Effectiveness of programme approaches to improve the coverage of maternal nutrition interventions in South Asia

Abstract The nutritional status of women before pregnancy, during pregnancy, and after delivery has far reaching consequences for maternal health and child survival, growth, and development. In South Asia, the high prevalence of short stature, thinness, and anaemia among women of reproductive age underlie the high prevalence of child undernutrition in the region, whereas overweight and obesity are rising concerns. A systematic review of evidence (2000–2017) was conducted to identify barriers and programme approaches to improving the coverage of maternal nutrition interventions in the region. The search strategy used 13 electronic bibliographic databases and 14 websites of development and technical agencies and identified 2,247 citations. Nine studies conducted in Bangladesh (n = 2), India (n = 5), Nepal (n = 1), and Pakistan (n = 1) were selected for the review, and outcomes included the receipt and consumption of iron and folic acid and calcium supplements and the receipt of information on dietary intake during pregnancy. The studies indicate that a range of barriers acting at the individual (maternal), household, and health service delivery levels affects intervention coverage during pregnancy. Programme approaches that were effective in improving intervention coverage addressed barriers at multiple levels and had several common features: use of formative research and client assessments to inform the design of programme approaches and actions; community‐based delivery platforms to increase access to services; engagement of family members, as well as pregnant women, in influencing behavioural change; actions to improve the capacity, supervision, monitoring, and motivation of front‐line service providers to provide information and counselling; and access to free supplements.

birth outcomes, including low birth weight and preterm delivery Liu et al., 2017). Severe anaemia and calcium deficiency increase the risk of postpartum haemorrhage and hypertensive disorders, respectively (Rahman et al., 2016;World Health Organization [WHO], 2013), two leading causes of maternal death globally (Say et al., 2014). Despite this compelling evidence, the performance of maternal nutrition programmes in low-and middle-income countries is generally poor. For example, coverage data from 59 low-and middle-income countries indicate that the mean proportion of women who consume iron folic acid (IFA) supplements for at least 90 days during pregnancy is only 31% (Development Initiatives, 2017).
Comprehensive maternal nutrition interventions can promote the health of pregnant women and contribute to healthy fetal growth and development Dewey, 2016). In November 2016, the WHO released guidelines on improving antenatal care (ANC) for women, including a comprehensive set of recommendations on nutrition interventions during pregnancy (WHO, 2016). Although these new guidelines provide an opportunity to reignite attention on maternal nutrition, concerted efforts are needed to ensure that policy and programme actions translate to nutrition services that reach women with quality and equitable coverage. Women in low-and middleincome countries face substantial barriers to accessing nutrition interventions and adopting recommended nutrition behaviours, including low education and knowledge, lack of family support, and inadequate coverage and quality of maternal health and nutrition services (Kavle & Landry, 2017;Victora et al., 2012).
In South Asia, there has been progress on improving women's nutrition. Our review of data from national demographic, health, and nutrition surveys conducted in the last 15 years indicate that low stature (<145 cm) in women of reproductive age has declined by at least 20% in three out of four countries with available data, and low body mass index (BMI; <18.5 kg/m 2 ) has fallen by at least 30% in all four countries. However, progress is not swift enough; we estimate that about one in 10 women in South Asia has a low stature and one in five has a low BMI. This is of immense concern because low maternal stature and BMI have been shown to predict child stunting in South Asia (Kim, Mejia-Guevara, Corsi, Aguayo, & Subramanian, 2017), in addition to other adverse maternal and newborn outcomes. Furthermore, the region is home to over one third of the world's anaemic women, and no country in the region is on track to meet the World Health Assembly target to halve anaemia in women of reproductive age by 2025 (Development Initiatives, 2017).
Given the close links between maternal and child nutrition, efforts to improve the nutritional status of women are critical to attaining the global nutrition targets of the Sustainable Development Goals and World Health Assembly and in unleashing the developmental potential of children and the economic development of South Asian nations (Aguayo & Menon, 2016;Vir, 2016;Development Initiatives, 2017).
Long-running interventions such as IFA supplementation are reaching far too few pregnant women in the region, although there is considerable variation both between and within countries. The most recent demographic and health surveys (DHS) show that the proportion of women receiving IFA supplements for at least 90 days during their most recent pregnancy ranges from 7% in Afghanistan (DHS 2015) to 71% in Nepal (DHS 2016), whereas interstate variation in India ranges from 4% in Nagaland to 82% in Lakshadweep (National Family Health Survey 2015. In light of the poor maternal nutrition indicators in South Asia and poor coverage of evidence-based interventions, this review was conducted to (a) identify drivers and barriers to the coverage of maternal nutrition interventions and (b) examine the evidence on the effectiveness of programme approaches and actions to improve coverage. In addition, grey literature, including programme evaluations and unpublished literature, was sourced from the organizational websites of 14 agencies, academic institutions, and technical bodies using an adapted search strategy for individual websites (Table S1) and was solicited from development partners working in the South Asia region.

| Study selection
One researcher performed the database and website searches using the aforementioned criteria and search strategy. All titles and abstracts of study reports were screened for inclusion by the first author, and a randomly selected sample was checked by the second author. The full text of potentially eligible studies was retrieved for screening against the eligibility criteria, and the selection of the final set of studies was confirmed by the second and third authors.

| Data extraction and synthesis
The selected studies were examined separately by the first three authors. A standardized data extraction template was used to extract data including the title, authors, publication year, country, study design, intervention and comparison, target population, sample size, outcome measure, and results. The study aims and recommendations were also recorded. Data were synthesized by type of maternal nutrition intervention and type of barrier to coverage of or adherence with the intervention. Barriers were identified at three levels: individual, household, and health service delivery.

| Quality assessment
The quality of evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation criteria based on the eight known assessment criteria (Atkins et al., 2004;Walker, Fischer-Walker, Bryce, Bahl, & Cousens, 2010). Two authors reviewed the evidence and made a judgement about quality.

| Equity assessment
We aimed to assess equity by using the PROGRESS framework (place of residence, race/ethnicity, occupation, gender, religion, education, socio-economic status, and social capital) for the included studies (O'Neill et al., 2014).

| Characteristics of the included studies
The searches identified 2,247 citations ( Figure 1). After removing duplicate citations and screening titles and abstracts, 37 articles were considered potentially eligible. After a full-text review, 28 studies were excluded because the study participants were not pregnant women and/or the study did not examine predictors or the coverage of a WHO-recommended maternal nutrition intervention. The remaining nine studies met the inclusion criteria and included two studies conducted in Bangladesh (Nguyen et al., 2017a;Nguyen et al., 2017b), five in India (Balakrishnan et al., 2016;Ghanekar, Kanani, & Patel, 2002;Prinja et al., 2017;Shivalli, Srivastava, & Singh, 2015;Wendt et al., 2015), one in Nepal (Sharma et al., 2016), and one in Pakistan (Nisar, Dibley, & Mir, 2014). These eligible studies included one cluster RCT (Nguyen, Kim, et , 2017b;Nisar et al., 2014;Wendt et al., 2015), and one qualitative study (Ghanekar et al., 2002). All the nine studies examined the receipt and/or consumption of IFA supplements during pregnancy, two studies examined the receipt and/or consumption of calcium supplements (Nguyen, Kim, et al., 2017a;Nguyen, Sanghvi, et al., 2017b), and one study examined the receipt of information on maternal diet (Nguyen, Kim, et al., 2017a). There were no studies that reported the coverage of other maternal nutrition interventions. A summary of the included studies is provided in Table 2 for cross-sectional studies and Table 3 for all intervention studies.   Wendt et al., 2015;Balakrishnan et al., 2016;Sharma et al., 2016;Nguyen, Kim, et al., 2017a;Nguyen, Sanghvi, et al., 2017b;Prinja et al., 2017).

| Barriers and enablers to the receipt and consumption of supplements
Household factors include husband/family support and encouragement to the pregnant woman to seek health care and to purchase and consume supplements, husband's education, and household wealth (Ghanekar et al., 2002;Nisar et al., 2014;Shivalli et al., 2015;Wendt et al., 2015;Sharma et al., 2016;Nguyen, Kim, et al., 2017a).
Health service delivery factors include the timing of the pregnant woman's first ANC visit, the number of ANC visits, health worker capacities, quality of dietary counselling, supply of IFA/calcium supplements to distribution sites, and provision of free supplements (Nisar et al., 2014;Wendt et al., 2015;Balakrishnan et al., 2016;Sharma et al., 2016;Nguyen, Kim, et al., 2017a;Prinja et al., 2017).
Only two intervention studies, one on the nutrition-focused Maternal Neonatal Child Health (MNCH) in Bangladesh (Nguyen, Kim, et al., 2017a) and the other on community-based maternal health promotion intervention in Nepal (Sharma et al., 2016) aimed to address barriers at all three levels; all other studies addressed barriers at two levels.

| Quality of evidence
The overall quality of evidence was assessed for five intervention studies using Grading of Recommendations, Assessment, Development, and Evaluation criteria for the receipt or consumption of IFA supplements, the most common outcome indicators, and was found to be low to moderate quality; the qualitative study, Ghanekar et al. (2002), was excluded from this assessment. The five studies (four quasi-experimental and one cluster RCT) included 20,066 participants.
Several study design biases limited the quality of the evidence: In Balakrishnan et al. (2016), there was a potential performance bias as the data from the intervention areas were compared with the data from the rest of Bihar and historical data from previous year; in Prinja et al. (2017) and Sharma et al. (2016), there was potential selection due to the lack of randomization; and in Shivalli et al. (2015), the small sample size was calculated using coverage parameters that were very different from the results. There was heterogeneity in the study designs and methods; only one study used an RCT and others were quasi-experimental with various designs. Clinical heterogeneity was noted due to the difference in types of interventions and outcomes used (Table S2). In terms of consistency, effects on the IFA coverage indicator were positive, inconclusive, or negative (Table 3). There was additionally imprecision in the effect size due to large confidence interval (Balakrishnan et al., 2016). As there was only one RCT included in this review (Nguyen, Kim, et al., 2017a), we downgraded a starting rating of "high quality" evidence (for RCT) by two levels for very serious concerns about risk of bias, clinical and methodological heterogeneity, inconsistency, and imprecisions.
There was variation across the seven interventions studies in the dose of iron, folic acid, and calcium in supplements and the definition of the outcome indicators on IFA and calcium supplementation with respect to the recall period (previous month of pregnancy or the entire  (Table S2). This variation renders it difficult to draw coverage comparisons across the studies.

| Equity
Although most studies reported the socio-economic status and demographic information of the women at baseline, only Sharma et al.
(2016) disaggregated the outcome variables by wealth and education and showed that having higher education and being wealthier were significant factors in the consumption of IFA supplements during pregnancy. We were therefore not able to assess equity for the outcomes on the receipt or consumption of IFA supplements.

| Receipt of supplements
The only study that reported a positive impact on the receipt of supplements was the nutrition-focused MNCH study in Bangladesh, which reported significant effects on the receipt of free IFA supplements and free calcium supplements by pregnant women in Bangladesh (Nguyen, Kim, et al., 2017a). In this programme, nutritionfocused interpersonal counselling, community mobilization, and the distribution of free IFA and calcium supplements to women at home were integrated into an existing MNCH programme and compared with standard ANC services and dietary counselling. The programme approaches were informed by formative research that indicated that early initiation of supplementation, a higher number of ANC visits (more than four visits versus four visits and less), and the provision of free supplements were potential drivers of the receipt of supplements (Nguyen, Sanghvi, et al., 2017b). Health workers and volunteers were highly trained and closely supervised, and health volunteers    other health services to pregnant women in the intervention group.
They indicated that the lack of impact on IFA supplement receipt was due to unresolved barriers, including a shortage of IFA supplements to distribution sites in the intervention area and the fact that the front-line workers using the mhealth application were not responsible for the supply of IFA supplements to pregnant women.  (Table 3).

| Consumption of supplements
The nutrition-focused MNCH programme in Bangladesh (Nguyen, Kim, et al., 2017a) included a number of initiatives to increase the consumption of supplements by pregnant women in the intervention areas.
Women were counselled during home visits on the benefits of IFA and calcium supplements, consequences of IFA or calcium deficiency, doses and duration of IFA and calcium supplementation that should be followed, side effects of supplements and ways to minimize them, and foods that may inhibit their absorption. Husbands and other family members were engaged to ensure women took sufficient supplements during home visits and community mobilization events such as husband forums.
In addition, women were requested to preserve empty strips of supplements to monitor how many they took. The increase between baseline and end line in the proportion of women who reported to consume supplements during pregnancy was greater in the intervention group than comparison group for both IFA (DDE 9.8 pp, p < 0.001) and calcium (DDE 12.8 pp, p < 0.001). Significant effects were also observed for the number of supplements consumed for both IFA (additional 46 supplements) and calcium (additional 50 supplements) in the intervention group (p < 0.001). The authors attribute these results to the high priority given to ensuring adequate supplies of free supplements to women at home, the high quality of training and supervision of front-line workers, the performance-based incentives for community volunteers, and high quality of counselling involving home visits and husbands.
In Nepal, a participatory community-based maternal health promotion intervention that focused on community groups to improve the uptake of maternal health services significantly increased the proportion of women who reported to consume IFA supplements during pregnancy between baseline and end line by 9.4 pp (p < 0.001); there was no significant increase in the comparison group (Sharma et al., 2016). The community groups comprised women aged 15-49 years with at least one child under the age of 2 years, some mixed with mothers-in-law groups, and separate men's group. The interventions consisted of 24 one-hr health promotion sessions covering a range of issues related to health care during pregnancy, delivery, and the postnatal period, including the importance of IFA supplementation.
Participatory activities with visual cards and role playing were conducted, and efforts were made to encourage group attendance by empowering lower castes to attend and providing small incentives to women on completion of four ANC visits. The results suggest that community-based health promotion groups involving mothers-in-law and husbands as well as pregnant women can improve the consumption of IFA supplements during pregnancy. Shivalli et al. (2015) examined the effectiveness of trials for improved practices to increase the consumption of IFA supplements during pregnancy in Varanasi, India. The trials for improved practices involved an assessment home visit to ascertain the perceptions and practices related to IFA consumption from pregnant women and her family members, followed by a negotiation home visit based on the findings of the assessment, in which pregnant women were asked to select and try individually tailored practices and were given home-based reminder tools. The proportion of women who reported to consume at least 100 IFA supplements during pregnancy was 85% in the intervention group, compared with only 38% in the comparison group. The authors reported that pregnant women in the intervention group appreciated the counselling, reminder materials, and family support, which together helped to motivate and sustain IFA consumption throughout pregnancy.
A study conducted in Vadadara, India (Ghanekar et al., 2002), had no baseline or comparison groups, and so it was not possible to determine the effect of the programme interventions on IFA supplement consumption with certainty. However, the study offers important insights into approaches to address barriers to IFA supplement con-

| Counselling on maternal diet
The nutrition-focused MNCH programme in Bangladesh (Nguyen, Kim, et al., 2017a)  Predictors of the receipt and consumption of IFA and calcium supplements in Bangladesh, India, and Pakistan included early and more frequent ANC visits, higher maternal education, higher paternal education, higher maternal knowledge, increased household wealth, access to counselling services, and a higher level of support from husbands.
These enablers confirm previous findings on IFA supplementation that point to the role of women's knowledge and empowerment, household resources, women's access health services, and the quality of health services in determining whether women receive and consume supplements and adopt positive dietary practices (Kavle & Landry, 2017;Kavle & Landry, 2018;Siekmans, Roche, Kung'u, Desrochers, & De-Regil, 2017;Sununtnasuk, D'Agostino, & Fiedler, 2016).
All programmes that delivered information and counselling to women and their family members, often in combination with other programme actions, significantly improved the receipt and consumption of IFA and calcium supplements and the receipt of information on a diverse diet during pregnancy (Ghanekar et al., 2002;Nguyen, Kim, et al., 2017a;Sharma et al., 2016;Shivalli et al., 2015). These programmes had several common features. First, they used formative research and ethnography to adapt the design of programme actions to the context-relevant barriers faced by women in accessing services or adopting recommended nutrition behaviours. This included the design of locally relevant information and counselling materials. Second, they engaged with and enlisted the support of influential family members, including husbands and mothers-in-law, to ensure pregnant women accessed services and adopted positive behaviours, thereby countering women's low decision-making authority. Third, they introduced home visits and community forums to increase the access of pregnant women, influential family members, and other community leaders to interventions. Fourth, they developed the capacity of front-line workers to provide information and counselling to women, family members, and other community members, and in Bangladesh (Nguyen, Kim, et al., 2017a), added mechanisms to supervise, monitor, and motivate their performance.
The two programmes that examined the effectiveness of mhealth applications to improve service delivery by front-line workers were not effective in increasing the receipt or consumption of IFA supplies (Balakrishnan et al., 2016;Prinja et al., 2017). It is likely that mhealth approaches alone will not be effective if other barriers to receipt or consumption of IFA, such as a lack of IFA supplies, are not addressed.
These findings confirm previous research that highlight the value of community-based intervention packages for improving maternal care and neonatal outcomes (Lassi & Bhutta, 2015). A recent review of global evidence found that community-based distribution of IFA supplements provides an effective platform to improve knowledge about anaemia, provide counselling to improve compliance, and should be coupled with actions to ensure the consistent supply of supplements at facility and community levels to improve both access and coverage (Kavle & Landry, 2018). In addition, a separate review highlighted the importance of tailored, culturally appropriate nutrition education and counselling during pregnancy (Kavle & Landry, 2018).
There are considerable gaps in the body of evidence on the effectiveness of programme approaches to improve the coverage of maternal nutrition interventions in South Asia. Most studies examined only IFA supplementation, and only one examined calcium supplementation and access to information on dietary intake during pregnancy, even though all countries in the region are delivering multiple nutrition interventions to women during pregnancy. We found no published evidence that met the review criteria from Afghanistan, Bhutan, Maldives, and Sri Lanka, only one study in Nepal and Pakistan, and two in Bangladesh. Most of the studies reviewed lacked detailed programme impact pathways making it challenging to assess which approaches, actions, or processes contributed to the changes in receipt or consumption of the interventions. Design issues and nonstandardized use of indicators among the included studies meant that it was not possible to compare the results between studies. The quality of the evidence on IFA supplementation coverage (receipt and consumption) was assessed as low to moderate due to risk of bias, clinical and methodological heterogeneity, inconsistency, and imprecision.
Greater attention is needed in all country settings in South Asia to resourcing and conducting formative and implementation research to better understand the pathways, enablers, and barriers to improving the access to all maternal nutrition interventions rec-