Reaching Mothers and Babies with Early Postnatal Home Visits: The Implementation Realities of Achieving High Coverage in Large-Scale Programs

Background:Nearly half of births in low-income countries occur without a skilled attendant, and even fewer mothers and babies have postnatal contact with providers who can deliver preventive or curative services that save lives. Community-based maternal and newborn care programs with postnatal home visits have been tested in Bangladesh, Malawi, and Nepal. This paper examines coverage and content of home visits in pilot areas and factors associated with receipt of postnatal visits.Methods:Using data from cross-sectional surveys of women with live births (Bangladesh 398, Malawi: 900, Nepal: 615), generalized linear models were used to assess the strength of association between three factors – receipt of home visits during pregnancy, birth place, birth notification – and receipt of home visits within three days after birth. Meta-analytic techniques were used to generate pooled relative risks for each factor adjusting for other independent variables, maternal age, and education.Findings:The proportion of mothers and newborns receiving home visits within three days after birth was 57% in Bangladesh, 11% in Malawi, and 50% in Nepal. Mothers and newborns were more likely to receive a postnatal home visit within three days if the mother received at least one home visit during pregnancy (OR2.18, CI1.46-3.25), the birth occurred outside a facility (OR1.48, CI1.28-1.73), and the mother reported a CHW was notified of the birth (OR2.66, CI1.40-5.08). Checking the cord was the most frequently reported action; most mothers reported at least one action for newborns.Conclusions:Reaching mothers and babies with home visits during pregnancy and within three days after birth is achievable using existing community health systems if workers are available; linked to communities; and receive training, supplies, and supervision. In all settings, programs must evaluate what community delivery systems can handle and how to best utilize them to improve postnatal care access. © 2013 Sitrin et al.

In each country, districts were selected for SNL-supported implementation in partnership with governments. Implementation areas were four rural unions in Faridpur District, Bangladesh with a population of 98,000 people; portions of three districts in Malawi (Chitipa, Dowa, Thyolo) with a population of 711,000; and Bardiya District in the Terai area of Nepal with a population of 460,000. Programs were designed to fit into health systems and used existing government authorized and trained workers or volunteers, supervisory and monitoring systems, facilities, and equipment and supplies [18], [19], [20]. SNL and other partners provided training in maternal and newborn care, add-on supplies, and support for supervision, monitoring, and community engagement. The existing community cadres were all providing some care to mothers or children, but there were important differences across countries in terms of variations in their characteristics and how they were incentivized, outlined in Table 2. In Bangladesh, three cadres of community workers were trained to deliver home visits for women and newborns; all three were actively delivering health services under different government programs. Community workers in Bangladesh and Malawi were government salaried employees or volunteers paid regular stipends, while Nepal used volunteers incentivized with a performance-based scheme [21]. In all countries, CHWs were supposed to live in their catchment areas, and they were recruited from communities in Bangladesh and Nepal. However, community workers in Malawi were assigned posts by District Health Teams and only 47% interviewed in pilot districts reported living in their catchment areas [22]. For the new programs, CHWs received in-service training on maternal and newborn care, ranging from five to nine days. CHWs were trained to identify pregnancies, visit women during pregnancy, and make three or four postnatal home visits, including a visit on the first day after birth (only for home births in Malawi) and two more visits within the first seven or eight days. Home visit schedules and content are shown in Table 2. Approximately 100 CHWs in Bangladesh, 600 in Malawi, and 850 in Nepal were trained in pilot areas. In all countries, programs aimed to strengthen CHWs’ capacity to identify pregnancies through training and supervision. In Nepal, Family Community Health Volunteers (FCHVs) identified pregnant women during existing mothers’ group meetings and household visits (visit frequency depended on needs identified by mothers’ groups). In Bangladesh, Family Welfare Assistants (FWAs) identified pregnant women during routine, monthly household visits (though visits often occurred less frequently) and during regular outreach clinic sessions (each FWA has about 8 per month). Health Assistants (HAs) and Community Nutrition Promoters (CNPs) identified pregnant women during routine activities, such as family planning counseling. In Malawi, Health Surveillance Assistants (HSAs) made lists of women of child bearing age and were instructed to update the list every two months to identify pregnancies. Community “core groups” were also encouraged to report pregnancies to HSAs. To facilitate early postnatal visits, pregnancy visits included counseling on when and how to notify the CHW about the birth. In Nepal, FCHVs were trained to attend home deliveries or accompany women to facilities, so pregnant women were instructed to call the FCHV at onset of labor. The Bangladesh program designed a mobile phone birth notification system and CHWs’ phone numbers were written on Mother’s Cards kept by families; families were instructed to notify CHWs after delivery. In Malawi, HSAs were instructed to encourage families to notify them after delivery, but counseling did not include specific notification instructions. Content of home visits was similar across countries and included promotion of optimal maternal and newborn care and routine facility services, counseling on danger signs and care-seeking for mother and baby, screening for newborn danger signs by a physical assessment including checking the baby’s temperature and breathing and weighing the baby (if not previously weighed at a facility in Malawi and Nepal), and referral to a health facility when needed. In Nepal, some curative care was included; FCHVs administered an oral antibiotic for danger signs suggesting possible severe bacterial infection and referred the baby to health posts for a seven day course of injectable antibiotic. CHWs were equipped with counseling cards, thermometers, scales, registers to record visits, and other supplies. In Nepal, FCHVs were given co-trimoxazole to treat suspected infection and a bag and mask and DeeLee suction to resuscitate babies that did not breathe immediately after birth. Add-on supplies were provided by SNL or partners. In all countries, CHW supervisors received training to oversee home visits. In addition, SNL staff were involved in supervision visits, done in partnership with MOH staff when possible, to monitor implementation progress and identify gaps or problems. Program monitoring used existing reporting systems with the addition of new tools to collect maternal and newborn data. In Bangladesh, micro-planning meetings coordinated different CHW cadres by analyzing performance and mapping pregnancies to share home visit responsibilities. In addition, programs included efforts to increase community engagement and support. In Bangladesh, community leaders attended orientation sessions to learn about the program. In Malawi, HSAs organized “core groups” for planning and decision-making. In Nepal, FCHVs discussed maternal and newborn health with existing mothers’ groups. Start of full implementation to final data collection was 14 months in Bangladesh (April 2009–June 2010), 12 months in Malawi (June 2010–June 2011), and 17 months in Nepal (January 2010–June 2011). Home visits may have started earlier in some areas, depending when training was conducted. Data were from cross-sectional household surveys of women with a live birth in the previous 12 months (3–12 months in Bangladesh). Sample size calculations were based on expected changes in key indicators. In Bangladesh, pilot unions were divided into 12 clusters; six were selected randomly. Using probability proportionate to size (PPS) procedures, live births were randomly selected from lists of all live births collected prior to the survey. In Malawi, thirty clusters in each district were selected proportional to size using the 2008 population census (sampling restricted to villages where the program was implemented). Clusters were divided into segments, one segment selected, all houses numbered within the segment, and an index house randomly selected. Eligible women (age 15–49 with a live birth within the last 12 months) were interviewed. The next closest household was visited until 10 eligible women were interviewed per cluster. In Nepal, thirty clusters were selected proportional to size using the 2001 population census (excluding urban municipalities). Household lists were developed with key informants; the first household was selected randomly and subsequent households identified by spinning a bottle and all households within 20 meters of either side of a straight line were included until 21 eligible women (had a birth in the last 12 months) were interviewed. Only Nepal interviewed women with stillbirths; these cases were excluded from analysis. Response rates were 90% in Bangladesh, 100% in Malawi, and 91% in Nepal. Data were collected by teams of trained interviewers, and each team had a data quality supervisor. Data can be made available upon request. The objective of this analysis was to examine the strength of association between three factors of programmatic interest –mother received a home visit during pregnancy, birth place, and CHW notified of the birth – and receipt of postnatal home visits from CHWs. The dependent variable was receipt of a postnatal home visit within three days after birth from a CHW. In Bangladesh and Malawi, women were asked about home visits before and after birth. In Nepal, women were not asked specifically about postnatal home visits. Instead, the woman was asked if she and the newborn were checked before discharge (or before the birth attendant left after a home birth), along with questions about the location and provider of the first two post-discharge checks on her health and the first three post-discharge checks on the newborn’s health. Since the woman was asked about only the first two post-discharge checks on her health, the percentage of women visited at home within three days after the birth appears lower compared to the percentage of newborns visited (41.3% versus 49.6%). Therefore, questions on post-discharge care for the baby were used to calculate the dependent variable in Nepal. A visit occurred within three days if it was reportedly done within 72 hours after birth or on day 0, 1, 2, or 3. Currently, the global indicator for postnatal care includes care received within two days from any provider [5], [23]. However, this analysis focuses on home visits, and national policy in Malawi requires a postnatal home visit within three days for facility births. We used a visit within three days in all countries for comparability. The dependent variable was measured dichotomously, where 0 indicated no postnatal home visit within three days from a CHW and 1 indicated receipt of a visit. The main independent variables, measured dichotomously, were defined as receipt of at least one home visit from a CHW during pregnancy (yes/no), place of birth (facility/non-facility), and whether a CHW was notified of the birth (yes/no). Questions on birth notification varied due to programmatic differences. In Bangladesh and Malawi, birth notification was based on whether women reported the CHW was notified of the birth. Since we examine the association between birth notification and receipt of home visits within 3 days, we classified birth notification as ‘no’ if the mother reported the CHW was notified more than 3 days after the birth. In Nepal, FCHVs were expected to attend deliveries, so birth notification was based on whether women reported that an FCHV was called at onset of labor. All analyses controlled for maternal age (<20, 20–29, 30+) and maternal education (any vs. none). Maternal age was made into a categorical variable because there was not a linear relationship between age and the log risk of receipt of a postnatal visit. The newborn content of postnatal home visits was also analyzed, using standard metrics [24]: 1) checked the cord, 2) counseled on breastfeeding (including demonstration or observation), 3) checked the baby’s temperature, 4) weighed the baby, and 5) counseled on newborn danger signs. Questions about maternal content were not consistent across countries, and women in Nepal were only asked what was done for her health during the first post-discharge check. In addition, content questions were prompted in Malawi and Nepal, but unprompted in Bangladesh. For comparability, only newborn content data from Malawi and Nepal are presented. Descriptive statistics were used to describe characteristics of interviewed women in the three countries. Pairwise correlation and collinearity among variables were evaluated. Generalized linear models were used to assess the relationship between the three primary independent variables – mother received a home visit during pregnancy, birth place, and birth notification – and receipt of a postnatal home visit within three days. Because data were collected within communities (i.e. clusters), not accounting for clustering may lead to incorrect statistical inference, such as underestimated standard errors and biased point estimates [25], [26]. We thus controlled for clustering in all analyses and appropriate standard error estimates were produced using the Taylor linearization method [27]. Relative risks (RR) and 95% confidence intervals (CI) were obtained. To understand whether similar trends were seen across countries, we generated adjusted pooled relative risk estimates of each primary independent variable on the main outcome. We used meta-analytic techniques, and pooled estimates were adjusted for maternal age and education as well as the other primary independent variables. The pooled relative risk estimates the average weighted association between the main outcome and each of the three independent variables [28]. We fitted both fixed effect models using the inverse-variance fixed-effect method and random effects models using the DerSimonian and Laird method to determine if there was heterogeneity among countries; that is, whether the true relationship between exposures of interest and the likelihood of a postnatal home visit is not the same in each country [29]. The resulting I2 statistic, the percentage of between-study heterogeneity attributable to variability in the true relationship, and the heterogeneity chi-squared test indicated there was statistically significant heterogeneity between countries for the relationship between birth notification and postnatal home visits (p = 0.000), so the fixed effect model was inappropriate for this case. There was no heterogeneity between countries for the relationship between pregnancy home visits or birth place and postnatal home visits, and the summary statistic and confidence intervals were the same whether fixed or random effects models were used. Therefore, we report results using the random effects models. We also did a sensitivity analysis for birth notification. Such meta-analyses have been used in the literature of cross-sectional studies [28]. STATA 11.0 was used for all analyses [30]. Programs were implementing national policy through routine systems. Ethical clearance was obtained from the Bangladesh Medical and Research Council and the National Health Sciences Research Committee in Malawi. Per approved protocols, women gave oral consent to participate in surveys due to high levels of illiteracy. To operationalize the National Neonatal Health Strategy [31], the Nepal Ministry of Health and Planning initiated the development of the Community-Based Newborn Care Package, which outlined the role of Save the Children in supporting the government to develop and test the package [32]. Data collection was completed as part of routine programmatic activities. Relevant district authorities granted permission and all respondents provided oral consent upon being informed of the purpose of data collection. Consent was documented by interviewers on the questionnaires.

N/A

• Dima Health