Report No. 39855-IN India Achieving the MDGs in India's Poor States Reducing Child Mortality in Orissa May 2007 Human Development Unit South Asia Region Document of the World Bank This report was written by Christine Allison, Lead Human Development Specialist in the South Asia Region with substantial support @om Ihsan Ajwad (Economist). Peter Berman, Maitreyi Das, Srivatsa Krishna, Samik Sunder Das and Priti Kumar participated in select missions to Orissa and other poor states. An in-depth field report for Bolangir District, Orissa, was prepared by Susanne van Dillen (University of Bonn). Gertrude Cooper provided administrative and production support to the report. The Peer Reviewers for the work are Mariam Claeson, Abdo Yazbeck and Viviana Mangiaterra, and we have benefited from their advice and support throughout the work Julian Schweitzer (Sector Director), Mansoora Rashid and Anabela Abreu (Sector Managers) have also provided support and advice. Various other colleagues in the WorldBank have contributed ideas. An earlier version of the report was presented to a meeting of the Orissa Government's multi-sectoral steering group on child mortality (Februaly 2006).Also in India, interactions with UNICEF, DflD and other stakeholders have helped inform the work. Contents Executive Summary.......................................................................................................................... i 2. 1. Introduction.............................................................................................................................. 1 Orissa ....................................................................................................................................... 3 4. 3. Recent Trends inInfant and Child Mortality, Orissa............................................................... 8 5.A Applying the Frameworkto Orissa........................................................................................ The Correlates of Infant and Child Mortality: The Framework............................................. 12 . The IndividualWoman........................................................................................................ 16 17 (i) expectantmother....................................................................................................... The 17 (ii) birth.......................................................................................................................... (iii) ofNewbornandyoungchildren............................................................................ The 18 Mothers' education - one of the strongest factors influencingchild mortality..............21 Care B.(iv) Family........................................................................................................................... 23 The 24 C. The Community ................................................................................................................... 28 6.D State Heterogeneity -District Profiles .................................................................................. . Services................................................................................................................................ 29 31 A Individual Woman andMother............................................................................................ . B. The Family ........................................................................................................................... 32 33 7. C Conclusions andPolicy Implications-The Roadmap to Lower ChildMortality.................34 . Services................................................................................................................................ Bibliography .................................................................................................................................. 37 Annex Tables ................................................................................................................................. 51 53 Boxes Box 2.1: Specific Characteristics of Poverty inOrissa:................................................................... Box 3.1: Why do different datasets produce different IMR estimates?......................................... 5 10 Box 5.1: Ante-natal and Post-natal Care and Practices: Findingsfrom a survey of eight districts27 Box 6.1: Mayurbhanj: Achieving rapidreduction inchild mortality ............................................ 36 Box 7.2: Joined up thinking.joined up action. The Case of ANKUR........................................... Box 7.1: Navajyoti: Reducing neonatal mortality and morbidity.................................................. 44 45 Box 7.3: A conditional cash transfer for Maternal and Child Health- Orissa style...................... 50 Figures Figure 3.1: Infant Mortality Rate . and Orissa (1981. ............................................... India 1999) 8 Figure5.1: Proportionof mothers receivingantenatalcare and tetanus toxoid vaccines across castes and welfare quintiles ........................................................................................................... 19 quintiles.......................................................................................................................................... 20 Figure 5.2: The proportion of deliveries conducted at home across caste and household welfare Figure 5.3: Birthattendants across caste and household welfare groups....................................... 21 Figure 5.4: BCG, DPT3 and measles immunization rates of children between 1and 2 years of age Figure 5.5: Child, Infant and Neonatal mortality rates by mother's educational attainment.........23 across caste and welfare quintiles.................................................................................................. 24 Figure 5.6: Proportionof mothers receiving antenatal care and tetanus toxoid vaccines across Figure 5.7: Variation inchild and infant mortality across social identity and welfare quintiles...24 mother's education......................................................................................................................... 28 Figure 6.1: Child mortality rates across districts ........................................................................... 31 Figure7.1: Simulated reductions inInfant Mortality from changes infour policy variables: Numberof infant deathsprevented................................................................................................ 41 Tables Table 2.1: Headcount Index inRegions of Orissa by Social Group (1999-2000) ........................... Table 1.1: Infant Mortality Rates across SelectedIndian States. 1980-1999 .................................. 1 4 Table 2.2: SelectedHuman Development and BPL Indicators by District ..................................... Table 3.1: Trends in some core indicators. RCHIand RCHI1....................................................... 6 9 Table 4.1: Frameworkfor Assessing Impact of Various Factors on ChildMortality.................... 14 32 Table 6.2: District level performance with IMR: Good and bad performers................................. Table 6.1: Child. Infant. andNeo-natal MortalityRates inSelect Districts .................................. Table 7.1: The Framework Revisited: What the Orissa evidence suggests ................................... 35 37 Table.7.2: A summary of the most important factors. direct and indirect. impacting Neo-natal. Table 7.3: The Multi-Sectoral Menu of Interventions, ranked by difficulty and time- horizon....40 Infant and Child Mortality inOrissa.............................................................................................. 38 Number of infant deaths prevented................................................................................................ Table 7.4: Simulated reductions inInfant Mortality from changes infour policy variables: 41 Table 7.5: Existing Interventions in the Priority Areas: A Scorecard........................................... 43 As we have been undertaking the work associated with this report (which started in early 2005) the Government of Orissa has shown a determination to raise theprofile of child mortality, and to make a concerted effort to accelerate a reduction in the number of infant and child deaths. The launch in April 2005 of "navajyoti"- a strategy to improve maternal and child care, with a particular focus on districts with high IMR - represents such a move, building on the experience of the earlier IMR mission; and in more recent times the launch in Orissa of the national rural health mission - an approach which brings together under one umbrella a number of centrally sponsored schemes. The latter is particularly noteworthy as it seeks to reach out beyond the health sector and involve nutrition, water and sanitation in the actions required to reduce infant and child mortality. It also recognizes the importance of district level plans by way of building on local resources and responding to local conditions. The state government has constituted a cross- departmental committee to bring together the various sectors which have a role to play in addressing child mortality, and we have had thepleasure of interacting with that committee in the course of preparing this report. These developments represent an important base upon which to deepen the dialogue around the spec@ strategies which could be adopted, and the subsequent implementation of aprogram which will ensure a reduction in child mortality. Achieving TheMDGs in India's Poor States EXECUTIVESUMMARY This report builds on the World Bank's earlier report on the Millennium Development Goals (MDGs) in India (2004), which highlighted the challenges Indiafaces in meeting a number of the MDGs. That report drew attention to the variation in MDG indicators across the states of India, and showed how much of the "gap" rests with a handful of poor states, of which one is Orissa. Although most of the MDG targets will be a stretchfor Orissa -and other poor states - it is child mortality where the challenge is perhaps greatest', for three reasons: first, there is a complex mix of policy issues to address root causes involving food, health, education, water, sanitation, and infrastructure, to namejust some of the core sectors; second, it requires a number of different actors - individuals, households, communities as well as policy makers and service providers - all to do their part -and providing effective services to women and children living in remote areas, many of whom are extremely poor and illiterate, is particularly challenging; and third, while events such as starvation deaths, avian flu, outbreaks of polio make the news and draw the attention of top-level policy makers, the steady state of large numbers of children dying before theirfijlh birthday rarely tops the headlines.As such, although many would consider child mortality a leading indicator of development more generally, it is yet tofind its way to the top of the country's political agenda and to enjoy the support of highprofile leaders. The Millennium Development Goalfor child mortality establishes a two-thirds reduction in under-five mortality between 1990 and 2015. For India, this translates into a target of around 32 deaths per 1000 live births, or a corresponding IMR of 27/1000 (startingfrom a figure of 80/1000 in 1990). The 11thFive Year Plan agenda would bringforward these targets to the year 2012. Using the same metricfor Orissa, the infant mortality MDG target is 41/1000 in 2015, and the child mortality target is 49/1000. Within theframework of the National Rural Health Mission, the Governmentof Orissa has setfor itself an IMR target of 50/1000 by 2010. Considering recent estimatesplace child mortality in the 80s (87/1000 according to the SRS, and 81 according to the NFHS), these targets should be considered a "stretch goal". Yet by establishing them it has already served to raise theprofile of child mortality in the state and stimulated discussion about what needs to done and how best it can be done. This report aims to help inform that debate and lay the groundwork for action which will set the state well on the road to achieving - and going beyond - the child mortality MDG. Some of the highlights of Orissa's child mortality situation can be summarized as follows: During the past few years, infant and child mortality hasfallen across the State - the exact numbers are subject to some variation across different data sources. I Should data be available on maternal mortality, that too in all likelihood would also pose a major MDG challengefor Orissa. i Achieving TheMDGs in India's Poor States Rising incomes in the coastal districts, increased completion of post-primary schoolingfor females, delayed age of marriage andfirst birth, and improved ante- natal care are some of thefactors associated withfalling infant and child mortality. 0 There are a number of other factors known to be associated with maternal and child health where progress in the state has been less good: environmental conditions (water, sanitation, indoor air pollution), hygiene practices, vector born disease (especially malaria), maternal and child nutrition, births attended by trained medical person (either at home or in a clinichospital), and emergency obstetric care. Of particular note is stagnant rates of immunization coverage. 0 There remains considerable variation across districts, with coastal districts in general having lower IMWchild mortality and interior/KBK districts having much higher rates. Malkangiri, the district with the highest child mortality rate, has more than 120 child deaths per 1000 live births compared to Balasore (46/1000), the district with the lowest child mortality rate; there is less variation .with neonatal mortality rates than withpost-neonatal infant and child mortality.; Districts with the highest infant and child mortality rates are generally poorer, the Cfemale)population is less educated, the population lives in more sparse settlements with poor road connections, making access to health and other services more difSicult, and service provision is weak. However, there are districts such as Sambalpur, Deogarh, Mayurbhanj, and Bhadrak which have lower mortality rates than their socio-economic status would suggest, and these offer interesting sitesfor further investigation. 0 Although childhfant mortality rates are higher among the Scheduled Tribe population of Orissa, this is largely afunction of poverty (lower levels of income and assets,)low levels of education, andpoor access/utilization on health services. Many of thefindings of the investigation underpinning this report confirm what is more broadly known about the child mortality situation in India, but some of the findings are more unique to Orissa (and potentially other poor states). Examples of the latter include the- seasonalitv of infant mortalitv during the rainy season (due to fever, malaria, diarrhea and pneumonia) and the cold season (indoor air pollution), the role of malaria, and the very uoor environmental health pi-actices (hygiene, cooking fuel). These more Orissa-specific factors, coupled with the high rates of uovertv, low levels of (female) literacv, the exceptionally poor state of public services (including health services) and infrastructure, especially roads, all contribute to high levels of neonatal, infant and child mortality. Drawing from the broader framework for assessing the impact of the various factors influencing child mortality (Table 4.1 in the body of the report), thefigure below summarizes the evidencefor Orissa, and begins to tease out thosefactors, practices, and services which have the strongest association with child mortality. Thefigure below is organized intofour quadrants - (i) characteristics and behaviors associated with the individual womadmother, (ii) the family within which the woman lives and into which the child is born, (iii) the community, and (iv) service provision. Each quadrant represents an important segment of the production function of child survival, and while shown as separate spheres, there are significant interactions and inter- dependences across all of them. This underscores the highly complex nature of child survival and the requirementfor inputs and actions in a number of different policy spheres. .. 11 Achieving TheMDGs in India's Poor States The Frameworkfor UnderstandingChildMortality: Actors and Actions A. Individual woman/mother B. Thefamily Education (esp. post-primary) Income and wealth Nutrition (during pregnancy and breastfeeding) Intra-household dynamics Age at first birthand spacing between births Water (piped into house) C. The Community survival D.Services Environmental health practices Basic healthhutritionservices in village/outreach to (water, sanitation, solid waste) households Beliefs and practices (e.g. at the birth) Access to health facilities for emergency obstetric Women's self-help groups and sick child care Active PRIs Other services, such as school, roadtransport and electricity Whatinterventions are needed? Looking to thefuture and a policy agenda, which is most likely to support an acceleration in the rate of decline of child mortality, there is an inclination to say that many things need to change, and indeed that is the case. But on the basis of quantitative and qualitative analysis (detailed in the report), we can suggest a shortlist of twelve interventions which seem to be particularly important. We have divided these into three categories: relatively quick and easy; more involved and needing more time; and longer term - but critical. Regarding the interventions in the latter category - measures which reduce income poverty for the poorest, and educating girls beyond primary education (ideally through secondary school) - the potential impact on lowering infant and child mortality is very significant. For example,policy simulations show that if all women of child bearing age had at least a primary education, the infant mortality rate in Orissa wouldfall by around 35% of its present value. Comparative datafrom other countries in the region confirm this finding. Likewise, a doubling of the asset index (a proxy for household income) would lead to a 17%reduction in infant mortality. The interventions (listed in the table on the next page) are clearly quite different in nature, involving direrent actors and actions in a variety of places - at home, by the household, by local communities, through outreachfrom diferent key sectors, and/or through access tofixed and referral services - span a very varied time-frame, involve different levels and types of resources (human andfinancial), and require diferent delivery strategies. Also, as illustrated in the Table, the role of these interventions varies across neo-natal, infant and child mortality and as such the optimal mix of interventions could change depending on the mortality profile. For example, a district with high post-neonatal mortality (such as Koraput) might need to place equal emphasis on the interventions in the right hand column of the table, whereas a district with high ... 111 Achieving TheMDGs in India's Poor States neonatal mortality but low post-neonatal mortality (such as Khorda) might need to place more emphasis on interventions in thefirst column. " _-~ ~ - ~ -- I 1_ - - I -I_"f I Neo-natal mortality Infant and child mortality (firstmonth) Hierarchy of Interventions I.Quickandeasy Improved hygiene practices, especially Improved hygiene practices, hand washing with soap especially hand washing with soap Insecticide treated nets Insecticide treated nets 2. More involved, needing more time and resources Quality ante-natal and Quality child care, with an emphasis post-natal care, strong emphasis on on home practices, treatment of sick maternal educatiodhome child and immunization practiceslnutrition Birth, including at home, attended by trained midwife or doctor Access to (physical, financial) Access to (physical, financial) emergency obstetric and newborn emergency care for sick child, care, and reliable service provision immunization and other preventative programs Exclusive breastfeeding Exclusive breastfeeding from Day One for 6 months I Spacing between births of Spacing between birthsof I 24 months or more 24 months or more Nutrition education supported by Supplementary nutrition supplementary nutrition for lactating women; weaning and for pregnant and lactating women child feeding practices Clean cooking fuels Clean cooking fuels 1 PiDed water into house Piped water into house 3. Morelong-term but critical Measures which promote Measures which promote empowerment and reduce income empowerment and reduce poverty for the poorest income poverty for the poorest Post-primary education, Post-primary education, especially for girls emeciallv for girls Together, the Government of India and the Government of Orissa support a very rich policy agenda addressing many of the priority areas identified in this Table. In the past year alone a number of major new initiatives have been launched by GoI, including the National Rural Health Mission, a second Reproductive and Child Health Program, the National Rural Employment Guarantee Scheme, and a rural infrastructure program. These complement other centrally sponsored schemes in the area of nutrition (PDS and ICDS), education (SSA), and vector-borne diseases (including malaria). At the state level, there is a whole host of programs including "missions" for rural drinking water and sanitation, a women's empowerment/rural credit scheme (Mission Shakti), school scholarship program for ST/SC girls, an IMWneo-natal iv Achieving TheMDGs in India's Poor States "mission", and the Orissa Health Sector Program. Overall, it could be concluded that Orissa is adequately provided for in terms of the essential policies and programs needed to bring about a further reduction in infant and child mortality. Moreover, some of the shortcomings identified by past reviews - such as inflexible centrally sponsored schemes and over-centralized planning, lack of attention to state and district heterogeneity -are being addressed in the new programs. But there are problems and major challenges. First, while there may be a perception infederal and state government circles that there is afull array of policies and programs to address the multi-sectoral nature of child survival, the reality for people living in the remoter rural areas is quite different. In parts of Orissa there is a near absence of services: public service providers and critical supplies are ofen absent due to both a shortage of providers and weak governance of resources in the system),private providers arefew due to limited market sizehahe, and there are relatively few non-governmental agencies. Heavy seasonal rains and poor infrastructure present additional barriers to regular service provision in rural Orissa. But it is not just a problem of physical access and fickle service suppliers. The poor are ofen faced with significant costs (user charges) when accessing public services, especially health services.A study undertaken in Bolangir District recentlyfound fee for service arrangements in place with Anganwadi workers (for ante-natal care), ANMs (attending births), and in government medicalfacilities (for many services). Moreover, patients reported not only facing considerable expenses in seeking healthcare from a governmentfacility but were also exposed to a great deal of arbitrariness and ill-will at the time of seeking medical care. The challenge to expand and sustain effective coverane/out-reach of services to Orissa's interior districts and scattered populations - in particular the ST population - cannot be understated. Adautinn uronrams to different situations and needs, and considering different delivery strategies -especially involving communities,localgovernmentsandNGOs-steps recently initiatedby the Orissa Government - seempromising areas. Second, some service providers are carrying too many responsibilities and their impact on any one priority is limited. A good example would be the Anganwadi worker who is the frontline service provider of many programs for women and children. Although such an arrangement helps promote convergence across programs, oneperson can only do so much. The advent of a second community based worker under the Rural Health Mission - the ASHA - is a good initiative and should help free up time for the Anganwadi worker to focus her core responsibilities. Line management of the two different functionaries - who report to two different government departments - will remain a challenge. Introduction of fixed health and nutrition days, at which service providers from different line departments converge in a village and provide integrated services, is one initiative to overcome this problem. Another is the increasing involvement of PRI bodies in the oversight of local service delivery. Third, building on recent efforts to respond to national and state level heterogeneity, further sharpening thefocus on those districts and blocks where child mortality is highest, and varying the "input package" of centrally sponsored schemes to reflect local conditions and the relative levels of neonatal, infant and child mortality is still needed. Preparing district (and block) plans, empowering communities and decentralizing resources, responsibilities and accountabilities to local government are all core elements of getting the right mix. Fourth, too many programs are run as self contained "silos" and miss critical synergies: too often potential synergy is lost due to complementary inputs occurring at different times and in different places. There are, for example, strong potential complementarities across sectors, such as those between health and nutrition, or between supply-side investments in service delivery and V Achieving TheMDGs in India's Poor States demand-side benefits from community mobilization. Finding ways to achieve greater inter- sectoral integratiodconvergence in order to benefit from the strong positive interactions across sectors and across programs should be given higher priority than launching any more new programs. The example of ANKUR, a joint initiative of the Orissa Government, and UNICEF is an interesting attempt topromote more multi-sectoral collaboration. Joined up thinking, joined up action: The Case of ANKUR ANKUR is a joint initiative of the Government of Orissa and UNICEF, under implementation in the District of Koraput (one of the poorest KBK districts). It has four inter- related goals: 0 Improve the quality of education for tribal children andgirls 0 Improvenutrition and development of children under 3 years of age 0 Improve infant survival 0 Support implementation of state-wide reforms for rural drinkingwater and sanitation. At the heart of the program is the District Plan of'Action for the young child, which identifies where action i s neededand by whom. There are no new activities, bur rather a focus on establishing functional linkages between existing programs to maximize their impact for the young child. This involves joint planning, implementation, and assessment aimed at changing community and family level behavior, strengthening service delivery, and ensuring strong interface between community and service providers. District level coordination i s replicated at block, GramPanchayatandvillage levels. ANKUR was launchedin2004 and is a four year program, coveringall blocks ina phased manner. The bulk of activities involve support for micro-planning, capacity building and training of front-line functionaries, and frequent coordination meetings. In two pilot blocks additional support for implementationof integrated actions in health, water and sanitation, education, child development and nutritioni s also envisaged. Thefinal area where the existing array of government policies and programs might be considered lacking is the strong emphasis given to supply side and hardware interventions at the expense of insuficient attention to demand side, "the soft side of services", information and behavioral aspects. A quick survey of existing interventions in Orissa (Table 7.5 in the text)finds the need for more emphasis on IEC, BCC in a number of programs (especially water and sanitation), a greater focus of promotive/preventative healthcare of women and young children, and support to individuals, households and communities to demand and access services. Puttingit all together: What might a child mortality reducing programlooklike? In so far as child mortality is linked to multiple determinants, there is a compelling argument for a more holistic, or integrated approach designed to bring to bear all relevant factors in one place at onepoint in time. But a "one sizefits all" approach will not work: what is neededfor one situation may differfrom what is needed in another situation. As such, a range of interventions should be considered depending on local conditions and ideally subject to close monitoring and periodic evaluations in order to provide feedback on what is working. In addition, in order for local preferences to be followed, placing purchasing power to "pick and mix" from the various interventions with individuals and communities would be an important vi Achieving TheMDGs in India's Poor States development. This could be in theform of vouchers or block grants, for example. This approach speaks to "thinkingmulti-sectorally, acting locally". One should also be mindful of the danger that trying to address everything in one package becomes difSused and unwieldy, and impact is minimal. Asking ourselves thefollowing set of questions is key: "of which sectors does it make the most sense to develop linkages and what benefits do these linkages produce; do these linkages add value over and above what these sectors are already doing on their own - and/or could be doing if better implemented; and which linkages are likely to be most cost-effective". Moreover, to be effective, multi-sectoral action requires the cooperation and coordination of a variety of actors and stakeholders. Many of these requirements run counter to public sector culture and practice in India, and as such present another major hurdle. This leads us to "thinking multi-sectorally, acting locally and selectively ". The Orissa state government has constituted a cross-departmental committee to bring together the various sectors that have a role to play in addressing child mortality, under the leadership of the Development Commissioner. This an important stepforward in acknowledging the multi-sectoral nature of child mortality. Moving forward, Go0 will need to consider alternative mechanisms to deliver any new and additional interventions. Given the importance of focusing energy at the local level, and reaching out more effectively to poor households and communities, working in conjunction with Mission Shakti (the self-help group movement) PRIs and NGOs would seem central to the design. The Orissa Poverty Reduction Mission is another promising organizational model to deliver on a multi-sectoralprogram. Building on the above, any new such program could be organized into two groups of activities: (i) cross-cutting support to enhance individual and household behavior and promote demandfor services, and (ii) innovative sewice delivery strategies involving public and private sectors, and NGOs, together with mechanisms that support greater accountability. These are detailed below. I. Cross-cuttingSupporttoPromoteBehavioralChangeandEnhanceDemand a. IEC/BCC - a substantial package of activities to increase knowledge, awareness and practices regarding behaviors that have a significant impact on child mortality. This might involve hygiene (hand washing with soap), water, sanitation, indoor air pollution (use of clean fuels, improved ventilation), malaria (use of bed nets), nutrition and feeding practices, early diagnosis of medical problems, and service provision (what, where and how). These messages would be designedfor a variety of stakeholders, viz. households, community leaders, PRIs, SHGs, and government functionaries. This would build on the new right to information bill and citizen charter, and aim to take information to women in their villages and homes through a variety of media. b. Addressing poverty and demand for critical child survival inputs. Many of the interventions identified as high priority involve a financial outlay on the part of households. Included here would be food (quantity and variety), cooking fuels other than wood or dung, soapfor hand washing, insecticide treated nets, costs involved in sending (girl) children to post-primary school together with the transport and user charges incurred when using either public or private health services. The absence of resources to meet these costs leads women and theirfamilies to adopt strategies which can have dire consequencesfor their own health and that of their children. vii Achieving TheMDGs in India's Poor States In the long run, poverty reducing economic growth should address these problems, although the extent to which growth will penetrate the remote rural areas where the extremepoor live is an open question. But, in any event, shorter-term action is calledfor and there are a number of instruments that could be used to provide financial support to poor households. Vouchers, stipends, scholarships are examples of instruments which provide poor people with additional purchasing power for key services. India's own JSY2 is an example of a scheme designed to ease the financial constraint which prevents pregnant women for giving birth in a hospital. Another instrument is the conditional cash transfer (CCT). This instrument has been successful in Latin America (especially Mexico) and is gaining interest elsewhere in the world. A CCT could support two critical aspects of child survival. The cash transfer to elinible women would help address income vovertv and the various ways in which it manifests itself into sub-optimal behavior, including the early return to wage employment and the cessation of exclusive breast-feeding. By combining the cash transfer with rewired actions on the behavioral side, the program would support attendance at information, education, and training programs (such as promoting better hygiene and feeding practices, as detailed in a. above) as well as possibly full utilization of ante-natal, birthing and post-natal services, growth monitoring of babies. (This would require critical improvement in the provision of services, behaviors of service providers, and other complementary actions.) Theprogram could be started as a pilot in a few of the poorest blocks, be accompanied by a robust effort to improve services, and include a monitoring and evaluation framework to allowfor a review of its effectiveness. Over time it could expand toprovide support tofamilies with post-primary aged girls who struggle to afford schooling. Given the very strong effect of secondary education on maternal and child health this would be a very strategic investment on thepart of government. 11. InnovativeService Delivery Strategies getting services to poor women - a. Community Investment and Innovation Fund (CIIF). This could be managed by a consortium of SHGs and front line service providers (ANM, Anganwadi, ASHA, RWSS), and would provide a pool of jlexible funding to support local area innovation and initiative. Thefunds could be channeled through PRIs in theform of block grants. Some examples of activities that might befunded through the CIIF include: More significant supplemental nutritionfor pregnant and lactating mother. e Health risk fund, a combination of savings and subsidy, to meet costs of health carehealth insurancefor thepoor, especially at time of delivery. Bulk purchase of insecticide anti-mosquito netsfor pregnant women and young children in high risk areas. Supplementary maintenance of AWC, health sub-center and provision of missing essential supplies; adequate water and sanitation. 0 Improved community water source and training of community in water quality testing and monitoring. Support towards cost of individual household latrines (IHL)for poorest of poor families (with bonus schemefor communities which achievefull elimination of open defecation). Janani Suraksha Yojana (JSY) is an initiative under the National Rural Health Mission whereby a cash grant is paid to all BPL women upon the delivery of a live baby. An additional payment is made to mothers who deliver in healthfacilities. ASHAs supporting women during their pregnancy and at the time of their delivery also receive apayment under the scheme. ... Vlll Achieving The MDGs in India's Poor States Scholarship schemefor ST girls where heavy drop-outfrom primary school is a problem. b. Strengthened ServiceDelivery (hardware and software) by line departments - selective investments in a number of areas such as: Supportfor front line serviceprovision in key sectors (health, nutrition, hygiene, water and sanitation); rotation of workers, additional training, etc. Build upfunctional referral capacities for child and maternal health outcomes with investment, restructuring, incentives. Management contracts for running of health facilities and other public-private partnership innovations. Measures to address lack of motivation and instill drive for performance in governmentfunctionaries, measure and reward results. Strengthen supervision, training, and management especially at district and block levels. Case manager to support poor womedfamilies derive services from public healthfacilities, especially at the time of the infant's birth - servicesfrom which they are often excluded. Seasonality - additional out-reach care for babies born in the rainy season (July,August, September)and the cold months (November-January). c. Modification to some serviceprovision packages, e.g., ICDS, to give greater priority to under-malnutrition in 0-36 month age children, including exclusive breast feeding and micronutrient supplementation: also women's nutrition status before, during pregnancy and in early months after birth; targeting RWSS to areas with greatest epidemiological need; pilot measures to address indoor air pollution reduction. d. Technical support and training to SHGs and PRIs for performance monitoring of service providers, promoting greater accountability through instruments such as scorecards; also to encourage greater participation of other institutions, e.g., NGOs, universities, in social audits. e. More evidence-based Planning, Monitoring and Evaluation - support for data collection to undelpin evidence based planning, including at the local, operational level as well as at district and state level; also to support monitoring and evaluation of inputs and activities vis-&vis child mortality outcome goals. Support would be provided for capacity building in (local level) data analysis and planning as well as data collection. Many of these areas would require considerable background work to assess their feasibility and to undertake costing. In so far as there are similar activities underway in other states - for example, integrated child development pilot project in Madhya Pradesh, an information campaignfor school management committees in Uttar Pradesh, a PRI block grant program in Karnataka, community-based development program in Andhra Pradesh - as well as useful pilots in Orissa (such as ANKUR) - a detailed review of these would enhance the knowledge base. Finally, there is growing experiencefrom around the world with multi-sectoral programs, especially regarding nutrition and HIV/AIDs, which offer important insights into elements and strategies associated with success. ix Achieving TheMDGs in India's Poor States 1. INTRODUCTION "Past shortcomingsnotwithstanding, we can reachthe Millennium Development Goals for IMR and MMR by the end of the 11thPlan" Approach Paperfor the 11thFive Year Plan, Planning Commission, Government of India, June 14,2006. 1.1 The Millennium Development Goal (MDG) for child mortality refers to a two-thirds reduction inunder-five mortality between 1990 and 2015. For India, this translates into a target of around 32 deaths per 1000 live births, or a corresponding IMR of 2711000. An ambitious 11th Five Year Plan agenda - quoted above - would bring forward these targets to the year 2012. For Orissa, this translates into a target IMR of 41/1000. Within the framework of the National Rural Health Mission, the Government of Orissa has set for itself an equally ambitious IMR target of 50/1000 by 2010 - a very significant reductionfrom the 2002 figure (of 87/1000 according to the SRS, and 81 according to the NFHS). 1.2 An important characteristic of child and infant mortality in India is the huge variation between state^.^ At the time of the 1998/99 National Family and Health Survey (NFHS), the all- India under-five child mortality rate stood at 100/1000. At one extreme, the state of Kerala had a rate of 17/1000, while at the other extreme Madhya Pradesh had a rate of 159/1000. (The corresponding IMR figures were 16 for Kerala and 103 for MP.) Orissa, another poor performing state, had very similarly highrates of both child and infant mortality in 1998/99, and depending on the data source and the year, Orissa often has among the highest rates of child and infant mortality. When data can be further disaggregated into district level estimates - permissible only with the Reproductive and Child Health Surveys - significant intra-state variations in child and infant mortality rates also emerge: within Orissa, for example, child mortality ranged from a low of 46/1000 (Balasore District) to a highof 124/1000 (Malkangiri) in2002/04. Table 1.1: InfantMortality Ratesacross Selected Indian States, 1980-1999 Source: SampleRegistrationSystem1980and 1990;NFHS for 1999. See figure 11.4 inWorld Bank (2004),"Attainingthe MillenniumDevelopmentGoals inIndia". 4Reduceby two-thirds,between1990and2015, the infantmortalityrate. Introduction 1 Achieving TheMDGs in India's Poor States 1.3 Thus, as India preparesto meet the MDGinfant, child (and maternal) mortality targets by the end of the 1lth Plan period, the biggest challenge rests with some of the poorest performing states - which also are home to a large share of the Indian population. Four states - UP, MP, Bihar and Rajasthan - together accounted for slightly more than one half of all infant deaths in India in the year 2000. Although not a large population state (37 million, or around 3.5% of India's population), Orissa - the focus of this particular report - accounted for 5 percent of all infant deaths in 2000. Further disaggregating to the district level, the mortality reduction challenge rests with some hundred or so districts throughout India, including about a dozen of Orissa's thirty district^.^ Evidence from around the developing world as well as progress from other Indian states suggests that these targets are within the bounds of achievement, but extraordinary effort will be required in a number of areas. The question i s what to do and how best to do it. This is the subject of this report. 1.4 The report focuses on the state of Orissa. The rationale for this choice of state i s threefold: first, among the poor states it has some of the worst human developmentMDG indicators (see the UNDP sponsoredHuman Development Report, Orissa, 2005); second, there i s considerable intra-state variation, which shows that better human developmenthealth outcomes are possible in such a poor state; and third, there is considerable interest on the part of the Government of Orissa to take on the challenge of rapidly reducing child and infant mortality. While some of the findings and recommendations of this paper are specific to Orissa, and indeed some are specific tojust parts of Orissa, reflecting the particular circumstances of the state, other findings andrecommendations are applicable elsewhere. 1.5 The report i s organized as follows: the next chapter provides a brief background on Orissa. This i s followed by a review of recent trends with infant and child mortality in the state. Chapter 4 introduces a framework for assessing the multitude of factors which have a bearing on infant and child mortality, dividing them into four groups: the individual womadmother; the family; the community and service provision. Chapter 5 applies the analytical framework to Orissa and this is followed by a chapter 6 which looks at district level patterns. The final chapter of the report pulls it altogether and attempts to answer three questions: what is needed to bring down child mortality rates in Orissa and achieve the 11" Five Year Plan and MDG goal; how well are existing interventions placed to do the job; and where are the gaps and how can they best be filled. The report ends with an outline of a possible multi-sectoralprogram designed to reduce child mortality inOrissa. According to the NFHS data during 1994-99a fifth of the country's districts and villages accounted for one half of all infant deaths inIndia. Introduction 2 Achieving TheMDGs in India's Poor States 2. ORISSA 2.1 With a per capita income of around US$250, Orissaranks as India's secondpoorest state. The populationis mostly rural, with 87 percent of the 37 millionpeople living inrural areas. The state has one of the lowest population densities (236 versus 324 persons per square kilometer) in the country, and certain parts of the state have fewer than 100people/square kilometer. The state also has a low population growth rate, the result of a relatively low fertility rate (3.3) and a fairly high mortality rate. The state is home to one of the highest rates of poverty of all-India (47% versus the all India average of 26%, 1999/2000, NSS6), and poverty i s more prevalent in rural areas (48 percent) than in urban areas (43 percent). There i s considerable variation within the state, with the coastal areas generally being more developed and having a lower poverty rate; the interior is less developed and has very high rates of poverty, in excess of 80% in some places. Scheduled tribes also figure prominently in the state's population (23% versus the India average of 8%), and within the state have a very high incidence of poverty (73%). Scheduled Tribes dominate in certain districts (such as Gajapati, Kandhamal, Keonjhar, Koraput, Malkangiri, Mayurbhanj, Nabarangpur, Rayagada and Sundargarh). Among the non-income indicators of welfare, education i s central: the state has one of the lower levels of literacy (64% in 2001), but the trend is positive, albeit showing some deceleration. There is a significant gender gap in school enrolment and literacy rates (female literacy is 51% compared to male literacy of 76%), but that too is improving. 2.2 Poverty in Orissa is closely associated with the lack of economic growth in the state. Whereas in 1980per capita income in Orissa was 27% lower than the rest of India, in 1997 it was 70% lower'. By 1999-2000 Orissa's agricultural wages - a major source of income for the poor - were the lowest in India.' Other factors contributing to poverty in Orissa include high inequality in asset ownership (especially land), significant dependence on forest products, low levels of literacy and, until recently, little private (foreign) investment. The state i s vulnerable to various large-scale natural hazards, such as cyclones - e.g., the super cyclone of 1999 - floods, drought, and malaria is rife. Ingeneral, the coastal areas are prone to floods, while the rest of the state is vulnerable to drought as a result of the combined impact of lack of irrigation facilities and erratic rainfall. It has been estimated that over a seven year period from 1996/97 to 2002/03, drought induced crop losses of over 50% occurred in five years and affected 261 out of 314 blocks of the state. 2.3 Regional disparities. Orissa is an overwhelmingly rural state. Across rural areas, inequalities are pronounced between the relatively wealthy coastal area (32 percent poverty) and inland areas (the rates range from 50 percent in the northern region to a staggering 87 percent in More recentpoverty estimatesfrom the 2004 NSS suggest a decline inpoverty. Mackinnon(2002).Assessing the Impact of Fiscal and Structural Reforms on Poverty in Orissa. (Oxford PolicyManagement.) 'A Deaton and J Dreze. `Poverty and Inequality in India: A Re-examination. Economic and Political Weekly, September2002. Orissa 3 Achieving TheMDGs in India's Poor States the southern regiong). While poverty declined in the rural coastal region during the 1990s, rural areas in the southern region witnessed an unprecedented increase in poverty from 69 percent in 1993-94 to 87 percent in 1999-00 (poverty also increasedslightly inthe northern region). 2.4 Social identity. Orissa has the third highest concentration (after MP and Maharashtra) of scheduled tribes (ST) population; Scheduled caste (SC) and ST constitute respectively 16.2 and 22.2 percent of the total population. Poverty among the SC and especially the ST population is strikingly higher than among other population groups. Thus, while the ST population represents 22 percent of the populationof Orissa, they constitute more than 40 percent of the total number of poor. The relative disadvantage of the scheduled tribe population is a remarkably robust feature of the profile of poverty in Orissa: many social indicators (education, health) for this group are considerably worse than for the majority population, as well as compared to the ST population in other parts of India. - - - - - ~ - ~ - - Table2.1: HeadcountIndexinRegionsof Orissa by Social Group (1999-2000)'0 ~ " ~ " ~ - ~ ~ . - --"ll^ll_- Rural Urban Region ST sc Other All ST sc Other All Coastal 66.6 42.2 24.3 31.7 63.5 75.7 34.3 41 Southern 92.4 88.9 77.7 87.1 72.3 85.0 24.6 43 Northern 61.7 57.2 34.7 49.8 54.4 63.1 37.8 46 Orissa 73.1 --_52.3 33.3 .o 59.4 72.0 34.2 II--~- Y I-.--r--&-,-^u~rrrrr- _^_j*_ _l__-_-.__.~___u 2.5 When taken together, regional and social identity characteristics present a very striking picture: no less than 92 percent of the ST population living in the rural areas of the southern region" of the state is poor. Regarding the urban areas, the highest incidence of poverty i s among the SC populationliving inthe Southernregion. 2.6 Structural and economic factors. Orissa's extensive poverty reflects a number of structural and economic features. Agricultural yields in Orissa are lower than in the rest of India, reflecting a low share of irrigated area (only 22 percent of agricultural land i s irrigated as per the Agricultural Census) and low use of commercial inputs. The proportion of villages connected by all weather roads is also lower than in the rest of India, with farmers not well connected to markets and processing infrastructure, thereby leading to an unusually high share of paddy in output and limited diversification. Orissa i s characterized by a high incidence of subsistence production and traditional land tenure. Beyond already skewed land ownership patterns, there is documented evidence on processes of land alienation and the expropriation of tribal groups, suggesting that improvements in the property rights of the poor is likely to be good for poverty reduction. Orissa's forest and tourist potential are under exploited, and as a result of mismanagement, total forested area, constituting a central livelihood asset for tribal households, are continuously being depleted.12 Estimates of poverty from small scale household studies intwo of the southern region districts suggest a somewhat lower poverty rate in the region of 70-75% in 2000/01. loOfficial NSS Data as calculated by A Dubey and referenced inDe Haan (2005). l1 This comprises the districts of Boudh, Balangir, Kalahandi, Kandhamal, Koraput, Malkangiri, Nabarangpur, Nawapara, Rayagada, and Sonepur. 12According to the Forest Survey of India, the rate of deforestation i s much higher in Orissa than at the all India level. Orissa 4 Achieving TheMDGs in India's Poor States Box 2.1: Specific Characteristicsof PovertyinOrissa: As identifiedby Departmentof Planningand Coordination(Economic Survey2004) Spatial disparities e.g., Southern Orissa Vulnerability to recurrent natural calamities Erratic precipitation which constrains the development of agricultural livelihoods Poor development of irrigation facilities Marginalization and poor connectivity in Southern and Western hills (restricts emergence of and access to centers of growth and service delivery) leading to extreme poverty and vulnerability Stagnation of agricultural growth Despite significance of forest resources, poor management and acute degradation undermines the possibility of sustainedprofitability Absence of adequatestatewide infrastructure base, thwarting private sector development. Communicable diseases (especially the deadly form of malaria) which spread easily given poor housing, prevalence of malnutrition and extremes of weather. Public health infrastructure is thinly spread and malfunctioning (with poor equipment, vacancies and absenteeism) Large-scale development projects (dams, river valley projects, open cast mining, etc.) and associateddisplacement, particularly acute inST hilly areas High sensitivity to price fluctuations, which PDS is only partially able to address Shock prone labor markets due to lack of economic diversification (out of agriculture) and limited 2.7 Human Development indicators. While Orissa's human development indicators have improved steadily inrecent years, but they still fall behind the rest of India. Life expectancy at 61 (for both men and women) i s low; maternal mortality (362 per 100,000 live births) i s muchhigher than in many neighboring states (e.g., 154 in AP and 264 inWest Bengal). More than half of all children are underweight, and- incontrast to other major states - there has been little progress in Orissa over the past decade inreducingchild malnutrition rates. Ineducation, there are heartening signs of catching-up with the rest of India, both in terms of enrollment and literacy rates for both sexes. However, evidence suggests that health and education outcomes are considerably worse among the poor, perpetuating a vicious cycle of debilitating ill-health, illiteracy, and poverty. 2.8 There i s considerable variation across districts in both health and education indicators. On the basis of the census data and information taken from a variety of other sources (published inOrissa's HumanDevelopment Report, 2004), the data presentedinTable 3 show an interesting pattern. While the state-wide literacy rate rose from 49% in 1991 to 64% in 2001, literacy levels inthe southernregionremained around 45%. Health indicators also show considerable intra-state variation. Initial efforts to correlate various indicators for human welfare deprivation (e.g., D e Haan, 2005) appear to suggest that people inthe worst off economic districts suffer from multiple deprivations. Orissa 5 Achieving The MDGs in India's Poor States Table2.2: SelectedHumanDevelopmentandBPLIndicatorsby District ---I-----a-- I*-"*,u"-. - ~ - - - - " --~-~--~--.~ ~ ~ ~ ~ __ ------I_y ~~~" I\IxI*--.yI. Region/ Literacy Literacy IMR Children BPLfamilies District (2001) (2001) (2002/04)* Completely (2002 Census) (2000 all women Immunized (%Io) povertyrates) (%) (%) (2002/04) * Coastal(32%) 71 56 58 62 Balasore 71 60 45 74 74 Bhadrak 75 64 44 54 66 Cuttack 76 66 45 87 52 Gajapati 42 29 58 46 61 Ganjam 63 48 71 53 55 Jagatsinghpur 80 70 50 46 53 Jajpur 72 61 71 37 60 Kendrapara 77 67 70 58 60 Khurda 80 71 51 59 59 Nayagarh . - 71 58 52 62 68 Puri 78 68 56 73 69 Northern (50%) 64 54 57 66 Angul 69 56 52 57 59 Bargarh 64 50 54 71 60 Debagarh 61 48 42 48 79 Dhenkanal 70 59 53 60 63 Jharsuguda 72 59 59 71 49 Keonjhar 60 47 68 35 77 Mayurbhanj 52 38 53 46 78 Sambalpur 67 55 40 72 60 Sundergarh 65 54 64 59 65 Southern (87%) 46 74 51 75 Boudh 58 40 70 70 80 Bolangir 55 39 76 65 61 Kalahandi 46 30 69 50 63 Kandhamal 53 36 69 59 78 Koraput 36 25 69 31 84 Malkangiri 31 21 100 40 82 Nabarangpur 34 21 74 45 74 Nawapara 42 26 67 43 78 Rayagada 36 24 88 46 72 Sonepur 64 47 61 60 73 Orissa(47%) 64 51 64 55 ----~----- 66 *These data are taken from the secondRCH survey which was carriedout in two phases: 15 districts in 2002 and 15 districts in 2004. Estimates of IMR from these data are comparableacross districts but the underlyingmethodology of the RCHswey preventscomparabilitybetweenthese andother data (e.g., NFHS). Source: Census 2001, BPLCensus 2002, NationalCommissionon Population, RCH2002 and 2004. Orissa 6 Achieving TheMDGs in India's Poor States 2.9 Health risksare among the most common risksfaced by the population in0 r i ~ s a .There l ~ are a number of communicable diseases - malaria, filariasis, tuberculosis, gastro-intestinal, respiratory infections, and leprosy - which are widespread in Orissa and which contribute to poverty, morbidity and mortality. According to the WHO, the region comprised of western Orissa and Chhattisgarh accounts for the highest number of registered malaria cases in India. With an annual average of 480,000 cases reported between 1998 and 2001, of which on average 380 were fatal, Orissa was inthe lead by a large margin. More recent surveillance data suggest a decline in the number of reported cases (to around 300,000 in 2004), but nonetheless Orissa continues to account for fifty percent of India's malaria burden. After malaria, in terms of both morbidity and mortality, tuberculosis poses a significant threat to life and livelihoods inOrissa. Aggravating the situation is the fact that the public health infrastructure is thin and patchy, especially in the southern and western parts of the region. The services it provides are reputed to be unreliable and beset with problems, such as poor equipment, vacancies, absenteeism, informal payments and lack of supplies. There are also problems on the uptake side. Interviews carried out in connection with the van Dillen Bolangir survey (see Annex 2 for details) revealed that poor infrastructure and difficulties associatedwith transportation of the sick person, and the cost of treatment (public and private, with clear preference for private medical institutions largely due to the more predictable nature of care and costs) were frequent reasons for seeking treatment only when the infectionreached an advanced stage and the patient was already ina life-threatening condition. 2.10 The region known as Orissa's `KBK districts' i s not only the poorest part of Orissa, but also one of the poorest and most vulnerable parts in the whole of India. It i s remote from any major city, and parts become inaccessible during the rainy season (and for as much as 6 months of the year). Inaddition to the health risks noted above (especially malaria), the inhabitants of the `KBK'districts are most sensitive to fluctuations inboth consumer and producer prices. Because there is very little diversification out of agriculture, the labor market in Orissa's `KBKdistricts' i s also prone to shocks, which are transmitted through scarcity of employment opportunities. Whatever diversification there i s mostly takes the form of non-agriculturalcasual wage. l3Inarecent survey of households inOrissa, close to 30% of households reportedexperiencing a health shock inthe past year, and among the poorest households this was the most frequent of all shocks. (Mahendra Dev et al. CESS. 2006) Orissa 7 Achieving TheMDGs in India's Poor States 3. RECENTTRENDS ININFANTAND CHILD MORTALITY,ORISSA 3.1 Infant mortality rates (IMR) in Orissa have been among the highest for all-India during the past two decades, but like elsewhere seemto be falling. (Different data sets -NFHS, SRS and RCH all generate different estimates of IMR - see Box 3.1 for a discussion of these differences - but all agree on one thing: IMR in Orissa is falling.) According to data from the Sample Registration System, in 1980, the IMR for Orissa was 135 deaths per 1000 live births: this compared with an all-India figure of 110, a nationwidehighof 150inUP and anationwide low of 37 in Kerala. By the year 2000, the all-India figure had fallen to 68/1000, the rate in UP had fallen to 83/1000, while the rate in Orissa was 86/1000.'4 Despite the decline, IMR in Orissa in the year 2000 was the highest of all Indian states, almost seventimes the rate inKerala. Figure3.1: InfantMortality Rate India and Orissa (1981- 1999) - 5w 160 140 0 120 2 0 0 v) 100 I- I= 80 1- n 60 40 -z2 I- 2 20 0 YEAR Source: As for Table 1.1. 3.2 The RCH data likewise show a significant declining trend in IMR. Between the two rounds of data collection (1998/99 and 2002/04), there has been a decline in the rate from 77/1000 to 64/1000. The decline inseen inmost districts. (Annex Table Al). Notwithstandingthe confidence intervals with which district level estimates of IMR can be taken (arising from relatively small numbers of observations), there does appear to be very positive falling IMR in a number of districts, viz. Balasore, Mayurbhanj, Gajapati and Jagatsinghpur. Child mortality cannot be estimated from the first RCHround, but the trend from the NFHS figure for 1998/99to the RCH figure for 2002/04 suggests a very significant decline. Moreover, the relatively small difference between child mortality and infant mortality inthe second RCH survey (i.e., 73 and 64 l4By 2004, SRS data give Orissa anIhTR of 77/1000 Recent Trends in Infant and ChildMortality, Orissa 8 Achieving TheMDGs in India's Poor States respectively, a gap of 9 deaths/1000 births among the age group 1-5 years) suggests rapid progress inreducingmortality inthe 1-5 year age group. Likewise, there would appear to be some decline inneo-natal mortality in this period, though there i s less reliability around the robustness of any estimate. However, of more significant note i s the highpreponderance of neo-natal in the overall infant and child mortality rates: state-wide RCH data (2002/04) show a neo-natal rate of 44/1000 - in other words, two-thirds of infant deaths are in the first month of life and sixty percent of under five, childmortality deaths are inthe neo-natal period. Table 3.1: Trends insome core indicators, RCHIand RCHI1 ~- _1___-_1_ _. Change (1998/99) (2002/04) (33.7) Infant mortalityrate( er 1000) (13.0) Educationlevelof mothers: % literate 11.5 Marriagebefore 18 birthday: % marriedbefore age 18 years- 32.2 23.1 (9.1) Ante-natal care during pregnancy: % at least one ANC visit 72.9 76.0 3.1 % full ANC (3 visits+TT+IFA) 32.5 58.4 25.9 % tetanus toxoid 79.7 86.0 6.3 Delivery of baby: % home delivery 76.2 65.0 (12.2) % safe deliveries* 32.7 58.0 25.3 Breastfeeding % breastfed within first 2 hours of 13.9 44.8 30.9 birth % exclusive breastfeeding till 6 4.9 20.3 15.4 month Immunization: % of 12-36 month children: Fully Immunized 57.8 55.4 (2.3) -,.".*---- Immunized aAainst measles +* ---."-...- ---- 64.1 67.9 3.9 "-_^*_-v -*-.--- "~- I NFHS 1998/99figure (no estimate of child mortality can be made from RCHI) Safe deliveries include institutional deliveries and home deliveries assisted by doctor/nurse/ANM. 3.3 The direct causes of infant and child mortality are well documented gl~bally,'~although less i s known about the direct causes of neonatal mortality than at other ages. The main clinical causes of death in India are diarrhea, pneumonia and infectious diseases (such as malaria), with underweightlmalnutrition often considered a major contributing factor. According to the Government of India's "causes of death, annual report" infant deaths inrural Orissa are attributed to pre-maturity (39%), pneumonia (15%), respiratory infection (9%), anemia (8%), bronchitis/asthma (5%), and tetanus (3%). Regarding children aged 1-5 years, the main causes of death are pneumonia (19%), diarrhea (14%), anemia (12%), jaundice (9%),typhoid (7%),malaria (7%), andmeningitis (5%). l5See, for example, Rutstein (2000) and The Lancet, special edition on child survival, summer 2003. ~ Recent Trends in Infant and Child Mortality, Orissa 9 Achieving TheMDGs in India's Poor States ~- ----."--_ --- - _I__ ----x--- -~-I i_ I- .- ~ -"- ." .-l"-l__ Box 3.1: Why do differentd ~ _ - _ . e different IMR estimates? ~I c1."-I__l_i~ There are three different sources of estimates for IMR for Orissa - and all produce a slightly different figure for the same year. The RCH data generate different estimates of infant mortality for the state of Orissa as compared to the Sample Registration System (SRS) and the National Family Health Survey (NFHS) - for similar years. The primary difference inestimates betweenthe RCH and the SRS i s the method of data collection. While the RCH conducts a one-off interview of mothers about births and deaths in the five years preceding the survey, the SRS uses part time local enumerators to conduct continuous enumeration of births and deaths. The differences in data collection between RCH and the NFHS, on the other hand, are quite minor. The former i s representativeat the district level, while the latter i s only representative at the state level due to sample size. At the state level, the two surveys produce similar IMRestimates for 1998/99-RCHproducesafigure of 77 deathsper thousand births,comparedto 81 inNFHS. The differences across data sources serve as a caution about using the absolute levels of IMR generated from the RCH data, especially the district level estimates where the number of observations become fewer. However, the RCH data are quite reliable for purposes of comparisonacross time periods --m (i.e., from RCH Ito RCH 11), across districts, and for analysis of the various factors influencing child o r t a l i t y . -v----Fm"-p-p *------------." 3.4 While there are a whole host o f factors, direct and indirect, which play a part in child survival, some o f the more important ones are known to include: women's education, age o f marriage and first birth, ante-natal check-ups during pregnancy, number o f pregnancies and birth spacing, births attended by a trained birth attendant, births in medical facilities, exclusive breastfeeding and immunization. At the state level, trends in some o f these indicators for which RCH data from both rounds exist are presented in Table 3.1. All indicators show a trend in the desired direction bar one. 3.5 On the positive side, (mean) age of marriage and first birth have risen (by half a year in the case o f marriage in the period 1999 to 2004), girls and mothers are more educated, and they are seeking more ante-natal care. On the RCH service delivery front, there appears to have been improvement in a number o f areas, notably ante-natal care (the incidence o f three ANC visit has risen from 33% to 58%, and tetanus toxoid coverage from 80% to 86%). On a less positive note, while there has been some gain inbirthsattended by a trainedbirthattendant and taking place in a health facility (public and private), both figures remain low. Post-delivery visits from a health worker have marginally improved (from 9% to 14% o f births). Exclusive breastfeeding for 6 months has increased from 5% to 20% o f mothers but i s still far below the desired level. (Physical) access to health facilities remains very low, although ICDS centers have become much more widely available. One area where the trend i s not at all positive i s immunization. The proportion of children immunized against measles - a potentially fatal disease for young children -hasheldsteady,buttheproportionofchildrenimmunizedagainstallthechildhooddiseaseshas ifanything fallen inthe period 1998/99to 2002/04.Moreover, district level data suggest that this i s not an isolated trend: in twenty (out o f thirty) districts the coverage rate has fallen.I6 (Annex Table A2). 3.6 Looking at trends in the eight KBK districts, where poverty levels are high, women's education low, and the proportion o f ST population large, the absolute levels o f most indicators are poorer compared to other districts, but bar immunization rates, there has been good progress. By 2002/04, the aggregate IMR for the eight KBKdistricts was 15 deaths/1000 birthlower than it l6 For other indicators, district level data show that the inter-survey trends were inthe desireddirection: all districts showedimprovementsin women's literacy and delayedage of marriage, both reflecting expansion of schoolingfor girls. Births in a healthfacility and onset of breastfeedingafter the birthalso show solid trends. Recent Trendsin Infant and ChildMortality, Orissa 10 Achieving TheMDGs in India's Poor States had been in 1998/99 (from 91/1000 to 76/1000), and the inter-temporal fall in IMR was slightly higher for KBK districts than non-KBK districts. The declining trend in immunization rates can however be seen in the KBK districts as elsewhere: a particularly worrisome trend since in 1998/99 these districts compared well with the rest of the state (Annex Table A2). 3.7 There is wide-scale documentation of a preference for male children inIndia. Orissa data (RCHand in-depthstudy) show the typical pattern of a slightly higher number of boy childbirths (inthe ratio of 51:49) anda higher neo-natal mortality rate among male children, especially inthe rural areas (across both time periods). Post-neonatal infant mortality rates, on the other hand, are very similar across boy and girl children although there i s a slightly higher girl child mortality rate in the 1-5 year age range. Breastfeeding practices and immunization rates are very similar across boy and girl children, but boy children are often favored in the following way: health expenditure (treatment and medicine) can be higherfor male children, and male children are more closely watched during the first months of their lives, in part in the belief that they are more susceptible to disease. (van Dillen) On balance, because infant and child mortality rates are dominated by neo-natal mortality, survival rates are generally higher for girl children, especially inthe rural areas, although there is some variation. Recent Trendsin Infant and Child Mortality, Orissa 11 Achieving TheMDGs in India's Poor States 4. THE CORRELATES OFINFANTAND CHILD MORTALITY: THE FRAMEWORK 4.1 There exists a large body of empirical studies which focus on analyzing the determinants of child mortality in India, and elsewhere in the world." The key findings often indicate that household income, female education, access to health services, and immunization are some of the most important determinants of child mortality. This evidence indicates that public policies emphasizing improvement in access to school (especially for girls), promoting economic growth and improving access to health services are all well placed to help reduce child mortality. Several recent studies have also shown that environmental conditions, such as access to safe water and sanitation facilities, electricity, and use of clean cooking fuels have an important healthimpact on young children. A recent World Bank study on the role of public policy and service delivery supporting India's efforts to attain the MDGs found a strong association between infadchild mortality on the one hand, and government health spending, female literacy, access to roads, regular supply of electricity, sanitation, and ante-natal careltetanus immunization of women on the other hand. That same report also demonstrated that if poor states, such as Orissa, enjoyed levels of spending and services similar to that inthe non-poor states, infant mortality rates would fall by an estimated 36.5 deaths per 1000births.'* 4.2 As such, it i s thus widely documented that there are multiple factors which influence whether a baby or young child survives the early years of life. These factors relate to the mother, including her situation prior to the pregnancy as well as practices during the pregnancy, through the birthand into the early days and months of the child's life. Inso far as the mother lives within the domain of a family, the situation and practices of the household more generally are also important. Similarly, the community is also of relevance, as it too exhibits social and cultural norms around preferred practices impactingboth mother and child. The fourth category of factors that have a bearing on the survival of the young babyjchild are services available to the mother and her family. A framework setting out these various factors i s summarized below and developed in more detail in Table 4.1, where a brief explanation of how these four categories of factors - individual, family, community and services - might impact infant and child mortality can be found. l7See, for example, Pandey, Cheo, Luther, Sahu andChand (1998), Rozenzweigand Schultz (1982), Murthi,Guio andDreze(1995), Hughes, Lvovsky andDunleavy (2001), JalanandRavaliion(2003), Mishra andRetherford (1997), (2003), WHO (1997), and Wagstaff and Claeson(2004). '*World Bank (2004). The Correlatesof Infant and Child Mortality: TheFramework 12 Achieving TheMDGs in India's Poor States - ~ . ~ ~ - - ~ --,- -.*--.. - _.^_. --A X . ~ - ^ _ ^""* *, -*---,"-I ~, _(I__ -L "X I ff--I"- -,I_ Health Individual Household Community Service Provision Outcome Woman Health, nutrition Householdbehaviors Community Health services - and general well- (e.g., hygiene) resources, public and private; Lower being of woman prior to and other risk factors environment, Nutrition services; child and during pregnancy, values and School Mortality; affected by level of Householdresources norms; socio- Infrastructure Healthier education, and the (assets, income, economic Transport, access to Babies and woman's status within food, housing) structure of services and market young the family, among other community, Children factors. Maternal and formal and childcare practices informal of family members. institutions. 4.3 Improvement in child survival, especially for the poor, requires that the range of factors identified inthis framework be fully understood and assessed for purposes of identifying the entry points for change. Consider, for example, the situation of a child born into a poor family in remote, rural Orissa. The mother is likely to have - at best - a primary school education (of questionable quality), have married at a young age and born her first child before her eighteenth birthday. She may or may not have had more than one ANC check-up, but in all likelihood will have been given a tetanus toxoid vaccination. If the child survives the birth- most likely in the home with the support of relatives but no trained birth attendant - from a very early age he/she will be exposed to the risk of disease through poor water quality and quantity (poor people are less able to afford modem connections and usage charges), inadequate sanitation, poor hygiene practices, indoor air pollution (cheap fuels), poor housing conditions, and high exposure to disease vectors (especially malaria). The child is most likely to have lower resistanceto infectious diseases because he/she will be undernourished due to denial of colostrum at birth, insufficient breast-feeding, a diet deficient in one or more of the essential micronutrients, had low birth weight as a result of poor maternal nutrition, short birth intervals between children, and to have recurrent disease episodes. Once sick, this poor child i s less likely to be taken to an appropriate health-care provider, due to lack of knowledge on the part of the mother and/or advice from older female relatives, poor road access, and the costs associated with seeking modem health care. If they are fortunate, once there, the child is less likely to receive appropriate care because facilities serving poor communities are not as likely to have well-trained staff on hand or to be stocked with supplies and drugs needed to attend to the poor child. In this way, the poor child is more likely to die. Challenges to child survival thus exist at every step along the path from exposure and resistance to infectious disease in the home, through care- giving to care-seeking, to the probability that the child will receive prompt treatment with effective therapeutic agents. The Correlates of Infant and ChildMortality: TheFramework 13 Achieving TheMDGs in India's Poor States Table 4.1: Frameworkfor Assessing Impactof Various FactorsonChildMortality" Domainof Variables Factorsto consider How factors impactchild mortality,direct andindirect A. The IndividualWoman 1.The expectant Mother Education Knowledge of pregnancy, Age at first pregnancy nutrition and related matters Birth spacing Physical maturity/ ability of body Time-uselworklactivities to carry child Nutrition and health status Caring practices during Use of ante-natal services uremancvlbirthuremration 2. The birth (birth order, first Place of birth Non-hygienic birth and and subsequent): where and Delivery attended by trained birth complications during birth can with whom attendant result inneo-natal death of child Access to emergency obstetric (and mother) care, and essential neo-natal services 3. Care and treatment o f new Treatment o f new-born - Inappropriatecare of babylyoung born and child washing, early child can increase probability of breastfeedingkolorums, exclusive infant mortality; breastfeeding until 6 months; Failure to immunize child, seek feeding and weaning of child; medical care for sick child can immunizations; prevention and also contribute to child mortality. treatment of sick child Early return to work can interrupt Return to work breastfeeding and care B. The Family 4. Householdbehaviors and risk Hygienehanitation practices (e.g., Interaction of other family factors hand washing, use of latrine); members with motherlbaby Cooking with "clean" fuel Environmental health Disease prevention (esp. malaria, Access to health care diarrhea, pneumonia) Young mother may be subject to Migration. practices dictated by others Attitudes to pregnancy and child around her, especially older bearing; health care seeking women. Dractice 5. Householdresources Assets (land, business, savings, Role of pregnanthew mother in people), income. Availability of economy of family. Ability of food. Intra-householddistribution family to afford food, health care, of resources, esp. food. Type of and seek it for pregnant housing, ventilation, womadbabylyoung (girl) child. fueYelectricitv. water. sanitation. Oualitv of home environment. C. The Communitv 6. Physicalenvironment Infrastructure (roads, water, Economic conditions facilitating sanitation, electricity ) or hindering healthy lifestyle; School, other public institutions. access of pregnant Environmental health factors. womanlmother/young child to various services 7. Community attitudes Gender norms; religious beliefs; Social conditions facilitating or ethnic mix. hindering access of pregnant Attitude towards government and womanlmother to various "modern" sector. services. l9This framework i s adapted from World Bank (2001) Poverty Reduction Strategy Paper Sourcebook, 2nd edition. The Correlatesof Infant and Child MortaliQ: TheFramework 14 Achieving TheMDGs in India's Poor States 8. Local institutions -political Community groups (SHGs). Level of empowerment of women ~ ~~ and administrative; leadership Decentralized planning, political to influence resource allocation structures and financial support for key decisions pertinent to mothers. interventions (PRIs) Local control over resources for important services. D.Service Provision 9. Supply (and cost) of health Availability of facilities; staffing; Accessibility of health services to services quality o f service - public, women; (physical) distance, private, traditional; availability of social distance (esp. for poor); medicineddrugs, blood, etc. hours of operations; costs (Forman and informal) cost of associated with using the service; using service out-reach to remote areas 10.Supply (and cost) of other Education & training Knowledge, attitudebehavior services Transport (public, private) Access to services Water and sanitation Input to good health Electricity Lighting, cooling, Financial services Consumption smoothing, paying for emerEncies -y__cx__( I % ...-* The Correlatesof Infant and Child Mortality: The Framework 15 Achieving TheMDGs in India's Poor States 5. APPLYINGTHE FRAMEWORK TO ORISSA 5.1 Inthis part of the reportwe apply the framework presentedinthe previous section and apply it to data for the state of Orissa. The idea here i s to validate the existence, relevance and strength of the multiple factors impacting child survival with the view to better informing policy. Inthefirst case, we apply theframeworkto state-wide data(thischapter). Inthe secondinstance, we disaggregatethe analysis to the district level (chapter 6). The data 5.2 The primary data set used in this chapter is the Reproductive and Child Health (RCH) household survey. One of the major reasons these data are used (as compared to other household surveys) i s that the data are representative at the district level, as well as being ideally suited to analyze the socioeconomic correlates of child mortality in Orissa. Two rounds of data are available, spanning the years 1998 to 2004. Each of the two rounds were carried out in two phases, half the 30 districts were surveyed in each of the phases. The first and second phases of the first round were collected during 1998 and 1999 respectively, and for the second round the first phase was carried out in 2002 and the second phase in 2004. The data provide specific insights into RCH services, such as coverage of antenatal care and immunization services; extent of safe deliveries; utilization of government health services, and user satisfaction as well as broader householdlevel socio-economic variables." 5.3 By way of complementing the analysis that can be done with the RCH household data, the World Bank commissioned an in-depth study of child mortality in one of the KBK districts - Bolangir. This district was chosen since it afforded an opportunity to build on a five-year panel of household data, which had been collected by a research team very familiar with the local area. Although a small number of households Gust under 200), and in no way representative of Bolangir District let alone the state more generally, the richness of data gathered over a five year period and the additional insights gained from a dedicated "round" on child mortality are crucial to an understanding of household knowledge, practices and decision making as they impact child mortality. (Annex 2 carries more details on the in-depth study, hereafter called the Bolangir van Dillen study.) 5.4 The material presented in the following paragraphs draws on both quantitative data from the RCH household surveys as well as data from the van Dillen study. On the quantitative side, we use both linear relationships and multivariate analysis to assess the impact of the various factors. (Regressiontables are at the end of the paper, Annex Tables A8-All.) While a lot can be learned from these manipulations, and indeed can be used to informpolicy and programdesign, it i s crucial to stress that the explanatory power of any one of these relationships i s not very strong21 2o For a fuller description of theRCHdata, see http://www.rchindia.org.abtus.htm. 21 The R squared inthe multivariate analysis is low, which indicates that many of the factors impacting child mortality are not being captured by the model. Applying the Framework to Orissa 16 Achieving TheMDGs in India's Poor States - in other words, there remains a lot of unexplained variation in infant and child mortality outcomes. This is where the in-depthqualitative materialcomes in. A. THEINDIVIDUAL WOMAN 5.5 There are a number of characteristics of the woman bearing the child that have a direct (and indirect) relationship on the likely survival chances of the baby/young child, as well as the mother herself. These include the age of first pregnancy, birth spacing, nutritional and health status of the woman prior to and during pregnancy, activities and care during pregnancy - includinguse of health services, place where the birthtakes place, care of the newborn, feeding of the young child and preventative and curative interventions for the sick child. The education and socio-economic status of the woman play a very important role throughout. These various factors are discussedin turninthe following paragraphs. (i)Theexpectantmother 5.6 The average age at marriage (cohabitation) and age at first birth are important for child survival. Typically having a first birth at age less than 20 years is associated with higher infant mortality. This is true for bothrural and urban areas. While the average age of marriage has risen to 17.6 years in the second RCH survey (up from 17.1 years in the first survey), almost 12 percent of first births are to mothers who are 18 years of age or younger. (Table A9). 5.7 Physical work during pregnancy. The van Dillen Bolangir study found that women, on average, engagedinphysically demanding work up to the sixth month of pregnancy, longer if the household depended crucially on the income or seasonal needs dictated the labor need. Insights suggest that this practice has changed little in recent years. Interestingly there i s little variation across caste groups. Insome communities, the pregnant woman's relatives would arrive inthe 7th month of pregnancy, bearing gifts and staying till the birthof the baby, thus relieving the soon-to- be-mother of housework and assisting with the birth. In other communities, young women travel to their parents' village for the birthand stay there for some months after the birth. 5.8 Nutrition. Poor nutrition i s another key factor affecting maternal and child health - bearing children in poor nutritional condition further depletes a woman's body, with far-reaching effects on her health and that of her children. Poor nutrition among pregnant women in countries with a very highmaternal and neonatal mortality rate is found to be a major contributingfactor to those highrates. InIndia, anemia i s reported to be a factor in more than 60 percent of maternal deaths. Gender stratification and attitudes also contribute through household behavior that deprives poor women of adequate nutrition, which leads to small stature and higher risk of delivery complications. 5.9 The RCH data do not provide any insights into maternal nutrition, but we do have some information from the van Dillen Bolangir study. Attitudes, knowledge and practice vary considerably in western Orissa. Older women support the notion that pregnant women should eat as little as possible in order to have small babies and an easy birth, and also contend that dietary supplements are leading to more caesareanbirths. Younger women on the other hand seembetter informed about the importance of good nutrition during pregnancy, taking their information from ANM/Anganwadi and the media (including TV). Dependingon seasonalavailability, in about 50 percent of cases women report a varied diet, though somewhat lacking in animal products. This eating practice continues through the lactating period. There is an interesting association between expectant father's education and their knowledge of the importance of good nutrition during Applying the Framework to Orissa 17 Achieving TheMDGs in India's Poor States pregnancy: more educated fathers are more inclined to understand the benefit of good nutrition and do whatever they can to provide for their pregnant wives. 5.10 Health seeking behavior of mothers-to-be. Effective prenatal care can benefit mothers and unborn children. Complicated pregnancies can sometimes be flagged during these visits, thereby averting complications at the time of the birth. In addition, these visits help mothers, especially first time mothers, learn about what they should and should not do to ensure a safe delivery. The in-depth Bolangir District study suggests that more women are seeking medical assistance when they detect some form of complication with their pregnancy - an occurrence in around one quarter of all pregnancies. On the precautionary side, according to the RCH data, more than 70% of women record one antenatal check-up during their pregnancy, and a little more than 50% received three antenatal care checkups during their most recent pregnancy (some at home, some at a clinic). A very high proportion of pregnant women, about 86 percent, received the tetanus toxoid (TT) vaccine prior to their last birth - an important step forward in reducing deaths of mothers and babies due to tetanus. However, the fact that the rate of TT vaccination is higher than that of a single ANC visit suggests that opportunities to undertake check-ups are beingmissed. While health workers have a mandate to carry out immunizations, they may lack the skills, time, equipment, incentive and/or household "permission" to performANC check-ups. 5.11 Factors associatedwith greater number of ANC visits in Orissa are similar to elsewhere: first pregnancy, ethnicity, women's education, and wealthhcome. For example, while 43 percent of scheduled tribe mothers-to-be received antenatal care, 70 percent of non-backward castes received full three antenatal care checkups. Although there i s variation in tetanus toxoid vaccine coverage rates across caste groups, the variation is smaller than for ANC checkups. The same pattern emerges across poverty groups. Dividing the households into welfare quintiles" almost eighty percent of all mothers in the lowest welfare quintile (i.e. the poorest) received the tetanus toxoid vaccine, while 93 percent of mothers in the richest welfare quintile holds received the TT vaccine. However, the disparity across welfare quintiles is stark for the proportion of mothers receiving three antenatal care checkups. Only 41 percent of poor mothers receive 3 ANC checkups, while 82 percent of women from the top welfare quintile mothers receive 3 ANC checkups (Figure 5.1). Quality and the value of preventative care as well as the cost of accessing ante-natal services are likely to be factors inthis outcome. (ii) birth The 5.12 Seasonalfactors. In-depthfieldwork from Bolangir District shows considerable variation in mortality rates across seasons. A large proportion (45%) of child deaths occur injust three months of the year, July, August, September, i.e., the rainy season. Immediate causes of death are thought to be fever, malaria, diarrhea and pneumonia. Duringthis time of the year travel by road i s very difficult making medical assistance difficult to access if i s needed. Also during this time of the year, viral fever and colds are widespread, and are particularly harmful to new mothers and young children. The cold season (November-January) i s also associated with more than average deaths. This i s thought to be associatedwith the need for heating, which i s typically an open fire with either firewood or cow-dung, and if usedcan lead to serious indoor pollution from smoke. 22 This is done using an asset index as a proxy for household consumptiontincome (since the RCHsurveys do not collect income or consumption data, but do collect asset information - durable goods, housing characteristics). SeeFilmer and Pritchett (1998) for more details on this methodology and its robustness as a measure of household welfare. Applying the Framework to Orissa 18 Achieving TheMDGs in India's Poor States ~ ~ Figure 5.1: Proportion of mothers receiving antenatal care and tetanus toxoid vaccines acrosscastes and welfare quintiles 100% I-'--- 100% 90% 90% 80% 80% 70% 70% 60% 6O"io 50910 5036 40% 4036 30% 30% 20% 20% 10% 10% 0% 0% sc ST OBC Other Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Pregnant mothers who receivedat least 3 ANC' Pregnant mothers who receivedat least 3 ANCl Proportionoi mothers receiwg tetanus toxoid Proporlionoi mothers receiung tetanus toxoid , Source: WB estimation usingRCHI1data. 5.13 Birthorder andbirthspacing/previousbirthinterval. Firstbirthsare moreat riskthan subsequent births (even controlling for mothers age) and high order births (after 6 or 7) are also risky, but have very low frequency. There i s a higher incidence of neonatal deaths among first born children than subsequentbirths.But women carrying their first child are more likely to have ante-natal checkups and deliver in a health facility than for subsequent births.While the average birthinterval is inthe range of 24 months, 46% of births occuat a lower interval. Birthintervals of more than 24 months have a very significant (positive) impact on neonatal, infant and child mortality rates; by comparison, birth intervals of less than 24 months tend to be strongly associatedwith higher mortality rates (Tables A8, A9 and A10). 5.14 Place of delivery. Although falling, almost seven out of every ten deliveries are conducted at home in Orissa. The disparity in home based deliveries i s particularly strong across urban and rural areas. In urban areas, 41 percent of all deliveries were conducted in the home, while in rural areas 77 percent of all deliveries are conducted at home. Around three quarters of birthsfor scheduled tribe women are carried out inthe home, while for non-backward castes, 46 percent of births are carried out at home. There is a steady inverse relationship between the proportion of deliveries at home and the household's economic status. While 85 percent of all births in the poorest households are conducted at home, 21 percent of all births in the richest households are conducted at home. Furthermore, there i s a particularly sharp drop in the proportion of births carried out at home between quintile 4 and quintile 5 (figure 5.2). This suggests that delivery in a health facility is particularly valuedcan be afforded by the richest households inOrissa. Mother's education also influences place of birth, as does the availability of a facility. There is, however, no direct relationshipbetween place of birthand neonatal deaths. Applying the Framework to Orissa 19 Achieving The MDGs in India's Poor States Figure 5.2: The proportion of deliveries conducted at home across caste and household welfare quintiles I ( 1I90% 80% ` i 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 1 Scheduled Caste Scheduled Tnbe OBC Other Quintile 1 Quintile 2 Quintile 3 Quiniile 4 Quintile 5 I Source: World Bank estimation using RCHI1data. 5.15 Assistance with the birth. In and of itself, delivering a baby at home i s considered acceptable practice if the birth is attended by a trained birth attendant, the birthplace i s sanitary, and there is access to emergency obstetric and intensive neonatal care services, should they be needed. (Obviously being present in a health facility with emergency obstetric and intensive neonatal care prior to the actual birth i s a preferable arrangement, but not always possible.) Almost three quarters of all deliveries are conducted by untrained people, and there i s a higher rate of neo-natal mortality among babies delivered by an untrained person (45/1000) compared to a trained medical worker (37/1000). Relatives and friends conduct 44 percent of all home deliveries, with another 30 percent being conducted by untrained Dais. Births among scheduled tribes are the least likely to be attended by skilled birthattendants. Only one in seven scheduled , tribe births is carried out by a trained attendant, while 37 percent of all births of non-backward castes are carried out by skilled birth attendants. Similar disparities are seen across household welfare quintiles. For the poorest households in Orissa, about 16 percent of all births are conducted by skilled birthattendants, while more than half of all births in the richest households are conducted by skilled birth attendants (figure 5.3). The cost of paying for the services of a trained birth attendant run into the hundreds of rupees (anything from Rs. 200-600; van Dillen) and can be beyond the reach of poor families. For A N M s , the large geographical areas falling in their jurisdiction and their lack of mobility further constrain their ability to be present at home births. 5.16 The van Dillen study in Bolangir District found more than 90% of births took place at home, and only 10% of these were attended by a trained birth attendant: in 83% of the home deliveries an untrained female relative alone was at the birth, and in the balance a traditional healer was called in. Considering the state of communications, infrastructure and transport in the area-as well as the households' hygiene practices (discussedbelow), home deliveries of this sort are highly risky. Should complications arise, competent medical assistance is often sought only when mother and/or baby are ina critical condition, often coming too late.23 23 Van Dillen finds a high incidence of baby deaths occurring at the time of birth, or shortly thereafter, from complications such as breech position. Applying the Framework to Orissa 20 Achieving TheMDGs in India's Poor States Figure 5.3: Birthattendants across caste and household welfare groups 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% SC ST OBC Other Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: WorldBank estimation usingRCHI1data. (iii)CareofNewbornandyoungchildren 5.17 Field study data suggest that the newbornchild andthe mother are cloistered inthe house for the first one to two weeks following the birth, either in a separate roomor in a secludedplace. Contact with the baby and the mother are restricted. 5.18 Post-natal visits by ANM. Less than 15% of new mothers get a visit from the ANM after giving birth (RCH data). However, there i s an interesting pattern of visits: a greater incidence of post-natal visits by ANMs occur in the rural areas, among ST mothers, less educated and poorer women, than among others. Van Dillen's Bolangir study suggests that around 40% of children get at least one medical check-up duringthe first 6 months of their life, but this could be initially with a traditional healer rather than an allopathic doctor, and in response to an illness. Issues of accessibility and cost can lead mothers to traditional healers rather than an allopathic doctor. 5.19 Breastfeeding. Exclusive breastfeeding for the first 6 months of a child's life is closely associated with higher rates of child survival in the developing world.24Both the RCH data and the van Dillenstudy report that young mothers obtain information on breastfeeding mainly from female relatives, rarely from medical practitioners. Breastfeeding practices, therefore, are largely determined by household and community traditions, which i s not necessarily advantageous for the child's health. Inthis manner, colostrums i s often squeezed out (claiming it i s impure, and bad for the child), and the young baby i s fed cow's milk or sugar water untilthe mother's milkbegins to flow. There i s some evidence that more educated and better-off mothers start breastfeeding sooner and allow babies to have colostrum. However, less educated and poorer mothers are more inclined to breastfeed longer. ST women, reflecting these influences, are more inclined to breastfeed longer but avoid giving their babies colostrum. State-wide only 20 percent of mothers practice exclusive breastfeeding up till the age of 6 months (Table A6). 5.20 Breastfeeding on demand requires the mother's presence. For most women, this is possible during the first two months of the baby's life, but for poorer women who are requiredto 24The Lancet 362, 2003. Applying the Framework to Orissa 21 Achieving TheMDGs in India's Poor States return to work soon thereafter it can become a problem. Young babies are sometimes left with elderly relatives or a young sibling, with mother's returning from their work every 2-4 hours to feed the baby. This can involve a cut in wages. Other mothers take their young babies to the field with them and attend to them as needed. Women typically seek work most compatible with regular breastfeeding during the first 6-12 months, as much as possible, a practice more easily followed among the better off households(van Dillen). 5.21 Weaning. This reportedly can start any time between 3 months and 10months. The most common weaning foods are enriched milk powder, rice paste, rice cakes, biscuits and cows milk (van Dillen study). 5.22 Health problems of the young child. Newborns are frequently affected by three different afflictions: diarrhea, fever, and skin rashes. A number of other afflictions affect slightly older children and can contribute to their death; pneumonia, diphtheria, measles and malaria. (Fever related to malaria i s the number one cause of death among young children inthe Bolangir study area.) Concerning choice of treatment for young children, many households start with home remedies (such as turmeric and figs for the treatment of diarrhea), then turn to traditional healers (in the case of skin rashes in particular) then to a traveling quack (who is favored because he comes to the village, thus reliving the family of travel expenses and he will often provide services on credit). Only ifthe case looks sufficiently serious i s the baby taken to a public or private clinic. Reasonsfor not going to a clinic include (i) difficulties with transport - condition of the road and the expense of hiring a vehicle, (ii) cost of medical serviceshreatment, (iii) cost of the the medicines and pathological tests, and (iv) the requirement that subsequent visits occur. A high proportion (108 out of 173) of infant fatalities arising from sick children in the Bolangir study never found their way to medical assistance prior to dying. Much more attention i s clearly needed to the integration of preventative and treatment of sick children, starting from the home and then taking the child to a qualified healthcare provider. 5.23 Immunization.More than 60 percent of all children between the ages of 1and 2 years have received 3 doses of DPT, 3 doses of polio and a vaccine against measles, and 55% of children under the age of 3 years are fully immunized (RCH data). More than 80 percent of children have received the BCG vaccine, but only 28 percent of babies received their polio OPV 0 drops. Immunization rates vary predictably across caste groups and across household welfare groups (Figure 5.4). Scheduled tribes and scheduled castes respectively account for among the lowest immunization coverage rates. Similarly, immunization coverage rates increase with household welfare quintilesfor BCG, DPT3 and measles.The relatively low rate of immunization against measles is of particular interest to this paper, in so far as it can make an impact on reducing child mortality. Set against an overall rate of 68% coverage ~tate-wide,'~coverage of ST/SC and poor populations at around 60% i s of particularconcern, falling well below the target of universal immunization of one-year olds. 25Orissa's performancewith measles vaccinationin"middling" among states. See fig. 11.24inWorld Bank (2004). Applying the Framework to Orissa 22 Achieving TheMDGs in India's Poor States Figure 5.4: BCG,DPT3and measlesimmunization rates of children between 1and 2 years of age - across casteandwelfare quintiles 1 I120% 90% Wfo 100% 70% 80% 60% 50% 60% 40% 30% 40% 20% 200lO 10% 0% 0% sc ST OBC Other I Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 BCG iDPT3 DMeasles ~ ~ Source: World Bank estimation using RCHI1data. 5.24 Knowledge and treatment of diarrhea reflect education and income levels of the population, thus there is an urban bias, on the one hand, and poorer practice among ST population on the other hand. (There i s no apparent child gender dimension though.) But state-wide more than 80% of mothers claim knowledge of appropriate treatment to deal with diarrhea. (iv) Mothers' education -one of the strongestfactors influencing child mortality 5.25 Education of women affects child mortality indirectly, through a range of intermediate factors. Mother's educational attainment in Orissa is indeed an important factor, with higher levels of educational attainment contributing to lower levels of under 5, infant and neonatal mortality (figure 5.5). The literature i s abound with evidence on the link between good health outcomes, whether for the mother or for the child, and the educational attainment of the mother. Mothers who are more educated tend to delay marriage, which in turn delays child bearing, are more inclined to seek health care for themselves and their children, have a greater decision making role inthe family, practice better hygiene, nutrition and health management (treatment of new-born, washing and feeding practice; feeding and weaning of older child; immunizations; care and treatment of sick child, etc.) and seek health care when it i s needed There can also be an income effect through more educated mothers having higher earnings. In this respect, Orissa is at a distinct disadvantage. Almost 60 percent of mothers in the 15-44 age range have no education and 13 percent have less than a primary school education. Conversely, only a quarter of all mothers have at least primary school education The strongest education impact occurs among women with more than 5 years of education, and as shown in figure 5.5, childmortality rates are considerable lower with this level of education. Moreover, this relationship holds true when all other variables are held constant (Tables AS, A9 and A10). Applying the Framework to Orissa 23 Achieving TheMDGs in India's Poor States ~ ~ Figure 5.5: Child, Infantand Neonatal mortality rates by mother's educational attainment I No Education Primary or less More than pnmary I Source: WorldBank estimationusingRCHI1data. 5.26 The importance of women's education is further reflected in health seeking behavior during pregnancy. Only 40 percent of mothers with no education received 3 ANC checkups, while 72 percent of mothers who have more than a primary education received 3 ANC checkups. Again, the disparity in tetanus toxoid vaccine coverage rates is smaller, with 78 percent of mothers with no education receiving the TT vaccine, and more than 90 percent of mothers with some education receivingthe TT vaccine (figure 5.6). Figure 5.6: Proportion of mothers receiving antenatal care and tetanus toxoid vaccines across mother's education .. Md% 80% 60% 40% 20% 0% No Education Primary or less More than primary 10% mothers recervlngtetanus toxoidvaccine I 1% mothers receimg 3 ANC checkups 1 Source: WorldBankestimationusingRCHI1data. B. THEFAMILY 5.27 There i s a tendency to treat all members of a household, or family, as a single unit assuming that whatever benefits one member will benefit the entire household. This may not always be true. Intra-household differences in gender and age may significantly affect how Applying the Framework to Orissa 24 Achieving TheMDGs in India's Poor States decisions are made and whether a decision is beneficial for all members. Gender disparities in access to education, credit and political influence have considerable impact on how individual family members are valued and on the degree to which women (as well as men) have a voice in household decisions about health and health care. Recognition of the importance of individuals and households in reducing child mortality should lead policy makers to consider both the individual woman and the broader household. 5.28 There are a number of characteristics of the family in which the pregnant woman belongs and into which the child is born that have a direct (and indirect) relationship onthe likely survival chances of the baby/young child, as well as the mother herself. These include the assets and resources of the family: its size and composition, physical assets (land, livestock, business, property, house), financial assets and income, amenities - such as water, sanitation, electricity, telephone, and availability of food. A household's daily practices and routines also have an important influence on the condition of a pregnant woman, young mothers and young children. They include the attitudes and behaviors of the family members vis-his hygienehanitation, cooking and eating, the work burden carried by the pregnant (and lactating) mother, interaction with health-care providers during pregnancy, at the time of the birthand after the birthas well as the birth place, care of the newborn, feeding of the young child and preventative and curative interventions for the sick child. The socio-economic status of the woman within the household will also be an important factor. Some of these aspects for which Orissa data exist are discussed below. 5.29 Water and sanitation. Provision of safe water has contributed to declines inmortality in many developed countries, and poor sanitation and unsafe water, along with poor personal hygiene are known to be major factors in the wide prevalence of parasitic diseases. Waterborne diseases can undermine the health of pregnant women because they cause anemia, a risk factor for pregnant mothers as well as their unborn children. In a review of over two thousand publications, Fewtrell et al. (2005) summarized the evidence as follows: (i)water quality interventions reduce diarrheal illnesses; (ii)household water connections also reduce the incidence of diarrhoea; (iii) hygiene interventions, especially hand washing, are very important for reducing diarrheal illnesses and upper respiratory-tract infections, and can be effective in areas of poor water and/or sanitation; (iv) the impact of improved sanitation i s thought to be positive inreducingdiarrheal illnesses, but few studies are robust intheir design; and (v) multiple interventions consisting of water supply, sanitation and hygiene education do not necessary strengthen the impact over and above any one individual intervention, although improved water quality at point of consumption was rarely included inthe studiesaZ6 5.30 The link between water supply and health of mothers and children involves both household and community factors, and the ability of a household to translate piped water, for example, into lower morbidity and mortality is by no means straightforward. Indeed a study in India (Jalan and Ravallion, 2001) found that health gains from piped water largely bypassedpoor families, although reduction in diarrhea among children in better-off households was significant. Other work has noted that it i s the better hygiene practices (especially hand-washingz7) and sanitation (use of latrines) that has a biggerimpact on reducing child mortality. As far as these are facilitated by the convenience of piped water inthe house (for frequent hand washing) and a place to dispose of feces (latrines) the infrastructure i s an important factor for health outcomes, 26 Reduction in incidence of diarrhea across these five interventions i s as follows: (i) 15-39%; (ii) 19-23%; ''(iii)33-42%;example (iv) 24-36%; and (v) 30-33%. See, for Luby et al. (2005) "Effect to hand-washingon child health: a randomized controlled trial". InTheLancet, vol. 366, July 2005. Applying the Framework to Orissa 25 Achieving TheMDGs in India's Poor States especially the reduction of diarrhea and associated child mortality. Quantity of water is particularly key, in large part due to the increasedfeasibility of adopting safe hygiene practices. 5.31 The RCH surveys collect information on the source of water for the household: tap (inside and outside), hand-pump, covereduncovered well, open water source and any other. (There are no data on quantity of water nor on hygiene practices.) The data do show a positive relationship between tap water and (lower) child mortality, especially at older ages of children. The effect is most pronounced inthe urban areas where taps are more prevalent, especially inthe home. Inrural areas, when other factors - suchas householdincome and women's education - are taken into account the positive effect disappears. This i s not to say that access to clean water is unimportant, but other factors may be more important. Also, how clean water i s usedinbaby care and child rearing is hugely important. The effect of modem sanitation on infandchild mortality is similar to water: there i s a consistent positive impact on lower mortality as a linear relationship, but the relationship i s not significant where other factors are taken into account (as shown in the regression tables inthe annex of the paper). Inall likelihood, the accompanying behaviors are not present inmany rural households. 5.32 The van Dillen Bolangir study confirms that hygiene practices are woefully inadequate: few households boil water (not even for weaning foods); in virtually all households the women reported never washing their hands with soap before cooking (including the preparation of weaning foods); use of pit toilets, even where they exist i s rare, and hand washing after defecation i s also very rare (15% of households on an occasional basis). Results from a baseline study on health outcomes of sanitation project in Bhadrak District (2005)*' are somewhat more positive: hand washing among adults and children was quite common, although use of soap was likewise rarez9;use of pit latrines was low (10% of households) as was treatment of drinking water (boil, treat or filter, 11%). The incidence of diarrhea among children under five was high, with close to half of all children having had an episode of diarrhea inthe month prior to the survey (August and September) . 5.33 Cooking fuels. Various studies in India (Hughes and Dunleavy, 2002; van der Klaauw and Wang, 2004) have shown that indoor air pollutioncausedby certain cooking fuels contributes to infant and child mortality. Indoor air pollution is associated with acute respiratory infections, asthma, blindness, chronic pulmonary disease, and tuberculosis - all illnesses found extensively in Orissa. Other studies have modeled the gains in child survival from switching to cleaner cooking fuels (such as kerosene and liquid petroleumgas). The van Dillen Bolangir study reports that most households in rural Orissa are cooking with firewood, supplemented with cow-dung - both of which produce a lot of smoke, known to be a serious health hazard. While women are most affected, many young children are with their mothers whilst cooking (91/137 households) and inhalingthe smoke. 5.34 Malaria is widespread throughout much of Orissa, especially in the interior districts in the west of the state, and it i s one of the most serious threats to health and well-being. Orissa accounts for a very proportion of the all-India deaths associatedwith malaria, and the number of reported cases of malaria has remained stubbornly high (at around 400,000 p.a.). Young children are particularly at risk, for they are much less resistant than adults. The Government of India promotes a number of strategies to address malaria, including early diagnosis and treatment, indoor spraying, insecticide treated bed nets, larvivorous fish, and behavior change communication. According to a UNICEF study in eight districts of Orissa, less than 50% of ''Undertaken by RTIInternational, under the guidance of the Orissa Department for Rural Development. 29 Knowledge about the importance of soap was however very high among adults. Applying the Framework to Orissa 26 Achieving The MDGs in India's Poor States pregnant women took chloroquine, and only 2% of children slept under a treated bed-net. Likewise, less than half the households in the van Dillen Bolangir study area reported taking precautionary measures against malaria, with less than 15% of households using mosquito bed nets - the balance using smoke from straw, leaves and occasionally commercial coils. Van Dillen's data suggest a strong relationship between use of precautionary measures against malaria and lower infant and child mortality. Reasons cited for not usingprecautionary measures include lack of awareness of the danger of mosquito bites, the cost of mosquito nets, and the unpleasant feeling associatedwith sleeping under the nets (a feeling of suffocation). -- --- I__-p *---*- Box 5.1: Ante-nataland Post-natalCare and Practices: Findingsfroma survey of eightdistricts Data collected from households (1400/district), monitoring of practices at health sub-centers (30) and facility surveys inhealth institutions (lO/district) found the following: 0 Less than 5% of home deliveries attendedby a TBA are "clean" 77% of newborns are bathed within first 48 hours raising the riskof hypothermia Less than a half of infants receive some post-natal care: only 10% receive full post-natal care Only 50% of pregnant women take chloroquine 0 Less than 10% of women receive ANC inthe first trimester Less than half of infants are exclusively breastfeed for 6 months Less than 4% of children under 5 years benefits from safe water sanitatiodhygiene practices Less than 2% of children under 5 years sleep under a treated bed net at night 0 Less than one third of children with diarrhea receivedoral re-hydration treatment 5.35 Household resources assets and income. Poverty acts through many channels which - ultimately affect infant and child mortality. Chief among these factors are work load of women during pregnancy and lactation, nutrition status of women and children, affordability of healthcare and drugs, and quality of housing and amenities in the house. According to the RCH data, both under 5 and infant mortality rates vary considerably across household welfare groups. Unsurprisingly, there i s a monotonic decreasing relationship between child mortality and household wealth (assets). While the poorest households lose around 94 children before the age of five per 1,000 live births, the richest households lose around 43 children before the age of five. IMR too decreases with household wealth (assets), with 77 deaths per 1,000 live births in the richest households and less than half (38 deaths per 1,000 live births) in the richer households. Neonatal deaths also fall as income rises, though the difference across income groups i s less striking than with older children. Other indicators of mother's well-being duringpregnancy, birth and child-bearing-related behavior show a similar pattern with household wealth (assets): better- off households have a higher incidence of ANC check-ups, especially with a doctor, a higher proportion of pregnant women get tetanus toxoid injections, deliver their babies in either a government or private hospital with a trained medical practitioner, and have their children fully immunized (figure 5.7). Applying the Framework to Orissa 27 Achieving TheMDGs in India's Poor States Figure 5.7: Variation inchild and infant mortality acrosssocial identity and welfare quintiles Scheduled Schediiiedlribe Mherbackward Other Lk not bow Quintile1 Qiiintile le 3 Quinile 4 Guinlile5 Source: World BankestimationusingRCH I1data. 5.36 Social Identity. On the surface this is also an important factor in explaining different levels of childhnfant mortality. As shown in Figure 5.7 (above), the incidence of child and infant mortality i s higher among scheduled tribe families than any other group: child mortality at 100 per 1,000 live births, infant mortality at 80/1000 and neonatal mortality at 51/1000 are all well above the rates for other groups. By the same token, healthcare seeking behavior among pregnant ST women also shows fewer ANC visits, a very high incidence of home delivery with untrained persons and lower rates of child immunization. However, when other factors are taken into account, most importantly poverty and education of women, the mortality outcomes in ST families are not so different from any other groups of the population. In other words, if ST households had the same level of education and income as other households, in all likelihood their child survivallmortality rates would be similar to the rest of the population. But, in so far as ST populations are poorer, less educatedand live inremoter parts of the state, a concerted effort will be needed to compensate for these other factors in order to have a positive impact on reducinginfant and childmortality. 5.37 Household level behavior and risk factors are influenced and reinforced by conditions in the community. Community factors include both the values and norms that shape household attitudes and behaviors and the physical and environmental conditions of the community. Beyond the family into which the young child is born lies the community, also with various characteristics, resources, assets and institutions which can have a bearing on the survival probabilities of the child. Rural infrastructure such as roads, schools, health facilities, anganwadi center, etc., would all seem to be important factors which could influence infant and child mortality, as well as environmental conditions (water, air quality, weather, etc.). Likewise the institutions of the village - formal and informal - could also play a part in determining who (which people) get access to what facilities and services. Gender norms can also be influenced by the community, and the existence of community groups, the degree of social cohesion can also Applying the Framework to Orissa 28 Achieving TheMDGs in India's Poor States affect practices. While these are important relationships to probe, the RCH data have limited information upon which to develop an ~nderstanding.~' 5.38 Roads and transport. Delays in reaching treatment facilities pose critical life- threatening obstacles for women who experience an obstetric emergency. Such delays can be the result of physical accessibility factors such as a distance to a facility, the availability and cost of transport, and the conditions of the road. InZimbabwe, for example, a study showed that the lack of availability of transport accounted for 50 percent of maternal deaths related to hemorrhage. Research in Nigeria, Uganda and Tanzania finds that poor roads, lack of vehicles and high transport costs were major causes of delay in deciding to seek and in reaching emergency obstetric care. The van Dillen study in Bolangir District likewise found that one of the key reasons for taking a young child to hospital only when its condition was critical was difficulty with transport (road condition) and the expenditure associated with hiring a vehicle. Villagers identified "giving birth" as a risk because mothers could not reach health centers due to inadequateroad access, particularly difficult during the rainy season, if they neededto. 5.39 Inthe RCHdata the existence of an all-weather roadshows apositiverelationship with lower infant and child mortality, and is closely correlated with the full course of ante-natal check- ups andgiving birthina healthfacility inthe rural areas (Annex Tables A l l and A12). D. SERVICES 5.40 Health and Nutrition Facilities. Orissa is one of the most sparsely populated states in India, and with many facilities "norm based" this low density poses a particularly difficult service delivery challenge. Primary healthcenters, community healthcenters, and hospitals are within the same village for a very small fraction of households. However, it does appear that ICDS centers are relatively numerous, with more than 80 percent of households having an ICDS center within the village. Sub-centers too are within the village of around a thirdof all households inOrissa. As far as one can tell fromthe RCHdata, there is little or not relationshipbetween the existence of a healthfacility and infadchild mortality outcomes. The one exception is lower neonatal mortality rates where a private hospital i s close by (see Annex Table A9). Of course, these data tell us little about quality of service (are service providers present and knowledgeable about the services they are expected to provide, for example), and non-physical accessibility of the service to the potential clientele - all critical to having an impact on neonatal, infant and child mortality. On these matters, we have some insights from the in-depthstudy in Bolangir District, and these are presentedin the next few paragraphs. 5.41 Inconnection with a scheme under which pregnant women and their newborns are given free medicines (including folic acid) during pregnancy and vaccines inpublic healthfacilities, the van Dillenstudy recorded that the anganwadi worker, who i s suppose to do the registrationwould do so only if paid a bribe (rates of Rs. 10-100 were recorded). By the same token, A N M s who are suppose to assist with deliveries are reported to charge between Rs. 200 and Rs. 600 on a routine basis for such services, an even larger amount if the ANM i s called "out of proper working hours" on an emergency basis. 5.42 Absenteeism of staff working ingovernment health facilities i s a common problem, fully confirmedby the Van Dillen study. This i s true for both PHCs and hospitals, and involves a wide 30This is an area where further study would be helpful. Applying the Framework to Orissa 29 Achieving The MDGs in India's Poor States range of staff from the anganwadi worker, the ANM and doctors.31At best, staff adhere to core working hours and finding medicalhelp after regular hours i s a particularproblem. When a trip to a PHC or hospital involved the hiring of a vehicle suchvain attempts to see a doctor involves the waste of considerable expenditure. However, there are cases of committed health workers who performexcellent service and "charge only moderate fees". 5.43 Corruption in connection with the utilization of government medical facilities is a common theme emerging from the field study. It was often reported that the quality of medical care depended on the patient's ability to pay either larger-than-usual fees, or to pay fees for services where no fees should be charged. In some facilities, the entire medical staff, doctors, nurses, and even the peons are complicit in extracting payments from patients. Fees can be demanded either directly or indirectly through obvious neglect. Fees are unpredictable, and can be subject to negotiation. What curbs these practices is the fear of monitoring andcontrol. 5.44 Transport costs and fees for medical services, whether legitimate or not, are not the only reasons why parents hesitate to take their children, or pregnant relatives to hospital before their condition becomes clearly serious. The additional costs that arise in connection with the purchase of medicines and the need for pathological tests are also barriers to service. Doctors in government hospitals are reported to frequently refer patients to private pharmacy shops to buy medicines citing the absence of medicines or their "bad quality" in the hospital. Many pathologicaltests are also referred to private laboratories. 5.45 In sum, people not only face considerable expense in seeking healthcare from a government facility: patients are exposed to a high degree of arbitrariness and ill-will at a time when they most need support and reassurance. 5.46 For many of the reasons cited above, people often turn to the private sector for health care. Quacks and traditional healers are the first choice when the condition i s thought not to be life threatening. (The choice apparently does not reflect deep conviction - most people seem to appreciate the skills of a well-trained doctor - but i s rather a reflection of the obstacles and drawbacks associatedwith approaching a doctor). The quack comes to the village/patient, rather than vice versa, and will often accept payment on credit. 5.47 Once it i s clear that the patient is in a serious condition, according to van Dillen, quite a number of households turn to private physicians - well-trained and well-regarded doctors - if they are available. In contrast to dealing with the public sector, financial matters are straightforward. There are prices for services, and patients and their relatives do not have to put up with arbitrary negotiations about side payments. Although perhaps presenting higher fees, if transport and opportunity costs are fully considered, together with the more predictable nature of costs, turning to private physicians is a credible alternative to seeking health care from government facilities. Moreover, quality service for the most part is forthcoming, and the patient has some control over next steps - medicines, pathological tests and subsequent visits - should these be needed. 31A recent World Bank review of the ICDS/Supplementary Nutritionprogram inOrissa found a 30 percent absent rate among Anganwadi workers. (2006). Applying the Framework to Orissa 30 Achieving TheMDGs in India's Poor States 6. STATE HETEROGENEITY-DISTRICTPROFILES 6.1 We have now summarized the general pattern of relationships, and teased out those factors having the strongest effect on neo-natal, infant and childmortality in Orissa. We have also categorized the factors into four groups: individual woman, family, community and service provision. We can now take the analysis to more disaggregated geographical areas - to the level where action might best be taken. Ideally this could be at the level of the District, Block - even down to the village. Due to the large sample size and sampling frame, the RCH data afford the possibility of looking at district level patterns for some of the variables. Childmortality at the district level 6.2 Data for all 30 districts show considerable inter-district variation with child mortality rates (figure 6.1, and Annex Table Al). Child mortality rates vary from a high of 124 deaths per 1000 live births in Malkangiri to a low of 46/1000 in Balasore. Malkangiri also has the highest IMR (lOO/lOOO) while Sambalpur has the lowestIMR (40/1000). The ranking of districts by child mortality and infant mortality shows some variation - such as Nayagarh and Gajapati which have modest IMRs (52 and 58 respectively) but quite highchild mortality (72 and Sl), suggesting the presenceof mortality factors affecting children inthe 1-5 age range. Two other districts stand out as having high child mortality rates where the mortality i s particularly high in the 1-5 year age period (Kalanidhi and Koraput). In many districts where IMR is relatively low, child mortality is similarly low, reflecting considerable progress in reducing mortality among children in the 1-5 age range- these are typically the best performingdistricts. Figure6.1: Childmortalityratesacrossdistricts I Source: WB estimation usingRCHI1data. State Heterogeneity -District Profiles 31 Achieving TheMDGs in India's Poor States 6.3 There i s less variation with neo-natal mortality rates across districts - a low of 29/1000 (Bhadrak) to a high of 63/1000 (Boudh). In some districts (Khordha, Sambalpur, Sundargarh, Balasore and Dhenkanal) neo-natal deaths account for 80%+ of IMR, and in Boudh it i s more than 90%. Two districts, Rayagada and Malkangiri, stand out as having very high rates of child mortality - 118 and 124 respectively, yet even there the neo-natal mortality rates are not especially high(49 and 48 respectively). This pattern probably reflects the very highincidence of home births, the low incidence of trained medical personnel at the birth and a general reluctance to give birth in a medical facility unless it becomes critical. These patterns vary little across districts. Table 6.1: Child,Infant,and Neo-natalMortalityRatesinSelect Districts Neo-natalmortality Note: Numbers inparenthesis refer to district ranking by child mortality rate, with (1) being lowest and (30)being highest. A. INDIVIDUAL WOMANAM) MOTHER 6.4 Regarding education, there is considerable variation inthe education status of women of child-bearing age: Balasore, Cuttack, Jagatsinghpur, Kendrapara and Puri - all coastal districts have relatively high levels of education, whilst Rayagada, Nuapada and Malkangiri have low levels of education status. As expected, there is a close relationship between levels of education and district level mortality figures: generally speaking, districts with higher levels of education have lower child mortality (especially Cuttack and Jagatsinghpur), while districts having low levels of education have high child mortality (especially Malkangiri and Rayagada). (See Table A3.) As noted above, education of women affects child mortality indirectly through a range of channels: delayed age of marriage and first birth, health seeking behavior during pregnancy, at birth and for young child, decision making role/control of resources within the family, hygiene, nutrition and healthmanagement practices, and feeding and weaning practices to namejust some of the more important channels. 6.5 Antenatal care, delivery and post-natal care. Quite a high incidence of women are getting at least one ANC visit (76%) and 58% are getting three ante-natal check-ups. Considerable variation exists across districts - Bolangir, Boudh, Cuttack, Puri, Sambalpur, Sonapur and Bargarh are getting higher numbers of women into ANC; Malkangiri and Nabarangpur, for example, are not doing so well. However, all districts are doing very well on tetanus toxoid, even the ones that struggle on other services (86% average coverage: Annex Table A4). StateHeterogeneity-District Profiles 32 Achieving TheMDGs in India's Poor States 6.6 Despite high levels of ANC visits few women deliver their babies outside the home - 65% of all deliveries in Orissa are in the home. InKendujhar, Rayagada, Koraput, Gajapati and Malkangiri, around 80% of all deliveries are at home.32Most home deliveries are assisted by either an untrained dai (30%) or a relative/friend (44%). Malkangiri and Koraput have a particularly highnumber of births assisted by an untrained person (Annex Table A5). As to the reason for these patterns, the over-whelming majority of women say it i s "not necessary" to go to a health facility. Inthree districts (Bhadrak, Kendujhar and Rayagada) cost was cited as an issue though transport in and of itself was rarely mentioned; poor service quality was of concern in close to ten percent of situations, with four districts recording a higher level of service quality concern (Sambalpur, Anugul, Bargarh and Bhadrak). The extent to which this pattern of over- whelming home delivery in the hands of untrained health workers i s a major factor inexplaining the very highlevels of neonatal mortality throughout the state remains an open question, but there i s evidence to suggest that it i s an important factor. Only a small percentage of new mothers receive a visit from the ANM (14% state wide). In a few Districts, the percentage of visits i s higher - Rayagada (32%), Malkangiri (25%) and Kendujhar (19%) - and in these same districts there i s evidence of additional visits (in Rayagada, for example, 23% of new mothers receiving a second visit). 6.7 Breastfeeding.Breastfeeding practice of the newborn child shows some variation across districts. On average, 45% of babies are breastfed within the first 2 hours of delivery (a significant increase since the first RCH survey) - more than a half in eight districts, with Puri topping the league with over 60%; by comparison less than a third of babies born in Angul, Sambalpur and Bargarh are breastfed within 2 hours of birth. Regarding the duration of breastfeeding, while some 81% of mothers breastfeed for a few months, only 16% continue with exclusive breastfeeding for 6 months. Angul, Jajapur, and Sundargarh have the highest incidence of longer term breastfeeding, while inCuttack, Khordha and Nayagada mothers tend to introduce other foods before the childreaches 6 months. InOrissa, as in many places in the sub-continent, the practice of squeezing and destroying colostrum (rather than allowing the baby to absorb it) i s widespread: 52% of mothers follow this practice (72% in Malkangiri): mothers in Nuapada, Khordha, Kandhamal and Ganjam are somewhat more inclined to allow the babies to have colostrumthan elsewhere. B. THEFAMILY 6.8 From the analysis in the previous section we know that household assetdwealth, along with mothers' education, are two of the strongest correlates of infant and child mortality in Orissa. District level data further confirm this relationship. The poorest districts of Malkangiri, Keonjhar and Nabarangpur have high child mortality rates - 124, 84 and 88 respectively - while the better off districts of Cuttack, Puri, Khorda and Anugulhave much lower rates (49, 58, 54 and 64 respectively). However, there are others - Sonepur, Mayurbhanj, Bargarh, Deogarh and Bhadrak, for example - which have a much lower mortality rate than their wealth status would suggest. At the other extreme Ganjam and Sundargarh have higher mortality rates than there wealthrank would suggest (Annex Table A3). 32 Only in Khordha District are a higher proportion of deliveries undertaken in a medical facility (either government or private), probably reflecting the availability of facilities inand around Bhubaneswar, and higher levels of education and income. State Heterogeneity -District Profiles 33 Achieving TheMDGs in India's Poor States 6.9 The ST/SC population mix of a district also has some bearing on infant and child mortality outcomes. But while it i s true that two districts with a high incidence of ST population also have high IMRs (Rayagada and Malkangiri), other districts with large ST populations have relatively low IMRs: Sambalpur, Mayurbhanji and Sundargarh. As noted above, the education level of the ST/SC population i s typically lower than among other social groups, and this combined with the higher incidence of poverty (Table 2.1) largely explain the observed higher infant and child mortality rates among the ST/SC population. C. SERVICES 6.10 Immunization. State-wide the "fully immunized" rate33in 2002/4 i s very similar to that recorded under the first RCH survey (1998/99): around 55% of children under the age of 36 months having received all the recommended immunizations. There is some variation in district performance between the two periods: four districts have significantly improved their performance - Balasore, Ganjam, Boudh and Cuttack - but a large number of districts have slipped, most notably Jajapur, Koraput and Sundargarh. Among the different vaccinations, BCG coverage i s generally good (81%) (Jajapur and Malkangiri slipping a little), DPT 3 and Polio 3 reasonable 62% and 61% respectively - with only Malkangiri standing out with consistently low rates of coverage and Sambalpur with consistent high rates (Annex Table A6). Regarding measles, the state-wide average i s 68% coverage of 12-14 month olds. The districts of Puri, Cuttack and Boudh achieve more than 80% coverage while Koraput, Malkangiri, Jajpur and Jagatsinghpur barely reachthe 50% mark. 6.11 Access to ICDS and health facilities. While access to ICDS centers is pretty comprehensive throughout the state (83% of households have access to an ICDS center), access to health facilities is very varied. At one extreme, the districts of Angul and Kendujhar have pretty good coverage of both public and private facilities (see Table A7). At the other extreme, districts such as Malgangiri and Sonapur have good access to neither public nor private health facilities. How well are districts performing? 6.12 A final part of the district level analysis involves an exercise whereby the actual IMRsfor the districts are compared with "predicted" IMR numbers: "predicted" estimates are made by aggregating a number of household socio-economic values, such as the wealth index, women's education, ethnicity of household, type of house, drinking water and sanitation, electricity, cooking fuel and other household characteristics known to be associated with child (infant) mortality. Inso far as these are some of the main determinants of mortality outcomes, they can be expected to predict the actual observed rates. It i s important to note that the formula used to predict the IMRs does not include any health facility variables - the only service provision data contained inthe RCH survey. This i s becausethese variables are not significant in explaining the observed variation in actual infant mortality rates (see Annex Table A9)34.The results for this exercise are presented in Table 6.2: lower numbers indicate an IMR performance better than household socio-economic characteristics would suggest; a higher number (more than 100%) indicate a poorer performance than household socio-economic characteristics would suggest. 33BCG, DPT (3 doses), Polio (3 doses) and measlesfor children under 36 months. 34I t i s possible that if more disaggregated data were available on health services, especially quality and timely availability of service, some significance vis-a-vis mortality rates would be observed. State Heterogeneity -District Profiles 34 Achieving TheMDGs in India's Poor States -----Table 6.2: Districtlevel performancewith IMR: Goodand badperformers Districts Actual IMR PredictedIMR* Actual IMR as a (RCH data) % predicted IMR* (lower figure indicates better performance) Bhadrak 40 59 69 Deogarh 46 66 69 Sambalpur 45 65 69 Gajapati 58 77 75 Balasore 45 56 79 Cuttack 45 52 87 Mayurbhanj 59 67 88 Koraput 68 77 89 Bargarh 60 67 89 Dhenkanal 53 58 91 Jharsuguda 58 63 93 Sonepur 62 66 94 Keonjhar 67 70 95 Angul 57 60 96 Kalahandi 69 71 96 Khurda 54 55 97 Jagatsinghpur 52 52 100 Sundargarh 67 65 102 Nayagada 61 59 103 Nuapada 74 70 105 Puri 55 52 106 Ganjam 71 66 107 Nabarangpur 73 68 107 Kandhamal 77 70 109 Bolangir 76 69 110 Rayagada 92 80 115 Boudh 70 60 116 Jajpur 75 63 120 Malkangiri 103 82 125 Kendrapara 69 54 128 Total 64 64 * Predicted IMRis basedon wealth (asset) index, mother's education, ethnicity of household and other household characteristics: lower numbers indicate an IMR performance better than household socio-economic characteristics would suggest; a higher number (more than 100%) indicate a poorer performance than household socio-economic characteristics would suggest. 6.13 Through this analysis, three districts stand out as having an actual IMR that i s considerably lower than their socio-economic circumstances would suggest - these districts are Bhadrak, Deogarh and Sambalpur; three others districts - Malkangiri, Jajapur and Kendrapara - are at the other end of the spectrum and have actual IMRs considerably higher than observed conditions would suggest they should be. These six districts - the good performers and the poor StateHeterogeneity-District Profiles 35 Achieving The MDGs in India's Poor States performers - could offer interesting sites for further investigation: the positive deviants and the negative deviants. In particular, household behavior and health service provision vis-&vis antenatal and post-natal care seem particularly important: Sambalpur, for example, does well with ante-natal care, deliveries ina medical facility, exclusive breastfeeding, and immunization despite being a district with average wealth and education, and a relatively highST population (35%). -- ------ Box 6.1: Mayurbhanj:Achievingrapidreductioninchildmortality ~~ Mayurbhanj is a district in the northern part of Orissa. It has a vast land area (10,000 sq.km.), a population of 2.2 million, of which more than 50% belong to schedule tribes and lead lives as huntedgatherers. This i s a poor district (77% of the population are classified as BPL), and the level of literacy quite low (38% among women). Malaria i s endemic in 21 out of 26 blocks. The district i s implementing a wide range of interventionsincluding early diagnosis and prompt treatment, vector control (spraying, bed nets and fish) and BCC. There has been some modest decline in the number of reported cases and deaths. There i s a high incidence of malnutrition, irritated by floods and drought, and pneumoniaand diarrhoea are widespread inthe district. Infant mortality has dropped from a high of 96/1000 inthe late 1990s (one of the highestinthe state at that time) to around53/1000 in2004 (RCH data). Mayurbhanj has been the focus of a number of innovative child survival programs, including CARE India's Integrated Nutrition and Health Project. More recently, with support from UNICEF, the IntegratedManagement of Neonatal and Childhood Illnesses was introduced in Mayurbhanj in 2004 and planning is to cover the entire district by the end of 2006. The linchpinof IMNCI is the Anganwadi workers: trained in a variety of child survival interventions including visits of newborns (1" visit within 24 hours; 3 visits within first 10 days); intense follow-up/case management of sick babies - referral, with associated transport money; food and hygiene practices; and close collaboration between the AWW and ANM andother healthdepartment functionaries. An important part of the IMNCI strategy is to "de-medicalise" vital aspects of child care, provide single standard protocols for identification, classification, referral and treatment by community healthhtrition workers. The program also promotes partnership with local NGOs, and uses "freelance" trainers and monitors - localyoung menand women with grassroots experience.At the Block level, there is regular convening of progress review meetings, facilitated by UNICEF/local NGOs, with a particular focus on planning and problem analysis. Like many community based programs, the success of this IMNCI rests with the quality and effectivenessof capacitybuildinghraining of community level workers, regular contact with those workers, the content of the message, availability of supplies, and support for community workers to reach out to marginalized neighborhoods. A review of CARE India's INHP found that the involvement of women's groups and PRI institutions, and coordination mechanisms between ICDS and MOHFW at block and district levels was instrumental in the success of its various child survival interventions (CARE India, 2004). -- --- State Heterogeneity -District Profiles 36 Achieving TheMDGs in India's Poor States 7. CONCLUSIONSAND POLICY IMPLICATIONS -THE ROADMAPTO LOWER CHILD MORTALITY 7.1 Thus far, this report has focused on two aspects of child mortality in Orissa, namely the various factors having an impact, either directly or indirectly on neo-natal, infant and child mortality, and the variation across districts inboth "inputs" and "outcomes". Many of the findings of this investigation confirm what i s more broadly known about the child mortality situation in India, but some of the findings are more unique to Orissa. Examples of the latter includet h e m seasonality of infant mortality during the rainy season, the role of malaria, and the very Door environmental health practices (hygiene, cooking fuel). These Orissa specific factors, coupled with the highrates of poverty, low levels of (female) literacy, and the exceptionally poor state of public services (including health services) and infrastructure, especially roads, result in child mortality rates that among the highest in the developing world. This final part of the report attempts to answer three questions: what is needed to bring down child mortality rates in Orissa and achieve the 1I* Five Year Plan and MDGgoal; how well are existing interventions placed to do the job; and where are the gaps and how can they best be filled. These are examined in this chapter, which culminates in an outline of a possible multi-sectoral program designed to reduce childmortality inOrissa. Table 7.1: The FrameworkRevisited:What the Orissa evidencesuggests A. Individual woman/mother B. Thefamily Education (esp. post-primary) Income and wealth Nutrition (during pregnancy and breastfeeding) Intra-householddynamics Age at first birth and spacing between births Water (piped into house) C. The Community D. Services Environmental health practices Basic healthhtrition services in village/outreach to (water, sanitation, solid waste) households Beliefs and practices (e.g. at the birth) Access to health facilities for emergency obstetric Women's self-help groups and sick child care Active PRIs Other services, such as school, roadtransport and electricity Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 37 Achieving TheMDGs in India's Poor States Summarizingthe evidence 7.2 Table 7.1 (above) and Table 7.2 (below) summarize the evidence discussed in previous chapters, and begin to tease out those factors, practices, and services which have the biggest impact on child mortality. Table 7.1 uses the same framework introduced earlier in chapter 4, but narrows the list to those factors which are validated by the Orissa specific data. This table is organized into four quadrants or spheres with each sphere representing an important segment of the production function of child survival. But while shown as separate spheres there are significant interactions and inter-dependences across all of them, and they are best seen as four parts of an integrated whole. Table 7.1 captures the variables contributing to all levels of child mortality, and Table 7.2 breaks the analysis down into neo-natal, infant and under 5 mortality. Table 7.2 further builds on the regression results (presented in the annex tables A8-10) and as such i s able to suggest the strength of the relationship - defined here as strong, moderate or weak -forthosevariablesforwhichRCHhouseholddataareavailable. -p--p-----m--^--I- 1 "_____^ Neo-natal Infant Under 5 mortality mortality year mortality Factors underscoredby quantitative data* Age of mother at first birth Weak Moderate Weak -* -----, Notes: See Tables A8 (under five mortality), A9 (infant mortality) and A10 (neonatal mortality) for * .---.--- detailed regression results, which i s the source of classification of strong, moderate and weak,+ Van Dillen study inBolangir District, and other qualitative data. Since these qualitative studies do not have sufficient measurement to classify the strength o f the effect, we have use moderate throughout. Conclusionsand Policy Implications - TheRoadmapto Lower ChildMortality 38 Achieving TheMDGs in India's Poor States What interventionsare needed? 7.3 One of the striking features of the factors, practices, and services i s their multi-sectoral nature, and the importance of actions in four separate, but closely related domains (the individual woman, family, community and service delivery). As such, progress with reduction of child mortality depends on a wide range of actors and actions. As "producers" of good healthoutcomes for themselves and their babieskhildren, individual and household behaviors are central to child survival; the income and wealth of households, community support systems, and nutrition, education, the environment, water, sanitation and a variety of other factors are all relevant in the quest for falling child mortality. Some of these factors also present themselves as obstacles to accessing healthcare critical to a healthy pregnancy and childbirth, including financial and physical barriers such as an all-weather road, distance to a health facility or the costs associated with getting there. On the supply side, a whole set of variables can undermine the effectiveness of care when it is sought, for example getting access to healthcare providers in a timely fashion, knowing ahead of time the payment schedule (should it apply), the availability of medicines, provider performance and appropriate treatment. 7.4 Buildingon the evidence we can narrow the list of important interventions to a short-list of twelve. We have further divided these into three categories: relatively quick and easy to implement; more involved and needing more time; and those requiring a longer time horizon. These are presentedinTable 7.3. 7.5 These interventions are clearly quite different in nature, span a very varied time-frame, and involve different levels and types of resources. Also, as illustrated in Table 7.3, the role of these interventions varies across neo-natal, infant and child mortality and as such the optimal mix of interventions could change depending on the mortality profile. For example, a district with high post-neonatal mortality (such as Koraput) might need to place equal emphasis on the interventions inthe righthand column of the table, whereas a district with highneonatal mortality but low post-neonatal mortality (such as Khorda) might need to place more emphasis on interventions inthe first column. 7.6 By way of illustrating the potential impact of different interventions on reducing infant and child mortality, we have undertaken some simulations with the RCH data. Starting with the results of the regression for infant mortality (Annex Table A9) we selected those four policy variables which are thought to have a significant impact on reducing infant mortality. These are girls post-primary education, household income, birth spacing and all-weather road. We then raised the value of these key variables - for example, assumed that all women of child bearing age have post-primary education, or all households have access to an all-weather road - while everything else stays the same, and assessed the impact of this on reducing infant mortality. In this exercise the decline ininfant mortality is attributed directly to the policy change and ignores any behavioral change as a response to the policy change. This might lead to an over-estimation of the positive effect of the policy change.35The results are summarized in Table 7.4 (for three scenarios) and figure 7.1 (one scenario). 36 35Lee, Rosenzweig and Pitt, 1997. 36 Similar exercises have been undertaken for all-India, for example, by van der Klaauw and Wang (2004), and Deolalikar (2004). Intheir paper, Subbarao and Raney (1995) find a reduction of 25 infant deaths per 1000 live births from a doubling of the female secondary school enrollment in a set of low-income countries. Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 39 Achieving TheMDGs in India's Poor States Table 7.3: The Multi-Sectoral Menu of Interventions, ranked by difficulty and time- horizon Conclusionsand Policy Implications - The Roadmapto Lower ChildMortality 40 Achieving TheMDGs in India's Poor States Table 7.4: Simulated reductions in Infant Mortality from changesinfour policy variables: Number of infantdeaths prevented DouEng the value Raisingthe value Utopia: raising the of the indicator of the indicator to value of the from its RCH2 the levelobserved indicator to 100% level in "best" district Education levelof women of child bearing age: at least completed 9 9 23 primary school Poverty - as measuredby the 11 13 n.a. wealth index Birthspacingof morethan24 8 2 10 2 3 26 35 64 64 38 29 41 41 y_-II_p __I__I_x"~~c_-L.yL-;__yI Notes: Present values of indicators are as follows: at least pi nary education, 34% f women; all-weather _x- road, 54% of villages; births spacedat morethan 24 months, 28%.-"Best" districts are as follows: schooling (Cuttack); Wealth index (Khurda); Proportion of mothers reporting a birth spacing of more than 24 months (Nuapada); Proportionof households with access to all weather roads (Cuttack). Figure 7.1: Simulated reductions inInfant Mortality from changes infour policy variables: Number of infant deaths prevented 8 9 -2 7- & -4 r a CT t: -6 .-c -8 YOchanae in variable Roportion of householdswith access to all weather roads Roportion of mothers with a birth spacing of 24 to 48 months 3tWealthindex Note: change i s variable is by +lo% intervals, reaching 100% (Le., doubling) at the right handend. 7.7 Perhapsthe most striking conclusion from this exercise i s that, on paper, the infant (child) mortality goal associated with the MDG 2015 targets is achievable for Orissa. One policy variable alone - post-primary schooling of girls - will go along way to achieving the child ~ ~ Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 41 Achieving TheMDGs in India's Poor States mortality goal (as well as achieving the two education MDGs). Combining this strategy with those interventions which enable girls' schooling to have its intended payoff vis-&vis lower child mortality, Le., improved nutrition, access to quality health services, information on the value of environmental hygiene and cooking practices, safe water, etc., should have a significant effect. Further combining the other policy variables from Table 7.4 - reducing poverty, in particular raising incomes among the poorest, improving the all-weather roadnetwork, and critical elements of the RCH program (birth spacing, exclusive breastfeeding, trained mid-wife/doctor at birth) - would further complement the schooling effect. Existing interventions what are they doing? - 7.8 Together, the Government of India and the Government of Orissa support a very rich policy agenda covering many of the priority areas identified in Table 7.3. In the past year alone a number of major new initiatives have been launched by GoI, including the Rural Health Mission, a secondReproductive and Child HealthProgram, the Rural Employment Guarantee Scheme, and a rural infrastructure program. These complement other centrally sponsored schemes in the area of nutrition (PDS and ICDS), education (SSA), and vector borne diseases (including malaria). At the state level, there is a whole host of programs including "missions" for rural drinking water and sanitation, a women's empowermenthural credit scheme (Mission Shakti), school scholarship program for ST/SC girls, an IMWneo-natal "mission", and the Orissa Health Sector Program. Table 7.5 presents a summary of the policies and programs for the priority interventions. The table also offers a summary situation analysis of the policies and programs. 7.9 National Rural Health Mission. One of the most important new initiatives of the Government of India is the National Rural Health Mission (NRHM).The program focuses on 18 states in which the public health indicators are poor. As such it includes Orissa. The goal of the program i s to support the reduction of infant and maternal mortality, further reduce fertility and achieve a major reduction in the prevalence of many diseases, through the provision of universal access to services in health (women and child, immunization), nutrition, water, and sanitation. NRHMwas launched inOrissa inJune 2005, and following the preparation of a state-wide plan, district level plans are underway. Major elements of the programincludea new community health worker -ASHA, various financial incentive schemes (for patients and the ASHA), the creation of village health and sanitation committees, strengthening of public health infrastructure and provision of mobile medical units for remote areas, convergence of various vertical health programs under one umbrella, close collaboration with nutrition programs, total sanitation campaign and rural water mission, and greater emphasis on monitoring outputs and outcomes. The programalso seeks to work closely with PanchayatRaj institutions - including a Rs. 10,000 fund into a joint account, ANM/Sarpanch, for minor repairs and maintenance at health sub- centers -as well as with NGOs and the private sector Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 42 Achieving TheMDGs in India's Poor States Table 7.5: Existing Interventions inthe Priority Areas: A Scorecard *-_^-______I_Lz_IyI--,,-,------ --p-?a--""pM- Intervention Core Elements Situation assessment Poverty reduction Income Growth promoting policies increasing Women's empowerment Wide-scale SHG movement, Mission Shakti, promising, but needs more support (viz. AP) Education Schooling of girls Rapid enrollment expansion, but issues of (esp. post-primary) retention and quality; scholarships for some SC/ST girls inplace Maternal and Child Health Services Ante-natal care RCHprogram, supplemented by NRHh4 State and district plans; convergence of services being promoted; ASHA Birth (home) Few home births attended by trained attendant: training of TBAs Emergency obstetric and Upgrading of CHCsPHCs underway; Newborn care RCHpromoting institutional deliveries but unreliable and poor quality service Post-natal care Few home visits Immunization Static coverage, needs new push Treatment of sick child Many barriers to accessing health services, physicaUfinancia1; quality o f care variable. Home Care of Mother and child Pregnandnursing woman Many good practices, but room for Baby/young child Improvement: IECBCC Hygiene practices BCC needsmuch more focus; (whole family) linkages with water, food, etc. need attention Nutrition Maternal nutrition (indirect) TPDS/AAY - poor targeting (direct) ICDS/SNP - limited impact Breastfeeding/weaning Insufficient IEC on diet, colostrum, early And exclusivebreastfeeding Micronutrient supplements Variable coverage Water Piped connections to homes Rural water "mission" More safe water IEC needed Sanitation Hygiene practices Total Sanitation Campaign Heavy infrastructure focus; more BCC needed Environment Indoor air pollution PDS kerosene/electrification not reaching the poor Malaria control Malaria program -needs big push on Preventative side, esp. bednets Roads All weather access New NREGSprogram offers potential - - " - ~ ~ - ~to improve access-inremoter parts ~ ~ ~ - e---"*- ~ ~ xI- - I~ - ~ - Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 43 Achieving TheMDGs in India's Poor States Building on the 2001 IMR Mission and launched in 2005, Navajyoti is an initiative of the Orissa State Government to strengthen community based care aimed at reducing neo-natal mortality. In recognition of the intra-state variation with mortality rates, the program focuses on 14 high infant mortality districts, but plans eventually to expand into all districts (under RCH2). Core elements of the home based care of mother and child program include: every delivery is attended by a skilled birth attendant (includingdais, suitably trained in hygiene, care of the new-born and mother, and identification of the need for emergency obstetric and newborn care), improved ANC and birth preparedness, home visits of the newborn, upgraded maternal and child services at Block PHC/CHC and referral hospitals, promotionof chemoprophylaxisand insecticidetreated bed-nets for protectionagainst malaria, and fixed monthly health and nutrition days (for ANM, Anganwadi and dais - and now including ASHA). The also aimsto link2 with the Mission Shaktiself-he -- ~ ~ -----__ s andPRIxo= ~ - --...-.- - ~ --^---- __ . ~ _ _ I I ._E I II-"I * 7.10 Overall, it could be concluded that Orissa is adequately provided for in terms of the essential policies and programs needed to bring about a further reduction in infant and child mortality. Moreover, some of the shortcomings identified by past reviews - such as inflexible centrally sponsored schemes and over-centralized planning, lack of attention to state and district heterogeneity - are being addressed in the new programs. But there are problems and major challenges. First,while there may be aperception infederal and state government circles that there is a full array of policies and programs to address the multi-sectoralnature of child survival, the reality for people living inthe remoter rural areas is quite different. Inparts of Orissa there is a near absence of services: public service providers and critical supplies are often absent (due to both a shortage of providers and weak governance of resources in the system), private providers are few due to limited market sizehahe, and there are relatively few non-governmental agencies. Heavy seasonal rains and poor infrastructure present additional barriers to regular service provision inrural Orissa. But it i s not just a problem of physical access and fickle service suppliers. The poor are often faced with significant costs (user charges) when accessing public services, especially health services. A study undertaken in Bolangir District recently found fee for service arrangements in place with Anganwadi workers (for ante-natal care), A N M s (attending births), and in government medical facilities (for many services). Moreover, patients reported not only facing considerable expenses in seeking healthcare from a government facility but were also exposed to a great deal of arbitrariness and ill-will at the time of seeking medical care. The challenge to expand and sustain effective coverage/out-reach of services to Orissa's interior districts and scattered populations - in particular the ST population - cannot be understated. Adauting programs to different situations and needs, and considering different delivery strategies - especially involving communities, local governments and NGOs - steps recently initiated by the Orissa Government - seem promising areas. Second, some service providers are carrying too many responsibilities and their impact on any one priority is limited. A good example would be the Anganwadi worker who i s the frontline service provider of many programs for women and children. Although such an arrangement helps promote convergence across programs, one person can only do so much. The advent of a second community based worker under the Rural Health Mission -theASHA-isagoodinitiativeandshouldhelpfreeuptimefortheAnganwadiworker to focus her core responsibilities. Line management of the two different functionaries - who report to two different government departments - will remain a challenge. Introduction of fixed health and nutrition days, at which service providers from different Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 44 Achieving TheMDGs in India's Poor States line departments converge in a village and provide integrated services is one initiative to overcome this problem. Another is the increasing involvement of PRI bodies in the oversight of local service delivery. 0 Third, building on recent efforts to respond to national and state level heterogeneity, further sharpening the focus on those districts and blocks where child mortality i s highest, and varying the "input package" of centrally sponsored schemes to reflect local conditions and the relative levels of neonatal, infant and child mortality is still needed. Preparing district (and block) plans, empowering communities and decentralizing resources, responsibilities and accountabilities to local government are all core elements of getting the right mix. 0 Fourth, too many programs are run as self contained "silos" and miss critical synergies: too often potential synergy i s lost due to complementary inputs occurring at different times and in different places. There are, for example, strong potential complementarities across sectors, such as those between health and nutrition, or between supply-side investments in service delivery and demand-side benefits from community mobilization. Findingways to achieve greater inter-sectoral integratiodconvergenceinorder to benefit from the strong positive interactions across sectors and across programs should be given higherpriority than launchingany more new programs. The example of ANKUR, ajoint initiative of the Orissa Government, and UNICEF is an interesting attempt to promote more multi-sectoralcollaboration. The final area where the existing array of government policies and programs might be considered lacking i s the strong emphasis given to supply side and hardware interventions at the expense of insufficient attention to demand side," the soft side of services", information and behavioral aspects. A quick survey of existing interventions in Orissa (Table 7.5) finds the need for more emphasis on IEC, BCC in a number of programs (especially water and sanitation), a greater focus of promotive/preventative healthcare of women and young children, and support to individuals, households and communities to demand and access services. Box 7.2: Joined up thinking, joined up action, The Case of ANKUR ANKUR is ajoint initiativeof the Government of Orissa and UNICEF, under implementation inthe District of Koraput (one of the poorest Kl3K districts). Ithas four inter-relatedgoals: Improve the quality o f education for tribal children and girls Improve nutrition and development o f children under 3 years of age Improve infant survival 0 Support implementation of state-wide reforms for rural drinking water and sanitation. At the heart of the program is the District Plan of Action for the young child, which identifies where action i s needed and by whom. There are no new activities, but rather a focus on establishing functional linkages between existing programs to maximize their impact for the young child. This involves joint planning, implementation, and assessment aimed at changing community and family level behavior, strengthening service delivery, and ensuring strong interface between community and service providers. District level coordination is replicated at block, Gram Panchayat and village levels. ANKUR was launched in2004 and is a four year program, covering all blocks ina phased manner. The bulk of activities involve support for micro-planning, capacity building and training of front-line functionaries, and frequent coordination meetings. In two pilot blocks additional support for implementation o f integrated actions in health, water and sanitation, education, child development and nutrition i s also envgised. " ` - ~ - ~ - w * - - - ~ - ~ - - ~ --~,--,-~~,--.--- ------___ ~ ~~ ~ Conclusionsand Policy Implications - TheRoadmap to Lower ChildMortality 45 Achieving TheMDGs in India's Poor States Putting it all together: What might a childmortality reducing program look like?37 7.11 Inso far as child mortality is linkedto multiple determinants, there is a compelling for a more holistic, or integrated approach designed to bringto bear all relevant factors in one place at one point in time. But a "one size fits all" approach will not work: it: what is needed for one situation may differ from what i s needed in another situation. As such, a range of interventions should be considered depending on local conditions and ideally subject to close monitoring and periodic evaluations in order to provide feedback on what is working. In addition, in order for local preferences to be demonstrated purchasing power to "pick and mix" from the various interventions should be place with individuals and communities wherever possible. This could be inthe form of vouchers or block grants, for example. This approach speaks to "thinking multi- sectorally, acting locally". 7.12 One should also be mindful of the danger that trying to address everything in one package becomes diffused and unwieldy, and impact i s minimal.Asking ourselves the following set of questions i s key: "of which sectors does it make the most sense to develop linkages and what benefits do these linkages produce; do these linkages add value over and above what these sectors are already doing on their own - and/or could be doing if better implemented; and which linkages are likely to be most cost-effective". Moreover, to be effective, multi-sectoral action requires the cooperation and coordination of a variety of actors and stakeholders. Many of these requirements run counter to public sector culture and practice in India, and as such present another major hurdle. This leads us to "thinking multi-sectorally, acting locally and selectively". 7.13 The Orissa state government has constituted a cross-departmental committee to bring together the various sectors which have a role to play in addressing child mortality, under the leadership of the Development Commissioner. This an important step forward in acknowledging the multi-sectoral nature of child mortality. Moving forward, Go0 will need to consider alternative mechanisms to deliver any new and additional interventions. Given the importance of focusing energy at the local level, and reaching out more effectively to poor households and communities, working in conjunction with Mission Shakti (the self-help group movement) PRIs and NGOs would seem central to the design. The Orissa Poverty Reduction Mission i s another promising organizational modelto deliver on a multi-sectoralprogram. 7.14 Building on the above, a new such program could be organized into two groups of activities: (i) cross-cutting support to enhance individual and household behavior and promote demand for services, and (ii) innovative service delivery strategies involving public and private sectors, and NGOs, together with mechanisms that support greater accountability. These are detailed below. I. Cross-cuttingSupporttoPromoteBehavioralChangeandEnhanceDemand a. ZEC/BCC - a substantial package of activities to increase knowledge, awareness and practices regarding behaviors that have a significant impact on child mortality. This might involve hygiene (hand washing with soap), water, sanitation, indoor air pollution (use of clean fuels, improved ventilation), malaria (use of bed nets), nutrition and feeding practices, early diagnosis of medical problems, and service provision (what, where and how). These messages would be designed for a variety of stakeholders, viz. households, community leaders, PRIs, SHGs, and government 37This potentialprogramdesignflows inpart from a World Bank TA mission to Orissa inFebruary2006. Conclusionsand Policy Implications - TheRoadmap to Lower ChildMortality 46 Achieving TheMDGs in India's Poor States functionaries. This would build on the new right to information bill and citizen charter, and aim to take information to women in their villages and homes through a variety of media. b. Addressing poverty and demand for critical child survival inputs. Many of the interventions identified as high priority involve a financial outlay on the part of households. Included here would be food (quantity and variety), cooking fuels other than wood or dung, soap for hand washing, insecticide treated nets, costs involved in sending (girl) children to post-primary school together with the transport and user charges incurred when using either public or private health services. The absence of resources to meet these costs leads women and their families to adopt strategies which have dire consequences for their own health and that of their children. Inthe long-run, povertyreducingeconomic growthshouldaddressthese problems, although the extent to which growth will penetrate the remote rural areas where the extreme poor live is an open question. But, in any event, shorter-term action i s called for and there are a number of instruments that can be used to provide financial support to poor households. Vouchers, stipends, scholarships are examples of instruments which provide poor people with additional purchasing power for key services. India's own JSY38i s an example of a scheme designed to ease the financial constraint which prevents pregnant women for giving birth in a hospital. Another instrument is the conditional cash transfer (CCT). This instrument has been successful inLatin America (especially Mexico) and i s gaining interest elsewhere in the world. A CCT could support two critical aspects of child survival. The cash transfer to eligible women would help address income poverty and the various ways in which it manifests itself into sub-optimal behavior, including the early return to wage employment be and the cessation of exclusive breast-feeding. By combining the cash transfer with rewired actions on the behavioral side, the program would support attendance at information, education, and training programs (such as promoting better hygiene and feeding practices, as detailed in a. above) as well as possibly full utilization of ante-natal, birthingand post-natal services, growth monitoring of babies. (This would require improvement in the provision of services, behaviors of service providers, and other complementary actions.) The program could be started as a pilot in a few of the poorest blocks, be accompanied by a robust effort to improve services, and include a monitoring and evaluation framework to allow for a critical review of its effectiveness. Over time it could expand to provide support to families with post- primary aged girls who struggle to afford schooling, given the very strong effect of secondary education on maternaland child health. (See Box 7.3 for further details). 11. InnovativeServiceDelivery Strategies -getting services to poor women c. Community Znvestment and Innovation Fund (CIIF). This could be managed by a consortium of SHGs and front line service providers (ANM, Anganwadi, ASHA, RWSS), and would provide a pool of flexible funding to support local area 38Janani Suraksha Yojana (JSY) i s an initiativeunder the National Rural Health Mission whereby a cash grant i s paidto all BPLwomen upon the delivery o f a live baby. An additional payment i s made to mothers who deliver inhealth facilities. ASHAs supporting women during their pregnancy and at the time of their delivery also receive a payment under the scheme. Conclusionsand Policy Implications - TheRoadmap to Lower Child Mortality 47 Achieving TheMDGs in India's Poor States innovation and initiative. The funds could be channeled through PRIs in the form of block grants. Some examples of activities that might be fundedthrough the CIIF include: More significant supplemental nutrition for pregnant and lactatingmother. Health risk fund, a combination of savings and subsidy, to meet costs of health care/health insurance for the poor, especially at time of delivery. Bulkpurchase of insecticide anti-mosquito nets for pregnant women and young children inhighrisk areas. Supplementary maintenance of AWC, health sub-center and provision of missingessential supplies; adequatewater and sanitation. Improved community water source and training of community in water quality testing and monitoring. Support towards cost of individual household latrines (IHL)for poorest of poor families (with bonus scheme for communities which achieve full elimination of open defecation). Scholarship scheme for ST girls where heavy drop-out from primary school is a prob1em . d. Strengthened Service Delivery (hardware and software) by line departments- selective investments ina number of areas such as: Support for front line service provision in key sectors (health, nutrition, hygiene, water and sanitation); rotation of workers, additional training, etc. Build up functional referral capacities for child and maternal health outcomes with investment, restructuring, incentives. Management contracts for running of health facilities and other public-private partnership innovations. Measures to address lack of motivation and instill drive for performance in government functionaries, measureand reward results. Strengthen supervision, training, and management especially at district and block levels. Case manager to support poor womedfamilies derive services from public health facilities, especially at the time of the infant's birth - services from which they are often excluded. Seasonality - additional out-reach care for babies born in the rainy season (July, August, September) and the cold months (November-January). e. Modification to some service provision packages, e.g., ICDS, to give greater priority to under-malnutrition in 0-36 month age children, including exclusive breast feeding and micronutrient supplementation; also women's nutrition status before, during pregnancy and in early months after birth; targeting RWSS to areas with greatest epidemiological need; pilot measures to address indoor air pollution reduction. f. Technical support and training to SHGs and PRIs for performance monitoring of service providers, promoting greater accountability through instruments such as scorecards; also to encourage greater participation of other institutions, e.g., NGOs, universities, in social audits. Conclusionsand Policy Implications - TheRoadmap to Lower ChildMortality 48 Achieving TheMDGs in India's Poor States g. More evidence-based Planning, Monitoring and Evaluation - support for data collection to underpin evidence-basedplanning, includingat the local, operational level as well as at district and state level; also to support monitoring and evaluation of inputs and activities vis-his child mortality outcome goals. Support would be provided for capacity buildingin (local level) data analysis and planning as well as data collection. 7.15 Many of these areas would require considerable background work to assess their feasibility and to undertake costing. In so far as there are similar activities underway in other states - for example, integrated child development pilot project in Madhya Pradesh, an information campaign for school management committees in Uttar Pradesh, a PRI block grant program in Karnataka, community-based development program in Andhra Pradesh- as well as useful pilots in Orissa (such as ANKUR) - a detailed review of these would enhance the knowledge base. Finally, there is growing experience from around the world with multi-sectoral programs, especially regarding nutrition and HTv/AIDs, which offer important insights into elements and strategies associatedwith success. Conclusionsand Policy Implications - The Roadmap to Lower Child Mortality 49 Achieving The MDGs in India's Poor States " _ _ - " - Box 7.3: A conditionalcashtransfer for Maternaland Conditional cash transfers (CCT) combine more traditional cash transfer programs with financial incentives for families to invest in human capital. Typically, cash transfers are disbursed to poor families conditional on the household engaging ina set of behaviors deemed important for humancapital formation. While most CCTs have focused on children's formal education, a few in Latin America have addressed malnutrition and high morbidity/mortality. Mexico's quite famous CCT program (Progresa -0portunidades) provides a cash transfer contingent on (i) family members using preventive health services, (ii) children aged 0-5 and lactating mothers attending growth monitoring clinics and receiving nutrition supplements as well as nutrition and hygiene education; and (iii) pregnant women attending prenatal care clinics (as well as other elements, including schooling). Studies of the Mexico program show significant improvement inthe health of both children and adults, and for purposes o f this paper, a reduction in child mortality (Gertler, 2004; Bardhan, 2004). For example, one municipality with full coverage o f PROGRESA-Oportunidades experienced a drop in its infant mortality rate to a level 11% lower than it would have had without the program. Other low income Latin American countries, such as Honduras and Nicaragua are also experimenting with CCTs for health and nutrition. In the case of Orissa, one could imagine a CCT designed to improve the nutrition and health status of pregnandlactating women and their young children. The cash transfer would help address (income) poverty and could support a number of actions associated with child mortality, viz. better maternal nutrition, costs of getting ANC, hiring a trained birth attendant and delivering in a medical facility (transport and user charges), exclusive breastfeeding for 6 months - postponing the return to wage labor, purchase of insecticide treated nets (against malaria), cleaner cooking fuel, and soap for hand washing, to name just some. On the side of behavioral change, the transfer could be conditional on some of the same elements as the Mexico program (listed above) - at least three ANC, TT (only from recognized service provider); birth attended by trained person; post-natal check-ups, growth monitoring of babies and young children - as well as attendance at information and education sessions addressing use of anti-malarial measures, clean cooking fuels, personal hygiene, exclusive breastfeeding and maternal and child nutrition. In order to promote the desired behavioralchanges, the CCT would need to be accompanied by lots of IEC and BCC. The program could be geographically targeted to the districtshlocks with the highest mortality rates, and further targeted to the poorer sections of society. Targeting criteria would need to be carefully considered. Inthose communities where the SHG movement is strong, this could be part of the targeting methodology (as it used inthe IKP project inAndhra Pradesh). Monitoring of actions and behaviors are an important part of the program design. Here there would need to be balance between promoting the desired home behaviors (such as hand washing and use o f insecticide treated bed nets ) and those actions which can be monitored by a thirdparty. The level of payment and the duration of payment would need careful consideration too, but a payment for at least one year (from minus 6 months of birth to age 6 months) - ideally longer - would seem sensible. Given the seasonal dimension of infant morbidity and mortality noted in this paper, a seasonal emphasis o f the program would be desirable. A crucial element of the success of the CCT would be significant improvement in the supply of services. This would be both out-reach IEC/BCC programs as well as in-situante-natal, maternal and child care programs. Some of the measures supported by the rural health mission (such as JSY, incentive system for ASHA) Eesent an e -^.' but mu --* -- llent startingpoint,""*^_I ----~-- ore would/.=-needto be done. _-._ I ._ 1 " *_ --.-"- - ~1~ lj__ I-* - " ~ ~ Conclusionsand Policy Implications - TheRoadmap to Lower ChildMortality 50 Achieving The MDGs in India's Poor States BIBLIOGRAPHY CARE India: Lessonsfrom the EarlyLearningPhase of INHP11.January 2004 :NewbornCareat the CommunityLevel De Haan, A. andA. Dubey. 2005. PovertyDisparitiesor the Developmentof Underdevelopment inOrissa.EPW.May-June2005. Deaton, A. andJean Dreze. 2002. "Poverty andInequalityinIndia:A Re-examination." Economic and Political Weekly,September 2002. Gertler, Paul. "Do conditionalcash transfers improvechild health?EvidencefromPorgresa's ControlRandomizedExperiment". InHealth, HealthCare andEconomicDevelopment, Vol. 94. No. 2, May 2004. Government of India: 2001. Census 2001. :2002. BPLCensus .2006. Formulationof District HealthAction Plans :ProcessManual.NationalRural Health Mission.Ministry of HealthandFamilyWelfare.June. Government of Orissa: HumanDevelopment Report 2004 :ANKUR -Linkagesfor the Young Child, May 2004. :2005. OrissaHealthSectorPlan2005-2010. :Navajyoti:A strategy to improvematernal andchildcare with focus onpreventionof neonatal mortality andmorbidity.April 2005. Guillot, M.andGupta, S. 2005. "Water Access, HouseholdSanitationand Child Survivalin India". Mimeo. Hughes, G.andDunleavy,M.2000. Why do babies andyoung childrendie in India?The role of the HouseholdEnvironment.Mimeo. InternationalInstitutefor populationStudies. (2001). Reproductiveand Child HealthProject. RapidHouseholdSurvey (Phase 1and2). 1998 and 1999. Conclusionsand Policy Implications - TheRoadmapto Lower ChildMortality 51 Achieving TheMDGs in India's Poor States Jalan, Jyotsna andMartinRavallion.2003. Does pipedwater improvechildhealth for poor families inrural India?Journal of Econometrics, 112: 153-173,February 2003. IFPRI.Final Report on the Impactof Progresaon Health. November 2000. Luby et al. 2005. "Effect to hand-washingon childhealth: a randomizedcontrolledtrial". The Lancet. Vol. 366, July 2005. Mackinnon,J. 2002.Assessing the Impact of Fiscal and Structural Reforms on Poverty in Orissa. Oxford PolicyManagement. MillenniumProject.2005. Who's got the power?Transforminghealthsystemsfor women and children.UNTask Forceon ChildHealthandMaternal Health. Mishra andRetherford.1997. NationalCommissiononPopulation, ReproductiveandChild HealthSurveys 2002 and2004. Pandey, A. andChoe, Luther,Sahu and Chand. 1998.Infant andChildMortality inIndia. NationalFamily Health Survey Subject Reports, Number 11,December 1998. RTIInternational.2005.EnvironmentalHealthImpacts ofWater Supply, SanitationandHygiene InterventionsinRuralOrissa, India.FinalStudy Protocol. The Lancet. 2003. Specialedition on child survival.Summer 2003. Van Der Klaauw,B.andWang, L.2004. "Child Mortality inRuralIndia". Mimeo. Van Dillen, Suzanne. 2006. Child HealthandMortality inWestern Orissa: A reportbasedon a longitudinalHousehold Survey inBolangirDistrict. Universityof Bonn.Mimeo. World Bank. 2001.Poverty Reduction Strategy Paper Sourcebook.2"*. Edition. .2004. India: Attaining the Millennium Development Goals in India. .2004. TheMillenniumDevelopment Goalsfor Health:Risingto the Challenges .2004. Reachingout to the Child: An integrated Approachto ChildDevelopment. .2004. Water, SanitationandHygiene:interventionsanddiarrhea.A systematic review andmeta-analysis.HNF' DiscussionPaper. August 2004. .2005. Health, NutritionandPopulationSector inOrissa. A policy note. World HealthOrganization.1997. Conclusions and Policy Implications - TheRoadmap to Lower Child Mortality 52 Achieving TheMDGs in India's Poor States ANNEX TABLES Annex Tables 53 Achieving The MDGs in India's Poor States -Districtsranked ate*District, RCHIandI1 I_ -" ~ l-______l-~--~ ~ - I ~~~ Infant Infant Neo-natal Child Mortality -lowesttohighest, by Child Mortality Mortality mortality Mortality fromRCH2 Rate, Rate, rate Rate, RCH2 RCH2 RCH2 1. Balasore 87 45 36 46 2. Sambalpur 67 40 37 49 3. Cuttack 46 45 33 49 4. Khurda 60 51 46 54 5. Bhadrak 51 44 29 56 6. Jagatsinghpur 85 50 31 56 7. Deogarh 68 42 35 57 8. Puri 51 56 38 58 9. Dhenkanal 53 53 46 60 10. Jharsuguda 81 59 46 62 11. Anugul 61 52 42 64 12. Mayurbhanj 96 53 38 65 13. Bargarh 75 54 36 68 14. Sonepur 81 61 47 68 15.Nayagarh 47 52 45 72 16.Boudh 90 70 63 72 17. Ganjam 93 71 46 74 18. Kendrapara 66 70 53 78 19.Jajpur 63 71 51 78 20. Nuapada 88 67 52 79 21. Sundargarh 84 64 55 79 22. Bolangir 78 76 58 79 23. Kandhamal 87 69 57 80 24. Kalahandi 92 69 50 80 25. Gajapati 111 58 31 81 26. Keonjhar 73 68 39 84 27. Nabarangpur 71 74 48 88 28. Koraput 93 69 37 93 29. Rayagada 101 88 49 118 30. Malkangiri 122 100 48 124 All districts 77.0 64.0 44.3 73.4 Annex Tables 54 Achieving TheMDGs in India's Poor States Marriagebefore 18 Proportionof years of age Birthsat home Districts mothersliterate(%) (%I (%I RCH I RCHI1 R C H I R C H I I R C H I RCH I1 Angul 35.9 63.4 40.0 83.6 62.1 Balasore 49.6 63.0 28.4 84.8 69.0 Bargarh 39.3 44.5 59.8 77.6 66.4 Bhadrak 47.7 57.7 17.7 80.7 62.9 Bolangir 22.8 40.1 57.7 87.1 61.1 Boudh 28.9 56.2 50.6 88.0 65.6 Cuttack 62.1 70.5 10.6 59.5 47.8 Deogarh 40.6 51.7 33.3 80.7 67.6 Dhenkanal 47.7 61.8 36.6 71.2 52.0 Gajapati 21.7 34.3 41.8 88.1 78.7 Ganjam 29.3 45.0 50.7 77.1 67.3 Jagatsinghpur 62.7 72.4 9.2 64.7 35.5 Jajpur 49.5 55.6 14.7 71.2 63.4 Jharsuguda 43.6 56.7 17.8 65.O 62.8 Kalahandi 16.7 35.0 59.4 89.6 69.4 Kandhamal 19.4 39.8 41.6 87.7 65.4 Kendrapara 61.5 67.8 15.8 78.5 57.7 Keonjhar 32.0 45.4 30.1 81.4 77.7 Khurda 61.6 71.5 23.4 44.7 44.3 Koraput 13.6 33.0 64.7 89.4 80.7 Malkangiri 10.6 13.3 56.0 92.6 89.7 Mayurbhanj 33.1 48.8 32.6 84.8 65.4 Nabarangpur 11.4 32.8 69.5 92.8 71.3 Nayagada 44.1 57.8 53.5 69.0 60.5 Nuapada 15.2 31.6 42.5 92.6 76.1 Puri 52.9 67.9 14.0 61.5 39.4 Rayagada 29.1 31.1 38.5 80.4 81.4 Sambalpur 40.3 52.4 29.5 70.8 53.8 Sonepur 31.4 48.1 39.2 86.2 68.1 Sundargarh 44.7 48.2 17.0 65.5 62.5 Annex Tables 55 Achieving TheMDGs in India's Poor States Table A2 (cont):Trends inBreastfeedingandImmunization, yI RCH1andI1 Childrenbreastfed within2 hoursof Children12-24months birth fully immunized Districts (%I (%I RCH 1 RCHI1 RCHI* RCH I1 Angul 7.6 29.2 68.9 56.7 Balasore 14.3 54.5 51.0 73.5 Bargarh 13.1 29.7 64.0 70.8 Bhadrak 8.2 40.9 54.7 54.0 Bolangir 11.3 40.4 72.2 65.1 Boudh 13.7 52.2 50.8 69.5 Cuttack 11.3 50.1 70.5 86.5 Deogarh 12.1 35.7 53.2 48.4 Dhenkanal 9.9 53.5 71.3 60.2 Gajapati 13.2 56.7 59.8 46.2 Ganjam 12.1 49.7 37.6 53.2 Jagatsinghpur 16.9 43.2 58.6 45.8 Jajpur 7.7 43.5 56.2 36.9 Jharsuguda 11.3 34.4 79.8 70.5 Kalahandi 18.8 44.6 54.9 49.9 Kandhamal 31.4 40.2 67.7 59.4 Kendrapara 12.7 55.5 53.9 58.1 Keonjhar 8.2 38.4 45.8 35.3 Khurda 8.9 42.1 61.2 59.2 Koraput 12.8 43.5 55.1 31.0 Malkangiri 9.5 49.7 50.0 39.9 Mayurbhanj 22.4 48.0 55.2 46.0 Nabarangpur 23.5 43.3 28.1 44.8 Nayagada 9.3 41.8 65.1 62.3 Nuapada 21.9 36.8 52.4 42.9 Puri 12.1 61.7 61.1 73.0 Rayagada 16.5 40.1 56.0 46.0 Sambalpur 20.8 27.2 76.6 72.4 Sonepur 9.9 49.3 66.7 60.4 Sundargarh 21.2 54.2 80.3 59.0 Total 13.9% 44.8% 57.8% 55.4% _I_--p- Annex Tables 56 Achieving TheMDGs in India's Poor States ---- Table A3: District-wise Child, Infant andNeo-natal mortality Women's Education and E t h n i c i t G C H I1Districts r a n k d c v e r t y -- --* --a- Districts Women's Ethnicity: ranked by Neonatal Infant Child education: % of household mortality Mortality mortality % with at population wealth index rate Rate rate least who is (wealthiest to primary scheduled poorest) schooling tribe Cuttack 32.5 44.8 49 52 4 Puri 38.3 55.3 58 46 1 Angul 42.3 57.4 64 43 13 Khurda 45.5 53.8 54 43 3 Balasore 36.2 44.6 46 47 12 Ganjam 45.5 71.1 74 26 5 Dhenkanal 45.7 52.8 60 43 10 Jagatsinghpur 30.7 52.3 56 53 1 Sambalpur 37.4 45.1 49 37 35 Nayagada 45.1 60.8 72 37 6 Jharsuguda 45.9 58.2 62 34 38 Sundargarh 54.9 66.8 79 35 49 Kendrapara 52.8 69.0 78 47 0 Jajpur 50.5 75.4 78 38 4 Bhadrak 28.6 40.5 56 39 3 Boudh 62.8 69.6 72 35 14 Bolangir 58.0 76.3 79 30 21 Gajapati 31.2 57.7 81 25 42 Rayagada 48.6 91.5 118 22 45 Koraput 37.4 67.9 93 25 46 Bargarh 36.2 59.6 68 26 24 Kandhamal 57.1 76.8 80 29 41 Deogarh 34.7 45.7 57 35 30 Kalahandi 50.0 68.6 80 25 27 Mayurbhanj 38.1 59.0 65 36 46 Sonepur 47.4 61.8 68 25 8 Nabarangpur 47.6 73.1 88 26 39 Nuapada 52.1 73.6 79 22 30 Keonjhar 38.9 66.8 84 31 40 Malkangiri 47.9 102.8 124 9 67 Annex Tables 57 Achieving TheMDGs in India's Poor States --...--p----."---p Table A4: Antenatal care at the District level *-------* ..- - -,..-",.d_^__l-l__ll -.*_ ---.- - _"-+** +**- Neo-natal % pregnant % pregnant % pregnant Districts ranked by Mortality women getting women getting women getting NNMR lowestto- Rate Tetanus toxoid one ANC three ANC highest Bhadrak 29 92 73 55 Jagatsinghpur 31 82 79 71 Gajapati 31 85 75 55 Cuttack 33 94 86 63 Debargarh 35 86 73 66 Bargarh 36 92 88 58 Balasore 36 86 68 65 Sambalpur 37 93 87 72 Koraput 37 79 72 45 Puri 38 89 84 68 Mayurbhanj 38 90 78 56 Kendujhar 39 78 58 27 Anugul 42 80 72 48 Nayagarh 45 81 72 51 Khordha 46 91 81 56 Ganjam 46 77 76 62 Dhenkanal 46 91 80 70 Jharsuguda 46 93 86 63 Sonapur 47 88 85 65 Nabarangpur 48 80 58 55 Malkangiri 48 69 54 41 Rayagada 49 86 72 55 Kalahandi 50 86 70 34 Jajapur 51 87 74 52 Nuapada 52 85 79 56 Kendrapara 53 93 80 65 Sundargarh 55 86 75 54 Kandhamal 57 85 69 51 Bolangir 58 92 88 61 Boudh 63 92 87 59 Annex Tables 58 Achieving TheMDGs in India's Poor States I----------.--- Table AS: M o r t a 9 r a t,"..--s andklaces of d e l i v e x -e __-~. -_ ~-,-,A- -_rI_-~l---_-l_n-./ r Districts rankedby Neonatal % deliveries in % deliveries at % home NNMR lowestto - mortality rate medicalfacility home deliveries by highest trained medical personnel (including Dai) Bhadrak 29 27 64 27 Jagatsinghpur 31 36 54 Gajapati 31 79 18 Cuttack 33 48 27 Debargarh 35 68 18 Bargarh 36 28 68 34 Balasore 36 69 24 Sambalpur 37 34 56 29 Koraput 37 81 7 Puri 38 39 34 Mayurbhanj 38 24 67 30 Kendujhar 39 79 19 Anugul 42 29 62 12 Nayagarh 45 61 14 Khordha 46 46 47 19 Ganjam 46 28 66 24 Dhenkanal 46 52 16 Jharsuguda 46 63 44 Sonapur 47 25 69 22 Nabarangpur 48 71 18 Malkangiri 48 9 89 8 Rayagada 49 15 80 27 Kalahandi 50 69 16 Jajapur 51 28 64 28 Nuapada 52 17 75 16 Kendrapara 53 58 45 Sundargarh 55 23 63 25 Kandhamal 57 19 67 22 Bolangir 58 61 32 Boudh 63 66 22 All districts 44.3 24 65 ~ ~ - ~ , - 23 --p---wpp-- Annex Tables 59 Achieving TheMDGs in India's Poor States Table A6: ChildMortality, Breastfeeding, Medical Knowledgeand Immunization --- at the District level ~ - .--- ~ _IIp % of children % of children Districts ranked by At least6 months Some knowledge aged 12-36 aged 12-24 ChildMortality -lowest exclusive of appropriate monthsfully months to highest breastfeeding diarrhea immunized immunized (%o) treatment againstmeasles 1. Balasore 19 86 74 79 2. Sambalpur 21 80 72 78 3. Cuttack 6 87 87 90 4. Khordha 5 100 59 65 5. Bhadrak 15 100 54 61 6. Jagatsinghpur 9 91 46 50 7. Debargarh 8 77 4s 66 8. Puri 19 89 73 82 9. Dhenkanal 8 84 60 76 10.Jharsuguda 17 67 71 78 11. Anugul 27 100 57 70 12. Mayurbhanj 9 65 46 74 13. Bargarh 17 100 71 79 14.Sonepur 23 100 60 70 15.Nayagarh 7 75 62 68 16.Boudh 13 85 70 82 17. Ganjam 14 100 53 66 18. Kendrapara 13 92 58 64 19.Jajapur 25 100 37 53 20. Nuapada 20 90 43 56 21. Sundargarh 35 100 59 70 22. Bolangir 15 81 65 75 23. Kandhamal 17 100 59 73 24. Kalahandi 14 82 50 71 25. Gajapati 9 44 46 66 26. Kendujhar 9 100 35 58 27. Nabarangpur 21 67 45 65 28. Koraput 24 44 31 51 29. Rayagada 12 51 46 64 30. Malkangiri 22 61 40 54 All districts 15.6 82.8 55.4 67.9 --------"- -....---- --- ----- --M-*I__x__l.- Annex Tables 60 Achieving TheMDGs in India's Poor States Table A7: District-wise Health Facilities: Districts ranked by Wealth -- -------- ----"-**wl- Percentage Percentage E E t 2 kPercentage h ? L Percentage Percentage of of of of of households households households households households having having having having having access to accessto access to accessto accessto ICDS health PHC CHC/RH private Center in sub-center clinic in the village inthe the village Malkangiri 86.7 13.7 3.3 0.0 2.4 Keonjhar 89.6 62.3 26.3 55.6 31.1 Nuapada 92.0 34.5 3.3 0.0 1.5 Nabarangpur 100.0 36.7 16.2 3.3 3.3 Sonepur 86.2 16.2 4.3 0.0 3.4 Mayurbhanj 91.0 33.5 3.3 5.7 9.4 Kalahandi 81.1 9.5 13.2 0.0 7.6 Deogarh 85.9 15.9 0.0 38.3 6.8 Kandhamal 80.7 30.8 14.6 12.5 14.1 Bargarh 86.4 36.7 24.5 13.7 23.1 Koraput 87.1 28.8 25.8 3.7 13.7 Rayagada 64.2 45.0 8.6 6.9 1.2 Gajapati 73.3 32.3 25.7 1.6 16.7 Bolangir 90.4 21.9 3.0 3.4 17.7 Boudh 92.1 47.5 0.0 2.4 5.1 Bhadrak 84.9 25.2 17.7 1.5 3.7 Jajpur 84.6 33.9 6.3 29.0 2.9 Kendrapara 92.2 22.1 8.8 0.0 12.5 Sundargarh 78.8 30.1 14.7 10.0 5.6 Jharsuguda 87.7 26.3 18.9 11.7 8.4 Nayagada 37.4 11.4 8.0 14.4 13.0 Sambalpur 83.9 30.5 18.9 17.6 21.8 Jagatsinghpur 88.5 26.4 10.5 0.0 16.9 Dhenkanal 85.1 29.5 10.5 3.9 11.8 Ganjam 84.7 30.6 12.6 4.2 19.9 Balasore 92.3 21.2 19.2 10.6 3.8 Khurda 89.8 21.9 0.0 27.0 0.0 Angul 90.4 55.1 34.9 60.4 27.9 Puri 90.1 16.8 14.8 0.0 0.0 Cuttack 100.0 11.0 2.5 0.0 5.4 Annex Tables 61 Achieving TheMDGs in India's Poor States ~ ~ - - ~ - ~ ~ - - " ~ ~ - - ~ . ~ --*-.- TableA8: Determinantsof Under FiveM o r t a l i i ~ I -I--- -I- -41^~1--~--1^---. 111^L^--..""..I -- . Rural & Urban RuralOnly FixedEffects Fixed Effects Rural 1.231691 ** 1.229558 ** Girl 0.905981 0.898685 * 0.864608 ** 0.85678 ** Birthorder 2-3 0.877404 0.885801 0.896669 0.902523 Birthorder > 3 0.817931 0.840953 0.787169 0.814531 Birthspacing24-47 0.69023 *** 0.691463 *** 0.643824 *** 0.647429 *** Birthspacing>47 0.76069 *** 0.762547 *** 0.707044 *** 0.709229 *** sc 1.298516 ** 1.304395 ** 1.39494 ** 1.329594 ** ST 1.280964 ** 1.326351 ** 1.266834 1.277518 * Other Backward 1.145221 1.127642 1.202275 1.129843 Age of mother at birth 20-29 0.836459 * 0.831691 * 0.835279 * 0.828787 * Age of mother at birth> 29 0.951299 0.934791 0.921378 0.894201 Age started living with husband 15-19 0.976459 0.974974 0.947998 0.932511 Age started living with husband> 19 0.98411 0.993187 0.95283 0.955273 Mother'seducation 1-5 0.798927 ** 0.774013 ** 0.805807 * 0.778248 ** Mother's education>5 0.506861 *** 0.494211 *** 0.522689 *** 0.507933 *** Semi-Pucca 1.252815 ** 1.294047 ** 1.219409 1.273186 ** Pucca 1.178961 1.225991 1.183951 1.2612 Tap inside 1.126107 1.121768 1.388261 1.327789 Tap public 0.723739 ** 0.702909 ** 0.830767 0.792312 Handpump 0.804734 0.789234 ** 0.822242 0.824757 Uncoveredwell 0.785968 ** 0.76252 ** 0.768959 * 0.752479 *** Own flush 0.958237 0.958581 1.041495 1.057868 Shared 1.433462 1.442487 1.620582 1.67734 Public Community 2.759855 2.725091 2.486833 2.65888 Light from electricity 1.386449 *** 1.37574 *** 1.457535 ** 1.480984 *** CookingLPG 1.036977 1.094795 1.554573 1.619706 Wealth index 0.697228 *** 0.677146 *** 0.63324 *** 0.601136 *** ICDS 1.032194 1.027216 Sub-center 0.863066 * 0.906232 PHC 0.949831 0.967435 CHC/RH 1.196672 1.18395 Gov. dispensary 0.849334 0.794674 ** Gov. hospital 1.122349 1.097421 Privateclinic 1.254078 * 1.232999 Privatehospital 1.096439 1.082474 ISM 1.065601 1.066964 All weather road 0.896338 0.888634 Obs 15413 15413 11621 11621 R-Squared 0.03 0.03 LR Chi --"...*-- 211 254 139 -- 172 ~ ~ ~ I I ^ _ _ ^ - I _ _ I I _ I I _ p ~ ~ - - ~ - ~ ~ - I-- I I ~ ~ ~ __I. * . ~_ Annex Tables 62 Achieving TheMDGs in India's Poor States - ~ ~ - -Table -- A9: Determinantsof InfantMortaliQ -- -P ,I -----I -"_"I1'-~I_-x-~_l___a__"--i~.y --~-----"---I -s_l_ L Rural8z Urban RuralOnly FixedEffects FixedEffects Rural 1.26115 ** 1.26893 ** Girl 0.881514 * 0.875261 ** @ti28929 ** 0.822901 *** Birthorder 2-3 0.940122 0.945779 0.961343 0.964565 Birthorder > 3 0.957477 0.973582 0.896 0.922782 Birthspacing24-47 0.669465 *** 0.672059 *** 0.619733 *** 0.624455 *** Birthspacing>47 0.727465 *** 0.727912 *** 0.65249*** 0.653235 *** sc 1.322585 ** 1.302557 ** 1.423981 ** 1.318025 * ST 1.259998 * 1.267212 * 1.293996 1.246092 Other Backward 1.2113 1.174694 1.298565 1.188049 Age of mother at birth 20-29 0.789083 ** 0.788365 ** 0.801111 *** 0.797906 ** Age of mother at birth > 29 0.933051 0.921796 0.960939 0.937395 Age started living with husband 15-19 1.032045 1.036328 1.030954 1.014751 Age started living with husband> 19 1.058196 1.069312 1.027725 1.027812 Mother's education 1-5 0.843123 0.819795 0.849942 0.826215 Mother's education>5 0.543625 0.529231 **** 0.545479 *** 0.529707 **** Semi-Pucca 1.2078 **** 1.244944 ** 1.14838 1.198948 Pucca 1.15469 1.202793 1.116459 1.195206 Tap inside 1.310229 1.318054 1.638755 1.552788 Tap public 0.795832 0.788011 0.904616 0.86845 Handpump 0.831045 0.826011 0.870438 0.873041 Uncoveredwell 0.835618 0.823012 0.824217 0.8 1813 Own flush 0.969237 0.968003 1.009073 1.031695 Shared 1.631748 1.648223 1.841794 1.898319 Public Community 3.021716 *** 3.021007 *** 2.688818 2.91946 Light from electricity 1.448197 *** 1.449635 *** 1.517667 *** 1.54799 **** Cooking LPG 1.139203 1.191499 1.695334 1.742716 Wealth index 0.667363 *** 0.649151 *** 0.628622 ** 0.60161 **** ICDS 1.022717 1.039764 Sub-center 0.926267 0.963662 PHC 0.95 1344 0.945491 CHC/RH 1.135686 1.137469 Gov. dispensary 0.821223 0.788547 ** Gov. hospital 1.236229 1.196563 Privateclinic 1.297217 * 1.281359 * Privatehospital 0.990345 0.979623 ISM 1.150343 1.126044 All weather road 0.914531 0.903327 Obs 15413 15413 11621 11621 R-Squared 0.03 0.02 LR Chi 181 --207 133 149 ---- _I ---------------* Annex Tables 63 Achieving The MDGs in India's Poor States Rural & Urban RuralOnly FixedEffects Rural 1.244299 1.2375 Girl 0.8316465 *** * FixedEffects 0.8284358 ** 0.7882731 *** 0.7839177*** Birthorder 2-3 0.8547438 0.8541115 0.8398245 0.8388396 Birthorder > 3 0.7203833 0.6923023 0.8291939 0.8069256 Birthspacing 24-47 0.658436 *** 0.6611983 0.6138048 *** 0.61912*** Birthspacing >47 0.6485577 *** *** 0.6474978 *** 0.5514253 *** 0.5553061*** sc 1.215318 1.209177 1.405899 * 1.311014 ST 1.115114 1.093178 1.175253 1.11441 Other Backward 0.9940293 0.9892796 1.104624 1.036351 Age of mother at birth20-29 0.8837218 0.8895782 0.9567957 0.9526084 Age of mother at birth > 29 1.141137 1.14836 1.277258 1.26075 Age startedliving with husband 15-19 1.Om41 1.031748 1.O 10227 0.992539 Age startedliving with husband> 19 1.035887 1.044062 1.004591 0.9869102 Mother's education 1-5 0.9056137 0.9062733 0.8638475 0.8591421 Mother's education>5 0.5826088 *** 0.5808786 *** 0.5600035 *** 0.5508344*** Semi-Pucca 1.OS6386 1.057825 0.9872321 1.00129 Pucca 0.9320914 0.9447824 0.9150839 0.9280743 Tap inside ' 1.54196 * 1.554392 * 1.766676 1.676467 Tap public 0.8864308 0.8918327 0.8855537 0.8411634 Handpump 0.8592659 0.8619408 0.9180189 0.9245441 Uncovered well 0.863812 0.902343 0.8331277 0.8715701 Own flush 1.079158 1.065352 1.440916 1.397649 Shared 0.8625252 0.9049891 0.5857609 0.6267976 Public Community 1.747703 1.847647 2.186049 2.556371 Light from electricity 1.384473 ** 1.37987 ** 1.437875 * 1.463763** Cooking LPG 0.9942741 1.01087 1.721408 1.757698 Wealth index 0.751 * 0.748912 * 0.7041879 0.6942337* ICDS 0.9710889 0.988428 Sub-center 0.9337616 0.9441103 PHC 0.9523993 0.9424679 CHC/RH 1.654299 *** 1.627417*** Gov. dispensary 0.7322567 ** 0.7220281** Gov. hospital 1.173042 1.1456 Private clinic 1.462458 ** 1.488728** Private hospital 0.8171256 0.7902655 ISM 1.050389 0.992321 All weather road 0.9252402 0.9327677 Obs 15413 15413 11621 11621 -LikelihoodRatio ---- R-Squared 0.02 0.02 109 -*-------" 113 I05 ---.-*-,---- 108 Annex Tables 64 Achieving The MDGs in India's Poor States I* - *--*---- Table All: Determinants of Women ReceivinLThree ANC c h e c_"k - - s -+-----.------ -----. ----_"-.l_ - ~ ~ -~- . z I I Rural& Urban RuralOnly Fixed Effects Rural 0.5539365 *** 0.5842903 *** Fixed Effects Birthorder 2-3 0.8777134 ** 0.8613205 ** 0.8342488 ** 0.8291355 ** Birthorder > 3 1.305582 1.241719 1.269873 1.246132 Birthspacing 24-47 0.8997561 * 0.9282051 0.9435802 0.9745538 Birthspacing >47 0.9193793 0.9253456 0.9478628 0.9663505 sc 1.137528 1.132458 1.033264 1.052031 ST 0.818824 ** 0.8522782 * 0.7908 188 ** 0.8458669 * Other Backward 1.O13546 1.048987 1.044887 1.096609 Age of mother at birth20-29 0.956473 0.950756 0.9838013 0.971083 Age of mother at birth > 29 0.7865246 ** 0.813628 * 0.8656846 0.9006995 Age started living with husband 15-19 1.083736 1.122136 1.116913 1.158525 Age started living with husband > 19 1.316101 ** 1.363053 *** 1.390866 ** 1.42863 *** Mother's education 1-5 1.32243 *** 1.316126 *** 1.372853 *** 1.357028 *** Mother's education>5 1.953944 *** 1.879868 *** 1.865505 *** 1.800402 *** Semi-Pucca 0.9265546 0.91977 1.031158 0.9997685 Pucca 0.9550965 0.9167023 1.004548 0.9355014 Tap inside 0.8008558 0.8233612 0.4443528 *** 0.4456377 *** Tap public 1.008856 1.083071 1.006317 1.095876 Handpump 1.009065 1.O 12785 1.101648 1.109552 Uncoveredwell 0.9201805 0.8818275 0.9873668 0.9631005 Own flush 0.8714917 0.9129845 0.7479093 0.7891337 Shared 0.9531882 0.8960151 0.8685403 0.8131566 Public Community 0.3998108 * 0.4538436 0.5389517 0.6566569 Light from electricity 1.032911 1.085717 0.8785058 0.9559993 Cooking LPG 0.9427036 0.9863077 0.8005186 0.8625824 Wealth index 1.589807 *** 1.528527 *** 1.747751 *** 1.64698 *** ICDS 1.283522 *** 1.273618 *** Sub-center 0.8667238 ** 0.8491296 ** PHC 0.9785382 0.992245 C H C N 1.253911 ** 1.089287 Gov. dispensary 0.9816365 1.032102 Gov. hospital 1.338132 1.42017 * Private clinic 1.220022 * 1.350682 *** Private hospital 0.8805915 0.7598123 ** ISM 0.9065839 0.9360434 All weather road 1.162469 ** 1.102902 * Obs 8180 8180 5753 5753 - R-Squared 0.11 0.07 LikelihoodRatio -1165 1181 ~ - - - . - - ~ - ~ - - - - ~ - " l " _ _ 464 532 - ~1.1 Annex Tables 65 Achieving The MDGs in India's Poor States ---- ".Table "-'..&--WI...M-uI A12: Determinants-~'---~.-~- --_.ofHomeDeliverx - --.-- - I- I_ I _-I __I ** -.,, ~---*.- "L f- xx- % Rural & Urban RuralOnly FixedEffects FixedEffects Rural 2.132039 *** 2.036278 *** Birthorder 2-3 1.548282 *** 1.568897 *** 1.670121 *** 1.689847 *** Birthorder > 3 1.479045 1.611086 ** 1.188842 1.277778 Birthspacing 24-47 1.205952 **** 1.194879 *** 1.118257 1.116732 Birthspacing >47 1.411812 *** 1.436272 *** 1.472093 *** 1.520653 *** SC 1.024486 1.063559 0.942322 1.011325 ST 1.624606 *** 1.860426 *** 1.508443 *** 1.812483 *** Other Backward 1.015819 1.061562 1.003284 1.07452 Age of mother at birth 20-29 0.984904 0.936151 0.978214 0.919208 Age of mother at birth > 29 0.908078 0.843395 0.966606 0.882945 Age started living with husband 15-19 0.8592 0.813225 ** 0.705759 *** 0.647196 *** Age started living with husband > 19 0.609719 *** 0.539628 *** 0.485689 *** 0.414676 *** Mother's education 1-5 0.805265 *** 0.743966 *** 0.893345 0.805393 ** Mother's education>5 0.491718 *** 0.479373 *** 0.472974 *** 0.450225 *** Semi-Pucca 1.056117 1.027092 1.076126 1.00812 Pucca 1.217071 ** 1.08533 1.207521 1.040077 Tap inside 1.I52948 1.064743 1.623833 * 1.369512 Tap public 0.804652 * 0.767125 ** 0.902962 0.838148 Handpump 0.836593 0.772136 ** 0.864469 0.830968 Uncoveredwell 0.870551 0.757027 ** 0.821714 0.723316 ** Own flush 0.981009 1.084612 1.276105 1.072654 Shared 0.854522 0.867798 0.835555 0.762509 Public Community 0.866723 0.907017 1.640389 1.462606 Light from electricity 0.99948 1 1.028475 1.0453 1.062263 Cooking LPG 0.75 1104** 0.8166 0.859454 0.978971 Wealth index 0.493992 *** 0.49286 **** 0.485618 *** 0.506286 *** ICDS 0.996889 1.048762 Sub-center 1.001927 1.04209 PHC 0.89164 0.920285 CHCRH 1.199059 * 1.073443 Gov. dispensary 1.051193 1.226015 ** Gov. hospital 0.40431 *** 0.445169 *** Private clinic 1.148062 1.06554 Private hospital 1.332552 ** 1.204349 ISM 1.169147 1.143568 All weather road 0.755018 **** 0.766764 *** Obs 12349 12349 9461 9461 R-Squared 0.24 0.16 LikelihoodRatio 3109 3580 1038 1549 Annex Tables 66 Achieving TheMDGs in India's Poor States Annex 2: Longitudinal Household Survey, Bolangir The Original Panel Survey The focus of the original survey was on individual and household activity and income portfolios. It was designed with the aim of generating a panel, with data collected over several periods. Three rounds of data collection have been conducted, starting in October 2001. For the first round (kharif 2000, rabi 2001, kharif 2001), the actual field work was carried out between October 2001 and February 2002; for the second round (kharif2002, rabi 2003, kharif2003), the field work was carried out between January and June 2004; for the third round (rabi 2004, kharif 2004), the field work was carried out between January and June 2005. Immediately following the completion of this last round, the special round - undertaken for this particular report - was carried out between August and November 2005. After a two-week period of training in the field, the lead researcher accompanied and supervised the original team throughout the first round of data collection in 2001/2002. Dr.B. K. Behera, the principal assistant and local project coordinator, independently organized the canvassing of the second and third rounds, and the special survey. During this time, the author made two more trips to the field, and, when not in Orissa, was in almost daily email contact with Dr. Behera. The fieldwork teams consisted of three to five young anthropologists, all from Sambalpur University, most of them with fresh masters' degrees. The members of the teams varied slightly from one round to the next. Since day one, however, the teams have been led by Dr. Behera. The teams contributed to the project extensive local knowledge and substantial anthropological competence. The mixture of male and female investigators proved to be advantageous and improved access to both sexes among our respondents. As befits the topic, all the field investigators for the special survey were female, Dr. Behera playing the role of coordinator on the spot. All interviews, whether conducted in connection with the panel or the special survey, took about two hours, and all villages (and most households) were repeatedly visited inorder to clarify and crosscheck information. Since the panel survey forms the 'backdrop' of the special survey, a few remarks about the nature of the data are inorder. Consumption and expenditure data cannot be reliably collected usinglong recall periods; but extensive experience with household surveys inIndia has taught us that people have a rather accurate idea of what they cultivated last year and how much they harvested, whether they worked in agriculture or on construction sites, or whether they participated in employment generation schemes, for how long they did all this and what they earned on average. Income calculations thus include earnings from wage labor and the various forms of self-employment reported by our respondents (own cultivation, collection of forest products, basket weaving, etc.). All inputcosts were covered, except the opportunity costs of own labor, land, draught animals and other fixed assets. Despite certain limitations, such 'small panels', if carefully designed and canvassed, can yield important insightsinto household behavior over time (seasonal and annual) and thus complement large-scale (non-panel) household surveys. As inthe special survey, data collection at the household level was supplementedinall rounds by a series of qualitative (topical) interviews with respondents from inside and outside the study villages. This source of information has been, and still is, extremely helpful in making sense of the quantitative data. 39This research was carried out by Susannevan Dillen, University of Bonn,Department of Geography. Annex Tables 67 Achieving TheMDGs in India's Poor States The Special Roundfor this Report@ The longitudinal survey i s based on an initial sample of 240 households which have been canvassed since 2001. The sample was drawn as follows: five contiguous Blocks were selected (Kesinga, Titlagarh, Saintala, Muribahal and Bongomunda) and a random sample of six villages was drawn from each Block, employing a stratification aimed at ensuring a fairly even spatial distribution. Six clusters of villages were defined in each block and one village drawn at random from each cluster. Eight households were selectedineach village, yielding 240 households in all. Again, only spatial stratification was employed, households being selected at random from different parts of the village settlement area. For the purposes of the special round, one Block (Kesinga) was dropped, leaving a sample of 192 householdsinall.41 The survey consists of four parts. In Part 1 all members of the sample households and their characteristics were recorded. In addition, we compiled a list of all the pregnancies of all women in the household, including each child's date of birth and, if the child died, the date and immediate cause of its death. Part 2 of the survey deals with household practices. It investigates attitudes and actual behavior in the realms of hygiene, nutrition, medical treatment, mobility and other issues that have a potentially important impact on maternal and child health and mortality. InPart 3, all pregnancy and child 'histories' are tracked individually. This includes information about the households circumstances and the existing health infrastructure at the time of pregnancy, childbirth and early infancy, supplemented by detailed information about when and why (or why not) medical assistancewas actually sought at the various stages of this period. The child's history of vaccination, diseases and medical treatments was also recorded in detail and, to close this section, mothers were asked to explain whether or not they were satisfied with their child's subsequent development. Part 4 deals exclusively with child mortality, each case being tracked separately. It seeks to capture the run-upto death and so to identify both immediate and long-term causes. Mothers were also asked whether they thought that such deaths could have been avoided, and if so, how. Turning to the interviewing itself, our respondents were all female, and some of the topics covered by the questionnaire were sensitive. Yet, women seemed to report their experiences freely and not to mind the considerable length of the interviews (2-3 hours, depending on the number of children). Here, three factors seemed to be important: first, they have been familiar with our team of investigators since 2001; second, all but one member of the team were female; third, our respondents seemed to have a strong interest in the matter at hand and many were infact eager to talk. One focus group interview was conducted in all but two villages, yielding a total of 22 such interviews. These were generally carried out on 'neutral ground' - that is, a place in the village accessible to members of all communities - and lasted for about two hours. Six to twelve women from different communities and age groups participated, more often than not joined by their curious men and children. The interview atmosphere was generally friendly and relaxed, which i s at least partly attributable to the fact that our survey team has become a familiar sight to most villagers, and the women seemed to speak freely even in public. The men, for their part, 40This round was carried out at the specific request of the authors of this paper and under contract to the World Bank. 41For further details of the original household survey's aims and design, see van Dillen (2005), 'Income and its Variability in a Drought-prone Region: Seasonality, Location and Household Characteristics', mimeo, University of Bonn, Department of Geography. Annex Tables 68 Achieving TheMDGs in India's Poor States seemed to understand that we were particularly interested in the women's views and made few attempts to take over the discussion. When they did so, they were politely put intheir place. Their contributions were also valuable, however; for maternal and child health issues ultimately concern the entire household. The interviews were concerned with four broad themes: (i) pregnancy and child birth; (ii) treatment of newborns (0-2 months); (iii) mothers' activities and breast-feeding practices during the first year; and (iv) experiences with maternal health and child care facilities. ---- A profile of Bolangirfrom the RCHdata Not one of the poorest districts (middling), with average share of ST population (21%). Slightly below average female literacy, primary school completion Mother education has increased, at 40% literacy, lower than state average o f 50% but not among the lowest NNMfell from 65/1000to 58/1000 inthe period 1998/99to 2002/04, butstill amongthehighest of all districts Little change inIMRduring this period, 76/1000, child mortality i s 79/1000 (one of the ten poorest performing districts) - but note bulk of mortality is inneo-natalperiod Increase inage of marriage between two RCH surveys Deliveries at home at 61%, lower than state-wide average o f 65%, and showed major decline from first RCH survey (87%) Also good trends with TT and ANC visits (above state-wide average) Incidenceof early breastfeeding also risen; duration of breastfeeding about average Immunization rate, 65% quite good by state average (55%) but has fallen since RCHl Annex Tables 69