Contribution of community-based newborn health promotion to reducing inequities in healthy newborn care practices and knowledge: Evidence of improvement from a three-district pilot program in Malawi

Background: Inequities in both health status and coverage of health services are considered important barriers to achieving Millennium Development Goal 4. Community-based health promotion is a strategy that is believed to reduce inequities in rural low-income settings. This paper examines the contributions of community-based programming to improving the equity of newborn health in three districts in Malawi. Methods. This study is a before-and-after evaluation of Malawi’s Community-Based Maternal and Newborn Care (CBMNC) program, a package of facility and community-based interventions to improve newborn health. Health Surveillance Assistants (HSAs) within the catchment area of 14 health facilities were trained to make pregnancy and postnatal home visits to promote healthy behaviors and assess women and newborns for danger signs requiring referral to a facility. “Core groups” of community volunteers were also trained to raise awareness about recommended newborn care practices. Baseline and endline household surveys measured the coverage of the intervention and targeted health behaviors for this before-and-after evaluation. Wealth indices were constructed using household asset data and concentration indices were compared between baseline and endline for each indicator. Results: The HSAs trained in the intervention reached 36.7% of women with a pregnancy home visit and 10.9% of women with a postnatal home visit within three days of delivery. Coverage of the intervention was slightly inequitable, with richer households more likely to receive one or two pregnancy home visits (concentration indices (CI) of 0.0786 and 0.0960), but not significantly more likely to receive a postnatal visit or know of a core group. Despite modest coverage levels for the intervention, health equity improved significantly over the study period for several indicators. Greater improvements in inequities were observed for knowledge indicators than for coverage of routine health services. At endline, a greater proportion of women from the poorest quintile knew three or more danger signs for pregnancy, delivery, and postpartum mothers than did women from the least poor quintile (change in CI: -0.1704, -0.2464, and -0.4166, respectively; p < 0.05). Equity also significantly improved for coverage of some health behaviors, including delivery at a health facility (change in CI: -0.0591), breastfeeding within the first hour (-0.0379), and delayed bathing (-0.0405). Conclusions: Although these results indicate promising improvements for newborn health in Malawi, the extent to which the CBMNC program contributed to these improvements in coverage and equity are not known. The strategies through which community-based programs are implemented likely play an important role in their ability to improve equity, and further research and program monitoring are needed to ensure that the poorest households are reached by community-based health programs. © 2013 Callaghan-Koru et al.; licensee BioMed Central Ltd. A land-locked East African country of over 13 million people [28], Malawi is ranked among the ten poorest countries in the world. The Gross National Income per capita is $290 [29] and development assistance accounts for a large part of the Malawi’s economy at $49 per capita [30]. The great majority of Malawi’s population can be considered poor by global standards, with over 90 percent of the population living on less than $2 per day [29]. Because of the overall poverty, general inequality measures for Malawi are not high when compared with many of Malawi’s richer neighbors [30]. However, there is evidence of inequities in several maternal and child health indicators, including skilled attendance at delivery (poorest: 65%; least poor: 90%), postnatal care for women (poorest: 38%; least poor: 52%), child mortality, and care seeking and nutritional status for children under age five [27]. The community-based maternal and newborn care pilot program was implemented in selected areas of three rural districts in Malawi—Thyolo, Dowa, and Chitipa. The major occupation in rural areas of Malawi is farming, and 64% of the population is considered literate [28]. Although the area is rural, the population is dense with an average of 139 persons per square kilometer [28]. Nationally, the neonatal mortality rate for Malawi is estimated at 31 deaths per 1,000 live births and under-five mortality is 112 per 1,000 [27]. A key challenge to improving maternal and newborn health in Malawi is access to quality health services. It is estimated that only 54% of the population resides within a 5-km radius of a health facility [31]. Patients who reach health facilities encounter a critical shortage of health staff in Malawi, where there are approximately 7 doctors and 37 nurses per 100,000 population [32], more than 20 times fewer than in the United States [33]. The expansion of Malawi’s existing cadre of community-based health workers, known as Health Surveillance Assistants (HSAs), is considered an important strategy for improving access to primary care for rural populations [31,34]. HSAs serve rural communities with a target catchment area of 1,000 population, although many HSA catchment populations exceed this target and can reach more than 2,500 population. MOH policy is that HSAs should be recruited from within the district where they will work, but many have been recruited centrally and are not from the communities where they are posted. HSAs’ main responsibility is to provide health education and hygiene promotion, and they receive 10 to 12 weeks of training for this role. In addition, they often receive in-service training for various vertical program activities, including community case management of childhood illness, family planning, tuberculosis drug distribution, water and sanitation, immunization and growth monitoring for children under age 5, providing community therapeutic feeding, voluntary counseling and testing for HIV/AIDS, and following up patients on antiretroviral therapy at community level [35]. Prior to this intervention, HSAs were not trained to provide home visits or counseling on newborn care. This study is a before-and-after evaluation of Malawi’s Community-Based Maternal and Newborn Care (CBMNC) program, a package of facility and community-based interventions to improve newborn health implemented by the Malawi Ministry of Health (MOH) with support from Save the Children’s SNL program and UNICEF. Three districts—Thyolo, Dowa, and Chitipa—were chosen for the pilot study in collaboration with the MOH. These districts represent Malawi’s three geographic regions (north, central, and south), and were also selected based on overall representation in relation to child and neonatal mortality indicators, progress in implementation of the Accelerated Child Survival and Development and Integrated Management of Childhood Illness programs, as well as district interest in participation. Seven health facilities in Dowa, 7 in Chitipa, and 8 facilities in Thyolo, were selected for the intervention, with a total catchment population of approximately 711,000. The selection criteria for participating health facilities within each district included the size of the facility’s catchment area (with a preference for larger catchment populations), the presence of at least two health workers on staff, and the interest of the facility staff in participation in the pilot. Among the 24 intervention facilities, eight were district or rural community hospitals and 14 were health centers. A decision was made not to include comparison areas in the study due to a simultaneous rapid scale-up of the same CBMNC package by the MOH and multiple partners across the three districts. At the facility level, health workers received a 21-day in-service training on integrated maternal and newborn care (IMNC), which included modules on newborn resuscitation, basic newborn care, (thermal protection, hygienic cord care, breastfeeding support, etc.), kangaroo mother care for low birth weight babies, and identification and referral of newborns with signs of infection. Ninety-six facility staff completed training on IMNC in the pilot areas between July and September 2009. At the community level, a total of 622 HSAs in the pilot areas received a 10-day training in Community Based Maternal and Newborn Care (CBMNC) with modules on care during pregnancy and delivery, immediate newborn care, postnatal care for mother and newborn, breastfeeding, identification of newborn danger signs, management of low birth weight babies and conducting home visits. Following training, HSAs were instructed to create a register of women of childbearing age in their catchment areas and update the list every two months through home visits and discussions with community leaders to identify current pregnancies. After identifying pregnant women, HSAs were expected to make 3 home visits during pregnancy (one per trimester) and 3 postnatal home visits for mothers and newborns (on days 1, 3 and 8 for all births). During home visits, HSAs promoted a package of health behaviors (see Table 1) and assessed women and newborns for danger signs requiring referral to a health facility. Seventy-five percent of the HSAs received additional community mobilization training to establish “core groups” of community members that would conduct health education, generate demand for services, and inform HSAs of pregnancies and deliveries. Trainings of HSAs in pilot areas started in July 2008 and were completed by October 2010. The Ministry of Health, Save the Children, and UNICEF conducted quarterly supervision of the CBMNC package to reinforce training and implementation. HSAs were selected for supervision visits based on assessments of their performance and reporting. During visits to HSAs’ communities, supervisors mentored HSAs in counseling during home visits and correctly completing registers. The District Executive Committees and District Development Committees were oriented to the program by the District Health Office, while HSAs were responsible for sensitizing the communities in which they worked to the new service through home visits and their meetings with Village Health Committees. Content of prenatal and postnatal home visits by HSAs A household survey was conducted at baseline and endline to measure the coverage of facility-based maternal and newborn care services, home visits by HSAs (endline only), knowledge about maternal and newborn danger signs, and newborn care practices. The baseline survey took place prior to the start of intervention activities in November-December 2007, and the endline survey was conducted in May-June 2011. Each survey included a random sample of 900 women, allowing us to detect, for the majority of indicators, a difference of at least 10 percentage points between baseline and endline with the pooled sample across districts and 20 percentage points within each district, accounting for a design effect of 2 with 80% power [36]. Households were selected for the survey following a two-stage cluster sampling design adapted from the Expanded Program on Immunization (EPI) survey methodology [37]. In the first stage, 90 enumeration areas, 30 per district, were randomly sampled with probability proportional to size. Within each cluster, teams interviewed 10 households with at least one woman 15–49 years of age who had a live birth in the last 12 months. Households were selected following the EPI random walk method, whereby teams randomly selected a starting household in the center of the cluster, and after choosing a random direction, proceeded to each adjacent household until identifying an eligible woman [38]. In the case that a selected household had more than one eligible woman, one woman was randomly selected to complete the interview. The sampling frame for clusters in the baseline survey was composed of all census enumeration areas within the three districts. For the endline survey, the sampling frame was restricted to only the census enumeration areas corresponding with the catchment areas of the 458 HSAs who had been trained by December 2009, to allow at least 15 months of program implementation. In both baseline and endline surveys, the sample was evenly divided between districts with 30 clusters per district. Data collection was conducted by 12 experienced interviewers who had either completed a university education or were university students. Data collectors were trained for 3 days and data collection took 13 days for both baseline and endline surveys. Each data collection team had a supervisor responsible for checking questionnaires for correctness and completeness and overseeing sampling and other survey procedures in the field. A full-time survey coordinator provided supervision and quality assurance for the overall implementation of the survey and SNL staff provided additional oversight and monitoring throughout the survey process. At baseline, data were entered into MS Access and at endline data were double entered into SPSS and discrepancies were reconciled with reference to the original survey form. All variables on coverage, knowledge, and practices were measured dichotomously. Postnatal home visits by HSAs were considered to be within three days if reported to have occurred within 72 hours after birth or on day 0, 1, 2, or 3. Skilled providers for antenatal care included doctors, clinical officers, nurses and midwives. Immediate breastfeeding was defined as initiation ≤1 hour after birth. Delayed bathing was defined as the newborn’s first bath given ≥6 hours after birth. Skin-to-skin contact was based on whether the mother reported the baby was placed in skin-to-skin contact with her 'as soon as s/he was born’. Records that were missing information for a specific indicator were excluded from the calculation of that indicator. Full indicator definitions are provided in Additional file 1: Table S1. In order to assign households to wealth quartiles, an asset score was generated using principal components analysis. All assets included in the index are presented in Table 2. The first principal component accounted for 18% of variation at baseline and 17% of variation at endline. Indicators were calculated by wealth quintile separately for baseline and endline with robust standard errors adjusting for clustering [39]. Background characteristics and assets included in the sample We calculated concentration indices for each indicator of interest as a single measure of inequality. The concentration index (CI) measures the area between the concentration curve and the line of perfect equality [40]. CI values range between -1 and +1; a positive value indicates inequality favoring the rich, a negative value indicates inequality favoring the poor, and a CI closer to zero indicates near perfect equality. Concentration indices have been increasingly used to assess inequalities in maternal and child health outcomes [41,42]. We generated concentration curves from the wealth scores using the generalized Lorenz curve approach. From these curves we derived the CIs using regression models with robust standard errors to account for clustering [42]. The change in CI between baseline and endline for each indicator was calculated and then tested using the t-test. All analyses were conducted in Stata 11 [43]. Oral informed consent was obtained from all survey respondents. Ethical review and approval was provided by the Malawi National Health Sciences Research Committee (protocol number 473).