OPERATIONALIZING A MULTI-SECTORAL APPROACH FOR THE REDUCTION OF STUNTING IN INDONESIA AN APPLICATION USING THE 2007 AND 2013 RISKESDAS OPERATIONALIZING A MULTI-SECTORAL APPROACH FOR THE REDUCTION OF STUNTING IN INDONESIA AN APPLICATION USING THE 2007 AND 2013 RISKESDAS FEBRUARY 2017 iv Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting v Acknowledgement The analysis in this report was carried out as part of the Cooperation Agreement between a team in the World Bank and the National Institute of Health Research and Development (NIHRD) of the Ministry of Health of the Republic of Indonesia. The Task Team Leader (TTL) was Emmanuel Skoufias, Lead Economist, Poverty and Equity Global Practice. The World Bank team included: Dr. Eko Setyo Pambudi (Health Analyst), Indira Maulani Hapsari (Research Analyst), Ratih Dwi Rahmadanti (Research Analyst) from the Bank’s Jakarta office, Dr. Katja Vinha (International Consultant), and Dr. Atmarita (International Consultant). The NIHRD team consisted of the following: Dr. Ni Ketut Aryastami (chairman of the NIHRD’s Team), Dr. Ingan Ukur Tarigan (member), Nunik Kusumawardani, PhD (member), Siti Isfandari, MA (member), Prisca Petty Arfines, MPH (member), Olwin Nainggolan, MKM (member), Djunaedi, SKM (member). The teams are grateful to Dr. Siswanto (Head of the NIHRD) for his support throughout the process of this report and to the Bank’s Health and Water teams in Jakarta for useful comments. This report was also was funded by the Water Global Practice and the Research Support Budget (RSB) of the Development Economics Vice Presidency of the World Bank. vi Executive Summary Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting vii Motivation development to strengthen its nation. Along parallel lines, initiatives within the World Bank and other development agencies and research institutions, aim Stunting is a widespread and persistent condition in to foster knowledge exchange and cross-sectoral Indonesia with more than one-third of young children collaboration and coordination at the project level for being stunted. The national stunting rate for under improving nutrition (World Bank, 2013). As of 2016, five-year-olds increased slightly from 36.8 percent the “Investing in the Early Years” initiative, adopts in 2007 to 37.2 percent in 2013, based on official a stepwise approach to the nature of intervention stunting rates reported by the Ministry of Health needed for the healthy physical and cognitive of the Government of Indonesia. During the same development of children by emphasizing the role of time period, the percentage of population in poverty reducing stunting and undernutrition for children reduced from 16.6 percent to 11.4 percent (World in their first 1,000 days of their lives (including 9 Bank), suggesting that the increased purchasing power months in utero), the role of education and stimulation did not translate to better nutritional outcomes for between 1,000 and 2,000 days and the role of social children. protection for the nutrition and health of children greater than 2,000 days. All these initiatives are based An acceleration of the progress towards reducing on the premise that the determinants of malnutrition stunting in Indonesia, requires enlisting more sectors, are multi-sectoral and that the solution to malnutrition in addition to the health sector, such as agriculture, requires multi-sectoral approaches. education, social protection, and water, sanitation, and hygiene in the effort to improve nutrition. Large scale Indonesia’s commitments to reducing stunting is “nutrition sensitive” interventions in these sectors will evidenced by the wide variety “nutrition-specific” and have to be able not only to address the key underlying “nutrition-sensitive” interventions. Nutrition-specific determinants of nutrition effectively, but also interventions are in place to address nutritional intensify the role of “nutrition-specific” interventions deficiencies at every point in the life-cycle beginning (Black et al., 2013).1 with folate and calcium supplementation, as well as supplemental feeding for malnourished pregnant In recent years there has been a significant increase in mothers, breastfeeding promotion and counseling the number of initiatives at the international as well for lactating mothers, growth monitoring, vitamin A as at the county level aiming to scale up nutrition- supplementation, iodization, supplemental feeding, sensitive interventions. One prominent example is fortification and therapeutic zinc supplements for the Scaling-Up Nutrition (SUN) movement, launched diarrhea management and deworming prevention in April 2010, whose framework is by now endorsed for children zero to five years of age, continuing by 57 developing countries, including Indonesia, and with immunization and school health programs, over a hundred partners and nearly 3000 community supplemental feeding and the promotion of healthy service organizations that are members of SUN. street food for school aged children. Finally, A number of countries are prioritizing nutrition additional nutritional services and reproductive as an investment in their growth, and recognizing health counseling, along with iron supplementation nutrition as an investment in economic and social are available for adolescents and elderly persons 1 Nutrition-specific interventions and programs address the immediate determinants of fetal and child nutrition and development- adequate food and nutrient intake, feeding, caregiving and parenting practices, and low burden of infectious diseases. Examples include: adolescent, preconception, and maternal health and nutrition; maternal dietary or micronutrient supplementation; promotion of optimum breastfeeding; complementary feeding and responsive feeding practices and stimulation; dietary supplementation; diversification and micronutrient supplementation or fortification for children; treatment of severe acute malnutrition; disease prevention and management; nutrition in emergencies. Nutrition-sensitive interventions and programs address the underlying determinants of fetal and child nutrition and development-food security; adequate caregiving resources at the maternal, household and community levels; and access to health services and a safe and hygienic environment-and incorporate specific nutrition goals and actions. Nutrition-sensitive programs can serve as delivery platforms for nutrition-specific interventions, potentially increasing their scale, coverage, and effectiveness. Examples include: agriculture and food security; social safety nets; early child development; maternal mental health; women’s empowerment; child protection; schooling; water, sanitation, and hygiene; health and family planning services. Executive Summary viii (Trihono et al. 2015). In parallel, Indonesia has a number of nutrition-sensitive programs in different Design & Methods sectors that are in the process of being scaled-up. For example, in the social protection sector, the This report lays the groundwork for more effective PNPM Generasi (National Program for Community multi-sectoral action on reducing stunting in Empowerment), and the conditional cash transfer Indonesia, by operationalizing the UNICEF conceptual program PKH Prestasi, target the most vulnerable framework.2 The UNICEF framework, first proposed children and women. At the same time the National in 1990 (UNICEF, 1990), was one of the first attempts Health Insurance program (JKN) aims to extend health at emphasizing food security, environment, insurance coverage to the poor and the near poor, health, and child care practices as the four main the self-employed, as well as those employed in the underlying determinants of child malnutrition in informal sector, consequently, allowing for better developing countries. A fundamental premise of this access to healthcare. conceptual framework, is that increases in access to adequate services in one or all of the four drivers of The effectiveness and ultimate success of such multi- malnutrition, say for example, food security alone, sectoral approaches towards reducing stunting cannot substitute for inadequate levels of access to depends on having a more holistic view of the the other determinants. While there is widespread inequities and gaps in access to adequate levels of the acknowledgment of the four key underlying underlying determinants of nutrition: Care, Health, determinants of nutrition there is limited quantitative Environment (or WASH), and Food Security (C H E information on the size and direction of the F). The interdependencies among the underlying interdependence among adequate (or inadequate) access determinants of nutrition are usually beyond the scope to food security, environment, health, and child care in or the control of any given sector. The integration of child nutrition. nutritional considerations in the agricultural sector, for example, is unlikely to take into account the fact Indicators for care, for health, for environment, and that water, sanitation and hygiene (W ASH) services for food security are constructed using the 2007 and and facilities may be poor and inadequate in some 2013 RISKEDAS surveys. Each indicator is comprised communities. As a consequence, the nutritional of various components based on availability in the impacts of such nutrition-sensitive interventions on survey, with the definition of “adequacy” based on key nutrition outcomes could be impeded considerably national and/or accepted international standards. by the absence of adequate W ASH facilities. On In consideration of the complexity of the linkages the other hand, the impacts of the same nutrition- between the underlying determinants of malnutrition sensitive agricultural interventions could be enhanced and the economic situation of the family, the analysis considerably if they were to be accompanied by is also carried out separately for urban and rural simultaneous improvements in the water and households, for resource-rich (top 60 per cent) and sanitation facilities in the same communities. Thus, a resource-poor (bottom 40 per cent) households as well more holistic approach to sector-specific “nutrition- as for districts with high stunting rates and those with sensitive” interventions is likely to be better able to low stunting rates. A more holistic view is provided address the key underlying determinants of nutrition to the extent to which adequate levels of the four key effectively, as well as reinforce the impacts of underlying determinants of nutrition— food security; nutrition-specific interventions. adequate caregiving resources at the maternal, household and community levels; access to health services; and a safe and hygienic environment—on their own as well as in combination are associated with better nutrition as measured by height-for-age z-scores (HAZ) and stunting rates. 2 The same conceptual framework also underpins the IPKM index (Indeks Pembangunan Kesehatan Masyarakat) of the Ministry of Health of the Republic of Indonesia. (IPKM, 2014). The IPKM essentially summarizes all the components of the underlying drivers of nutrition into one index using statistical methods. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting ix Results persist (Figures II through V). For For example, in 2013 the use of health care facilities was in general greater by wealthier households than poorer households. Slightly In spite of the considerable constraints imposed by less than two-thirds of the mothers of children in the data comparability issues across survey years, the poorest households were seen in prenatal visits (64 analysis of the trends in access to the four drivers percent), 63 per cent of the children were delivered by of malnutrition reveals that access to most of the a health care professional or had a post-natal checkup comparable components of care, health, environment (63 percent), whereas among children in the wealthiest and food security has improved between 2007 and households, 90 percent were seen in prenatal visits, 98 2013 (Figure I). percent were delivered by a health care professional and 88 percent had a post-natal checkup (See Figure III). In spite of the improvements over time, substantial Along similar lines, in 2013 children living in rural areas inequalities in access between rural and urban areas, were less likely to have access to adequate environment between districts with high and low stunting rates and than children living in urban areas. The differences were between poorer and wealthier households continue to greater for sanitation measures than for drinking water. FIGURE I. CHANGES IN ACCESS TO THE COMPONENTS OF ADEQUATE, CARE, HEALTH, ENVIRONMENT, AND FOOD SECURITY BETWEEN 2007 AND 2013 Health Care Components of adequacy in health Components of Adequacy in Care National, children less than 60 months National, children less than 60 months Immunizations Percentage of children who meet criteria Percentage of children who meet criteria Vitamin A* Mother's handwashing 76 No smoking 72 70 67 55 62 38 30 25 27 2007 2010 2013 2007 2013 Source: Author estimates based on 2007, 2010 and 2013 RISKESDAS. Source: Author estimates based on 2007 and 2013 RISKESDAS. Note: *Only for children 7 months or older Environment Food Components of Adequacy in Environment - Sanitation Components of Adequacy in Food National, children less than 60 months National, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria Improved Mom fruit/vegetable intake Improved Exclusive breastfeeding 64 53 60 44 15 12 2007 2013 2007 2013 Source: Author estimates based on 2010 and 2013 RISKESDAS. Source: Author estimates based on 2007 and 2013 RISKESDAS. Source: Authors’ calculations based on the 2007 and 2013 RISKESDAS. Executive Summary x In rural areas, 47 percent of children had access to basic premises), with about 11 percent of rural children having sanitation whereas in urban areas 74 percent had such access to safely managed water and 30 percent of urban access (see Figure IV). Also, urban children as a whole children having such access. Finally, in 2007, the year in had better access to basic sanitation with 62 percent of which this information is available, the consumption of children living in communities where at least 75 percent calories and protein per adult equivalent in the poorest of the households had access to basic sanitation, versus households less than in households at the fifth quintile only 27 percent of rural children. Access to improved and of the wealth distribution. Also, only six percent of the basic water was relatively high in both areas, with about mothers in the poorest households consumed both fruits 79 percent of rural children having such access and 94 and vegetables at least five days of the week, whereas 23 percent of urban children having access to improved percent of the mothers in the highest wealth quintile did and basic water. However, far fewer children had access so (Figure V). to safely managed water (effectively piped water to the FIGURE II. DIFFERENCES IN ACCESS TO THE COMPONENTS OF ADEQUATE, CARE BETWEEN RURAL AND URBAN AREAS AND ACROSS WEALTH QUINTILES IN 2013 Care by Rural vs. Urban Care by Wealth Quintile Components of Adequacy in Care, 2013 Components of Adequacy in Care, 2013 By urban/rural, children less than 36 months By wealth quantile, children less than 36 months Percentage of children who meet criteria Percentage of children who meet criteria 72 74 73 66 71 66 62 53 53 54 49 49 5050 45 46 44 45 40 38 39 38 37 38 38 36 33 35 36 32 32 26 24 27 23 Rural Urban 1 2 3 4 5 Early breastfeeding* Appropriate breastfeeding* Complementary** Handwashing Smoke-free Source: Authors’ calculations based on the 2007 and 2013 RISKESDAS. Notes: Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). FIGURE III. DIFFERENCES IN ACCESS TO THE COMPONENTS OF ADEQUATE HEALTH BETWEEN RURAL AND URBAN AREAS AND ACROSS WEALTH QUINTILES IN 2013 Health by Rural vs. Urban Health by Wealth Quintile Components of Adequacy in Health, 2013 Components of Adequacy in Health, 2013 By urban/rural, children less than 36 months By wealth quantile, children less than 36 months 95 98 94 92 Percentage of children who meet criteria Percentage of children who meet criteria 89 87 89 87 90 88 86 84 81 78 79 81 78 79 80 77 77 76 77 77 67 67 68 70 64 63 63 66 56 57 40 Rural Urban 1 2 3 4 5 Prenatals Assisted Growth Immunizations Vitamin A* Source: Authors’ calculations based on the 2007 and 2013 RISKESDAS. Notes: Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report) Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting xi A move away from an analysis of the individual one of the four determinants alone cannot substitute components of the four drivers of malnutrition towards for inadequacies in the other three underlying a more aggregate analysis of differences in access to determinants. Nevertheless, it is important to bear adequate levels in the four drivers of malnutrition, in mind that alternative criteria could be employed. care, health, environment and food security involves For example, adequate access to care or health or a variety of options. In this report, adequate access environment or food security could be defined as a child to care or health or environment or food security is having adequate access to at least one or at least two of defined as a child having adequate access to all of the the components of each of the four drivers of nutrition. components of each of the four drivers of nutrition. Alternatively, different weights could be applied to the Although this is admittedly a very strict criterion, it is distinct components of each nutrition driver, reflecting in line with the UNICEF conceptual framework that the preferences, values and/or information of the policy assumes that increases in access to adequate levels in making authority. FIGURE IV. DIFFERENCES IN ACCESS TO THE COMPONENTS OF ADEQUATE ENVIRONMENT BETWEEN RURAL AND URBAN AREAS AND ACROSS WEALTH QUINTILES IN 2013 Environment by Rural vs. Urban Environment by Wealth Quintile Components of Adequacy in Environment, 2013 Components of Adequacy in Environment By urban/rural, children less than 36 months By wealth quantile, in 2013 Percentage of children who meet criteria 94 94 95 97 96 Percentage of children who meet criteria 94 93 9090 79 78 86 74 81 80 74 62 68 64 62 63 47 46 30 35 27 26 26 19 18 11 13 7 2 4 Rural Urban 1 2 3 4 5 Sanitation Basic Community Drinking water Improved Basic Safely managed Source: Authors’ calculations based on the 2007 and 2013 RISKESDAS. Notes: Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). FIGURE V. DIFFERENCES IN ACCESS TO THE COMPONENTS OF ADEQUATE FOOD SECURITY BETWEEN RURAL AND URBAN AREAS AND ACROSS WEALTH QUINTILES IN 2013 Food Security by Rural vs. Urban Food Security by Wealth Quintile Components of Adequacy in Food Components of Adequacy in Food By Urban/Rural, children less than 60 months, in 2007 By wealth quintile, children less than 60 months, in 2007 Percentage of children who meet criteria Percentage of children who meet criteria 75 67 69 62 62 65 58 46 47 45 40 41 40 42 43 37 36 37 37 39 33 23 19 16 9 12 7 6 Rural Urban 1 2 3 4 5 Household calories Household protein Exclusively breastfed* Mom fruit/vegetable intake Source: Authors’ calculations based on the 2007 RISKESDAS. Executive Summary xii Application of the above mentioned definition for 46 percent.5 In 2013 children living in rural areas were adequate access leads to more stark differences less likely to have access to all of the four drivers of in access to adequate access to the four drivers of nutrition than urban children. The largest discrepancy nutrition. At the national level, access to adequate care was in the access to adequate environment with only was the driver with the lowest prevalence rate and around 40 percent of those in rural areas having access to adequate environment was the driver with access and around 70 percent of those in urban the highest prevalence rate. Nationally only seven areas having access (Figure VI). Adequate access to percent of the children had access to adequate care environment is also the nutrition driver with the (Figure VI)3. Similarly, very few children had access largest differences by wealth quintiles. Only one to adequate food, with only 14 percent having such percent of the children in the lowest wealth quintile access nationally.4 About 56 percent of the children had access to adequate environment whereas in the had access to adequate environment or basic drinking highest quintile 92 percent of the children had access water and improved sanitation simultaneously. The to an adequate environment (Figure VI). access to adequate health was also relatively high, at FIGURE VI. ACCESS TO ADEQUATE, CARE, HEALTH, ENVIRONMENT AND FOOD SECURITY IN 2013 National Adequacy status in 2013, by National Percentage of children who meet criteria 0 to 36 month olds 56 46 Food Care Environment 14 7 Health Rural vs. Urban Wealth Quintile Adequacy status in 2013, by Urban/Rural Adequacy met in 2013, by Wealth Quantile 0 to 36 month olds 0 to 36 month olds 92 Percentage of children who meet criteria Percentage of children who meet criteria 81 71 62 51 53 53 49 41 41 39 26 27 18 21 17 10 8 8 6 11 8 10 6 5 4 7 1 Rural Urban 1 2 3 4 5 Source: Authors’ calculations based on the 2013 RISKESDAS. Notes: Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). 3 Adequate care is defined as mother’s hand washing practices, and whether household is smoke free. For children zero to 23 months of age it also includes immediate breastfeeding after birth, and age appropriate breastfeeding at the time of survey. For children six to eight month of age the additional component of the child receiving complementary feedings is also included 4 Adequate food is based on mother’s fruit and vegetable consumption. For children zero to five months of age it also includes exclusive breast feeding. 5 Access to adequate health is defined by mother having had at least four prenatal checkups, the birth was assisted by a skilled professional, the child was seen at a post-natal checkup, and the child’s immunizations are up to date. For children 7 to 35 months of age the additional condition of having received vitamin A supplementation is added. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting xiii A more austere picture emerges through an analysis of isolated sector-specific nutrition-sensitive initiatives the extent to which children have simultaneous access may be constrained by inadequate access to the to two or more of the four underlying determinants of four underlying determinants of malnutrition: food nutrition. In 2013, 23 per cent of the children between 0 security, child care, environment, and health. and 3 years of age did not have access to adequate level in any of the four determinants of nutrition. In 2007, The preceding argument is reinforced by the result this fraction was 39 per cent. On the other end, less that simultaneous access to two or more of the four key than 1 per cent of children had access simultaneously to determinants of nutrition is associated with a decrease all four key underlying determinants of nutrition (see in the likelihood of a child being stunted (see Figures Figure VII). These differences in simultaneous access to VIII and IX). The likelihood that children between two or more drivers become larger between rural and 0 and 36 months of age are stunted is lower when urban areas and by quantile of wealth). children have simultaneous access to adequate levels to two of the four drivers of nutrition and even lower The low proportions of children with simultaneous among children with simultaneous access to adequate access to more than one of the drivers of nutrition levels to three of the four drivers of nutrition (Figure suggests that the integration of nutritional VIII). More importantly, the same pattern appears to considerations in any given sector may have hold separately in rural and urban areas as well as for nutritional impacts that are limited primarily by children in households at the top 60 percent of the inadequate access to the underlying drivers of wealth distribution (Figure IX). All in all, these results nutrition. The recent emphasis on sector-specific validate the importance of coordinated multi-sectoral nutrition sensitive interventions (World Bank, policies and suggest that the success of “sector-specific 2013) rightly emphasizes the synergies that can be nutrition-sensitive” initiatives can be enhanced by exploited within specific sectors such agriculture, better coordination and integration of multi-sectoral water and sanitation, or social protection. These interventions that address effectively the four findings suggest that the success of uncoordinated or underlying determinants of nutrition. FIGURE VII. SIMULTANEOUS ACCESS TO ADEQUATE LEVELS OF TWO OR MORE OF THE FOUR DRIVES OF MALNUTRITION IN 2013 National Percentage of children who meet criteria Adequacies status in 2013, National Adequate in none 0 to 36 month olds Adequate in 1 of 4 Adequate in 2 of 4 Adequate in 3 of 4 38 Adequate in all 4 30 23 8 1 Rural vs. Urban Wealth Quintile Number of adequacies met in 2013, By urban/rural Number of adequacies met in 2013, By wealth quantile Percentage of children who meet criteria Percentage of children who meet criteria 0 to 36 month olds 0 to 36 month olds 68 42 42 44 39 42 43 37 38 36 34 33 33 29 23 16 18 13 12 14 8 11 5 4 2 0 6 3 0 1 00 1 1 2 Rural Urban 1 2 3 4 5 Source: Authors’ calculations based on the 2013 RISKESDAS. Notes: Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Executive Summary xiv FIGURE VIII. SIMULTANEOUS ACCESS TO DRIVERS OF NUTIRITION AND THE PROBABILITY OF BEING STUNTED: 2013 Difference in Probability of Being Stunted, 2013 National, Relative to the Reference Group -0.052 One Nutrition Drivers -0.089 Two -0.134 Three -0.109 Four -0.200 -0.150 -0.100 -0.050 0.000 Percentage point difference in stunting rate Source: Authors’ calculations based on the 2013 RISKESDAS. Note: Showing point estimates and 95% CI. Reference group: Children with access to none, p=.414 FIGURE IX. SIMULTANEOUS ACCESS TO DRIVERS OF NUTIRITION AND THE PROBABILITY OF BEING STUNTED: 2013 Rural, Relative to the Reference Group Urban, Relative to the Reference Group -0.052 -0.015 One One Nutrition Drivers Nutrition Drivers -0.073 -0.056 Two Two -0.104 -0.096 Three Three -0.070 -0.071 Four Four -0.200 -0.100 0.000 0.100 -0.200 -0.150 -0.100 -0.050 0.000 0.050 Percentage point difference in stunting rate Percentage point difference in stunting rate Children in B40, Relative to the Reference Group Children in T60, Relative to the Reference Group -0.029 -0.012 One One Nutrition Drivers Nutrition Drivers -0.017 -0.041 Two Two -0.047 -0.081 Three Three -0.083 -0.051 Four Four -0.400 -0.200 0.000 0.200 0.400 -0.150 -0.100 -0.050 0.000 0.050 Percentage point difference in stunting rate Percentage point difference in stunting rate Source: Authors’ calculations based on the 2013 RISKESDAS. Note: Showing point estimates and 95% CI. Reference group: Children with access to none, p=.414 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting xv Policy Considerations The operationalization of the UNICEF conceptual Targeting and tailoring: The above also suggests that framework in this report offers the opportunity of “one size fits all” multi-sectoral programs are not likely serving as a basis for a more systematic monitoring of to be as effective as multi-sectoral programs that are the progress in access to the four main drivers of child tailored and targeted to specific age groups, specific malnutrition. It also serves as a practical diagnostic locations and/or low wealth groups. Investments in framework for identifying potential “binding nutrition-specific and nutrition-sensitive interventions constraints” in the Indonesian context towards the should focus on areas where access to adequate effort to reduce child stunting and malnutrition. levels in the four drivers of nutrition is relatively less prevalent. Measurement drives diagnosis and response: The analysis carried out highlights numerous critical Better coordination and integration: Progress towards data gaps in key components of the four underlying reducing stunting in Indonesia can be enhanced by determinants of child malnutrition. For example, coordinated multi-sectoral interventions that address the 2013 RISKESDAS survey, the one and only effectively the four key underlying determinants of survey containing anthropometric measures for nutrition. The poor performance of multi-sectoral children and adults at the national as well as at the projects in nutrition and health across the world district level does not include variables useful for provides the opportunity to learn a lot from the quantifying key components of food security, such as failures of the past, especially when it comes to setting dietary diversity and consumption or availability of the clarity of objectives and the role and responsibility calories and proteins. On the other hand, the annual of each sector involved in the renewed effort to reduce SUSENAS survey which collects these important stunting Indonesia. Clarity and prior agreement on variables but only at the household level, contains no the common yardstick to be used among the different information on child anthropometric measures and sectors involved can lead to significant improvements is relatively weaker in measures of child care. A more in the efficiency and efficacy of policies against child coordinated approach to data collection that closes undernutrition. the gaps in information collected by nutrition-related and other socio-economic surveys in Indonesia can have substantial benefits for efforts aimed at reducing child stunting through a more informed and better coordinated multi-sectoral approach. Context matters: The probability of being stunted with simultaneous access to two or more of the four underlying determinants of nutrition, varies by the wealth status of the household, and rural and urban areas. Executive Summary xvi Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting xvii Contents EXECUTIVE SUMMARY iii Motivation iv Design & Methods v Results xi Policy Considerations Introduction 1. 5. Synergies The Underlying Determinants of Nutrition 2. and Height-for-age in 2013 31 In Search of Synergies 34 Stunting and Fiscal Resource Allocation in Indonesia 6 Stunting, Fiscal Transfers to Districts, and District Expenditures 6. Summary of Findings 43 9 Methodological Framework 3. 7. Policy Considerations 47 Measures of Adequate Access to 4. REFERENCES 49 the Determinants of Nutrition APPENDIXES 51 Based on the RISKESDAS 13 Adequate Food Security 17 Adequate Care 21 Adequate Health 25 Adequate Environmental Services xviii Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 1 Chapter 1. Introduction Stunting is a widespread and persistent condition as at the county level aiming to scale up nutrition- in Indonesia with more than one-third of young sensitive interventions. One prominent example is children being stunted. The national stunting rate the Scaling-Up Nutrition (SUN) movement, whose for under five-year-olds increased slightly from 36.8 framework is endorsed by 30 developing countries. percent in 2007 to 37.2 percent in 2013, based on Indonesia, together with a number of other countries, official stunting rates reported by the Ministry of is prioritizing nutrition as an investment in its Health of the Government of Indonesia. During the growth, and recognizing nutrition as an investment same time period, the percentage of population in in economic and social development to strengthen poverty reduced from 16.6 percent to 11.4 percent its nation. Along parallel lines, initiatives within (World Bank), suggesting that although a smaller the World Bank and other development agencies share of households was below the poverty-line, the and research institutions, aim to foster knowledge increased purchasing power did not translate to better exchange and cross-sectoral collaboration and nutritional outcomes for children. coordination at the project level for improving nutrition (Shekar et al., 2013). All these initiatives By now, there is a wide consensus that economic are based on the premise that the determinants of growth is not sufficient for improving nutrition malnutrition are multi-sectoral and that the solution to outcomes. An acceleration of the progress towards malnutrition requires multi-sectoral approaches. reducing stunting in Indonesia, requires enlisting more sectors, in addition to the health sector, such as Indonesia’s commitments to reducing stunting is agriculture, education, social protection, and water, evidenced by the wide variety of “nutrition-specific” sanitation, and hygiene in the effort to improve and “nutrition-sensitive” interventions. Nutrition- nutrition. Large scale “nutrition sensitive” interventions specific interventions are in place to address in these sectors will have to be able not only to nutritional deficiencies at every point in the life-cycle address the key underlying determinants of nutrition beginning with folate and calcium supplementation, effectively, but also intensify the role of “nutrition- as well as supplemental feeding for malnourished specific” interventions (Lancet 2013).6 pregnant mothers, breastfeeding promotion and counseling for lactating mothers, growth monitoring, In recent years there has been a significant increase in vitamin A supplementation, iodization, supplemental the number of initiatives at the international as well feeding, fortification and therapeutic zinc supplements 6 Nutrition-specific interventions and programs address the immediate determinants of fetal and child nutrition and development— adequate food and nutrient intake, feeding, caregiving and parenting practices, and low burden of infectious diseases. Examples include: adolescent, preconception, and maternal health and nutrition; maternal dietary or micronutrient supplementation; promotion of optimum breastfeeding; complementary feeding and responsive feeding practices and stimulation; dietary supplementation; diversification and micronutrient supplementation or fortification for children; treatment of severe acute malnutrition; disease prevention and management; nutrition in emergencies. Nutrition-sensitive interventions and programs address the underlying determinants of fetal and child nutrition and development— food security; adequate caregiving resources at the maternal, household and community levels; and access to health services and a safe and hygienic environment—and incorporate specific nutrition goals and actions. Nutrition-sensitive programs can serve as delivery platforms for nutrition-specific interventions, potentially increasing their scale, coverage, and effectiveness. Examples include: agriculture and food security; social safety nets; early child development; maternal mental health; women’s empowerment; child protection; schooling; water, sanitation, and hygiene; health and family planning services. Chapter 1. 2 for diarrhea management and deworming prevention The effectiveness and ultimate success of such multi- for children zero to five years of age, continuing sectoral approaches towards reducing stunting with immunization and school health programs, depends on having a more holistic view of the supplemental feeding and the promotion of healthy inequities and gaps in access to adequate levels of the street food for school aged children. Finally, underlying determinants of nutrition: Care, Health, additional nutritional services and reproductive Environment (or WASH), and Food Security (C H E health counseling, along with iron supplementation F). The interdependencies among the underlying are available for adolescents and elderly persons determinants of nutrition are usually beyond the scope (Trihono et al. 2015). In parallel, Indonesia has a or the control of any given sector. The integration of number of nutrition-sensitive programs in different nutritional considerations in the agricultural sector, sectors that are in the process of being scaled-up. for example, is unlikely to take into account the fact For example, in the social protection sector, the that water, sanitation and hygiene (WASH) services PNPM Generasi (National Program for Community and facilities may be poor and inadequate in some Empowerment), and the conditional cash transfer communities. As a consequence, the nutritional program PKH Prestasi, target the most vulnerable impacts of such nutrition-sensitive interventions on children and women. At the same time the National key nutrition outcomes could be impeded considerably Health Insurance program (JKN) aims to extend health by the absence of adequate W ASH facilities. On insurance coverage to the poor and the near poor, the other hand, the impacts of the same nutrition- the self-employed, as well as those employed in the sensitive agricultural interventions could be enhanced informal sector, consequently, allowing for better considerably if they were to be accompanied by access to healthcare. simultaneous improvements in the water and sanitation facilities in the same communities. Thus, a more holistic approach to sector-specific “nutrition- sensitive” interventions is likely to be better able to address the key underlying determinants of nutrition effectively, as well as reinforce the impacts of nutrition-specific interventions. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 3 Chapter 2. Stunting and Fiscal Resources Allocation in Indonesia Stunting is distributed across the archipelago with were able to decrease stunting by more than four most provinces having districts with stunting rates percentage points, whereas a number of districts both above and below the national average--the with stunting rates above the national average saw exception being Yogyakarta with all five districts an increase in stunting. having stunting rates below the national average both in 2007 and in 2013. The maps in Figure 1 below show Furthermore, more children in the poorest households the prevalence of stunting across districts both in were stunted in 2013 than in 2007 rising from 43 2007 and 2013.7 percent in 2007 to 49 percent in 2013.8 Children in wealthier households were five percentage points less Between 2007 and 2013 stunting rates changed rather likely to be stunted in 2013 than in 2007. Therefore, erratically across districts. In the maps (Figure 2), whereas in 2007 the difference in stunting rates the upper panel shows changes in stunting rates for between children under 60 months from the those districts where the stunting rate in 2007 was poorest households and the wealthiest households above the national stunting rate of 36.8 percent and was 10 percentage points, in 2013 it had increased the lower panel shows changes in stunting rates for to 20 percentage points. That is, even though for those districts where the stunting rate in 2007 was children from certain households the likelihood of below the national stunting rate of 36.8 percent. being stunted decreased, children from the most In both panels, districts in green saw a more than vulnerable households, those with fewer resources, four percentage point decrease in stunting between the likelihood increased. 2007 and 2013 and those in red saw a more than four percentage point increase in stunting. Many of Between 2007 and 2013, the stunting rate decreased the districts with stunting rates above the national for both older children and younger children. The average in 2007 had lower stunting rates in 2013, prevalence of stunting decreased by six percentage and many of the districts with stunting rates below points in the younger age cohort and four percentage the national average in 2007 had higher a prevalence points in the older age cohort (Figure 3). The of stunting in 2013. Only a few districts that had prevalence of stunting is higher in older children than stunting rates lower than the national average in younger ones in each year. 7 The 2013 stunting rates are projected onto 2007 district boundaries. That is, if a district split between 2007 and 2013, we take the weighted average of the stunting rates in all the areas covered by the 2007 “mother” district. 8 These analyses,and all that follow,are based on theofficial HAZ scores calculated by theIndonesianMinistry of Health. Chapter 2. 4 FIGURE 1: PREVALENCE OF STUNTING BY DISTRICT, 2007 AND 2013 STUNTING RATE IN INDONESIA, 2007 16.7 – 36.8 36.8 – 67.4 STUNTING RATE IN INDONESIA, 2013 11.1 – 37.2 37.2 – 70.4 Source: Authors’ calculations based on the 2013 RISKESDAS Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 5 FIGURE 2: CHANGES IN DISTRICT STUNTING RATES, 2007 TO 2013 DISTRICT WITH STUNTING RATES ABOVE NATIONAL AVERAGE (36.8%) IN 2007 Stunting decreased (41.6 to 4.0 percentage points) Stunting remained constant (less than 4.0 percentage point change) Stunting increased (4.0 to 34.5 percentage points) Stunting in 2007 below 36.8% DISTRICT WITH STUNTING RATES BELOW NATIONAL AVERAGE (36.8%) IN 2007 Stunting decreased (41.6 to 4.0 percentage points) Stunting remained constant (less than 4.0 percentage point change) Stunting increased (4.0 to 34.5 percentage points) Stunting in 2007 above 36.8% Chapter 2. 6 FIGURE 3. STUNTING RATES IN INDONESIA 2007 TO 2013 Stunting in children 0 to 59 months Stunting in children 0 to 59 months By wealth quintile By Age Percentage of children with HAZ < -2 SD Percentage of children with HAZ < -2 SD 48 49 43 42 44 43 40 39 39 39 44 41 40 37 35 38 33 33 33 28 29 1 2 3 4 5 0-23 months 24-59 months 2007 2010 2013 Source: Author estimates based on 2007, 2010 and 2013 RISKESDAS. Stunting, Fiscal Transfers The general block grants (DAU) provided by the central government are the largest source of revenue to Districts, and District for most districts.9 The allocation of these grants has been based on a formula that aims to address Expenditures disparities between local expenditure needs and local fiscal potential (Agustina et al., 2012; Hofman et al. 2006). The first question we address is whether there Throughout the 1990s, economic progress and is a “needs-based” allocation of DAU block grants or demand for greater political autonomy across transfers to district governments (denoted by Td). It Indonesia saw growing pressures for greater is important to bear in mind that the DAU transfers democratization and decentralization. In June 1999, from the central government to the districts do not Indonesia’s first relatively free and fair elections in include the “deconcentrated” spending by the central 44 years were held, sweeping in a new batch of more government on poverty alleviation programs such as assertive civil servants into local legislatures (DPRDs). the household-level conditional cash transfer program In August 1999, two ground breaking decentralization (PKH) or the incentivized community-level block grant laws were passed. These laws, in effect, transferred program (PNPM Generasi).10 the bulk of basic service delivery to more than 300 district governments (as opposed to provinces and Recognizing that district “needs” can be defined in governors). They also folded the deconcentrated many different ways our analysis defines “need” structures into these district level government according to the stunting rate prevailing in the district. structures and provided them with a significant block grant as well as natural resource revenue sharing (World Bank, 2003; Skoufias et al., 2011). 9 In 2004, for example, DAU accounted for an average of 64 per cent of total revenues.. 10 PKH stands for Program Keluarga Harapan or Hopeful Family Program and PNPM Generasi stands for Program Nasional Pemberdayaan Masyarakat Generasi Sehat dan Cerdas or the National Program for Community Empowerment. Spending at the district level on PKH and PNPM Generasiis “deconcentrated” meaning that district level officials have responsibility for the implementation of these programs following the selection criteria and implementation guidelines provided by authorities in the central government. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 7 For example, we estimate a regression such as To complete the picture, the panels (c) and (d) display the correlations between stunting rates and district level expenditures per capita on health. In 2007 there is a significantly positive correlation12 between total per capita expenditures in health by districts and stunting rates at the district level. Thus, in 2007 districts with higher stunting rates spent more per capita on health-related matters. However, in 2013 this where Sd is the stunting rate in district d and Td is the positive correlation appears to be much weaker and not per capita DAU transfer received by (or expenditure statistically significant meaning that it is not possible to in the health sector) in district d. In this specification, reject the null hypothesis of zero correlation between the coefficient γ provides an estimate of the partial district level spending on health and stunting rates. correlation between fiscal transfers to the district and the stunting rate in the district. If central government In combination, the analysis carried so far reveals that: transfers are allocated based on need, one would • The stunting rate among children in the poorest expect a significantly positive correlation between households in 2013 increased to 49 percent (from the level of per capita transfers and needs, i.e. a higher 43 percent in 2007). allocation of fiscal transfers to districts with a higher • The design of the fiscal transfer system is stunting rate. In terms of the specification in equation targeting more financial resources (DAU per (X) above, a “needs-based” allocation of fiscal transfers capita) to districts with higher stunting rates implies that γ > 0. in both 2007 and in 2013. There is an alarming change in the spending patterns of districts on Panels (a) and (b) in Figure 4 below report the health. regression line obtained by regressing the district level • The fiscal expenditures per capita of districts stunting rates in 2007 an in 2013 against the amount on the health sector which includes funding of per capita DAU transfers (general purpose grant) for direct nutritional interventions were from the central government to the district in 2007 higher in districts with higher stunting rates and 2013. DAU is intended to bring equality in fiscal in 2007. By 2013, the positive relation between capacity among districts. Therefore, DAU allocation is district spending on health and stunting rates is a function of fiscal gap, defined as fiscal needs minus considerably weaker. fiscal capacity. In this regard, districts with high fiscal gap—fiscal needs exceeding the capacity—are likely to be the high-stunt districts. The regression estimates in Figure 4 reveal that there is a significantly positive correlation11 between total per capita transfers and district needs as defined by the higher stunting rate at the district level. Districts with higher stunting rates receive slightly higher transfers from the central government. In fact, the hypothesis that the slope of the regression line is equal to zero is strongly rejected (p-value 0.01). This is not so surprising since the formula used to allocate grants to the districts (DAU) takes into account the district’s poverty level as well as other human development indicators. 11 The p-value of the coefficient γ is less than 0.01 12 The p-value of the coefficient γ is less than 0.01 Chapter 2. 8 FIGURE 4. DISTRICT FINANCES AND DISTRICT STUNTING RATES (a): District-level DAU Transfer (per capita) and District Stunting (b): District-level DAU Transfer (per capita) and District Stunting in 2007 in 2013 stunt_data = -10.608 + 3.3358 lrpc_DAU R2 = 7.4% stunting = .38347 + 2.7302 lrpc_DAU R2 = 6.3% 80 80 Stunting Rate from Riskesdas Data 60 60 stunting 40 40 20 20 0 12 13 14 15 16 17 8 10 12 14 16 18 Log of Per capita DAU Log of Per capita DAU n = 432 RMSE = 8.7828774 n = 463 RMSE = 8.849954 (c): District-level Expenditure (per capita) on Health and District (d): District-level Expenditure (per capita) on Health and District Stunting Rates in 2007 Stunting Rates in 2013 stunt_data = 4.4239 + 2.6362 lrpc_hea~h R2 = 5.0% stunting = 32.602 + .51988 lrpc_hea~h R2 = 0.2% 80 60 Stunting Rate from Riskesdas Data 60 50 stunting 40 40 30 20 20 0 9 10 11 12 13 14 8 10 12 14 16 Log of Per capita Health Function Expenditure Log of Per capita Health Function Expenditure n = 295 RMSE = 8.7366433 n = 414 RMSE = 9.0539997 Source: District-level Stunting Rates for 2007 and 2013 published by NIHRD. District level DAU transfers and health expenditures in 2007 and 2013 obtained from the DAPOER data base The World Bank Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 9 Chapter 3. Methodological Framework The conceptual framework summarized in Figure 5 clusters: (a) inadequate household food security, (b) views malnutrition as the consequence of a variety inadequate care and feeding practices, (c) unhealthy of interlinked and interrelated events. The causes household environment, and (d) inadequate health of malnutrition are classified into three hierarchical services. The basic causes of malnutrition summarize categories: the immediate causes, the underlying the social, cultural, economic and political context causes, and the basic causes of malnutrition. In any and the prevailing inequalities in the distribution given context identification of the immediate causes of resources in the society. In combination these of malnutrition (disease or inadequate dietary intake) contextual or structural factors play a fundamental is useful for guiding policy actions especially in role in the extent to which there are inequalities situations of crises. However, disease and inadequate among households and their members in having dietary intake are typically consequences of a adequate food security, care and feeding practices, variety of underlying drivers that are interrelated. healthy environment and adequate health services For conceptual simplicity the underlying causes of (i.e., the underlying causes of malnutrition). malnutrition are themselves grouped into the four Intergenerational consequences FIGURE 5. DETERMINANTS OF CHILD NUTRITION Long-term consequences: Adult height, cognitive ability, economic Short-term consequences: productivity, reproductive performance, Mortality, morbidity, disability overweight and obesity, metabolic and cardiovascular diseases MATERNAL AND CHILD UNDERNUTRITION IMMEDIATE Inadequate dietary intake Diseases causes UNDERLYING Household food Inadequate care and Unhealthy household Inadequate health causes insecurity feeding practices environment services Household access to adequate quantity and quality of resources: Land, education, employment, income, technology BASIC Inadequate financial, human, physical and social capital causes Source: An adaptation of the UNICEF (1990) Social cultural, economic and political context “Strategy for Improved Nutrition of Children and Women in Developing Countries” Chapter 3. 10 Since its conception this conceptual framework has stress level, and self-confidence, (4) autonomy and been revised and extended in various dimensions. control of resources, (5) workload and time constraints, Various international organizations have adopted as (6) social support received from family and community. well as adapted this framework. For example, FAO (2011) discusses adaptation of this framework for The third underlying driver of nutrition is access FAO’s nutrition analysis. USAID - FANTA (Food and to a healthy environment. The driver measures Nutrition Technical Assistance) also adapted this the child’s exposure to pathogens in the physical framework (Riely et al., 1999). World Food Program environment where they live. The measure is based (WFP) refers to it as the Food and Nutrition Security on the definitions adopted by WHO / UNICEF Joint Conceptual Framework in its Emergency Food Monitoring Program (JMP) and as part of monitoring Security Assessment Handbook (WFP, 2009, pg. the Sustainable Development Goals and include 25). However, regardless of the adaptations and the components on (1) access to improved drinking water extensions to the original framework, the fundamental (2) access to improved sanitation, (3) adequate hand ideas regarding the critical interactions, interrelations washing practices, (4) adequate disposal of child’s and synergies among food security, environment and feces. Given that it is not only the child’s immediate health, and care have remained at the core. This is environment, i.e. the facilities at the dwelling unit, also very transparent in the framework for actions but also those in the immediate neighborhood which to achieve optimum fetal and child nutrition and affect the degree of exposure to pathogens, especially development extracted from the 2013 Lancet Maternal community-wide access to improved sanitation is a and Child Nutrition Series (see Figure 6). fundamental component. The first underlying driver of nutrition is access to The fourth underlying driver of nutrition is access adequate food security. A child is food secure when to adequate healthcare. This driver measures the at all times, they have physical, social and economic child’s access to skilled medical care to minimize the access to sufficient, safe and nutritious food that meets effects of illness and preventively address health their dietary needs and food preferences for an active issues, especially those linked with malnutrition, such and healthy life (FAO). The ideal measure encompasses as diarrheal diseases. The measure encompasses the three broad factors. First, a comprehensive measure availability and use of healthcare services for pre- takes into account the availability of food. In general, natal, birth and post-natal care. this component measures the supply of food at the national (or regional) level and is based on agricultural Although the framework is a holistic way of production and food trade balance relative to the conceptualizing nutrition it is also important to country’s size. Second, the measure captures the acknowledge some of its limitations. Prices, knowledge, household specific and individual specific access to the education, and household income all influence available food. That is, given household’s income and components of the four clusters of the framework, the prices of food at local markets, what range of food resulting in some overlap in the measures. The choices does the household have available for them. methodology is informative in finding the overall And within the household, how does the food get relationships, from which more focused and detailed distributed. Third, the measure captures the quality analyses can be carried out to determine more of the actual food choices made by the household. concretely the underlying causes. So for example, That is, it measures whether or not the diet and more detailed information would be needed to cooking methods provide all the necessary micro and determine whether food inadequacies were due to the macronutrients needed for healthy growth. cost of food relative to income, to lack of information on the importance of diversified diet, or due to some The second underlying driver of nutrition is access to other factor. The models estimated in this report adequate care. This driver measures the ability of the are not reduced form models (taking into account primary caregiver to provide a safe and appropriate budget constraints etc.) as done in Barrera (1990), but environment for the child to grow and develop. Ideally rather correlations between nutritional outcomes the measure is based on the child’s caregivers’ (1) as measured by height-for-age Z-scores and having knowledge, practices and beliefs regarding childcare, adequate access to the four drivers of nutrition. (2) health and nutritional status, (3) mental health, Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 11 FIGURE 6. FRAMEWORK FOR ACTIONS TO ACHIEVE OPTIMUM FETAL AND CHILD NUTRITION AND DEVELOPMENT Benefits during the life course Morbidity and Cognitive, motor, School performance Adult stature Work capacity mortality in childhood socioemotional development and learning capacity and productivity Obesity and NCDs Nutrition sensitive Optimum fetal and child nutrition and development programmers and approaches Nutrition specific interventions • Agriculture and food security and programmers • Social safety nets • Adolescent health and Breastfeeding, Feeding and caregiving • Early child development preconception nutrition Low burden of nutrient-rich foods, practices, parenting, • Maternal mental health infectious diseases • Maternal dietary and eating routine stimulation • Women’s empowerment supplementation • Child protection • Micronutrient supplementation or • Classroom education fortification Food security, Feeding and caregiving Access to and use • Water and sanitation • Breastfeeding and including availability, resources (maternal, of health services, • Health and family planning services complementary feeding economic access and household and a safe and hygienic • Dietary supplementation use of food community levels environment for children Building and enabling • Dietary diversification environment • Feeding behaviours and • Rigorous evaluations stimulation Knowledge and evidence Politics and governance • Advocacy strategies • Treatment of severe Leadership, capacity and financial resources • Horizontal and vertical coordination acute malnutrition Social, economic, political and environemntal context (national and global) • Disease prevention and • Accountability, incentives regulation, management legislation • Nutrition interventions • Leadership programmes in emergencies • Capacity investments • Domestics resource mobilisation Source: the Executive Summary of “The Lancet Maternal and Child Nutrition Series 2013.” Chapter 3. 12 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 13 Chapter 4. Measures of Adequate Access to the Determinants of Nutirition Based on the RISKESDAS We use data from the RISKESDAS 2007 and 2013 25 percent are categorized as low-stunt districts surveys collected by the Ministry of Health (NIHRD) and those in the top 25 percent are categorized as and which are representative at the district level. high-stunt districts. The list of high- and low-stunt Not all the information collected in each round was districts for 2007 and 2013 are given in Appendix A. similar and thus the adequacy measures developed also differ from 2007 to 2013 making it impossible to directly compare and contrast the measures Adequate Food Security themselves. However, if a particular component of an adequacy cluster is available in both surveys, the The Food and Agricultural Organization (FAO) defines evolution of the components between 2007 and 2013 food security as “a situation that exists when all people, is analyzed. Furthermore, we also utilize data from at all times, have physical, social and economic access the RISKESDAS 2010 for trends in the components to sufficient, safe and nutritious food that meets their common to 2010 and 2007 or 2013. However, the dietary needs and food preferences for an active and 2010 survey is only representative at the province healthy life”. This definition is a significant departure level and is not analyzed further. Following are from previous conceptualizations of food security descriptions of the components for each of the four which focused inordinately on the availability of food nutrition drivers developed for each year followed at the national or local level. But, in being broad and by a detailed discussion of the prevalence rates of all encompassing, this definition is also a difficult one the components. The prevalence rates are discussed to operationalize, as it emphasizes the importance at the national level as well as for urban and rural of access and utilization of food just as much as households, households in the bottom 20% of the availability (Barrett, 2009). wealth distribution (B20) and in the top 20% of the wealth distribution (T20), and high-stunt and low- What constitutes availability, access and utilization – stunt districts, as well as for two age cohorts zero to the three dimensions of the current thinking on food 23-month olds and 24- to 59-month olds. The wealth security? Availability is associated with the supply index used is that given by the Ministry of Health side of food, measured most often by the extent of in each of the datasets. Similarly, the height-for-age agricultural production and food trade balance relative Z-scores used are those calculated by the NIHRD. to the size of consumption for any given country. High- and low-stunt districts are assigned based on Access, on the other hand, brings in the demand the distribution of stunting rates across districts and element to the equation: conditional on what is based on the official NIHRD district stunting rates. available in the local market and the price at which it Those districts with stunting rates in the bottom is available, what is the range of food choices that are Chapter 4. 14 open to households given their incomes? Conceptually, average protein consumption is at least 80 percent of it is this dimension of food security that has the the 55-gram threshold (or 44 grams) the household strongest resonance with poverty and vulnerability is classified as having access to adequate protein not only because of its direct relationship with income, consumption. As the third measure, we consider but also because of its links to broader issues of social exclusive breastfeeding for children who are less than and political enfranchisement. Food security of six months of age. For this age group it is possible to individual household members, for example hinges determine a child specific component, given the WHO on their social standing within the household almost recommendation for exclusive breastfeeding until six as much as it does on the household’s overall ability months of age. The fourth component is constructed to procure enough food (vulnerable groups within so as to be able to compare across the two years. the household may include children, daughters, For this measure we use information on the child’s daughter-in-law, or the elderly). Finally, the utilization mother’s consumption of fruits and vegetables as a dimension brings to bear the quality dimension of the proxy for the dietary diversity of the household and accessed food. Do household’s make good use of the the child. Specifically we assess whether or not the food they are able to access? Are diets diverse enough child’s mother has consumed fruits and vegetables five to provide all the micro and macronutrients necessary out of seven days of the recall period. This variable is for healthy physiological and cognitive growth? Are not used in the construction of the adequacy measure cooking methods sanitary and healthy enough to for 2007, but presented when comparing the dietary preserve the nutritional attributes of the eaten food? conditions in 2007 and 2013. Of the three dimensions of adequate food security The 2013 survey lacks information on detailed food (availability, access, and utilization) we only have consumption by the household or the child but does variables that allow us to explore the utilization of have information on breastfeeding practices. The food either at the household or at the individual level. 2013 survey provides far less information on the There are various measures adopted by international food consumption than the 2007 survey. The only organizations that measure the quality of food child specific variable available is breastfeeding. We consumed. For children younger than 24 months construct a variable for the children younger than the most commonly used measure is the minimum six months of age to indicate whether the young acceptable diet based on a child’s dietary diversity infant was exclusively breastfed. Furthermore, for score and number of feedings (WHO, 2008). The all children we use information on their mother’s measure is based on a 24-hour recall period of all consumption of fruits and vegetables during the seven food items consumed which are then categorized day recall period to construct a proxy measure of into seven different food groups. However, such dietary diversity. The diet is considered diverse if the detailed information was not provided in the mother consumed fruits at least five of the seven days Riskesdas surveys and alternative measures of food and vegetables at least five of the seven days. consumption are used here. Access to adequate food improved from 2007 to 2013. The 2007 RISKESDAS provides information on It is possible to compare exclusive breastfeeding for average caloric and protein intake of the household children younger than six months of age and the but not detailed information on child specific food mother’s consumption of fruits and vegetables (Figure consumption, besides breastfeeding. The first 6). Nationally from 2007 to 2013, mothers’ fruit and component considered is the average caloric intake per vegetable consumption at least five days of the week person in the household. The total caloric consumption increased from 12 percent to 15 percent. In both years, in the household is divided by the number of adult the mother’s fruit and vegetable consumption was equivalents living in the household. In order for the higher in the low-stunt districts (the 25% of districts household to be considered consuming sufficient in that particular year with the lowest stunting rates) quantity of calories, the average has to be at least 84 than in the high-stunt districts (the 25% of districts in percent of the 2,100 kcal threshold, or 1,760 kcal. The that particular year with the highest stunting rates). In second component assesses the household’s protein the low- and high-stunt districts the measure improved consumption. Again, total protein consumption from 2007 to 2013, by two and three percentage points, is divided by adult equivalents and as long as the respectively. Urban mothers consumed more fruits and Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 15 vegetables than rural mothers in both years. Between five percent of the mothers in the poorest households 2007 and 2013 fruit and vegetable consumption consumed both fruits and vegetables at least five days increased three percentage points in the rural areas and of the week, whereas 29 percent of the mothers in one percentage point in the urban areas. However, with the highest wealth quintile did so (Figure 6). In fact prevalence rates of twenty percent of lower, the overall between 2007 and 2013 there was a slight decrease dietary diversity remains low. in the number of mothers consuming fruits and vegetables in the three lowest wealth quintiles. As Mother’s vegetable and fruit consumption is highly expected, differences in the mother’s fruit and vegetable correlated with the household’s wealth. In 2013 only consumption by the age of the child are minimal. FIGURE 7. COMPARABLE COMPONENTS OF ADEQUATE FOOD, 2007 AND 2013 Components of Adequacy in Food National, children less than 60 months Percentage of children who meet criteria 64 44 12 15 Components of Adequacy in Food Components of Adequacy in Food By Urban/Rural, children less than 60 months By district stunting, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 67 69 60 58 50 46 46 40 19 20 19 16 7 10 8 10 Rural Urban High stunt rate Low stunt rate Components of Adequacy in Food Components of Adequacy in Food By wealth quintile, children less than 60 months By Age, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 72 67 65 64 59 60 47 45 44 41 42 39 29 23 16 18 15 15 9 8 12 11 12 11 6 5 1 2 3 4 5 0-23 months 24-59 months 2007 Mom fruit/vegetable intake Exclusive breastfeeding 2013 Mom fruit/vegetable intake Exclusive breastfeeding Source: Authors’ calculations based on 2007 and 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Chapter 4. 16 Exclusive breastfeeding of children under six months breastfed than rural children (67 percent). In low- of age improved from 2007 to 2013. Besides slight stunt districts exclusive breastfeeding increased by improvements in mother’s fruit and vegetable twelve percentage points—an increase which was consumption, there was an increase of 20 percentage below the average national increase. The largest points in exclusive breastfeeding (Figure 7).13 Children increase was in households in the lowest wealth from both urban and rural areas saw a similar 20 quintile where exclusive breastfeeding of children percentage point jump, such that in 2013 urban less than six month of age increased by 25 percentage children (60 percent) were still less likely to be points to 72 percent. The smallest increase was in the FIGURE 8 COMPONENT OF ADEQUATE FOOD, 2007 Components of Adequacy in Food National, children less than 60 months, in 2007 Percentage of children who meet criteria 64 44 37 12 Components of Adequacy in Food Components of Adequacy in Food By Urban/Rural, children less than 60 months, in 2007 By district stunting, children less than 60 months, in 2007 Percentage of children who meet criteria Percentage of children who meet criteria 67 66 62 62 50 46 46 37 40 38 38 36 19 16 7 8 Rural Urban High stunt rate Low stunt rate Components of Adequacy in Food Components of Adequacy in Food By wealth quintile, children less than 60 months, in 2007 By Age, children less than 60 months, in 2007 Percentage of children who meet criteria Percentage of children who meet criteria 75 69 62 65 65 63 58 47 45 44 41 40 42 43 37 37 39 37 37 33 23 16 9 12 12 11 6 1 2 3 4 5 0-23 months 24-59 months Household calories Household protein Exclusively breastfed* Mom fruit/vegetable intake Source: Authors’ calculations based on 2007 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). 13 The rate considers only children less than six months of age. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 17 fourth wealth quintile where exclusive breastfeeding last 24 months and the first six months should be increased by 17 percentage point. The fourth wealth exclusive. That is, in the first six months the child quintile also had the lowest exclusive breastfeeding should not receive any other food besides mother’s rate in 2013 with 59 percent of children less than six breastmilk. Furthermore the WHO (2008) recommends months of age being breastfed. complementary feedings of solid or semi-solid foods for children six to eight months of age. This information In 2007, about 37 percent of children lived in a is available for the 2013 survey. From the 2010 household where per capita food consumption was RISKESDAS we are also able to construct measures of adequate and 64 percent in household were protein age-appropriate breast feeding and these are presented consumption was adequate. That is, only about a third for the national level to better gage the trends. of household had access to sufficient calories and two- thirds to sufficient amounts of protein. There were no Another component explored is the early initiation of marked differences by location of household or by the breastfeeding. That is, whether or not the child was age of the child for caloric intake but urban households put on the mother’s breast within one hour of birth. and households in low-stunt districts are slightly Namely, the early initiation of breast feeding ensures more likely to consume sufficient quantities of protein skin-to-skin contact which keeps the child warm than rural households or households in high-stunt (Moore, Anderson and Bergman, 2007). Furthermore, districts (Figure 8). However, there were differences children who are put on the breast early are more in both caloric and protein consumption by wealth. likely to be successfully breastfed (Bramson et al., A household in the bottom wealth quintile was ten 2010, Moore et al., 2007). Edmond et al. (2006) find percentage points less likely to have access to adequate that initiating breastfeeding within one hour of birth is calories and 17 percentage points less likely to have associated with better neonatal survival odds in Ghana, access to adequate amount of protein than a household and Mullaney et al. (2008) find similar results for in the top wealth quintile. However, less than half of Nepal. Both studies also found smaller, but statistically the households in the top wealth quintile consumed a significant, benefits from initiating breastfeeding within sufficient amount of calories. Of all the components of the first 24-hours versus a later initiation time. This adequate food considered access to adequate quantity information is only available for 2010 and 2013. of protein was most likely met among the Indonesian households. Least likely to be met was frequent The hygienic behavior of the mother can be measured consumption of vegetables and fruits, our proxy across the 2007 and 2013 RISKESDAS surveys. The measure for dietary diversity.  measure is based on the mother’s self-reported hand washing actions for three different events: (1) before eating, (2) before preparing a meal, (3) after defecating Adequate Care or after anal cleansing of a baby. For the mother to be considered knowledgeable she must wash her hands For adequate care we are able to construct measures with soap after each of these three events. In addition of the mother’s (and household’s) knowledge, practices for 2007 specific analyses knowledge of proper hand and beliefs regarding child care. Specifically, for both washing techniques also includes mother’s actions 2007 and 2013 we are able to construct a measure of after holding and touching animals. That is child is breastfeeding practices, mother’s stated knowledge considered to receive adequate care if the mother of proper hand washing practices, and smoking washes hands after all four events. In turn for 2013, behaviors by the household head. In 2007 it is possible the hygienic component includes the mother’s hand to determine breastfeeding behaviors for all children washing behavior: (1) whenever her hands are dirty, younger than 5 years of age. In 2013 the breastfeeding (2) after defecation, (3) after using pesticides or information is only collected for children less than insecticides and (4) before milk-feeding the baby as 24 months of age. The other two components are well as the three before mentioned events. available for all children under five years of age. The smoking habits of the household are the last The first behavior considered is the mother behavior considered. In comparing across 2007 and 2013 breastfeeding practices until the child reaches 24 surveys we determine the smoking status of the head of months of age. Based on the recommendations of the household. The head is considered a smoker if they the WHO (2008), a child should be breastfed for at currently smoke daily or sometimes. For 2013 we can Chapter 4. 18 ascertain if anyone in the household smokes and we also stunt districts than in low-stunt districts. In terms consider this measure in the year specific analyses. of wealth, consistently across the years the bottom wealth quintile has the highest breastfeeding rate and Breastfeeding behaviors have improved from 2007 the highest wealth quintile the lowest. However, in to 2013 with the majority of the improvement 2007 the difference between the two quintiles was occurring before 2010. Nationally, age-appropriate only eight percentage points but it had grown to 12 breastfeeding—that is children younger than six percentage points by 2013. months being exclusively breastfed and those six months to 23 months breastfed—went from 37 percent The early initiation of breastfeeding has also increased. to 69 percent from 2007 to 2013 (Figure 9). However, In 2010, twenty-nine percent of children were put on the the breastfeeding rate was even slightly higher in breast within the first hour of birth, whereas in 2013 the 2010 at 74 percent, suggesting a potential downward number had risen to 37 percent (Figure 9). The rates are trend in this behavior. Breastfeeding rates were higher similar in rural and urban settings and in high and low- in rural areas in both 2007 and 2013. However, in stunt districts. In 2010 fewer children from wealthier 2007 breastfeeding was one percentage point higher households were put on the breast immediately than in low-stunt districts than in high-stunt districts, but children from poorer households. However in 2013 the by 2013 breastfeeding had increased by 35 percentage reverse was true, such that 38 percent of children from points in the high-stunt districts and only by 28 the top wealth quintiles were put on breast whereas 35 percentage points in low-stunt districts such that percent of children from the bottom quintile received in 2013 breastfeeding was more prevalent in high- breastmilk within the first hour of birth. All in all, the FIGURE 9. COMPARISON OF FEEDING PRACTICES FOR ADEQUATE CARE, 2007, 2010 AND 2013 Components of Adequacy in Care Components of Adequacy in Care National, children less than 24 months By Urban/Rural, children less than 24 months Percentage of children who meet criteria Percentage of children who meet criteria 74 72 69 66 51 53 49 37 37 36 38 29 32 23 Components of Adequacy in Care Components of Adequacy in Care By district stunting, children less than 24 months By wealth quintile, children less than 24 months Percentage of children who meet criteria Percentage of children who meet criteria 71 80 79 65 74 73 73 71 6966 54 5962 48 53 54 46 49 50 36 36 36 38 37 38 30 30 32 35 31 30 29 28 29 28 26 26 23 High stunt rate Low stunt rate 1 2 3 4 5 2007 Breastfed* Source: Author estimates based on 2007, 2010 and 2013 RISKESDAS. 2010 Early breastfeeding Breastfed* Notes: *Appropriate fed: 0-5 month olds exclusive; **For 6-8 month olds. 2013 Early breastfeeding Breastfed* Complimentary** Source: Authors’ calculations based on 2010 and 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 19 rate of early initiation of breastfeeding remains low and For 2013 the adequate care component most lacking was the improvements have been much smaller than for age- living in a non-smoking household and most likely to appropriate breastfeeding even though the action itself is be met was age-appropriate breast feeding. Nationally of much shorter duration. 29 percent of children younger than 36 months of age lived in a household where no one smoked, 37 percent In 2013, about half of all children aged six to eight were breastfed within an hour of birth, 43 percent had months received complementary solid or semi-solid mothers who washed their hands after the six events foods. The rate was slightly higher among urban considered,14 and 69 percent are age-appropriately children and children from low-stunt districts (Figure breast fed (Figure 11). Of the six- to eight-month old 9). The highest rate of complementary feedings for six children 51 percent received complementary feedings. to eight month olds were in third and fourth wealth Immediate breastfeeding was similar across rural and quintiles. Reliable complimentary feedings are only urban households. Rural mothers were more likely available for the 2013 sample so it is not possible to to appropriately breastfeed than urban mothers, but determine how rate has changed through time. urban mothers were more likely to give complementary feedings to children aged six to eight months of age and There were significant improvements in mother’s to report proper hand washing practices. In high-stunt hygienic behavior between 2007 and 2013. Nationally, districts 71 percent of children zero to 24 months were mother’s reported hand washing in the three common breastfed whereas in low-stunt districts 65 percent events increased from 38 percent to 67 percent (Figure were breastfed. In low-stunt districts 54 percent of six- 10). In both years, mothers in high-stunt districts were to eight-month-olds received complementary feedings less likely to wash their hands than mothers in low- and 45 percent had mothers with adequate hygienic stunt districts. In both types of districts there was about behaviors. For the high-stunt districts the percentages 25 percent improvement in mothers’ hand washing were lower at 48 percent and 36 percent, respectively. behaviors between the two years. Furthermore, rural By wealth quintile, children in poorer households were mothers were less likely to report handwashing than more likely to be breastfed, and less likely to receive urban mothers but in both areas the reported hand complementary feedings or have mothers who washed washing practices improved 29 and 25 percentage their hands than children in the wealthier households. points, respectively. Mothers living in households in That is, in the more vulnerable populations—children in the top wealth quintile were more likely to wash their high-stunt districts or children in poorer households— hands than mothers living in households in the lowest access to complementary feedings, handwashing and a wealth quintile in both 2007 and 2013 (Figure 10). In smoke-free household were less likely than for children 2007 the gap was 13 percentage points and by 2013 the in low-stunt districts or from wealthier households. gap had grown to 27 percentage points. In 2013 mothers of younger children were slightly more likely to report washing hands with soap in the three comparison events than mothers of older children. Adequate Health Between 2007 and 2013, smoking by the head of the The adequate health measure is composed of measures household reduced by three percentage points. The of capturing access and use of health services during national smoking rate in 2007 was 73 percent and in the pre-natal, birth and post-natal periods. The WHO 2013 it was 70 percent (Figure 10). Household heads (2007) recommends that the mother be seen in at in rural areas or in a high-stunt rate districts were least four pre-natal visits prior to giving birth. For more likely to smoke than those in urban areas or the 2007 RISKESDAS we can only ascertain whether in low-stunt districts. The rate of smoking decreases or not the mother went to at least one pre-natal visit with wealth quintiles and for most subpopulations but for 2013 we are able to construct a measure based considered, smoking rates decreased between 2007 and on the actual number of visits. For 2013 it is possible 2013. However, there was a slight increase in smoking to determine if the child’s delivery was assisted by a in the bottom two wealth quintiles. health professional. Those considered as skilled health 14 For the 2013 analyses the hand washing behaviors are expanded to include hand washing prior to:(1) whenever her hands are dirty, (2) after using pesticides/insecticides, and (3) before milk-feeding the baby Chapter 4. 20 FIGURE 10. COMPARISON OF HOUSEHOLD CARE PRACTICES, 2007 AND 2013 Components of Adequacy in Care National, children less than 60 months Percentage of children who meet criteria 67 38 27 30 Components of Adequacy in Care Components of Adequacy in Care By Urban/Rural, children less than 60 months By district stunting, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 73 73 62 57 48 46 33 31 33 31 34 27 26 27 29 24 Rural Urban High stunt rate Low stunt rate Components of Adequacy in Care Components of Adequacy in Care By wealth quintile, children less than 60 months By Age, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 77 72 69 68 66 63 50 47 43 40 37 39 39 38 34 26 24 26 28 29 32 31 27 31 27 30 25 23 1 2 3 4 5 0-23 months 24-59 months 2007 Mother's handwashing No smoking 2013 Mother's handwashing No smoking Source: Authors’ calculations based on 2007 and 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 21 FIGURE 11. COMPONENTS OF ADEQUATE CARE, 2013 Components of Adequacy in Care, 2013 Components of Adequacy in Care, 2013 By urban/rural, children less than 36 months Percentage of children who meet criteria National, children less than 36 months Percentage of children who meet criteria 72 69 66 53 49 51 45 40 38 43 36 33 37 26 29 Rural Urban Components of Adequacy in Care, 2013 Components of Adequacy in Care, 2013 By district stunting, children less than 36 months By wealth quantile, children less than 36 months Percentage of children who meet criteria Percentage of children who meet criteria 71 74 73 71 65 66 62 54 53 54 48 49 50 50 45 46 44 45 36 36 36 36 39 38 37 38 38 34 35 32 32 27 24 27 23 High stunt rate Low stunt rate 1 2 3 4 5 Source: Author estimates based on 2013 RISKESDAS. Early breastfeeding* Appropriate breastfeeding* Complementary** Note: *For 0- to 24-month-olds; **For 6- to 8-month-olds Handwashing Smoke-free Source: Authors’ calculations based on 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). professionals are a doctor, a nurse or a midwife. For Between 2007 and 2013 there were large post-natal care there are several variables that are improvements in immunization rates. Of the adequate included. The first measure captures whether or not health components available for all children under 60 the child was seen for a neonatal checkup and it can months of age it is possible to compare adherence to be constructed for 2007 (for those under 12 months immunization schedules for all children and Vitamin of age) and for 2013 (for those under 36 months of A supplementation for those 7 months or older across age). The second post-natal component measures the two survey years (Figure 12). Nationally up to date adherence to the national immunization schedules for immunizations more than doubled from 25 percent BCG, oral polio, DPT, and measles and is available for of children to 62 percent of children, with the main all three years. In the construction of this variable we improvement occurring between 2007 and 2010 with follow the Immunization Guidelines by the Ministry of 55 percent of children immunized in 2010. Vitamin Health (1997). While immunizations themselves may A supplementation also increased from 72 percent not have a direct positive effect on a child’s height, the to 76 percent for children seven to 59 months of age. component is introduced as a proxy for the availability Vitamin A supplementation in 2010 was slightly lower, and use of health care services. The third post-natal at 70 percent, than in 2007. As with many of the component, available for all three years, measures other measures analyzed, children living in high-stunt whether or not a child older than six months of age districts had lower levels of access to these health has received a vitamin A supplement in the past six components than children living in low-stunt districts. months, as recommended by the Health Ministry. Chapter 4. 22 However, the improvements were greater in the than in 2007 and therefore the increases in the gap were wealthier households such that the gap in immunization not from a larger share of children in the high wealth rates between the lowest and highest wealth quintiles quintile receiving the supplementation in 2013 than in grew. In 2007 the gap in immunization rates between 2007, but from fewer children in the low wealth quintile the children in the lowest wealth quintile and the receiving the supplementation in 2013 than in 2007. highest wealth quintile was seven percentage points and by 2013 it had grown to 32 percentage points (Figure Both immunizations and vitamin A supplementation 12). For Vitamin A supplementation the gap increased rates were higher in urban areas than in rural areas from seven percentage points to 11 percentage points. and in low-stunt districts than high-stunt districts. Furthermore, in the lowest income quintile fewer Between 2007 and 2013 the immunization rate children received the Vitamin A supplement in 2013 increased by 32 percentage points in rural areas FIGURE 12. COMPARISON OF ADEQUATE HEALTH COMPONENTS, 2007, 2010 AND 2013 Components of Adequacy in Health National, children less than 60 months Percentage of children who meet criteria 76 72 70 62 55 25 Components of Adequacy in Health Components of Adequacy in Health By Urban/Rural, children less than 60 months By district stunting, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 75 77 75 78 74 70 67 69 69 69 56 51 29 31 22 17 Rural Urban High stunt rate Low stunt rate Components of Adequacy in Health Components of Adequacy in Health Percentage of children who meet criteria Percentage of children who meet criteria By wealth quintile, children less than 60 months By Age, children less than 60 months 79 79 767476 77 75 7273 7373 75 72 72 75 69 71 69 68 71 69 6465 65 63 62 61 5456 56 57 56 55 49 39 33 25 25 28 21 24 19 1 2 3 4 5 0-23 months 24-59 months Source: Author estimates based on 2007, 2010 and 2013 RISKESDAS. 2007 Immunizations Vitamin A* Note: *Only for children 7 months or older 2010 Immunizations Vitamin A* 2013 Immunizations Vitamin A* Source: Authors’ calculations based on 2007, 2010 and 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 23 and by 38 percentage points in urban areas, further districts (Figure 13). In high-stunt districts 82 percent of widening the immunization gap (Figure 12). In the low- mothers had gone to at least one prenatal visit whereas stunt districts the immunization rate improved by 38 in low-stunt districts 90 percent of mothers had been percentage points, whereas in the high-stunt districts seen by a medical professional prior to giving birth. by slightly less, 34 percentage points. Similarly, 62 percent of the infants had been seen by a doctor after birth in high-stunt districts whereas 70 In terms of the child’s age, the older children have percent had been seen by a doctor in low-stunt districts. seen greater improvements in both immunization These comparisons suggest that in 2007 the use of status and vitamin A supplementation from 2007 to health care infrastructure was greater in the low-stunt 2013. In 2007 only 19 percent of children 24 to 59 districts. Furthermore, the use of health care increased months of age were up to date on their vaccinations; with increased wealth such that 91 percent of mothers in 2013 the percentage had risen to 61 percent (Figure in the wealthiest quintile went to a prenatal checkup 12), or in other words, the rate more than tripled. whereas only 80 of mothers in the poorest households These are sizable improvements. However, still nearly did so. Similarly, 75 percent of children from wealthier 40 percent of children did not receive the complete households went to a post-natal checkup whereas only set of immunizations. 59 percent of children from the poorest households were evaluated after birth. These differences in use may also For 2007 it is possible to also look at prenatal visits reflect differences in access to health infrastructure. and post-natal checkup visits for children less than 12 months of age. In the high-stunt districts the prevalence In 2013, more than four-fifths of children less than 36 of the health components was lower than in low-stunt months of age were seen at least 4 times prenatally FIGURE 13. COMPONENTS OF ADEQUATE HEALTH, 2007 Components of Adequacy in Health Components of Adequacy in Health National, children less than 60 months, in 2007 By Urban/Rural, children less than 60 months, in 2007 94 Percentage of children who meet criteria 86 81 Percentage of children who meet criteria 74 75 72 70 67 62 35 28 25 Rural Urban Components of Adequacy in Health Components of Adequacy in Health By district stunting, children less than 60 months, in 2007 By wealth quintile, children less than 60 months, in 2007 91 Percentage of children who meet criteria Percentage of children who meet criteria 90 87 89 82 84 80 75 73 75 76 69 70 69 71 72 70 65 67 62 59 31 28 24 25 25 21 17 High stunt rate Low stunt rate 1 2 3 4 5 Source: Author estimates based on 2007 RISKESDAS. Prenatals Growth Immunizations Vitamin A* Note: *Only for over 7-month-olds Source: Authors’ calculations based on 2007 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Chapter 4. 24 (83 percent), were delivered by a health professional the use of health care facilities was in general greater (88 percent) and were seen within the first month for by wealthier households than poorer households. a post-natal check (82 percent). In 2013, 95 percent Slightly less than two-thirds of children in the poorest of the mothers were seen at least once during the households were seen in prenatal visits (64 percent), pregnancy, but were 83 percent were seen at least were delivered by a health care professional (63 percent) the recommended four times. As in 2007, in 2013 or had a post-natal checkup (63 percent), whereas of children in urban areas or in low-stunt districts tended children in the wealthiest households, 90 percent were to better use health care facilities than those in rural seen in prenatal visits, 98 percent were delivered by areas or high-stunt districts (Figure 14).15 Similarly, , a health care professional and 88 percent had a post- FIGURE 14. COMPONENTS OF ADEQUATE HEALTH, 2013 Components of Adequacy in Health, 2013 National, children less than 36 months 88 83 Percentage of children who meet criteria 82 77 62 Components of Adequacy in Health, 2013 Components of Adequacy in Health, 2013 By urban/rural, children less than 36 months By district stunting, children less than 36 months 94 93 Percentage of children who meet criteria Percentage of children who meet criteria 89 86 89 85 81 78 78 78 77 77 76 74 73 72 67 69 56 52 Rural Urban High stunt rate Low stunt rate Components of Adequacy in Health, 2013 Components of Adequacy in Health, 2013 By wealth quantile, children less than 36 months By age, children less than 36 months 95 98 92 Percentage of children who meet criteria Percentage of children who meet criteria 87 89 87 90 88 89 84 84 82 82 85 79 81 78 77 79 80 77 77 80 77 67 68 70 64 63 63 66 62 62 57 40 1 2 3 4 5 0-23 months 24-35 months Source: Author estimates based on 2013 RISKESDAS. Note: *Only for 7-month-olds or older Prenatals Assisted Growth Immunizations Vitamin A* Source: Authors’ calculations based on the 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 25 natal checkup. Based on the age cohorts, the younger For 2010 and 2013, the available components to derive cohort was slightly more likely to have benefitted an adequate environment indicator are in better from health facilities than the older cohort, suggesting alignment with the measures also used to evaluate some improvements in these through time. The fact progress made towards SDGs. The JMP definition that 88 percent of children were born with a skilled for improved sanitation is one “that effectively professional present suggests a potential avenue for separates excreta from human contact, and ensure promoting behaviors conducive for better nutrition, that excreta do not re-enter the immediate household such as immediate breastfeeding, age-appropriate environment” (WHO/UNICEF Joint Monitoring feeding practices, and handwashing practices. Program for Water Supply and Sanitation, 2015, pg. 20). Basic sanitation is defined as having access to a non-shared facility that is: (1) goose neck to sewage Adequate disposal facility (SPAL), (2) Pit/latrine with floor to SPAL, (3) Goose neck to septic tank, or (4) Pit/ Environmental Services latrine with floor to septic tank.16 In order to capture the child’s exposure to pathogens in her larger There is detailed information for 2010 and 2013 surrounding environment we also consider a measure regarding the various components of adequate of community sanitation which is based on 75% of environment, but not for 2007. For adequate households in the child’s kecamatan having access to environmental, the two more recent surveys have improved sanitation. sufficient information to construct variables in line with the Joint Monitoring Program for Water Supply As per JMP classification, improved drinking water and Sanitation (JMP) definitions as well as with the source is considered to be one that “protects drinking proposed components for evaluating progress towards water from outside contamination, especially from fecal the Sustainable Development Goal (SDG) 6. For 2007 the matter” (WHO/UNICEF Joint Monitoring Program for questions were worded differently and the sanitation Water Supply and Sanitation, 2015, pg. 21). Effectively and drinking water components are at best weak the source of drinking water is considered to be improved proxies for the JMP categorization. if it comes from one of the following: (1) piped water, (3) artesian well or pumped, (3) protected well or spring, or For 2007, the components explored for an adequate (4) rain water. Furthermore, bottled water is considered environment indicator are a modified measure of improved, but only if the main source of water for improved sanitation, a modified measure of safely washing and cleaning is from an improved source. We managed drinking water and a measure of community consider additional, stricter, definitions of acceptable sanitation. Sanitation is considered improved if the drinking water. For a household to have access to basic mother’s usual place of defecation is a latrine and drinking water, in addition to coming from an improved the waste water from the dwelling goes outside of source, as described above, the source must be within the house into a closed area. This definition is not in a 30 minute roundtrip (including any queuing) from line with the JMP classification as there is no detailed the dwelling. An even more stringent condition is a information on the type of sanitation facility used modification of safely managed water requiring the by the household. Community sanitation is defined water to be piped to premises, or if the drinking water is as 75% of households in the kecamatan have access bottled, refill or retail piped water then the main source to improved sanitation. Safely managed drinking of water must be piped to premises. water is defined by the source being at most a 30 minute roundtrip from the dwelling, self-reported In 2007, 20% of children less than 60 months lived in availability throughout the year, and self-reported lack households with access to improved sanitation, 5% of problems in the quality of the water. This definition of children live in communities where at least 75% is not in line with the JMP classification as there is no of the households had access to improved sanitation, information on the source of the drinking water. and 62% lived in households with access to safely 15 These figures may not only reflect the use of but also the availability of health services in the vicinity of the household. 16 We also consider the Government of Indonesia (GoI) definition but do not report the results here. Chapter 4. 26 managed water. That is, in 2007 access to adequate Both access to sanitation and drinking water is environmental infrastructure was low, especially in correlated with the household’s wealth. Only 13 percent terms of access to sanitation. Furthermore, as discussed of children in households in the bottom wealth quintile above, given the laxness in the definitions, these values had access to improved sanitation where as 33 percent are most likely overestimates of true access (Figure 15). of children in the highest quintile had such access (Figure 15). Similarly 57 percent of children in the In 2007 children living in low-stunt districts or in lowest wealth quintile had access to safely managed urban areas had better access to improved sanitation drinking water where as 70 percent of children in the and drinking water than children living in high-stunt highest wealth quintile had access. Access to adequate districts or rural areas. Children in urban areas were community sanitation was relatively constant, at five nearly four times as likely to have access to improved percent, across the five wealth quintiles. That is, even sanitation as children in rural areas and 16 percentage though children from wealthier households were more points more likely to have access to safely managed likely to have access to improved sanitation in their own water (Figure 15). Children in low-stunt districts were dwelling, in most cases less than three-fourths of the more than twice as likely to have access to improved surrounding dwellings had such access. sanitation as children in high-stunt districts, and 64 percent of the children in low-stunt districts had access Between 2010 and 2013 there were slight improvements to safely managed drinking water in comparison with in access to improved sanitation and drinking water, 57 percent of children in high-stunt districts. but the improvements were not distributed equally FIGURE 15. COMPONENTS OF ADEQUATE ENVIRONMENT, 2007 Components of Adequacy in Environment Components of Adequacy in Environment National, children less than 60 months, in 2007 By Urban/Rural, children less than 60 months, in 2007 Percentage of children who meet criteria Percentage of children who meet criteria 73 62 57 37 20 10 11 5 1 Rural Urban Components of Adequacy in Environment Components of Adequacy in Environment By district stunting, children less than 60 months, in 2007 By wealth quintile, children less than 60 months, in 2007 Percentage of children who meet criteria Percentage of children who meet criteria 70 64 62 65 57 57 60 30 33 26 17 20 12 10 13 4 4 5 5 6 1 High stunt rate Low stunt rate 1 2 3 4 5 Improved sanitation Improved community sanitation Safely managed drinking water Source: Authors’ calculations based on 2007 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 27 across income quintiles. For 2010 and 2013 it is possible children living in urban areas. The differences were to construct adequate environment variables that are greater for sanitation measures than for drinking water. more closely aligned with definitions used to assess the In rural areas, 47 percent of children had access to Sustainable Development Goals. basic sanitation whereas in urban areas 74 percent had such access (Figure 17). Also, urban children as a whole Nationally, access to improved sanitation increased had better access to basic sanitation with 62 percent of from 53 percent to 60 percent (Figure 16). Access to children living in communities where at least 75 percent improved drinking water improved from 83 percent of the households had access to basic sanitation, versus to 87 percent and access to basic water improved from only 27 percent of rural children. Access to improved 82 percent to 86 percent. However, fewer households and basic water was relatively high in both communities, in the bottom two wealth quintiles had access to with about 79 percent of rural children having such improved sanitation and fewer households in the access and 94 percent of urban children having access to bottom wealth quintile had access to improved and improved and basic water. However, far fewer children basic drinking water.17 had access to safely managed water (effectively piped water to the premises), with about 11 percent of rural In general, in 2013 children living in rural areas were children had access to safely managed water and 30 less likely to have access to adequate environment than percent of urban children having such access. FIGURE 16. COMPARISON OF ADEQUATE ENVIRONMENT COMPONENTS, 2010 AND 2013 Components of Adequacy in Environment - Sanitation Components of Adequacy in Environment - Sanitation National, children less than 60 months By wealth quintile, children less than 60 months 95 Percentage of children who meet criteria 86 82 Percentage of children who meet criteria 68 69 62 57 20 45 30 26 2 1 2 3 4 5 Components of Adequacy in Environment - Drinking water Components of Adequacy in Environment - Drinking water National, children less than 60 months By wealth quintile, children less than 60 months 97 94 90 8893 93 9296 Percentage of children who meet criteria 90 90 83 87 86 8281 8080 85 84 Percentage of children who meet criteria 82 73 71 64 62 3635 2826 21 18 17 21 21 11 13 7 1 2 3 4 5 2010 Improved Basic Safely managed 2013 Improved Basic Safely managed Source: Authors’ calculations based on 2010 and 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). 17 Given the poor overlap between the survey questions regarding sanitation and drinking water, it is not possible to look at changes between 2007 and 2013 in environmental service variables Chapter 4. 28 In general, in 2013 children living in high-stunt districts living in the poorest households had access to basic were less likely to have access to adequate environment sanitation, 95 percent of the children living in the than children living in low-stunt districts. In high-stunt wealthiest households had access to basic sanitation districts, 45 percent of children had access to basic (Figure 17). That is, almost no child in a poor household sanitation whereas in the low-stunt districts 72 percent had access to basic sanitation, whereas nearly all children had such access (Figure 17). Also, the communities in in the wealthiest households had access. As the wealth which the children in low-stunt districts live as a whole quintile increases so does the percentage of children had better access to basic sanitation with 60 percent who had access to basic sanitation. The wealthier the of children living in communities where at least 75 household the more likely it is that the child also had percent of the households had access to basic sanitation, access to improved, basic or safely managed water. versus only 28 percent of children in high-stunt However, even in the lowest wealth quintile 64 percent districts living in such communities. Similarly, access to of the children had access to an improved water source basic drinking water was higher in low-stunt districts at and 62 percent to basic drinking water. Thus access to the 91 percent than high-stunt districts at 77 percent. adequate environment components are the ones with most inequality in access in terms of wealth. In 2013 there were sizable disparities across wealth quintiles in terms of access to adequate environment components. Whereas only two percent of children FIGURE 17. COMPONENTS OF ADEQUATE ENVIRONMENT, 2013 Components of Adequacy in Environment, 2013 Components of Adequacy in Environment, 2013 National, children less than 36 months By urban/rural, children less than 36 months 94 94 Percentage of children who meet criteria 87 86 Percentage of children who meet criteria 79 78 74 61 62 47 45 27 30 21 11 Rural Urban Components of Adequacy in Environment Components of Adequacy in Environment By district stunting, children less than 60 months, in 2013 By wealth quantile, in 2013 Percentage of children who meet criteria Percentage of children who meet criteria 94 93 95 97 96 92 91 9090 86 79 81 80 77 74 72 68 64 62 63 60 45 46 35 28 26 26 26 17 19 18 13 7 2 4 High stunt rate Low stunt rate 1 2 3 4 5 Sanitation Basic Community Drinking water Improved Basic Safely managed Source: Authors’ calculations based on 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 29 Chapter 5. Synergies By definition, synergies are present in the event food security only (and inadequate care, health and that various elements are combined and interact to environment) and 0 otherwise. Similarly, B2 is 1 when produce a total effect that is greater than the sum of the households is adequate in environment only (and individual elements. In this section, an effort is made inadequate care, food, and environment) and is 0 to derive some quantitative estimates of the role of otherwise; B3 is 1 when the household is adequate in synergies associated with having simultaneous access health (and inadequate care, food, and environment), to adequate levels in one or more of the clusters of and B4 is 1 when the households is addequate in care food security, child care, and health and environment, (and inadequate food, heatlh, and environment). In the on child nutrition. same vein, B12 takes the value of 1 if the households has adequate food security and adequate in environment at The analysis is aimed at analyzing whether the mean the same time (and inadequate care and health), and B13 height-for-age Z-scores of children with simultaneous takes the value of 1 if the household has adequate food access to adequate levels in two or more of the security and adequate health (and inadequate care and underlying determinants of nutrition is equal to or environment), and so on. greater than the sum of the mean height-for-age Z-scores of children with access to adequate level in A comparison of the mean values of height-for-age only one of the determinants at a time. To explore the Z-scores in each of these groups of children can shed light potential synergies among the four determinants of on the extent to which simultaneous access to adequate nutrition outcomes, a simple regression model is used levels of two or more of the four clusters is associated to summarize in a parsimonious way the differences in with higher heights for age Z-scores of children. Estimates the mean height-for-age among children with access of the regression model A below can provide the answer to only or more of the four drivers of the underlying to this question: determinants of nutrition. On purpose, no additional controls are used in these regressions since including HAZi = α + β1B1 + β2B2 + β3B3 + β4B4 + such controls is likely to create the impression that an + β12B12 + β13B13 + β14B14 + β23B23 + β24B24 + effort is made to minimize the influence of confounding factors in the relationship between the dependent and +β123B123 + β124B124 + β234B234 + εi . independent variables in the regression, a practice common to all studies aimed at estimating causal In this specification the constant term α provides an regressions within an econometric framework. estimate of the mean value of HAZ scores for children without access to an adequate level for any of the Each child is assigned into one of the fifteen exclusive four underlying determinants of nutrition: food groups each group distinguished by whether the security (B1=0), environment (B2=0), health (B3=0) and household/child has access to an adequate level of care (B4=0). With E ( HAZi | B =1 or 0) denoting the only one or only two or only three or all four of the expected (or mean) value of height-for-age (outcome), underlying determinants of nutrition. For this purpose, conditional on having adequate access (B1=0) or 15 binary variables are constructed by taking into inadequate access (B=0), the expected height-for-age consideration whether the child has access to adequate for when the child does not have adequate access to levels in the other determinants. Specifically, the binary any of the four determinants is:18 variable B1 is equal to 1 if the households has adequate E ( HAZi | B1 = 1, B2 = 0, B3 = 0, B4 = 0 ) = α 18 It is also assumed that E (εi | B1, B2, ... , B1234 ) = 0. Chapter 5. 30 The coefficients βj where j =1, 2, 3, 4, yield estimates of coordination between the entities implementing the increase in the mean HAZ score of children when a interventions on j and k, for example, food and care child has access to adequate level in one of the clusters practices, and/or neglect of the impediments and only (or net of the potential gain in HAZ of having negative side effects associated with inadequate access access to an adequate level in one or more of the other to the other underlying determinants of nutrition clusters). That is: (sanitation and health services). E ( HAZi | B1 = 1, B2 = 0, B3 = 0, B4 = 0 ) = α + β1 As such, the estimates from model A above serve as a E ( HAZi | B1 = 0, B2 = 1, B3 = 0, B4 = 0 ) = α + β2 useful benchmark for policy in terms of highlighting the potential gains in height-for-age Z-scores that could E ( HAZi | B1 = 0, B2 = 0, B3 = 1, B4 = 0 ) = α + β3 be accomplished with having simultaneous access E ( HAZi | B1 = 0, B2 = 0, B3 = 0, B4 = 1 ) = α + β4 to adequate levels of the underlying drivers. This specification allows for the exploration of the patterns Specifically, the coefficient β1 yields an estimate of correlation between the various adequacy measures of the increase inthe mean HAZ score of children and nutritional outcomes as measured by height- (compared to the mean HAZ score of the reference for-age. That is, the model estimates the correlation group summarized by the constant term α ) have between adequacies and height-for-age for each set of access to adequate food security only (B1= 1) but do adequacies based on information in one time period. not have access to adequate environment, (B2 = 0), to adequate health (B2 = 0) or adequate care (B4 = 0). The The estimates of the mean differences in height-for- coefficients βjk , and have analogous interpretations for age among children in each of the 15 different groups environment, health and care, respectively. provide only indirect evidence, at best, on the presence or absence of synergies among the four determinants The coefficients βjk where j, k = 1, 2, 3, 4, yield estimates of nutrition. In an effort to get more direct evidence of the potential complementarities or synergies on synergies, Appendix C also reports estimates from associated with having simultaneous access to an alternative econometric specification (model B). adequate levels in more than one of the clusters of The estimates in Appendix C provide no evidence of underlying determinants of nutrition. Specifically, the significant positive synergies, neither for the sample mean HAZ score of children having access to adequate as a whole nor for urban and rural children separately. food security and adequate care only at the same time Thus, simultaneous access in more than one of the (and thus inadequate care and health), is summarized key determinants of nutrition does not appear to be by the expression associated with an increase in the height of children beyond the sum of the gains in height associated with E ( HAZi | B14 = 1 ) = α + β14 , or more generally by adequate access to one determinant at a time. E ( HAZi | Bjk = 1 ) = α + βjk. It is important to bear in mind that the model In an ideal context, if synergies are present one employed above does not allow for causal inferences would expect that βjk > βk + βk . For example, if there on the effects of having access to adequate levels in the significant synergies between food security and child various clusters adequacy components on nutrition care practices, ceteris paribus, one would expect nor does it provide a formal test of the UNICEF the increases in the HAZ score of children with conceptual framework. A more rigorous causal simultaneous access to adequate food security and analysis would require the use of methods aimed at child care practices to be greater than the sum of the addressing the endogeneity bias associated with the gain in the mean HAZ score of children that have fact that many of the components themselves are to access to adequate child care practices only (i.e., β14 > a large extent choice variables (e.g. such as child care β1 + β4 ). variables, immunizations, and visits for prenatal care) as well as the inclusion of additional control variables Furthermore, in case βjk ≤ βk + βk , one can infer that aimed at reducing or eliminating the impact of other the potential synergies associated with simultaneous contextual variable omitted from the regression access to two of the underlying determinants of (omitted variable bias). nutirition are not realized either because of poor Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 31 The Underlying one of the four determinants alone cannot substitute for inadequacies in the other three underlying Determinants of Nutrition determinants. Nevertheless, it is important to bear and Height-for-age in 2013 in mind that alternative criteria could be employed. For example, adequate access to care or health or environment or food security could be defined as a child In this section we analyze the correlations between having adequate access to at least one or at least two of nutrition drivers and nutrition outcomes for 2013. the components of each of the four drivers of nutrition. First, the prevalence of access to the four nutrition Alternatively, different weights could be applied to the drivers are discussed, followed by a description of the distinct components of each nutrition driver, reflecting correlations between drivers and outcomes. Similar the preferences, values and/or information of the policy analyses are carried out using the 2007 RISKESDAS. making authority. These results are provided in Appendix D. The 2013 analyses includes children less than 36 PREVALENCE OF ACCESS TO ADEQUATE CHEF IN 2013 months of age. Table 1 summarizes the components used to define the nutrition drivers. Given data A move away from an analysis of the individual availability the components vary by age cohort. For components of the four drivers of malnutrition towards children less than 24 months of age breastfeeding a more aggregate analysis of differences in access to practices are included in access to adequate care. For adequate levels in the four drivers of malnutrition, older children only mother’s hand washing practices care, health, environment and food security involves as well as the household’s smoking behavior are a variety of options. In this report, adequate access included. to care or health or environment or food security is defined as a child having adequate access to all of the In 2013 access to adequate care was the driver with components of each of the four drivers of nutrition. the lowest prevalence rate and access to adequate Although this is admittedly a very strict criterion, it is environment was the driver with the highest in line with the UNICEF conceptual framework that prevalence rate. Nationally only seven percent of the assumes that increases in access to adequate levels in children had access to adequate care (Figure 18). For TABLE 1. COMPONENTS OF ADEQUATE ACCESS, 2013 ADEQUACY COMPONENTS USED (AGE GROUP APPLICABLE) 0 to 23 months 24 to 35 months Exclusive breast feeding (for 0 to 5 month olds only) √ Food Mother’s consumption of fruit and vegetables √ √ Immediate breastfeeding after birth √ Appropriate breastfeeding √ Care Complementary feedings (6 to 8 month olds only) √ Mother washes hands with soap after six events √ √ Household is smoke free √ √ At least four prenatal checkups √ √ Birth assisted by a skilled professional √ √ Health Post-natal checkup √ √ Immunizations up to date √ √ Vitamin A supplementation (7 to 35 month olds) √ √ Access to improved sanitation √ √ Environment Access to basic drinking water √ √ Chapter 5. 32 the younger cohort it was only three percent and for Similarly, very few children had access to adequate the older cohort 14 percent but for the older cohort the food, with only 14 percent having such access measure is only based on mother’s handwashing and nationally. About 56 percent of the children had access household smoking behaviors. These prevalence rates to adequate environment or basic drinking water and are surprising low given that the component with the improved sanitation. The access to adequate health lowest prevalence rate was a smoke-free household was also relatively high, at 46 percent. at around 30 percent. When only feeding practices are included for the younger cohort, the access to In 2013 rural children or children living in districts adequate care rises to 26 percent for the cohort. with high stunting rates were less likely to have FIGURE 18. ACCESS TO NUTRITION DRIVERS, 2013 Adequacy status in 2013, National 0 to 36 month olds Percentage of children who meet criteria 56 46 14 7 Adequacy status in 2013, By urban/rural Adequacy status in 2013, By district stunting 0 to 36 month olds 0 to 36 month olds Percentage of children who meet criteria Percentage of children who meet criteria 71 69 51 52 41 41 40 35 18 18 10 8 10 9 6 6 Rural Urban High stunt rate Low stunt rate Adequacy status in 2013, By wealth quantile Adequacy status in 2013, By age 0 to 36 month olds 0 to 36 month olds 92 Percentage of children who meet criteria Percentage of children who meet criteria 81 62 57 56 53 53 49 47 44 39 26 27 21 17 14 15 14 8 6 11 8 10 5 4 7 1 3 1 2 3 4 5 0-23 months 24-35 months Food Care Environment Health Source: Authors’ calculations based on 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 33 access to all of the four nutrition drivers than urban adequate environment whereas in the highest quintile children or children living in a low-stunt district. The 92 percent of the children had adequate access. largest discrepancy was in the access to adequate environment with only around 40 percent of those Very few children have access to three or more of in rural areas or high-stunt districts having access the nutrition drivers. Nationally only one percent and around 70 percent of those in urban areas or had access to four drivers and eight percent to three low-stunt districts having access (Figure 18). Adequate drivers (Figure 19). Children in high-stunt districts access to environment is also the driver with the or in rural areas were even less likely to have access largest differences by wealth quintiles. One percent of to three or more drivers, four and five percent the children in the lowest wealth quintile had access to respectively. In comparison 13 percent of children in FIGURE 19. NUMBER OF NUTRITION DRIVERS, 2013 Adequacy status in 2013, National 0 to 36 month olds Percentage of children who meet criteria 38 30 23 8 1 Number of adequacies met in 2013, By urban/rural Number of adequacies met in 2013, By district stunting 0 to 36 month olds 0 to 36 month olds Percentage of children who meet criteria Percentage of children who meet criteria 37 38 36 38 37 36 36 34 23 21 13 12 15 12 5 4 0 1 0 1 Rural Urban High stunt rate Low stunt rate Number of adequacies met in 2013, By wealth quantile Number of adequacies met in 2013, By age 0 to 36 month olds 0 to 36 month olds Percentage of children who meet criteria Percentage of children who meet criteria 68 42 42 44 43 39 42 39 36 33 33 30 29 29 23 23 16 18 14 11 8 7 10 4 6 3 0 0 2 0 1 1 2 0 2 1 2 3 4 5 0-23 months 24-35 months Adequate in none Adequate in 1 of 4 Adequate in 2 of 4 Adequate in 3 of 4 Adequate in all 4 Source: Authors’ calculations based on 2013 RISKESDAS. Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Chapter 5. 34 low-stunt districts and urban areas had access to at least three facets. The differences are, however, more In Search of Synergies in 2013 striking when comparing those without access to any driver. In rural areas 34 percent of children had Simultaneous access to two or more of the four key no access, whereas in urban areas 13 percent had no determinants of nutrition is associated with gains in access. In high-stunt districts 38 percent of children height for age z-scores (see Figures 21 and 22). Mean had no access whereas in low-stunt districts 15 percent height for age z-scores are higher among children did not have access to any driver. with simultaneous access to adequate levels to two of the four drivers of nutrition and even higher among By wealth, the differences in access are even more children with simultaneous access to adequate levels dramatic. In the lowest wealth quintile households, to three of the four drivers of nutrition (Figure 20). 68 percent of the children were adequate in none, More importantly, the same pattern appears to hold whereas in the top wealth quintile only three percent separately in rural and urban areas as well as for were adequate in none (Figure 19). To underline the children in households at the top 60 percent of the difference even further, in the bottom quintile there wealth distribution (Figure 21). All in all, these results were effectively no children who had access to three validate the importance of coordinated multi-sectoral or four of the drivers simultaneously. In the top wealth policies and suggest that the success of “sector-specific quintile 20 percent of the children had such access. nutrition-sensitive” initiatives can be enhanced by better coordination and integration of multisectoral By age, the older cohort is more likely to have access interventions that address effectively the four to one or more adequacy measure than the younger underlying determinants of nutrition. cohort. The result is driven by the fact that the adequate food and adequate care conditions have fewer components for the older age cohort. FIGURE 20. SIMULTANEOUS ACCESS TO DIFFERENT COMBINATIONS OF CHEF AND MEAN HEAIGHT-FOR =-AGE Z-SCORE: 2013 Average Difference in Height-for-Age Score, 2013 Relative to the Reference Group 0.196 Adequate in 1 Adequancy Grouping 0.303 Adequate in 2 0.503 Adequate in 3 0.251 Adequate in 4 -0.200 0.000 0.200 0.400 0.600 Source: Authors’ calculations based on RISKESDAS 2013. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 35 Peeking Underneath the that any observed correlations need to be interpreted with caution given the small percentage of children on Hood which they are based. For example, there are only 53 children (from the sample total of 30,051 children) who The analysis so far did not distinguish among are adequate in food and care only and who have valid the specific drivers of nutrition. A more detailed height information. Furthermore, in this particular investigation of the distribution of children across group, the children who have the lowest Z-scores are the different combinations of the specific drivers of also the children who come from households who nutrition reveals that some combinations (categories) are weighted more, and thus the weighted average have very few children. Less than three percent of HAZ-score results very low. That is, when unweighted children are either adequate in food alone or in care means are compared, those who are adequate in care alone (Figure 22). Furthermore, except for access to and food are no statistically significantly different both health and environment, five percent or less of from those who are adequate in none, but once the sample are adequate in each of the specific two- the sample is weighted, the average Z-score is 0.72 access categories. These low percentages are reminders standard deviations lower. The estimate for the smaller FIGURE 21. SIMULTANEOUS ACCESS TO DIFFERENT COMBINATIONS OF CHEF AND MEAN HEAIGHT-FOR =-AGE Z-SCORE: 2013 Rural areas Urban areas Average Difference in Height-for-Age Score, 2013 Average Difference in Height-for-Age Score, 2013 Rural, Relative to the Reference Group Urban, Relative to the Reference Group 0.214 0.072 Adequate in 1 Adequate in 1 0.255 0.202 Adequate in 2 Adequate in 2 Adequacy Grouping Adequacy Grouping 0.468 0.364 Adequate in 3 Adequate in 3 0.200 0.113 Adequate in 4 Adequate in 4 -0.500 0.000 0.500 1.000 -0.400 -0.200 0.000 0.200 0.400 0.600 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score B40 percent of the distribution T60 percent of the distribution Average Difference in Height-for-Age Score, 2013 Average Difference in Height-for-Age Score, 2013 Children in B40, Relative to the Reference Group Children in T60, Relative to the Reference Group 0.105 0.071 Adequate in 1 Adequate in 1 0.039 0.144 Adequacy Grouping Adequate in 2 Adequate in 2 Adequacy Grouping 0.080 0.329 Adequate in 3 Adequate in 3 -0.078 0.058 Adequate in 4 Adequate in 4 -1.000 -0.500 0.000 0.500 1.000 -0.400 -0.200 0.000 0.200 0.400 0.600 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score Source: Authors’ calculations based on RISKESDAS 2013. Note: Showing point estimates and 95% CI. Chapter 5. 36 FIGURE 22. PERCENTAGE OF CHILDREN IN EACH ADEQUACY GROUPING, BY VARIOUS SUBPOPULATIONS, 2013 Children in Each Adequacy Grouping in 2013 None 23 F 2 C 1 E 22 H 13 FC 0 FE 4 FH 2 CE 1 CH 1 EH 22 FCE 1 FCH 0 CEH 2 FEH 6 FCHE 1 0 10 20 30 40 50 Percentage of children Rural Children in Each Adequacy Grouping in 2013 Urban Children in Each Adequacy Grouping in 2013 None 34 None 13 F 2 F 1 C 1 C 1 E 18 E 26 H 16 H 10 FC 0 FC 0 FE 2 FE 5 FH 2 FH 1 CE 1 CE 2 CH 1 CH 1 EH 17 EH 27 FCE 0 FCE 1 FCH 0 FCH 0 CEH 1 CEH 2 FEH 3 FEH 8 FCHE 0 FCHE 1 0 10 20 30 40 50 0 10 20 30 40 50 Percentage of children Percentage of children Children from B40 in Each Adequacy Grouping in 2013 Children from T60 in Each Adequacy Grouping in 2013 None 54 None 9 F 3 F 1 C 2 C 1 E 7 E 29 H 25 H 8 FC 0 FC 0 FE 0 FE 5 FH 2 FH 1 CE 0 CE 2 CH 2 CH 1 EH 4 EH 30 FCE 0 FCE 1 FCH 0 FCH 0 CEH 0 CEH 3 FEH 1 FEH 8 FCHE 0 FCHE 1 0 10 20 30 40 50 0 10 20 30 40 50 Percentage of children Percentage of children Children from High-Stunt Districts in Each Adequacy Grouping in 2013 Children from Low-Stunt Districts in Each Adequacy Grouping in 2013 None 38 None 15 F 3 F 1 C 1 C 1 E 19 E 24 H 14 H 11 FC 0 FC 0 FE 3 FE 4 FH 2 FH 1 CE 1 CE 2 CH 1 CH 1 EH 14 EH 27 FCE 0 FCE 1 FCH 0 FCH 0 CEH 1 CEH 3 FEH 3 FEH 8 FCHE 0 FCHE 1 0 10 20 30 40 50 0 10 20 30 40 50 Percentage of children Percentage of children Source: Authors’ calculations based on RISKESDAS 2013. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 37 subpopulations is based on an even fewer number of Most children adequate in environment and some children such that, for example, only seven children other nutrition driver(s) are statistically significantly comprise the group of adequate in food and care in the taller than children with access to none. Those who high-stunt districts. have access to adequate food and environment are 0.25 standard deviations taller than those without In 2013, at the national level, access to adequate food any access and those with access to environment only, adequate health only, or environment only are and health are 0.36 standard deviations taller (Figure associated with taller children. A child with access 23). Those with access to adequate environment to adequate food is 0.38 standard deviations taller and two other nutrition drivers are between 0.39 than a child without access to any nutrition driver and 0.55 standard deviations taller those without (Figure 23).19 A child with only access to adequate any such access. The results suggest that access to environment is taller by 0.20 standard deviations adequate sanitation and drinking water (or another and a child with access to adequate health only is 0.19 characteristic associated with such access) play a standard deviations taller, or about half of the average significant role in nutrition outcomes.20 difference of those with access to adequate food only. Appendix B (Table B.5) presents the results when using The underlying determinants of nutrition, as defined stunting status as the dependent variable. The results in this study, were more likely to be correlated with corroborate the findings from using the height-for-age better nutrition outcomes for rural children than Z-score as the dependent variable. urban children. Whereas for both cohorts of rural FIGURE 23: CORRELATION BETWEEN ADEQUACY GROUPINGS AND HEIGHT-FOR-AGE Z-SCORE NATIONAL, 2013 Average Difference in Height-for-Age Score, 2013 Relative to the Reference Group 0.383 F -0.049 C 0.199 E 0.187 H -0.720 F, C 0.248 F, C, E Adequacy Grouping 0.049 F, C, E, H 0.115 C, E 0.156 C, E, H 0.355 E, H 0.505 F, C, E 0.245 F, C, E, H 0.387 C, E, H 0.548 F, C, E, H 0.251 F, C, E, H -1.500 -1.000 -0.500 0.000 0.500 1.000 Difference in Height-for-Age Z-score Source: Authors’ calculations based on RISKESDAS 2013. Notes: Showing point estimates and 95% CI Groups with less than 100 observations are not shown. 19 The regression results are presented in a tabular form in Appendix B. Tables B.1 – B.3 correspond to Figures 22–24 20 The statistically significant negative coefficient on access to both adequate food and care is most likely arises from the small sample of children who fall into this category. Chapter 5. 38 children access to either food only or environment from households in the top 60 percent of the wealth only was associated with better nutrition outcomes, distribution are taller if they have access to health only, no nutrition driver singly was associated with higher environment and health only or food, environment HAZ scores for urban children (Figure 24). and health only. In wealthier households, access or use of health services is associated with better nutrition Analyzing by wealth does not result in many positive outcomes. The results suggest that children tend to have associations. For children from the bottom 40 percent more similar associations by their location (i.e. rural / of the wealth distribution access to food only or urban or highstunt/low-stunt district) than by the access to environment and health only are positively household’s wealth.21 associated with height (Figure 24). Similarly, children FIGURE 24. CORRELATION BETWEEN ADEQUACY GROUPINGS AND HEIGHT-FOR-AGE Z-SCORE, RURAL VS. URBAN AND B40 VS T60 CHILDREN, 2013 Average Difference in Height-for-Age Score, 2013 Rural, Relative to the Reference Group Urban, Relative to the Reference Group 0.403 0.280 F F -0.079 -0.068 C C 0.214 0.060 E E 0.213 0.083 H H Adequacy Grouping Adequacy Grouping 0.107 0.228 F, C, E, F, C, E, H 0.285 -0.178 F, C, E, H F, C, E, H 0.045 C, E 0.007 0.232 C, E, H C, E, H 0.028 0.227 C, E, H E, H 0.420 0.348 F, C, E E, H 0.356 0.180 C, E, H C, E, H 0.374 0.610 F, C, E, H 0.113 F, C, E, H F, C, E, H -0.500 0.000 0.500 1.000 -1.000 -0.500 0.000 0.500 1.000 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score B40, Relative to the Reference Group T60, Relative to the Reference Group 0.271 0.362 F F -0.290 C 0.033 0.056 E C 0.204 Adequacy Grouping Adequacy Grouping H 0.001 0.065 E F, C, E 0.067 F, C, E, H 0.112 0.006 H C, E 0.289 C, E, H -0.147 0.169 F, C, E, H E, H 0.315 -0.075 F, C, E C, E, H 0.216 C, E, H 0.377 0.283 F, C, E, H E, H 0.058 F, C, E, H -0.500 0.000 0.500 1.000 -1.000 -0.500 0.000 0.500 1.000 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score Source: Authors’ calculations based on RISKESDAS 2013. 21 Appendix B also reports the results for these subpopulations by age cohort. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 39 Estimating separately by age, suggests that the FIGURE 25. CORRELATION BETWEEN ADEQUACY differences in height by adequacy category are GROUPINGS AND HEIGHT-FOR-AGE stronger in the older cohort. Given that the definition Z-SCORE, BY AGE, 2013 of adequate care is quite different for zero- to 23- month olds and 24- to 35-month olds, it is useful to Average Difference in Height-for-Age Score, 2013 analyze by age cohorts. The first thing to note is that By age, Relative to the Reference Group the average child without access to any nutrition driver in the 0 to 23 month old cohort had a HAZ 0.365 F score of -1.16 standard deviations whereas in the 0.393 older cohort it was -1.85 standard deviations (Figure 0.302 C 25). This fact reflects the steady drop in the average 0.169 HAZ score with age. Given that height-for-age is a 0.149 reflection of chronic nutrition conditions, the older age E 0.290 cohort where the growth curves have stabilized better 0.120 captures the ultimate long term associations between H 0.282 nutrition drivers and nutrition outcomes. However, -0.994 the result depends on the degree to which their access F, C -0.216 to the nutrition drivers has changed since birth. As 0.225 long as the access has remained similar since birth, the F, C, E 0.353 estimates for the older cohort reflect the accumulated 0.029 associations. The analysis reveals that for the older F, C, E, H -0.104 Adequacy Grouping cohort almost any combination of nutrition drivers 0.483 was associated with better nutrition outcomes than C, E 0.314 having access to none (Figure 25). The only exceptions 0.415 were access to adequate care only, access to adequate C, E, H 0.353 care and food only, and access to adequate food and 0.301 health only. The coefficient estimates for having access E, H to three or more nutrition drivers (Z-scores from 0.57 0.407 0.714 to 1.13 standard deviations) were greater than the F, C, E coefficient estimates for those having access to one 0.814 -1.053 or two drivers (up to 0.41 standard deviations). That F, C, E, H is, having access to the nutrition drivers is associated 1.129 0.373 with nutrition outcomes and the greater number C, E, H of drivers that a child has access to, the larger the 0.742 0.506 average difference in height between the group and F, C, E, H those without access to any driver. 0.572 0.224 F, C, E, H Investigating potential differences in rural and urban 0.615 areas depending on the age of the child, reveals that -3.000 -2.000 -1.000 0.000 1.000 2.000 the average height difference in the older cohort is Difference in Height-for-Age Z-score greater than in the younger cohort (Figure 26). In rural areas, access to adequate level in only one of the 0 to 23 months 24 to 35 months four drivers of nutrition among older children (24 to 35 months old) appears to be associated with better Source: Authors’ calculations based on RISKESDAS 2013. nutrition outcomes than the nutrition outcomes of Notes: Showing point estimates and 95% CI younger children (0 to 23 months old) in rural areas. This difference between age groups may reflect the Chapter 5. 40 fact that the beneficial effects of access to adequate In urban areas, access to any of the nutrition drivers level in any one of the four drivers of nutrition may alone does not appear to be associated with better become more apparent as the child ages, rather than nutritional outcomes (Figure 27). This may reflect the contemporaneously. If one were to assume that the fact that there were far fewer children in urban areas access at 24 to 35 months reflects the access the without access to any driver (13 percent) as well as child had since birth, having access to adequate the fact that these children were on average taller environment in rural areas is associated with a 0.25 than reference group of children in the rural areas (i.e. standard deviation increase in average HAZ score, without access to any of the drivers of nutrition). In access to adequate health with a 0.33 standard fact, in urban areas the children without access to any deviation increase in average HAZ score and access to were more similar in height to those with access to adequate food with a 0.54 standard deviation increase only one driver than in rural areas, i.e. 0.24 and 0.29 in average HAZ score after completing two years of standard deviations taller for the younger and older age (Figure 26). cohorts, respectively. FIGURE 26. CORRELATION BETWEEN ADEQUACY GROUPINGS AND HEIGHT-FOR-AGE Z-SCORE, RURAL CHILDREN BY AGE, 2013 Average Difference in Height-for-Age Score, 2013 For Rural 0 to 23 month olds, Relative to the Reference Group For Rural 24 to 35 month olds, Relative to the Reference Group 0.342 0.542 F -0.024 F 0.247 C 0.194 C 0.245 E 0.127 E 0.326 H -0.894 H -0.173 F, C F, C Adequacy Grouping Adequacy Grouping 0.371 0.163 F, C, E, -0.293 F, C, E, -0.107 F, C, E, H 0.538 F, C, E, H 0.186 C, E 0.399 C, E 0.153 C, E, H 0.367 C, E, H 0.241 E, H 1.615 E, H 0.329 F, C, E 1.763 F, C, E 1.257 F, C, E, H -0.074 F, C, E, H 0.674 C, E, H 0.518 C, E, H 0.727 F, C, E, H 0.147 F, C, E, H 0.590 F, C, E, H F, C, E, H -2.000 0.000 2.000 4.000 -1.000 0.000 1.000 2.000 3.000 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score Source: Authors’ calculations based on RISKESDAS 2013. Notes: Showing point estimates and 95% CI. Groups with less than 100 observations are not shown Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 41 However, urban children aged 24 to 35 months of Although having access to more than one nutrition age showed most consistent correlations with access driver in many cases is associated with better to three of the four nutrition drivers and nutrition nutrition outcomes, in general the analysis has been outcomes. For the older urban cohort, access to any able to uncover any evidence of significantly positive three of the four nutrition drivers was associated synergies associated with having simultaneous access with better nutrition outcomes (Figure 27). These to two or more drivers of nutrition. In fact estimating range from 0.35 standard deviations for access to Model B yielded quite a few statistically significant food, environment and health to 0.89 standard negative synergies, suggesting some substitutability deviations for food, care and health. However, even among the four different drivers. The synergy results with access to food, care and health, the average child are presented in Appendix C. in this category still had a negative HAZ at -0.76 standard deviations. The results suggest that there are additional drivers that are not captured by the analyses which contribute to nutrition outcomes. FIGURE 27. CORRELATION BETWEEN ADEQUACY GROUPINGS AND HEIGHT-FOR-AGE Z-SCORE, URBAN CHILDREN BY AGE, 2013 Average Difference in Height-for-Age Score, 2013 For Urban 0 to 23 month olds, Relative to the Reference Group For Urban 24 to 35 month olds, Relative to the Reference Group 0.314 0.099 F 1.030 F -0.020 C -0.000 C 0.171 E 0.051 E 0.136 H -1.280 H -0.310 F, C F, C Adequacy Grouping Adequacy Grouping 0.033 0.280 F, C, E, 0.456 F, C, E, -0.214 F, C, E, H 0.326 F, C, E, H 0.248 C, E 0.366 C, E 0.494 C, E, H 0.137 C, E, H 0.364 E, H 0.331 E, H 0.817 F, C, E -1.479 F, C, E 0.891 F, C, E, H 0.443 F, C, E, H 0.627 C, E, H 0.358 C, E, H 0.351 F, C, E, H 0.104 F, C, E, H 0.450 F, C, E, H F, C, E, H -4.000 -2.000 0.000 2.000 4.000 -2.000 -1.000 0.000 1.000 2.000 Difference in Height-for-Age Z-score Difference in Height-for-Age Z-score Source: Authors’ calculations based on RISKESDAS 2013. Notes: Showing point estimates and 95% CI. Groups with less than 100 observations are not shown Chapter 5. 42 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 43 Chapter 6. Summary of Findings 1. Between 2007 and 2013, the mother’s fruit and 5. Between 2010 and 2013 access to improved vegetable consumption as well as their self-reported sanitation drastically fell in the bottom two hand washing practices improved nationally. Fruit wealth quintiles. In the lowest wealth quintile and vegetable intake increased from 12 percent to it fell from 30 percent of households having 15 percent. These percentages are very low overall access to improved sanitation in 2010 to only 2 given the lax requirement of consuming both fruit percent having such access in 2013. In the second and vegetables on five out of seven days. In 2013 wealth quintile the drop was not as drastic, but the prevalence rates were higher in urban areas nevertheless the access nearly halved from 45 than rural areas, and in low-stunt districts than in percent to 26 percent. Also access to improved high-stunt districts. Self-reported handwashing drinking water fell in the bottom wealth quintile increased from 38 percent to 67 percent. In 2013, from 73 percent to 64 percent. The survey 62 percent of rural mother and 73 percent of questionnaires were similar, so it is not clear urban mothers washed their hands. Also in high- to what the decrease in the access to these stunt districts handwashing was less frequent (57 environmental components is attributable. percent) than in low-stunt districts (73 percent). 6. While there has been in increase in the access 2. Breastfeeding practices have improved between to most of the components analyzed, for some 2007 and 2013. Age-appropriate breastfeeding components there has also been an increase in increased from 37 percent to 69 percent, although the inequality of access between children from there has been a slight decline in breastfeeding poorer households and wealthier households. For between 2010 and 2013. However, early example, although immunization rates rose for all breastfeeding rates improved between 2010 wealth quintiles between 2007 and 2013, the gap and 2013, rising from 29 percent to 37 percent between the first and fifth wealth quintile rose nationally. In 2013 breastfeeding was more from seven percentage points to 32 percentage prevalent in rural areas and in high-stunt districts points. Furthermore between 2010 and 2013 the than in urban areas and low-stunt districts. Early immunization rate in the bottom quintile fell 10 breastfeeding rates were similar across high- and percentage points. Such a large drop in the span low-stunt districts and rural and urban areas. of three years is of concern. Similarly the gap 3. Immunization rates also improved substantially between mother’s handwashing practices rose between 2007 and 2013, but vitamin A from 13 percentage points to 27 percentage points. supplementation remained relatively constant. 7. Even more worrisome is the fact that some pro- Nationally immunization rate rose from 25 percent nutrition behaviors decreased in the lowest wealth to 62 percent, an improvement of 37 percentage quintile. For example, the gap in household head points. In 2013 immunization rates were higher in smoking was six percentage points in 2007 but it urban areas and in low-stunt districts than in rural rose to 17 percentage points with a two percentage areas and high-stunt districts. point increase is smoking in the bottom quintile 4. Smoking by the head of the household decreased between the years. Similarly, the gap in vitamin between 2007 and 2013, but not in the lower A supplementation was only six points in 2007 wealth quintiles. In 2007, 73 percent of household but rose to 10 percentage points in 2013 with heads smoked whereas in 2013 the percentage had a four percentage point decrease in vitamin A dropped slightly to 70 percent. In 2013 smoking supplementation in the poorest wealth quintile. was more prevalent in rural areas and in high- The gap in the mother’s fruit and vegetable stunt districts. consumption increased from 17 percentage points Chapter 6. 44 to 24 percentage points with one percentage point with access to food, environment and health. The decrease in fruit and vegetable consumption by rest were composed of four percent or less. mothers in the lowest quintile. 11. In both years in the national sample, having 8. The only driver where the poorer households access to an adequate level to adequate food only had more gains than the wealthier household or adequate environment only is associated with was age-appropriate breastfeeding. In the bottom taller children. In 2007 the gains were about 0.10 wealth quintile age-appropriate breastfeeding rose standard deviations for adequate food and 0.36 from 31 percent to 74 percent, a 43 percentage standard deviations for adequate environment. point increase, whereas in the top wealth Adequate in care was associated with an increase quintile it rose from 23 percent to 62 percent, a of 0.13 standard deviations in the average height- 39 percentage point increase. However, the gap for-age Z-score. In 2013 the correlations were in age-appropriate breastfeeding is closing. The 0.38 standard deviations for either adequate food most recent trend, from 2010 to 2013, was a six and 0.20 for adequate in environment and 0.19 percentage point drop in breastfeeding in the for adequate in health. Also, in both years access bottom quintile and a three percentage point to both food and environment was associated increase in the top quintile. with higher Z-score. In 2007 the difference was 9. Even though in both years, those with access to a 0.37 standard deviations and in 2013 it was 0.25 single nutrition driver were the largest group, in standard deviations.23 At the national level access both years there was a sizable number of children to these two drivers seems to be consistently without access to any of the nutrition drivers. In associated with better nutrition outcomes. 2007 thirty-nine percent and in 2013 twenty- 12. However, the results regarding adequate food three percent of children did not have access to and adequate environment are not consistent any nutrition dimension.22 In 2007 forty-three across different subpopulations. Namely, in 2007, percent had access to one driver and in 2013 adequate food is only correlated with HAZ scores in thirty-eight percent. Given that the definitions urban areas, where those with access to adequate are based on different sets of components, the food were 0.21 standard deviations taller. Access two numbers are not comparable, and we cannot to adequate food was not correlated with height in estimate by how much the “no access” group has rural areas, in high-stunt districts or for children shrunk. Nonetheless the percentages do indicate from households in the bottom 40% of the wealth a widespread lack of access to the underlying distribution. The lack of correlation may arise determinants of nutrition. from the fact that the food adequacy measure 10. In general, only a small percentage of the is based on a per capita measure and is not child children have access to the each of the specific specific. Especially if the allocation of food within combinations of nutrition drivers. For many the family is not equitable, then the measure may combination of nutrition drivers there are only not reflect the child’s true food availability. In few children fitting each description. Of children rural areas, access to adequate environment was with access to multiple drivers, in 2007 the largest associated with a 0.51 standard deviation increase group was composed of children with access to in the Z-score whereas in urban areas it was only both food and health only, at five percent. The 0.21 standard deviations. In fact rural children percentage of children with access to each of the with access to adequate environment were on categories with three or four drivers was one average slightly taller than urban children with percent or less. In 2013 the multiple nutrition access to adequate environment, -1.028 and -1.161 driver categories had slightly larger shares. The standard deviations, respectively. Furthermore, largest share was that of children with access to access to adequate environment was not correlated environment and health only at 22 percent and with height in low-stunt districts. In 2013, neither the second largest, at six percent, were children food nor environment was associated with better 22 Both the 2007 and 2013 values are based on definition 2, the definitions used for most of the analyses. 23 However, the gain or increases in child HAZ scores are not directly comparable across years given the different subcomponents and wording in the survey for the four nutrition factors in the different years. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 45 nutrition outcomes in urban areas or for children It is of interest to note that in the older cohort all in low-stunt districts or children from the wealthier sets of nutrition drivers, except for adequate in households. That is, access to adequate food only food, care and health only, yielded average height- was correlated with height for rural children, for for-age scores more than one standard deviation children from high-stunt districts (0.72 standard below zero. That is, even with many of the pro- deviations) and from the bottom 40% of wealth nutrition components present, the distribution of distribution (0.36 standard deviations). Given that heights was to the left of the expected distribution adequate food in 2013 was based on the mother’s by one standard deviation. consumption of fruits and vegetables, it may be 15. In 2013, across the samples studied, access that while in the more vulnerable populations adequate environment and health only was it is correlated with the quantity and quality of consistently associated with taller children. the child’s food intake, in the better off areas or Nationally children with access to both households the relationship may not hold and environment and health were 0.36 standard thus we do not observe any statistically significant deviations taller where the younger cohort had correlations. Access to adequate environment was a slightly smaller point estimate (0.30 standard associated with better nutrition outcomes in rural deviations) than the older cohort (0.41 standard areas and for children from high-stunt districts deviations). For the subpopulations analyzed, the (0.29 standard deviations). correlation ranged from 0.17 standard deviations 13. Access to adequate care alone was rarely in low-stunt or in wealthier households to 0.37 associated with better nutritional outcomes. standard deviations in the younger cohort of rural Only for the national sample in 2007 was there a children. The only subpopulation where it was not statistically significant correlation between access statistically significantly different from zero is in adequate care only and nutrition outcome. For the younger cohort of urban children. none of the subpopulations was the correlation 16. The results for location based subpopulations statistically significant. In 2013 the coefficient showed stronger correlations between nutrition estimate was never statistically significant. drivers and nutrition outcomes than for 14. In 2013, the adequacy measures were more subpopulations based on wealth. This finding significantly correlated with nutrition outcomes suggests that policies based on location, such as for older children (24 months to 35 months) than the district, are probably better than those based for younger children (0 to 23 months). If the on wealth as the drivers associated with outcomes adequacy status of the older cohort has remained in location based populations are similar. constant since birth, the older cohort results 17. There is no systematic evidence of the presence reflect the cumulative associations. The correlation of pairwise synergies among the four underlying coefficient is of similar magnitude for food only determinants of nutrition. That is, in estimating in the two cohorts (0.37 standard deviations Model B, we do not find any statistically for the younger cohort and 0.39 for the older significant positive correlations. These findings cohort), but for adequate environment only it have numerous interpretations. It may be that was about twice as large in the older cohort (0.29 such synergies are not present as having access standard deviations) than in the younger cohort to each component singly is already associated (0.15 standard deviations). In the younger cohort, with sizable gains in the average Z-score. In access to adequate health was not statistically fact, it could be that there is some degree of significantly correlated with nutrition outcomes. substitutability among the drivers. However, it However, in the 24- to 35-month cohort there was is also possible that for the synergies to become a positive correlation (0.28 standard deviations) apparent the adequacy measures need to be more between access to adequate health and nutrition finely defined or that there are synergies among outcomes. In the older cohort, having access certain components of different measures that a to three or four of the adequacies was always more comprehensive measure such as the ones associated with taller children ranging from 0.57 used here are not able to detect. standard deviations to 1.13 standard deviations. Chapter 6. 46 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 47 Chapter 7. Policy Considerations The operationalization of the UNICEF conceptual Context matters: The gain in height for age z-scores framework in this report offers the opportunity of associated with simultaneous access to two or more of serving as a basis for a more systematic monitoring of the four underlying determinants of nutrition, varies the progress in access to the four main drivers of child by the wealth status of the household, and rural and malnutrition. It also serves as a practical diagnostic urban areas. framework for identifying potential “binding constraints” in the Indonesian context towards the Targeting and tailoring: The above also suggests that effort to reduce child stunting and malnutrition. “one size fits all” multi-sectoral programs are not likely to be as effective as multi-sectoral programs that are Measurement drives diagnosis and response: The tailored and targeted to specific age groups, specific analysis carried out highlights numerous critical locations and/or low wealth groups. Investments in data gaps in key components of the four underlying nutrition-specific and nutrition-sensitive interventions determinants of child malnutrition. For example, the should focus on areas where access to adequate 2013 RISKESDAS survey, the one and only survey levels in the four drivers of nutrition is relatively less containing anthropometric measures for children prevalent. and adults at the national as well as at the district level does not include variables useful for quantifying Better coordination and integration: Progress towards key components of food security, such as dietary reducing stunting in Indonesia can be enhanced by diversity and consumption or availability of calories coordinated multi-sectoral interventions that address and proteins. On the other hand, the annual SUSENAS effectively the four key underlying determinants of survey which contains these important variables only nutrition. The poor performance of multi-sectoral at the household level but not at the individual level, projects in nutrition and health across the world contains no information on child anthropometric provides the opportunity to learn a lot from the measures and is relatively weaker in measures of failures of the past, especially when it comes to setting child care. Undoubtedly, the costs of collecting all the the clarity of objectives and the role and responsibility necessary information in one national survey that is of each sector involved in the renewed effort to reduce representative at the district level in a country such stunting Indonesia. Clarity and prior agreement on as Indonesia, are prohibitively high. However, the the common yardstick to be used among the different costs of a more systematic approach to data collection sectors involved can lead to significant improvements that closes the gaps in data collected by nutrition- in the efficiency and efficacy of policies against child related and other socio-economic surveys need to undernutrition. be carefully weighed against the benefits that could accrue to Indonesia’s efforts towards reducing child malnutrition based on more informed and thus better coordinated multisectoral approaches. Chapter 7. 48 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 49 References Alderman, H., Hoogeveen, H., & Rossi, M. (2005). Edmond, KM, et al. (2006). “Delayed breastfeeding “Reducing child malnutrition in Tanzania initiation increases risk of neonatal mortality,” – combined effects of income growth and Pediatrics 117 (3), e380-3386. program interventions,” World Bank Policy Research Working Paper 3567. The World Bank, Elfindri and Gouranga Lal Dasvarma (1996). ‘Child Washington, DC. Malnutrition in Indonesia’, Bulletin of Indonesian Economic Studies 31 (1): 97–111. Barrera, A. (1990): “The Role of Maternal Schooling and Its Interaction with Public Health Programs in FAO (2004). “Human energy requirements. Report Child Health Production,” Journal of Development of Joint FAO/WHO/UNU Expert Consultation,” Economics 32: 69–91. Food and Nutrition Technical Report Series, No. 1. Accessed February 2016 from ftp://ftp.fao.org/ Barrett, C. B. (2009): “Measuring Food Insecurity”, docrep/fao/007/y5686e/y5686e00.pdf. Science, 327, 825(2010) FAO, (2011): Developing A Response Analysis Framework Behrman, J., Deolalikar, A., (1987). “Will developing For Food Security Emergencies, Discussion Paper country nutrition improve with income? A case FAO. Accessed June 6, 2013 from http://www.fao. study for rural south India,” Journal of Political org/fileadmin/templates/tc/tce/pdf/Response_ Economy, 95, 492–507. Analysis_Framework_Discussion_Papers.pdf . Black, R.E., C.G. Victora, S.P. Walker, Z.A. Bhutta, P. Haddad, L., Alderman, H., Appleton, S., Song, L., & Christian, M. de Onis, M. Ezzati, S. Grantham- Yohannes, Y. (2003). “Reducing Child Malnutrition: McGegor, J. Katz, R. Martorell, R. Uauy, and How Far Does Income Growth Take Us?” World the Maternal and Child Nutrition Study Group, Bank Economic Review, 17(1), 107–131. (2013) “Maternal and Child Undernutrition and Overweight in Low-Income and Middle-Income Hoddinott, J., Rosegrant, M., & Torero, M. Countries, The Lancet, 382 (Issue 9890): 427-51. (2012). “Investments to reduce hunger and http://dx.doi.org/10.1016/S0140-6736(13)60937-X undernutrition.” Copenhagen Consensus Challenge Paper, prepared for 2012 Global Copenhagen Bramson, Leslie, Jerry W. Lee, Elizabeth Moore, Consensus Challenge, Copenhagen Consensus Susanne Montgomery, Christine Neish, Khaled Centre, Copenhagen. Bahjri, and Carolyn Lopez Melcher (2010). “Effect of Early Skin-to-Skin Mother—Infant Contact Horton S., Shekar M., McDonald C., Mahal A.; During the First 3 Hours Following Birth on Krystene B. J. (2010): Scaling Up Nutrition: What Exclusive Breastfeeding During the Maternity Will it Cost?. Washington, DC: World Bank. Hospital Stay,” Journal of Human Lactation, 26(2), Accessed June 6, 2013 from http://hdl.handle. 130-137. net/10986/2685. Christiaensen, L., and Alderman, H. (2004). “Child IPKM (2014) I ndeks Pembangunan Kesehatan malnutrition in Ethiopia: Can maternal knowledge Masyarakat 2013. –Jakarta : Badan Penelitian dan augment the role of income?” Economic Pengembangan Kesehatan. 2014 Development and Cultural Change, 52(2), 287–312. Jann, B. (2013). coefplot: Stata module to plot regression Deolalikar, A. (2008). “Malnutrition and hunger.” coefficients and other results. Available from http:// Copenhagen consensus 2008 perspective paper. ideas.repec.org/c/boc/bocode/s457686.html. Copenhagen: Copenhagen Consensus Centre. 50 Ministry of Health, Republic of Indonesia and JICA Skoufias et al (2015). “Synergies in Child Nutrition: (1997). Buku Kesehatan Ibu dan Anak. Jakarta. Interactions of Food Security, Health and Accessed from http://www.depkes.go.id/resources/ Environment, and Child Care.” Washington, DC, download/info-terkini/BUKU%20KIA%20 The World Bank. 20_03%202016.pdf. Strauss, John, and Thomas D., (1995). “Empirical Moore, Elizabeth R., Gene C. Anderson, and Nils Modeling of Household and Family Decisions”, in Bergman (2007). “Early skin-to-skin contact for J. Behrman and T.N. Srinivasan (eds), Handbook mothers and their healthy newborn infants.” of Development Economics, 3A, North Holland, Cochrane Database Syst Rev 3. Amsterdam. Mullany, LC, Katz J, Li YM, Khartry SK, LeClerq SC, Trihono et al. (2015) “Pendek (Stunting) di Indonesia, Darmstadt GL, Tielsch JM (2008). “Breast-feeding Masalah dan Solusinya,” Jakarta: Badan Penelitian patterns, time to initiation, and mortality risk dan Pengembangan Kesehatan. among newborns in southern Nepal,” Journal of Nutrition, 138 (3), 599-603. UNICEF (1990). “Strategy for Improved Nutrition of Children and Women in Developing Countries.” Rajaram, S., Zottarelli, L. K., & Sunil, T. S. (2007). ”Individual, household, programme and community: WHO (2007). “Standards for Maternal and Neonatal Effects on childhood malnutrition in rural India,” Care.” Department of Making Pregnancy Safer. Maternal and Child Nutrition, 2007(3), 129–140. Accessed March 2016 from http://apps.who.int/ iris/bitstream/10665/69735/1/a91272.pdf. Riely, F., Mock, N., Cogill, B., Bailey, L. & Kenefick, E. (1999): “Food Security Indicators and Framework WHO (2008). “Indicators for assessing infant and for Use in the Monitoring and Evaluation of young child feeding practices – part I: definition.” Food Aid Programs”, Food Security and Nutrition Accessed from March 2016 from http:// Monitoring (IMPACT) Project for the U.S. Agency www.who.int/maternal_child_adolescent/ for International Development, Arlington, V A. documents/9789241596664/en/. Accessed on June 6, 2013 from http://www. fantaproject.org/downloads/pdfs/fsindctr.PDF . WHO/UNICEF Joint Monitoring Program for Water Supply and Sanitation (2015). “JMP Green Paper: Sahn, D. and Alderman H. (1997). “On the Global monitoring of water, sanitation and hygiene Determination of Nutrition in Mozambique: post-2025,” Draft, Updated October 2015. Accessed The Importance of Age-Specific Effects,” World on March 24, 2016 from http://www.wssinfo.org/ Development, 25(4): 577-588. fileadmin/user_upload/resources/JMP-Green- Paper-15-Oct-2015.pdf. Shekar M, Ruel-Bergeron J, Herforth A.(2013) Module A. Introduction. In: Improving Nutrition World Bank (2013). “Improving Nutrition Through through Multisectoral Approaches. Washington, Multisectoral Approaches.” Washington, DC, The DC, International Bank for Reconstruction World Bank. and Development, International Development Association of The World Bank, 2013. World Food Programme, (2009). “Emergency Food Security Assessment Handbook.” Rome, WFP, Skoufias, E. (1999). “Parental Education and Child Accessed on June 24, 2013 from http://documents. Nutrition in Indonesia,” Bulletin of Indonesian wfp.org/stellent/groups/public/documents/ Economic Studies, Vol.35, No. 1, (April), pp. 99-119. manual_guide_proced/wfp203244.pdf. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 51 Appendixes 52 Appendix A. High and Low Stunting Districts 2007 LOW-STUNT DISTRICTS HIGH-STUNT DISTRICTS 1172 1407 3175 3315 3575 6372 7325 1101 1213 2104 5206 6110 7306 8201 1203 1408 3209 3372 3576 6402 7372 1102 1214 3204 5207 6203 7314 8202 1217 1505 3211 3373 3577 6406 7403 1103 1215 3278 5272 6204 7315 8203 1273 1571 3216 3401 3578 6471 7406 1104 1216 3313 5301 6205 7316 8272 1277 8271 3272 3471 3671 7104 9107 1106 1271 3316 5304 6206 7371 9103 1301 1706 3273 3501 5102 7105 9171 1110 1304 3317 5305 6210 7404 9105 1308 1771 3275 3503 5103 7172 9401 1112 1309 3329 5306 6211 7407 9106 1310 1805 3275 3503 5103 7172 9401 1113 1601 3520 5307 6301 7502 9108 1311 1871 3275 3503 5103 7172 9401 1116 1602 3527 5310 6303 7603 9410 1371 1872 3277 3506 5171 7209 9404 1117 1610 3528 5312 6306 7604 9416 1373 1903 3279 3514 6104 7304 9411 1173 1701 3601 5314 6308 8102 9417 1375 2171 3301 3516 6171 7307 9419 1201 1707 3604 5315 6311 8103 9418 1377 2172 3302 3525 6172 7311 9420 1202 1802 5202 6101 6407 8104 9426 1404 3171 3309 3571 6208 7313 9471 1205 1804 5203 6102 6408 8105 9427 1405 3173 3311 3572 6212 7317 1210 1808 5203 6105 7203 8107 1406 3174 3312 3573 6310 7322 1211 2102 5204 6108 7305 8171 2013 LOW-STUNT DISTRICTS HIGH-STUNT DISTRICTS 1104 1571 2105 3278 3404 3578 6208 7313 1101 1217 1506 3529 5316 6307 7504 9109 1171 1610 2171 3309 3471 3603 6401 7407 1106 1219 1603 5201 5317 6308 7601 9171 1173 1611 2172 3310 3502 3671 6402 7505 1109 1220 1609 5202 5319 7107 7602 9402 1174 1671 3171 3311 3504 3672 6403 8108 1112 1221 1702 5204 5320 7110 7604 9404 1274 1674 3172 3312 3505 3673 6404 8171 1113 1222 1704 5205 6101 7201 7605 9409 1307 1707 3173 3313 3507 3674 6405 8271 1117 1224 1805 5208 6107 7203 8104 9414 1308 1771 3175 3314 3515 5102 6410 9401 1175 1225 1809 5301 6109 7207 8105 9417 1372 1806 3201 3320 3516 5103 6471 9410 1201 1277 1872 5302 6111 7309 8106 9418 1373 1807 3210 3371 3519 5105 6472 9415 1202 1278 3217 5303 6204 7315 8109 9429 1374 1901 3216 3372 3520 5171 6474 9419 1203 1302 3315 5304 6209 7316 8172 9430 1375 1902 3271 3373 3521 5207 7104 9426 1204 1309 3316 5307 6210 7317 8201 9431 1376 1904 3273 3374 3525 6103 7105 9426 1205 1310 3321 5308 6212 7318 8203 9434 1405 1905 3274 3376 3572 6110 7106 9436 1208 1312 3327 5312 6302 7401 8204 9435 1406 1906 3275 3401 3573 6171 7172 1213 1407 3511 5313 6303 7404 9105 1408 1971 3276 3402 3574 6172 7173 1215 1501 3513 5314 6304 7408 9106 1507 2101 3277 3403 3575 6207 7312 1216 1505 3524 5315 6306 7472 9108 A district is considered to be a high-stunt district if For the 2013 district stunting rates published by the in the particular year it had a stunting rate that was Ministry of Health were used. For 2007 the district in the top 25 percent of the district stunting rates stunting rates were based on the HAZ calculated by and a low-stunt district if in the particular year it the authors. Following are the district codes for the had a stunting rate that was in the bottom 25% of districts assigned to each category. the district stunting rates. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 53 Appendix B. Additional models for RISKESDAS 2013: HAZ and stunting as measures of nutrition Tables B.1 – B.3 presents the regression results depicted in Figures 22 -24. TABLE B.1: Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A NATIONAL NATIONAL, DEFINITION 2 Defn 1 Defn 2 0-23 month olds 24-35 month olds 0.386* 0.383*** 0.365** 0.393** Adequate in: Food only (0.212) (0.141) (0.184) (0.196) -0.059 -0.049 0.302 0.169 Adequate in: Care only (0.106) (0.172) (0.385) (0.172) 0.154** 0.199*** 0.149** 0.290*** Adequate in: Environment only (0.078) (0.054) (0.068) (0.087) 0.072 0.187*** 0.120 0.282** Adequate in: Health only (0.086) (0.063) (0.075) (0.111) 0.072 -0.720** -0.994** -0.216 Adequate in Food and Care only (0.332) (0.311) (0.475) (0.403) 0.171 0.248** 0.225 0.353** Adequate in Food and Environment only (0.155) (0.110) (0.139) (0.163) -0.069 0.049 0.029 -0.104 Adequate in Food and Health only (0.205) (0.136) (0.167) (0.164) 0.083 0.115 0.483* 0.314** Adequate in Care and Environment only (0.111) (0.130) (0.255) (0.144) 0.191 0.156 0.415 0.353* Adequate in Care and Health only (0.116) (0.179) (0.309) (0.214) 0.273*** 0.355*** 0.301*** 0.407*** Adequate in Environment and Health only (0.074) (0.054) (0.065) (0.086) 0.396 0.505** 0.714 0.814*** Adequate in: Food, Care and Environment only (0.244) (0.222) (0.437) (0.228) 0.055 0.245 -1.053 1.129*** Adequate in: Food, Care and Health only (0.282) (0.464) (1.044) (0.379) 0.310*** 0.387*** 0.373 0.742*** Adequate in: Care, Environment and Health only (0.114) (0.142) (0.275) (0.155) 0.609*** 0.548*** 0.506*** 0.572*** Adequate in: Food, Environment and Health (0.140) (0.098) (0.118) (0.160) 0.164 0.251 0.224 0.615*** Adequate in: All Four (0.171) (0.189) (0.341) (0.226) -1.143*** -1.383*** -1.164*** -1.852*** Constant (0.052) (0.038) (0.046) (0.055) Observations 20,034 30,051 20,034 10,017 R-squared 0.005 0.005 0.005 0.012 F-Stat 2.651 5.113 2.973 4.208 Prob>F 0.001 0 0 0 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1. 54 TABLE B.2. Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A RURAL URBAN 0-23 month 24-35 month 0-23 month 24-35 month olds olds olds olds 0.342* 0.542* 0.314 0.099 Adequate in: Food only (0.204) (0.284) (0.327) (0.240) -0.024 0.247 1.030 -0.020 Adequate in: Care only (0.318) (0.242) (0.883) (0.249) 0.194** 0.245** -0.000 0.171 Adequate in: Environment only (0.090) (0.113) (0.117) (0.146) 0.127 0.326*** 0.051 0.136 Adequate in: Health only (0.087) (0.122) (0.144) (0.202) -0.894 -0.173 -1.280* -0.310 Adequate in Food and Care only (0.562) (0.376) (0.689) (1.047) 0.371 0.163 0.033 0.280 Adequate in Food and Environment only (0.242) (0.282) (0.183) (0.215) -0.293 -0.107 0.456 -0.214 Adequate in Food and Health only (0.185) (0.269) (0.297) (0.196) 0.538 0.186 0.326 0.248 Adequate in Care and Environment only (0.435) (0.260) (0.322) (0.181) 0.399 0.153 0.366 0.494* Adequate in Care and Health only (0.453) (0.304) (0.404) (0.296) 0.367*** 0.241** 0.137 0.364** Adequate in Environment and Health only (0.087) (0.108) (0.114) (0.143) 1.615 0.329 0.331 0.817*** Adequate in: Food, Care and Environment only (1.073) (0.332) (0.481) (0.291) 1.763*** 1.257** -1.479 0.891* Adequate in: Food, Care and Health only (0.298) (0.568) (1.062) (0.512) -0.074 0.674*** 0.443 0.627*** Adequate in: Care, Environment and Health only (0.271) (0.226) (0.371) (0.220) 0.518*** 0.727*** 0.358** 0.351* Adequate in: Food, Environment and Health only (0.193) (0.268) (0.160) (0.210) 0.147 0.590 0.104 0.450 Adequate in: All Four (0.630) (0.410) (0.412) (0.281) -1.237*** -1.940*** -0.997*** -1.650*** Constant (0.051) (0.062) (0.098) (0.112) Observations 10,569 5,274 9,465 4,743 R-squared 0.006 0.010 0.004 0.010 F-Stat 4.508 2.029 1.316 1.905 Prob>F 0 0.011 0.183 0.019 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 55 TABLE B.3: Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A STUNTING WEALTH Low High B40 T60 0.195 0.721** 0.362* 0.271 Adequate in: Food only (0.334) (0.285) (0.185) (0.211) -0.099 0.107 0.056 -0.290 Adequate in: Care only (0.379) (0.221) (0.217) (0.283) 0.117 0.291*** 0.001 0.033 Adequate in: Environment only (0.107) (0.107) (0.133) (0.088) 0.082 0.304*** 0.112 0.204* Adequate in: Health only (0.136) (0.115) (0.075) (0.115) -1.115 -0.927* -0.736* -0.819* Adequate in Food and Care only (0.780) (0.525) (0.429) (0.460) 0.315* 0.032 -0.048 0.065 Adequate in Food and Environment only (0.185) (0.250) (0.502) (0.131) -0.079 -0.178 -0.147 0.067 Adequate in Food and Health only (0.248) (0.316) (0.164) (0.205) -0.053 0.524* -0.653** 0.006 Adequate in Care and Environment only (0.194) (0.297) (0.296) (0.153) 0.375 -0.190 -0.075 0.289 Adequate in Care and Health only (0.232) (0.429) (0.268) (0.230) 0.174* 0.294** 0.283** 0.169* Adequate in Environment and Health only (0.105) (0.128) (0.129) (0.087) 0.565* 1.142 0.434 0.315 Adequate in: Food, Care and Environment only (0.338) (0.815) (1.358) (0.232) -0.463 0.219 0.823 -0.252 Adequate in: Food, Care and Health only (0.798) (0.388) (0.568) (0.615) 0.400* -0.053 -0.106 0.216 Adequate in: Care, Environment and Health only (0.230) (0.258) (0.298) (0.161) 0.427*** 0.594* -0.132 0.377*** Adequate in: Food, Environment and Health only (0.158) (0.307) (0.390) (0.119) 0.181 0.299 -0.078 0.058 Adequate in: All Four (0.281) (0.728) (0.488) (0.203) -1.096*** -1.833*** -1.458*** -1.186*** Constant (0.087) (0.061) (0.042) (0.078) Observations 8,201 6,804 11,033 19,018 R-squared 0.005 0.008 0.003 0.003 F-Stat 1.337 1.831 1.529 1.899 Prob>F 0.171 0.026 0.086 0.019 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 As with urban and rural children, there are be more likely statistically significant in the older differences in the correlations between nutrition cohort. Furthermore, when the sample is divided by drivers and nutrition outcomes by age cohort when wealth of the household, especially for the poorer analyzing by the stunting status of the district households, there are very few nutrition drivers that or by household income (Table B.4). The results are correlated with outcomes. underline the fact that the correlations tend to 56 TABLE B.4: Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A Low-stunt district High-stunt district Bottom 40% Top 60% 0-23 24-35 0-23 24-35 0-23 24-35 0-23 24-35 months months months months months months months months 0.355 -0.075 0.684** 0.835 0.390 0.194 0.256 0.351 Adequate in: Food (0.481) (0.195) (0.331) (0.571) (0.238) (0.224) (0.279) (0.297) 0.654 0.067 -0.133 0.553** 0.331 0.251 0.230 -0.091 Adequate in: Care (1.029) (0.331) (0.418) (0.259) (0.437) (0.217) (0.776) (0.252) 0.041 0.290* 0.281** 0.286* 0.078 -0.160 0.028 0.081 Adequate in: Environment (0.140) (0.157) (0.133) (0.160) (0.171) (0.177) (0.113) (0.120) 0.003 0.285 0.212 0.389** 0.024 0.222* 0.197 0.223 Adequate in: Health (0.165) (0.243) (0.139) (0.194) (0.089) (0.118) (0.140) (0.201) -1.447** -0.613 -1.402*** 0.271 -0.882 -0.315 -1.197** -0.302 Adequate in Food and Care (0.600) (1.084) (0.454) (0.831) (0.780) (0.485) (0.478) (0.646) Adequate in Food and 0.256 0.480** 0.113 -0.180 0.483 -0.652 0.084 0.125 Environment (0.240) (0.243) (0.318) (0.366) (0.489) (0.780) (0.166) (0.182) 0.080 -0.345 -0.440 0.427 -0.227 -0.108 0.129 -0.289 Adequate in Food and Health (0.324) (0.226) (0.384) (0.506) (0.201) (0.245) (0.253) (0.218) Adequate in Care and 0.340 0.130 -0.645 1.267*** -1.010* -0.102 0.531* 0.090 Environment (0.407) (0.204) (0.581) (0.316) (0.548) (0.296) (0.282) (0.170) 0.199 0.802*** 0.277 -0.190 0.676 -0.259 -0.001 0.743** Adequate in Care and Health (0.370) (0.293) (0.433) (0.625) (0.454) (0.183) (0.354) (0.291) Adequate in Environment and 0.093 0.335** 0.376** -0.059 0.287* 0.188 0.170 0.156 Health (0.133) (0.156) (0.159) (0.185) (0.161) (0.173) (0.109) (0.116) Adequate in: Food, Care and 1.050* 0.766** 1.720*** 1.371 4.052*** -0.394 0.440 0.602** Environment (0.617) (0.343) (0.346) (0.877) (0.965) (0.790) (0.441) (0.239) Adequate in: Food, Care and -2.581*** 1.162* 0.767 0.435 0.655 1.296** -1.577 0.794* Health (0.841) (0.633) (0.893) (0.358) (0.564) (0.590) (1.157) (0.430) Adequate in: Care, Environment 0.534 0.663*** -0.553 0.562* -1.926*** 0.819*** 0.302 0.470*** and Health (0.458) (0.230) (0.454) (0.300) (0.355) (0.282) (0.293) (0.179) Adequate in: Food, Environment 0.386* 0.519** 0.738** 0.350 -0.170 -0.153 0.395*** 0.326* and Health (0.204) (0.215) (0.357) (0.394) (0.335) (1.022) (0.147) (0.174) -0.142 0.680** 0.696 0.548 -0.028 0.215 0.092 0.347 Adequate in: All Four (0.618) (0.285) (1.036) (0.888) (1.303) (0.169) (0.356) (0.242) -0.905*** -1.533*** -1.617*** -2.268*** -1.219*** -1.943*** -1.028*** -1.579*** Constant (0.115) (0.105) (0.076) (0.088) (0.051) (0.063) (0.099) (0.097) Observations 5,403 2,798 4,525 2,279 7,342 3,691 12,692 6,326 R-squared 0.007 0.015 0.010 0.026 0.005 0.008 0.003 0.007 F-Stat 1.809 2.610 3.650 1.924 4.316 1.841 1.759 1.862 Prob>F 0.028 0.001 0 0.018 0 0.025 0.035 0.022 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 Results using the child’s stunting status as the likely to be stunted. In line with the results with HAZ dependent variable, corroborates the findings score as the dependent variable, there were more when using HAZ as the dependent variable (Tables statistically significant correlations in the older age B.5and B.6). That is children with access to adequate cohort than in the younger cohort. environment only or adequate health only were less Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 57 TABLE B.5. Estimates of the marginal decline in the stunting rate with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A NATIONAL NATIONAL, DEFINITION 2 Defn 1 Defn 2 0-23 month 24-35 month -0.171 -0.276** -0.175 -0.471** Adequate in: Food only (0.199) (0.131) (0.165) (0.221) 0.070 0.045 -0.145 -0.183 Adequate in: Care only (0.097) (0.145) (0.280) (0.177) -0.200*** -0.245*** -0.192*** -0.347*** Adequate in: Environment only (0.071) (0.052) (0.064) (0.092) -0.168** -0.185*** -0.150** -0.216** Adequate in: Health only (0.080) (0.059) (0.070) (0.105) -0.030 0.532 0.472 0.302 Adequate in Food and Care only (0.260) (0.369) (0.601) (0.480) -0.116 -0.306*** -0.131 -0.669*** Adequate in Food and Environment only (0.145) (0.103) (0.125) (0.173) -0.001 -0.157 -0.124 -0.089 Adequate in Food and Health only (0.199) (0.147) (0.174) (0.289) -0.138 -0.195 -0.732*** -0.299* Adequate in Care and Environment only (0.108) (0.137) (0.278) (0.164) -0.087 -0.229 -0.475 -0.393* Adequate in Care and Health only (0.105) (0.167) (0.298) (0.212) -0.356*** -0.439*** -0.361*** -0.565*** Adequate in Environment and Health only (0.070) (0.053) (0.063) (0.091) -0.206 -0.591** -0.641 -0.895*** Adequate in: Food, Care and Environment only (0.207) (0.238) (0.431) (0.281) -0.240 -0.698 0.653 -1.721*** Adequate in: Food, Care and Health only (0.314) (0.471) (0.762) (0.529) -0.302*** -0.516*** -0.373* -0.878*** Adequate in: Care, Environment and Health only (0.103) (0.134) (0.226) (0.170) -0.597*** -0.624*** -0.541*** -0.764*** Adequate in: Food, Environment and Health only (0.129) (0.097) (0.115) (0.172) -0.362* -0.476** -0.495 -0.757*** Adequate in: All Four (0.185) (0.201) (0.365) (0.247) -0.553*** -0.346*** -0.533*** 0.041 Constant (0.047) (0.035) (0.043) (0.061) Observations 20,022 30,036 20,022 10,014 F-Stat 3.299 6.916 3.461 5.509 Prob>F 0 0 0 0 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 58 TABLE B.6. Estimates of the marginal decline in the stunting rate with access to each one of the four underlying determinants of nutrition separately and in combination, 2013. Model A RURAL URBAN 0 - 23 month 24 - 35 month 0 - 23 month 24 - 35 month -0.157 -0.418 -0.084 -0.474 Adequate in: Food only (0.198) (0.265) (0.287) (0.416) -0.268 -0.062 0.147 -0.218 Adequate in: Care only (0.337) (0.220) (0.508) (0.301) -0.200** -0.197* -0.008 -0.255 Adequate in: Environment only (0.083) (0.116) (0.113) (0.158) -0.175** -0.364*** -0.026 0.090 Adequate in: Health only (0.080) (0.121) (0.138) (0.194) 0.269 0.387 1.165 0.104 Adequate in Food and Care only (0.707) (0.628) (0.962) (0.792) -0.250 -0.269 0.117 -0.669*** Adequate in Food and Environment only (0.196) (0.267) (0.172) (0.245) 0.065 0.030 -0.381 -0.061 Adequate in Food and Health only (0.211) (0.370) (0.307) (0.449) -0.465 -0.181 -0.706** -0.184 Adequate in Care and Environment only (0.472) (0.255) (0.331) (0.234) -0.467 -0.085 -0.386 -0.682** Adequate in Care and Health only (0.382) (0.283) (0.481) (0.334) -0.404*** -0.365*** -0.152 -0.510*** Adequate in Environment and Health only (0.084) (0.120) (0.113) (0.155) -1.190 -0.625 -0.298 -0.783** Adequate in: Food, Care and Environment only (1.073) (0.415) (0.488) (0.376) -1.571** 1.097 -1.740** Adequate in: Food, Care and Health only (0.680) (0.795) (0.839) -0.080 -0.868*** -0.332 -0.686*** Adequate in: Care, Environment and Health only (0.400) (0.243) (0.269) (0.246) -0.299 -0.652** -0.416*** -0.565** Adequate in: Food, Environment and Health only (0.183) (0.285) (0.161) (0.230) -0.490 -0.500 -0.286 -0.621* Adequate in: All Four (0.560) (0.391) (0.456) (0.318) -0.432*** 0.157** -0.775*** -0.224* Constant (0.046) (0.068) (0.096) (0.125) Observations 10,560 5,272 9,460 4,742 F-Stat 2.001 2.313 1.517 2.276 Prob>F 0.0144 0.00283 0.0901 0.00342 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 59 TABLE B.7. Estimates of the marginal decline in the stunting rate with access to each one of the four underlying determinants of nutrition separately and in combination, 2013,. Model A STUNTING WEALTH Low High B40 T60 -0.097 -0.395** -0.288* -0.099 Adequate in: Food only (0.327) (0.200) (0.163) (0.206) 0.009 -0.238 0.123 0.031 Adequate in: Care only (0.328) (0.266) (0.167) (0.273) -0.252** -0.291*** -0.068 -0.028 Adequate in: Environment only (0.118) (0.096) (0.108) (0.084) -0.064 -0.263** -0.136* -0.146 Adequate in: Health only (0.137) (0.109) (0.070) (0.106) 0.709 2.181** 1.068** 0.147 Adequate in Food and Care only (0.717) (0.934) (0.536) (0.534) -0.308 -0.043 -0.241 -0.064 Adequate in Food and Environment only (0.201) (0.207) (0.387) (0.124) -0.027 -0.032 -0.045 -0.077 Adequate in Food and Health only (0.301) (0.313) (0.206) (0.210) -0.157 -0.465* 0.739** -0.061 Adequate in Care and Environment only (0.257) (0.280) (0.366) (0.159) -0.897** 0.090 0.229 -0.694** Adequate in Care and Health only (0.349) (0.381) (0.220) (0.272) -0.283** -0.348*** -0.318** -0.206** Adequate in Environment and Health only (0.116) (0.109) (0.132) (0.084) -0.486 -0.393 0.440 -0.404 Adequate in: Food, Care and Environment only (0.418) (0.661) (0.877) (0.255) -0.197 -1.291* -0.957 -0.357 Adequate in: Food, Care and Health only (0.776) (0.719) (0.647) (0.630) -0.420* -0.158 -0.351 -0.280* Adequate in: Care, Environment and Health only (0.226) (0.336) (0.532) (0.152) -0.598*** -0.447 -0.008 -0.404*** Adequate in: Food, Environment and Health only (0.172) (0.288) (0.446) (0.117) -0.409 -0.314 -0.347 -0.230 Adequate in: All Four (0.325) (0.487) (0.866) (0.214) -0.705*** 0.120** -0.253*** -0.593*** Constant (0.095) (0.053) (0.038) (0.073) Observations 8,197 6,800 11,024 19,012 F-Stat 1.585 1.982 1.682 1.899 Prob>F 0.0700 0.0135 0.0473 0.0189 Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 60 Appendix C. An Analysis of Synergies using RISKESDAS 2013 (Model B) The estimates of the mean differences in height-for-age among children in each of the 15 different groups provide only indirect evidence, at best, on the presence of synergies among the four clusters of determinants of nutrition. In an effort to get more direct evidence on the presence of synergies, the following econometric specification (model B) can also be estimated 444 HAZi = α + ∑ βj Aj + ∑ ∑ γjk ( Aj * Ak ) + j=1 j=1 k=j+1 (Model B) 4 4 4 ∑ ∑ ∑ γjkm ( Aj * Ak * Am ) + γ1234 (A1*A2*A3*A4 ) + εi j=1 k=j+1 m=j+2 where HAZi is the Height-for-Age Z-scores for the access (A=0) to cluster A, the expected height-for-age child i, and Ai denotes access to the four adequacies, for when the child does not have adequate access to for each child i. Namely, A1 is 1 when the household any of the four clusters is: 24 is adequate in food and is 0 otherwise; A2 is 1 when the household is adequate in environment E ( HAZi | A1 = 1, A2 = 0, A4 = 0, A4 = 0 ) = α and is 0 otherwise; A3 is 1 when the household is adequate in health and equals 0 otherwise; and, A4 The coefficients βj yield estimates of the increase in is 1 when the household is adequate in care and is 0 the mean HAZ score of children when a child has otherwise. These binary variables are constructed access to adequate levels in one of the clusters only without any consideration to whether the child (or net of the potential gain in HAZ of having access has access to adequate levels in the other three to an adequate level in one or more of the other clusters. It is also important to keep in mind that clusters). That is: there are no additional control variables used in the regression because the objective here is simply E ( HAZi | A1 = 1, A2 = 0, A3 = 0, A4 = 0 ) = α + β1 to compare mean values in HAZ among children in E ( HAZi | A1 = 0, A2 = 1, A3 = 0, A4 = 0 ) = α + β2 these different sub-groups of children defined by the extent to which they have access to one or more E ( HAZi | A1 = 0, A2 = 0, A3 = 1, A4 = 0 ) = α + β3 of the pillars. E ( HAZi | A1 = 0, A2 = 0, A3 = 0, A4 = 1 ) = α + β4 In this specification the constant term α provides Specifically, the coefficient β1 yields an estimate an estimate of the mean value of HAZ scores for of the increase in the mean HAZ score of children children without access to adequate food security (compared to the mean HAZ score of reference (A1= 0), adequate environment (A2= 0), adequate group summarized by the constant term, α) who health (A3= 0) and adequate care (A4= 0). That is, have access to adequate food security only (A1= 1) with E ( HAZ | A=1 or 0) denoting the expected (or but do not have access to adequate environment, mean) value of height-for-age (outcome), conditional (A2= 0), to adequate health (A3= 0) or adequate care on having adequate access (A=1) or inadequate (A4= 0). The coefficients β2 , β3 and β4 have analogous 24 It is also assumed that E ( εi | A1, A2, A3, A4 )=0. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 61 interpretations for environment, health, and care, and care, or food and health, or environment and respectively.25 care, etc.) are similarly defined. The coefficients γjk yield estimates of the synergies The mean HAZ of children from having access to associated with having access to adequate levels three components (i.e. adequate food security (Ai1 = in more than one of the cluster of underlying 1) and adequate environment (Ai2 = 1) and adequate determinants of nutrition. Specifically, the mean health (Ai3 = 1)) is given by the expression HAZ score of children having access to adequate food security (A1= 1) and adequate environment (A2= E (HAZi | A1 = 1, A2 = 1, A3 = 1, A4 = 0) = α + β1 + β2 + β3 1) is summarized by the expression + γ12 + γ13 + γ23 + γ123 E (HAZi | A1 = 1, A2 = 0, A3 = 0, A4 = 0) = α + β1 + β2 + γ12 with the coefficient γ123 summarizing the potential synergies from simultaneous access to the three The expression for the mean value of HAZ scores components. These are synergies in addition to any of children in households with access to adequate synergies from pairwise interactions. And similarly food security and adequate environment consists of the mean HAZ of children from having access to all the sum of three components: the first component four components is given by the expression is the increase in HAZ scores associated with children in households with adequate food security E (HAZi | A1 = 1, A2 = 1, A3 = 1, A4 = 1)= α + β1 + β2 + β3 only (i.e., β1); the second component (i.e., β2) is the +β4 +γ12 + γ13 + γ14 + γ23 + γ24 + γ34 + γ123 + γ124 + γ134 + increase in HAZ scores associated with children in γ234 + γ1234 households with adequate environment only, and (Model B) the third component (i.e., γ12) is the increase in HAZ scores associated with children in households that There were no systematic positive pairwise have access to both adequate food security and synergies in 2013 among the nutrition drivers in adequate environment. Thus the coefficient γ12 the populations considered. That is, although access yields information on whether there are additional to more than one nutrition driver can be associated (extra) gains in HAZ scores derived from having with taller children than those with access to just simultaneous access to adequate food and adequate one, there were no additional gains from access environment. A significant and positive value to two drivers. In fact the drivers tended to be of the coefficient γ12 implies synergies from the substitutes to some degree. Table C.1 presents the simultaneous access to adequate food security and results at the national level, Table C.2 for the urban adequate environment in the production of child and rural samples and Table C.3 for high/low-stunt nutrition. The mean HAZ of children from having districts and by household wealth. access to other two adequacies (for example, food 25 It should be noted that the coefficient estimates from model A are linear combinations of the coefficient estimates in model B. However, if some interaction terms are grouped together, such as including an indicator variable of whether or not the child is adequate in any three clusters, instead of including separately all four 3-way interaction terms, γjkm, then the correspondence between the coefficients in model A and B no longer holds. 62 TABLE C.1. Synergies associated with simultaneous access to two or more of the underlying determinants of nutrition, 2013, Model B NATIONAL NATIONAL, DEFINITION 2 Defn 1 Defn 2 0-23 month 24-35 month -0.255 -1.054*** -1.661*** -0.779 Adequate in Food and Care (0.403) (0.370) (0.630) (0.475) -0.370 -0.334* -0.289 -0.329 Adequate in Food and Environment (0.270) (0.176) (0.227) (0.257) -0.527* -0.522*** -0.455* -0.779*** Adequate in Food and Health (0.296) (0.198) (0.251) (0.269) -0.013 -0.035 0.032 -0.145 Adequate in Care and Environment (0.161) (0.219) (0.466) (0.226) 0.178 0.017 -0.007 -0.099 Adequate in Care and Health (0.160) (0.248) (0.495) (0.275) 0.047 -0.031 0.032 -0.165 Adequate in Environment and Health (0.116) (0.085) (0.103) (0.147) 0.552 1.395*** 1.815** 1.215** Adequate in: Food, Care and Environment (0.470) (0.457) (0.819) (0.557) 0.260 1.283** 0.283 1.940*** Adequate in: Food, Care and Health (0.533) (0.651) (1.261) (0.658) -0.069 0.099 -0.255 0.409 Adequate in: Care, Environment and Health (0.231) (0.322) (0.618) (0.350) 0.847** 0.666*** 0.584* 0.880** Adequate in: Food, Environment and Health (0.367) (0.246) (0.307) (0.358) -1.039 -1.952*** -0.791 -2.668*** Adequate in: All Four (0.637) (0.748) (1.437) (0.785) Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 63 TABLE C.2. Synergies associated with simultaneous access to two or more of the underlying determinants of nutrition, 2013, Model B Urban Rural 0 - 23 month 24 - 35 month 0 - 23 month 24 - 35 month -1.212* -0.962* -2.624** -0.390 Adequate in Food and Care (0.660) (0.522) (1.158) (1.097) -0.165 -0.624 -0.280 0.010 Adequate in Food and Environment (0.325) (0.406) (0.360) (0.316) -0.762*** -0.975** 0.091 -0.450 Adequate in Food and Health (0.276) (0.403) (0.444) (0.338) 0.368 -0.306 -0.704 0.096 Adequate in Care and Environment (0.541) (0.361) (0.944) (0.300) 0.296 -0.420 -0.715 0.378 Adequate in Care and Health (0.555) (0.382) (0.974) (0.395) 0.047 -0.331* 0.087 0.057 Adequate in Environment and Health (0.135) (0.181) (0.172) (0.242) 2.111 1.187* 2.596** 0.850 Adequate in: Food, Care and Environment (1.350) (0.697) (1.302) (1.156) 2.996*** 2.499*** 0.374 1.137 Adequate in: Food, Care and Health (0.855) (0.853) (1.641) (1.262) -1.082 0.913* 0.695 -0.191 Adequate in: Care, Environment and Health (0.762) (0.525) (1.076) (0.475) 0.735* 1.543*** 0.095 0.328 Adequate in: Food, Environment and Health (0.421) (0.568) (0.490) (0.447) -3.825** -3.294*** -0.906 -1.762 Adequate in: All Four (1.624) (1.113) (1.823) (1.369) Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 64 TABLE C.3. Synergies associated with simultaneous access to two or more of the underlying determinants of nutrition, 2013, Model B STUNTING WEALTH Low High B40 T60 -1.211 -1.755*** -1.154** -0.800 Adequate in Food and Care (0.885) (0.631) (0.502) (0.547) 0.003 -0.980** -0.411 -0.239 Adequate in Food and Environment (0.368) (0.393) (0.549) (0.237) -0.356 -1.203*** -0.621** -0.407 Adequate in Food and Health (0.427) (0.429) (0.250) (0.299) -0.071 0.125 -0.710* 0.262 Adequate in Care and Environment (0.425) (0.367) (0.380) (0.316) 0.392 -0.601 -0.243 0.375 Adequate in Care and Health (0.445) (0.495) (0.342) (0.368) -0.026 -0.301* 0.170 -0.068 Adequate in Environment and Health (0.167) (0.181) (0.187) (0.131) 1.631* 2.633** 2.290 1.078* Adequate in: Food, Care and Environment (0.972) (1.106) (1.560) (0.613) 0.534 2.646*** 2.311*** 0.396 Adequate in: Food, Care and Health (1.291) (0.933) (0.817) (0.914) 0.004 0.021 0.508 -0.300 Adequate in: Care, Environment and Health (0.559) (0.635) (0.561) (0.433) 0.411 1.761*** 0.255 0.584* Adequate in: Food, Environment and Health (0.479) (0.588) (0.712) (0.334) -1.426 -3.470** -3.004 -1.040 Adequate in: All Four (1.408) (1.547) (1.839) (0.992) Source: Author calculations based on Riskesdas 2013. Notes: Standard errors in parenthesis. *** p<0.01, ** p<0.05, * p<0.1 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 65 Appendix D: Stunting and the Underlying Determinants of Nutrition using RISKESDAS 2007 THE UNDERLYING DETERMINANTS OF NUTRITION Furthermore, we do not consider exclusive AND HEIGHT-FOR-AGE IN 2007 breastfeeding or mother’s consumption of fruit and vegetables as components of adequate food or This section explores the access to the four hand washing for adequate care. The difference nutrition dimensions and their correlation to between the two definitions is the inclusion of the height-for-age, or nutrition outcomes in 2007. non-smoking component in adequate care in the Furthermore, it discusses any synergies among second set. the nutrition drivers. However, before analyzing the synergy models, it is instructive to explore Based on the first set of definitions all nutrition the prevalence of simultaneous access to the four drivers have prevalence rates less than 40 percent. clusters of determinants. The highest prevalence rate is for adequate care (based on breastfeeding practices) at 37 percent PREVALENCE OF ADEQUATE ACCESS IN 2007 and the lowest prevalence rate is for adequate environment at 15 percent (Figure D.2). Adding the For 2007, the analyses are based on all children additional condition of no smoking by household less than 60 months of age. Table D.1 summarizes head reduces the prevalence of adequate care to the components used for the construction of the ten percent. adequacy measures. Given that the prenatal and post-natal checkups are only available for children The nutrition drivers with large differences less than 12 months, these components are not between children from rural versus urban areas, included in adequate health. Furthermore, due to as well as between children living in high-stunt the very low prevalence of community sanitation, versus low-stunt districts, are access to adequate this component is also excluded from consideration environment. Access to adequate environment is as part of the adequate environment measure. three times greater in urban areas than in rural TABLE D.1. Components of Adequate Access, 2007 COMPONENTS USED Definition ADEQUACY (AGE GROUP APPLICABLE) 1 2 Household meets caloric needs √ √ Household meets protein needs √ √ Food Exclusively breast fed (0-5 month olds) Mother’s consumption of fruit and vegetables Breastfed for 24 months √ √ Exclusively breast fed for 6 months √ √ Care Household head does not smoke √ Mother washes hands with soap after 4 events Immunizations up to date √ √ Health Vitamin A supplementation (7 months and older) √ √ Access to basic sanitation √ √ Environment Access to safely managed drinking water √ √ 66 areas and two times greater in low-stunt districts are three percentage points more likely to have than high-stunt districts (Figure D.2). Furthermore, access than those in the top wealth quintile. That in low-stunt districts children are almost twice as is, with the exception of access to adequate care, likely to have access to adequate health as children wealth quintiles predict access to the underlying in high-stunt districts. determinants of nutrition. Except for adequate care, as the wealth quintile The younger cohort of children is more likely to have increases so does the percentage of children who access to adequate health than the older cohort, but are adequate in the underlying drivers of nutrition. less likely to have access to adequate care. Although Comparing those children in the lowest wealth younger children have had less time to get their quintile with those in the highest wealth quintile, vaccinations, they are more likely to have their there were six, ten, and sixteen percentage point immunization up to date then older children. This differences in access to adequate health, food and result may signal improved outreach to get children environment, respectively (Figure D.2). For access vaccinated. The fact that the older cohort tends to adequate care, those in the lowest wealth quintile to have better access to adequate care may reflect FIGURE D.2. Access to nutrition drivers, 2007 Adequacy definitions 1: Adequacy status in 2007 Percentage of children who meet criteria National, children less than 60 months 35 37 22 15 Adequacy definitions 1: Adequacy status in 2007 Adequacy definitions 1: Adequacy status in 2007 Percentage of children who meet criteria Percentage of children who meet criteria By Urban/Rural, children less than 60 months By district stunting, children less than 60 months 39 36 36 37 35 35 34 35 28 25 27 20 22 15 6 9 Rural Urban High stunt rate Low stunt rate Adequacy definitions 1: Adequacy status in 2007 Adequacy definitions 1: Adequacy status in 2007 Percentage of children who meet criteria Percentage of children who meet criteria By wealth quintile, children less than 60 months By Age, children less than 60 months 38 38 38 36 41 41 34 35 37 35 35 35 31 28 30 22 25 24 18 20 20 22 15 15 14 16 9 12 1 2 3 4 5 0-23 months 24-59 months Source: Author estimates based on 2007 RISKESDAS. Food Care Environment Health Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 67 recall error. The adequate care is based on mother’s determinants and less than one percent had recall of months of exclusive breastfeeding and adequate access to all four drivers. While 32 percent breastfeeding with complementary foods. As the of children in high-stunting districts had access to child ages, it may be more difficult for the mother no nutrition driver, only 24 percent of children in recall the exact age and she may overestimate the low-stunting districts had no access. Similarly, eight length of breastfeeding. percent of children in low-stunting districts had access to three or more drivers whereas only three Using the first set of definitions, 28 percent of percent had such access in high-stunting districts. the children had inadequate levels for any of the By wealth quintile the results are as expected with determinants of nutrition in 2007. The largest share children in the bottom wealth quintile having far of children, or 44 percent, had access to adequate less access than those in the top quintile. There was level in just one of the four determinants (Figure an eleven percentage point gap between the top and D.3). Only five percent of the children nationally bottom quintile in terms of access to no nutrition had access to adequate levels in three of the four driver (Figure D.3). FIGURE D.3. Number of nutrition drivers, 2007 Adequacy definitions 1: Number of adequacies met in 2007 National, children less than 60 months Percentage of children who meet criteria 44 28 23 5 0 Adequacy definitions 1: Number of adequacies met in 2007 Adequacy definitions 1: Number of adequacies met in 2007 By Urban/Rural, children less than 60 months By district stunting, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 45 42 45 42 30 32 27 24 27 21 23 19 7 7 3 0 1 3 0 1 Rural Urban High stunt rate Low stunt rate Adequacy definitions 1: Number of adequacies met in 2007 Adequacy definitions 1: Number of adequacies met in 2007 By wealth quintile, children less than 60 months By Age, children less than 60 months Percentage of children who meet criteria Percentage of children who meet criteria 44 45 43 42 43 44 44 32 29 28 27 25 26 28 27 23 24 21 23 24 20 5 5 6 8 5 3 0 1 1 1 4 0 0 0 1 2 3 4 5 0-23 months 24-59 months Adequate in none Adequate in 1 of 4 Adequate in 2 of 4 Adequate in 3 of 4 Adequate in all 4 Notes: Districts with high (low) stunting rates are the districts with the top (bottom) 25% stunting rates at the district level. Wealth quintiles calculated by the NIHRD using polychoric prinipal components analysis (pp. 61-62 of 2013 report). 68 MULTIVARIATE ANALYSIS OF NUTRITION DRIVERS to any nutrition driver. About ten percentages IN 2007 points fewer children have no access to even one nutrition driver if they are from a low-stunt district, A more detailed analysis of the distribution of from a non-poor household (top 60 percent of the children into the different categories with access to wealth distribution) or from an urban area. adequate level reveals that some of the categories have very few children (Table D.2). Without the In describing the results we first present the total smoking condition in adequate care (Definition differences in the average height-for-age Z-scores 1) three (out of sixteen) categories have less than for the children in the fifteen different adequacy two percent of children. With the inclusion of groups in comparison to those children who are not household head’s smoking behavior in the adequate adequate in any nutrition driver. That is, we present care definition the number of categories with less the results from Model A, where the categorization the two percent of the children rises to seven. It is of children is exclusive, such that they are identified important to keep in mind this uneven distribution in only group. For example, a child adequate in food of children in the various categories when and care would only be identified in the “adequate interpreting the correlations between adequacy in food and care” category. categories and children’s nutritional outcomes. We present results at the national level using both sets At the national level having access to adequate of definitions, but present results using the second food, adequate care, or adequate environment was set of definitions for all subpopulation analyses associated with taller children. The correlation given the importance of smoking in Indonesia. In between the nutrition drivers and the nutritional terms of the subpopulations, the low-stunt and outcome is stable across the two definitions of rural populations appear to be similar in terms adequate care considered. The strongest correlation of the percentage of children with access to the between a nutrition driver and height is with access sixteen adequacy categories. About 44 percent of to adequate environment. Children with access children from high-stunt districts, from poorer to adequate environment were on average 0.35 households, or bottom 40 percent of the wealth standard deviations taller than the 39 percent of distribution, and rural children do not have access children who did not have access to any nutrition TABLE D.2. Distribution of children for Model A, 2007 DEFN 1 DEFN 2 DEFINITION 2 Low- High- National B40 T60 Rural Urban stunt stunt NONE 28% 39% 44% 33% 43% 35% 44% 32% Food 16% 22% 26% 19% 22% 22% 25% 17% Care 16% 4% 5% 3% 4% 4% 5% 3% Environment 5% 7% 3% 10% 5% 8% 3% 12% Health 7% 10% 8% 11% 10% 9% 10% 10% Food and Care 9% 2% 2% 2% 2% 2% 3% 2% Food and Environment 3% 4% 3% 6% 3% 5% 2% 8% Food and Health 4% 5% 4% 6% 5% 6% 5% 5% Care and Environment 2% 1% 0% 1% 1% 1% 0% 1% Care and Health 3% 1% 1% 1% 1% 1% 1% 1% Environment and Health 2% 2% 1% 4% 1% 3% 1% 4% Food, Care and Environment 2% 1% 0% 1% 0% 1% 0% 1% Food, Care and Health 2% 1% 0% 1% 0% 1% 1% 0% Care, Environment and Health 1% 0% 0% 0% 0% 0% 0% 1% Food, Environment and Health 1% 1% 1% 2% 1% 2% 1% 2% In all 4 0% 0% 0% 0% 0% 0% 0% 0% Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 69 driver. Those children who had access to adequate significantly different from those without access food only or adequate care only were about 0.11 to any nutrition driver. The categories with two or standard deviations taller than children without three nutrition drivers that include both care and access to any driver. Appendix B presents the environment no longer are statistically significantly results when instead of HAZ, an indicator for different. That is, once the household head’s stunting is used as the dependent variable. The smoking habits are included there are no statistically results underline the strong positive correlation significant differences in height of those adequate between access to adequate environment and better in both care and environment, or in food, care and nutrition outcomes. environment or in care, environment and health and those not adequate in any nutrition driver. However, In rural areas the correlation between malnutrition those who had access to all four nutrition drivers and adequate environment was greater than in were on average 0.56 standard deviations taller than urban areas and the correlation between adequate those without any access. It is important to note that food and malnutrition was present only in the for most of the coefficient estimates the statistical urban areas. Analyzing the sample separately for significance of the correlations needs to be taken rural and urban children suggests that although with caution, given the small percentage of children on average child without access to any nutrition in any of these categories. driver in rural areas was shorter than the average child in urban areas, with HAZ scores 1.53 and 1.37, For children in high-stunt districts, only access respectively, an average child in rural area with to adequate environment alone was associated access to adequate environment was slightly taller with better nutritional outcomes. The average than an average child in urban area, with HAZ child without access to any nutrition driver in a scores of 1.03 and 1.16, respectively. In fact in urban high-stunt district was 1.06 standard deviations areas, access to adequate food as associated with shorter than the average child without access a similar positive difference in HAZ than access from a low-stunt district (Table D.4). The only to adequate environment. In rural areas access to nutrition dimension associated with better nutrition adequate food alone was not associated with taller outcomes in high-stunt districts was adequate children. That is, even if on average each adult environment which is associated with a 0.44 equivalent household member is consuming at least standard deviation increase in the mean height-for- 1,760 kcal and 44 grams of protein, the children age. It is not clear which are the associated drivers were no taller than children in households where with having access to both care and environment the threshold was not met. (only) that make the group on average 1.23 standard deviations shorter than those without access to Under the first set of definitions, children adequate any nutrition driver. Although the majority of the in environment and at least one other driver were point estimates for groups with access to two or taller than children without access to any driver. An more drivers are positive, they are not statistically average child was at least 0.26 standard deviations significant. Again this may be due to the limited taller than a child without access to any nutrition number of children in each of these categories. driver if they were adequate in environment and one or two other drivers. A child not adequate in As above, for children living in a household in the environment but in a combination of food, health bottom 40 percent of the wealth distribution, access and care was not statistically significantly taller to adequate environment was associated with better than those without access to any dimension. The nutrition outcomes. The difference in the height of only exception was access to only food and care child without access to any driver in the bottom 40 simultaneously where the average child was 0.17 percent of the wealth distribution and the top 60 standard deviations taller than a child without percent of the wealth distribution was 0.21 standard access to any dimension. This difference was slightly deviations. Children from poorer households with greater if the child was adequate in food only or access to adequate environment were 0.34 standard care only. Changing the definition of adequate care deviations taller than children from poorer to include the smoking behavior of the household’s household without such access. This difference head alters which groups of children are statistically is quite similar for children from the wealthier 70 TABLE D.3. Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2007. Model A NATIONAL URBAN RURAL Defn 1 Defn 2 Definitions 2 0.110** 0.104** 0.213*** 0.054 Adequate Food (0.048) (0.041) (0.071) (0.050) 0.110** 0.130* 0.202 0.093 Adequate Care (0.043) (0.079) (0.127) (0.099) 0.344*** 0.356*** 0.212*** 0.506*** Adequate Environment (0.070) (0.062) (0.077) (0.117) -0.011 -0.027 0.045 -0.090 Adequate Health (0.055) (0.049) (0.079) (0.061) 0.166*** 0.280** 0.227 0.303** Adequate in Food and Care (0.056) (0.115) (0.195) (0.143) 0.341*** 0.370*** 0.348*** 0.187 Adequate in Food and Environment (0.092) (0.081) (0.102) (0.130) 0.077 -0.017 0.003 -0.049 Adequate in Food and Health (0.069) (0.061) (0.092) (0.081) 0.260*** 0.107 -0.031 0.210 Adequate in Care and Environment (0.099) (0.174) (0.212) (0.306) -0.060 -0.079 0.032 -0.150 Adequate in Care and Health (0.073) (0.137) (0.262) (0.153) 0.398*** 0.399*** 0.335** 0.334** Adequate in Environment and Health (0.110) (0.104) (0.131) (0.165) 0.431*** 0.214 -0.021 0.555** Adequate in: Food, Care and Environment (0.111) (0.166) (0.205) (0.272) -0.092 0.142 -0.206 0.321 Adequate in: Food, Care and Health (0.087) (0.167) (0.249) (0.216) 0.421** 0.388 0.312 0.344 Adequate in: Care, Environment and Health (0.164) (0.253) (0.293) (0.509) 0.255** 0.212* 0.179 0.057 Adequate in: Food, Environment and Health (0.124) (0.117) (0.138) (0.232) 0.216 0.559* 0.357 0.876 Adequate in: All Four (0.191) (0.286) (0.320) (0.607) -1.506*** -1.482*** -1.373*** -1.534*** Constant (0.028) (0.024) (0.039) (0.031) Observations 34,006 31,698 12,346 19,352 R-squared 0.003 0.003 0.004 0.003 F-Stat 4.954 4.978 2.000 2.826 Prob>F 0 0 0.012 0 Source: Author calculations based on Riskesdas 2007. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 71 TABLE D.4. Estimates of the marginal increase in mean HAZ score with access to each one of the four underlying determinants of nutrition separately and in combination, 2007. Model A DISTRICT STUNTING WEALTH Low High B40 T60 0.157* 0.027 0.091 0.096* Adequate Food (0.082) (0.079) (0.056) (0.058) 0.136 -0.143 0.173 0.068 Adequate Care (0.197) (0.139) (0.107) (0.116) -0.068 0.442*** 0.344*** 0.303*** Adequate Environment -0.113 (0.147) (0.089) (0.085) -0.193** -0.113 -0.010 -0.060 Adequate Health (0.095) (0.109) (0.065) (0.071) 0.346* -0.054 0.322** 0.208 Adequate in Food and Care (0.196) (0.224) (0.151) (0.172) 0.217 0.196 0.228* 0.380*** Adequate in Food and Environment (0.139) (0.168) (0.126) (0.104) -0.187 0.037 0.012 -0.085 Adequate in Food and Health (0.118) (0.148) (0.088) (0.082) 0.113 -1.236** -0.073 0.136 Adequate in Care and Environment (0.345) (0.489) (0.319) (0.201) 0.214 0.272 -0.198 0.031 Adequate in Care and Health (0.347) (0.270) (0.166) (0.213) 0.224 0.379 0.263 0.407*** Adequate in Environment and Health (0.191) (0.321) (0.182) (0.126) -0.250 0.505 0.633** -0.107 Adequate in: Food, Care and Environment (0.256) (0.526) (0.314) (0.203) -0.321 0.164 -0.084 0.273 Adequate in: Food, Care and Health (0.310) (0.364) (0.249) (0.220) -0.213 0.489 0.261 0.401 Adequate in: Care, Environment and Health (0.482) (0.329) (0.416) (0.316) -0.178 0.119 0.173 0.140 Adequate in: Food, Environment and Health (0.179) (0.388) (0.211) (0.141) 0.326 0.715 0.811 0.357 Adequate in: All Four (0.387) (1.005) (0.533) (0.332) -0.873*** -1.935*** -1.573*** -1.360*** Constant (0.049) (0.048) (0.032) (0.035) Observations 7,526 7,553 16,424 15,274 R-squared 0.005 0.004 0.003 0.004 F-Stat 1.819 1.713 2.175 2.874 Prob>F 0.027 0.042 0.005 0 Source: Author calculations based on Riskesdas 2007. 72 households where the difference was 0.30 standard children separately. Although access to some deviations. However, unlike for the subsample of combinations of two or more nutrition drivers children from high-stunt districts, children from are associated with taller children there are no households in the bottom 40 percent of the wealth significant positive synergies among the nutrition distribution are taller if they have simultaneous drivers. That is, in evaluating Model B we do not access to adequate food and care (0.32 standard find any positive synergies between nutrition deviations), or food and environment (0.23 standard drivers. These results are presented in Table D.5. deviations), or food, care and environment (0.63 standard deviations). Access to health care alone or That is, being adequate in more than one nutrition with other nutrition drivers was not associated with driver is not associated with taller children beyond taller children. any additive differences from being adequate in more than one driver. In fact there are some The estimates provide no evidence of positive negative correlations in the synergy variables, synergies amongst the nutrition drivers, neither indicating that being adequate in both care and for the sample as a whole nor for urban and rural health (under the first set of adequacy definitions) TABLE D.5. Synergies associated with simultaneous access to two or more of the underlying determinants of nutrition, 2007. Model B NATIONAL URBAN RURAL Defn 1 Defn 2 Definitions 2 -0.053 0.046 -0.188 0.156 Adequate in Food and Care (0.076) (0.140) (0.240) (0.172) -0.113 -0.090 -0.077 -0.373** Adequate in Food and Environment (0.115) (0.101) (0.133) (0.171) -0.022 -0.094 -0.255* -0.013 Adequate in Food and Health (0.092) (0.080) (0.130) (0.103) -0.194 -0.379* -0.445* -0.389 Adequate in Care and Environment (0.119) (0.195) (0.252) (0.327) -0.159* -0.181 -0.215 -0.152 Adequate in Care and Health (0.094) (0.163) (0.300) (0.188) 0.065 0.070 0.078 -0.082 Adequate in Environment and Health (0.134) (0.125) (0.160) (0.206) 0.228 0.046 0.062 0.508 Adequate in: Food, Care and Environment (0.191) (0.290) (0.397) (0.455) -0.066 0.165 -0.008 0.274 Adequate in: Food, Care and Health (0.152) (0.264) (0.451) (0.322) 0.266 0.419 0.435 0.458 Adequate in: Care, Environment and Health (0.230) (0.344) (0.468) (0.601) -0.117 -0.107 -0.037 0.056 Adequate in: Food, Environment and Health (0.211) (0.197) (0.252) (0.344) -0.171 0.101 0.336 -0.130 Adequate in: All Four (0.364) (0.535) (0.704) (0.969) Source: Author calculations based on Riskesdas 2007 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 73 or in care and environment (under adequacy positive synergies. For children in high-stunt definitions 2) children are not as tall as expected districts access to both care and health was if contributions from each driver were additive. associated with positive synergies but the synergy In fact one can think of these drivers as being just countered the negative correlations between complementary to some degree. access to adequate care only and access to adequate health only such that a child with access to both Although for the considered sub-groups of children, was as tall as a child without access to any nutrition some adequacies are positively correlated with driver. Similarly the synergies on food care and height-for-age, there are no systematic synergies environment, and care, environment and health among the four drivers of nutrition. Homogenizing counter some of the negative coefficients and lower the groups by district characteristics yields some level synergies. TABLE D.6. Synergies associated with simultaneous access to two or more of the underlying determinants of nutrition, 2007. Definition 2 STUNTING WEALTH Low High B40 T60 0.053 0.062 0.058 0.044 Adequate in Food and Care (0.283) (0.274) (0.191) (0.202) 0.128 -0.272 -0.207 -0.020 Adequate in Food and Environment (0.178) (0.217) (0.156) (0.132) -0.152 0.123 -0.069 -0.121 Adequate in Food and Health (0.160) (0.187) (0.114) (0.114) 0.045 -1.535*** -0.589* -0.235 Adequate in Care and Environment (0.406) (0.521) (0.346) (0.240) 0.271 0.529* -0.361* 0.022 Adequate in Care and Health (0.412) (0.316) (0.204) (0.250) 0.485** 0.051 -0.071 0.163 Adequate in Environment and Health (0.231) (0.359) (0.204) (0.158) -0.701 1.924** 0.764 -0.363 Adequate in: Food, Care and Environment (0.522) (0.774) (0.510) (0.372) -0.595 -0.321 0.034 0.223 Adequate in: Food, Care and Health (0.573) (0.538) (0.362) (0.376) -0.890 1.259* 0.776 0.139 Adequate in: Care, Environment and Health (0.705) (0.741) (0.581) (0.471) -0.537 -0.138 0.095 -0.222 Adequate in: Food, Environment and Health (0.336) (0.566) (0.320) (0.249) 2.185** -1.179 -0.216 0.318 Adequate in: All Four (0.941) (1.475) (0.936) (0.695) Source: Author calculations based on Riskesdas 2007. 74 MULTIVARIATE ANALYSIS OF NUTRITION DRIVERS another nutrition driver is associated with better IN 2007 – STUNTING AS DEPENDENT VARIABLE nutrition outcomes. With the inclusion of the non-smoking household head condition, access to When stunting is used as the dependent both care and environment no longer is statistically variable, instead of height-for-age Z-scores, the significantly associated with stunting. However, results are similar (Table D.7). There is a strong with the second set of definitions, access to any positive correlation between access to adequate three or all four are all associated with lower environment and non-stunting. Again under the probability of stunting. first set of definitions access to environment and TABLE D.7. Estimates of the marginal decline in the stunting rate with access to each one of the four underlying determinants of nutrition separately and in combination, 2007. Model A NATIONAL URBAN RURAL Defn 1 Defn 2 Definitions 2 -0.026 -0.048 -0.145** -0.007 Adequate Food (0.042) (0.036) (0.068) (0.042) -0.055 -0.070 -0.180 -0.018 Adequate Care (0.040) (0.068) (0.127) (0.082) -0.426*** -0.457*** -0.246*** -0.625*** Adequate Environment (0.066) (0.059) (0.077) (0.105) 0.004 -0.006 -0.016 0.028 Adequate Health (0.052) (0.047) (0.079) (0.058) -0.112** -0.140 -0.172 -0.114 Adequate in Food and Care (0.050) (0.094) (0.171) (0.113) -0.345*** -0.337*** -0.199** -0.307** Adequate in Food and Environment (0.088) (0.075) (0.097) (0.123) -0.124* -0.077 -0.098 -0.032 Adequate in Food and Health (0.066) (0.060) (0.105) (0.074) -0.281*** -0.169 0.040 -0.304 Adequate in Care and Environment (0.091) (0.158) (0.201) (0.259) -0.007 -0.032 0.156 -0.121 Adequate in Care and Health (0.072) (0.132) (0.225) (0.163) -0.309*** -0.355*** -0.214* -0.336** Adequate in Environment and Health (0.108) (0.099) (0.125) (0.165) -0.463*** -0.364** -0.101 -0.644** Adequate in: Food, Care and Environment (0.106) (0.172) (0.210) (0.307) -0.033 -0.285* -0.006 -0.423** Adequate in: Food, Care and Health (0.096) (0.172) (0.292) (0.213) -0.453*** -0.438* -0.187 -0.736 Adequate in: Care, Environment and Health (0.155) (0.252) (0.294) (0.500) -0.489*** -0.427*** -0.339** -0.267 Adequate in: Food, Environment and Health (0.134) (0.126) (0.155) (0.221) -0.321 -0.719** -0.512 -0.873 Adequate in: All Four (0.199) (0.316) (0.372) (0.604) -0.237*** -0.243*** -0.425*** -0.157*** Constant (0.025) (0.022) (0.041) (0.027) Observations 33,953 31,649 12,329 19,320 F-Stat 6.496 6.707 1.486 3.815 Prob>F 0 0 0.101 0 Source: Author calculations based on Riskesdas 2007 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 75 Appendix E: An Analysis of Synergies using SUSENAS 2007 and 2013 For the following components (Food, Care, Health, and Environment) of adequacy, the individual dummies are collapsed at district level. Then the threshold is set at median value of all districts. District adequacy dummy equals 1 if mean of that particular district is above the median. DISTRICT ADEQUACY DUMMY DEFINITION Children who are: 1. Under 6 months: are exclusively breastfed 2. Between 6-23 months: are breastfed or formula-fed and having a Dietary Diversity Score (DDS)* of 4 or greater (see below for DDS definition) 3. Between 24-59 months: having a DDS of 3 and above Adequate in: Food However, the downside of this measure is: this measure only uses total consumption data, instead of calories intake (because SUSENAS 2007 and 2013 do not provide such data). *See below for definition of Dietary Diversity Score Children who are: 1. Under 6 months: are exclusively breastfed 2. Above 6 months up to 59 months: are continued breastfed up to 24 months Adequate care excludes requirement for complementary feedings (solid/soft food) in this definition since the consumption of food data is HH consumption data and Adequate in: Care hence assuming all HH member gets the same amount of consumption. This will then overestimate the number of children who are adequate in food and in care (as food consumption is part of each definition). In the end, all children will be assumed to have consumed solid food in the HH since there must have been some food eaten in the household by someone. Therefore, there will be no variation in the indicator. Children under 59 months who get complete vaccination* as required schedule and their births were assisted by health professionals. Adequate in: Health *See below for complete vaccination schedule Children in a HH who have access to safe water and safe sanitation and 75% HH in the community have access to adequate safe sanitation. The definition of safe Adequate in: Environment (GoI - water and safe sanitation are based on GoI definition*. Including Community Sanitation) *See below for definition of access to safe water and sanitation Children in a HH who have access to safe water and safe sanitation and 75% HH in the community have access to adequate safe sanitation. The definition of safe Adequate in: Environment (JMP - water and safe sanitation are based on JMP definition*. Including Community Sanitation) *See below for definition of access to safe water and sanitation 76 VARIABLE DEFINITION Districts with both Food and Care adequacy mean are above median, Adequate in: Food and Care irrespective of their mean value in Health and Environment adequacy. Adequate in: Food and Districts with both Food and Environment adequacy mean are above median, Environment irrespective of their mean value in Health and Environment adequacy. Districts with both Food and Health adequacy mean are above median, Adequate in: Food and Health irrespective of their mean value in Care and Environment) adequacy. Adequate in: Care and Districts with both Care and Environment adequacy mean are above median, Environment irrespective of their mean value in Food and Health adequacy. Districts with both Care and Health adequacy mean are above median, Adequate in: Care and Health irrespective of their mean value in Food and Environment adequacy. Adequate in: Environment and Districts with both Environment and Health adequacy mean are above median, Health irrespective of their mean value in Food and Care adequacy. Adequate in: Food, Care and Districts with Food, Care and Environment adequacy mean are above median, Environment irrespective of their mean value in Health adequacy. Adequate in: Food, Environment Districts with Food, Environment and Health adequacy mean are above median, and Health irrespective of their mean value in Care adequacy. Adequate in: Food, Care and Districts with Food, Care and Health adequacy mean are above median, Health irrespective of their mean value in Environment adequacy. Adequate in: Care, Environment Districts with Care, Environment and Health adequacy mean are above median, and Health irrespective of their mean value in Food adequacy. Districts with only Food adequacy mean that is above median; while their Care, Adequate in: Food only Health, and Environment adequacy mean should not be above median value. Districts with only Care adequacy mean that is above median; while their Food, Adequate in: Care only Health and Environment adequacy mean should not be above median value. Districts with only Environment adequacy mean that is above median; while Adequate in: Environment only their Care, Health and Food adequacy mean should not be above median value. Districts with only Health adequacy mean that is above median; while their Adequate in: Health only Care, Food and Environment adequacy mean should not be above median value. Districts with both Food and Care adequacy mean are above median; while their Adequate in: Food and Care only Health and Environment adequacy mean should not be above median value. Adequate in: Food and Districts with both Food and Environment adequacy mean are above median; Environment only while their Health and Care adequacy mean should not be above median value. Adequate in: Food and Health Districts with both Food and Health adequacy mean are above median; while only their Care and Environment adequacy mean should not be above median value. Adequate in: Care and Districts with both Care and Environment adequacy mean are above median; Environment only while their Food and Health adequacy mean should not be above median value. Districts with both Care and Health adequacy mean are above median; while Adequate in: Care and Health only their Food and Environment adequacy mean should not be above median value. Adequate in: Environment and Districts with both Environment and Health adequacy mean are above median; Health only while their Food and Care adequacy mean should not be above median value. Adequate in: Food, Care and Districts with Food, Care and Environment adequacy mean are above median; Environment only while their Health adequacy mean should not be above median value. Adequate in: Food, Environment Districts with Food, Environment and Health adequacy mean are above median; and Health only while their Care adequacy mean should not be above median value. Adequate in: Food, Care and Districts with Food, Care and Health adequacy mean are above median; while Health only their Environment adequacy mean should not be above median value. Adequate in: Care, Environment Districts with Care, Environment and Health adequacy mean are above median; and Health only while their Food adequacy mean should not be above median value. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 77 VARIABLE DEFINITION Children who are adequate in all four components: Care, Environment, Health Adequate in: All Four and Food Exclusive breastfeeding is when a child under 6 months of age is only fed Exclusive Breastfeeding breastmilk. Continued breastfeeding is when a child under 24 months of age is still Continued Breastfeeding breastfed. The complete vaccine/immunization schedule is based on national schedule, gathered from Buku Kesehatan Ibu dan Anak (Buku KIA) as follows: • BCG: at 1 month Complete Vaccines • DPT: at 2, 3, 4 months • Polio: at 1, 2, 3, 4 months • Measles: at 9 months • HepB: at birth, 2, 3, 4 months old An assisted birth is when the child was born in the presence of doctors, or Assisted Birth midwives, or other trained health professionals. DDS is a measure of the nutritional quality of the food consumed. In SUSENAS 2007 and 2013, we can only use total HH consumption as a proxy of DDS— because SUSENAS in these years do not have information on HH calories intake. Dietary Diversity Score (DDS) First off, the HH food consumption during the past week is grouped into: (1) grains, roots and tubers; (2) legumes and nuts; (3) dairy products; (4) flesh foods including organ meats; (5) eggs; (6) Vitamin A rich fruits and vegetables including orange and yellow vegetables; (7) and other fruits. For every child, the sum of these food groups will reflect the child’s DDS which then will determine if the child meets minimum acceptable diet. A. SUSENAS 2007 Access to safe drinking water is defined as follows, when • HH source of drinking water is either from piped meter/piped retail/rain water • HH source of drinking water is either from borehole/pump/protected well/ protected spring and the distance to nearest feces containment > 10 meters B. SUSENAS 2013 Access to safe drinking water is defined as follows, when • HH source of drinking water is either from piped meter/piped retail/rain Access to Safe Water (GoI) water • HH source of drinking water is either from borehole/pump/protected well/ protected spring and the distance to nearest feces containment > 10 meters • HH source of drinking water is either from borehole/pump/protected well/ protected spring and the distance to nearest feces containment <= 10 meters and source of bathing/washing water is either from piped meter/piped retail/borehole/pump/protected well/protected spring/rain water • HH source of drinking water is from bottled water/unprotected well/ unprotected spring/river water/other and source of bathing/washing water is either from piped meter/piped retail/borehole/pump/protected well/ protected spring/rain water 78 VARIABLE DEFINITION Access to safe drinking water is defined as follows: when HH source of drinking Access to Safe Water (JMP) water is either from piped meter/piped retail/bottled water/borehole/pump/ protected well or spring/rain water Access to safe sanitation is defined as follows, when • Type of closet HH used is goose neck Access to Safe Sanitation (GoI) • HH uses either a private or shared defecation facility • Final disposal site is in a tank/septic tank 18) SUSENAS 2007 Access to safe sanitation is defined as follows, when • Defecation final disposal is located in a tank/septic tank, while type of closet HH used is either goose neck/pit toilet/squat toilet/none • Final disposal site is on the ground Access to Safe Sanitation (JMP) 19) SUSENAS 2013 • Access to safe sanitation is defined as follows, when • Type of closet HH used is either goose neck/pit toilet/squat toilet; or when HH does not use closet but the defecation facility is privately used • Final disposal site is located in a tank/septic tank; or when final disposal site is located on the ground but HH uses a goose neck closet For the community to have improved sanitation, at least 75% of HH in the Community Toilet 75% community must have improved sanitation. In SUSENAS, the ‘community’ is defined as a census block. Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 79 TABLE E.1. Correlation of District-level Stunting Rates with Simultaneous Access to Adequacies, SUSENAS 2007 and 2013. Model A 2007 2013 VARIABLES GoI JMP GOI JMP -0.5 -0.6 2.2** 1.8* Adequate in: Food only (1.0) (1.2) (1.0) (1.0) 1.8** 2.2** 3.0*** 0.6 Adequate in: Care only (0.9) (0.9) (0.9) (1.0) -1.0 -1.2 0.9 1.6** Adequate in: Environment only (0.9) (1.0) (0.8) (0.8) -0.9 -0.7 -0.4 -0.1 Adequate in: Health only (0.9) (0.9) (0.9) (0.9) -0.3 -0.4 2.2** 3.5*** Adequate in: Food and Care only (0.9) (1.0) (0.9) (1.0) -1.3 -3.4*** -2.8*** -2.4*** Adequate in: Food and Environment only (1.0) (1.2) (1.0) (0.9) -1.9* -1.2 -1.1 1.1 Adequate in: Food and Health only (1.0) (1.0) (1.0) (0.9) 0.3 -0.6 0.5 2.8*** Adequate in: Care and Environment only (1.9) (1.3) (1.1) (0.9) -1.8 -0.1 3.8*** 3.3* Adequate in: Care and Health only (2.6) (3.0) (1.4) (1.8) -2.2** -1.4 0.8 -0.9 Adequate in: Environment and Health only (1.0) (0.9) (0.9) (0.9) -4.1*** -3.1*** -1.1 -1.4* Adequate in: Food, Care and Environment only (0.9) (1.0) (0.9) (0.8) -1.4 -2.4** 1.4 0.0 Adequate in: Food, Environment and Health only (1.1) (1.1) (1.2) (1.1) -1.4 -0.7 0.3 1.6* Adequate in: Food, Care and Health only (0.9) (0.9) (0.9) (0.9) 1.8 -0.2 -1.7 1.7 Adequate in: Care, Environment and Health only (4.3) (3.2) (1.8) (1.3) -0.0 -0.5 -4.0*** -3.8*** Adequate in: All Four (0.9) (0.9) (1.1) (1.1) 44.6*** 45.3*** 38.8*** 37.6*** Constant (1.6) (1.8) (1.6) (1.5) Observations 438 438 497 497 R-squared 0.2 0.2 0.2 0.3 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 80 TABLE E.2. Stunting Rates and Synergies in Simultaneous Access to Adequacies, SUSENAS 2007 and 2013. Model B 2007 2013 VARIABLES GoI JMP GOI JMP -2.3*** -1.9** -3.0*** -3.0*** Adequate in: Food only (0.8) (0.8) (0.8) (0.8) -1.3 -1.8 3.6* 3.5* Adequate in: Care only (2.8) (2.7) (1.9) (1.8) 1.9 -4.5** -4.2* -5.6*** Adequate in: Environment only (3.5) (2.3) (2.4) (1.8) -1.6 -1.4 4.9* 7.1** Adequate in: Health only (3.7) (3.7) (2.8) (2.8) 1.8 1.4 -1.0 -0.3 Adequate in: Food and Care only (2.8) (2.8) (2.0) (1.9) -3.8 0.2 -0.6 -0.8 Adequate in: Food and Environment only (3.5) (2.3) (2.4) (1.8) -0.2 -0.2 -6.4** -6.6** Adequate in: Food and Health only (3.7) (3.6) (2.8) (2.9) -2.6 -4.0 7.1** -0.3 Adequate in: Care and Environment only (3.5) (5.9) (3.6) (2.7) 3.8 3.1 -1.7 -3.4 Adequate in: Care and Health only (6.3) (6.2) (2.9) (3.0) -4.2 0.2 5.2 0.4 Adequate in: Environment and Health only (2.6) (2.4) (3.6) (2.8) -1.2 1.5 -7.3** -1.1 Adequate in: Food, Care and Environment only (3.4) (5.9) (3.6) (2.7) 2.5 -2.1 -3.0 -0.2 Adequate in: Food, Environment and Health only (2.5) (2.5) (3.7) (2.9) -5.4 -4.3 1.6 2.5 Adequate in: Food, Care and Health only (6.3) (6.3) (2.9) (3.0) 1.5 1.2 -3.6 5.9* Adequate in: Care, Environment and Health only (1.1) (1.3) (6.3) (3.1) - - -0.7 -8.0*** Adequate in: All Four (6.2) (2.9) 43.8*** 44.5*** 43.9*** 44.3*** Constant (0.8) (0.8) (0.7) (0.7) Observations 438 438 497 497 R-squared 0.2 0.2 0.2 0.3 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Operationalizing a Multi-Sectoral Approach for the Reduction of Stunting 81