39690 Document of The World Bank Report No. 39690-IND HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES JED FRIEDMAN PETER F. HEY WOOD GEOFF MARKS FADIA SAADAH YOONJOUNG CHOI JANUARY 20, 2006 Health, Nutrition, and Population Unit East Asia and Pacific Region The International Bank for Reconstruction and Development Washington, D.C. HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES TABLE OF CONTENTS I. INTRODUCTION 1 II. ACCOMPLISHMENTS AND REMAINING NUTRITION CHALLENGES FOR INDONESIA 4 III. MULTI-DIMENSIONAL DIVERSITY IN INDONESIA'S NUTRITION CHALLENGES 8 DISTRICT LEVEL DIVERSITY IN NUTRITION OUTCOMES 8 PROVINCE LEVEL DIVERSITY IN NUTRITION OUTCOMES 11 PROVINCE LEVEL DIVERSITY IN NUTRITION SERVICE UTILIZATION 12 IV. WHICH TYPES OF NUTRITION PROGRAMS ARE MOST COST-EFFECTIVE? 12 V. ASSESSING EXISTING INSTITUTIONAL ARRANGEMENTS OF NUTRITION SERVICE DELIVERY 15 VI. WHERE TO FROM HERE. 16 REFERENCES 18 ANNEX A. OVERVIEW OF NUTRITIONAL SITUATION IN INDONESIA -- OUTCOMES AND SERVICES 19 ANNEX B. REGIONAL HETEROGENEITY IN NUTRITION OUTCOMES AND SERVICES 39 ANNEX C. COST-EFFECTIVENESS OF NUTRITION PROGRAMS 53 ANNEX D. INSTITUTIONAL ANALYSIS: VARIATION IN NUTRITIONAL CAPACITIES ACROSS REGIONS 62 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES iv I. INTRODUCTION Malnutrition, which occurs to varying degrees in all provinces of Indonesia, is a significant and preventable risk factor affecting the quality of human resources. The effects of malnutrition range in severity from growth retardation, reduced resistance to illness, and learning impairment to severe disability and early mortality. In the last 25 years, Indonesia has made considerable progress through its efforts to reduce malnutrition. Nevertheless, certain nutrition problems persist through today. In the past, Indonesia has mainly operated centrally designed and managed large scale nutrition projects. Some of the successful examples include the expansion of the Posyandu in the 1970s and the creation of the Bidan Di Desa program in the 1990s. Both of these programs led to gains in the nutritional status of children. Now, however, the national Ministry of Health no longer plays the role of front line provider of nutritional services. Instead it serves as an advisor to districts and provinces as the local governments now engage the local nutritional situations found throughout Indonesia. Against the backdrop of gradual improvements in population nutrition, Indonesia's recent program of political decentralization poses new opportunities and challenges for the health sector in general and nutrition delivery in particular. There is considerable diversity in nutritional outcomes across provinces, and at district level within provinces, driven by local factors as well as average socioeconomic status. Under decentralization, local governments have become the focal point for health care provision. This shift is an opportunity to make public spending more responsive to the varying local conditions of malnutrition and disease. However, the process of decentralization faces challenges from institutional issues evident in the past implementation of nutrition processes. These issues include the lack of coordination and suitability in government structures and bodies, inadequate skills of district level staff, and insufficient planning devoted to the process. The movement to decentralized, local-level programs may also be associated with the loss of economies of scale. There is a tension between the need for nutrition programs to address increasing regional disparities and the possibility that local governments may not have adequate capacities or resources to recognize and address their local nutritional issues. 1 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES This policy concept paper is intended to assist the center navigate the tension between opportunities and challenges as activities are adapted to the decentralized national nutrition policy, and to help guide districts and provinces in the conduct of locally appropriate nutrition programs. This paper synthesizes the findings of an extensive study undertaken by the World Bank and presented in four annexes to this report. The annexes provide extensive data and analysis to shed light on the opportunities and challenges in the new institutional environment. Annex A describes the current state of nutrition outcomes and service utilization in Indonesia today through the analysis of recent large scale household surveys. It indicates that whilst there has been gradual improvement in several national-level indicators, at the disaggregated level achievements have been unequal. Specifically, nutrition outcomes are poorer for individuals with lower socioeconomic and educational status. Annex B also makes use of data from the large scale household surveys, and is the first of two annexes analyzing local diversity. It focuses on the diversity in nutrition outcomes and service use across Indonesian districts and provinces. Annex C begins with a review of the costs and effectiveness of existing nutrition programs such as growth monitoring and promotion, supplementary feeding, and vitamin A, iron, and iodine supplementation programs. The review of cost-effectiveness is extended to a careful analysis of the substantial regional heterogeneity in both the costs and effectiveness of delivery. The results inAnnexes B and C are important for the decentralization program since regions vary in the nutritional conditions of local populations and the ability to service these populations. Consequently the cost effectiveness of these programs will vary within Indonesia. Annex D focuses on the challenges inherent in the institutional setting as it exists today, as well as the demands of the new decentralized service delivery environment. It concludes by discussing how the existing setting can be adapted to better serve the new environment. The main sections of the paper below present a summary of the results and conclusions discussed in detail in the annexes. It begins with a review of the accomplishments and remaining nutrition challenges for Indonesia. It then turns to look at the regional diversity in Indonesia's nutrition challenges and asks which type of nutrition programs are most cost-effective. It concludes with an assessment of the existing institutional arrangement for nutrition service delivery, and discusses steps Indonesia can take to further improve population nutrition and health. 2 II. ACCOMPLISHMENTS AND REMAINING NUTRITION CHALLENGES FOR INDONESIA Nutritional status in Indonesia has improved markedly over the past decades. In particular, Indonesia has achieved a large decrease in protein-energy malnutrition. Figure 1 below shows that the prevalence of underweight among children under-5 years was 27% in 2001, a reduction by about one third from prevalence in the late 1980s (Marks 2003). In addition, Indonesia is one of the first developing countries to identify micronutrient deficiency problems and launch successful micronutrient intervention programs (GOI 2003). By the early 1990s, severe vitamin A deficiency (VAD) declined to a level where it was no longer a public health problem (Helen Keller International, 1998). The total goiter rate (TGR) also declined from 28% in 1990 to 10% in 1998. In 14 iodine deficiency disorder (IDD) endemic districts, TGR decreased further from 44% in 1996-98 to 25% in 2003 (World Bank 2004). Thanks to improved availability of iodized salt, in part due to the IDD project, Indonesia now has the highest coverage of iodine fortified salt use in South East Asia (UNICEF 2005). Figure 1. Trend of underweight prevalence among children under-5 years, Indonesia 45 40 ) %( ecne 35 al evr P 30 25 1985 1990 1995 2000 2005 Year (Source: WHO Global Database on Child Growth and Malnutrition, SUSENAS 2001) However, Indonesia still faces many challenges. Continuing improvement in nutrition has been threatened by the economic crisis in 1997/98, declining resources for nutrition, and increasing diversity in food intake. Indeed there has recently been a slight upsurge in the percentage of children underweight, partly due to the income shocks of the recent crisis. More than a quarter of children are still underweight, and about 33% and 11% of children are stunted (a long-term malnutrition indicator) or wasted (a short-term malnutrition indicator) in 2000, respectively. Regional comparisons show that prevalence of underweight and stunting in Indonesia is roughly similar with those in Myanmar, Philippines, and Viet Nam (UNICEF 2005). However, Indonesia has a high wasting prevalence relative to its under-five mortality rate in comparison (Figure 3), implying a more urgent problem of short-term malnutrition. 3 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Figure 3. Prevalence of wasting and Under-five mortality rate for selected Southeast Asian countries, 2000* 20 200 Wasting U5MR %)( 15 150 ec en al 10 100 ev pr )0001rep( gnit R M 5 50 U5 was 0 0 Cambodia Indonesia Myanmar Philippines Viet Nam (SOURCE: UNICEF http://www.childinfo.org, accessed on March 18 2005, and Indonesian Family Life Surveys 2000) * The wasting estimate for the Philippines is from year 2001. Efforts to promote exclusive breastfeeding appear to be flagging. The health benefits of breastfeeding for both mother and child are well documented, and WHO and UNICEF recommend that infants are exclusively breastfed for the first 6 months of life (WHO 2001). However, only 38% of children under-6 months of age had exclusive breastfeeding in Indonesia (BPS and ORC Macro 2003). Furthermore, exclusive breastfeeding rates have stagnated or decreased steadily since the late 1980s (Figure 4). Improvements in child nutrition can be realized through the continued promotion of exclusive breastfeeding and Indonesia should further utilize the village midwife program to carry out such promotion (Frankenburg 2003). Figure 4. Trends of exclusive breastfeeding rates by age group 100 1987 1991 1994 1997 2002 (%) 80 60 breastfeeding 40 20 Exclusive 0 0-1 2-3 4-5 Age (month) (Source: ORC Macro, 2005. MEASURE DHS STATcompiler: http://www.measuredhs.com, accessed on March 18 2005) 4 Chronic energy deficiency (CED) among women is still a prevalent problem in Indonesia.Among women between 15 and 50 years, about 14% are estimated to have CED1 in 2000. In particular, CED is more prevalent among younger women between 15 and 30, who are responsible for about 70% of child births (BPS and ORC Macro 2003), and is possibly an important risk factor for low birthweight. At the same time, overweight and associated chronic diseases have increased in Indonesia. About 21% of women between 15 and 50 were estimated to be overweight in 2000. The overweight prevalence is substantially higher for those in the 30s and 40s and for those in urban areas, indicating an emerging need for chronic disease prevention in these populations2. For micronutrients, about 19% of women 15-49 years and 53% of children 1-4 years still suffer from anemia3. In particular, nearly 70% of children between 12-23 months were estimated to have anemia. In terms of subclinical VAD, no national-level data are available. Severe clinical VAD is minimal in Indonesia, and population-level data on subclinical VAD is needed to properly monitor and evaluate vitamin A supplementation programs. Iodine deficiency remains prevalent in some parts of the country. With respect to micronutrient programs, 80% of pregnant women received antenatal iron supplementation; 43% of post-partum women and 75% of children 6 to 59 months received vitamin A supplementation; and 85% of households consumed iodine fortified salt4 (BPS and ORC Macro 2003). Beyond national-level accomplishments and challenges, nutritional outcomes and service utilization vary greatly by socioeconomic background. Assuring equity of opportunity is a key role for government and inequalities in nutrition outcomes and service utilization will hopefully motivate new policy approaches. Women and children in socio-economically disadvantaged groups -- those living in a rural area or a poor household, and with less education -- have poorer outcomes and lower program utilization, indicating greater nutritional needs.5 Figure 5 shows the wealth gradient in selected nutritional outcomes, holding various demographic and socioeconomic characteristics constant. Women and children from the poorest quintile households are more likely to be malnourished -- for both micronutrients and energy -- compared to those in the three middle wealth quintiles, whereas women and children from the richest quintile are better off than those in the middle group. 1Measured by Body Mass Index below 18.5 2The national health and household surveys conducted in 1995 and 2001 reported increasing cardiovascular disease-specific mortality rates among adults (GOI 2003). 3Source: Indonesian Family Life Survey 2000. Anemia refers to serum hemoglobin level below 11 g/dl (12g/dl for pregnant women). 4Iodized salt coverage was estimated using SUSENAS 2003. 5Two exceptions to the rural/poor disadvantage are exclusive breastfeeding and overweight prevalence. 5 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Figure 5. Relative odds of selected nutritional problems by household wealth status 2 1.8 poorest quintile middle richest quintile 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Anemia among Night blindness CEM among Stunting among Wasting among women among pregnant women children under-5 children under-5 women Results from multivariate analyses (Annex A). The reference group is the three middle quintiles. All disparities are statistically significant, except odds of night blindness for the richest quintile and odds of wasting among the poorest quintile. Inequalities in service utilization by educational attainment are also evident (Figure 6). Compared to those who completed junior high school or more, women and children in lower educated groups are less likely to use a program even after controlling for household resources. Given the observed socioeconomic gradients, strategies targeting increases in service utilization and nutritional intake among the less educated and/or the poor are essential in order to sustain nutritional improvement in Indonesia. Figure 6. Relative odds of service utilization by educational attainment 1.4 none primary junior high or more 1.2 1 0.8 0.6 0.4 0.2 0 Iron supplement Vitamin A Vitamin A Iodized salt Posyandu visit among pregnant supplement supplement, (adequately among children women among post among chlidren fortified) under-5 partum women 6-59 months use among households Results from multivariate analyses (Annex A). The reference group is junior high school or more. All disparities are statistically significant. Mother's educational status Household head's educational status 6 III. MULTI-DIMENSIONAL DIVERSITY IN INDONESIA'S NUTRITION CHALLENGES As previously mentioned, Indonesia's recent program of political decentralization now poses new challenges and presents new opportunities for the health sector in general and nutrition delivery in particular. The share of local governments in total public health spending increased from 10% prior to decentralization to 50% in 2001. This shift in fiscal responsibility will hopefully make public spending more responsive to the local conditions of malnutrition and disease. However it may also cause increasing regional disparities if local governments do not have adequate capacities or resources to recognize and address local nutritional issues. Measuring and assessing district-level heterogeneity is a pre-requisite for guiding and coordinating local nutrition policies at the central and provincial level, and this study uses existing data sets to accomplish this task. District Level Diversity in Nutrition Outcomes Nutrition conditions vary widely across districts in Indonesia. A first step to inform nutrition policy makers at a local level is to understand variation in nutritional outcomes and service utilization across districts and provinces. This section presents the magnitude and patterns of this regional variation seen in large scale social surveys. Indonesia has 33 provinces and more than 440 districts and municipalities. There is significant variation in nutritional status across districts. For instance, the district-level prevalence of underweight6 among children under-5 years ranged from 3% to 81%, while the national average was about 27% (Figure 7). In addition, although about 13% of women were estimated to have CED7, this prevalence varied from 0% to 60% at the district-level.About 34% of districts had CED prevalence of 15% or higher. Figure 7. Distribution of district-level prevalence of children 0-4 years underweight (%) .1 8 .0 st icrt 6 isdfo .0 n iotroporp 4 .0 2 .0 0 0 25 50 75 100 Prevelance of children underweighted (%) (Source: SUSENAS 2001. Red line indicates the national average) 6Children whose weight-for-age is less than 2 standard deviation below median of a standard international distribution 7Women whose Middle Upper Arm Circumference is less than 23cm. 7 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Similarly, although the national iodized salt coverage rate is about 82%, and 58% of districts reached coverage 90% or higher, coverage at the district level varied greatly from 17% to 100%. Variation is much greater for adequately fortified iodized salt use: compared to the national average of 66%, the district- level coverage ranged from 9% to 100%. About 21% of districts had coverage below 50%, while 24% of districts had coverage of 90% or higher (Figure 8). In addition, most provinces with low utilization of iodized salt are concentrated in eastern Java Island, Bali, and southern Sulawesi, where small-scale salt farming is prevalent. This indicates the need for interventions targeted to small-scale producers of salt or reorganization of the salt industry. Figure 8. Distributions of district-level coverage of iodine fortified salt (%) a) Iodized salt b) Adequately fortified iodized salt .3 .3 5 5 .2 .2 stcirt st .2 icrt .2 isdfo isdfo 5 5 n .1 n .1 oi rtoporp .1 iotroporp .1 5 5 .0 .0 0 0 0 25 50 75 100 0 25 50 75 100 household coverage (%) household coverage (%) (Source: SUSENAS 2001. Red line indicates the national average) A majority of this variation in nutritional outcomes and program utilization is explained by the socioeconomic characteristics of each district (See Table B2,Annex B, for more information on multivariate analyses). Districts with a higher percent of households living under the official poverty line were more likely to have a higher malnutrition prevalence among women and children, even after controlling for other district-level socioeconomic conditions such as income inequality, education, and urban/rural status (Figure 9). 8 Figure 9. Predicted* district-level prevalence of underweight among children 0-4 years, by region and district poverty level (10th percentile, median, and 90th percentile of the district poverty distribution) 45 10th percentile median 90th percentile 40 35 ) 30 %( 25 ecne 20 al 15 evr P 10 5 0 Java & Bali Sumatra Nusa Kalimantan Sulawesi Tenggara, Maluku, & Irian Jaya (Source: SUSENAS 2001) Poverty level refers to percent of population living under the poverty line. Poverty levels are 1%, 6%, and 20% for the 10th percentile, 50th percentile, and 90th percentile of the district-level distribution, respectively. * Unweighted average of district-level mean was used for other variables (education, GINI coefficient, and residential area). More information on district-level multivariate analyses is available in Table B2 (Annex B). ** Differences by poverty level are statistically significant. *** Differences between Sumatra and Java & Bali are not significant. Predicted probabilities in Nusa Tenggara, Maluku, and Irian Jaya, Kalimantan, and Sulawesi are higher than that in Java & Bali at a significant level. Households in urban districts are more likely to consume adequately fortified iodized salt than households in rural districts, even after controlling for socioeconomic characteristics (Figure 10). Districts with a higher proportion of adults who have completed primary education also had higher coverage of iodized salt use. However, even within similar socioeconomic level, districts tend to perform differently depending on the region. For example, districts in Kalimantan, Sulawesi, Nusa Tenggara, Maluku, and Irian Jaya had higher prevalence of underweight among children than districts in Java and Bali (Figure 9). Districts in Nusa Tenggara, Maluku, and Irian Jaya had lower coverage of iodized salt than those in Java and Bali (Figure 10). 9 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Figure 10. Predicted* district-level coverage of adequately fortified iodized salt among households, by region and district type 100 80 ) %( 60 Rural agerev 40 Urban co 20 0 Java & Bali Sumatra Nusa Kalimantan Sulaw esi Tenggara, Maluku, & Irian Jaya (Source: SUSENAS 2001) * Unweighted average of district-level mean was used for other variables (poverty level, GINI coefficient, and education). More information on district-level multivariate analyses is available in Table B2 (Annex B). ** Differences by residential area are statistically significant. *** Differences between districts in Java & Bali and districts in Nusa Tenggara, Maluku, and Irian Jaya are statistically significant. Province Level Diversity in Nutrition Outcomes Substantial variation in nutritional outcomes exists at the province-level as well. Relative differences across provinces exceed two fold for the prevalence of underweight, stunting, and wasting among children 0-4 years. The prevalence of wasting -- an indicator of short-term malnutrition -- ranged from 3% in Bali to 16% in South Sumatra8. Anemia prevalence among women also varied from 12% in West Sumatra to 32% in South Sumatra9. The majority of provincial variation in most nutritional outcomes can again be explained by socioeconomic characteristics across provinces. The prevalence of both energy malnutrition and anemia increase with the percent of households living under the poverty line (Figure 11-12). 8 Among 13 provinces included in the Indonesian Family Life Survey 2000 9 Among 13 provinces included in the Indonesian Family Life Survey 2000 10 Figure 11. Malnutrition prevalence and poverty at the province-level a) CED among women 15-49 years b) Underweight among children 0-4 years 40 50 40 30 ) ) %( %( 30 ec 20 ecne enlave al evr 20 P Pr 10 10 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Population living under poverty (%) Population living under poverty (%) Solid line is a fitted line from a simple linear regression (Source: SUSENAS 2001) CED refers to women whose Middle Upper Arm Circumference less than 23 cm Underweight refers to children whose weight-for-age is less than 2 standard deviation below median of a standard international distribution Figure 12. Anemia prevalence and poverty at the province-level a) Anemia among women 15-49 years b) Anemia among children 1-4 years 40 80 30 70 ) ) %( %( ecne ecne 20 60 al al evr evr P P 10 50 0 40 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Population living under poverty (%) Population living under poverty (%) Solid line is a fitted line from a simple linear regression (Source: Indonesian Family Life Survey 2000) Anemia refers to serum hemoglobin level below 11 g/dl (below 12/g/dlf or pregnant women) Province Level Diversity in Nutrition Service Utilization Nutrition service utilization also varies across provinces.Antenatal iron supplementation coverage ranged from 58% in Central Kalimantan to 98% in DI Yogyakarta. Vitamin A supplementation among children 6 to 59 months varied from 51% to 80%. Adequately iodized salt use at home ranged from 19% in NTB to 96% in Jambi. However, the relationship between service utilization and provincial poverty levels varies 11 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES by program. Antenatal iron supplement coverage is negatively correlated with the provincial poverty level. For vitaminAsupplementation, however, there was no clear coverage pattern by poverty indicating vitamin A supplementation programs were implemented to a similar extent across provincial poverty levels (Figure 13). In addition, as observed in the district-level, there was no clear relationship between Posyandu use among children 0-4 years and poverty at the province-level (result not shown). Finally, iodized salt use did not show a linear relationship with provincial poverty, but the scatter plot showed a rather unique convex shape, indicating relatively lower coverage in provinces with both low and high poverty (result not shown). Figure 13. Vitamin A supplement program coverage and poverty at the province-level 0 10 ) 80 %( ega er ovctne 60 children 6-59 months postpartumwomen melppus 40 A ni matiV 20 0 0 5 10 15 20 25 30 Percent of population living under poverty (Source: Demographic Health Surveys 2002/3) Given the significant amount of variation in nutritional outcomes and service utilization across districts and provinces, a one size fits all approach to the nutrition policy will most likely not address the diverse problems in each district or province. More individually tailored programs for each local area (either a district, a group of districts, or province) based on local conditions should now be the model with decentralization. However, ensuring equity across the nation will be the challenge for decentralization. The analysis presented here confirms that the resource poor regions are those with the greatest nutritional needs. If the goal is to raise the nutritional status of every Indonesian, the central government may consider supporting socio-economically disadvantaged areas with more resources and/or technical support. IV. WHICH TYPES OF NUTRITION PROGRAMS ARE MOST COST-EFFECTIVE? The determinants of the most effective use of health funds are an important input into policy design. The cost-effectiveness review summarized here pays careful attention to the substantial regional heterogeneity in both the costs and effectiveness of delivery and to the determinants of such heterogeneity. Regions vary in the nutritional conditions of local populations as demonstrated above. This is a major cause for the observed heterogeneity in program cost effectiveness. Another important factor is district capacity 12 for implementing nutrition policy. Recognizing and understanding this heterogeneity in capacity is an important step in determining the effective conduct of policy in diverse settings. Estimating the costs of actual nutrition interventions is a necessary first step for the assessment of cost- effectiveness. To get a better sense of the types and the range of costs borne by nutrition interventions, this study fielded and analyzed a cost survey of existing nutrition programs and activities. An activity- based costing framework was used for the survey design. Five districts were purposively sampled and visited in this study -- one in the province of Lampung, one in Yogyakarta, two in East Java, and one in East Nusa Tenggara. The survey focused on five main nutrition activities currently undertaken by the Government of Indonesia: (1) Child growth monitoring and promotion, (2) Supplementary feeding, (3) VitaminAsupplementation, (4) Iron supplementation, and (5) Iodine supplementation. Information was collected on the cost of implementing selected nutritional programs in each district. Respondent sources included staff members from both Posyandu and Puskesmas facilities in each district, selected at random. In addition, each District Health Department was visited by the surveyors as well as the Provincial Health Departments in the relevant provincial capitals and the National Department of Health in Jakarta. The annual cost per recipient, child or woman -- also known as total unit cost -- in each of the five districts is summarized in Table 1. This cost includes variable, fixed, off-budget, and 'extra activity' components (see Table C3, Annex C). The annual cost per recipient varies substantially across the five districts. This should not be surprising since many factors that determine program costs vary across districts, such as the overall local price levels, the size and density of the catchment area served by the program, and the manner in which the program is managed. Table 1. Annual cost per recipient for the five selected programs, per person/year (in Rupiah) Program Kota Surabaya Lumajang Kupang Gunung Kidul Lampung Selatan Vitamin A supplementation Infants 6-11months 2057 2513 1317 2315 1540 Children 12-59 months 1886 2455 1360 2894 3241 Post-partum mothers 3467 4603 2910 3298 3103 Iron supplementation Pregnant women 3139 3518 2899 2714 2804 Iodine supplementation School age children - 1300 - 1200 - Growth monitoring Children under 5 years 24220 22126 15317 20171 18473 Therapeutic feeding (Puskesmas) Children under 5 years 257049 215256 408234 191452 - Complementary feeding (Posyandu) Children under 5 years 183264 188305 221882 207521 - International estimates show that nutrition programs are amongst the most cost-effective health interventions available (World Bank 1993). However both the costs and effectiveness of any programs can vary widely according to the health conditions in a specific location and how the programs are 13 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES implemented. Because the district is now the focal point in Indonesia for decisions on health priorities, which programs are to be implemented and how these are supported, it is important to consider the cost- effectiveness (CE) of nutrition programs at the district-level. We use the cost information for the five districts described above to estimate CE for the nutrition programs at the district-level. Cost-effectiveness is measured in terms of Rupiah per DALY averted and summarized in Table 2 (Details of this approach are given in Annex C). Table 2: Estimated cost effectiveness of selected the nutrition interventions at district-level (in '000 Rupiahs/DALY averted) District Intervention Target Population Kota Lumajang Kupang Gunung Lampung Surabaya Kidul Selatan Vitamin A supplementation Children 6-59 months 158 235 126 232 306 Iron supplementation Pregnant women 74 94 82 63 78 Growth monitoring & Children 6-12 months 10536 10916 15466 11817 - complementary feeding The range in CE across districts within intervention type is greatest for Vitamin Asupplementation, with a ratio of 2.44:1 for the highest incremental cost per DALY averted to the lowest (235 in Lampung Selatan vs. 126 in Kupang). The corresponding ranges for iron supplementation and growth monitoring/ complementary feeding are 1.50 and 1.47 respectively. Surabaya ranks as the first or second most cost- effective district across the three interventions, that is the least cost per DALY averted, partly because Surabaya is an urban area and the population density reduces the cost of serving the catchment area. The other districts rank quite differently depending on the intervention. For example, Kupang is the most CE for vitamin A, but near the least CE for the others. The three interventions are quite distinct in their relative CEs in the five districts, with no overlap between the ranges and a clear ranking in CE - iron supplementation is clearly the most CE, followed by vitamin Asupplementation, then growth monitoring/ complementary feeding, in a ratio of 1:2.7:155.8. Compared to growth monitoring and complementary feeding, the two micronutrient supplementation programs are significantly more cost-effective. Due to varying methods of cost estimation, there are few other data with can be directly compared with the estimated absolute dollar amount per DALY averted10. However, WHO also estimated that vitamin Aand iron supplementation programs were far more effective than the growth monitoring and complementary feeding program in the SoutheastAsia region: 128, 230, and 3478 I$/DALY averted, respectively (WHO-CHOICE 2003)11. 10Our cost survey only focused on the marginal cost of implementing a specific intervention. However WHO's costs included the cost of an overall health center visit. In addition, WHO included the cost of 3-month postpartum iron supplementation for iron supplementation while we restricted to cost within the antenatal period only (WHO-CHOICE 2003). 11International dollar (I$) is based on year 2000, and WHO's estimates are based on program coverage comparable to what observed in the five districts. 14 V. ASSESSING EXISTING INSTITUTIONAL ARRANGEMENTS OF NUTRITION SERVICE DELIVERY Decentralization has raised significant questions regarding the relationships between different levels of government in Indonesia and the roles of key institutions and agencies in nutrition delivery. This section summarizes a systematic study, conducted in the spring of 2005, of the divisions of roles and responsibilities between levels of government and other institutions involved in nutrition program delivery. The work aims to provide a comparative institutional analysis across administrative levels and regions. Four districts from three provinces were purposively sampled for these case studies. They are Kota Surabaya and Lumajang in East Java, Gunung Kidul in Yogyakarta, and Kota Kupang in East Nusa Tenggara (seeAnnex D for details). These are among the same districts selected for the CE exercise and so will provide complementary information. Our review identified five important institutional issues limiting Indonesia's response to nutrition problems. First, government structures and processes are unsuited to tackling nutrition problems in a large and diverse country in which the type and degree of nutrition problems also varies widely across districts. The most critical issue is the contested authority between the various levels of government in thewakeoftheinitialdecentralizationanditscontinuingmodification.Theresultofthiscontestedauthority is considerable variation across districts with no clear relationship between the local structure and the nature and extent of local nutrition problems. While this variation is particularly noticeable across districts, it is also apparent across provinces. Additionally, the distribution of responsibilities within districts, and between districts and provinces, has become opaque or overlapping. Some districts still preserve an explicit nutrition section operating under the DHO; some have merged the nutrition section into more broadly defined health sections; and some do not have any nutrition related section or sub-section at all.Acombination of these three varieties can be seen across the 38 districts in East Java. On the other hand all 16 districts in NTT have a nutrition section. These opaque and overlapping responsibilities are, in part, due to the lack of a coordinating body, and each district or province has created redundant or duplicate roles and responsibilities. For example both the Bappeda and the SKPE in NTT are responsible for the collection of nutrition data. In addition, no section or sub-section covers all the roles and responsibilities necessary for nutrition policy and service delivery. For instance, health surveillance (including nutrition) is under provincial authority12, but few PHOs have a defined role for surveillance. These problems result in structures and staffing levels that are not clearly related to the nutrition and health problems of individual districts. In addition there currently appears to be limited possibility to make changes to provide approaches and structures more aligned to addressing the local issues. The second institutional issue regards the overall low skill levels of district level staff. There is a mismatch between the required skills, especially for program planning and evaluation, and those available at the district and provincial levels. District staff who previously were expected to merely follow the central 15 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES instructions are now, suddenly, expected to plan, implement and evaluate nutrition programs, tasks for which they are ill-prepared. Equally important, there is an almost complete absence of in-service training for staff at all levels, especially at the district and provincial levels. Third, as a result of the low level of human resources devoted to nutrition and the skills mismatch for those staff that are available, planning and implementation of nutrition programs is inadequate. Crucially, the lack of monitoring and evaluation at all levels means that the evidence base for program planning is very thin. The result is that the overall coverage and effectiveness of most nutrition programs is low. Fourth, there are limited resources for nutrition programs. This occurs partly due to delays in the release of funds under new budget processes and partly to the low priority accorded to nutrition by district administrators and parliaments. There are indications that these resource issues may be, worse in those areas with the greatest nutrition problems, though more analysis is required on this subject. And fifth, there is substantial variation in the collaboration with groups outside government at the district level. Collaboration at the central level is good, especially with national companies (e.g. Kimia Pharma, Indo Pharma and Gizindo) and national NGOs, especially the Indonesian Coalition for Fortification. Collaboration at the district and provincial levels is much more limited, a notable exception being the close coordination between the LPKS and Kota Surabaya. 12Based on PP No. 25/2000 16 VI. WHERE TO FROM HERE As the first part of this paper shows, Indonesia has done well in controlling nutrition problems to date. For example, in the last decades protein-energy malnutrition has been reduced by a third, vitamin A deficiency has been controlled to the point where it is no longer a public health problem, and there have been important reductions in the extent of iodine deficiency.At the same time challenges remain -- protein- energy deficiency is still important, anemia is widespread, important pockets of iodine deficiency remain, and new problems associated with lifestyle and nutrition are now important in some groups. With significant old and new nutrition challenges remaining, the nutrition situation in Indonesia is characterized by considerable diversity. This diversity is manifest in variation between districts in the level of malnutrition and considerable variation between regions and with socioeconomic status and maternal education. But there are also equal levels of diversity between districts in program implementation, coverage and impact. And finally, there is considerable diversity in the way in which districts are now organized to combat malnutrition. One of the important results of this multi-dimensional diversity is variation in the cost-effectiveness of the main nutrition control programs at the district level. In recent years Indonesia has undergone significant administrative changes, with decentralization of major resources and responsibilities to the district level being the most important institutional change in the last 50 years. However, contrary to expectations, decentralization has not so far resulted in local governments and programs that are more responsive to local conditions and variations in malnutrition. Further, there has been a loss in the ability to deal with inter-district issues. These are the same as the issues which affect all parts of the health system. The challenge now is to address these remaining nutrition problems within the context of the new, decentralized Indonesia. The wide variation between districts in nutrition problems, their solutions and in the cost-effectiveness of current interventions, calls for locally differentiated approaches. The new decentralized structure contains the potential for the flexibility needed. To realize this potential the government now needs to: · Reform government structures and processes so that they are suited to tackling nutrition in a large and diverse country. This will involve delineation of clear institutional roles at each level of government where the provinces and the center specialize in critical public nutrition functions and districts assume primary responsibility for nutrition sector performance in their jurisdictions using local staffing structures and interventions that are responsive to the local nutrition problems and capacity; · Augment human resources so that there is a close match between the required skills and those available, particularly at the district and provincial levels. This will involve not only rationalization of staffing structures and roles, but also attention to pre-service and in-service training to ensure that the skills required for the various roles are available; 17 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES · Strengthen planning and implementation of nutrition programs. Apart from a renewed emphasis on training, this will involve collaboration with other levels of government, a range of government departments, universities and NGOs; · Ensure adequate financial resources for nutrition programs, especially in the worst affected areas. Apart from continuing reform of government budget processes at all levels, this will involve active advocacy with the district authorities for increased resources for nutrition in accordance with a locally relevant evidence base that includes the nature and extent of the local problems and the local effectiveness of interventions; · Promote collaboration at all levels with groups outside government in the design, delivery and evaluation of nutrition programs. 18 REFERENCES Badan Pusat Statistik (BPS), ORC Macro. 2003. Indonesia Demographic and Health Survey 2002-2003. Calverton, Maryland, USA: BPS and ORC Macro. Dijkhuizen, M.A., F.K. Wieringa, C.E. West, Muhilal Muherdiyantiningsih. 2001. ìCurrent Micronutrient Deficiencies in Lactating Mothers and their Infants in Indonesia.î American Journal of Clinical Nutrition 73:786-91 Frankenberg, E., W. Suriastini, D. Thomas. 2003. ìCan ExpandingAccess to Basic Health Care Improve Children's Health Status? Lessons from Indonesia's 'Midwife in the Village' Program.î Department of Sociology, University of California at Los Angeles. Mimeo. Frankenberg, E. 2004. ìIndonesia's Posyandu Program: Trends in Use and Quality.î Department of Sociology, University of California at Los Angeles. Mimeo. Government of Indonesia (GOI). 2003. ìSituationAnalysis of Nutrition Problems in Indonesia: Its Policy, Programs and Prospective Development.î Directorate of Community Health, Ministry of Health, Government of Indonesia. Mimeo Hadi, H., R.J. Stoltzfus, M.J. Dibley, L.H. Moulton, K.P. West, C.L. Kjolhede, T. Sadjimin. 2000. ìVitamin A Supplementation Selectivity Improves the Linear Growth of Indonesian Preschool Children: Results from a Randomized Controlled Trial.î American Journal of Clinical Nutrition 71:507-13 Helen Keller International. 1998. ìRe-emergence of the Threat of VitaminADeficiency.î Indonesia Crisis Bulletin Year 1, Issue 2: October 1998. Horton, S. 1999. ìOpportunities for Investments in Nutrition in Low-income Asia.î Asian Development Review 17: 246-273 Marks, G. 2003. ìProtein-energy Malnutrition in Indonesia: Key Challenges and Options.î Australian Centre for International Tropical Health and Nutrition (ACITHN). Mimeo UNICEF. 2005. Monitoring the Situation of Children and Women. Website http://www.childinfo.org, accessed on March 18 2005. World Bank. 1993. World Development Report 1993: Investing in Health. New York: Oxford University Press. World Bank. 2004. Implementation Completion Report for the Intensified Iodine Deficiency Control Project. Report No: 29511. Washington, D.C.: The World Bank. World Health Organization. 1982. Control of VitaminADeficiency and Xerophthalmia: Report of a Joint WHO/ UNICEF/USAID/Helen Keller International/IVACG Meeting. WHO Technical Report Series No. 672:5- 64. Geneva: World Health Organization. World Health Organization. 2001. The Optimal Duration of Exclusive Breastfeeding. Report of an Expert Consultation. Geneva: WHO. WHO-CHOICE. 2003. CHOosing Interventions that are Cost-Effective. Website http://www.who.int/choice/ en/ accessed on December 22 2006. 19 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES 20 ANNEX A. OVERVIEW OF NUTRITIONAL SITUATION IN INDONESIA -- OUTCOMES AND SERVICES InAnnexA, we examine (1) the overall nutritional situation in Indonesia and (2) inequalities in nutrition outcomes and service utilization by socioeconomic status and residential area. Data This overview involves the analysis of three major data sources including the National Socio-Economic Household Survey (SUSENAS) 2001 and 2003, the Indonesian Family Life Survey (IFLS) 2000, and the Indonesian Demographic and Health Survey (DHS) 2002/3. Although these surveys differ in sampling frames and questionnaire content, we maximize the available information by using a survey where that particular survey is strongest. This means that for a particular indicator the data source may vary, but in general a more recent and/ or nationally representative survey is used when multiple sources are available (Table A1). SUSENAS is a series of large-scale multi-purpose annual socioeconomic surveys initiated in 1963. It has included a nationally representative sample, typically of about 200,000 households since 1993. Each survey contains specialized modules covering about 60,000 households that are rotated over time to collect specific information such as health care and nutrition. SUSENAS 2001 and 2003 contained health modules which collected anthropometric measures: weight for children and mid-upper arm circumference for adult women. The large sample sizes facilitate estimations at a district-level, which is typically not possible in IFLS and DHS. IFLS is a continuing longitudinal socioeconomic and health survey. It is based on a sample of households representing about 83% of the Indonesian population living in 13 of the 26 provinces in 1993: Sumatera Utara, Sumatera Barat, Sumatera Selatan, Lampung, Jakarta, Jawa Barat, Jawa Tengah, Yogyakarta, Jawa Timur, Bali, Nusa Tenggara Barat, Kalimantan Seletan, and Sulawesi Seltan. The first wave (IFLS1) was administered in 1993 to individuals living in 7,224 households, followed by subsequent surveys in 1997, 1998, and 2000. IFLS 2000, which covered 43,650 individuals, included information on serum hemoglobin level, the only serum biomarker available in our analysis, and various anthropometric measures for women and children. Finally, DHS is a nationally-representative household survey on the areas of population, health, and nutrition. It is a standardized survey which has been conducted across more than 70 countries. Typically, women between 15 and 49 from sampled households are interviewed on various health and population issues. For children of these women between 0-4 years of age detailed maternal and child health care information -- including nutrition -- is obtained. In Indonesia, five surveys have been conducted since 1987. DHS 2002/3 is the latest survey with a sample size of 29,483 ever married women from 33,088 households. Nutritional outcome and service utilization indicators We focus on two broad areas of nutrition programs in this report: (1) prevention of micronutrient deficiency including iron, vitaminA, and iodine, and (2) growth monitoring and promotion.Abrief overview of nutritional programs is presented in Box A1. For each program area, we examine both outcome and service utilization indicators. Definitions of the indicators employed are explained in detail in Table A1. For micronutrient deficiency prevention programs, limited outcome indicators and relatively rich information on service utilization are available. Serum hemoglobin level is the only biologic indicator included. Night 21 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES blindness during pregnancy -- which is relatively rare and symptomatic -- is included, but no vitamin A deficiency measure among children is available at the national-level. No outcome indicator on iodine deficiency is included in our analysis. Various anthropometric measures are available as outcome indicators for growth monitoring and promotion. For adult females, a Body Mass Index (BMI) below 18.5 and Mid-Upper Arm Circumference (MUAC) less than 23 cm are considered indications of malnutrition. ABMI of 25 or higher is considered overweight. For children, standardized Z-scores, or number of deviations below the median, are calculated based on weight, height, and age. Children whose Height-for-Age Z score is less than -2 are categorized as stunted, whose Weight-for-Age Z- score is less than -2 are underweight, and those with a weight-for-Height Z-score less than -2 are classed as isted. Box A1. Brief overview of nutritional programs in Indonesia In terms of micronutrient deficiency control there are three general strategies: (1) to target micronutrient supplements to endemic area or vulnerable populations, (2) to fortify micronutrients in the diet, and (3) to improve diets containing micronutrients. Indonesia pursues each of these strategies to at least a limited extent, however much of the focus in micronutrient deficiency control centers on supplementation programs. A major program for iron deficiency control is the antenatal iron supplementation program, which aims to cover all pregnant women who are reached through Community Health Centers and Posyandus. Distribution of iron syrup to children under-5 years has been implemented in the eastern part of the country since 1996. Fortification programs have not reached the population effectively. For vitamin A deficiency control, vitamin A capsules are distributed nation wide to children 6 to 59 months old in April and August, twice a year. Vitamin A supplements are distributed to postpartum women as well. Diet improvement through nutrition education is integrated into routine nutrition programs. Finally, universal salt iodization is the major intervention strategy for iodine deficiency control. In addition, iodized capsule distribution in iodine deficiency disorder endemic areas targets women of reproductive age and young children. In terms of growth monitoring and promotion, the Posyandu program -- originated from the UPGK (Family Nutrition Improvement Program) in the 1970s -- is the main national community nutrition program and focuses on early child and maternal nutrition. Growth monitoring of young children by their mothers though monthly weighing at the village posyandu post is one of the major components. A growth monitoring chart is used to record weight and to identify target children for food supplementation (explained shortly below). The Posyandu also can provide nutrition education on infant ands child feeding, and is also used for vitamin A supplement distribution for women and children. Another major program for growth monitoring and promotion is a supplementary feeding program. JSP-BK (Social Safety Net) is the largest program with several components including supplementary food, launched after the economic crises in 1998. Although target households are to be identified by using a household wealth classification system, supplementary food does not preferentially target the poor due to program complexity. In addition, supplementary food for children and pregnant women is typically implemented through Posyandu, and target children are identified using growth monitoring charts. A school feeding program has been implemented to the limited extent. More detailed information on each program is available elsewhere (GOI 2003). 22 Table A1. Definitions and data sources for service utilization and outcome indicators by program area Indicator Definition Data source Iron Supplementation Service utilization Antenatal iron supplement Percent of women who took antenatal iron DHS 2002 supplement, based on the latest pregnancy within the last 5 years before the interview Hemoglobin level measured Percent of women who checked hemoglobin level IFLS 2000 during pregnancy during pregnancy, based on the latest pregnancy within the last 5 years before the interview Outcome Anemia prevalence, women Percent of women 15-49 years with serum hemoglobin IFLS 2000 15-49 years level below 11g/dl (for pregnant women, serum hemoglobin level below 12g/dl) Anemia prevalence, Percent of children 1-4 years with serum hemoglobin IFLS 2000 children 1-4 years level below 11g/dl Anemia prevalence, Percent of children 5-14 years with serum hemoglobin IFLS 2000 children 5-14 years level below 11g/dl Vitamin A Supplementation Service utilization Post-partum vitamin A Percent of women who received a vitamin A dose in the DHS 2002 supplement first two months after delivery, among those who gave birth in the five years preceding the survey Vitamin A capsule received, Percent of children 6-59 months old who received DHS 2002 children 6-59 months vitamin A supplement within the last 6 months before the interview Vitamin A rich food intake, Percent of children 0-2 years who consumed vitamin DHS 2002 children 0-2 years A-rich fruits and vegetables (pumpkins, carrots, red sweet potatoes, green leafy vegetables, mangoes, papayas, and other locally grown fruits and vegetables that are rich in vitamin A) within the day or night before the interview Outcome Prevalence of night blindness Percent of women who suffered from night blindness DHS 2002 during pregnancy during pregnancy, based on the latest pregnancy in the five years before the interview Iodine Fortification Service utilization Iodized salt use at home Percent of households using iodized salt for cooking SUSENAS 2003 Iodized salt use at home, Percent of households using adequately fortified iodized SUSENAS 2003 adequately fortified salt for cooking Growth Monitoring and Promotion Service utilization Visited Posyandu (1 month), Percent of children 0-4 years who visited Posyandu within SUSENAS 2001 children 0-4 years the last month before the interview Visited Posyandu (2 months), Percent of children 0-4 years who visited Posyandu within SUSENAS 2001 children 0-4 years the last two months before the interview Complementary meal received Percent of children who received feeding from Posyandu, IFLS 2000 at the Posyandu13 among those who visited Posyandu within the last month before the interview 13IFLS2000measuredthepercentofchildrenreceivingcomplementarymealsasarewardwhenvisitingthePosyandu,notthe"supplementary food for children and pregnant women" program which targets children with growth problems defined from a growth chart. 23 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Weight measured at Percent of children whose weight is measured at IFLS 2000 the Posyandu Posyandu, among those who visited Posyandu within the last month before the interview Behavioral Outcome Exclusive breast feeding, Percent of children 0-3 months who are exclusively DHS 2002 children 0-3 months breastfed Exclusive breast feeding, Percent of children 0-5 months who are exclusively DHS 2002 children 0-5 months breastfed Protein rich food intake, Percent of children 0-2 years who consumed protein rich DHS 2002 children under 0-2 years food (meat, fish, shellfish, poultry, and eggs) within the day or night before the interview Outcome BMI <18.5, women Percent of women 15-49 years whose Body Mass Index* IFLS 2000 15-49 years is less than 18.5 BMI >=25, women Percent of women 15-49 years whose Body Mass Index* IFLS 2000 15-49 years is equal to or greater than 25 MUAC < 23 cm, women Percent of women 15-49 years whose Mid-Upper Arm SUSENAS 2001 15-49 years circumference is less than 23 cm Stunting, children 0-4 years Percent of children 0-4 years whose Height-for-Age Z- IFLS 2000 score is below 2 Underweight, Percent of children 0-4 years whose Weight-for-Age Z- SUSENAS 2001 children 0-4 years score is below 2 Wasting, children 0-4 years Percent of children 0-4 years whose Weight-for-Height IFLS 2000 Z-score is below 2 * Body Mass Index = body weight (kilograms)/height squared (meters) 24 1. OVERVIEW OF CURRENT NUTRITION SITUATION IN INDONESIA Nutritional outcomes and service utilization levels at a national level are summarized by program area in Table A2. Table A2. Nutritional status and utilization of nutrition programs in Indonesia (%) Indicator National SE N** Source average* Iron Supplementation Service utilization Antenatal iron supplement 79.9 0.59 13349 DHS 2002 Hemoglobin level measured during pregnancy 43.8 1.12 2459 IFLS 2000 Outcome Anemia prevalence, women 15-49 years 18.8 0.43 10273 IFLS 2000 Anemia prevalence, children 1-4 years 53.0 1.04 2671 IFLS 2000 Anemia prevalence, children 5-14 years 18.9 0.55 6316 IFLS 2000 Vitamin A Supplementation Service utilization Post-partum vitamin A supplement 42.5 0.73 13349 DHS 2002 Vitamin A capsule received, children 6-59 months 75.4 0.61 13670 DHS 2002 Vitamin A rich food intake, children 0-2 years 67.4 0.85 8867 DHS 2002 Outcome Prevalence of night blindness during pregnancy 1.7 0.17 13349 DHS 2002 Iodine Fortification Service utilization Iodized salt use at home 84.8 0.10 204822 SUSENAS 2003 Iodized salt use at home, adequately fortified 69.7 0.13 204822 SUSENAS 2003 Growth Monitoring and Promotion Service utilization Visited Posyandu (1 month), children 0-4 years 39.8 0.33 28651 SUSENAS 2001 Visited Posyandu (2 months), children 0-4 years 56.0 0.33 28651 SUSENAS 2001 Complementary meal received at the Posyandu 71.1 1.22 1537 IFLS 2000 Weight measured at the Posyandu 98.6 0.30 1537 IFLS 2000 Behavioral Outcome Exclusive breast feeding, children 0-3 months 50.7 2.66 1057 DHS 2002 Exclusive breast feeding, children 0-5 months 37.7 2.09 1638 DHS 2002 Protein rich food intake, children under 0-2 years 55.7 0.90 8867 DHS 2002 Outcome BMI <18.5, women 15-49 years 14.3 0.38 10306 IFLS 2000 BMI >=25, women 15-49 years 20.7 0.45 10306 IFLS 2000 MUAC < 23 cm, women 15-49 years 13.1 0.14 76806 SUSENAS 2001 Stunting, children 0-4 years 33.2 0.81 3889 IFLS 2000 Underweight, children 0-4 years 27.2 0.45 11785 SUSENAS 2001 Wasting, children 0-4 years 10.5 0.53 3874 IFLS 2000 * Estimates based on IFLS 2000 are representative for only 13 provinces. ** Unweighted number of observations 25 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Prevention of micronutrient deficiency About 80% of pregnant women received antenatal iron supplements, although only 44% of pregnant women had serum hemoglobin level measured during pregnancy. However anemia prevalence is still moderately high in Indonesia. Among women between 15 to 49 years of age, 19% had anemia -- a serum hemoglobin level below 11g/dl (below 12g/dl for those pregnant). Prevalence is slightly lower for women 15-19 years, but it is by and large similar across age groups (Table A3). More than half of children 1-4 years had low hemoglobin levels consistent with iron deficiency anemia. In particular, 70% of children 12-23 months old had anemia, while prevalence decreased with age (Table A3). About 19% of children 5-14 years had anemia. Table A3. Age patterns of selected outcome indicators (%) Women Indicator Age group (years) 15-19 20-24 25-29 30-34 35-39 40-44 45-49 Prevention of micronutrient deficiency Anemia prevalence, women 15-49 years 16 19 18 19 20 19 22 Growth monitoring and promotion BMI <18.5, women 15-49 years 26 19 13 10 7 10 11 BMI >=25, women 15-49 years 6 11 16 25 30 34 32 MUAC < 23 cm, women 15-49 years 28 17 11 8 7 7 8 Children Indicator Age group (months) 0-11 12-23 24-35 36-47 48-60 Prevention of micronutrient deficiency Anemia prevalence, children 1-4 years - 70 56 47 41 Growth monitoring and promotion Stunting, children 0-4 years 21 40 38 34 37 Underweight, children 0-4 years 12 29 34 31 30 Wasting, children 0-4 years 10 15 13 8 7 Regarding prevention of vitaminAdeficiency, 43% of post-partum women received vitaminAsupplementation during the post-partum period. Approximately 75% of children between 6 to 59 months received vitamin A supplements within 6 months before the survey, and 67% of children under age 3 years are reported to take vitamin-A rich fruits and vegetables in the last 24 hours. There are no vitamin A deficiency (VAD) measures available for children under age 5 at the national-level, although community-level studies have estimated VAD levels. In a study conducted in rural Central Java in the early 1990s, while xeropthalmia is almost absent from the study sample (n=1437), 15% and 67% of children between 6-48 months had plasma retinol level less than 0.35 ?mol/L and 0.70 µmol/L, respectively14 (Hadi 2000). In rural West Java in 1996, with 155 study observations, 54% of infants had plasma retinol level less than 0.70 µmol/L (Dijkhuizen 2001) Although severe clinical vitamin A deficiency (VAD) such as xeropthalmia has 14 WHO identifies vitamin Adeficiency as a public-health problem when prevalence of vitamin Adeficiency (plasma retinol level less than 0.35 µmol/L) is 5% or higher (WHO 1982). 26 decreased significantly in Indonesia (HKI 1998), monitoring population-level prevalence of subclinical VAD will be essential for evaluating vitamin A supplementation. Regarding iodine fortification, about 85% of households used iodine fortified salt in 2003, significantly improved form 63% in 1997, thanks to increased availability of salt iodization that in part resulted from the World Bank funded IDD control project (World Bank 2004). However only 70% of households consumed adequately fortified iodized salt. In terms of measured outcomes, iodine deficiency disorder (IDD) has decreased from 28% in 1990 to 9% in 1998 (GOI 2003). In more recent data the total goiter rate, as assessed through the difficult to standardize palpation method, is 11% from the IDD evaluation survey in 2003, unchanged from the prevalence in 1998 (GOI 2003). Nevertheless, in 14 IDD endemic districts identified in 1998, the total goiter rate decreased from 44% in 1996-98 to 25% in 2003 (World Bank 2004). Growth Monitoring and Promotion About 40% of children under age 5 visited the Growth Education and Promotion Center - Posyandu - within a month before the survey. Among those who visited the Posyandu, 71% received a complementary meal and 99% had their weight measured. A protein rich diet for all children (Posyandu attendees and others) remains relatively uncommon. Only 56% of children under-3 years had protein rich food in the last 24 hours. The international evidence for the benefits of exclusive breastfeeding is growing and WHO guidelines recommend that it be promoted (WHO 2001). However only 51% and 38% of children under 4 and 6 months of age, respectively, are exclusively breastfed. Furthermore exclusive breastfeeding rates among children 0-1 and 2-3 months declined between 1987 and 1991 and have stagnated since then. The rates among children 4-5 months decreased steadily overall (FigureA1). One Indonesian program that appears to effectively extend the exclusive breastfeeding period and improve nutritional status among children is the village midwife program (Frankenberg et al. 2004). Figure A1. Trends of exclusive breastfeeding rates by age group 100 ) %( 80 gnideeft 198 7 60 199 1 ase 199 4 br 40 199 7 evi 200 2 uslcx 20 E 0 0-1 2-3 4-5 age in months (Source: ORC Macro, 2005. MEASURE DHS STATcompiler. http://www.measuredhs.com, accessed on March 18 2005) 27 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Anthropometric measures also highlight prevalent malnutrition in Indonesia. About 14% and 13% of women between 15 to 49 years had BMI lower than 18.5 and MUAC below 23 cm, respectively. In particular, malnutrition -- measured by both BMI and MUAC -- is more prevalent among younger women (Table A3). On the other hand, 21% of women are estimated to be overweight. Overweight prevalence is substantially higher for women in the 30s and 40s, indicating an emerging need for chronic disease prevention (Table A3). The prevalence of underweight among children under 5 years was 27% in 2001, a reduction by about one third from the prevalence in the late 1980s (FigureA2). However, about 33% and 11% of children still suffered stunting (a long-term malnutrition indicator) and wasting (a short-term malnutrition indicator), respectively, in 2000. Figure A2. Trend of underweight prevalence among children under-5 years, 05 0 04 ra Ye 032 01 0 1985 1990 1995 2000 2005 Prevalence (%) (Source: WHO Global Database on Child Growth and Malnutrition, SUSENAS 2001) Regional cross-national comparisons shows that the prevalence of underweight and stunting prevalence in Indonesia is roughly similar with those in Myanmar, Philippines, and Viet Nam (FigureA3). However, Indonesia exhibits a high wasting prevalence relative to its under-five mortality rate (Figure A4), implying a more urgent problem of short-term malnutrition. Figure A3. Prevalence of malnutrition among children under-5 for some Southeast Asian countries, 2000 50 40 ) %( 30 underw eight ecne stunting al 20 w asting evr P 10 0 Cambodia Indonesia Myanmar Philippines Viet Nam (SOURCE: UNICEF http://www.childinfo.org, accessed on March 18 2005, and SUSENAS 2001) * The estimates for Indonesia and the Philippines are for year 2001. 28 Figure A4. Prevalence of wasting, and Under-five Mortality Rate for some Southeast Asian countries, 2000* 16 160 14 140 12 120 0) ) 10 100 (% gnits Wasting 8 80 100rep( U5MR R wa 6 60 M5 U 4 40 2 20 0 0 Cambodia Indonesia Myanmar Philippines Viet Nam (SOURCE: UNICEF http://www.childinfo.org, accessed on March 18 2005, and Indonesian Family Life Surveys 2000) * The wasting estimate for the Philippines is from year 2001. 2. INEQUALITIES IN OUTCOMES AND PROGRAM UTILIZATION BY HOUSEHOLD CHARACTERISTICS Beyond national-level accomplishments and challenges, understanding disparities in the nutritional situation by subpopulation is essential for developing effective strategies and ensuring equity across groups. In the rest of AnnexA, we discuss differentials by socioeconomic condition (Annex B discusses differentials by geographic region). We focus on three socioeconomic characteristics: residential area (urban/rural), household wealth, and education. In general, the analysis shows that women and children in socio-economically disadvantaged groups -- those living in a rural area, living in a poor household, and with less education -- have greater nutritional needs than their counterparts. Differentials in outcome and service utilization by residential area In 2003, about 59% of households were in rural areas15, and residential patterns of outcomes and service utilization by urban/rural status are summarized in Table A4. Overall, women and children in rural areas exhibit poorer nutritional outcomes as well as lower program utilization, compared with their counterparts in urban areas. Disparities by residential area are particularly noticeable for: night blindness during pregnancy and stunting prevalence among children 0 to 4 years. Differences in service utilization are also large for antenatal hemoglobin measurement, postpartum vitamin A supplementation, and complementary meal receipt among children who visited Posyandu. In particular, the difference in complementary meal receipt among Posyandu attendees (79% in urban vs. 65% in rural) suggests a differential in quality of service supplied by these facilities. Two exceptions to the rural disadvantage are exclusive breastfeeding and overweight prevalence. Exclusive breastfeeding among children under 4 and 6 months is slightly higher in rural areas. Overweight prevalence among women is about 40% higher in urban areas (25% in urban vs. 17% in rural). 15The estimate is based on SUSENAS 2003 analysis results. 29 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Table A4. Differentials in nutrition outcome and service utilization by residential area Indicator (%) National Residential area average* Urban Rural Iron Supplementation Service utilization Antenatal iron supplement 79.9 82.8 74.5 Hemoglobin level measured during pregnancy 43.8 * 49.3 39.5 Outcome Anemia prevalence, women 15-49 years 18.8 * 18.1 19.4 Anemia prevalence, children 1-4 years 53.0 * 54.0 52.2 Anemia prevalence, children 5-14 years 18.9 * 17.8 19.7 Vitamin A Supplementation Service utilization Post-partum vitamin A supplement 42.5 47.6 38.0 Vitamin A capsule received, children 6-59 months 75.4 80.2 71.3 Vitamin A rich food intake, children 0-2 years 67.4 70.1 65.0 Outcome Prevalence of night blindness during pregnancy 1.7 1.4 2.1 Iodine Fortification Service utilization Iodized salt use at home 84.8 88.5 82.1 Iodized salt use at home, adequately fortified 69.7 74.8 66.1 Growth Education and Promotion Service utilization Visited Posyandu (1 month), children 0-4 years 39.8 40.6 39.3 Visited Posyandu (2 months), children 0-4 years 56.0 55.4 56.5 Complementary meal received at the Posyandu 71.1 * 78.7 64.7 Weight measured at the Posyandu 98.6 * 98.7 98.5 Behavioral Outcome Exclusive breast feeding, children 0-3 months 50.7 50.0 51.3 Exclusive breast feeding, children 0-5 months 37.7 36.3 39.1 Protein rich food intake, children 0-2 years 55.7 59.6 52.1 Outcome BMI <18.5, women 15-49 years 14.3 * 12.8 15.7 BMI >=25, women 15-49 years 20.7 * 24.7 17.2 MUAC < 23 cm, women 15-49 years 13.1 11.7 14.3 Stunting, children 0-4 years 33.2 * 26.7 38.3 Underweight, children 0-4 years 27.2 24.5 29.1 Wasting, children 0-4 years 10.5 * 10.1 10.7 * Estimates based on IFLS 2000 are representative for only 13 provinces. An estimate is statistically different (p-value<0.05) between urban and rural areas (Standard errors are not shown in the table). Differentials in outcome and service utilization by household resources Two related types of household resource information are available: an asset and housing index from DHS and per capita household expenditure from IFLS and SUSENAS. Based on these continuous measures, household wealth or consumption quintiles are created16, weighted for survey sampling-weights. Disparities in outcomes and program utilization by household wealth quintile are summarized in Table A5. Overall, women and children in the lowest resource quintile had poorer nutritional outcomes as well as lower program utilization, compared to the middle quintiles. On the other hand, women and children in the highest resource quintile had better nutritional outcomes as well as higher program utilization than the middle. All 16The wealth quintile variable is included in the DHS dataset. 30 indicators showed 20% or higher relative differences between the lowest (poor) and the highest resource quintile (rich), but disparities between the poor and the rich are most marked for following outcomes: - Anemia among women 15-49 years (15% in rich vs. 23% in poor), - Night blindness during pregnancy (0.9% in rich vs. 3.1% in poor), - BMI below 18.5 among women 15-49 years (11% in rich vs. 16% in poor), - MUAC below 23 cm among women 15-49 years (10% in rich vs. 17% in poor), - Stunting among children 0-4 years (20% in rich vs. 43% in poor), and - Wasting among children 0-4 years (7% in rich vs. 12% in poor), and Table A5. Differentials in nutrition outcome and service utilization by household resource quintile Indicator (%) National Household resource quintile** average* Lowest second middle second highest lowest highest Iron Supplementation Service utilization Antenatal iron supplement 79.9 63.7 77.3 80.6 85.4 87.4 Hemoglobin level measured during pregnancy 43.8 * 40.8 40.9 41.0 49.2 51.1 Outcome Anemia prevalence, women 15-49 years 18.8 * 22.8 19.4 18.8 17.2 15.3 Anemia prevalence, children 1-4 years 53.0 * 56.3 58.6 57.4 42.8 43.7 Anemia prevalence, children 5-14 years 18.9 * 20.7 17.6 21.6 17.2 15.7 Vitamin A Supplementation Service utilization Post-partum vitamin A supplement 42.5 32.5 35.7 43.2 49.0 54.1 Vitamin A capsule received, children 6-59 months 75.4 62.7 75.8 78.3 82.8 80.6 Vitamin A rich food intake, children 0-2 years 67.4 61.7 65.0 67.4 71.7 72.2 Outcome Prevalence of night blindness during pregnancy 1.7 3.1 1.7 1.7 1.1 0.9 Iodine Fortification Service utilization Iodized salt use at home 84.8 78.4 83.3 85.8 87.4 89.1 Iodized salt use at home, adequately fortified 69.7 62.0 67.3 70.4 72.7 76.3 Growth Education and Promotion Service utilization Visited Posyandu (1 month), children 0-4 years 39.8 39.3 42.2 41.8 38.6 35.8 Visited Posyandu (2 months), children 0-4 years 56.0 56.0 58.7 57.5 55.5 50.2 Complementary meal received at the Posyandu 71.1 * 72.4 71.2 64.7 72.8 76.1 Weight measured at the Posyandu 98.6 * 98.2 98.5 99.5 98.1 98.9 Behavioral Outcome Exclusive breast feeding, children 0-3 months 50.7 54.5 58.2 52.6 52.2 31.7 Exclusive breast feeding, children 0-5 months 37.7 40.3 44.8 38.3 39.3 23.5 Protein rich food intake, children 0-2 years 55.7 48.4 49.3 54.7 62.9 64.3 Outcome BMI <18.5, women 15-49 years 14.3 * 16.2 17.5 13.8 13.1 10.7 BMI >=25, women 15-49 years 20.7 * 14.9 18.3 20.9 24.0 26.6 MUAC < 23 cm, women 15-49 years 13.1 17.1 14.3 12.7 11.0 9.6 Stunting, children 0-4 years 33.2 * 42.7 37.2 32.9 24.3 19.6 Underweight, children 0-4 years 27.2 29.4 28.5 28.0 25.5 20.6 Wasting, children 0-4 years 10.5 * 11.8 10.9 11.4 9.7 7.1 * Estimates based on IFLS 2000 are representative for only 13 provinces ** Quintiles are created using sampling-weight, based on an asset-housing condition index (DHS) and per capita expenditure (IFLS and SUSENAS). Household wealth quintiles are not fully comparable across indicators due to different wealth measures by data source. Data source for each indicator is listed in Table A1. An estimate is statistically different (p-value<0.05) from that in the lowest quintile (Standard errors are not shown in the table). 31 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Relative differences in program utilization exceeded 25% or more for antenatal iron supplementation, postpartum vitaminAsupplementation, and protein rich food intake for children. Interestingly, even though the poor had a greater prevalence of underweight, there is little to no difference by resource quintiles in terms of Posyandu usage, the receipt of a complementary meal, or measured weight at the Posyandu. In addition, exclusive breastfeeding and overweight prevalence are again two exceptions of the poor disadvantage. Overweight prevalence among women is about 80% higher among the rich (27% in the highest vs. 15% in the lowest quintile), indicating different needs for health education by wealth status. Differentials in outcome and service utilization by education In certain settings, education can be an important determinant of nutrition, independent of wealth. We categorize educational attainment into five levels: (1) no education or less than primary school completion; (2) primary school completion; (3) junior high school completion; (3) senior high school completion; and (5) college and more. For women's indicators, we obtained educational attainment for each woman. For indicators among children under 5 years, we used educational attainment of the mother (indicators from DHS 2002) or a household head (indicators from IFLS 2000 and SUSENAS 2001). In 2003, about 31% of household heads had no education or did not complete primary education.Approximately 33%, 14%, 19%, and 5% of household heads completed primary school, junior high school, high school, and junior college or more, respectively17. Among mothers of children under 5 years, 20% did not finish primary education.About 36% of mothers completed only primary school.About 18%, 21%, and 6% of mothers completed junior high school, high school, and junior college or more, respectively18. Table A6 summarizes disparities in outcomes and service utilization by educational attainment. By and large, women and children in lower educational categories show poorer outcomes and lower program utilization.Among outcome indicators, anemia in women and children, night blindness in women, and stunting and underweight in children show clear gradients in prevalence by educational attainment. Exclusive breastfeeding rate is highest among children whose mother completed primary school (higher than that among children whose mother did not complete primary school), and it decreased as mother's educational attainment increased. For women's anthropometric indicators and children's wasting, there are also concave shapes of malnutrition prevalence by education. Women who completed primary or junior high education have a higher prevalence of low BMI than women in lower or higher education categories (although this may be attributed to the secular increase in education among younger women). In addition, children whose mother completed primary or junior high education have a higher prevalence of wasting. These patterns are further examined using multivariate analyses below. For all micronutrient program utilization rates, educational gradients are apparent -- the more educated the mother, the more likely is the mother or child to receive the supplement. Differentials in children's food intake are relatively moderate, compared to those in micronutrient program use. For Posyandu use, the gradients are 17Educational attainment of a household head is analyzed using SUSENAS 2003. 18Mother's educational attainment is analyzed using DHS 2002/03. 32 less clear. Children from the three middle groups (children whose household head completed primary, junior high, or senior high school) utilized Posyandu more than children from both the lowest and highest education group. Table A6. Differentials in nutrition outcome and service utilization by educational attainment Indicator (%) Highest education completed ** National Less Primary Junior Senior Beyond average* than high high high high primary Iron Supplementation Service utilization Antenatal iron supplement 79.9 61.1 77.6 82.3 88.6 91.8 Hemoglobin level measured during pregnancy 43.8 * 39.6 41.2 42.8 47.9 55.6 Outcome Anemia prevalence, women 15-49 years 18.8 * 21.7 19.6 16.5 17.5 14.5 Anemia prevalence, children 1-4 years 53.0 * 56.9 53.0 54.0 48.2 46.9 Anemia prevalence, children 5-14 years 18.9 * 20.1 21.5 17.5 14.6 13.9 Vitamin A Supplementation Service utilization Post-partum vitamin A supplement 42.5 28.8 40.5 44.8 53.4 54.1 Vitamin A capsule received, children 6-59 months 75.4 62.1 76.3 79.2 81.1 82.9 Vitamin A rich food intake, children 0-2 years 67.4 64.2 64.4 69.7 71.5 72.9 Outcome Prevalence of night blindness during pregnancy 1.7 2.9 1.7 1.7 1.1 0.8 Iodine Fortification Service utilization Iodized salt use at home 84.8 76.5 86.0 90.0 91.1 93.4 Iodized salt use at home, adequately fortified 69.7 59.0 69.8 77.4 79.6 83.0 Growth Education and Promotion Service utilization Visited Posyandu (1 month), children 0-4 years 39.8 33.9 42.5 41.0 42.3 38.8 Visited Posyandu (2 months), children 0-4 years 56.0 50.1 59.6 57.6 57.2 53.7 Complementary meal received at the Posyandu 71.1 * 67.2 68.1 70.8 80.0 75.9 Weight measured at the Posyandu 98.6 * 99.7 98.5 97.1 98.6 99.0 Behavioral Outcome Exclusive breast feeding, children 0-3 months 50.7 44.4 58.0 48.5 49.5 25.1 Exclusive breast feeding, children 0-5 months 37.7 39.0 42.5 34.9 33.4 22.5 Protein rich food intake, children 0-2 years 55.7 48.2 50.7 58.7 64.9 65.4 Outcome BMI <18.5, women 15-49 years 14.3 * 13.2 14.2 18.7 12.4 11.5 BMI >=25, women 15-49 years 20.7 * 22.7 21.4 16.1 20.9 22.4 MUAC < 23 cm, women 15-49 years 13.1 12.3 13.6 16.3 11.0 7.7 Stunting, children 0-4 years 33.2 * 40.5 34.7 33.5 26.3 19.0 Underweight, children 0-4 years 27.2 30.6 27.7 26.7 24.5 20.8 Wasting, children 0-4 years 10.5 * 10.7 11.7 11.8 9.2 6.9 * Estimates based on IFLS 3 are representative for only 13 provinces ** For children's indicators, education by a mother (DHS 2002) or a household head (IFLS 2000 and SUSENAS 2001) is used. Data source for each indicator is listed in Table A1. An estimate is statistically different (p-value<0.05) from that in a less than primary education category (Standard errors are not shown in the table). 33 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Multivariate analysis of outcomes and utilization Finally, we conduct multivariate weighted-regression analyses in order to examine adjusted differentials in indicators controlling for individual demographic and household socioeconomic characteristics. For most women's indicators, we adopt this specification: Yij = ij + age*Aij + R*Xij + P*Pi + ij, where Y is a particular binary outcome or utilization measure, A is age in years, X is a vector of covariates, and the P's are province-level dummy variables. The covariates in X are all binary variables: (1) rural area households -- with a reference category of urban area; (2) the lowest quintile households and (3) the highest quintile households -- with a reference category of the three middle quintile households; and (4) less than primary school completion and (5) the completion of only primary school -- with a reference category of completing junior high school or more. The specification is estimated as a logistic regression. For antenatal iron and postpartum vitamin A supplement, we included two age dummy variables instead of a continuous variable, age in year, in order to control for lower maternal health care utilization among younger and older women. The model is: Yij = ij + age20*Aage20 + age35*Aage35 + R*Xij + P*Pi + ij, where Y is logit of the particular binary utilization measure, Aage20 is a binary variable for women younger than 20 years at the time of delivery, Aage35 is a binary variable for women 35 or older at the time of delivery, X again is a vector of covariates same as in the previous model, and the P's are province-level dummy variables. For children's indicators, we adopt the specification: Yij = ij + age*Aij + sex*Sij + R*Xij + P*Pi + ij, where Y is a particular binary outcome or utilization measure, A is age in months (in years for anemia), S is sex of a child, X is a vector of the same covariates as in a women's model, and Ps are a province dummy variable. Finally, for household iodized salt use, we used a model; Yij = ij + R*Xij + P*Pi + ij, where Y is logit of the particular utilization measure, X is a vector of the same covariates as above, and P are province-level dummy variables. Standard errors are adjusted for clustering at the provincial level. Statistical significance is indicated for p-value less than 0.05. 34 (a) Adjusted differentials in outcome indicators Prevention of micronutrient deficiency Adjusted odds ratios of outcomes in micronutrient deficiency prevention are summarized in Table A7. Women with lower socioeconomic status (living in rural areas, living in the lowest wealth quintile households, and less schooling) are more likely to be anemic. Women from the poorest wealth quintile are 26% more likely to have anemia while those living from the richest quintile are 22% less likely to have anemia compared to those in the three middle quintiles. Women's education is inversely associated with anemia. Among children, female children between 1 to 4 years have 13% lower odds of having anemia than male children. Odds ratios by household wealth status among children are not statistically significant, while household head's education is negatively associated with the odds of being anemic. Interestingly, for children 1 to 4 years old, those in rural areas are 20% less likely to be anemic than children in urban areas. Table A7 Odds Ratios of outcomes in micronutrient deficiency prevention, by individual and household socioeconomic characteristics: Multivariate regression analyses Indicator OR SE* p-value N source Anemia, women 15-49 years Age (year) 1.01 0.00 0.00 10208 IFLS 2000 Rural 0.99 0.06 0.88 Poorest quintile 1.26 0.09 0.01 Richest quintile 0.78 0.09 0.00 Less than primary school 1.19 0.06 0.01 Primary school completed 1.13 0.05 0.02 Anemia, children between 1-4 years Age (year) 0.67 0.03 0.00 2654 IFLS 2000 Female 0.87 0.04 0.00 Rural 0.80 0.09 0.02 Poorest quintile 1.08 0.17 0.66 Richest quintile 0.71 0.19 0.06 Less than primary school 1.29 0.08 0.00 Primary school completed 1.14 0.20 0.52 Anemia, children between 5-14 years Age (year) 0.79 0.01 0.00 6249 IFLS 2000 Female 0.92 0.08 0.34 Rural 0.98 0.09 0.84 Poorest quintile 1.07 0.11 0.55 Richest quintile 0.98 0.16 0.88 Less than primary school 1.44 0.12 0.00 Primary school completed 1.57 0.07 0.00 Night blindness, women 15-49 during pregnancy Age (year) 1.01 0.02 0.57 13344 DHS 2002 Rural 0.98 0.18 0.90 Poorest quintile 1.63 0.24 0.04 Richest quintile 0.68 0.52 0.46 Less than primary school 1.63 0.20 0.02 Primary school completed 1.05 0.28 0.86 * Standard Error of the coefficient estimated from logistic regression analyses, adjusted for clustering on province. (Estimates include province dummy variables, not shown) 35 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES For antenatal night blindness, there are typical socioeconomic gradients. However, since the condition is rare (1.7% among women between 15 to 49 years), the magnitude of differentials are higher than those in other indicators. Women in the poorest quintile are 63% more likely to have night blindness than women in the middle quintiles. Women with less than primary education are 66% more likely to have night blindness, compared to women who completed junior high school or more. Growth monitoring and promotion Women of lower socioeconomic status are more likely to be malnourished (TableA8). Women in rural areas are 26% more likely to have a BMI less than 18.5. Women living in the poorest quintile households are 22% more likely to have MUAC less than 23 cm, compared to women in the middle three quintiles. Women living in the richest quintile are less likely to have malnutrition than women from the middle three quintiles (odds ratio 0.83 for MUAC and odds ratio 0.75 for BMI). For women with less than completed primary education, odds of having malnutrition increased by 26% (MUAC) and 34% (BMI), compared to women who completed junior high school or more. On the other hand, women with higher socioeconomic status are more likely to be overweight. Women in urban areas are 54% more likely to be overweight. Women from the richest quintile households are 20% more likely to be overweight, while those living in the lowest quintile are 30% less likely to be overweight, compared to women in the three middle quintiles. The effect of educational attainment is not statistically significant once adjusted for all control variables. Children from a lower socioeconomic status are more likely to be malnourished. Living in a rural area is associated with a 42% increased odds of being stunted. For children living in the richest quintile households, the odds of being underweight, stunted, and wasted are lower by 40%19, 40%, and 34%, respectively, compared to children living in the three middle quintiles. On the other hand, for children living in the poorest quintile households, the odds of being underweight, stunted, and wasted are higher by 40%20 and 37%, respectively. Children whose household head did not complete primary school are about 35% more likely to be underweight or stunted, compared to children whose household heads completed junior high school or more. 19Analysis with SUSENAS 2001 suggests 24% lower odds of underweight for the richest quintile. 20Odds of underweight for the poorest quintile is not significant using SUSENAS 2001. 36 Table A8. Odds Ratios of outcomes in growth monitoring and promotion, by individual and household socioeconomic characteristics: Multivariate regression analyses Indicator OR SE* p-value N source Women's nutritional status, between 15-49 years Thin: MUAC < 23 cm Age (year) 0.93 0.00 0.00 76806 SUSENAS 2001 Rural 1.10 0.08 0.22 Poorest quintile 1.22 0.05 0.00 Richest quintile 0.83 0.06 0.00 Less than primary school 1.34 0.03 0.00 Primary school completed 0.98 0.02 0.28 Thin: BMI <18.5 Age (year) 0.95 0.01 0.00 10235 IFLS 2000 Rural 1.26 0.04 0.00 Poorest quintile 1.06 0.05 0.27 Richest quintile 0.75 0.05 0.00 Less than primary school 1.26 0.08 0.05 Primary school completed 0.96 0.08 0.66 Overweight/Obesity: BMI >=25 Age (year) 1.07 0.00 0.00 10235 IFLS 2000 Rural 0.65 0.13 0.00 Poorest quintile 0.70 0.10 0.00 Richest quintile 1.20 0.07 0.01 Less than primary school 0.87 0.10 0.19 Primary school completed 1.14 0.07 0.06 Children's nutritional status, between 0-4 years Underweight Age (month) 1.02 0.00 0.00 11119 SUSENAS 2001 Female 0.90 0.07 0.16 Rural 1.10 0.05 0.07 Poorest quintile 1.01 0.10 0.89 Richest quintile 0.76 0.06 0.00 Less than primary school 1.23 0.07 0.01 Primary school completed 1.11 0.06 0.07 Stunting Age (month) 1.01 0.00 0.00 3822 IFLS 2000 Female 0.99 0.06 0.86 Rural 1.42 0.12 0.00 Poorest quintile 1.37 0.05 0.00 Richest quintile 0.60 0.16 0.00 Less than primary school 1.36 0.07 0.00 Primary school completed 1.18 0.09 0.07 Wasting Age (month) 0.99 0.00 0.01 3808 IFLS 2000 Female 1.08 0.10 0.40 Rural 0.99 0.17 0.97 Poorest quintile 1.10 0.09 0.28 Richest quintile 0.66 0.13 0.00 Less than primary school 1.00 0.15 0.99 Primary school completed 1.12 0.11 0.28 * Standard Error of the coefficient estimated from logistic regression analyses, adjusted for clustering on province. (Estimates include province dummy variables, not shown) 37 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES (b) Adjusted differentials in service utilization indicators Prevention of micronutrient deficiency Adjusted odds ratios of service utilization in micronutrient deficiency prevention are summarized in Table A9. Differentials by residential area are not statistically significant in all three micronutrient programs (iron supplementation, vitaminAsupplementation, and iodine fortification) implying that unadjusted differences observed in Table A4 are associated with wealth and educational gaps between urban and rural areas rather than an access gap across residential areas. For both antenatal iron and postnatal vitaminAsupplementation, women from the poorest quintile are less likely to receive supplementation (odds ratio 0.5 and 0.79 for iron and vitamin A, respectively) while those from the richest quintile are more likely to receive supplementation (odds ratio 1.14 and 1.24 for iron and vitaminA,respectively),comparedtothoseinthethreemiddlequintiles.Women'seducationisalsopositively associated with service use. For antenatal iron supplementation, younger (less than 20 at delivery) or older (35 and above at delivery) women are less likely to receive the supplementation. For children's vitamin A supplementation, children from the poorest quintile are 43% less likely to receive the supplement than children in the middle group. However, children living in the richest quintile are also less likely to receive vitaminAsupplementation. This is less of concern, since children in the richest quintiles are more likely to have vitamin A rich food and, probably, less likely to have vitamin A deficiency (though no evidence is currently available in Indonesia). Nevertheless, this suggests low participation in the vitamin A campaign among the richest households perhaps due to changing consumer preference resulting from low service quality or perhaps due to the Posyandu Revitalization Program, which differentially increased utilization among lower income households (Frankenberg, 2004). Households with lower socioeconomic status (rural areas, the lowest wealth quintile, and less schooling of household head) are less likely to use iodine fortified salt. However, differentials by household wealth quintile are smaller than those observed in iron and vitamin A supplementation programs. Growth monitoring and promotion For Posyandu use among children under 5 years, the utilization patterns by socioeconomic characteristics are rather unique. There is no differential in visit by residential area. However, among those who visited Posyandu, children in rural areas are 52% less likely to receive a complementary meal than urban children, indicating possible supply and quality issues in rural areas. Children from the richest quintile are less likely to visit a Posyandu. Among those who visited a Posyandu, children from the poorest quintile are more likely to receive a complementary meal. Household head's educational attainment is negatively associated with Posyandu use, adjusted for other control variables. Protein rich food intake among children under 3 years is positively associated with household socioeconomic status. Children from the poorest quintile are 26% less likely to take protein rich food, while those from the richest quintile are 27% more likely to take it compared to those in the three middle quintiles. Children whose mother did not complete primary education are less likely to have protein rich food than their counterparts, even after controlling for household resources. 38 Table A9. Odds Ratios of outcomes in micronutrient deficiency prevention, by individual and household socioeconomic characteristics: Multivariate regression analyses Indicator OR SE* p-value N Source Antenatal iron supplement receipt Age at delivery, below 20 years 0.67 0.12 0.00 13344 DHS 2002 Age at delivery, 35 and above 0.73 0.10 0.00 Rural 1.01 0.10 0.89 Poorest quintile 0.50 0.07 0.00 Richest quintile 1.14 0.06 0.04 Less than primary school 0.30 0.12 0.00 Primary school completed 0.60 0.08 0.00 Postnatal vitamin A supplement receipt Age at delivery, below 20 years 0.81 0.13 0.11 13344 DHS 2002 Age at delivery, 35 and above 0.88 0.07 0.08 Rural 0.87 0.11 0.24 Poorest quintile 0.79 0.07 0.00 Richest quintile 1.24 0.08 0.01 Less than primary school 0.49 0.06 0.00 Primary school completed 0.78 0.06 0.00 Vitamin A supplement receipt, children under 6-59 months Mean age (month) 1.00 0.003 0.53 13664 DHS 2002 Female 1.04 0.058 0.51 Rural 0.78 0.135 0.07 Poorest quintile 0.58 0.058 0.00 Richest quintile 0.79 0.097 0.01 Less than primary school, mother 0.40 0.146 0.00 Primary school completed, mother 0.76 0.069 0.00 Vitamin A rich food intake, children 0-2 years Mean age (month) 1.15 0.01 0.00 8864 DHS 2002 Female 1.17 0.06 0.01 Rural 0.94 0.16 0.68 Poorest quintile 0.77 0.13 0.04 Richest quintile 1.12 0.10 0.27 Less than primary school, mother 0.66 0.17 0.02 Primary school completed, mother 0.79 0.08 0.00 Household has iodized salt, any level Rural 0.62 0.14 0.00 204822 SUSENAS 2003 Poorest quintile 0.80 0.05 0.00 Richest quintile 1.24 0.10 0.03 Less than primary school, household head 0.37 0.07 0.00 Primary school completed, household head 0.65 0.06 0.00 Household has adequately iodized salt Rural 0.67 0.06 0.00 204822 SUSENAS 2003 Poorest quintile 0.84 0.03 0.00 Richest quintile 1.22 0.08 0.01 Less than primary school, household head 0.44 0.05 0.00 Primary school completed, household head 0.67 0.03 0.00 * Standard Error of the coefficient estimated from logistic regression analyses, adjusted for clustering on province. (Estimates include province dummy variables, not shown) 39 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Table A10. Odds Ratios of service utilization in growth monitoring and promotion, by individual and household socioeconomic characteristics: Multivariate regression analyses Indicator OR SE* p-value N Source Posyandu visit, children under 5 years Mean age (month) 0.97 0.00 0.00 28651 SUSENAS 2001 Female 1.05 0.03 0.09 Rural 1.04 0.07 0.51 Poorest quintile 0.92 0.05 0.10 Richest quintile 0.78 0.07 0.00 Less than primary school, household head 0.61 0.06 0.00 Primary school completed, household head 0.89 0.03 0.00 Supplementary food receipt from Posyandu, conditional on Posyandu visit Mean age (month) 1.03 0.01 0.00 1519 IFLS 2000 Female 1.08 0.10 0.46 Rural 0.48 0.19 0.00 Poorest quintile 1.38 0.09 0.00 Richest quintile 0.92 0.29 0.79 Less than primary school, household head 0.74 0.14 0.03 Primary school completed, household head 0.66 0.07 0.00 Protein rich food intake, children 0-2 years Mean age (month) 1.12 0.01 0.00 8864 DHS 2002 Female 1.04 0.05 0.44 Rural 0.97 0.07 0.71 Poorest quintile 0.74 0.11 0.00 Richest quintile 1.27 0.05 0.00 Less than primary school, mother 0.50 0.11 0.00 Primary school completed, mother 0.66 0.06 0.00 * Standard Error of the coefficient estimated from logistic regression analyses, adjusted for clustering on province. (Estimates include province dummy variables, not shown) 40 ANNEX B. REGIONAL HETEROGENEITY IN NUTRITION OUTCOMES AND SERVICES In the context of decentralization where local governments have increasing influence over public nutrition policy, it is essential to look beyond national mean outcomes and to understand the magnitude and patterns of variation across districts and provinces. 1. District-level heterogeneity In this annex we first examine district-level estimates for five selected indicators using SUSENAS 2001: (1) mid- upper arm circumference among women 15 to 49 years; (2) underweight among children under 5 years; (3) iodized salt use at home; and (4) Posyandu visit among children under 5 within a month prior to the survey. Information from a total of 328 districts are available from SUSENAS 2001, but only districts with 25 or more un-weighted observations are included in analyses in order to minimize the influence of measurement error. Table B1 summarizes population weighted district-level estimates and presents heterogeneity across districts. Table B1. Summary of district-level estimates of selected indicators for outcomes & service utilization (%) Indicator Median Mean* SD Min Max Number Median of number of districts** observations within a district** Outcome MUAC <23cm among women 15-49 yrs 12 13 8 0 60 316 190 Underweight among children 0-4 yrs 28 29 12 3 81 197 45 Service utilization Iodized salt use 94 84 20 17 100 306 606 Iodized salt use, adequately fortified only 75 70 23 9 100 306 606 Posyandu visit among children 0-4 years 38 40 19 0 100 280 78 Source: SUSENAS 2001 * Unweighted average of district-level estimates ** Districts with unweighted observations less than 25 are excluded. Posyandu visit within 1 month prior to the survey About 13 % of women between 15-49 years exhibit a MUAC less than 23 cm at the national level. However, district-level estimates ranged from 0 % (Malinau, Bulingan, and Nunukan in East Kalimantan; Selayar in South Sulawesi; and Dumai in Riau) to 60 % in T.T. Utara in Nusa Tenggara Timur.About 34% and 7 % of districts have prevalence over 15 % and 25 % respectively (Figure B1). There is even greater heterogeneity across districts for underweight among children 0-4 years. While the national prevalence is about 27 %, district-level estimates ranged from 3 % in Tabanan, Nusa Tenggara Barat to 81 % in Barito Selatan, Kalimantan Tengah. About 39 % of districts have a prevalence of 30 % or higher (Figure B2). In addition, clusters of districts show high malnutrition prevalence for both women and children. Districts were categorized into 3 equal-size groups based on ranking of malnutrition prevalence: low, moderate, and high prevalence group, and about 30 districts have high malnutrition prevalence for both women and children (Table B2). Looking over this list, it is quite apparent that poor nutrition outcomes are clustered in a handful of regions, most notably NTT with pockets in NTB, South Sulawesi, and East and Central Java. 41 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Figure B1. Distribution of district-level estimates of percent of women 15-49 years with MUAC < 23 cm. 6 .1 4 .1 2 .1 st icrt .1 isdfo 8 n .0 iotroporp 6 .0 4 .0 2 .0 0 0 25 50 75 100 precent of women with MUAC<23cm Figure B2. Distribution of district-level estimates of percent of children 0-4 years with underweight .1 8 .0 st icrt isdfo 6 .0 n iotroporp 4 .0 2 .0 0 0 25 50 75 100 precent of children, under-weighted Turning now to the consumption of iodized salt, while 82 % of households at the national level consume iodized salt, the coverage within each district varies greatly from 17 % to 100 %. About 10 % of districts have a coverage below 50 %, while 58 % of districts have a coverage of 90 % or higher. Heterogeneity is even greater for adequately fortified iodized salt use. Compared to the national average of 66 %, the district-level coverage ranges from 9 % to 100%. About 21 % of districts have a coverage below 50 %, while 24 % of districts have 90 % or higher (Figure B3). We can also look at the extent of geographic clustering of districts with low rates of iodization coverage as seen in Figure B5. In this figure we sort districts by iodization coverage into three equal sized groups (tertiles) depending on whether the districts have low, medium, or high coverage rates. Most of the lowest tertile districts are concentrated in eastern Java, Nusa Tenggara, and southern Sulawesi (Figure B4), where small-scale salt farming occurs. This indicates the need for programs that address iodization issues among small-scale salt producers. 42 Finally, Posyandu use also shows significant variation across districts. At the national level, about 40 % of children under-5 years visit Posyandu within a month before the survey. However, district-level estimates vary from 0 % to 100 %. Posyandu utilization rates are below 25 % in about a quarter of districts, whereas about 7 % of districts have the utilization rates higher than 70 %. Posyandu utilization shows a more widely spread distribution pattern compared to three indicators above, as well as no clear geographic pattern. Figure B3. Distribution of district-level estimates of percent of households using adequately fortified iodized salt .1 8 .0 st icrt isdfo 6 .0 n iotroporp 4 .0 2 .0 0 0 25 50 75 100 precent of households with adequately fortified iodized salt Figure B4. Distribution of district-level estimates of percent of children 0-4 years utilizing Posyandu .1 8 .0 st icrt 6 isdfo .0 n iotroporp 4 .0 2 .0 0 0 25 50 75 100 precent of children visited Posyandu 43 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Table B2. Districts with high malnutrition prevalence* for both women and children Province District Prevalence (%) Children Women Riau INDRAGIRI HILIR 43.9 21.4 South Sumatra MUSI RAWAS 44.5 16.2 Lampung LAMPUNG UTARA 34.9 16.2 West Java BEKASI 33.1 15.8 Central Java GROBOGAN 32.4 16.9 Central Java JEPARA 48.8 18.2 Central Java KENDAL 31.9 21.2 East Java PROBOLINGGO 33.8 20.7 East Java GRESIK 33.3 20.8 East Java SAMPANG 64.7 18.8 East Java PAMEKASAN 38.9 16.8 Banten PANDEGLANG 36.4 20.3 Banten SERANG 36.3 16.5 West Nusa Tenggara LOMBOK TENGAH 32.7 22.0 West Nusa Tenggara LOMBOK TIMUR 31.8 18.0 West Nusa Tenggara DOMPU 44.6 29.2 East Nusa Tenggara SUMBA BARAT 59.7 27.7 East Nusa Tenggara KUPANG 41.1 35.4 East Nusa Tenggara T.T. SELATAN 60.6 53.3 East Nusa Tenggara T.T. UTARA 48.0 60.0 East Nusa Tenggara BELU 40.0 46.5 East Nusa Tenggara ALOR 56.0 45.9 East Nusa Tenggara KUPANG 32.9 34.6 West Kalimantan KAPUAS HULU 46.2 19.5 Central Kalimantan BARITO SELATAN 80.6 17.6 Central Sulawesi DONGGALA 42.6 15.9 South Sulawesi GOWA 43.2 18.8 South Sulawesi MAMUJU 36.0 20.4 South Sulawesi PARE-PARE 46.7 15.9 Southeast Sulawesi MUNA 45.3 21.2 * Districts in the highest tertile based on prevalence ranking. Only 197 districts with malnutrition prevalence for both women and children were included. 44 Figure B5. District quartiles based on the rate of adequately fortified iodized salt use at home Outcome and service utilization patterns by selected socio-economic characteristics at the district-level We further examine outcome and service utilization patterns by socioeconomic characteristics at the district- level. We accomplish this by conducting multivariate regression analysis in order to examine differentials in outcome and service utilization indicators, adjusted for selected socioeconomic characteristics. All control variables are district-level weighted averages of selected characteristics as measured in the SUSENAS 2001. In order to reduce the influence of data points that may be unduly influenced by mismeasurement, we include districts where the unweighted number of observations for an outcome or service utilization indicator is 25 or higher. The simple estimation model employed is the following: Y = + U*U + P*P + GINI*GINI + E*E + R*R + where Y is an outcome or service utilization indicator measured in proportionate terms, U is a dummy variable for urban districts, P is the proportion of district households living under the official poverty line, GINI is the GINI coefficient of inequality based on observed inequality in household consumption, E is the proportion of adults 15-59 years who did not complete primary education, and R is a regional dummy variable. We divide districts into five regions: (1) Sumatra Island, (2) Java Island and Bali, (3) Nusa Tenggara, Maluku, & Irian Jaya, (4) Kalimantan, and (5) Sulawesi. The combined region of Java and Bali is used as a reference group in analyses. Standard errors are adjusted for clustering on region. Statistical significance is defined for p-value less than 0.05. 45 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Malnutrition among women (MUAC < 23cm) and children (underweight) have a significant positive association with poverty: districts with more poor people witnessed greater malnutrition (Table B3). Districts that are more unequal also tend to have a higher proportion of stunted children. Adjusted for other control variables, there is no significant difference in women's and children's malnutrition between urban and rural districts. However there are strong regional differences in mean nutrition outcomes. Prevalence of women's malnutrition is lower in Sumatra, Kalimantan, and Sulawesi than in Java and Bali, while it is higher in Nusa Tenggara, Maluku, and Irian Jaya. Underweight prevalence among children 0-4 years is higher in Kalimantan, Sulawesi, and Nusa Tenggara, Maluku, and Irian Jaya, compared to that in Java and Bali. The current specification is admittedly parsimonious and additional indicators, such as public nutrition program expenditure information, would be valuable for further examination of district-level differentials in malnutrition. Table B3. Differentials in outcome and service utilization at the district-level, by selected socioeconomic characteristics: Multivariate linear regression analyses coefficient standard error p-value N R-squared Proportion of women 15-49 year with MUAC <23cm Urban district 0.017 0.015 0.257 253 0.29 Proportion of population under poverty line 0.202 0.080 0.013 Gini coefficient -0.091 0.135 0.501 Proportion adults without primary education 0.002 0.091 0.980 region: Sumatra -0.059 0.011 0.000 region: Nusa Tenggara, Maluku, & Irian Jaya 0.096 0.035 0.006 region: Kalimantan -0.030 0.015 0.046 region: Sulawesi -0.037 0.013 0.006 Constant 0.156 0.041 0.000 Proportion of children 0-4 years underweight Urban district -0.001 0.032 0.972 165 0.22 Proportion of population under poverty line 0.231 0.110 0.036 Gini coefficient -0.510 0.264 0.055 Proportion adults without primary education -0.099 0.189 0.603 region: Sumatra 0.009 0.023 0.684 region: Nusa Tenggara, Maluku, & Irian Jaya 0.101 0.037 0.007 region: Kalimantan 0.091 0.036 0.013 region: Sulawesi 0.105 0.035 0.004 constant 0.377 0.084 0.000 Proportion of households using adequately fortified iodized salt Urban district 0.062 0.027 0.024 253 0.59 Proportion of population under poverty line 0.163 0.161 0.314 Gini coefficient -0.115 0.274 0.676 Proportion adults without primary education -0.625 0.177 0.000 region: Sumatra 0.270 0.021 0.000 region: Nusa Tenggara, Maluku, & Irian Jaya -0.287 0.041 0.000 region: Kalimantan 0.331 0.022 0.000 region: Sulawesi 0.034 0.044 0.437 constant 0.702 0.085 0.000 Proportion of children 0-4 years visited Posyandu Urban district -0.033 0.044 0.447 222 0.23 Proportion of population under poverty line 0.109 0.144 0.448 Gini coefficient 0.007 0.352 0.983 Proportion adults without primary education -0.395 0.239 0.099 region: Sumatra -0.201 0.031 0.000 region: Nusa Tenggara, Maluku, & Irian Jaya 0.037 0.078 0.632 region: Kalimantan -0.115 0.034 0.001 region: Sulawesi -0.154 0.037 0.000 Constant 0.551 0.110 0.000 46 Turning to iodized salt use, we see it is significantly higher in urban districts. The proportion of adults without primary education completion shows a strong negative association with iodized salt use, indicating a potential gap in reaching low socio-economic households. Again, there is a significant regional gap. Districts in Sumatra and Kalimantan have higher iodized salt used than districts in Java and Bali, while those in Nusa Tenggara, Maluku, and Irian Jaya have lower use. This is consistent with the observation that regions containing small- scale salt producers exhibit the lowest rates of iodized salt use. Finally, for Posyandu use among children 0-4 years, there is no significant differential by socioeconomic characteristics, except perhaps for districts that had a high proportion of adults without primary education. However, districts in Sumatra, Kalimantan, and Sulawesi have significantly lower Posyandu usage. 2. Province-level heterogeneity For a wider range of nutrition indicators we explore province-level heterogeneity using DHS 2002/3 and IFLS 2000 as well as Susenas 2001. The small sample sizes at the district-level restrict this analysis only to indicators at the province level. IFLS 2000 provided information on 13 out of 26 provinces prior to the new administrative area change in 2001. DHS 2002/3 included 26 out of 30 provinces. A weighted average for each province is calculated using the relevant sampling-weight. Table B4 summarizes province-level estimates and presents heterogeneityacrossprovinces.Detailedresultsofprovincialestimatesanddatasourcesarepresentedbyindicator in Table B5. Awide range of provincial level heterogeneity is observed.Absolute differences in indicators (highest provincial estimate -- lowest provincial estimate) range from 6 percentage points in weight measurement among children visiting Posyandu to 78 percentage points in adequately fortified iodized salt use at home. Relative differences exceed 3 fold in 8 indicators: - prenatal hemoglobin level measurement (18 % in South Sumatra vs. 66 % in South Sulawesi); - anemia prevalence among children 5-14 years (9 % in North Sumatra vs. 34 % in South Sumatra); - adequately fortified iodized salt use at home (19 % in West Nusa Tenggara vs. 96 % in Jambi); - Posyandu visit among children under 5 years (18 % in North Sumatra vs. 55 % in East Nusa Tenggara); - exclusive breastfeeding among children under 4 months (22 % in DKI Jakarta vs. 80 % in North Sulawesi); - exclusive breastfeeding among children under 6 months (15 in DKI Jakarta % vs. 62 % in South Sulawesi); - women's MUAC less than 23 cm (7 % in North Sulawesi vs. 32 % in East Nusa Tenggara); and - isting among children under 5 years (3 % in Bali vs. 16 % in South Sumatra) Provincial performance varies substantially across Indonesia, yet the ranking of provinces does not consistently identify the same underperforming provinces. For example, provinces in Kalimantan, Sulawesi, and Nusa Tenggara typically show a high prevalence of protein energy malnutrition for women and children, while those 47 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES in Sumatra have generally lower prevalence (Figure B3)21. The same regional pattern, however, does not hold for micronutrient supplementation program coverage. Provinces in Sumatra with low malnutrition prevalence in fact show relatively low coverage of micronutrient programs (Figure B5)22. Table B4. Summary of province- level estimates (%) of outcome and service utilization indicators Indicator Median Mean* SD Min Max Number of provinces Iron Supplementation Service utilization Prenatal iron supplementation received 79 77 11 58 98 26 Hemoglobin level measured during pregnancy 45 43 12 18 66 13 Outcome Anemia prevalence, women 15-49 years** 16 18 5 12 32 13 Anemia prevalence, children 1-4 years** 51 53 7 42 70 13 Anemia prevalence, children 5-14 years** 16 18 7 9 34 13 Vitamin A Supplementation Service utilization Post-partum vitamin A supplementation received 41 42 9 26 59 26 Vitamin A capsule received, children 6-59 months 77 76 9 51 89 26 Vitamin A rich food intake, children 0-2 years (1-day recall) 66 66 6 57 80 26 Outcome Prevalence of night blindness during pregnancy 2 2 1 0 6 26 Iodine Fortification Service utilization Iodized salt use at home 97 86 18 36 100 29 Iodized salt use at home, adequately fortified 89 75 22 19 96 29 Growth Education and Promotion Service utilization Visited Posyandu within last month, children under 5 32 36 11 18 55 27 Visited Posyandu within last 2 months, children under 5 50 52 11 32 73 27 Supplementary food received at the Posyandu 63 66 16 37 93 13 Weight measured at the Posyandu 98 98 2 95 100 13 Behavioral Outcome Exclusive breast feeding, children under 4 months 52 53 15 22 80 26 Exclusive breast feeding, children under 6 months 37 39 13 15 62 26 Protein rich food intake, children 0-2 years (1-day recall) 58 58 8 40 70 26 Outcome Body Mass Index <18.5 (Thin), women 15-49 years 15 15 3 10 18 13 Body Mass Index >=25 (Overweight), women 15-49 years 19 20 4 14 27 13 Mid-Upper Arm Circumference < 23 cm, women 15-49 years 12 13 5 7 32 27 Stunting (Height-for-Age), children under 5 years 35 36 8 24 56 13 Underweight (Weight-for-Age), children under 5 years 28 29 6 18 42 27 Isting (Weight-for-Height), children under 5 years 10 10 3 3 16 13 * Unweighted average of provincial estimates ** Anemia refers to serum hemoglobin level below 11g/dl, except for pregnant women (serum hemoglobin level below 12g/dl). Latest pregnancy within the last 5 years, among women between 15 to 49 Latest delivery within the last 5 years, among women between 15 to 49 Among children who visited Posyandu 21A total of 26 provinces are ranked based on sum of a women's malnutrition prevalence (percent of women 15-49 years with MUAC less than 23 cm) rank and a children underweight prevalence (percent of children 0-4 years whose Weight-for-Age Z-score is below 2) rank. Then provinces are categorized into tertiles: the higher the rank, the higher malnutrition prevalence. 22For the micronutrient program, a total of 26 provinces are ranked based on sum of a children's vitamin A supplementation coverage rank and a prenatal iron supplementation coverage rank: the higher rank the higher program coverage. Then provinces are categorized into tertiles: the higher the rank, the higher coverage. 48 Outcome and service utilization by socio-economic characteristics at the province-level We further explore outcome and service utilization patterns by poverty level at the province-level. Figures B6- B7 show scatter plots of malnutrition and anemia prevalence by the percent of households living below the poverty line23. Both malnutrition and anemia prevalence show a significant positive association with poverty at the province-level24. Figure B6. Scatter plot of children underweight and poverty at the province-level 50 ) %( sr eay 40 4-0 nerdilhc 30 ong 20 ma htgie wr 10 Unde 0 0 5 10 15 20 25 30 Percent of population living under poverty (Solid line is a fitted line from a linear regression: slope = 0.36 (p-value = 0.027), R-squared = 0.21, N = 23) Figure B7. Scatter plot of anemia among women and poverty at the province-level 40 ) %( sr eay 30 49 5-1 ne 20 om w ngo ma 10 ai emn A 0 0 5 10 15 20 25 30 Percent of population living under poverty (Solid line is a fitted line from a linear regression: slope = 0.58 (p-value = 0.017), R-squared = 0.42, N = 13) 23As in the district-level analysis, proportion of households living below the poverty line is estimated using SUSENAS 2001. 24Variation in age distributions is a potentially important factor contributing to district- or province-level heterogeneity, since malnutrition and anemia prevalence is highly correlated with age (Table A3, Annex A). However, due to small age-district specific or age-province specific sample sizes, we are not able to estimate age-specific prevalence of malnutrition or anemia. Nevertheless, our preliminary analyses showed fairly comparable age distributions between 15 to 49 years across districts, and we believe district-level or province-level prevalence estimates are reasonably comparable. 49 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES On the other hand, service utilization patterns by province poverty level vary by program. For prenatal iron supplementation, there is a marginally significant negative relationship between the utilization rate and poverty (Figure B9). For vitamin A supplementation, however, there is no clear coverage pattern by province poverty levels (Figure B10), indicating vitamin Asupplementation programs are implemented to a similar extent across provincial poverty levels. Iodized salt use did not show a linear relationship with provincial poverty, but the scatter plot shows a rather unique convex shape, indicating relatively lower coverage in provinces with both low and high poverty (Figure B11). Figure B9. Scatter plot of prenatal iron supplement coverage and poverty at the province-level 0 10 ) %( egarevoctne 80 60 em pl ups 40 onrilataner 20 P 0 0 5 10 15 20 25 30 Percent of population living under poverty (Solid line is a fitted line from a linear regression: slope = -0.46 (p-value = 0.126), R-squared = 0.11, N = 23) Figure B10. Scatter plot of vitamin A supplement coverage and poverty at the province-level 0 10 ) %( ega 80 er ovctne 60 melppus 40 A ni mati 20 V 0 0 5 10 15 20 25 30 Percent of population living under poverty children 6-59 months postpartumwomen 50 Figure B11. Scatter plot of iodized salt use and poverty at the province-level 0 10 ) 80 %(e m hota 60 seutl sa 40 d zeidoI 20 0 0 5 10 15 20 25 30 Percent of population living under poverty Finally, while there is no clear relationship between Posyandu use among children 0-4 years and poverty at the provincial level (result not shown), there is a significant negative association between poverty and complementary meal receipt among those children who attended Posyandu (Figure B12). This implies provinces with higher poverty are less likely to have supplementary food at Posyandu, thus missing the full opportunity to treat malnutrition among children. Figure B12. Scatter plot of poverty and complementary meal receipt among children 0-4 years who attended Posyandu at the province-level 0 ) 10 %( andu 80 syo P morftpiecer 60 odof 40 yratne 20 mel pp Su 0 0 5 10 15 20 25 30 Percent of population living under poverty (Solid line is a fitted line from a linear regression: slope = -1.94 (p-value = 0.006), R-squared = 0.51, N = 13) 51 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Table B5. Province-level estimates (%) of outcome and service utilization indicators rat rat i rat a a a b a rta v ulu tung a m m m m k a pung k J u Riau a a t National Su S J Su m Beli J s Program Indicator t a average* s th Beng La k rth We DKI No We Sou Bang Iron Supplementation Service utilization Prenatal iron supplementation received Ü 79.9 59 85 71 59 79 85 81 66 91 76 Hemoglobin level measured during pregnancy Ü 43.8 * 35 48 - - 18 - 40 - 49 45 Outcome Anemia prevalence, women 15-49 years*** 17.1 * 14 12 - - 32 - 25 - 18 19 Anemia prevalence, children 1-4 years*** 53.0 * 44 42 - - 70 - 61 - 51 53 Anemia prevalence, children 5-14 years*** 18.9 * 9 11 - - 34 - 31 - 15 22 Vitamin A Supplementation Service utilization Post-partum vitamin A supplementation received á 42.5 29 41 45 52 38 34 26 29 52 42 Vitamin A capsule received, children 6-59 months 69.5 51 72 73 76 78 81 78 71 75 78 Vitamin A rich food intake, children under 3 years (1-day recall) 36.9 68 59 68 59 70 71 69 58 76 64 Outcome Prevalence of night blindness during pregnancy Ü 1.7 2.8 3.9 1.8 1.8 0.7 1.7 1.3 4.7 0.4 1.6 Iodine Fortification Service utilization Iodized salt use at home 84.8 99 100 97 100 99 100 96 99 73 88 Iodized salt use at home, adequately fortified 69.7 91 92 90 96 92 94 85 92 55 68 Growth Education and Promotion Service utilization Visited Posyandu within last month, children under 5 39.8 18 31 38 32 29 30 38 21 35 48 Visited Posyandu within last 2 months, children under 5 56.0 32 45 53 51 45 47 53 38 54 64 Supplementary food received at the Posyandu ÜÜ 71.1 * 52 73 - - 37 - 55 - 76 61 Weight measured at the Posyandu ÜÜ 98.6 * 95 96 - - 98 - 95 - 99 98 Behavioral Outcome Exclusive breast feeding, children under 4 months 49.9 44 35 35 76 55 65 67 36 22 53 Exclusive breast feeding, children under 6 months 37.1 28 26 26 61 37 44 56 27 15 40 Protein rich food intake, children under 3 years (1-day recall) 30.5 56 59 61 44 63 58 52 55 69 54 Outcome Body Mass Index <18.5 (Thin), women 15-49 years 14.3 * 10 11 - - 17 - 12 - 15 13 Body Mass Index >=25 (Overweight), women 15-49 years 20.7 * 24 24 - - 18 - 14 - 27 20 Mid-Upper Arm Circumference < 23.0cm, women 15-49 years 13.1 8 9 8 10 10 8 13 18 9 12 Stunting (Height-for-Age), children under 5 years 33.2 * 41 38 - - 40 - 35 - 24 29 Underweight (Weight-for-Age), children under 5 years 27.2 29 23 28 25 30 19 27 26 24 22 Wasting (Weight-for-Height), children under 5 years 10.5 * 7 8 - - 16 - 11 - 10 11 * Estimates based on IFLS 3 are representative for only 13 provinces. *** Anemia refers to serum hemoglobin level below 11g/dl, except for pregnant women (serum hemoglobin level below 12g/dl). Latest pregnancy within the last 5 years, among women between 15 to 49 Latest delivery within the last 5 years, among women between 15 to 49 Among children who visited Posyandu (Standard error not shown) 52 i ra ra i i i s a a s s s v rta v n ant ant ant ant a a ul J k a et n n n n we we a a a we we talo a a J a a y y t Bali m m m m ul ralt s Tengga Ban Tengga Sula ron ak a a Sula Sula Sula t o a ka M J Data source Yog Ea s s s Kali Kali Kali Kali G u u t th Cen ht t rth talr Ma htr DI N s N ralt s o Irian t t No Sou thea s s We Ea N Sou Cen Cen We Ea Sou 89 98 88 59 88 87 78 66 58 84 82 92 65 73 72 78 - - - DHS 2002 39 55 47 - 44 45 - - - 28 - - - 66 - - - - - IFLS 2000 14 18 15 - 18 16 - - - 15 - - - 16 - - - - - IFLS 2000 51 51 54 - 51 58 - - - 49 - - - 53 - - - - - IFLS 2000 16 13 20 - 18 16 - - - 14 - - - 12 - - - - - IFLS 2000 38 49 59 34 35 42 45 38 37 36 55 51 32 42 45 55 - - - DHS 2002 80 87 84 70 81 89 80 76 58 80 74 81 58 77 73 84 - - - DHS 2002 72 80 72 69 66 66 59 68 64 62 63 64 67 68 57 57 - - - DHS 2002 1.4 1.0 0.7 1.5 0.7 1.8 5.9 3.1 0.3 4.2 3.8 1.1 1.7 2.0 1.0 3.8 - - - DHS 2002 78 90 85 82 60 36 46 99 100 99 100 100 99 77 78 98 48 84 99 SUSENAS 2003 57 67 72 59 42 19 34 89 92 89 96 95 89 63 64 91 39 67 91 SUSENAS 2003 54 55 43 27 32 30 55 27 39 31 41 53 43 25 39 23 - - 30 SUSENAS 2001 69 73 60 43 50 47 70 44 62 44 57 69 58 43 48 40 - - 48 SUSENAS 2001 80 93 82 - 79 63 - - - 59 - - - 46 - - - - - IFLS 2000 99 100 100 - 100 97 - - - 99 - - - 95 - - - - - IFLS 2000 48 39 41 51 42 66 62 44 43 62 39 80 61 72 72 62 - - - DHS 2002 36 23 31 34 26 60 38 30 33 45 34 49 49 62 58 40 - - - DHS 2002 45 70 57 55 64 51 40 53 69 67 63 65 66 65 56 59 - - - DHS 2002 15 17 15 - 13 17 - - - 17 - - - 18 - - - - - IFLS 2000 19 19 23 - 19 15 - - - 18 - - - 16 - - - - - IFLS 2000 16 12 15 12 8 19 32 12 10 13 12 7 13 12 13 9 - - 17 SUSENAS 2001 32 29 33 - 33 56 - - - 43 - - - 39 - - - - - IFLS 2000 26 23 27 29 18 32 37 36 36 30 26 42 34 37 30 36 - - - SUSENAS 2001 11 11 10 - 3 6 - - - 12 - - - 9 - - - - - IFLS 2000 53 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES ANNEX C. COST-EFFECTIVENESS OF NUTRITION PROGRAMS 1. Cost of nutrition programs In order to better understand the cost of nutrition delivery and the cost-effectiveness (CE) of existing nutrition services, the World Bank recently collected information on the cost of nutritional service delivery in five Indonesian districts. These five districts were purposively selected in order to capture reasonable diversity in geography and socioeconomic levels: Surabaya and Lumajang in East Java, Kupang in East Nusa Tenggara, Gunung Kidul inYogyakarta, and Lampung Selatan in Lampung. Except Kota Surabaya, all districts are primarily rural. An overview of the socio-economic characteristics of each district is presented in Table C1. Table C1. Socioeconomic characteristics of five districts for the cost-effectiveness analysis Socioeconomic characteristics Kota Gunung Lampung Surabaya Lumajang Kupang Kidul Selatan Population distribution by residential area Urban 100.0 20.6 2.0 4.9 13.1 Rural 0.0 79.4 98.0 95.1 86.9 Household head's sex Male 85.4 84.5 88.3 83.7 92.0 Female 14.6 15.5 11.7 16.3 8.0 Household head's education Less than primary education completion 8.5 49.0 46.7 48.0 40.3 Primary education completed 23.4 35.0 33.2 29.4 35.5 Distribution of household resources, using consumption quintile at the national-level Lowest 4.7 39.4 54.5 35.7 37.8 Second lowest 8.7 25.8 17.3 24.7 25.8 Middle 16.4 18.2 11.7 21.6 19.7 Second highest 27.9 10.8 9.9 13.3 11.4 Highest 42.2 5.7 6.5 4.8 5.2 (Source: SUSENAS 2003) The cost survey focused on five programs: (1) Vitamin A supplementation for children and women, (2) Iron supplementation for pregnant women, (3) Iodine supplementation for school children, (4) Growth monitoring for children under-5 years, and (5) Feeding programs. Program activities and target populations are summarized in Table C2. 54 Table C2. Main activities by program Program Activity Vitamin A supplementation Infants 6-59 months Distribution of high dose vitamin A in Posyandu Distribution of high dose vitamin A door to door by cadres Postpartum women Distribution of high dose vitamin A through midwives Distribution of high dose vitamin A door to door by cadres/ Iron supplementation Pregnant women Distribution of Iron supplementation through cadres/midwives Distribution of Iron supplementation door to door by cadres/ midwives Iodine supplementation School age children Distribution of Iodine syrup at school. Growth monitoring Children under 5 years Weight measurement and record in a growth chart Consultation on child growth by the cadre based on child's growth chart Therapeutic feeding from Puskesmas Children under 5 years Distribute the supplementary food from the government to the undernourished U-5 children of poor family Complementary feeding from Posyandu Children under 5 years Distribute the complementary food from the government to the U-5 children of poor families who visit to Posyandu InAugust 2004, two trained interviewers familiar with nutrition programs and government accounting practices collected information on the cost of implementing the selected five nutritional programs in each district. Respondent sources included staff from the Posyandu and Puskesmas in each district, selected at random. In addition each District Health Department was visited by the surveyors as well as the Provincial Health Departments in the relevant provincial capitals and the National Department of Community Health and Nutrition in Jakarta. Cost information was collected separately for each nutrition activity. Reported program delivery costs were classified into four broad categories: (1) variable or recurrent costs (such as for various supplies and the costs for meetings, reports, and staff monitoring); (2) fixed costs (such as for equipment and initial staff training); (3) off- budget cost (which attempts to value donated services such as volunteer times); and (4) the cost of additional relevant extra activities that vary across districts. Table C3 depicts the annual cost per child or woman by district for each program. There are several points to be discussed. First, the unit cost for a particular program varies greatly across the 5 districts. For vitamin A supplementation, relative differences between lowest and highest estimates are 1.9 fold, 2.4 fold, and 1.6 fold for infants 6-11 months, children 12-59 months, and pregnant women, respectively. Growth monitoring cost for children under-5 years also varies by 1.6 fold, and therapeutic feeding cost varies by 2.1 fold. This should not be surprising since many factors that determine program costs vary across districts, including the local price levels and the size and density of the catchment area served by the program. Another point to note is that the off-budget costs, largely made up by the opportunity costs of volunteer services, makes up a substantial component of total costs for micronutrient distribution programs. In particular, the proportions of off-budget cost range from 58% in Lumajang to 81% in Kupang for postpartum vitamin A 55 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES supplementation. Similarly, for prenatal iron supplementation, proportions of off-budget cost range from 58% in Lumajang and 83% in Lampung Selatan. This suggests that cost estimates based on budget only may lead to significant overestimation of cost-effectiveness for some programs. Table C3. Program cost for the five selected programs, per person/year (in Rupiah) Program Kota Lumajang Kupang Gunung Kidul Lampung Surabaya Selatan Vitamin A supplementation Infants 6-11months Variable 980 1948 823 1450 710 Fixed 144 103 112 105 104 Off-budget 725 462 382 760 727 Extra activity 208 0 0 0 0 TOTAL 2057 2513 1317 2315 1540 Children 12-59 months Variable 1016 1748 877 1608 1597 Fixed 144 103 112 107 104 Off-budget 517 603 371 1179 1541 Extra activity 208 0 0 0 0 TOTAL 1886 2455 1360 2894 3241 Post-partum mothers Variable 835 1933 541 1158 472 Fixed 84 3 12 47 44 Off-budget 2547 2668 2356 2094 2587 TOTAL 3467 4603 2910 3298 3103 Iron supplementation Pregnant women Variable 621 1081 523 627 647 Fixed 84 3 12 75 70 Off-budget 2433 2433 2364 2013 2087 TOTAL 3139 3518 2899 2714 2804 Iodine supplementation School age children Variable - 1300 - 1200 - TOTAL - 1300 - 1200 - Growth monitoring Children under 5 years Variable 16146 16564 12708 17718 15783 Fixed 6247 4578 1827 1584 1435 Off-budget 1507 921 782 869 1255 Extra activity 321 63 0 0 0 TOTAL 24220 22126 15317 20171 18473 Therapeutic feeding (Puskesmas) Children under 5 years Variable 241240 196115 402774 188868 - Fixed 6309 9641 1901 1584 - Off-budget 9500 9500 3558 1000 - TOTAL 257049 215256 408234 191452 - Complementary feeding (Posyandu) Children under 5 years Variable 175455 177339 219108 204843 - Fixed 6247 9653 1902 1584 - Off-budget 1563 1313 873 1094 - TOTAL 183264 188305 221882 207521 - 56 Finally, in comparison to unit costs in selected low income Asian countries (Horton, 1999)25 , average unit cost 26 of vitamin Asupplementation program among five districts (0.22 USD per infant 6-11 months and 0.26 USD per child 12-59 months) appears to be at a similar level (0.20 USD per child under 3 years). However, postpartum vitamin A supplement cost is relatively higher in the study districts (0.39 USD per pregnancy vs. 0.10 USD per pregnancy), although the cost is still modest. On the other hand, average cost for prenatal iron supplement is significantly lower in the study districts (0.34 USD per pregnancy) compared to 2.50 USD per pregnancy. 2. Nutrition program effectiveness International estimates show that nutrition programs can be amongst the most cost-effective health interventions available (World Bank 1993). However both the costs and effectiveness of any program can vary widely according to the health conditions in a specific location and how the program is implemented. Annex B of this report addresses the substantial regional heterogeneity in nutrition conditions and the extent of successful nutrition program delivery. With decentralization, it is important to consider the cost-effectiveness of nutrition programs at the district level since now districts are the focal point for decisions on health priorities -- which programs are to be implemented and how these are supported. This section uses the cost information for the five districts described above to estimate cost-effectiveness (CE) for the nutrition programs at the district level. The study applies an adapted version of the Generalized CE Analysis framework developed by WHO, a standardized approach that can be applied to all interventions in different settings (WHO-CHOICE 2003). It evaluates the CE of interventions relative to a ìnullî scenario in which no interventions are made. The approach has been used to derive global and regional estimates for a broad range of health interventions, with the detailed assumptions underlying the modeling published on the internet to enable analysts to update and modify the estimates as more appropriate data becomes available. The impact of the interventions is assessed using a population model to simulate the lifespan of individuals in a population, allowing individuals to be categorized into one of three mutually exclusive health states. Population health is expressed as the number of healthy years lived, while differences due to the interventions are estimated in terms of DALYs (disability adjusted life-years) averted. Use of the same metric to describe the health impact of interventions for different diseases and risk factors allows direct comparison across different types of interventions and different health conditions. WHO has defined 'Epidemiologic sub regions', grouping countries into 14 regions according to mortality profile. Indonesia is grouped with Thailand and Sri Lanka in SEAR B, described as 'low child mortality, low adult mortality'. As outlined in Table C4, a population model and aggregated disease profile has been developed to describe the disease experience and estimate the maximal attainable impact of health interventions for the region. The coverage of the nutrition interventions in each of the five study districts (estimated from SUSENAS 2003) has been used to estimate actual health impact in each district. This is combined with the cost data described earlier to estimate the CE of selected nutrition programs in the five districts. Unfortunately health impact estimates 25Countries included in the estimates are Bangladesh, Cambodia, China, India, Laos, Nepal, Pakistan, Sri Lanka, and Viet Nam (Horton, 1999). 26Cost comparison is in 1999 USD (1USD=7813Rp), adjusted for inflation (14.8 % between 1999 and 2004) (Source: BPS Statistics Indonesia, 2005. Monthly Indonesia's Consumers Price Indices and Inflations, 1999 - 2005. http://www.bps.go.id/sector/cpi/table3.html, accessed on August 15, 2005). 57 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES were available for only six of the eight interventions for which cost data were collected (Table C5). To estimate CE these data were collapsed into three interventions in order to reconcile the collected cost data with the available effectiveness information: (1) vitamin A supplementation for children 6-59 months, (2) prenatal iron supplementation, and (3) growth monitoring and complementary feeding for children 6-12 months27. Table C4: Data elements used in the CE modeling and their sources. Model elements Variables Assumptions Sources Population model Total population size WHO regional-level age and sex GBD 2000 Database, for the risk factor Population mortality rate specific estimates are applicable SUSENAS 2003 Population fertility rate to the study districts. Health state valuation WHO regional-level age, sex and WHO 2002 disease specific estimates are applicable to the study districts 'Observed' Incidence, case-fatality, Determined using DISMOD II GBD 2000 Database epidemiology for remission, prevalence, equations, based on WHO-region the risk factor duration, co-morbidities specific data Intervention effects Effect WHO region specific estimates of Ezzati et al 2004 the attributable burden for the nutrient are applicable to the study districts. They were estimated by recent meta-analyses, assumed to be the maximal attainable health effect with an optimal program. Coverage -- SUSENAS 2003, IFLS 2000, DHS2002. Intervention costs Variable, fixed, off-budget, -- Primary data collected, extra activities 2004 Table C5: Nutrition interventions considered Program Interventions Cost data Health impact being implemented collected (effectiveness) estimates available Vitamin A supplementation Vitamin A supplementation, Y Combined 6-59 months only infants 6-11 months Vitamin A supplementation, Y Combined 6-59 months only children 12-59 months Vitamin A supplementation, Y N post-partum mothers Iron supplementation Iron supplementation, Y Y pregnant women Iodine supplementation Iodine supplementation, Y N school aged children Growth monitoring, Y Onlycombinedwithcomplementary children under 5 years feeding, 6-12 months Complementary feeding, children under 5 years Y 6-12 months only Therapeutic feeding, children under 5 years Incomplete For underweight children 6-12 months 27Growth monitoring was combined with complementary feeding in children under-5 years, as generally is the case in Posyandu services. 58 Table C6 describes the estimated health impact (effectiveness) of the three nutrition interventions in the five districts, expressed in DALYs averted. It is a function of district population, maximal health impact, and program coverage, and there are a few contrasts and patterns to note. First, the maximal potential impact28 is greatest for vitaminAsupplementation, followed by iron supplementation. The potential impact of growth monitoring and complementary feeding is significant but relatively much less than other two interventions. Second, while supplementation coverage for both vitamin Aand iron is generally high across districts, the growth monitoring and complementary feeding reached only about half of the children. This coverage gap by program results in a significant difference in the estimated actual effect across programs. For example in Surabaya, the ratio of the estimated actual effect of vitamin A vs. growth monitoring and complementary feeding is about 5.5 (3090 vs. 561), while the ratio of the maximal potential effect is about 3.2 (3395 vs.1058). Finally, the estimated actual effect is greatly affected by population size and it is difficult to directly compare across districts due to a wide range of population size (from 0.3 million in Kupang to 2.7 million in Kota Surbaya). Table C6: Estimated actual effect of the nutrition interventions at district level Intervention District District DALYs averted/ yr, Coverage b DALYs averted, actual Population maximal effect a (%) effect, average 1 yr c Vitamin A supplementation, 6-59 months Surabaya 2,692,638 3,395 91 3,090 Lumajuang 1,000,260 1,261 79 996 Kupang 332,840 420 81 340 Gunung Kidul 686,732 868 92 799 Lampung Selatan 1,195,376 1,507 75 1,130 Iron supplementation, pregnant women Surabaya 2,692,638 2,647 94 2,488 Lumajuang 1,000,260 983 83 816 Kupang 332,840 327 78 255 Gunung Kidul 686,732 675 96 648 Lampung Selatan 1,195,376 1,175 80 940 Growth monitoring & complementary feeding, children 6-12 months Surabaya 2,692,638 1,058 53 561 Lumajuang 1,000,260 393 52 204 Kupang 332,840 131 41 54 Gunung Kidul 686,732 270 52 140 Lampung Selatan 1,195,376 470 49 230 a Based on 100% program coverage in a population with SEAR B age/ sex profile and prevalence of deficiencies. b Assumes that coverage for age sub-groups applies to overall age group (eg 6-12 mo versus 13-59 mo). cActual effect = (Maximal effect x coverage) The cost and health impact data are combined in Table C7 to give CE estimates. As might be expected from the ranges across districts in the data used to derive these estimates, the cost per DALY averted varies widely across districts for each intervention and the relative CE for each intervention is quite different. The district-level variation in CE is greatest for Vitamin A supplementation, with a ratio of 2.44:1 for the least cost-effective district to the most, in Lampung Selatan and Kupang, respectively. The corresponding ranges for iron 28 With the methods used, maximal DALYs averted are a direct fraction of population size, with a different fraction for each intervention ('intervention effects' in Table C4) 59 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES supplementation and growth monitoring and complementary feeding are 1.50 and 1.47 respectively. Surabaya generally ranks as the first or second most cost-effective district across the three interventions, that is the least cost per DALY averted. The other districts rank quite differently depending on the intervention. For example, Kupang is the most cost-effective for vitamin A supplementation, but near the least for the other interventions. In addition, the three interventions are quite distinct in their relative CEs, with no overlap between the ranges and a clear ranking in CE: iron supplementation is clearly the most cost-effective (31 I$ /DALYaverted), followed by vitamin A supplementation (84 I$ /DALY averted), then growth monitoring and complementary feeding (4813 I$ /DALY averted). Although vitamin A supplementation has the highest maximal potential effect (Table C6), prenatal iron supplementation is estimated to be most cost-effective intervention due to the relatively smaller but higher risk target group. Table C7: Estimated cost-effectiveness of the nutrition interventions at district level Intervention District District Average cost/ DALYs averted, Cost / Cost / Population beneficiary/ actual effect, DALY DALY in age year (Rupiah) average 1 yr b averted averted c group a (Rupiah) (International $, 2004) Vitamin A supplementation, children 6-59 months Surabaya 256,366 1,905 3,090 158,051 62 Lumajuang 95,235 2,461 996 235,315 93 Kupang 31,690 1,355 340 126,294 50 Gunung Kidul 65,384 2,830 799 231,585 91 Lampung Selatan 113,812 3,054 1,130 307,594 122 Average 211,768 84 Iron supplementation, pregnant women Surabaya 58,700 3,139 2,488 74,059 29 Lumajuang 21,806 3,518 816 94,012 37 Kupang 7,256 2,899 255 82,491 33 Gunung Kidul 14,971 2,714 648 62,703 25 Lampung Selatan 26,059 2,804 940 77,733 31 Average 78,200 31 Growth monitoring & complementary feeding, children 6-12 months Surabaya 28,488 207,484 561 10,536,193 4162 Lumajuang 10,582 210,431 204 10,915,592 4312 Kupang 3,521 237,199 54 15,466,253 6109 Gunung Kidul 7,266 227,692 140 11,817,215 4668 Lampung Selatan 12,647 - 230 - - Average 12,183,813 4813 aBased on district total population (Susenas 2003) and SEAR B population age/ sex profile (6-11 mo = 0.01058 of total; 6-59 = 0.09521 of total; pregnant = 0.0218). b For vitamin A calculated using cost per beneficiary weighted (11% infant cost; 89% 12-59 cost); for growth monitoring & complementary feeding, costs of individual services are added; otherwise costs as reported in Table C3. cExchange rate: International Dollar 1 = Rp. 2531.526 (Source: IMF) 60 Unfortunately, there are few other data with which to directly compare these results. WHO-CHOICE's CE estimates for the SoutheastAsia Region are 128, 230, and 3478 (I$ /DALYaverted) for vitaminAsupplementation, iron supplementation, and complementary feeding and growth monitoring, respectively (WHO 2003). Part of 29 the discrepancy is explained by the difference in cost measurement. Our cost survey focused on marginal cost of a specific intervention only, whereas WHO's cost included the cost of an overall health center visit. In addition, WHO included the cost of a 3-month postpartum iron supplementation for iron supplementation while we restricted the cost to within the prenatal period only. Horton (1999) estimated CE in terms of the cost per death averted for interventions focusing on protein energy malnutrition (PEM), iron, and vitamin A supplementation in the Asian region. A considerable diversity across countries was suggested: the highest vs. lowest ratios were 10.8, 18.9 and 6.5, respectively. As expected these cross national ranges are greater than the inter-district ranges we observe for Indonesia. In contrast to the findings presented in the tables above, the cost per death averted was highest for iron interventions, in some cases more than 80 times of the cost for PEM and vitaminA. The estimates in Horton's analysis will be more highly influenced by population age/sex profiles and prevalence of disorders than this study, where these were constrained to be similar for all districts30. We further explore any association between CE and population characteristics across the five districts. Figure C1 shows the scatter plot of CE and coverage by program. Even though there are a maximum of five data points, we see a negative relationship between CE and coverage for prenatal iron supplementation and growth monitoring. We did not observe any significant association between CE and target population size or wealth profile. Our findings are preliminary and we intend to refine these analyses by assessing how sensitive the estimates are to each set of assumptions, and to include discounting of costs and benefits. Nevertheless, the findings already have important implications related to decentralization. The cost-effectiveness of nutrition interventions is likely to range by a factor of about 1.5 to 2.5 across districts, depending on the intervention. Any national estimate of CE will mask this considerable regional heterogeneity. VitaminAsupplementation programs are the most variable in CE across districts. In addition, iron supplementation programs are the most CE across all districts, followed by vitamin A supplementation and growth monitoring/ complementary feeding being the least CE. 29International dollar computations (I$) are based on year 2004. WHO results employ a program coverage estimate comparable to what observed in the five districts: 80% for vitamin A and iron supplementation and 50 % for complementary feeding and growth monitoring. 30District-level prevalence was available for only underweight among children under-5 years. 61 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES Figure C1. Scatter plot of the cost-effectiveness and coverage by program a) Vitamin A supplementation for children 6-59 months 0 35 0)00 0 n'i 30 Lampung Selatan d,etre 0 av 25 YL A Lumajuang Gunung Kidul D 0 h/a 20 pi Ru( E 0 C 15 Surabaya Kupang 0 10 60 70 80 90 100 programcoverage (%) correlation coefficient -0.44 (p-value 0.46) b) Iron supplementation for pregnant women 0 10 0)00 Lumajuang n'i 90 d,etre av 80 Kupang YL Lampung Selatan A D Surabaya h/a 70 pi Ru( E C 60 Gunung Kidul 50 60 70 80 90 100 programcoverage (%) correlation coefficient -0.71 (p-value 0.18) c) Growth monitoring and complementary feeding for children 6-12 months 00 160 Kupang 0)00 00 n'i 140 d,etre av 00 YL A 120 D Gunung Kidul h/a Lumajuang pi Ru( 00 Surabaya E 100 C 0 800 30 40 50 60 70 programcoverage (%) correlation coefficient -0.98 (p-value 0.02) 62 ANNEX D. INSTITUTIONAL ANALYSIS: VARIATION IN NUTRITIONAL CAPACITIES ACROSS REGIONS This annex involves a systematic review of the division of roles and responsibilities between levels of government and other institutions involved in nutrition program delivery. This review is achieved through the summary of detailed case studies conducted in different regions in Indonesia. Four districts from three provinces were purposively sampled for these case studies. They are Kota Surabaya and Lumajang in East Java, Gunung Kidul in Yogyakarta, and Kota Kupang in East Nusa Tenggara. These are among the same districts selected for the cost-effectiveness analysis in Annex C and so will provide additional information. This review is conducted with a focus on the effect of decentralization, the most important institutional change in Indonesia in the last 50 years, on the delivery and effectiveness of nutrition services. Decentralization poses significant and still unresolved questions regarding the relationships between different levels of government in Indonesia and the roles of key institutions and agencies in nutrition delivery. Role differentiation and improved management capacity are important open issues today that will eventually need to be addressed. Although a possible future that rationalizes institutional roles is one where the provinces and the center specialize in critical public nutrition functions and districts assume primary responsibility for nutrition sector performance in their jurisdictions, the current situation is quite different. Understanding this current situation in some detail is a crucial step in the rationalization of roles and responsibilities of nutrition delivery. The annex is divided into three sections. First we compare institutional responsibilities across administrative levels and regions. Other institutions involved in nutrition delivery are then briefly reviewed. Finally the current strengths and weaknesses of each institution, and how these strengths and weaknesses should affect the course of the reform process, are discussed. I. COMPARATIVE ANALYSIS OF INSTITUTIONAL RESPONSIBILITIES ACROSS ADMINISTRATIVE LEVELS AND REGIONS In this section, we examine three aspects of institutional functioning at the central, provincial, and district level: (1) the institutional structure; (2) the defined institutional responsibilities; and (3) structural and related issues that inhibit the institution from fulfilling these responsibilities. Information on structure and responsibilities was obtained from official documents governing each institution: the Peraturan Pemerintah (PP) (government regulation) at the central level and the Peraturan Daerah (PD) (local government regulation) at the province- or district-level. A. Institutional Structure The new autonomy granted to local governments in 2001, resulted in diverse institutional structures with respect to nutrition policy and services. This variation is particularly noticeable across districts although it is also apparent across provinces. Some districts still preserve an explicit nutrition section in the local government; some have merged the nutrition section into more broadly defined health sections; and some do not have any nutrition related section or sub-section at all. All three of these situations can be seen across the 38 districts in East Java. (On the other hand all 16 districts in NTT have a nutrition section.) 63 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES At the province-level, each of the Provincial Health Offices (PHO) in East and Central Java and East Nusa Tenggara has a nutrition section. In NTT there are actually two sections responsible for nutrition programs: the Improvement in Nutrition Section and the Food and NutritionAlertness Section (SKPG). These two sections are under the coordination of different divisions and there is overlap between the two in their functions and responsibilities. B. Institutional responsibilities Areview of the defined institutional responsibilities indicates that the differentiation of nutrition roles between the planning and health departments seen at the central level is echoed at the province and district levels. The planning department (Bappenas at the central level and the Bappeda at the provinces and districts) generally takes responsibility for overall policy and planning, and hence budgeting, while the health department (Depkes at the central level and Dinkes at the local level) is the technical arm of the government in the health sector. The health department formulates technical standards, provides technical assistance, and evaluates nutrition policies. In addition to these technical responsibilities, the provincial health offices (PHO) are also charged with coordinating nutrition policies across the districts within their purview. The district health offices (DHO), on the other hand, play less of a specific technical role and instead focus their efforts on implementing the family and community nutrition programs in their local areas and otherwise operationalizing the policy directives they receive from above. In practice, the roles and responsibilities of the different levels are not sufficiently differentiated and clear in the minds of the relevant public officials. One result of the decentralization process is that the province has lost much of its authority over the districts -- districts now often ignore requests from the centre and the province, something that was not possible in the past. As a result, provinces and the centre frequently have little idea of the status of program implementation. This is particularly true for programs that do not receive deconcentration funds. Another important effect of decentralization is that local nutrition sections have changed their roles and responsibilities, resulting in considerable diversity across districts and provinces. As there is less pressure to coordinate programs across districts and no one body has the authority to direct such coordination, each department in each district or province has set out its own roles and responsibilities. These roles and responsibilities frequently duplicate or overlap with those of another section or department. For example, in NTT, a small province with only a limited number of well trained staff, the PHO has two sections which deal with nutrition -- one for 'Improvements in Nutrition' which is charged with delivering programs and producing reports on activities, and a second group, the 'Food and Nutrition Alertness' section which is responsible for monitoring all aspects of nutrition and implementing some programs. At the same time the Bappeda is also responsible for monitoring and evaluating programs as well as planning and coordination. However, while there are nutrition staff positioned at all Puskesmas, there is no provincial budget (or district budget in most cases) for data collection and surveys. While the case of NTT is more extreme than in most provinces, it nevertheless reflects the overall lack of coordination between departments, the overlap in responsibilities between sections and departments, and the generally inadequate budget for monitoring, surveilling and evaluating the activities in nutrition at either the provincial or district levels. 64 C. Structural and related issues that inhibit the fulfillment of these institutional responsibilities There are a number of problems that prevent government institutions from successfully executing all of their stated nutrition responsibilities. Looking broadly across all possible administrative levels and regions, four types of barriers were most prominently suggested by study respondents: issues in program implementation, human resources, planning and budgeting, and health information systems. Issues in program implementation Provincial level As stipulated by PP No. 25 in 2000, there are 11 roles and responsibilities of the provincial government in the health sector (including nutrition). However, only two of these roles are currently implemented as programs and therefore receive a budget. These budgeted programs are: 1. Epidemiological surveillance (including nutrition surveillance) 2. Provision of essential pharmaceuticals for basic health needs (including those for micronutrient supplementation programs) The other nine roles and responsibilities are implemented only in the form of standards determination, permission provision, and monitoring (including standardization of nutrient value and guidelines for certification of health and nutrition technology) and receive no budgetary support at the provincial level.As a result, the involvement of the provincial level in nutrition programs is now very limited. Virtually all implementation is left to the districts. District level The District Health Office (DHO), through its Nutrition Section, is now the entity most responsible for implementing nutrition programs. Within the district, the Puskesmas is typically seen as the front line institution for implementation. There has recently been a revision of the roles and responsibilities of Puskesmas that reduced the health center mandate from 18 to 6 major programs. One of these six major programs is community nutrition. PP No 25 in 2000 is based on UU No. 22 in 1999.31 However, the extent to which the Puskesmas can fulfill this new nutrition role is limited by a lack of relevant information, a shortage of staff in most districts and Puskesmas, and limited funds allocation by the district government. One pressing issue in nutrition program design and implementation is the lack of relevant information, and limited use of the scant existing information when institutions formulate nutrition programs. This problem exists at all levels of government. For example Bappenas formulated the RPJM (Five-year development plan) in nutrition based on the vision and mission of the newly elected president, rather than on the current nutrition situation. Neither the center nor the provinces have high quality nutrition information as neither level receives complete data from districts, sub-districts, and villages. This breakdown in reporting arose after decentralization when the reporting requirements and responsibilities were changed. For example, the ability of the Dinas in East Java to compile data, including data on nutritional status and program performance, is limited since districts no longer send regular reports to the province. The reports received from the districts are restricted to programs funded only through the de-concentration budget or provincialAPBD, and the reports are sometimes incomplete and/or late. 31Since UU No. 32 in 2004 revised UU No. 22, there will be another revision for PP No. 25. This revision may change the roles and responsibilities of the province (and districts) in the health sector, including nutrition. 65 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES As a consequence of this lack of relevant nutrition information, the nutrition section of each PHO prepares program plans which are not firmly based on evidence. In certain provinces the program plan derives from the previous plans with little change and without an explanation of any changes that are made. At the same time, technical policy (still essentially set by the centre) does not adequately reflect existing nutritional needs or take into account the problems at lower administrative levels. Issues in human resources At the central level of government, the Directorate of Community Nutrition has adequate numbers of well qualified staff. Ironically, nine additional staff were recruited in 2005, after decentralization, while the provinces and districts suffer from limited human resources. A common problem in the three provinces visited is that the heads of the nutrition sections do not have educational backgrounds in nutrition. Additionally, numerous staff in the province offices have only completed the senior level of high school. However the largest staffing problem is found at the Puskesmas level, the front line of nutrition program implementation. East Java confronts the most serious human resource situation: only 503 out of 923 Puskesmas have nutrition staff properly trained in nutrition32. Among these 503 Puskesmas with properly trained staff, 37 have nutrition staff who are only honorary employees and another 37 only have voluntary staff. These honorary and voluntary members commonly leave the Puskesmas when they receive a permanent job elsewhere. In Surabaya and Lumajang districts, each DHO has only two nutrition staff. Furthermore, some of these staff are expected to leave soon since they will continue study in other areas. Yogyakarta generally has a better situation in human resources than East Java, except for the district of Gunung Kidul. Nevertheless, only 16 out of 29 Puskesmas have nutrition staff, but two of these Puskesmas with nutrition staff are expected to lose them soon. This problem may grow even worse over time as other PHC nutrition staff will retire in the near future and those staff who continue their education at the university level will most likely not return to Puskesmas after completing their degrees. The staffing situation in NTT is comparatively better. The Puskesmas in NTT have two nutrition staff on average, although staff training is frequently not adequate for the required tasks. Each DHO in NTT has more than two nutrition staff, except Kota Kupang where the Nutrition section does not have any staff at all. The rate of skill upgrading has stalled since decentralization. All training programs, which prior to decentralization were run by the center and funded out of the central budget, have now been transferred to the districts as a part of the decentralization process. The training budget is now included in the DAU (dana alokasi umum --general allocation budget), and in the four districts studied, proposed training by the nutrition section was rejected by the planning sub-division at the DHO or rejected in the discussion stage of Satuan Empat (see Box 2).As a result, there has been little nutrition training at the district-level since decentralization, even though the district staff, now fully responsible for program delivery, generally have inadequate skill levels for the tasks they are now expected to carry out. 32 Proper training defined as those with Nutrition Diploma 1 and/or Diploma 3. 66 Issues in planning and budgeting Prior to decentralization, sectoral allocations were decided by the central government. Resources were transferred from the central to the regional (provincial and district) governments through multiple earmarked grants. Most health and nutrition funds were channelled through the MOH with smaller amounts through other ministries; other funds were allocated directly to the districts and provinces.After decentralization in 2001, central-regional transfers have continued to be the primary means of transferring resources from the center to the district but they are no longer earmarked for specific activities in any sector. Almost all district funds derive from the Balancing Fund (dana perimbangan). This fund includes a general grant (DAU -- dana alokasi umum), shared taxes, natural resource and revenue shares (SDA, sumber daya alam) and a special sector-specific allocation grant (DAK -- dana alokasi khusus). The share of DAU and natural resources allocated to the health sector is determined by the district governments themselves. DAK (special allocation grants for health) are distributed by the central government according to specified allocation criteria. Planning and budgeting processes at each administrative level are explained in more detail in Boxes D1-D3. In regard to central-level allocative planning, the process is clearly defined although there are some weaknesses which can create an imbalance in the de-concentration budget across provinces. Provinces with qualified nutrition staff and strong collaboration with universities (such as all of the provinces in Java and some in Sumatera and Sulawesi) have a big advantage in obtaining resources through the submission of well developed proposals; while more likely to successfully obtain resources, the provinces with these advantages may not have the greatest nutritional needs. At the provincial level, there are two major issues. First, there is little consultation with outside parties, especially the DHOs, in the planning and budgeting process and therefore the provincial nutrition programs at times work at cross-purposes with district programs. Secondly, typically there is poor coordination across provincial level institutions (the Bappeda, PHO, and other institutions responsible for food and nutrition programs such as the Food Security Board, the Women Empowerment Board, and PKK) and with the DPRD. Developed programs often overlap in area and responsibility. In addition, programs proposed by theAPBD can be eliminated by the Bappeda or the DPRD. At the district-level, the lessons learned from Lumajang and Surabaya should be noted. These two districts have established good coordination between the DHOs (including the nutrition section), other governmental institutions, and NGOs. This coordination has resulted in an increased budget allocation for the nutrition sector in theAPBD. This is especially so in Lumajang, where successful collaboration originated from a good relationship between the Director of DHO and the Bupati.33 In Surabaya, the nutrition section has built a successful collaboration mechanism: NGOs are responsible for conducting nutrition surveys; survey results are disseminated by NGOs to various parties including the mayor, Bappeda, and DPRD; NGOs, BAPPEDA and DPRD then discuss nutrition problems with the nutrition section. As a result, nutrition programs proposed by DHO are easily accepted by the Bappeda and DPRD. 33 Nevertheless, the nutrition section at DHO does not necessarily perform well. This is not only because the section is headed by a midwife with little background in nutrition but also because nutrition is no longer a stand alone division but now a part of maternal and child health/family planning activities. 67 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES In summary, there are two sets of issues related to health sector planning and budgeting, both of which are important for nutrition programs. The first issue derives from the change in budget processes that have come with decentralization (Box D1-D3). Whereas the funds for nutrition programs were essentially determined by the centre -- largely based on historical grounds -- before decentralization, now they are almost entirely determined by the district as part of a general transfer of funds from the centre. Thus, it is the district health office, the Bappeda and the DPRD who now decide how the district funds will be allocated between and within the sectors. The centre and provinces have little influence on the outcome. Largely as a result of the way this process has developed -- the inexperience of the district sector officials in the cut and thrust of a competitive budget process and the inability of the newly elected politicians to place sectoral concerns in the broader context of the overall budget or to develop a set of criteria for allocation of public funds -- funds for health (including nutrition) are frequently less than had been the case before decentralization. Funds may not be allocated to health and nutrition, and cross-district issues -- including training, monitoring and evaluation, and the like -- tend to receive much lower priority than activities that will bring a more immediate political impact. The second, and related, issue pertains to the more technical questions of the most appropriate ways to intervene in nutrition problems once they are on the political agenda. Here, it is not only the need for information that is important but also applying the lessons from elsewhere in Indonesia and other countries, and the availability of technical expertise at the district level. This requires a much more deliberate and medium term (beyond the life of the local parliamentary term) approach, consideration of the role of information, application of lessons learned in other settings, and a focus on results. All of these considerations were not part of the previous approach and are not yet established in the new, decentralized approach. The district by itself may not be the best placed entity to take these issues into account; it may be that some can be more effectively dealt with at the provincial level, in collaboration with the districts. These institutional issues apply to the health sector generally and to other sectors as well. The recent revisions to the basic laws encompassing decentralization, particularly those related to reconsideration of the respective roles of the province and district, mean that further change is underway. The result may well be that the improved collaboration between districts and with provinces required to deal with many health and nutrition problems is now more possible. Box D1. Planning and budgeting process at the central level Bappenas and MOH together propose the nutrition plan and budget, which is approved by parliament. The nutrition planning and budgeting process will follow these steps in the next years: 1. Priority assessment by Ditzi (Direktorat Gizi Masyarakat -- Directorate of Community Nutrition) will refer to the RPJM (Rencana Pembangunan Jangka Menengah -- Midterm Development Plan). There is a meeting where all units in Ditzi will discuss planning and budgeting needs related to the role of MOH under de-concentration. 2. At the same time, the PHOs prepare nutrition program proposals for each province under the de-concentration budget released through MOH. 3. A meeting between Ditzi and the PHOs is conducted to discuss proposals submitted from the PHOs and then to determine the nutrition program and de-concentration budget allocation for coming years. 4. The dissemination of results of the meeting between Ditzi and PHO to Dirjen Binkesmas (Director General of Public Health), then to be compiled with programs from others directorates and other units within organization's structure of Dirjen Binkesmas. 68 5. Compilation by Dirjen Binkesmas is proposed to Sekjen Depkes (Secretary General of MOH) and the Planning Bureau of MOH. Results of compilation and discussion at Sekjen Depkes is established as RKA-KL (Rencana Kerja Anggaran-Kementerian Lembaga -- action plan for ministry and board). 6. Sekjen Depkes proposes RKA-KL to Bappenas. 7. Bappenas develops the RKP (Rencana Kerja Pemerintah -- government action plan) draft based on proposed RKA-KL from all ministries and boards. 8. Bappenas conducts Musrenbangpus (Musyawarah Perencanaan Pembangunan Pusat -- deliberation on central development planning) to improve the RKP. Ministries and boards at the central level participate in this. 9. At the same time, Musrenbangda (Musyawarah Perencanaan Pembangunan Daerah - deliberation on regional development planning) is conducted at the district level to formulate the provincial and central development plans that will be proposed to Bappenas. 10. Bappenas conducts Musrenbangnas (Musyawarah Perencanaan Pembangunan Nasional- deliberation on national development planning) by inviting Governors and central institutions to synchronise central and regional planning. 11. Results of Musrenbangnas are revised by Bappenas for use in the cabinet assembly and will be established as the RKP (Rencana Kerja Pemerintah -- government action plan). 12. The RKP is determined through the President's Regulation and disseminated to DPR as the basis for RAPBN development. 13. The Government and DPR review RAPBN corresponding to RKP and determine the APBN, which includes central and regional budget allocation. Box D2. Planning and budgeting process at the provincial level There are two channels for planning and budgeting at the provincial level: Channel One: Planning and budgeting to determine the part of the provincial development plan that will be supported from provincial APBD. There exists only slight variations in the planning and budgeting process at PHO among the three provinces. The differences in process arise after the plan is sent to local governments. The planning and budgeting occurs as follows: 1. The nutrition section prepares program plans. These program plans are typically not evidence based. In certain provinces, the program plan is based on the previous plan with little change and with no rationales given. 2. The results of this formulation are then delivered to Head of Planning Sub-division at PHO. 3. The planning Sub-division compiles and selects program plans from all units. The sub-division selects the program without any consultation with the original proposers. 4. The compilation and selection results from the sub-division are then delivered to Bappeda to be discussed at the Tim Anggaran Eksekutif (Executive Budget Team)34 level. 5. Satuan Empat synchronises and coordinates with the proposer institution (for example, PHO) however one of the surveyed provinces directly selects the proposal without consultation with the proposer, 34Tim Anggaran Eksekutif is a group consisted of local government institutions: Bappeda, Development Division, Financial Bureau and Regional Monitoring Board 69 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES 6. Results from Satuan Empat manifest as Provincial Annual Action Plan to be discussed in the DPRD forum. In the observed case of NTT Province, the Financial Bureau revised the report from Satuan Empat by deleting the nutrition program proposal without informing other members of Satuan Empat. 7. The Provincial Annual Action Plan is realized as provincial APBD through the DPRD assembly. Channel Two: Planning and budgeting to determine the part of the national nutrition program where MOH provides support with the de-concentration budget. Based on experience from previous years, the PHO collects information on the amount of de-concentration budget allocated for its province. Based on this information, the nutrition section develops a program plan. In almost all provinces, planning is conducted independently by the nutrition section without involving other parties. The planning formulation from nutrition section is then delivered by the Director of PHO to MOH. Box D3. Planning and budgeting process at the district-level 1. The head of the planning sub-division distributes pre-designed program plan forms to be filled in by each section including the nutrition section. Usually there is no prior discussion of the strategies and priorities of the district. 2. The form is returned back to head of planning sub-division and includes proposed budget allocations. 3. The planning sub-division compiles and selects the proposed programs. At this stage, the head of the planning sub-division does not discuss the proposal again with the proposer sections before selecting the programs. Thus at times programs considered essential to a section are eliminated. In all districts sampled, the programs usually eliminated are training programs and so no training for nutrition staff at the district level has been conducted in the sample districts. 4. The compilation and selection results from the planning sub-division are delivered to head of development division in the Bupati/Major office. At this stage, the proposals are re-evaluated by the head of the development division. 5. Selection results from development division are then discussed by the district level Tim Anggaran Eksekutif. At this stage the DHO, represented by the Head of Administration Division and the head of planning sub- division, has to explain and justify the proposed programs. Common at this stage, all nutrition programs are rejected because members of Satuan Empat do not understand nutrition problems and DHO representatives are not able to present supporting evidence or adequately explain the motivations for proposing these programs. An extreme case is Kupang in 2005. All nutrition programs proposed by the Kupang DHO were rejected 6. The results from Satuan Empat manifest as the Regional Development Plan based on a formal official letter from the Bupati. 7. The Regional Development Plan is discussed at the DPRD with local government. In this forum, the Director of DHO participates as a part of the local government. At this stage there is a possibility of rejection of certain programs or the introduction of entirely new programs. In the case in Kupang, although no nutrition programs were included in the regional development plan, DPRD members agreed to include a nutrition program in the form of the complementary food program (PMT) for poor families. This program became the only nutrition program of the DHO of Kupang in 2005. 8. Result of discussion on Regional Development Plan is realized as APBD (Regional Income and Expenses Budget Allocation) through regional regulation (Perda). 9. Perda APBD becomes the basis for regional development for the coming year. 70 Issues in health information systems To assure the proper functioning of the nutrition governance structure, information has to flow smoothly both between the health department and the planning department at every level of government as well as from the district to the province and from the province to the center. This information then has to be analysed and utilized in the planning process and in monitoring and evaluation of programs. These flows of information frequently do not occur. At the same time, districts and provinces need access to the results of national surveys, such as SUSENAS, in planning and evaluating their activities; in general districts have little ready access to these surveys and even if they did they rarely have the skills to analyse and use the information. Prior to decentralisation, nutrition information was collected and disseminated through the SKPG (Sistem Kewaspadaan Pangan dan Gizi -- Food and Nutrition Alertness System). The SKPG survey was conducted annually by related sectors. MOH was responsible for conducting community nutritional status surveys and household nutrition consumption surveys. There was considerable variation between districts and provinces in the quality of the information and in the extent to which it was utilized to set priorities, monitor program implementation and evaluate effectiveness. Since decentralization, SKPG surveys fall under the authority of each district. The SKPG budget now is included in the DAU (dana alokasi umum -- general allocation budget) and most districts do not allocate budget for it. Among the 4 districts studied, Kupang does not allocate any budget for SKPG, and the other 3 districts allocate budget only in a very limited amount. At the provincial level, budget for SKPG is only available for the SKPG development process. There is no doubt that lack of information hinders not only the planning process but also deprives program design of an evidence base on which to build. But the mere availability of information is not enough to improve the planning and implementation of programs and the effectiveness of interventions. The skills to use the information as well as the incentives to do so are also required. These skills and incentives do not exist at the district level now, nor were they there in the past. The required skills were present in limited supply in some provinces and at the centre but the main programming decisions were, in fact, made in a way that took little account of the differences in nutrition problems between districts, the skills of staff, or the performance of programs in the past. The continuing development of decentralization provides an opportunity to define and implement a new institutional structure in nutrition, and the health sector generally, which is more capable of using an improved evidence base and taking inter-district differences into account when program planning and implementation decisions are made. II. OTHER INSTITUTIONS INVOLVED IN NUTRITION DELIVERY In addition to the planning and health offices of the governments, two other institutions are important for nutrition at the district level - the Dewan Perwakilan Rakyat Daerah (DPRD-Local Parliament) and local NGOs. 71 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES A. DPRD (Dewan Perwakilan Rakyat Daerah-Local Parliament) All the DPRDs at the provincial- and district-level35 behave in a similar fashion with respect to nutrition. No DPRD has a committee devoted solely for nutritional issues. Nutrition is only a small part of the health sector, and the health sector itself is one of the integrated sectors in the Commission of Social Welfare. At the provincial level, the health sector is located in Commission E while it is in Commission D at district level. Interviews and discussions with DPRD members indicate that they typically have a limited understanding of nutrition. In general, nutrition is considered only a problem for observably underfed children, and all provinces and districts visited have budget allocation for PMT programs although the amount varies36. In addition, members of DPRD commonly recognize nutritional problems at a community-level only when marasmus or kwashiorkor, a type of severe clinical malnutrition commonly referred to as busung lapar37, is found. After decentralization, the DPRDs now have great power in determining the district budgets. In some provinces and districts such as East Java, Surabaya and Lumajang, the DPRD has a stated interest in nutrition because of pressures from local NGOs such as LPKS and continuous advocacy conducted by health offices with support from universities. For example, the head of the Nutrition Section at PHO in East Java and head of the DHO in Surabaya play an important role in building a good relationship with NGOs and Universities, including the Academy of Nutrition. In these districts, budget allocation for nutrition programs is relatively higher than other areas. B. Local Non Governmental Organizations Governments at all administrative levels collaborate with the NGOs PKK38 and Pokja IV39. One major program of PKK is growth monitoring through the Posyandu. PKK in selected areas also has specific programs that complement other nutrition programs (Box D4). However, collaboration between a Health Office and domestic NGOs is uncommon, with the exception of Surabaya where DEPKES works closely with KFI (Koalisi Fotifikasi Indonesia-Coalition for Fortification of Indonesia)40. DEPKES in Surabaya has collaborated with two other domestic NGOs, LPKS (Surabaya Consumer Protection Agency) and WVI (Indonesian Vision Mode) (Box D5). Among international NGOs, CARE has worked in NTT since 1992 with the initial nutrition and health program intended for earthquake victims. Currently, six nutrition and health programs managed by CARE are implemented in 5 out of the 16 districts in NTT. 35The DPRD of Lumajang could not be visited because they were involved in APBD discussions 36In Kota Kupang, the PMT program is the only nutrition program in 2005. 37Busung lapar, which refers to a condition with marasmus and/or kwashiorkor symptoms, is a non medical term, but is widely used among local people as in the recent outbreak in NTT. 38PKK is not a typical NGO, but an organization formed and fully supported by government. Members of PKK are wives of civil servants. The head of PKK is the Governor's wife at the province level, and Bupati or Mayor's wife at district level. 39Pokja IV includes health and nutrition sectors which are generally headed by the wife of PHO or DHO's director. 40KFI is a non-profit and independent foundation, an active advocator and government partner in fortification. 72 Box D4. Examples of PKK activities in selected areas PKK of East Java has a malnutrition program with local budget support. In this program, PKK provides additional budget for districts where severe protein-energy malnutrition is prevalent among children. With this additional support, some districts were able to implement PEM control programs. PKK in Lumajang is the most active PKK team, and received the biggest budget among all the PKK teams in areas studied, across administrative levels. Some of its activities include (1) field testing the local PMT (locally made complementary feeding) in collaboration with DHO, Food Security Board Office, and Brawijaya University and (2) facilitating Posyandu services with increased budget. Box D5. NGO activities in Surabaya: LPKS and WVI LPKS is an influential NGO in Surabaya not only in the health and nutrition sector but also in other sectors. All local nutrition surveys have been conducted by LPKS since 2004. LPKS also disseminates results of the nutrition surveys to the government and the public, and advocates for budget increases in nutrition programs. As a result, APBD budget allocation for nutrition programs increased from Rp 1.440.500.500 in 2004 to Rp. 1.736.916.260 in 2005. The Nutrition Section in Surabaya also collaborates with WVI for PEM control. WVI operates a family development program with nutrition education and also focuses on development and improvement in sanitation and clean water facilities, entrepreneurship, and healthy food preparation training for food vendors. WVI has relatively sufficient budget support from the international NGO World Vision. III. CURRENT STRENGTHS AND WEAKNESSES OF EACH INSTITUTION THAT MAY AFFECT THE COURSE OF THE REFORM PROCESS In summary, the issues which impair the ability of the system to deliver improved nutrition of the community are similar to those which affect other parts of the health system. The most significant are: 1. Government structures and processes unsuited to tackling nutrition in a large and diverse country. The most important issues are: * Contested authority between the various levels of government in the wake of the initial decentralization and its continuing modification. The result is: * The opaque and overlapping responsibilities of those responsible for nutrition at all levels. * Overlap between the central, provincial and district descriptions as they seem to be formulated by each level independently of the other levels. * Most are not working according to the job descriptions due to financial constraints and lack of skills. * Inadequate collaboration within districts and between districts and provinces. * Structures and staffing levels which are not clearly related to the nutrition and health problems of the district and province or the responsibilities at each level. 73 HEALTH SECTOR DECENTRALIZATION AND INDONESIA'S NUTRITION PROGRAMS: OPPORTUNITIES AND CHALLENGES * Very limited flexibility in the ability of districts and provinces to structure their staffing levels and skills mix to meet local needs. * Leadership is a critical issue at all levels with considerable differences between provinces and across districts 2. Human resources in which there is a mismatch between the required skills and those available, particularly at the district and provincial levels * Nutrition has a lower priority than before decentralization as judged by: At the district-level: First, a number of districts no longer have a nutrition section (28 out of 38 in Jatim). Second, the person heading the nutrition section often does not have a nutrition background. Finally, the number of staff working in nutrition is small. At the Puskesmas level: First, almost half Puskesmas do not have nutrition staff. Second, many of the staff who are now there will leave soon as they are voluntary and are looking for a permanent position. Finally, many nutrition staff are approaching retirement and it is not clear that they will be replaced given the restrictions on recruitment. * Generally low skill levels of staff at the district level, especially for program planning and evaluation. * An almost complete absence of in-service training for staff at all levels, but particularly at the district and province levels. 3. Inadequate planning and poor implementation of nutrition programs. There is: * A limited evidence base, and related staff skills, on which to base program planning and to assess the effectiveness of programs. * Considerable overlap between the various departments in their responsibility for nutrition with little coordination vertically, between centre, province and district, or horizontally between departments at any level (e.g. Bappeda, Food Security Office, Depkes, DPRD, Women's Affairs etc) in planning or implementation. * However, in those districts in which there was coordination (e.g. Surabaya) nutrition programs seemed to have bigger budgets, and perhaps better implementation, at least as measured by CE. * Overall nutrition programs have low coverage. * Monitoring and evaluation (M&E) is limited due to both staff shortages and lack of skills: * It is now more difficult for the centre and the provinces to do M&E as districts no longer feel an obligation to provide reports. Evidence based policy making is even more difficult in this environment of reduced evidence. * Most of that monitoring which is being done is on a project basis, rather than a program basis, * but Surabaya seems to be an exception where a limited staff is making a good effort to monitor program implementation 4. Limited financial resources, especially in the worst affected areas: * New budget processes have delayed availability of funds, and program implementation at each level of government until well into the financial year. * Limited understanding of, and consequent low priority accorded to nutrition and health issues by district governments. * A lack of resources, in some districts, to actually implement nutrition programs. 74 5. Limited collaboration with groups outside government in delivery of nutrition programs * Limited involvement of central research institutions in applied research which supports planning and evaluation of nutrition programs. * Limited involvement of the private sector in nutrition programs. * At the central level there is good cooperation with: * national companies e.g. Kimia Pharma, Indo Phama, Gizindo, and * national NGOs e.g. Indonesian Coalition for Fortification. * At the provincial level there is: * limited cooperation between NGOs and provincial health office * limited cooperation between provinces and private sector or NGOs-- an example of an exception is the collaboration between the Jatim and the iodized salt producers * At the district/kota level there is: * limited collaboration between NGOs and district/kota -- the best example is Surabaya where there is good cooperation with LPKS (who are paid by the province to carry out nutrition surveys) and Wahana Visi Indonesia (who implement some nutrition programs). * PKK has good collaboration with health at all levels. At the same time, decentralization offers a new opportunity for new institutional links and structures which address nutrition problems in ways that: (a) Redefine government structures and processes so that they promote a renewed partnership between all levels of government based on realistic and agreed roles and responsibilities with enough flexibility to take account of local variation in nutrition problems and capacity. (b) Strengthen the information base for program planning, implementation and evaluation. (c) Improve the capacity of staff at all levels of government and across departments to use information on nutritional status and program effectiveness for planning and implementation. (d) Increase collaboration with * the district administration and parliament for increased attention to nutrition problems. * NGOs and the private sector as partners with government in delivery of nutrition programs. 75 The World Bank Group 1818 H Street N.W. Washington, D.C 20433, U.S.A Telephone 1-202-477-1234 Facsilmile 1-202-477-6301 http://www.worldbank.org The World Bank Office, Jakarta Jakarta Stock Exchange Building Tower 2, 12th Floor Jl. Jendral Sudirman Kav. 52-53 Jakarta 12190, Indonesia Telephone 62-21 5299-3000 Facsimile 62-21 5299-3111 http://www.worldbank.org/id