Study design and participants
The data presented were collected as part of the baseline evaluation for SM2015, which was established to address the health issues faced by the poorest quintile of the population in El Salvador, Guatemala, Honduras, Nicaragua, Belize, Costa Rica, Panama, and Mexico. Surveys were conducted in households and health facilities in each country. We conducted our own censuses within each selected primary sampling unit, a segment of approximately 150 households, in order to identify eligible households. This ensured we used the correct denominator in indicator estimation and allowed us to account for the potential movement of the population in the study areas since the last national census. Among eligible households, a randomly selected subset was chosen for the household survey.
The household survey had three components. A household questionnaire captured information on assets, wealth, and characteristics of the home. A maternal health questionnaire collected demographic, health behavior and reproductive health information on women of reproductive age (15–49 years). A child health questionnaire on health, diet, and vaccination history was completed for children 0- to 59-months old. Physical measurements and anemia tests were conducted for children.
To assess maternal education, women were asked if they have ever attended school and if they have ever completed a literacy course. Women who responded that they have attended school were asked about the highest level of schooling that they attained: primary (elementary school), secondary (middle school), preparatory or university. To assess reproductive health indicators, women were asked to answer questions about their birth history in the last five years. For each birth, women were asked if they had received at least one antenatal care (ANC) visit. Women who had received at least one ANC visit were further asked about the number of visits attended. For each ANC visit, women were asked to indicate the person who provided them with care. Interviewers were instructed to have women specify the most qualified attendant during each of these visits. To assess skilled birth attendance (SBA) and in-facility delivery, women were asked to identify each person who provided them with attention during birth and to indicate where they gave birth. Women were also asked if they used any family planning method after each birth. Women who had used family planning, were asked what method was used and how soon after birth did they start using this method.
To assess post-natal care for each child in the last five years, women were asked if the child was examined by a health provider at some time after birth and to indicate how many hours, days, or weeks after birth the child had a first post-natal care. In reference to each child, women were then asked if they breastfed at least one time. To assess early initiation, women were asked how soon after birth they breastfed for the first time. To assess exclusive breastfeeding, questions were asked about a 24-hour dietary recall for each child born in the six months prior to the date of the survey.
To assess immunization coverage, interviewers reviewed child vaccination cards and recorded the vaccines and dates marked on the cards for each child under five years old. Vaccination recall was assessed by asking women to indicate all of the vaccines that each child had received. Questions about vaccines were asked in adherence to national vaccination schemes for each country.
The SM2015 surveys were conducted using a computer‐assisted personal interview (CAPI) by trained interviewers. Data were continuously monitored by the Institute for Health Metrics and Evaluation (IHME). All data were collected after obtaining informed consent. The field surveyors explained the purpose of this study to participants. Then, written informed consent was obtained from all study participants who agreed to participate prior to data collection. The study received institutional review board (IRB) approval from the University of Washington, partnering data collection agencies, and the Ministry of Health in each country to ensure that the data were collected in an appropriate and ethical manner. Baseline surveys were conducted from 1 March 2011 to 31 August 2013. We used Stata 12.1 and Stata 13.1 for the analyses. All estimates are computed using survey weights, unless otherwise noted. Additional details on SM2015 design, sampling, methodology, and implementation are available elsewhere [14].
Definitions
Household monthly expenditure was computed as the sum of reported weekly, monthly, or semi-annual expenditures after being converted to monthly totals: food, alcohol and tobacco, education-related expenses, household utilities; clothing and footwear, transportation, communication, out-of-pocket health care costs, social security premiums, private insurance premiums, and associated health care costs. Households that spent 25 % or more on health care were considered to have incurred catastrophic health expenditure in the past month.
Key child health indicators were also computed. Adherence to national vaccination schemes for all vaccines and for measles, mumps, and rubella (MMR) were estimated based on caregiver recall and vaccination card information. Anthropometric measurements of children were used to calculate the prevalence of wasting and stunting, defined as −2 standard deviations below the mean height-for-age and weight-for-height according to World Health Organization (WHO) criteria, respectively [15]. Additionally, we assessed whether children with signs of diarrhea in the past two weeks received proper oral rehydration salt (ORS) treatments.
Reproductive health indicators included services received during the antenatal period, delivery care, and breastfeeding. Among deliveries in the two years prior to the survey, we estimated coverage of ANC and SBA with a doctor or nurse. We focused on SBA and in-facility deliveries as they are strongly associated with a reduction in maternal and infant mortality [16]. Exclusive breastfeeding during the first six months of life was estimated using a 24-hour dietary recall; all children 0- to 5-months old who consumed exclusively breast milk, as reported by a caregiver, were considered adherent.
A sample of mother-child pairs was used to compute a composite coverage score of select maternal, newborn, and child health indicators. Data were linked for each child, mother, and corresponding birth history and restricted to each woman’s youngest child born in the two years prior to the survey. This score is equal to the summed presence of eight select health indicators: one ANC visit with a skilled attendant, four ANC visits with a skilled attendant, SBA, use of ORS treatment for recent diarrhea, initiation of breastfeeding within 24 hours of birth, complete childhood vaccination based on age and national scheme, absence of stunting, and absence of wasting. Segment-level coverage was computed for each subcomponent in order to calculate correlation with segment average wealth. The highest possible score is 8 and was converted to a proportion for some analyses. For each mother/child pair we present the health indicators as a continuum of care (ANC1, ANC4, SBA, in-facility delivery, breastfeeding initiation within one hour, skilled post-natal care for baby within one week, use of modern contraceptive, and complete immunization) by education and expenditure.
Statistical analyses
The surveys were conducted in communities that were designated as the poorest areas in each country. Even so, substantial income variability was found within these communities, with monthly household expenditure per month ranging from USD 3 to 1,200 per month. In order to examine the variation of wealth and health indicators within our sample, we computed prevalence and uptake of select indicators at the segment-, municipality-, and country-level. In addition, the sample was stratified by maternal education level and by household expenditure quintile to examine variation by country in the uptake of select health indicators in the continuum of maternal and child care. Average uptake of health-seeking behaviors from prenatal through early childhood care is reported for each subgroup.
We used multivariable logistic regression to measure the association between delivering in a health facility and select household and maternal characteristics. Data for each mother-child pair were pooled across countries, and a country-specific fixed effect was added to account for different patterns of SBA between countries. Model 1 covariates include within-country household expenditure quintile, asset index, attained maternal education level (no education, primary education, or secondary or higher education), maternal age in years at the time of the survey, and maternal parity. An alternative asset-based metric of wealth was computed as a factor score and the results are similar (data available upon request). Model 2 incorporated additional covariates of household characteristics and women’s autonomy, including household size, head of household gender, maternal occupational status, and maternal marital status. Model 3 included information on potential barriers to care: whether the mother is insured, travel time to the usual health facility, whether the mother received counseling from a community health worker during the past three months, whether the mother was exposed to media (newspaper, radio, or television) during the past week, and mother-reported barriers to care. If travel time to the usual health facility is missing, we used the travel time to the closest health facility. If that too was missing, we used the median travel time to the usual facility among households in that segment. Barriers to care were reported among women who had a recent illness but did not seek medical care, so dummy variables were added to reflect women who were not recently ill or were ill but did seek care.
Less than 7 % of observations were excluded from the regression analysis because they lacked information about one or more of the independent variables. We used self-reported barriers to care in our model to account for potential bottlenecks from the women’s side in seeking required health care. Alternative models using a dependent variable of SBA and in-facility delivery with SBA were also conducted, but the results were similar (data available upon request). In order to assess whether health seeking patterns varied by country in relation to health insurance, we tested for interaction between country and health insurance in our models. We found no statistically significant interaction between the two variables.
Role of the funding source
The funders of this study had no role in study design, data collection, data analysis, interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.