MIND OVER MATTER IN THE PHILIPPINES: A STUDY OF KEY STAKEHOLDERS' PERCEPTIONS OF CHILDHOOD STUNTING DISCUSSION PAPER JUNE 2020 Iman Sen Nkosinathi Mbuya Gabriel Demombynes Varun Gauri / MIND OVER MATTER IN THE PHILIPPINES? A Study of Key Stakeholders’ Perceptions of Childhood Stunting Iman Sen, Nkosinathi Mbuya, Gabriel Demombynes, Varun Gauri June 2020 Health, Nutrition, and Population (HNP) Discussion Paper This series is produced by the Health, Nutrition, and Population Global Practice of the World Bank. The papers in this series aim to provide a vehicle for publishing preliminary results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. For information regarding the HNP Discussion Paper Series, please contact the Editor, Martin Lutalo at mlutalo@worldbank.org or Erika Yanick at eyanick@worldbank.org. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because the World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street, NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. © 2020 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 All rights reserved. ii Health, Nutrition, and Population (HNP) Discussion Paper Mind over Matter in the Philippines? A Study of Key Stakeholders’ Perceptions of Childhood Stunting Iman Sena Nkosinathi V. N b Gabriel Demombynes c Varun Gauri d a Development Economics, World Bank, Washington DC, United States b Health, Nutrition and Population Global Practice, World Bank, Washington DC, United States c Human Development, World Bank, Manila, Philippines d Princeton University, Princeton, United States The Government of Japan provided financial support for this work through the Japan Trust Fund for Scaling Up Nutrition Abstract: Declines in rates of child stunting in the Philippines have decelerated, making it hard for the country to achieve its targets on nutritional outcomes. The knowledge base, beliefs, and practices of caregivers have been extensively researched, but little is known about how health workers and policy makers fare in comparison. We conduct qualitative interviews, striving to preclude bias as we capture these stakeholders’ views on factors that affect stunting, and go on to compare and contrast these perceptions. We subsequently investigate the importance of the different factors in detail through a large-scale quantitative survey with frontline health and nutrition workers. The findings suggest that while most workers’ knowledge and beliefs are consistent with accepted practices, important deviations from consensus views exist, and these are correlated with worse self-reported service delivery outcomes at local health centers. The findings suggest that in the Philippines any endeavor to further improve service delivery must take into consideration the beliefs of frontline workers. Keywords: Survey methods, Information and Knowledge, Beliefs, Health E-mails: isen@worldbank.org, nmbuya@worldbank.org, gdemombynes@worldbank.org, vgauri@worldbank.org. Disclaimer: The findings, interpretations, and conclusions expressed in the paper are entirely those of the authors, and do not represent the views of the World Bank, its Executive Directors, or the countries they represent. Correspondence Details: Nkosinathi Mbuya, World Bank, 1818 H Street, NW, Washington, DC, USA iii TABLE OF CONTENTS RIGHTS AND PERMISSIONS ........................................................................................... II ACKNOWLEDGMENTS .................................................................................................... VI I. BACKGROUND ........................................................................................................... 1 A. Measuring Beliefs and Mental Models ............................................................................................. 3 B. Qualitative Data Collection ............................................................................................................... 4 Pre-interview Listing and Grouping Exercise ................................................................................. 4 Focus Group Discussions ................................................................................................................. 4 In-depth Interviews among Key Informants.................................................................................... 4 Characteristics of Respondents ....................................................................................................... 5 Qualitative Data Collection Sample ................................................................................................. 5 C. Limitations of the Study .................................................................................................................... 7 II. FINDINGS FROM QUALITATIVE INTERVIEWS ....................................................... 9 Caregiver and Health Worker Beliefs about the Causes of Stunting ......................................... 9 Beliefs about the Consequences of Stunting and Monitoring Growth...................................... 14 Beliefs about Weight Gain during Pregnancy .............................................................................. 15 Beliefs about Stunting among Policy Makers............................................................................... 15 Discussion and Takeaways from the Qualitative Interviews ..................................................... 16 III. FINDINGS FROM QUANTITATIVE HEALTH WORKER SURVEYS ...................... 18 A. Knowledge levels of Barangay Nutrition Scholars and Barangay Health Workers ......................... 18 B. Beliefs about Child Height Held by Barangay Nutrition Scholars and Barangay Health Workers .. 21 C. Beliefs on Causes of Stunting .......................................................................................................... 22 D. Interpreting Mental Models of Health Workers Through Response Patterns ............................... 24 E. How Knowledge and Beliefs Relate to Service Delivery at Local Health Centers ........................... 26 F. Profiles of Health Workers in the Philippines ................................................................................. 32 IV. DISCUSSION ............................................................................................................. 36 V. CONCLUSIONS AND RECOMMENDATIONS ........................................................ 39 REFERENCES .................................................................................................................. 41 ANNEX .............................................................................................................................. 43 A1. Barangay Health Workers and Nutrition Scholars: Background....................................................... 43 A2. Additional Tables and Regressions ................................................................................................... 45 A3. Reliability and Validity of Belief Indexes........................................................................................... 52 A4. Comparison with Regional Level Stunting Rates .............................................................................. 53 iv ABBREVIATIONS AND ACRONYMS BHC Barangay health center BHW Barangay (local village) health worker BNS Barangay nutrition scholar CCM Cultural Consensus Modeling DOH Department of Health FGD Focus group discussion HNP Health, Nutrition, and Population IDI In-depth interview IFA Iron and folic acid IHGLP Institutionalizing Health Leadership and Governance Program IPV Intimate partner violence LCA Latent class analysis LGU Local government unit NNC National Nutrition Council NNS National Nutrition Survey OPT-Plus Operation Timbang Plus PPAN Philippine Plan of Action for Nutrition RHM Rural health midwife SDG Sustainable Development Goal v ACKNOWLEDGMENTS The report was prepared by a World Bank team comprising Nkosinathi V. N. Mbuya, Iman Kaylan Sen, Gabriel Demombynes, Sharon Faye Alario Piza, and Varun Gauri. The authors would like to thank the Philippine Survey and Research Center for implementation of the qualitative work, Sharon Piza for managing the implementation of the quantitative survey, and Toby Monsod for assistance with survey instrument design and analysis. The authors are grateful to the World Bank peer reviewers, Cecilia Acuin, Leslie Elder, and Ana Maria Munoz Boudet for their detailed feedback and for providing technical guidance and quality review of this report. The authors are grateful to the World Bank for publishing this report as an HNP Discussion Paper. vi I. BACKGROUND The past 25 years have witnessed only limited progress toward substantially reduced malnutrition in the Philippines. Between 1993 and 2015, the prevalence of stunting among children under five years decreased by just 5 percentage points—from 38.8 to 33.4 percent. At that rate, the country could wait another 80 years before it meets its 2025 target of 18.2 percent. The immediate causes of high levels of chronic malnutrition are not disputed. As the Philippine Plan of Action for Nutrition (PPAN) states, “Poor infant and young child feeding in the first two years of life coupled with bouts of infection can explain high levels of stunting.” In 2015, only 48.8 percent of children 0 to 5 months of age were exclusively breastfed, with rates declining from 68.0 percent at the first day of life to 24.7 percent by the fifth month, and only 15.5 percent of infants 6 to 23 months old enjoyed a “minimum acceptable diet.” 1 The document also highlights the poor state of maternal nutrition—maternal nutrition and health during pregnancy are intimately linked to child nutrition trajectories; low birthweight is one of the most prominent risk factors for stunting—noting that “the prevalence of nutritionally-at-risk 2 women has not improved over the years, with a rate between 24 to 26 percent since 2008” (NNC 2017). What is less clear is why high levels of suboptimal infant and young child feeding and maternal nutrition continue to persist, despite more than four decades of state-sponsored programs to address them directly. 3 Interestingly, income poverty is not the main reason. The PPAN recognizes that a blanket attribution of the problem to income poverty would be a simplistic view, pointing out, for instance, that while higher wealth quintiles have higher proportions of children 6 to 23 months old that meet the minimum acceptable diet criteria, these numbers are still low at less than 20 percent. Global evidence suggests that effective delivery of a package of evidence-based health and nutrition interventions focused on the first 1,000 days of life can have an impact on childhood stunting. A critical starting point is to ensure access and utilization of these interventions, which can be delivered by coordinated primary health care services through strong maternal and child health and nutrition programs. Unfortunately, in countries with a high stunting burden, including the Philippines, access to such services is often limited, and levels of uptake can be disappointingly low. Several factors are known to hinder access to health and nutrition care services, and they stand in the way of appropriate diagnosis and prevention of chronic childhood undernutrition. These include cultural beliefs and social norms, inadequate knowledge of signs and symptoms of undernutrition and services available, cost of services, lack of transport options, and poor quality 1. A composite measure of both the minimum feeding frequency and minimum dietary diversity for different age groups. 2. The definition of “nutritionally-at-risk pregnant women” is based on an analysis of the Philippine National Nutritional Survey results (Magbitang et al. 1988), where women were classified according to their risk of delivering a low birthweight baby based on the adequacy of their weight gain during pregnancy. This has been used as the standard reference for assessing weights of pregnant women in the country. 3. As Briones et al. (2017) recount, the country’s pursuit of good nutrition goes back more than four decades, when in 1974 the first Philippine Nutrition Program was formulated (“to coordinate the various nutrition interventions” that had arisen since independence in 1946), followed by more systematic targeting enabled by the first National Nutrition Survey in 1978. Groups targeted at that time for improvements in nutrition outcomes—infants and preschoolers (0–6 years), school children (7–14 years), pregnant women, and lactating mothers—are largely the same now, and the main interventions/approaches adopted then—food support and assistance, health protection, nutrition information, and education—continue to anchor the Philippine Plan of Action for Nutrition national action plan now. 1 of service. Quality of care, in particular, has been identified as the key reason for high morbidity and mortality in some countries, despite substantial increases in coverage by health services (Graham and Varghese 2012). One aspect that has not yet been fully explored is the role of the beliefs and attitudes about important markers of child undernutrition, such as stunting, held by caregivers on the one hand and policy makers and frontline policy implementers on the other. Globally, there is a better understanding of the importance of the beliefs and behaviors of those who care for infants and young children regarding child nutrition, especially in the realm of feeding practices. For example, strong notions of “good” versus “bad” foods exist for pregnant women and young children across countries (Raman et al. 2016). However, there is a paucity of evidence of how the knowledge and beliefs of caregivers on markers of undernutrition, such as stunting, may influence optimal feeding and other caregiving and health monitoring practices. Similarly, the knowledge, beliefs, and practices of health workers are important. These may have considerable influence over how caregivers and their families perceive and experience health and nutritional care and consequently influence decisions to seek and access care. For instance, care that is not framed or explained by providers in a way that leads to acceptance by caregivers and families may lead to dissatisfaction with the health system. In addition, beliefs of health workers can directly influence the quality of care and the effectiveness of efforts to promote maternal and child health and nutrition (Mannava et al. 2015). The strength of a health worker’s beliefs will affect whether or not there is consistent monitoring of growth, counseling on optimal infant and young child feeding and caring practices, or the provision of essential micronutrients such as iron and folic acid (IFA), vitamins, and zinc. From an international perspective, the quality of health workers’ knowledge about nutrition—and, by extension, their counseling skills—has been a concern (Sunguya et al. 2013). Historically, medical training has lacked adequate and updated nutrition training, and as a result, health workers trained at most of the medical training institutions have lacked adequate knowledge of nutrition. Such health workers may also lack the competence and skills to provide basic nutrition advice about nutritional care to their clients. Evidence is available on how in-service nutrition training has a positive impact on health workers’ nutrition knowledge, nutrition counseling skills, and management of child undernutrition. However, there is little evidence on whether health workers’ beliefs affect nutrition service delivery, and if so, how. In the Philippines, the beliefs of policy makers on childhood undernutrition have scarcely come to light, even though they must be understood on account of their major impact on support for legislation and programs promoting better nutrition. Among government officials, several individual cognitive biases have been documented, all of which obstruct objective and impartial decision making. In a recent paper, Banuri, Dercon, and Gauri (2017) carefully document several biases among civil servants and find evidence of confirmation bias, which suggests that policy makers are likely to interpret information and favor policies in a manner consistent with their existing beliefs. Therefore, existing beliefs and views of policy makers with respect to health and nutrition may affect the extent of their support for nutrition programs. This paper discusses the knowledge, beliefs, and practices of health workers, caregivers, and policy makers with respect to undernutrition among children in the Philippines—in particular, beliefs about the root causes of stunting—using primary data collected through qualitative interviews. A large-scale quantitative survey is then implemented among health and nutrition workers to delve deeper to understand the worldviews held by frontline workers and how these beliefs relate to service provision in local health centers in the Philippines. 2 METHODS AND DATA COLLECTION A. MEASURING BELIEFS AND MENTAL MODELS Nutrition-related behaviors may be influenced by deeply internalized beliefs about how the world works, also known as cultural schema or mental models (DiMaggio 1997). These encompass the default associations, categories, concepts, identities, prototypes, stereotypes, causal narratives, and views that we use to make sense of the world. When we think, we utilize concepts that reflect the shared beliefs and understandings of others in our community, rather than inventing these concepts de novo. When these concepts are reflective of an outlook shared by everyone around us, we tend to unquestioningly adhere to them. Thus, mental models come to shape how we interpret the world around us and enable or constrain specific thoughts and actions—by shaping perceptions and filtering the “facts” that people believe in and are able to understand, with profound implications for decision-making (World Bank Group 2015). To measure mental models related to stunting, we adapted and applied a method used by cognitive anthropologists, called cultural consensus modeling or CCM (Romney, Weller, and Batchelder 1986). CCM is a mixed-method technique to estimate the “culturally correct” answers to a series of belief questions, to test if there is a dominant group belief, as well as to calculate each respondent’s degree of sharing and agreement with the group (Weller 2007). A key feature of CCM is its emic approach, meaning the adoption of the respondent’s perspective, using his or her own words to understand culture and construe meaning. The CCM method involves a number of steps, as shown in Figure 2.1. The first is a free (random) listing of key terms by respondents (e.g., all possible causes of stunting). The second requires the respondents to sort these terms into broader categories. These steps are completed during one or more phases of qualitative data collection. Subsequently, respondents rate these categories on several dimensions such as importance, controllability, and changeability, in quantitative surveys. If the group harbors a singular belief system, or mental model, the pattern of responses to the series of questions will reveal it. Figure 2.1. Sequence of Steps to Measure Mental Models (Source: authors) STEP 1: FREE LIST OF RELEVANT TERMS STEP 2: SORT TERMS INTO BROADER CATEGORIES STEP 3: RATE TERMS (various yardsticks) 3 B. QUALITATIVE DATA COLLECTION Caregivers, frontline health workers, and politicians were interviewed in qualitative fieldwork to understand and compare beliefs with respect to stunting. 4 Qualitative specialists from the local research firm, the Philippine Survey and Research Center, were contracted to facilitate focus group discussions (FGDs) and in-depth interviews (IDIs) with these stakeholders. To ensure that all pertinent topics were covered, discussion guides were prepared in the form of a semistructured questionnaire. All sessions were recorded for transcription and quality control. The FGDs were held in a conveniently and centrally located venue; this afforded participants a comfortable environment conducive to sharing and discussion. Meanwhile, IDIs were conducted at the most convenient location for informants, such as their offices. Several methods were employed to collect data from a number of different regions in the country. Pre-interview Listing and Grouping Exercise Before the actual interviews, respondents were handed out sheets for a listing and grouping exercise as per Steps 1 and 2 (Figure 2.1). This was implemented at the very outset to preclude the risk of priming respondents or exerting any sort of influence on them. Respondents were asked to “write down as many factors, things, events, and relationships that you can think of that you think affect a person’s height” before trying to group all the factors listed into broader categories. The aim of this exercise was to capture the unprimed views of respondents before the discussions. This written exercise was completed by health workers, caregivers, and policy makers. Focus Group Discussions That written exercise was followed by group discussions with about five respondents per group. These groups consisted of women caregivers of children in both the age categories (under twos and two to fives, respectively) as well as frontline health workers in barangays (villages as administrative units or mini-districts). These small groups allowed for more in-depth discussion of personal experiences than would have been possible within a larger format. Researchers were able to record the totality of personal experiences in detail. At the same time, the group dynamics naturally encouraged the sharing of opinions and the building of ideas. The discussions explored the causes of stunting reported by respondents during the written exercise. Other topics included the ability to discern if a child is stunted, the consequences of stunting, and how—in their roles as caregivers or frontline health workers—the respondents could best support the healthy growth of children by overcoming the various obstacles highlighted in the discussions. In-depth Interviews among Key Informants IDIs, rather than group discussions, were conducted with government officials involved in policy making and health service delivery in the local government. This enabled the researchers to probe a respondent’s role in community nutrition and draw out his or her personal views on how best to improve existing policies, programs, and processes. 4. These stakeholders were purposely selected from provinces with better or poor nutrition on the basis of per capita income and stunting rates. Among the frontline health workers and local policy makers were health professionals such as physicians. 4 Characteristics of Respondents The main categories of respondents who participated in the qualitative study included the following: • Caregivers or mothers caring for children under five years old, segmented into caregivers or mothers of children under two years and caregivers or mothers of children ages two to five. • Public health workers from the barangay health unit, which included barangay health workers (BHWs), midwives, nurses, and physicians. • Local public officials with considerable (often decisive) influence on health policies and programs, such as the governor, mayor of the capital city, or mayor of an adjacent municipality. In the absence of a governor or mayor, municipal administrators at provincial or city level were interviewed. At the national level, interviewees included a member of the Senate Committee on Health and Demography and a Department of Health (DoH) executive. Qualitative Data Collection Sample A countrywide sample was drawn, as shown in Table 2.1, including provinces with better and poorer overall outcomes in terms of nutrition. Purposive sampling was completed using a network of recruiters. In total, 20 FGDs and 17 IDIs were completed, as listed in Table 2.1. Table 2.1. Qualitative Data Collection Number Number Area Specifications Province City or Municipality of FGDs of IDIs Capital 2 1 Bontoc Poor nutrition municipality Mt. Province province Adjacent 2 2 Sagada North municipality Capital city Ilagan 2 1 Better nutrition Isabela Adjacent 2 2 province Naguilan municipality Capital city Borongan 2 1 Poor nutrition Eastern province Samar Adjacent 2 2 Maydolong South municipality Capital city Digos 2 1 Better nutrition Davao del Sur Adjacent 2 2 province Santa Cruz municipality Cotabato City 2 1 (Provisional Autonomous capital and part Region in Capital city of ARMM and Muslim Maguindanao part of Mindanao Maguindanao (ARMM) geographically) Adjacent 2 2 Sultan Kudarat municipality NCR Metro Manila Manila 2 Source: Focus group discussions and in-depth interviews with caregivers, health workers, and policy makers conducted as part of this study 5 Note: NCR = National Capital Region; FGDs = Focus group discussions; IDIs = In-depth interviews. Health and nutrition data in the Philippines are mostly gathered through household surveys. Information on attitudes and beliefs always derives from the perspective of caregivers. The objective of the Philippines Health Worker Survey was to fill an information gap: the perspective of frontline service providers. The survey aimed to fairly represent barangay health workers (BHWs) and barangay nutrition scholars (BNSs) across the country. 5 A self-administered questionnaire was completed by a sample of 5,053 in various regional events held for BHWs and BNSs from September to December 2018. 6 These were coordinated through the Department of Health Central Office (for BHWs) and the National Nutrition Council (for BNSs). The survey was undertaken in 10 regions comprising 36 provinces. The distribution of the sample across regions and between BHWs and BNSs is shown in Figure 2.2. Figure 2.2. Survey Sample XIII - NCR 256 427 I - Ilocos (Ilocos Sur only) Luzon III - Central Luzon 475 441 IV - CALABARZON 474 444 V - Bicol 402 VI - Western Visayas Visayas VII - Central Visayas 414 461 XI - Davao (Davao del Norte only) 114 Mindanao XII - SOCCSKSARGEN 771 374 XV - ARMM BHW BNS 5. Barangay nutrition scholars (BNSs) are volunteer community workers who are trained in implementing health and nutrition programs in a locality. BNSs are expected to identify and monitor malnourished pregnant and lactating mothers and malnourished children and provide counseling on good nutrition practice to them. Barangay health workers (BHWs) are also volunteers, who render primary care services in areas such as maternal, newborn, and child health in the community. They provide information, education, and motivation services for primary health care, maternal and child health, child rights, family planning, and nutrition and may administer immunizations and regular weighing of children. They often assist midwives in providing birthing services. As of August 2017, there were 46,293 BNSs and 276,919 active BHWs deployed across the country (https://www.doh.gov.ph/sites/default/files/publications/2017_FHSIS_Final_0.pdf). See the Department of Health (DOH) website, https://www.doh.gov.ph/health-program/BNS. 6. Questionnaires were translated into four local languages: Tagalog, Hiligaynon, Cebuano, and Bicolano. Respondents took between 15 to 20 minutes to complete the survey. 6 Source: Survey of health workers conducted as part of this study Note: NCR = National Capital Region; ARMM = Autonomous Region in Muslim Mindanao; BHW = Barangay health worker; BNS = Barangay nutrition scholar. The vast majority of health workers are middle-aged women (BHWs are slightly older). On average, BHWs have served longer (15 years) than BNSs (9 years). Both groups have a similar education profile, the majority having completed secondary education. Figure 2.3. Demographic Characteristics 97% 98% 52% 52% 48% 47% 48% 39% 41% 30% 16% 8% 9% 3% 2% 2% 1% 2% 3% 3% male female 15-24 25-44 45-64 65 and up Elementary High school College Post graduate Sex Age group Education BHW BNS Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; BNS = Barangay nutrition scholar. C. LIMITATIONS OF THE STUDY We note three limitations to the study. First, during the qualitative data collection process, earlier discussions presented a “person’s height” as the core concept, whereas this later transitioned to discussion of “child height.” The distinction is significant, since while genetics may influence average adult height, the scientific consensus is that it does not affect child height up to the age of 24 months. It is possible that the framing influenced responses; however, this is not clear from the responses. For example, participants indicated that nutrition mattered more and genetics less in the initial written exercise when considering a “person’s height” and were likely, subsequently, to acknowledge the importance of genetics when discussing “child height.” In the quantitative study, although questions were framed around child height, it is important to note that certain questions used the word “stunting,” which may have caused confusion. In the discussion below of findings and results, any change of framing is noted. Second, in our opportunistic implementation of the quantitative survey, we were unable to randomly sample health and nutrition workers to interview. However, to alleviate selection concerns, we ensured that all workers who attended the conferences completed the survey 7 instrument. These concerns notwithstanding, it should be noted that responses were ultimately collected from a large sample of workers across 10 regions. Finally, we measured beliefs and services offered in the health center in the same survey instrument, which may lead to common source bias. These efforts ought to be supplemented by data on services from other, more objective sources. Meanwhile, we were able to reduce social desirability bias in reporting on services by asking about service delivery at the local health center, rather than by the individual respondent. 8 II. FINDINGS FROM QUALITATIVE INTERVIEWS The analysis framework used for the qualitative study reflects the relationships between the different components that affect perceptions and practices toward stunting. In examining these in detail, it may be possible to identify key misconceptions that could be addressed. The framework assumes that knowledge, beliefs, and practices have a linear relationship whereby an individual’s knowledge of stunting feeds into her or his beliefs about stunting, which are in turn manifested in practices. These three components are also heavily affected by a person’s sociocultural context, which therefore cannot reasonably be ignored when considering perceptions toward stunting. • Context: The sociocultural and political environment, encompassing geography, culture, the economy, politics, and individual socioeconomic status. • Knowledge: A respondent’s current mental picture of stunting. This includes what he or she believes causes stunting, the indicators or cues for stunting, and the potential short- term and long-term effects of stunting. • Beliefs: The respondent’s mindset with regard to the issue of stunting, including the importance ascribed to it. This affects how people view overall health, in particular child height, but also such factors as weight, illnesses, or activity levels, in relation to their role as caregivers, policy implementers, and policy makers. • Practices: The resulting actions that respondents take in dealing with the issue of stunting. This will include caregivers’ practices in supporting their children’s development, efforts by health workers to implement existing programs and policies that may affect stunting, and the policies prioritized by policy makers. Caregiver and Health Worker Beliefs about the Causes of Stunting The FGDs showed that both caregivers and health workers believe genetics or hereditary factors to be an important cause of stunting, along with undernutrition. However, there was a discrepancy between findings from the initial written exercise (the request to list all factors that affect stunting) and the discussions that followed. Respondents considered both the mother and father’s height and race (and in some cases those of grandparents or other members of the family) to be important hereditary factors. During the discussions, caregivers often strongly attributed height to genetics, while health workers considered genetics and nutrition to be evenly matched determinants of child height. For caregivers, undernutrition was defined as having insufficient vitamins or nutrients for the body to achieve proper development. For most, the salient indicator of undernutrition was being underweight. However, in the initial written exercise, the majority of caregivers and health workers (as well as politicians) listed undernutrition as the cause of stunting. Undernutrition is aggregated for terms reflecting both the quality and quantity of food (for example, mentions of lack of food, nutrients, vitamins, and vegetables). The second most commonly listed cause was heredity (e.g., terms reflecting genes, heredity, and race), followed by lack of sleep, and then low income or poverty (see Table A2.1 for a full list of causative factors and frequency). An important finding is the inconsistency shown by a large number of caregivers who initially attribute stunting to undernutrition (in the written exercise), but during subsequent discussion conclude that genetics is a more powerful factor. One interpretation of this discrepancy is that it is a form of “blame avoidance” by caregivers anxious not to admit, especially not to a group, that having a malnourished child (as they actually suspect) is a consequence of inadequate parenting. 9 An alternative explanation is that they hesitate to voice their true internalized beliefs about the genetic causes of stunting until somebody else in the group has mentioned it first (and thereby socially sanctioned it). “ Sa amin din kasi iyong height ko ay nakuha ko rin sa mama ko—medyo maliit, iyong papa ko, matangkad. For me, I got my height from my mother—she’s kind of short; my dad is tall. … Doon ko rin naisip na baka nga namamana nga. May nagsasabi rin kasi, kahit na doon sa lola or sa mga aunty, pwede rin daw doon mamana. Maaring hindi lang rin sa magulang namamana. Kahit sa kamaganak, nakukuha rin. That is how I thought that maybe it is hereditary. There are also those who say, even among grandmothers and aunts, that it can be passed down. It’s possible that it’s not only from the parents. Even from other relatives, it can be passed down. - Caregiver of child less than two years old, Digos City Davao “ Based ko doon sa mga anak ko, genes talaga yun. Yung 15 years old ko, mas matangkad pa sa tatay. Matangkad talaga siya, pati yung 13 years old ko kasing tangkad ko lang. Eh yung 10 years old ang… hindi naman as in na pandak, pero pandak talaga na malaki lang ang katawan. Sabi ko, ‘Saan ito nagmana?’ Tas sabi ng tatay, ‘Nagtataka ka pa? Doon sa mga kapatid mo eh’, sabi niyang ganon. Kasi yung mga kapatid ko and yung mama ko kasi hindi gaano ka tangkad. Judging by my children, it really is genes. My 15-year-old is taller than my husband. He is really tall; even my 13-year-old is just as tall as me. But my 10-year-old… it’s not that he’s just short, it’s just that… well, actually he is really short, and fat. I said, “Where did he get that from?” Then my husband said, “You’re still asking that? He got that from your siblings.” It’s because my siblings and my mom are not that tall. Pag over age na kasi yung bata hindi pa matangkad, ibig sabihin sa lahi na niya yon. If the child is already beyond a certain age, then the child will not grow taller; that is because of their race. - Caregiver of child less than two years old, Cotabato City, Maguindao “ Siyempre lahat, kahit hindi bata, even tao, adult. Kailangan natin ng pagkain para mabuhay. Kasi iyon ang rason kaya nag su-survive tayo dito, lalong-lalo na iyong mga bata, kailangan talaga ng nutrition kasi kailangan pa nilang lumaki at---para maayos iyong paglaki. Of course, it’s all of us, not just children, even people, adults. We need food to survive. That is the reason why we survive here, especially the children, they really need nutrition because they use it to grow, so that they grow properly. Marami pang vitamins or minerals na kailangan ng katawan nila kaya nutrition. There are many vitamins or minerals that the body needs: that’s why nutrition is important. - Caregiver of child less than two years old, Eastern Samar “ Di nakadepende yun kung matangkad ka man o hindi. Kasi ang importante ay yung hindi malnourish. Wala kasi yun sa kung pandak ka o matangkad basta ang importante kumakain ka nang wasto. Di bale kung pandak ka basta wala kang sakit, basta malusog kang bata, okay na yon. It doesn’t matter if you are tall or not. What is important is that you are not undernourished. It’s not a matter of being short or tall, what’s important is that you are eating well. It doesn’t matter if you are short so long as you have no illnesses; so long as the child is healthy, that is okay. - Caregiver of child two to five years old, Sagada, Mt. Province 10 After genetics and nutrition, caregivers mentioned illness as a factor that directly inhibits child growth. Respondents claimed that this may indirectly affect nutrition since children who are sickly will have a poor appetite and therefore a limited intake of nutrients. For caregivers, certain factors may increase a child’s susceptibility to disease, such as unclean environments and poor diet and eating habits. “ Kapag madumi ang kapaligiran, siyempre maraming mga sakit na dadapo sa mga bata. Mas makaka apekto sa paglaki niya. Of course, if the environment is dirty, then clearly the child will contract many diseases. That will have a larger effect on their growth. - Caregiver of child less than two years old, Borongan, Eastern Samar Among the indirect causes of shortness in height, poverty is the most salient concern for caregivers since it has a huge impact on their ability to provide ample nutritious food. Caregivers reported that when under extreme financial pressure, their primary goal shifted, of necessity, to the simple alleviation of hunger rather than the provision of nutritious food. At these times they give their children food that is filling regardless of nutritional content. They prioritize rice in their diet since it promises sufficient satiety, especially when paired with affordable but nutrition-lacking viands (customary companions to rice) such as instant noodles, dried fish, oil, soy sauce, fish paste, and even packets of chips. In more depressed areas, such as Isabela and Maguindanao, sleep becomes a way for families to cope with hunger when food is not available. “ Kung minsan po sir alas dose, tapos kung minsan nagkakape lang po ako sa umaga at nagtitinapay. Pero kung minsan po sa hapon sir nagluluto lang ako ng lugaw, kasi wala po akong pambili ng ulam, kaya kung minsan po pinagkakasya ko nalang po sa aming lahat. Sometimes, Sir, [we get up] at 12 noon, so in the late morning I’ll just have coffee and bread. But sometimes in the afternoon, Sir, I’ll cook porridge because I have nothing to buy viands with, so sometimes I do what I can to make enough for all of us. - Caregiver of child less than two years old, Ilagan, Isabela Although knowledge was not spontaneously mentioned, caregivers did consider it a potential factor that may hamper a child’s growth, particularly when caregivers have inadequate knowledge of proper child nutrition. Caregivers perceived that the consumption of junk food (such as potato chips or sugary drinks) showed such a lack of knowledge and understanding. “ Kapag walang kaalaman din iyong nag-aalaga ng bata katulad ng unang anak, kapag hindi pa sanay. If the person taking care of the child lacks knowledge, [this is] like how it is with the first child, when the parent is still not used to it. O first time, hindi niya alam kung ano ipapakain niya sa anak niya. Kung anong mga kailangan niyang ibigay sa anak niya, kung ano iyong dapat nakukuha ng tama ng bata. For first timers, they do not know what to feed their child. They do not know what should be given to their child, what are the right things the child should be getting. - Caregiver of child less than two years old, Borongan, Eastern Samar Caregivers believed that children who lack parental supervision are likely to have a combination of poor hygiene, diet, and lifestyle, which hinders a child’s physical development. Poor hygiene may result in a child contracting diseases or parasitic worms, leading in turn to poor appetite. Lack of parental supervision may lead children to eat at inappropriate times and consume unhealthy food. On occasion, caregivers recognize that infant nutrition is sacrificed when mothers decide to 11 shift from breastfeeding to bottle-feeding (as soon as they are obliged to return to work). Without sufficient parental care, children also develop a poor lifestyle involving inadequate exercise or sleep, hampering growth. “ May mga batang hindi masyadong naalagaan ng magulang dahil sa kadahilanang busy sila sa trabaho. There are children who are not taken care of by their parents, who say this is because they are busy at work. …Yung hindi na maasikaso ang pagpapakain sa kanya, yung pagpaligo kaya yun na puros na lang naga laro, kaya hindi maasikaso, naaapektuhan din ang kalusugan niya. Minsan yon kaya hindi siya tumatangkad. When the parents are no longer able to sort out food for the children, let alone baths, then all the children do is play: that’s why they are not able to take care of them, and the child’s health is affected. That is the reason why sometimes they do not grow taller. - Caregiver of child less than two years old, Cotabato, Maguindanao “ Kapag sinabi pong kulang sa alaga, iyong paghahanda po ng pagkain, kapag pinapaliguan mo siyempre kapag hindi mo naasikaso iyong anak mo, magiging malnourish. When they say that the children lack proper care, like having food prepared, or getting baths… of course, when you are not able to take care of your child, the child becomes malnourished. …Sakitin. Madaling kapitan ng mga bacteria. Sickly. The bacteria will easily latch on to them. …Hindi pinapaliguan magiging masakitin at mabaho. When they are not given baths, they become sickly and smelly. ….Caregiver of child less than two years old, Borongan, Eastern Samar As implementers of existing national and local policies and programs on health, BHWs were more knowledgeable than BNSs. Even more knowledgeable were the nurses and midwives who oversee the BHWs. However, health workers mentioned a similar set of reasons for stunting as caregivers. During the FGDs, the health workers stated that both undernutrition and genetics were among the most important causative factors behind stunting, followed by disease, heavy labor, and lack of sleep, in that order. “ Tapos, para sa akin iyong pinaka number one na factor talaga na nakaka-affect sa tangkad ng isang bata is genes. Genes talaga kasi iyong mga sinasabi nating pobre, may matatangkad talagang mga pobre. Parang mas ano talaga na genes ang nakakaapekto sa pagtangkad ng isang bata. For me, the number one factor that really affects height is the genes of the child. Genes really because if we look at those who we say are poor, we soon see that there are tall people among them. So it really seems that genes is what affects the height of a child. - BHW, Borongan, Eastern Samar Yes, we accept [that we are short]. Particularly the Itas. Maybe the few who are not Itas will be tall. But if they are Itas, that is the race, that is their body build, so we’re in the same boat as indigenous people. - Midwife, Sagada, Mt. Province Other possible factors also came up during the discussion with varying degrees of importance as shown by the word cloud in Figure 3.1. 12 Another reason for stunting given by health workers (but not by caregivers) was a lack of adequate nutrition during pregnancy, as well as the consequences of teenage pregnancy in general. Based on these and other factors identified during the FGDs, a conceptual framework drawn from the discussion with health workers is shown in Figure 3.2, indicating the relationship between the different reasons mentioned (the diagram for caregivers is very similar to the one for health workers in Figure 3.2). Figure 3.1. Word Cloud of Causes Perceived by Health Workers Based on Intensity of Their Responses Source:Focus group discussions with health workers conducted as part of this study Note: Figure based on discussions with 56 health workers in 10 FGDs. 13 Figure 3.2. Diagram of Relationships between Health Workers’ Perceived Causes of Stunted Growth Source: Authors Note: Direct causes are arranged based on perceived importance, from most important (left) to least important (right). Beliefs about the Consequences of Stunting and Monitoring Growth The beliefs of caregivers and health workers about the consequences of stunting included longer- term impacts such as low self-esteem and exclusion from employment opportunities involving height requirements. Caregivers mentioned worse prospects for marriage, while health workers mentioned lower cognitive abilities and issues during pregnancy for women. It must however be noted that neither group regarded these as consequences of great concern. Respondents highlighted counterexamples of individuals who were short, but had achieved great success in life. 7 When evaluating the growth of the child, greater importance is attached to monitoring the weight of the child than the height. Weight is accorded more importance because the effects of wasting are perceived to be more serious than any consequence of stunting (which is accurate in the sense that wasting, not stunting, is associated with increased risk of mortality). In their assessment of a child’s development and well-being, other indicators that either caregivers or 7. This is indeed an accurate counterpoint to a usually simplified (and false) notion that being short is a negative outcome/occurrence and will cause damage to an individual’s life course. The sole negative physical outcome for low height (causally related) is an increased risk for cephalopelvic disproportion (CPD) among women. It is critical that accurate understanding of stunting in the first two years of life as a proxy for underlying negative conditions in the child’s environment that are associated with the condition of stunting (e.g., delayed cognitive development) should be the focus of all programming/training on child stunting. Linear growth retardation and stunting are associated with— but based on available evidence do not cause—delayed child development, reduced earnings at adulthood, and chronic diseases. 14 health workers monitor include immunity from sickness, a child’s physical activity, and possibly cognitive functions. There is also a potential link to health worker behavior—for example, some caregivers claimed that health workers only monitored a child’s weight and rarely the child’s height, therefore strengthening these beliefs. Beliefs about Weight Gain during Pregnancy The qualitative interviews also uncovered the advice that health workers give to expecting mothers on the subject of weight gain during pregnancy. They explain the importance of weight monitoring during pregnancy—as this serves as an indicator of the health of the unborn child. Although mothers are unsure of the exact weight they are expected to gain each month, they do expect to be admonished by health workers if the weight gain appears to be excessive. Some respondents described an increase of about 5 to15 kilograms (kgs) as normal and an increase of over 20 kgs as excessive. Most health workers in the different survey areas agreed that a weight gain of up to a kilogram every month is ideal. The mothers, however, reported that health workers would often ask them to go on a diet if they had gained roughly three kilograms in a month, claiming that putting on too much weight may lead to complications during birth. Health workers corroborate the finding that the weight of pregnant women is monitored monthly during routine checkups. The “ideal” recommended weight gain is one kilogram per month, and greater gains may trigger dietary recommendations from health workers. Beliefs about Stunting among Policy Makers The policy makers interviewed had varying degrees of involvement in the nutrition and health programming in their jurisdictions. Less interested policy makers often delegated decisions on health and nutrition policy to health officers, whereas those who were more interested sought to extend current programs and policies, including actively spotlighting malnutrition and stunting. Some of the most involved officials mentioned the need to expand prenatal care programs to include maternal malnutrition more broadly, rather than focusing narrowly on anemia. Policy makers who showed greater interest and involvement also typically had some medical training. Discussions with policy makers found strong associations between local government officials’ beliefs about stunting and how involved they were in the policy making and implementation process. While all policy makers mentioned nutritional factors behind stunting, less involved policy makers were more likely to emphasize genetics as the primary factor, noting that further intervention would thus probably be in vain. In stark contrast, the most interested and involved policy makers consciously denied genetics were a cause of stunting (see difference in mentions in Figures 3.3 and 3.4 below). Policy makers also mentioned fewer factors affecting stunting beyond malnutrition and genetics than did caregivers and health workers. The policy makers discussed poverty, lack of knowledge, food insecurity, and strenuous labor as other causes of stunting. Both involved and less involved policy makers mentioned food insecurity, while the less involved also mentioned strenuous labor. When asked about consequences of stunting, comparatively less involved policy makers mentioned bullying and exclusion from jobs with height requirements, whereas policy makers with greater involvement mentioned consequences such as slower cognitive development, worse performance at school, and lower work productivity. 15 Figure 3.3. Word Cloud of Causes Perceived by Policy Makers with Low Involvement Source: In-depth interviews with policy makers conducted as part of this study Note: Data based on in-depth interviews with eight policy makers. Figure 3.4. Word Cloud of Causes Perceived by Highly Involved Policy Makers Source: : In-depth interviews with policy makers conducted as part of this study Note: Data based on in-depth interviews with four policy makers. Discussion and Takeaways from the Qualitative Interviews During the qualitative interviews, respondents often attributed stunting to nutritional factors. However, further in-depth discussions also uncovered beliefs around genetics, ethnicity, and race. These sometimes conflicted beliefs came to light during discussions with both caregivers and health workers. While caregivers were more likely than health workers to attribute stunting to genetic factors, this was still a disputed topic of discussion among health workers. Discussions 16 with stakeholders often highlighted structural factors, such as poverty and the disease environment, as well as factors related to a child’s cognitive development, such as parental knowledge, care, and attention. In-depth discussions with policy makers led to the following finding: officials who had a health or medical background (about a quarter of all officials interviewed) were most likely to believe in nutrition-related causes of stunting and disavow genetic or other immutable characteristics in a stronger manner than health workers and caregivers. These same officials were also more involved in the formulation and implementation of health and nutrition policy and activities in their jurisdictions. This suggests that those policy makers with medical or scientific knowledge of the true causes of stunting were more likely to believe that their work could have an impact on reducing stunting. Conversely, policy makers lacking such knowledge tended to excuse their lack of involvement by attributing stunting to factors entirely beyond their control. 17 III. FINDINGS FROM QUANTITATIVE HEALTH WORKER SURVEYS The qualitative exercises and discussions helped collect a number of reasons that relevant stakeholders believe affect child stunting. However, the discussions also highlighted tensions between which of these beliefs are of primary importance and how they might relate to each other. We therefore use this exhaustive list of reasons to design and test, at scale, a ratings exercise to uncover the most important determinants in child stunting, as well as the underlying mental models among health workers. This chapter discusses findings from the quantitative surveys completed by over 5,000 frontline health and nutrition workers— BHWs and BNSs. Respondents rated the level of importance of each factor that may affect child height on a Likert scale. All such factors from the qualitative discussions and written exercises were included. Respondents also answered additional knowledge questions and reported the extent of various types of service provision in their local barangay health centers (BHCs). Section A) discusses the findings from the knowledge questions asked of health workers, followed by a detailed discussion of the beliefs held by health workers, including the causes and consequences of malnutrition and stunting, and an analysis of the patterns of responses across the set of belief questions. The next section presents findings on the correlation between beliefs held by health workers and services offered at their local health centers. The following section validates our findings, by relating the knowledge, beliefs, and services reported in the survey to regional stunting rates in the Philippines. The final section analyzes the distribution of beliefs (mental models) among health workers, uses these to identify health worker types or profiles, and considers how these profiles relate to service delivery. A. KNOWLEDGE LEVELS OF BARANGAY NUTRITION SCHOLARS AND BARANGAY HEALTH WORKERS All respondents were asked a series of eight questions focused primarily on infant feeding practices and nutrition (Table 4.1) to determine their level of knowledge. The questions tested basic information with which all health workers were supposed to be reasonably familiar. BNSs typically provided more accurate responses compared to BHWs, answering one additional question correctly on average (4.6 compared to 3.7). Nearly two-thirds of BNSs answered five or more questions correctly (out of eight), and the same percentage of BHWs answered at least half the questions correctly (four out of eight). Each question was answered accurately by more than half of all workers, with one striking exception: a majority of BNSs and BHWs thought that breast milk alone is sufficient for children ages six to nine months. In fact, this is a crucial stage in the transition to complementary feeding. Workers were asked the rate of under-five stunting in the Philippines, which is 33 percent. The responses of BNSs were closer to the actual rate than those of BHWs. The BNSs on average believed the rate to be 36.6 percent, while BHWs on average believed it was higher, at 43.7 percent, as shown in Table 4.2. However, when asked about weight gain during pregnancy, the results were more mixed. 8 BHWs more often recommended that pregnant women “eat less food than normal” compared to BNSs (see Table 4.2). When asked about the “recommended” weight gain for women during pregnancy, 8. According to the Clinical Practice Guidelines on Maternal Nutrition and Supplementation (2013) made by the Philippine Obstetrical and Gynecological Society (POGS), weight gain for Filipino pregnant women should be as follows (based on pre-pregnant weight status): if of normal weight, 25 to 35 pounds; if underweight, 28 to 40 pounds; if overweight, 15 to 25 pounds; if obese, 11 to 20 pounds. Therefore, a weight gain of 9.0 kg (19.8 pounds) for a woman of normal weight would not meet these guidelines, but 12.3 kg does (27.0 pounds). 18 nearly 47 percent of respondents said they did not know or did not respond to the question. Another 13 percent mentioned that there is no recommended weight gain during pregnancy for women. More BHWs did not respond to the question or mentioned there is no recommended weight gain. However, for the remaining 40 percent of workers who responded, BNSs on average recommended a weight gain of 9.0 kg, compared to 12.3 kg by BHWs. We also observe differences in knowledge between BNSs and BHWs from different regions. To show geographical differences for BNSs and BHWs, we compared those regions where sufficient numbers of each category of worker were interviewed. 9 For example, in the set of eight knowledge questions, BHWs in region V (Bicol) answered the fewest questions correctly (2.5 on average), whereas those in region IV (Calabarzon) answered 4.6 questions correctly on average, a difference of over 2 questions. The variation across BNSs was lower, with those in region XII (Soccsksargen) getting the fewest correct (4.1) and those in region XIII (NCR) scoring the highest number of accurate answers (5.6), with a difference of 1.5 questions between these regions. The variation in responses for both BNSs and BHWs across regions is shown in Table A2.3. BNSs in region XII (Soccsksargen) come closest to accurately estimating the under-five stunting rate in the Philippines (quoting 33.1 percent), and BHWs in region IV (Calabarzon) are the closest (quoting 37.5 percent). Table 4.1. Responses to Knowledge Index Questions by Barangay Nutrition Scholars and Barangay Health Workers Question BNS BHW Lack of rice (% yes) 11.80 16.20 Worm infestation (% yes) 20.10 18.70 What are the causes of Lack of iron-rich food (% yes) 75.60 63.20 anemia among children? Lack of fat (% yes) 6.40 4.90 Lack of iodine (% yes) 21.40 16.50 (Multiple choice) Unsure (% yes) 0.90 1.00 Only selected “Lack of iron-rich food” (%) 46.50 40.10 Lack of calcium (% yes) 5.50 11.10 What causes night Lack of rice (% yes) 2.50 4.33 blindness in children? Lack of vitamin A (% yes) 84.00 70.70 Lack of vitamin C (% yes) 7.30 11.90 (Multiple choice) Unsure (% yes) 1.00 1.40 Only selected “Lack of vitamin A” (%) 77.30 59.50 When should a mother Within one hour of giving birth (% yes) 71.10 62.10 initiate breastfeeding of her Between 2 and 3 hours of giving birth (% yes) 12.00 10.90 newborn child? After 5 hours of giving birth (% yes) 1.80 1.80 On the second day of giving birth ( % yes) 0.20 0.60 (Only one option selected) Only selected “Within one hour…” (%) 71.10 62.10 A child should be fed extra food when he/she has a True (% correct) 71.80 65.50 cold, fever, or diarrhea It is recommended that a mother breastfeed her child True (% correct) 79.60 67.70 for at least 2 years 9. We include regions where at least 100 workers from each category were interviewed. There are six such regions in total. 19 Question BNS BHW Breast milk alone is an adequate source of nutrition for a child older False (% correct) 40.10 21.10 than 6 months and less than 9 months Milk formula or canned milk is good for infants who are False (% correct) 80.10 65.70 less than 6 months old A child who is short for her age in the first two years can catch up in height with True (% correct) 72.70 68.10 appropriate nutrition during adolescence Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHS = Barangay health worker. Table 4.2. Differences in Knowledge between Barangay Nutrition Scholars and Barangay Health Workers Question Full Sample BNS BHW What is your best guess on the Mean 39.90 36.60*** 43.70 percentage of children under age 5 SD (26.60) (25.62) (27.19) who are severely short for their age in the Philippines? (%) N 3,505 1,849 1,656 What is your best guess on the Mean 34.60 31.30*** 38.20 percentage of children under age 5 SD (26.65) (25.20) (27.72) who are severely short for their age in your province? (%) N 3,405 1,791 1,614 Mean 79.60 88.30*** 69.60 "Stunting is a problem" (%) SD (40.33) (32.19) (46.02) N 4,067 2,173 1,894 Recommended weight that a pregnant Mean 10.40 9.10*** 12.30 woman should gain during her SD (20.87) (18.97) (23.17) pregnancy (kg) N 1,781 1,042 739 Always 34.60 30.40 39.40 Very often 25.00 24.70 25.30 How often do you recommend that a Sometimes 18.20 18.90 17.50 pregnant woman eat less food than normal during pregnancy? Only a few times 7.70 7.50 7.80 Never 14.60 18.50 10.00 N 3,968 2,115 1,853 Mean 4.16 4.56*** 3.74 Index of knowledge question (# correct out of 8)a SD (2.09) (1.94) (2.16) N 5,053 2,557 2,496 Source: Survey of health workers conducted as part of this study Note: SD = Standard deviation. a. Ranging from 0 to 8 correct answers. ***p < 0.01, **p < 0.05, *p < 0.1. 20 B. BELIEFS ABOUT CHILD HEIGHT HELD BY BARANGAY NUTRITION SCHOLARS AND BARANGAY HEALTH WORKERS A set of questions on beliefs was posed to facilitate the inference of mental models of health and nutrition workers in the Philippines. Views about the causes and consequences of child height were elicited through a number of questions and included the factors identified during the qualitative discussions. Respondents were asked a few individual questions, such as to what extent they agreed with the statement, “Filipinos are in general short by nature”(Figure 4.1) and “Do you think stunting is a problem?” They were also asked to rate 17 factors that may affect a child’s height and select the three most important (Tables 4.3 and 4.4) and rate how likely it is that certain outcomes will be experienced by a child who is “short” (in effect, the consequences of being stunted10); these were posed in an easy-to-administer table format with Likert scale responses. Respondents were also presented with vignette questions to obtain the perceived relative importance of income, parental height, and nutritious food. BNSs were less likely to concur with the assertion that Filipinos are in general short by nature, although there is clear evidence here of a bimodal distribution (twin peak pattern). For BHWs, the distribution is more skewed toward agreement. Nearly 80 percent of all health and nutrition workers believe that stunting is a problem in the Philippines. However, BNSs are about 20 percentage points more likely to acknowledge this compared to BHWs. Figure 4.1. Distribution of Responses in Agreement to the Statement “Filipinos Are in General Short by Nature” Strongly agree Filipinos are in general short by nature Agree bhw Neither agree nor disagree Disagree Strongly disagree Strongly agree Agree bns Neither agree nor disagree Disagree Strongly disagree 0 10 20 30 40 Percent Source: Survey of health workers conducted as part of this study 10. The language was adapted to make the survey instrument as easy as possible to understand, hence the use of regular language such as “short” compared to “stunted.” It is assumed that the consequences are perceived to be similar regardless of the choice of descriptive term. 21 C. BELIEFS ON CAUSES OF STUNTING The five most important factors affecting stunting, according to aggregated responses (shown in Table 4.3), indicate that the beliefs of respondents align with globally accepted views about immediate causes (food consumed, disease environments), underlying causes (such as prenatal care, parental care and attention), and basic causes (poverty) of child height. Ethnicity and height of mother or father are not rated consistently as important factors, as compared to findings from the qualitative interviews. When ratings of the 17 factors listed under the question, “To what extent does each of the following factors affect a child’s height?” are aggregated and scored, “affliction of disease,” “poverty,” “quantity and quality of food consumed,” “prenatal care,” and “parental care and attention” are the top five factors ranked by both BNSs and BHWs (Table 4.3, the full set of Likert scale responses is shown in Table A2.2). BNSs and BHWs also share the bottom seven factors with only slight differences. The bottom seven factors may be described as having to do with genetics and ethnicity (i.e., “height of mother,” “height of father,” and “ethnicity”) and faith or fate (“destiny,” “faith,” “God,” and “luck”). However, the rankings of the factors change when respondents are asked to choose the three most important factors, and this difference is driven by the choices of BHWs. Specifically, prenatal care is ranked most important, followed by quantity and quality of food consumed, and ethnicity of parents (Table A2.6). For the subsample of BNSs alone, however, although prenatal care and quantity and quality of food consumed are still the two most important factors, exclusive breastfeeding is the third most important factor. For the subsample of BHWs, quantity and quality of food consumed (330), prenatal care (284), and ethnicity of parents (234) were the three that were ranked first most frequently, while poverty was sixth (144). Vignettes are short stories that present different scenarios about a main character in a story. They are followed up by questions about the most likely outcome based on the information provided. This indirect way of presenting questions has been demonstrated to help overcome social desirability bias and to elicit more honest responses from respondents. The response elicited is taken to be indicative of the beliefs and choices of respondents unfettered by self-censorship. We tested the relative importance of three of the main factors identified during the qualitative data collection through vignettes—having access to sufficient and nutritious food, or not; having tall or short parents; and being better-off or poor (regression results shown in Table A2.4). The analysis shows that nutrition still plays the most important role in determining if the child is likely to be stunted, for both BNSs and BHWs. However, some of the main effects appear to be counterintuitive (e.g., having tall parents or being well-off is positively correlated with stunting— Columns 1, 2, 3, Table A2.4). To unpack the main effects, we show the full model including the interactions between these factors (Columns 4, 5, 6). Here, we find that better nutrition is negatively correlated with the likelihood of stunting in all specifications and is the only factor that affects stunting according to BNSs. However, height of parents is also negatively correlated with stunting for BHWs, showing that some importance is also given to genetics by BHWs, as was found during the qualitative discussions. The main effects are consistent across regions, as shown in Table A2.5. The results can be interpreted as follows across the different analyses (aggregation and ranking of individual beliefs and vignettes). For BNSs, it appears that beliefs are better aligned with 22 accepted factors and understanding of child malnutrition, including nutritional deficiency, and prenatal and postnatal parental care. However, when asked to choose or determine the most important factors (in the ranking and vignette exercises), BHWs also indicate that ethnicity and height of parents are important factors. In summary, we find evidence of factors related to ethnicity and genetics emerge in certain quantitative tasks, specifically for BHWs, as was inferred from the qualitative data. Table 4.3. “To What Extent Does Each of the Following Factors Affect a Child’s Height?” Overall BNS BHW Obs Meana Rank Obs Meana Rank Obs Meana Rank Disease 4,584 0.683 1 2,386 0.700 1 2,198 0.664 1 Poverty 4,568 0.642 2 2,380 0.655 2 2,188 0.628 2 Food consumed 4,641 0.544 3 2,389 0.571 4 2,252 0.516 3 Prenatal care 4,539 0.527 4 2,363 0.572 3 2,176 0.478 4 Parental care 4,550 0.511 5 2,368 0.553 5 2,182 0.465 5 and attention Access to toilets 4,544 0.489 6 2,359 0.531 6 2,185 0.444 7 and clean water Exclusive 4,591 0.436 7 2,374 0.476 7 2,217 0.393 10 breastfeeding Sufficient sleep 4,544 0.429 8 2,370 0.444 8 2,174 0.412 9 Carry heavy 4,559 0.424 9 2,375 0.404 10 2,184 0.445 6 loads Help family 4,543 0.414 10 2,360 0.413 9 2,183 0.415 8 livelihood Destiny 4,499 0.291 11 2,334 0.258 11 2,165 0.326 11 Height of mother 4,604 0.283 12 2,390 0.258 11 2,214 0.310 12 Height of father 4,589 0.282 13 2,391 0.258 11 2,198 0.309 13 Faith 4,497 0.252 14 2,324 0.232 14 2,173 0.273 14 God 4,481 0.214 15 2,308 0.195 15 2,173 0.235 16 Ethnicity 4,646 0.207 16 2,408 0.180 16 2,238 0.236 15 Luck 4,491 0.182 17 2,332 0.156 17 2,159 0.210 17 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHS = Barangay health worker. a. Mean of normalized scores, which are computed as follows: Does not affect = 0.0; Slightly affects = 0.33; Affects = 0.67; Greatly affects = 1.0. Table 4.4. Ranking of Top Three Beliefs about the Causes of Stunting Factors Joint sample BNS BHW Freq Rank Freq rank Freq rank Prenatal care 1,047 1 763 1 284 2 Quantity and quality of food consumed 660 2 330 2 330 1 Ethnicity of parents 364 3 130 7 234 3 Parental care and attention 331 4 178 4 153 5 Exclusive breastfeeding 296 5 189 3 107 7 Sufficient sleep 293 6 136 6 157 4 23 Factors Joint sample BNS BHW Freq Rank Freq rank Freq rank Poverty 284 7 140 5 144 6 God's will 146 8 58 8 88 8 Helping with family livelihood 103 9 40 10 63 9 Affliction of disease 93 10 41 9 52 10 Destiny 43 11 19 12 24 11 Faith and prayer on part of parents 37 12 22 11 15 14 Carrying heavy loads 33 13 15 14 18 12 Height of mother 33 14 16 13 17 13 Access to toilets and clean water 20 15 6 15 14 15 Height of father 6 16 2 16 4 16 Luck 3 17 1 17 2 17 Total 3,792 2,086 1,706 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHW = Barangay health worker. D. INTERPRETING MENTAL MODELS OF HEALTH WORKERS THROUGH RESPONSE PATTERNS The section above presents aggregated and ranked results of individual responses to each of the belief questions along with additional vignette analysis to further elicit true beliefs of health workers. It should be noted that separate analysis of responses to a single belief question will fail to pick up patterns of responses across the entire set of belief questions. Interpretation of such patterns could well prove useful, especially when they are driven by underlying worldviews that are not otherwise amenable to direct observation. One way of measuring such underlying constructs is through factor analysis. In factor analysis, it is assumed that such latent constructs cannot be directly measured, but can instead be evaluated indirectly through the relationships observed in response patterns across a set of variables. Factor analysis of the responses to the set of belief questions on the causes of stunting would uncover the key underlying constructs (factors). These are in effect worldviews or mental models held by BNSs and BHWs, and they explain the patterns we observe. Factor analysis confirms there is a dominant worldview among health workers in the Philippines. The overall number of factors or response patterns that emerge from the responses are between two and four (see Figure A2.1). The ratio of the first factor to the second is close to 3, which as a rule of thumb means that there is a dominant pattern of response and consequently a dominant worldview. However, the second through fourth factors are also precisely those that warrant inclusion in the analysis, since these depict other worldviews held by BNSs and BHWs that represent deviations. Therefore, we extract the maximum number of four factors for this analysis. 24 Based on the factor loadings, four factors (mental models) emerge based on the items that load onto these factors (see Table A2.6). • Prescribed Practices (Factor 1): “Quality and quantity of food,” “Parental care and attention,” “Prenatal care,” “Breastfeeding,” “Access to toilet and water,” and “Sufficient sleep” 11 • Genetics (Factor 2): “Ethnicity of parents,” “Height of mother,” “Height of father” • Poverty (Factor 3): “Poverty,” “Helping with family livelihoods,” “Carrying heavy loads,” “Affliction of disease” • Faith and Fate (Factor 4): “Destiny,” “God’s will,” “Faith and prayer on part of parents,” “Luck” Indexes created from these groups of beliefs tell two very different stories for BNSs and BHWs, across different regions of the Philippines. 12 Figures 4.2 and 4.3 show the standardized means of these four belief indexes for BNSs and BHWs. For BNSs in most regions (except for region XII), the prescribed practice and poverty indexes are positive, indicating greater agreement with the items in these indexes. The faith and fate index is usually negative, indicating the least agreement with these factors for BNSs. Agreement with genetics-related causes is also low and similar to the faith/fate index, with exceptions in regions IV and XI, where it is higher. The figure for BHWs is starkly different; across all regions, agreement with the faith/fate and genetic indexes is always higher than with the prescribed practices and poverty indexes. Figure 4.2. Means of Standardized Belief Indexes for Barangay Nutrition Scholars by Region Standardized Belief Indexes, BNS IV - CALABARZON VII - Central Visayas XI - Davao XII - SOCCSKSARGEN Prescribed XIII - NCR Genetics Faith and Fate Poverty -.4 -.2 0 .2 .4 11. Sufficient sleep was also an important primary factor often mentioned in the formative research. 12. Indexes are created by aggregating the items that load onto each of the four factors and standardizing them. 25 Source: Survey of health workers conducted as part of this study Note: : NCR = National Capital Region; BNS = Barangay nutrition scholar. Figure 4.3. Means of Standardized Belief Indexes for Barangay Health Workers by Region Standardized Belief Indexes, BHW I - Ilocos IV - CALABARZON V - Bicol VI - Western Visayas XI - Davao Prescribed Genetics XV - ARMM Faith and Fate Poverty -.5 0 .5 1 Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; ARMM = Autonomous Region in Muslim Mindanao. Interpreting response patterns through factor analysis revealed four different underlying constructs, which we interpret as mental models or worldviews. The first and most consequential factor includes prescribed practices typically recommended for better child health and can be interpreted as the main worldview for health workers. However, several additional views emerged from the factor analysis, including views related to genetics, poverty, and faith/fate that are held by health workers. It also appears that BHWs are more likely to hold these additional views than BNSs. In summary, while most health workers report accepted determinants of child health, some report deviations from conventional views, and these deserve further investigation. E. HOW KNOWLEDGE AND BELIEFS RELATE TO SERVICE DELIVERY AT LOCAL HEALTH CENTERS While the uncovering of deviations from the expected worldview is in and of itself important, can these worldviews also influence health worker behavior, such as services offered at local (barangay) health centers? This section first describes summary statistics of the self-reported services offered in health centers and subsequently explores the predictive validity of the different worldviews and mental models held by BNSs and BHWs (along with knowledge and other control 26 variables) regarding whether these services are typically delivered at local health centers. We asked workers if the services are delivered in the health centers where they work, to avoid asking directly if they had performed these services themselves (and thereby minimize social desirability bias in self-reported service delivery behavior). However, reported service delivery is still interpreted here as being influenced by whether individual health workers delivered the services themselves more often and is therefore regarded as indicative of individual health worker behavior. Table 4.5 shows that BNSs report greater service delivery on average at the health centers than do BHWs, and these responses are significantly different. BNSs and BHWs answered eight questions about services typically provided at the health facility where they work. Questions were asked about advice given to pregnant women and mothers of young children on breastfeeding and complementary feeding; provision of IFA and vitamin A supplements; and if height, weight, and growth monitoring measurements were conducted. According to the responses, the two least reported services carried out at local health centers were height measurements of children under two years old and vitamin A supplements for children under five. However, there was high agreement with measuring the weight of children and women, as well as the distribution of other supplements, such as IFA tablets. As an indicator of health and well-being, more frequent weight measurement (rather than height measurement) is consistent with the findings from the qualitative data collection. There is also a high level of agreement with the assertion on advice about exclusive breastfeeding and complementary feeding. Table 4.5. Services at Health Facility Where Respondent Works Question BNS BHW “Pregnant women were Agree or strongly agree (%) 97.50 93.89 given iron/folic acid Number 2,272 2,061 supplements” “Weight measurements were Agree or strongly agree (%) 95.33 95.23 taken for all children under 2 Number 2,271 2,057 years old" “Height measurements were Agree or strongly agree (%) 77.88 59.02 taken for all children under 2 Number 2,229 1,989 years old" “All the children under 2 Agree or strongly agree (%) 92.07 91.01 years old have growth charts Number 2,256 2,025 used to track their growth" “All the children under 5 Agree or strongly agree (%) 80.39 64.39 years old were given vitamin Number 2,249 2,005 A supplements” During prenatal / antenatal care visits (%) 97.40 95.25 Women’s weight was If the woman was sick or unwell measured (%) 20.50 16.60 If the woman looked underweight / 27.60 25.20 overweight (%) Advice to breastfeed Yes (%) 90.34 83.90 exclusively Number 2,516 2,367 Correct advice on Yes (%) 92.00 85.77 complementary feeding (%) Number 2,551 2,488 27 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHS = Barangay health worker. Then, to explore the variation in self-reported service delivery, a standardized index was created for all eight services reported by health workers. Figure 4.4 shows that this is indeed quite different across provinces, with workers in NCR, Central Visayas, and Davao reporting the most service delivery, and those in the ARMM, Ilocos, and Bicol provinces reporting the least. When comparing the three regions reporting higher service delivery (NCR, Central Visayas, and Davao) with the three regions with lower service delivery (ARMM, Ilocos, and Bicol), it would be helpful to consider the role of local government units (LGUs) in the implementation of national government programs. While policies and program frameworks emanate from national levels, resources to actually deliver the services to intended recipients have to come from provincial, municipal, and city health offices. ARMM and Bicol are among the poorest regions with high undernutrition rates. Both experience frequent disturbances—in peace and order (ARMM) or natural calamities (Bicol). The Ilocos situation bears further investigation, as the region does not have a high undernutrition burden. Figure 4.4. Variation in Service Delivery by Region Standardized Service Index across Regions .4 Standardized Service Index .2 0 -.2 -.4 M as as ol on s EN ao N R co ZO C M ic ay ay uz av G -N -B Ilo R s is AR R lL D -A Vi lV SA I- V II - tra AB XI rn XI XV tra SK en te AL en es -C C -C C C W SO III I- IV - VI VI I- XI Source: Survey of health workers conducted as part of this study Note: NCR = National Capital Region; ARMM = Autonomous Region in Muslim Mindanao. 28 Table A3.2 shows regressions results of standardized belief and knowledge indexes on a standardized index of all the service questions to explore the relationship across these variables. As described, belief indexes are created based on the factor analysis—standardized indexes are created for each of the following groupings: “Prescribed Practices,” “Genetics,” “Faith and Fate,” and “Poverty.” The regression results in Table 4.6 show that beliefs conforming to prescribed practices have a significant and strong positive association with greater reported service delivery. Conversely, mental models that conform to faith- and fate-based causes have significant negative associations. These results are driven by the beliefs of both BHWs and BNSs. Belief indexes for poverty and genetics as causes of stunting have weaker positive and negative associations, respectively, with service delivery (this relationship is only true for BNSs). The knowledge index has significant and strong positive associations with service delivery for both BHWs and BNSs. The regression specification also includes several control variables, including additional beliefs (“if stunting is a problem”), poverty (a measure of the provincial poverty incidence), and individual characteristics of health workers (including age, education, and years of service). As expected, salience of stunting as a problem and education are positively associated with service delivery, while age and incidence of poverty are negatively associated. Years of service do not appear to be associated with reported service delivery. Figures 4.5 and 4.6 show the individual coefficients on the relationship between each service delivery question, and each of the belief and knowledge indexes. (Table A2.7 also shows a breakdown of the service index and relationship with belief and knowledge indexes). Increased reporting of service delivery quite clearly correlates with belief in prescribed practices and greater knowledge, respectively, and this holds true across most service delivery questions for health workers. The lowest coefficient here is weighing mothers in different circumstances. Also, advice given on exclusive breastfeeding appears to still be low for BNSs. Stronger beliefs in poverty-related causes are weakly but still positively associated with greater service delivery, as noted before. This is driven by increased height and weight measurement of children, greater provision of IFA supplements, and greater advice on breastfeeding by BNSs. For BHWs this is driven primarily by height measurement of children, alongside advice on breastfeeding and complementary feeding. As discussed, faith/fate and genetic beliefs are more negatively correlated with service delivery indicators for BNSs than for BHWs. The pattern is fairly consistent for BNSs and rather mixed for BHWs. For BNSs, the strong negative correlation with faith- or fate-related beliefs is driven by increased supplement provision and body measurements taken. The regressions confirm interesting relationships between health workers’ views and knowledge indexes and service delivery at local health centers. However, since these relationships are a two- way street; it is also possible that improved service delivery at centers influences health workers’ beliefs and knowledge or that another unobserved variable (such as ability or motivation of health workers) explains this relationship. Another limitation is that service delivery is self-reported, and beliefs, knowledge, and services are recorded in the same survey instrument, leading to potential common source bias. 29 Table 4.6. Regressions of Belief and Knowledge Indexes on Services Index (1) (2) (3) (4) service_index_ service_index_ service_index_ service_index_ std std std std Sample Full Full BHW BNS prescribed_practice_index_std 0.182*** 0.168*** 0.163*** 0.167*** (0.0161) (0.0168) (0.0287) (0.0203) genetics_index_std −0.0302* −0.0322* −0.0339 −0.0340 (0.0164) (0.0166) (0.0249) (0.0226) faith_fate_index_std −0.0599*** −0.0474*** −0.0108 −0.0766*** (0.0161) (0.0163) (0.0263) (0.0219) poverty_index_std 0.0774*** 0.0704*** 0.0744*** 0.0657*** (0.0169) (0.0178) (0.0264) (0.0241) knowledge_index_std 0.135*** 0.122*** 0.130*** 0.114*** (0.0142) (0.0148) (0.0221) (0.0200) Controls Provincial poverty −0.00611*** −0.0160*** −0.00301 (0.00204) (0.00452) (0.00228) Age −0.00727*** −0.00897*** −0.00618*** (0.00176) (0.00284) (0.00216) Education 0.0294** 0.0417** 0.0140 (0.0132) (0.0192) (0.0176) Years of service 0.00284 0.00537** 0.00192 (0.00206) (0.00271) (0.00301) Observations 4,555 4,145 1,973 2,172 R-squared 0.140 0.141 0.132 0.101 Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; BNS = Barangay nutrition scholar. A standardized service index is created using all seven service questions put to health workers. This is regressed on standardized indexes created for the four belief indexes described above. Robust standard errors in parentheses clustered at the barangay level. All specifications include region-fixed effects. *** p<0.01, ** p<0.05, * p<0.1. 30 Figure 4.5. Relationship between Belief and Knowledge Indexes for Each Service Delivery Indicator among Barangay Nutrition Scholars Standardized values of (prescribed) Standardized values of (genetics) Standardized values of (faith) IronSupplement WeightMeasured Standardized values of (poverty) HeightMeasured GrowthCharts VitaminASupplement WeighedMother Standardized values of (kindex_all) BreastFeeding ComplementaryFeeding -.1 0 .1 .2 Source: Survey of health workers conducted as part of this study 31 Figure 4.6. Relationship between Belief and Knowledge Indexes for Each Service Delivery Indicator among Barangay Health Workers Standardized values of (prescribed) Standardized values of (genetics) Standardized values of (faith) IronSupplement WeightMeasured Standardized values of (poverty) HeightMeasured GrowthCharts VitaminASupplement WeighedMother Standardized values of (kindex_all) BreastFeeding ComplementaryFeeding -.2 -.1 0 .1 .2 Source: Survey of health workers conducted as part of this study F. PROFILES OF HEALTH WORKERS IN THE PHILIPPINES A previous section uncovered four worldviews or mental models held by health workers using patterns of responses to questions about factors that may affect stunting. This section explores the presence and distribution of these different mental models in health workers. It interprets differences in the prevalence of these worldviews to identify different groups or profiles of health workers in the Philippines. We use a statistical method called latent class analysis (LCA) 13 to profile and classify health workers into different groups based on response patterns to the four belief indexes described above (“Prescribed Practices,” “Genetics,” “Faith and Fate,” and “Poverty”). The analysis was also conducted using the “Knowledge” index, but this is not shown as knowledge levels were found to be consistently high across all groups. 13. Latent class analysis (LCA) is a statistical method that is commonly used to identify unobserved class membership among respondents, using observed responses and characteristics. In this case we use the four continuous mental model index variables to identify the latent groups. Such analysis, using continuous variables is often called latent profile analysis. LCA estimates class membership based on response patterns and agreement (in this case to the four indexes), as well as proportion of workers in each class. 32 Figures 4.7 and 4.8 show the results from the LCA for three profiles or groups for each category of health worker (BHW or BNS). 14 Each group shows the probabilities for each of the four belief indexes for individuals in that group. For example, the first group has a high probability for the Poverty index but low for the others. It is interesting to note that there is a separate group of health workers who believe that the main causes of stunting relate to disease and child labor, which are beliefs in the Poverty index. The second group shows higher marginal probabilities for both Prescribed Practices and Poverty indexes but lower for the others. The third group shows higher marginal probabilities for all the belief indexes, including the Faith and Fate and the Genetics indexes. This indicates that views based on faith and fate or genetics do not necessarily exist independently of the other views. It is interesting to note that the patterns within the groups are very similar for both BHWs and BNSs, and the only difference is that the estimated probabilities are higher for Prescribed Practices and Poverty and lower for the Faith and Fate and Genetics indexes for BNSs. Figure 4.7. Profiles of Barangay Health Workers Based on Response Patterns to the Belief Indexes Class 1 Class 2 Class 3 3.5 3.5 3.5 3 3 3 Predicted mean 2.5 2.5 2.5 2 2 2 1.5 1.5 1.5 1 1 1 ed ed ed te ty rty s s e ty s te ic ic ic t er er Fa Fa Fa rib ib rib ve et et et v v cr en en en Po Po Po sc sc d d d es an an an G G G e e Pr Pr Pr ith th ith i Fa Fa Fa Source: Survey of health workers conducted as part of this study 14. In LCA, the number of groups is prespecified, and the resulting model fit can be assessed by looking at the Akaike Information Criterion (AIC) and Bayesian information criterion (BIC) and other such goodness-of-fit criteria. We tested for a number of different groups, but the three patterns shown in Figures 4.7 and 4.8 remain the same (i.e., the levels of the different probabilities of the indexes change with additional groups, but the pattern does not change). 33 Figure 4.8. Profiles of Barangay Nutrition Scholars Based on Response Patterns to the Belief Indexes Class 1 Class 2 Class 3 3 3 3 2.5 2.5 2.5 Predicted mean 2 2 2 1.5 1.5 1.5 1 1 1 ed ed ed s te rty te rty s rty s te ic ic ic Fa Fa Fa ib ib ib ve ve ve et et et cr cr cr en en en Po Po Po d d d es es es an an an G G G Pr Pr Pr ith ith ith Fa Fa Fa Source: Survey of health workers conducted as part of this study The LCA also estimates the proportion of the different profiles or groups in the population of health workers. Table 4.7 shows the predicted proportions of the BHW and BNS populations in each of the groups shown above. It is interesting to note that Group 3, which has greater agreement with the genetics and faith and fate–based views, accounts for just below 20 percent for both BNSs and BHWs. However, overall proportions appear to be similar across the different types of workers. Table 4.7. Group Membership based on Mental Models Group BHW BNS 1 0.44 0.42 2 0.39 0.41 3 0.17 0.17 Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; BNS = Barangay nutrition scholar. Finally, this study considers whether these groupings are meaningful. Figure 4.9 shows the average self-reported service delivery by BHWs and BNSs (using the standardized service delivery index). The graph confirms a few findings discussed above, such as the fact that reported 34 service delivery is much higher for BNSs than for BHWs overall. Furthermore, Group 2 reports the highest service delivery—this is the same group that reports most agreement with prescribed practices and poverty views and least with the geneticsand faith and fate views. Also, when agreement with genetics and faith and fate are higher, reported service delivery is lower, as in the case of Group 3. The interesting insight here is from Group 1. When agreement with only poverty- related views is high (and not with prescribed practices), this group also reports lower service delivery. Figure 4.9. Mean of Standardized Service Delivery Index by Profile/Group of Health Workers .4 Mean of Standardized Service Index .2 0 -.2 -.4 1 2 3 1 2 3 BHW BNS Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; BNS = Barangay nutrition scholar. 35 IV. DISCUSSION Stunting rates have not declined in recent years in the Philippines. We employ methods drawn from cognitive anthropology to measure and interpret the deep-seated beliefs of different stakeholders. Initial qualitative discussions with caregivers, health workers, and policy makers showed that while factors related to a lack of nutrition and appropriate maternal and child care were often cited as important reasons for child malnutrition, genetic and racial reasons were also foremost for several stakeholders—and the discussions revealed strong tensions between the two, especially for health workers. To understand the importance ascribed to the different beliefs uncovered in these discussions, self-administered surveys were administered to over 5,000 frontline health workers (BNSs and BHWs). Analysis of the response patterns suggests that there are four underlying worldviews (or mental models) at play. The most dominant worldview adheres to conventional prescribed practices in improving child health outcomes and includes nutrition alongside prenatal care and attention paid to infants by parents. However, additional views held by respondents encompassed genetic and racial factors, faith- and fate-based factors, and factors related to poverty and the disease environment. We subsequently show that these views, along with knowledge of health workers, are strongly correlated with self-reported service provision at the local BHCs—views predicated on prescribed practices and poverty are positively correlated with service provision, while views predicated on genetics and faith are negatively correlated. These findings demonstrate the societal implications of the levels of knowledge and the different worldviews held by health workers. While most health workers command sufficient knowledge and subscribe to views that are consistent with conventional practices in improving child health, a minority of workers also agree with views on child stunting that include genetic and racial factors, alongside faith- and fate-based factors. BHWs and BNSs who have more robust, evidence-based knowledge and beliefs are more likely to work in facilities that provide better (self-reported) maternal and child nutrition services. Conversely, BHWs and BNSs who also see child height as a question of genetics, faith, or fate are less likely to work in facilities that provide better maternal and child nutrition services. We find suggestive evidence in comparisons with actual stunting rates from the 2018 National Nutrition Survey (NNS) that the mental models, knowledge, and services are better in provinces with lower-than-average stunting rates and conversely worse in provinces with higher stunting rates (see Annex A4). Profiling of health workers based on the different worldviews suggests that views on genetics and faith/fate can coexist with conventional wisdom on child health. This suggests that it is the relative intensity of these beliefs, rather than the presence or absence of a set of beliefs, that can influence service delivery at local health centers. The analysis also highlighted differences between the worldviews of BHWs and BNSs. BHWs were much more likely to agree with views ascribing child height to factors predicated on genetics and faith or fate than were BNSs. These findings are, however, not unexpected. The more focused training on nutrition of the BNSs reflects on their belief models and knowledge responses, which tend to be more aligned with health practices. Not having this benefit, BHWs more often reflect the belief systems of the communities they come from. These results may have implications, given the roles played by BHWs and BNSs in local health systems in the Philippines. While BNSs carry out more focused work based on their training and task assignments on nutrition services, BHWs are often asked to work on a much broader range of health services. Because there are more BHWs than BNSs, in practice, both groups of community health workers participate in each other’s programs, depending on the need for manpower. Hence, BHWs assist in providing vitamin A supplements, and BNSs, in turn, help out during immunization days. It is not inconceivable for 36 BHWs to be called upon to take on nutrition-related assignments, even with minimal or no training on the task and with supervision most likely from the midwife, or even the BNS. Changing beliefs and mental models of health workers (as well as caregivers and policy makers) is likely to require introducing innovative exercises at trainings, will need to be carefully tested, and is a useful direction for future research. Recent literature from the behavioral sciences offers some guidance; however, such interventions will still need to be customized and adapted to the local context and topic. The channels through which mental model change can occur includes critical reflection and discussion on the topic and exposure to alternative worldviews. Different scenarios are often discussed in an open environment, where participants are able to discuss their beliefs and actions, and are also able to observe the beliefs and actions taken by others. These sessions can be made more engaging if participants can role-play reactions to different scenarios presented to them or view entertaining audiovisual materials with compelling narratives. It is important for such sessions to be highly participative, as well as to show that many others are also being exposed to this new set of ideas. For example, changing men’s mental models in the domains of maternal and reproductive health, particularly in the domain of intimate partner violence (IPV), can be challenging, but recent studies have shown promising results—advocacy campaigns, community outreach, and workshops with men and boys in India, Brazil, and Chile lead to statistically significant changes in attitudes toward gender-based violence and decrease in IPV (Instituto Promundo 2012). A recent couples intervention in Rwanda increased male involvement in maternal and reproductive health and reduced IPV through critical reflection, dialogue, and participation in small group sessions (Doyle et al. 2018). An evaluation of interventions to promote male involvement in Bangladesh increased both micronutrient intake and dietary diversity for pregnant women through direct counseling, husbands’ forums, and videos (Nguyen et al. 2018). Different forms of “edutainment” have also been shown to be useful in changing mental models and influencing real life behavior. For example, access to a TV network in Brazil that was dominated by soap operas with independent female characters with few or even no children has been linked to the country’s rapid drop in fertility (La Ferrara, Chong, and Duryea 2012), and watching the program “MTV Shuga” in Nigeria increased the likelihood of being tested for HIV, as well as a reduction of sexually transmitted diseases among female participants (Banerjee et al. 2019). Furthermore, audience participation and role-playing (as well as observing others getting involved) can be transformational, as shown in a forthcoming paper where villagers exposed to street theater in India, including the chance to participate, were seen to have improved attitudes toward IPV. 15 An unequivocal finding from the qualitative interviews is that the beliefs of policy makers are important—those who held stronger beliefs about nutrition-related causes of stunting (and rejected genetic reasons) were also more likely to be actively involved in improving and implementing nutrition policy in their jurisdiction. Unsurprisingly, such policy makers were also likely to have a background in health or medicine, including physicians. This finding leads directly to the challenge of how to garner the support of policy makers who do not have such a background. 15. A paper based on the work of Karla Hoff at the World Bank should be available shortly. 37 Policy makers with backgrounds in health could serve as role models for other officials to help change social norms and mental models, which have proved to be effective when the audience can identify with the role models (e.g., Bernard et al. 2014). Officials can be further incentivized with tools such as social recognition of improved performance, as individuals may often be motivated by status and recognition (see Gauri et al. 2018, for an example). The Department of Health (DOH) has in fact already realized the importance of this for local policy makers. From 2013 to 2018, the DOH conducted a Health Governance and Leadership Program with the Zuellig Family Foundation for Mayors of municipalities identified as priorities by the National Anti-Poverty Commission. This program continues as the Institutionalizing Health Leadership and Governance Program (IHGLP) involving selected provinces and regions of the country. The program focuses on “capacitating local leaders to undertake health system reforms.” 16 The initial phase of the program has been able to demonstrate substantial reductions in child and maternal mortality in the LGUs enrolled in the program. In the past two years, its model has also included a nutrition component with the aim of reducing stunting. 16. ZFF 2016 38 V. Conclusions and Recommendations In conclusion, this study reveals a strong and urgent need for more innovative and competency- based training of BHWs and BNSs in the Philippines. It is imperative that these health workers command adequate knowledge of nutrition, shift beliefs toward acceptable practices, and are able to communicate their knowledge effectively so as to catalyze sound behavioral change. Education and communication on nutrition must be investigated and improved at the level of the training curriculum as well as of in-service education, such that any beliefs that deviate from conventional practices around child nutrition can be addressed. How might these results be of use to the Department of Health, which is ultimately responsible for the country’s health outcomes and on whose shoulders rest the burden of reducing childhood stunting? We propose a number of possible actions, particularly in the crisis situation brought about by the COVID-19 pandemic. As the world learns to live with the SARS-coronavirus-2, nutrition could be positioned as a way of strengthening the immune system so that people can be better protected and can more appropriately respond to treatments, including the anticipated COVID vaccine. It is crucial for frontline health workers such as the BNSs and BHWs to be appropriately equipped with updated nutrition knowledge and skills to respond to these emerging health and nutrition needs. The training and supervisory support provided to BNSs is far more systematic than that provided to BHWs. The DOH might consider providing nutrition training to the BHWs through the National Nutrition Council (NNC). This would ensure consistency in training content and methods and might be more efficient than providing a separate training process for the BHWs. The DOH could support cross-training to enable BHWs to attend BNS training or support regional-level training to be delivered by academic institutions. The latter is an approach that the NNC has used for its BNS training. Post-training, these BHWs could be supervised by the Nutrition Action Officer(NAO) for nutrition-related activities and performance. Given the extent of power local policy makers have over their jurisdictions, it is not surprising that their belief systems would strongly influence their priorities and programs. It would also not be surprising to find that these local policy makers’ beliefs influence their health workers, who are beholden to them for their benefits. The DOH needs to continue its ongoing programs to improve local health systems performance, involving local policy makers such as mayors and governors. The DOH can include nutrition modules and discussions during the sessions, particularly on child growth and stunting. Provincial and municipal models have been developed by DOH partners in this regard. 17 Mayors and governors could be motivated to integrate nutrition in their food production, environment, and related projects as part of the Philippine government’s localization efforts for the Sustainable Development Goals (SDGs). The PPAN 18 2017–2022 provides the policy mandate for the “intensified mobilization of local governments” as well as complementing of national and local government actions in the delivery of nutrition outcomes. There are best practices from the NNC’s National Nutrition 19 Awardees. 17. DOH and Zuellig Family Foundation’s Health Leadership and Governance Program. 18. National Nutrition Council 2017. 19. Consistent Regional Outstanding Winner in Nutrition Award, Nutrition Honor Award, and the National Outstanding Barangay Nutrition Scholar Award. 39 While training will be necessary, it may not be sufficient to ensure that these health workers are capacitated to implement stunting reduction programs. In this regard, the DOH could consider reviewing the roles and expectations of community health workers such as the BNSs and the BHWs. They play vital roles in the health system but are not given the recognition and benefits they deserve, and this affects the low motivation that was apparent in poor nutrition regions. Issues that need to be addressed could include ensuring their security of tenure as well as formulating and enforcing standards for their qualifications, recruitment, training, performance, and retention. The study focused on the beliefs, knowledge, and practices of BNSs and BHWs regarding the nutrition-specific actions they undertake. BNSs, as part of their community functions, are also expected to participate in nutrition-sensitive programs, and they need to be supported in this role. The DOH, together with other government agencies, can provide incentives for the attainment of nutrition-related milestones, particularly those related to the SDGs, such as reductions in stunting and wasting. The incentives can be cross-cutting, for example, to support the achievement of better diet diversity through an agricultural crop-diversification program. The latter is an existing Department of Agriculture program that has been enhanced in response to the effects of COVID- 19 on the food supply chain. The DOH could complement this initiative by including the incentive to link up this program to a complementary feeding program that BNSs/BHWs could implement and also use to monitor nutritional outcomes of the program’s beneficiaries. The power of beliefs and knowledge to influence action can be enabled through a broader approach involving both national and local governments. The scope of and support for nutrition actions go beyond the DOH. Enhanced training of BNSs and BHWs will mark a key first of many actions toward NNC’s “all of government, all of society” call for better nutrition to be achieved. 40 References Banerjee, Abhijit, Eliana La Ferrara, Orozco Olvera, and Victor Hugo. 2019. “The Entertaining Way to Behavioral Change: Fighting HIV with MTV (English).” Policy Research Working Paper No. WPS 8998, World Bank Group, Washington, DC. Banuri, S., S. Dercon, and V. Gauri. 2017. “Biased Policy Professionals.” World Bank Policy Research Working Paper 8113, World Bank, Washington, DC. Bernard, Tanguy, Stefan Dercon, Kate Orkin, and Alemayehu Seyoum Taffesse. 2014. “The Future in Mind: Aspirations and Forward-Looking Behaviour in Rural Ethiopia.” Centre for the Study of African Economies, University of Oxford, Oxford. Briones, Roehlano, Ella Antonio, Celestino Habito, Emma Porio, and Danilo Songco. 2017. Strategic Review: Food Security and Nutrition in the Philippines. DiMaggio, P. 1997. “Culture and Cognition.” Annual Review of Sociology 23: 263–87. doi:10.1146/annurev.soc.23.1.263 Doyle, K., R. G. Levtov, G. Barker, G. G. Bastian, J. B. Bingenheimer, S. Kazimbaya, et al. 2018. “Gender-transformative Bandebereho Couples’ Intervention to Promote Male Engagement in Reproductive and Maternal Health and Violence Prevention in Rwanda: Findings from a Randomized Controlled Trial.” PLoS ONE 13 (4): e0192756. https://doi.org/10.1371/journal.pone.0192756. Gauri, Varun, Julian C. Jamison, Nina Mazar, Owen Ozier, Shomikho Raha, and Karima Saleh. 2018. “Motivating Bureaucrats through Social Recognition: Evidence from Simultaneous Field Experiments.” Policy Research Working Paper No. 8473, World Bank, Washington, DC. https://ssrn.com/abstract=3238307. Graham, W. J., and B. Varghese. 2012. “Quality, Quality, Quality: Gaps in the Continuum of Care.” The Lancet 379 (9811): e5–6. Instituto Promundo. 2012. Engaging Men to Prevent Gender-based Violence: A Multi-country Intervention and Impact Evaluation Study. Report for the UN Trust Fund. Washington, DC: Promundo. La Ferrara, Eliana, Alberto Chong, and Suzanne Duryea. 2012. “Soap Operas and Fertility: Evidence from Brazil.” American Economic Journal: Applied Economics 4 (4): 1–31. Mannava, P., K. Durrant, J. Fisher, M. Chersich, and S. Luchters. 2015. “Attitudes and Behaviors of Maternal Health Care Providers in Interactions with Clients: A Systematic Review.” Global Health 11: 36. doi: 10.1186/s12992-015-0117-9. Magbitang, J., J. Tangco, E. de la Cruz, E. Flores, and F. Guanlao. 1988. “Weight-for-Height as a Measure of Nutritional Status in Filipino Pregnant Women.” Asia Pacific Journal of Public Health 2 (2): 96–104. doi:10.1177/101053958800200204. 41 NNC (National Nutrition Council). 2017. Philippine Plan of Action for Nutrition 2017–2022. Executive Summary (final). http://www.nnc.gov.ph/downloads/technical- papers?download=870:philippine-plan. Nguyen, P. H., E. A. Frongillo, T. Sanghvi, G. Wable, Z. Mahmud, L. M. Tran, B. Aktar, K. Afsana, S. Alayon, M. T. Ruel, and P. Menon. 2018. “Engagement of Husbands in a Maternal Nutrition Program Substantially Contributed to Greater Intake of Micronutrient Supplements and Dietary Diversity during Pregnancy: Results of a Cluster-Randomized Program Evaluation in Bangladesh.” J Nutr. 148 (8): 1352–63. doi:10.1093/jn/nxy090. PubMed PMID: 29931108; PubMed Central PMCID: PMC6075465. Raman, Shanti, Rachel Nicholls, Jan Ritchie, Husna Razee, and Samaneh Shafiee. 2016. “Eating Soup with Nails of Pig: Thematic Synthesis of the Qualitative Literature on Cultural Practices and Beliefs Influencing Perinatal Nutrition in Low and Middle Income Countries.” BMC Pregnancy and Childbirth 16 (192). 10.1186/s12884-016-0991-z. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture and Consensus: A Theory of Culture and Informant Accuracy. American Anthropologist 88 (2): 313–38. Sunguya, B. F., K. C. Poudel, L. B. Mlunde, D. P. Urassa, J. Yasuoka, and M. Jimba. 2013. “Nutrition Training Improves Health Workers’ Nutrition Knowledge and Competence to Manage Child Undernutrition: A Systematic Review.” Frontiers in Public Health 1: 37. doi: 10.3389/fpubh.2013.00037 Weller, Susan C. 2007. “Cultural Consensus Theory: Applications and Frequently Asked Questions.” Field Methods 19 (4): 339–68. World Bank Group. 2015. World Development Report 2015: Mind, Society, and Behavior. ZFF (Zuellig Family Foundation) 2016. Journey towards Sustainable Development Goals: Sustainability Report 2014–2015. Parañaque City, the Philippines: ZFF. 42 ANNEX A1. BARANGAY HEALTH WORKERS AND NUTRITION SCHOLARS: BACKGROUND Barangay Nutrition Scholars (BNSs) are volunteer community workers who are trained in implementing health and nutrition programs in a locality. BNSs undergo “didactic training” (approximately 10 days), which includes food production, food fortification, nutrition information and education, maternal and child health and nutrition, and livelihood assistance, as well as 20 days of practical work, during which they learn how to weigh preschoolers accurately, interview mothers comprehensively on child-rearing practices, collect and analyze data using family and barangay profile forms, and formulate an action plan. When their training is concluded, BNSs are considered to be well qualified to identify and monitor pregnant and lactating mothers and malnourished children and provide counseling on good nutrition practice. Presidential Decree No. 1569 of 1978 mandated the deployment of one BNS in every barangay in the country to monitor the nutritional status of children and/or link communities with nutritional and related service providers. As of August 2017, there were 46,293 BNSs deployed across the country. Barangay health workers (BHWs) are likewise volunteers who undergo a basic training program (about five weeks) under an accredited government or nongovernment organization and render primary care services in the community (maternal, newborn, and child health). BHWs are accredited to function as such by the local health board in accordance with the guidelines promulgated by the Department of Health (DOH) (Section 3 of Republic Act No. 7883). BHWs live in the communities they serve and are expected to act as change agents in their communities. They provide information, education, and motivation services for primary health care, maternal and child health, child rights, family planning, and nutrition. They may administer immunizations and regular weighing of children and often assist midwives in providing birthing services. As of the end of 2017, there were 276,919 active BHWs across the country. Both BNSs and BHWs take part in Operation Timbang Plus (OPT-Plus), an annual weight-taking activity conducted by local government units (LGUs) since the 1970s, but which, since January 2012, included length/height-taking as well. At the barangay level, a midwife or a rural health midwife (RHM) leads the OPT-plus team, assisted by the BNS and BHW, and, when requested, the barangay committee chair on health and nutrition, Sangguniang Kabataan (Youth Council) chairperson, other BHWs, day care workers, and teachers-in-charge as ad hoc members of the team. Leaders of zones (puroks) or mothers’ groups, other community leaders, and representatives from civic organizations can also be part of the team. It is the BNS, however, who is tasked with preparation for OPT-Plus and follow-up. The BNS will first inform the community of the OPT-Plus schedule, prepare equipment and supplies, and print the previous year’s master list. Follow-up will entail various tasks, including encoding or manually consolidating results for presentation to the Barangay Nutrition Council and on a monthly basis weighing all preschoolers identified as underweight during the OPT operations. The BNS is also expected to weigh all infants age 0 to 24 months every month to monitor their growth and, to the extent possible, weigh every child age 25 to 71 months on a quarterly basis. The regular weighing provides the basis for corrective actions, which may include referral to the appropriate service or implementation of nutrition projects in coordination with the community. BNSs and BHWs have been described as “ill equipped” to handle the caseloads of malnourished children within their communities (Briones et al. 2017). Among the challenges identified were 43 inadequate training, lack of equipment, multiple roles, and poor pecuniary incentives. For instance, although BNSs are ideally supposed to have 30 days of training, resource constraints can, in some cases, shrink the 30-day training period to 3 to 4 days, with the practicum (practical work) phase incorporated into the BNS’s service period. Ad hoc OPT-Plus team members (and even BNSs) may also lack the skills to conduct anthropometric measurements, affecting the quality of the data with which BNSs are expected to work (Villorente 2019). Also, many assigned nutrition workers have designations and responsibilities in addition to their nutrition work; BNSs assist municipal health, nutrition, and social workers as well as agriculture workers bring services to more remote and traditional elements of society. However, BNSs only receive a transportation allowance from NNC and an additional monthly allowance from the LGU, which ranges from PHP 50.00 in poor barangays or municipalities, to PHP 3,000 in rich cities (Briones et al. 2017). Although other incentives exist—for instance, second-grade civil service eligibility for those who have completed two years of satisfactory service—a BNS has no security of tenure and serves only at the pleasure of the incumbent (Briones et al. 2017). BHWs are slightly better off than BNS with respect to benefits and incentives. This is due to R.A. 7883, the Barangay Health Workers’ Benefits and Incentives Act of 1995. Among the incentives provided by the law are hazard allowance and subsistence allowance; training, education, and career enrichment programs; preferential access to loans; and civil service eligibility. Amounts in R.A. 7883 are observed to be outdated, however, with BHWs still receiving a monthly honorarium at the discretion of local leaders. 44 A2. ADDITIONAL TABLES AND REGRESSIONS Table A2.1. Frequencies of Factors Mentioned in Written Exercise during Qualitative Data Collection by Stakeholder Davao del Sur Eastern Samar Isabela Maguindanao Mt. Province Factors C HW P C HW P C HW P C HW P C HW P Total Nutrition 9 9 3 8 9 3 8 12 3 10 8 2 9 10 3 106 Hereditary 0 3 2 5 6 1 0 4 2 0 1 0 3 8 2 37 Sleep 4 1 1 3 0 0 0 1 1 3 0 0 1 1 0 16 Income 0 0 0 1 2 1 0 6 0 0 5 0 0 0 0 15 Parental care 0 1 0 0 2 0 1 6 0 1 3 0 1 0 0 15 Labor 0 0 1 2 1 0 0 0 0 1 1 0 0 0 1 7 Environment 0 0 1 1 1 0 0 1 0 0 0 1 2 0 0 7 Disease 0 0 0 0 1 0 1 0 3 0 0 0 0 1 0 6 Exercise 1 2 0 0 0 0 0 0 0 1 0 0 1 0 0 5 Pregnancy 0 0 0 0 0 0 0 2 0 0 0 0 0 1 0 3 Source:Written exercises conducted as part of qualitative data collection in this study Note:C = Caregiver; HW = Health Worker; P = Policymaker Table A2.2. Likert Scale Responses to Belief Questions by Barangay Nutrition Scholars and Barangay Health Workers BNS BHW Does not Does not Slightly Affects Greatly affect Slightly Affects Greatly affect (%) affects (%) (%) affects (%) (%) affects (%) (%) affects (%) Ethnicity 63.79 20.47 13.70 2.03 55.18 22.88 18.01 3.93 Food 15.82 20.76 39.64 23.78 19.05 25.00 38.14 17.81 Destiny 53.47 20.95 20.39 5.18 43.37 23.56 24.94 8.13 Sufficient sleep 35.40 14.35 31.77 18.48 38.09 17.07 27.92 16.93 Help family 31.78 24.62 31.44 12.16 31.06 25.52 31.29 12.14 Carry load 31.41 27.24 30.06 11.28 25.14 29.26 32.55 13.05 Poverty 13.40 14.16 35.00 37.44 14.76 16.82 33.59 34.83 God 68.93 11.53 11.66 7.89 62.36 13.67 15.19 8.79 Parental care 26.39 13.09 28.67 31.84 32.95 16.27 29.15 21.63 Disease 8.30 12.91 39.27 39.52 9.46 17.29 37.72 35.53 Access 26.28 13.95 33.83 25.94 35.42 16.43 27.69 20.46 Faith 60.03 17.13 15.96 6.88 55.68 16.34 18.36 9.62 Height of mother 49.25 27.95 18.95 3.85 42.64 27.33 24.53 5.51 Height of father 49.31 27.85 18.86 3.97 43.45 26.39 24.29 5.87 Breastfeed ing 40.56 6.82 21.90 30.71 46.41 10.55 21.79 21.24 Luck 72.21 13.08 10.46 4.25 62.76 17.46 13.90 5.88 Prenatal care 27.55 11.09 23.61 37.75 34.05 14.25 25.83 25.87 45 Source: Survey of health workers conducted as part of this study Note: BNSs = Barangay nutrition scholars; BHWs = Barangay health workers. Table A2.3. Differences in Knowledge for Barangay Nutrition Scholars and Barangay Health Workers across Regions BNS BHW Region Region IV VII XI XII XIII I IV V VI XI XV “Filipinos are in general short by nature”a 2.89 2.94 3.23 3.04 3.38 2.87 2.90 2.36 2.94 2.74 2.55 What is your best guess on the percentage of children under age 5 who are severely short for their age in the Philippines? (%) 37.46 41.00 41.84 33.12 36.43 42.03 37.52 49.05 43.72 46.66 44.91 What is your best guess on the percentage of children under age 5 who are severely short for their age in your province? (%) 29.11 37.76 37.60 29.51 30.83 37.48 30.51 43.09 40.95 42.99 35.10 "Stunting is a problem" (fraction) 0.90 0.89 0.75 0.86 0.93 0.68 0.69 0.88 0.61 0.59 0.80 Recommended weight that a pregnant woman should gain during her pregnancy (kg) 9.29 13.92 9.19 7.06 7.43 14.23 10.07 14.57 10.26 12.65 14.38 How often do you recommend that a pregnant woman eat less food than normal during pregnancy?b 3.09 1.96 1.74 2.26 3.48 2.49 2.63 2.07 2.25 1.73 2.15 Index of knowledge question (# correct out of 8) +++ 4.81 4.51 4.23 4.08 5.62 3.42 4.60 2.53 4.28 3.92 3.72 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHS = Barangay health worker; Only includes regions where at least 100 workers were interviewed. a. Rating, from 1 = Strongly disagree, to 5 = Strongly agree. b. Rating from 1 = Never, to 5 = Always. +++ Ranging from 0 to 8 correct answers. Region legend: I = Ilocos, III = Central Luzon, IV = Calabarzon, V = Bicol, VI = Western Visayas, VII = Central Visayas, XI = Davao, XII = Soccsksargen, XIII = NCR (National Capital Region), XV = ARMM (Autonomous Region in Muslim Mindanao). ***p < 0.01, **p < 0.05, *p < 0.1. 46 Table A2.4. Vignette Analysis of Main and Interaction Effects of the Three Factors (Tall, Well off, Nutritious food) (1) (2) (3) (4) (5) (6) Stunted Stunted Stunted Stunted Stunted Stunted Sample Full BHW BNS Full BHW BNS tall 0.0138 −0.0266 0.0532*** −0.0248 −0.0652* 0.0189 (0.0133) (0.0207) (0.0169) (0.0230) (0.0360) (0.0293) well_off 0.0498*** 0.0630*** 0.0386** 0.0378 0.0313 0.0474 (0.0133) (0.0207) (0.0169) (0.0230) (0.0361) (0.0292) nutritious −0.527*** −0.435*** −0.609*** −0.523*** −0.423*** −0.600*** (0.0133) (0.0207) (0.0169) (0.0234) (0.0376) (0.0291) tall + well_off 0.0537** 0.0779* 0.0357 (0.0266) (0.0415) (0.0338) tall + nutritious 0.0224 −0.00575 0.0328 (0.0266) (0.0414) (0.0339) well_off + nutritious −0.0303 −0.0179 −0.0513 (0.0266) (0.0415) (0.0339) Constant 0.748*** 0.701*** 0.790*** 0.760*** 0.717*** 0.794*** (0.0134) (0.0212) (0.0168) (0.0178) (0.0287) (0.0221) Observations 4,089 1,895 2,194 4,089 1,895 2,194 R-squared 0.279 0.192 0.372 0.280 0.194 0.374 Source: Survey of health workers conducted as part of this study Note: BHS = Barangay health worker; BNS = Barangay nutrition scholar; Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. R-squared is the proportion of variance in the dependent variable that is explained by the independent variables. 47 Table A2.5. Vignette Analysis of Main Effects by Region (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Region Central Western Central Ilocos Luzon Calabarzon Bicol Visayas Visayas Davao Soccsksargen NCR ARMM Outcome Stunted Stunted Stunted Stunted Stunted Stunted Stunted Stunted Stunted Stunted tall −0.0550 0.283*** −0.0334 −0.0229 −0.0888* −0.0416 −0.00358 0.0594* 0.0317 −0.00174 (0.0533) (0.0406) (0.0239) (0.0603) (0.0537) (0.0419) (0.0420) (0.0332) (0.0383) (0.0531) well_off 0.0615 0.0942** 0.0123 0.0144 0.157*** 0.0877** −0.0136 0.0453 −0.0184 0.142*** (0.0536) (0.0406) (0.0239) (0.0611) (0.0537) (0.0419) (0.0420) (0.0332) (0.0380) (0.0529) −0.827* nutritious −0.469*** −0.437*** −0.726*** −0.429*** −0.164*** −0.601*** −0.423*** −0.586*** ** −0.345*** (0.0499) (0.0406) (0.0239) (0.0606) (0.0537) (0.0419) (0.0420) (0.0332) (0.0389) (0.0527) 314 426 835 221 329 360 466 597 240 301 0.231 0.284 0.526 0.189 0.058 0.374 0.181 0.347 0.660 0.148 Source: Survey of health workers conducted as part of this study Note: NCR = National Capital Region; ARMM = Autonomous Region in Muslim Mindanao; Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. 48 Table A2.6. Factor Analysis of Causes of Stunting Variable Factor1 Factor2 Factor3 Factor4 Uniqueness f1_ethni 0.3757 0.7962 f2_food 0.4996 0.3387 0.6107 f3_destiny 0.3737 0.7426 f4_sufsleep 0.6243 0.5287 f5_helpfam 0.4623 0.6135 f6_carry 0.5135 0.6654 f7_poverty 0.5866 0.5959 f8_god 0.5539 0.5762 f9_parentc~e 0.7252 0.3999 f10_disease 0.3543 0.4681 0.6469 f11_access 0.5129 0.6385 f12_faith 0.3482 0.5856 0.4727 f13_hmother 0.9359 0.0883 f14_hfather 0.9352 0.0828 f15_breastf 0.7626 0.3439 f16_luck 0.5815 0.5963 f17_prenatal 0.7743 0.3445 (blanks represent abs(loading)<.3) Source: Survey of health workers conducted as part of this study Note: Factors 1,2,3,4 show the factor loadings of the different belief questions after exploratory factor analysis and rotation Figure A2.1. Screen Plot Showing That Four Factors Are Relevant Scree plot of eigenvalues after factor 5 4 Eigenvalues 3 2 1 0 0 5 10 15 20 Number Source: Survey of health workers conducted as part of this study 49 Table A2.7. Regression Showing Different Service Index Outcomes (1) (2) (3) (4) (5) (6) supplement supplement_in advice_inde advice_inde measurement_ measureme Variables _index_std dex_std x_std x_std index nt_index Sample BHW BNS BHW BNS BHW BNS prescribed_practice_i ndex_std 0.132*** 0.156*** 0.0920*** 0.0685*** 0.0716*** 0.0740*** (0.0327) (0.0213) (0.0316) (0.0209) (0.0154) (0.0125) genetics_index_std 0.00300 −0.0335 0.00728 −0.0428* −0.0332** −0.00481 (0.0270) (0.0220) (0.0261) (0.0221) (0.0142) (0.0144) faith_fate_index_std −0.0237 −0.0693*** −0.0348 −0.0265 0.0134 −0.0366*** (0.0284) (0.0232) (0.0289) (0.0206) (0.0147) (0.0140) poverty_index_std 0.0408 0.0673*** 0.0950*** 0.0479** 0.0173 0.0206 (0.0274) (0.0257) (0.0283) (0.0240) (0.0145) (0.0137) knowledge_index_std 0.120*** 0.0994*** 0.121*** 0.0805*** 0.0521*** 0.0431*** (0.0267) (0.0265) (0.0246) (0.0226) (0.0118) (0.0121) Controls Provincial poverty −0.0128*** −0.000416 −0.0113** 7.19e−05 −0.00661*** −0.00255* (0.00467) (0.00259) (0.00520) (0.00218) (0.00234) (0.00136) Age −0.00550* −0.00530** −0.00824*** −0.00398* −0.00370** −0.00249* (0.00310) (0.00225) (0.00275) (0.00224) (0.00162) (0.00127) Education 0.0550*** 0.0375** 0.0373* −0.00918 0.00900 0.00409 (0.0198) (0.0177) (0.0193) (0.0179) (0.0109) (0.0110) Years of service 0.00209 0.00345 0.00573* −0.00141 0.00211 0.00126 (0.00304) (0.00287) (0.00296) (0.00302) (0.00147) (0.00185) Observations 1,748 1,993 1,945 2,161 1,967 2,172 R-squared 0.114 0.102 0.063 0.055 0.079 0.051 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHW = Barangay health worker; Robust standard errors in parentheses clustered at the Barangay level. All specifications include region fixed effects. ***p < 0.01, **p < 0.05, *p < 0.1. 50 Table A2.8. Descriptive Statistics on Consequences of Child Height Ran Total sample Rank BNS Rank BHW k Bullied in school 3.81 1 3.94*** 1 3.66 1 Complete secondary school 3.62 2 3.64 2 3.60 2 Get married as an adult 3.46 3 3.50** 3 3.42 3 Get a good job as an adult 3.40 4 3.41 4 3.40 4 Get good grades in school 3.30 5 3.31 6 3.30 5 Get very sick as an adult 3.22 6 3.32*** 5 3.12 6 See well in the dark 3.15 7 3.18* 7 3.12 6 Win the lottery as an adult 3.02 8 3.03 8 3.01 7 Play professional basketball 2.53 9 2.46*** 9 2.62 8 Source: Survey of health workers conducted as part of this study Note: BNS = Barangay nutrition scholar; BHW = Barangay health worker; + Rating from 1 = Much less likely, to 5 = Much more likely. ***p < 0.01, **p < 0.05, *p < 0.1 Table A2.8 shows agreement with different scenarios if a child is short for his or her age. Health workers do not perceive any serious consequences arising from children being short (this confirms similar findings from the focus groups). Respondents rated the likelihood (ranging from “much less likely” to “much more likely”) of a shorter child experiencing a set of conditions later in life, and aggregated responses are shown in the table. There was agreement between BNSs and BHWs on responses to this set of questions. Completing secondary school and getting a good job as an adult were ranked second and fourth, respectively; however, these were scored similar to being bullied in school (first) and getting married as an adult (ranked third). 51 A3. RELIABILITY AND VALIDITY OF BELIEF INDEXES We examine the reliability of the belief measurements and calculate Cronbach’s alpha score. For response patterns, the score serves as a measure of internal consistency, showing how closely related a set of items are as a group. Table A3.1 shows high alpha scores and therefore reliability for all the belief items together, as well as the subsets of items that fall under Prescribed practices, Genetics, Faith and fate, and Poverty. Table A3.1. Cronbach’s Alpha for Belief Items and Indexes Items Cronbach’s alpha All items 0.87 Prescribed practices 0.85 Genetics 0.79 Faith and fate 0.69 Poverty 0.66 Source: Survey of health workers conducted as part of this study Table A3.2 shows construct validity of the four mental models (Prescribed Practices, Genetics, Faith and Fate, and Poverty) derived from factor analysis. We correlate with responses to the statement “Filipinos are in general short by nature” (answer options are 1 = Strongly agree, to 6 = Strongly disagree) in Columns 1 to 3. As expected, we see that those who think that not following prescribed practices are likely causes of stunting are also more likely to disagree with the statement. Conversely, those who think that reasons based on faith/fate and genetics are more likely causes of stunting are also more likely to agree with this statement; however, it is interesting to note that the faith- or fate-based relationship is true for BHWs, and the genetic reasoning is true for BNSs. There is also a weak negative association for BNSs on poverty-based causes and agreement with this statement. Columns 4 to 6 show that correlation with the statement “Stunting is a problem” yields similar results (answer options were 1 = Yes, 0 = No). Table A3.2 shows construct validity for the mental model (indexes). Health workers who are more likely to acknowledge that “stunting is a problem” in the Philippines are also more likely to agree with factors related to prescribed practices and poverty and to be more knowledgeable and less likely to agree with genetics and faith and fate–related factors. Table A3.2. Construct Validity of Belief and Knowledge Indexes (1) (2) (3) “Stunting is a problem” Full sample BHW BNS prescribed_practices_index_std 0.0488*** 0.0397*** 0.0298*** (0.00810) (0.0150) (0.00858) genetics_index_std −0.0167** 0.00184 −0.0104 (0.00733) (0.0129) (0.00796) Faith_Fate_index_std −0.0485*** −0.0625*** −0.0196** (0.00737) (0.0127) (0.00813) poverty_index_std 0.0300*** 0.0431*** 0.0128 (0.00737) (0.0121) (0.00829) knowledge_index_std 0.0202** −0.00438 0.0144 (0.00799) (0.0131) (0.00922) Observations 3,922 1,812 2,110 52 R-squared 0.028 0.023 0.012 Source: Survey of health workers conducted as part of this study Note: BHW = Barangay health worker; BNS Barangay nutrition scholar; Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. A4. COMPARISON WITH REGIONAL LEVEL STUNTING RATES How do the survey measures of knowledge, mental models, and services compare with actual stunting rates across regions in the Philiipines? This comparison is useful for at least two reasons; first, it helps validate the measurements in the health worker survey, and, second, it establishes that stunting rates in regions do in fact covary with (aggregated) health worker knowledge, beliefs, and service delivery. However, since stunting rates are available at the province level only for some provinces from the most recent NNS, we make use of this to broadly classify regions as better nutrition (where under- five stunting rates are less than 30 percent, the Philippine stunting rate average in 2018 from the NNS) or poor nutrition (stunting rates greater than 30 percent). 20 The stunting rates from 2018 are the average provincial rates from the regions surveyed in the NNS in 2018, which was completed just before the health worker survey. 21 As a comparison, the NNS stunting rates in these regions are also shown for 2015. Table A4.1. Regional Stunting Rates in the Philippines in 2018 and 2015 Region Regional level stunting Regional level stunting “Better” Nutrition rates from 2018 NNS rate from 2015 NNS Region Classification I - Ilocos . 21.5 Omitted III - Central Luzon 23.7 16.6 Yes IV - Calabarzon 29.6 19.0 Yes V - Bicol 29.5 28.4 Yes VI - Western Visayas 33.1 26.6 No VII - Central Visayas 23.8 22.8 Yes XI - Davao 35.9 20.7 No XII - Soccsksargen 43.7 25.8 No XIII - NCR 23.8 15.1 Yes XV - ARMM 30.3 24.9 No Source:National Nutrition Surveys conducted in the Philippines in 2015 and 2018 Note: NCR = National Capital Region; ARMM = Autonomous Region in Muslim Mindanao; NNS = National Nutrition Survey. Figure A4.1 shows that standardized indexes of the health worker survey measures of knowledge, beliefs, and services vary as expected for the most part. For example, knowledge is very different, and services and prescribed practices are also higher as expected in better nutrition regions. Belief in faith and fate is strikingly higher across this classification, and poverty-related beliefs are 20. The most recent NNS is a multiyear survey, and the data are from provinces where the survey was carried out in 2018 only (which are also the most recent data available and for about a third of total provinces in the country). The survey was continued in 2019 (but data are still unavailable and paused in 2020 because of COVID-19). 21. Unlike earlier surveys, the 2018 Expanded National Nutrition Surveys (eNNS) generated provincial-level stunting rates. However, since the provinces do not necessarily coincide with the provinces in the survey of health workers, we aggregate and carry out regional-level comparisons. Region 1, Ilocos, is excluded from this analysis as no provinces were surveyed in the NNS from this region in 2018. 53 also higher in poorer nutrition regions, as expected. However, the beliefs related to genetics are higher in the better nutrition regions, somewhat counterintuitively. Figure A4.1. Comparison of Knowledge, Mental Models, and Services across Better and Poorer Nutrition Regions .2 .1 0 prescribed genetics faith poverty -.1 knowledge services Poor nutrition Better nutrition Source: Survey of health workers conducted as part of this study 54 Declines in rates of child stunting in the Philippines have decelerated, making it hard for the country to achieve its targets on nutritional outcomes. The knowledge base, beliefs, and practices of caregivers have been extensively researched, but little is known about how health workers and policy makers fare in comparison. We conduct qualitative interviews, striving to preclude bias as we capture these stakeholders’ views on factors that affect stunting, and go on to compare and contrast these perceptions. We subsequently investigate the importance of the different factors in detail through a large-scale quantitative survey with frontline health and nutrition workers. The findings suggest that while most workers’ knowledge and beliefs are consistent with accepted practices, important deviations from consensus views exist, and these are correlated with worse self-reported service delivery outcomes at local health centers. The findings suggest that in the Philippines any endeavor to further improve service delivery must take into consideration the beliefs of frontline workers. ABOUT THIS SERIES: This series is produced by the Health, Nutrition, and Population Global Practice of the World Bank. The papers in this series aim to provide a vehicle for publishing preliminary results on HNP topics to encourage discussion and debate. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they represent. Citation and the use of material presented in this series should take into account this provisional character. For free copies of papers in this series please contact the individual author/s whose name appears on the paper. Enquiries about the series and submissions should be made directly to the Editor Martin Lutalo (mlutalo@ worldbank.org) or HNP Advisory Service (askhnp@worldbank.org, tel 202 473- 2256). For more information, see also www.worldbank.org/hnppublications. 1818 H Street, NW Washington, DC USA 20433 Telephone: 202 473 1000 Facsimile: 202 477 6391 Internet: www.worldbank.org E-mail: feedback@worldbank.org