogumaniha’s design. This study is an extension of the monitoring and evaluation component of the Ogumaniha project and analyzes cross-sectional survey data collected at the project’s conclusion (April and May 2014). The survey tool, developed by an interdisciplinary team of researchers, is a 500-item questionnaire covering eight dimensions and includes many questions borrowed from previous national surveys in Mozambique such as the Demographic Health Survey and Multiple Indicator Cluster Survey. The survey was designed to gather information from the female head of household on topics such as household demographics; economic status; health knowledge, attitudes, and practices; access to health services and products; access to improved water and sanitation; nutritional intake; and others. The female head of household was selected because, in Zambézia Province, she is thought to be the person most familiar with the health and caretaking of all household members (e.g., nutrition, water and sanitation, health events, and health-care access). There was a potential for bias in the case of polygamous families because the principal or eldest wife was selected, whereas the younger wives and their children may have been even more disadvantaged. A child health and immunization module collected child-specific information on up to two children 6–59 months of age per household. Mobile survey teams conducted interviews in 262 enumeration areas (EA). EA were selected in two-stage cluster sampling using the national census results as a sampling frame; first, we stratified by urban/rural grouping, and then sampled with probability proportional to size. Three districts (Alto Molócuè, Namacurra, and Morrumbala) were oversampled to improve accuracy for the primary project evaluation at reduced cost. A smaller, less dense sample of the remaining 14 districts was collected to provide survey-weighted estimates representative of the entire province. Further details about the sampling methodology, electronic data collection using mobile phones, Open Data Kit, and management protocols have been published elsewhere.32 In an effort to identify independent predictors of health-care utilization for children 6–59 months of age, we focused on the association between utilization and maternal education, household income, decision-making authority of the female head of household, and distance (in minutes) to a health facility. Covariates were selected a priori based on extensive literature review, and include child’s age and sex, respondent’s age, marital status, whether the respondent understands Portuguese, household size, mode of transport to a health-care facility, rural or urban location, ownership of a radio, and whether the respondent had delivered her last child in a health facility. Descriptive statistics were calculated for continuous variables as weighted estimates of median (interquartile range [IQR]) and for categorical variables as weighted percentages, with each observation being weighted by the inverse of the household or child sampling probability. Outcomes of interest included health facility utilization after three common childhood illnesses: fever, diarrhea, and respiratory illness (symptoms included cough, difficulty breathing, or fast/shallow breathing). No definition for fever or diarrhea was provided by the study interviewer in order for the participant to respond based on their understanding of these conditions. Multivariable logistic regression models were used with robust covariance to account for clustering of children within households and EA. In each model, we disaggregated children with illness reported in the past 30 days and modeled the probability of health facility utilization. The significance level for all testing was two-sided and set at 0.05. If there was evidence of nonlinearity (P < 0.10) of continuous covariates with the log-odds of health facility utilization, then that variable was modeled using a restricted cubic spline. Multiple imputation was used to account for missing survey responses in covariates. We used the Hmisc package in R which used predictive mean matching to take random draws from imputation models; 10 imputation data sets were used in the analysis.33 Data analysis was conducted in Stata 13.1 and R version 3.2.2.34,35 Household participation in the survey was voluntary and without incentives. Written informed consent was obtained from the female head of household before the interview and child measurements were conducted. The study was approved by the Inter-institutional Bioethics Committee for Health of Zambézia Province (CIBS-Zambézia), Mozambique, and the Vanderbilt University Institutional Review Board.
N/A
