38705 Africa Region Working Paper Series No. 99 IMPROVING NUTRITION IN ETHIOPIA ­ A MULTI-SECTORAL CHALLENGE Jesper J. Kühl1 July 2006 1Study conducted by Jesper Kuhl (consultant) under supervision of a World Bank team consisting of Luc Christiaensen (Senior Economist), Harold Alderman (Lead Human Development Economist), and Meera Shekar (Senior Nutrition Specialist) and Iqbal Kabir (UNICEF). The findings, interpretations, and conclusions expressed are entirely those of the author, and they do not necessarily represent the view of the World Bank or UNICEF, its Executive Directors, or the countries they represent. IMPROVING NUTRITION IN ETHIOPIA ­ A MULTI-SECTORAL CHALLENGE Africa Region Working Paper Series No. 99 July 2006 Abstract Child malnutrition follows from a host of factors, including food insecurity, disease, limited maternal education and poor nutritional knowledge and practices. Using the baseline survey for the evaluation of the Child Growth Promotion Component (CGPC), this paper describes malnutrition outcomes, determinants of malnutrition at the individual, household and community level for 5700 children in southern Ethiopia, as well as program indicators for the CGPC. Malnutrition rates are in general similar to findings from national surveys, and expected signs of causation are found with respect to gender and illnesses. The survey shows varying quality of the caregivers' knowledge and practices on child nutrition and health, and only 58% of the caregivers correctly assess their child's true nutritional status. The surveyed households have a low resource base, with a high prevalence of shocks. The communities have very low levels of basic health, transport and communication services, and child-related relief programs are only available for a minority of the households. Even though the health personnel of the CGPC shows better knowledge and practices on child malnutrition than the surveyed households in general, two-third of them think that their training for the program had not been sufficient for their job. The Africa Region Working Paper Series expedites dissemination of applied research and policy studies with potential for improving economic performance and social conditions in Sub-Saharan Africa. The series publishes papers at preliminary stages to stimulate timely discussions within the Region and among client countries, donors, and the policy research community. The editorial board for the series consists of representatives from professional families appointed by the Region's Sector Directors. For additional information, please contact Momar Gueye, (82220), Email: mgueye@worldbank.org or visit the Web Site: http://www.worldbank.org/afr/wps/index.htm. The findings, interpretations, and conclusions in this paper are those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors, or the countries that they represent and should not be attributed to them. Authors'Affiliation and Sponsorship Luc Christiaensen, Senior Economist, Poverty Reduction Economy Management Team, The World Bank lchristiaensen@worldbank.org Jesper Kuhl Consultant, jesper_kuehl@yahoo.dk ACKNOWLEDGEMENT This study is the result of a successful partnership between staff from different institutions and their support is gratefully acknowledged. The core study team consisted of Jesper Kuhl, Luc Christiaensen, Harold Alderman, Meera Shekar (World Bank) and Iqbal Kabir (UNICEF). The team further benefited tremendously from the guidance and support provided by the steering committee including representatives of the World Bank, UNICEF, the Food Security Bureau of the Government of Ethiopia, the Department for International Development, Save the Children ­ UK, and Development Cooperation Ireland. The team would also like to thank Jean Delion and Laketch Mikael Imru (overall and local World Bank Task Team Leaders for the Food Security Project respectively) for their support of this initiative. The helpful cooperation and assistance of the Federal Food Security Bureau of the Government of Ethiopia, as well as of the regional food security and health authorities in The Southern Nations, Nationalities and People's Region throughout the study is especially acknowledged. The survey has been conducted by JHA Health & Nutrition Consultancy under supervision of Jemal Haidar and Girma Akalu. Finally, financing for both the survey and study have been gratefully provided by UNICEF Ethiopia, from the Bank Netherlands Partnership Program Trust Fund managed by the Human Development Network of the World Bank, and from a Danish trust fund managed by the PREM network of the World Bank. In-kind contributions from Save the Children ­ UK both in terms of technical assistance during the survey design and training as well as follow up comments on the study are also gratefully acknowledged. Table of Contents ACKNOWLEDGEMENT.............................................................................................................. 1 LIST OF FIGURES ..................................................................................................................... 3 LIST OF TABLES....................................................................................................................... 4 ACRONYMS ............................................................................................................................. 6 EXECUTIVE SUMMARY...................................................................................................... 7 1. INTRODUCTION............................................................................................................ 9 2. THE PROGRAMMATIC BACKGROUND ............................................................... 11 THE FOOD SECURITY PROJECT.............................................................................................. 11 THE CHILD GROWTH PROMOTION COMPONENT.................................................................... 12 3. EVALUATION OF THE CHILD GROWTH PROMOTION COMPONENT....... 13 DESIGN OF CGPC EVALUATION ............................................................................................ 14 DESCRIPTION OF THE BASELINE SURVEY ............................................................................... 15 4. NUTRITIONAL FINDINGS......................................................................................... 16 Z-SCORES .............................................................................................................................. 18 DATA HANDLING................................................................................................................... 19 NUTRITIONAL STATUS........................................................................................................... 20 NUTRITIONAL STATUS OF PARENTS ....................................................................................... 25 5. HEALTH......................................................................................................................... 26 ILLNESS IN LAST 2 WEEKS...................................................................................................... 26 ILLNESS AND MALNUTRITION ................................................................................................ 29 HEALTH SEEKING BEHAVIOR................................................................................................. 29 6. ATTITUDES AND PRACTICES................................................................................. 31 PREGNANCY & BIRTH ........................................................................................................... 32 CHILD FEEDING PRACTICES ................................................................................................... 33 HYGIENE ............................................................................................................................... 39 CARE..................................................................................................................................... 39 7. KNOWLEDGE............................................................................................................... 40 CAREGIVERS' ASSESSMENT OF MALNUTRITION ..................................................................... 41 KNOWLEDGE ON SPECIFIC CHILD HEALTH AND NUTRITION TOPICS........................................ 42 SOURCES OF KNOWLEDGE ..................................................................................................... 44 8. HOUSEHOLD CHARACTERISTICS ........................................................................ 46 HOUSEHOLD COMPOSITION ................................................................................................... 46 SCHOOLING........................................................................................................................... 47 ASSET OWNERSHIP ................................................................................................................ 49 LAND OWNERSHIP ................................................................................................................. 51 SHOCKS TO HOUSEHOLD CONSUMPTION................................................................................ 53 9. KEBELES ....................................................................................................................... 55 HEALTH INSTITUTIONS .......................................................................................................... 55 INFRASTRUCTURE.................................................................................................................. 55 2 WATER & ELECTRICITY ........................................................................................................ 56 BUSINESS INSTITUTIONS........................................................................................................ 57 KEBELE PROFILES.................................................................................................................. 58 10. FSP-CREDITS............................................................................................................ 58 11. HEALTH WORKERS & PROMOTERS................................................................ 61 PERSONAL PROFILES.............................................................................................................. 61 SCHOOLING & TRAINING....................................................................................................... 63 HEALTH WORKERS AND PROMOTERS AS ROLE MODELS ......................................................... 66 12. OTHER PROGRAM ACTIVITIES......................................................................... 68 THE ENHANCED OUTREACH STRATEGY - EOS...................................................................... 68 SAFETY NET / PUBLIC WORKS ............................................................................................... 68 NGOS/CBOS ACTIVE IN COMMUNITY ................................................................................... 71 13. CONCLUSIONS......................................................................................................... 71 REFERENCES....................................................................................................................... 73 APPENDIX A ­ SURVEY METHODOLOGY................................................................... 75 A1. SAMPLE SIZE CALCULATIONS FOR AN EVALUATION OF THE CHILD GROWTH PROMOTION COMPONENT.......................................................................................................................... 75 A2. SELECTION OF SAMPLING HOUSEHOLDS.......................................................................... 76 APPENDIX B - KEBELE PROFILES................................................................................ 78 APPENDIX C - HOUSEHOLD QUESTIONNAIRE......................................................... 80 APPENDIX D ­ KEBELE QUESTIONNAIRE.................................................................. 96 APPENDIX E ­ ANIMATOR & HEALTH WORKER QUESTIONNAIRE................ 100 List of Figures Figure 1: Causes of child malnutrition..................................................................................................................... 10 Figure 2: Malnutrition across age groups ................................................................................................................ 22 Figure 3: Exclusive breastfeeding, by age groups................................................................................................... 34 Figure 4 Percentage of children still breastfed ........................................................................................................ 35 Figure 5: Bottle-feeding, by age groups................................................................................................................... 35 Figure 6: Complementary foods............................................................................................................................... 36 Figure 7: Care of child.............................................................................................................................................. 40 Figure 8: Introduction of complementary foods...................................................................................................... 42 Figure 9: Schooling by gender ................................................................................................................................. 49 Figure 10: Land holdings ......................................................................................................................................... 52 Figure 11: Timing of credits..................................................................................................................................... 59 Figure 12: Repayment of loans ................................................................................................................................ 60 Figure 13: Introduction of complementary food (month) ....................................................................................... 67 3 List of Tables Table 1: Programmatic interventions....................................................................................................................... 12 Table 2: Survey sample size..................................................................................................................................... 16 Table 3: Range control for anthropometric variables.............................................................................................. 19 Table 4: Sources for missing anthropometric Z-scores........................................................................................... 20 Table 5: Aggregated malnutrition rates across woredas.......................................................................................... 21 Table 6: Malnutrition rates across survey design groups ........................................................................................ 22 Table 7: Malnutrition by gender & age.................................................................................................................... 23 Table 8: MUAC by age groups ................................................................................................................................ 23 Table 9: Diagnostic ability of MUAC...................................................................................................................... 24 Table 10: Adult malnutrition by BMI...................................................................................................................... 25 Table 11: BMI comparisons..................................................................................................................................... 26 Table 12: Illness and treatment ................................................................................................................................ 28 Table 13: Malnutrition and illness ........................................................................................................................... 29 Table 14: Reason for no BCG vaccination .............................................................................................................. 30 Table 15: Medical treatments................................................................................................................................... 31 Table 16: Number of visits to health facility during pregnancy............................................................................. 32 Table 17: Assistance at birth .................................................................................................................................... 32 Table 18: Why was child never breastfed?.............................................................................................................. 33 Table 19: Food Group Frequencies.......................................................................................................................... 38 Table 20: Type of toilet............................................................................................................................................ 39 Table 21: Care of child if main caregiver is out of the house ................................................................................. 40 Table 22: Diagnostic ability..................................................................................................................................... 41 Table 23: Diverge in knowledge and practice for complementary foods ............................................................... 43 Table 24: Stated causes of illnesses ......................................................................................................................... 43 Table 25: Right and wrong answers for diarrhea..................................................................................................... 44 Table 26 : Most important source of knowledge on child nutrition........................................................................ 45 Table 27: Radio & TV ownership (% of hhd.s)....................................................................................................... 45 Table 28: Sources of knowledge on diarrhea and malaria....................................................................................... 46 Table 29: Household size ......................................................................................................................................... 46 Table 30: Age profile for children under 2 years of age.......................................................................................... 47 Table 31: School achievements................................................................................................................................ 47 Table 32: Reading abilities....................................................................................................................................... 48 Table 33: Distance to primary and secondary school.............................................................................................. 49 Table 34: Asset ownership now and before............................................................................................................. 51 Table 35: Development in land ownership.............................................................................................................. 52 Table 36: Comparison of land ownership and cultivation....................................................................................... 53 Table 37: Land ownership in treatment and control group ..................................................................................... 53 Table 38: Shocks in the previous 3 years................................................................................................................. 54 Table 39: Effect of most recent shock on household consumption......................................................................... 54 Table 40: Distance to health institutions.................................................................................................................. 55 Table 41: Infrastructure............................................................................................................................................ 56 Table 42: Access to utilities ..................................................................................................................................... 57 Table 43: Business institutions................................................................................................................................. 58 Table 44: Use of credit ............................................................................................................................................. 60 Table 45: Alternative credit sources ........................................................................................................................ 61 Table 46: Personal characteristics of health workers and promoters...................................................................... 62 4 Table 47: Formal education of health workers and promoters................................................................................ 63 Table 48: Training on specific topics....................................................................................................................... 64 Table 49: Was the training sufficient for your current work?................................................................................. 65 Table 50: Sufficiency of training, by HWP ............................................................................................................. 66 Table 51: First substance to newborn? - HWP ........................................................................................................ 66 Table 52: Causes of diarrhea - HWP........................................................................................................................ 67 Table 53: Participation in public works, by woreda................................................................................................ 69 Table 54: Remuneration of public works................................................................................................................. 69 Table 55: Free handouts and public works participation......................................................................................... 70 Table 56: Public works and free handouts, by male labor in household................................................................. 70 Table 57: Dependency ratio statistics ...................................................................................................................... 71 Table 58: NGO/CBO activities ................................................................................................................................ 71 5 Acronyms BMI - Body-Mass Index CGPC - Child Growth Promotion Component DID-approach - difference-in-difference approach EDHS 2005 - Ethiopian Demographic and Health Survey 2005 ETB - Ethiopian birr FDRE - Federal Democratic Republic of Ethiopia FFSCB - Federal Food Security Coordination Bureau FSP - Food Security Program HEP - Health Extension Program of the Government of Ethiopia HP - Community health promoter of the CGPC HW - Community health worker of the CGPC HWP - Community health workers and promoters of the CGPC MUAC - Mid-Upper Arm Circumference NCHS - US National Center for Health Statistics NGO - Non-governmental organization SNNPR - Southern Nations, Nationalities and People's Region of Ethiopia UNICEF - United Nations Children's Fund WHO - World Health Organization 6 EXECUTIVE SUMMARY Malnutrition in the first years of life impairs the physical and mental abilities and can permanently reduce an individual's capacity and well-being. Child malnutrition follows from a host of factors, including lack of income and food insecurity, disease, limited maternal education and poor nutritional knowledge and practices. The Government of Ethiopia in 2005 initiated the Child Growth Promotion Component (CGPC) as part of the broader Food Security Project (FSP). The CGPC seeks to improve the nutritional status of children below 2 years of age through growth monitoring and advice to the caregivers. As the first step towards an evaluation of the CGPC, a team of researchers supported by the World Bank and UNICEF in July and August 2005 conducted a baseline survey of 5706 children in 7 woredas/70 kebeles of SNNPR. The survey finds 19% of the children below 2 wasted, 33% stunted and 34% underweight, with substantial differences between the 7 woredas. The stunting and underweight-figures are similar to the EDHS 2005-findings, while wasting is higher, possibly due to the implementation of the survey during the hunger season. Boys are found to have a steeper increase in stunting as they get older than girls. The measurements of Mid-Upper Arm Circumference (MUAC) are found to be poor predictors of child malnutrition, with many false negatives. The MUAC measures detected only 42% of the wasted children, while they correctly assessed 77% of the not wasted children. Approximately one-third of the 0 to 2 year old children have had diarrhea, malaria and/or pneumonia/cold during the 2 weeks prior to the survey interviews. Cross-tabulations with the wasting index show effects of the illnesses on malnutrition, while correlations with stunting suggest the existence of reverse causality going from malnutrition to a larger incidence of illnesses. Only 57.9% of the caregivers correctly assessed their child's true nutritional status as expressed by the height-for-age index. This finding is robust to alternative malnutrition indicators and cut-off values, and quite closely matches other results from Ethiopia. The quality of the caregivers' knowledge on child nutrition and health is also mixed. The dietary recommendations for infants on e.g. exclusive breastfeeding the first 6 month and the introduction of complementary foods at the age of 6 month, are largely not followed. Less than 50% of the 0 to 3 month old infants are exclusively breastfed, and between 7 and 15% of the children in the different age groups are bottle-fed. More caregivers seem to know the recommendations on the introduction of complementary foods than the analysis of their practices indicate. Asked about the causes of malaria and diarrhea a majority knows the right answers, but many also state wrong and misleading causes. Older family members and other relatives are by far the mostly used general source of information on child health, nutrition and care, while health institutions and 7 personnel are seen as important sources of information on specific illnesses such as diarrhea and malaria. Schools are in general not stated as sources of information, but 62% of the household members older than 5 years have not completed any level of schooling. This figure even increases to 79% for the mothers of the 0 to 2 year old children. The effectiveness of the CGPC hinges on the work of the local health workers and promoters. Two-third of them think that their training for the program had not been sufficient for their job. The health workers and promoters however do show a better knowledge and health behavior than the surveyed households in general, and can act as role models for the surrounding community. Only 3% of the surveyed households state that they do not own any land, which is in line with the large share of the Ethiopian population linked to agriculture and the implementation of the survey in rural areas. The share of asset-owning households has increased in the 3 years prior to the survey for most asset categories. A high share of households are hit by ecological, price and health shocks, with for instance 95% of the household experiencing a drought in the 3 years prior to the survey. The opportunity set of households to improve their livelihood is also shaped by the external society. Woreda and kebele profiles document long distances to various institutions in the rural areas. While most of the surveyed kebeles have basic health facilities, only about half of the kebeles have a bore hole for water and the access to transport and communication infrastructure is in general low. The overall Food Security Project aims to increase the resource base of poor households and provides the initiation capital for local rotating credit schemes. These credits are seemingly very needed and popular, since over half of all interviewed households had applied for a loan. 32% of the surveyed households (in the treatment kebeles) have received a credit and 75% of all credits are used for the purchase of an ox or other livestock. Only 17 of the 70 kebeles had active child-related NGO/CBO-programs in the community, with a large part being supplementary feeding programs. The Enhanced Outreach Strategy-program by UNICEF and WFP was implemented in all the woredas in which the CGPC baseline survey was undertaken, but only 9% of the surveyed households stated that they had participated in the most recent EOS campaign. 67% of the surveyed households had participated in public works in the 6 month prior to the survey, and had mainly been paid through food handouts. Free food handouts are intended for households that are not able to participate in public works, but the survey finds a strong correlation between public work participation and the receipt of free cash or food. There is a very small targeting of public works towards households with a higher dependency ratio, while no such distinction is found for free handouts. The CGPC baseline survey was for the purpose of the overall CGPC evaluation divided into a treatment and a control group, where the CGPC is only introduced in the former. A comparison of the two groups across the various topics addressed in the questionnaire and this report shows that the control group areas have lower malnutrition rates and are in general more affluent with larger asset and land holdings, while health parameters and coverage are similar. 8 Early child capacity, delays th In combination, th from top to botto malnutrition level findings across ge underlying causes use of health facil the health seeking child malnutrition, Sections 7 the caregivers on c proper action. Sect schooling, as they 2. T The Child current section wil The Food S the Government o base of poor hous community-based (FDRE, 2002) for Table 1: Programma Woreda Boloso Sore Konso Damot Woyde Kucha Kedida Gamela Uba Debre-Tsehay Burji Note: In line with the gene where the program is implem The CGPC spans two lines of administration. At the woreda level the health office administers the CGPC, but the funding is channeled through the food security coordination bureau (at the federal, regional and woreda level), together with the overall funding for the FSP. The regional Food Security Coordination Office embraces a Food Security Program Coordinator. The CGPC complements the household-level approach to child nutrition and care with a community-wide problem analysis and action initiated by the health worker. The communities are supposed to discuss the overall growth data and decide on actions at the community and program level to support and enhance child nutrition, for which the program can provide grants. 3. EVALUATION OF THE CHILD GROWTH PROMOTION COMPONENT The objectives of the evaluation of the CGPC are fourfold: 1) Assess quantitatively the net impact of the CGPC on the nutritional outcomes of children, i.e. the impact of better caring practices imparted through improved nutritional knowledge; 2) Investigate empirically the additional impact that accrues from key exogenous factors, e.g. increased household income, maternal education, better health service delivery, physical isolation and sanitation; 3) Explore the conditions under which a child growth promotion program is most effective; in particular, examine through primary data analysis if income and nutritional knowledge act as substitutes or complements and if this relation holds across different income levels; 4) In realizing objectives 1)-3) generate Ethiopia specific empirical evidence to: a) Provide feedback to the project during its implementation (especially based on the analysis of the baseline); b) Contribute to the overall discussions and fine tuning of the national nutrition strategy, in particular regarding the scope for child growth promotion and maternal education programs as a timely and potentially cost effective intervention to reduce child malnutrition; c) Inform the broader ongoing debate on nutrition in Ethiopia. It is important to note that this evaluation does not seek to identify what the impact of a `perfect' child growth promotion project would be on nutritional outcomes. Rather, it specifically seeks to identify the impact of an actual program, with all the inherent capacity constraints and difficulties that go with it. Such feedback from experience on the ground is especially useful to inform discussions about extending and scaling up of the CGPC. The 13 government is currently considering including child growth promotion as one of the key activities of the Health Extension Program (HEP). Their experience as captured during the evaluation would also provide useful feedback as the HEP is rolled out over the coming years. DESIGN OF CGPC EVALUATION The design of the overall evaluation and baseline survey is determined by the aim to estimate the net effect of the CGPC on pre-school children's nutritional status over and above the effect of other factors. Differences in malnutrition rates between a treatment group where the CGPC is implemented and a control group without the CGPC can result from three factors: 1. Differences attributable to the program itself, 2. Initial differences between the kebeles/households in the two groups, 3. Differential changes over time across the two groups unrelated to the program. The focus of the CGPC evaluation is on estimating the first effect. It is unlikely that the treatment and control groups will be the same (point 2) and it is also unlikely that the evolution in their characteristics over time will be the same (point 3). One would optimally randomize the treatment across persons or households in order to average out all initial differences between the treated and non-treated groups and consequently also the likelihood of differential changes over time among these groups. Yet, in our case, such an experimental evaluation design would require that the kebeles that will receive the CGP component as well as the control kebeles should be randomly selected across the different regions. This is not the case. As a result, it is unlikely that the treatment and control groups will be the same (point 2) and it is also unlikely that the evolution in their characteristics will be the same over time (point 3). However, under a few plausible assumptions a difference in difference (DID)-approach complemented with data collection on key observable determinants of child malnutrition can substitute for the randomization of treatment across communities. Under the DID-approach, the change in malnutrition rates before and after the CGP program in the control group is subtracted from the change in malnutrition rates before and after the CGP program in the treatment group to control for changes over time unrelated to the CGP program, assuming they are identical across both groups. This may not be the case, because the CGP and FSP woredas and kebeles have been selected following particular criteria. By collecting additional data on key individual, household and community determinants of child malnutrition (e.g. gender, household income, paternal education, sanitary environment) as well as some of the selection criteria and using a regression framework one can further control for initial differences between the treatment and control groups as well as differential changes in the characteristics over time. This would reduce the potential contamination of the estimate of the net treatment effect3. Under plausible assumptions, DID can also remove the influence of unobservable community factors if these are constant over time since their net influence on the outcome measures is part of the observed information at the baseline. By differencing the changes in the project period, this effect is removed. 3See Appendix A1 for a more detailed and technical exposition. 14 As a first s the World Bank a woredas of SNNPR The survey from the treatmen evaluation group th interventions show 40 kebeles in the 4 control group is s Table 2: Survey sample size Woreda Number of Designed number Surveyed number Attrition rate kebele of households of households (%) Treatment group Boloso Sore 10 750 770 -2.7 Damot Woyde 10 750 625 16.7 Kucha 10 750 705 6.0 Konso 10 750 752 -0.3 Treatment group total 40 3000 2852 4.9 Control group Kedida Gamela 10 1000 992 0.8 Uba Debre-Tsehay 10 1000 922 7.8 Burji 10 1000 940 6.0 Control group total 30 3000 2854 4.9 TOTAL 70 6000 5,706 4.9 The surveyed households in each kebele were selected by systematic random sampling, surveying only households with children below 2 years of age. See Appendix A2 for a detailed description of the household sampling approach at the kebele level. Questionnaires The household questionnaire had an emphasis on the nutritional status of children below 2 years of age and the knowledge, attitudes and practices related to their care. It however also covers household composition, the income generating activities of the households, household assets and housing, participation in the FSP, and shocks to the households' livelihood. The questionnaire allowed for the recording of up to two children below the age of 2 in the child-specific modules on health and nutrition. The majority (99%) of the households however only had 1 child below 2 years of age. Besides the household survey a kebele questionnaire on community characteristics, infrastructure and institutions was administered to a group of kebele leaders. Also the health workers were surveyed in a separate health personnel questionnaire on their personal characteristics, training and knowledge, if there is a health station in the kebele. In the treatment group the same questionnaire was also administered to 3 to 4 community health promoters in each kebele. 4. NUTRITIONAL FINDINGS Anthropometrics is used to assess and predict performance, health and survival of individuals. The measurement of various body dimensions and their combinations can assist in the diagnosis of (the type of) malnutrition and its consequences. Using the age, height and weight of the children, three indices can be calculated that express specific characteristics of malnutrition (Cogill, 2003; WHO, 1997). We use the nutrition indices to characterize the malnutrition for children and their parents in the CGPC baseline survey. The following 16 paragraphs offer short descriptions of these indices. Box 1 gives the concise definitions provided by the World Health Organization (WHO, 1997). Weight-for-height helps to identify children suffering from current or acute malnutrition, and a low weight-for-height index is termed wasting. Wasting stems from (combinations of) inadequate food intake, incorrect feeding practices, diseases and infections and is a short-term measure from which children can recover if fed and cared for appropriately. Wasting in individuals or population groups can change rapidly, and is therefore highly susceptible to seasonal variations in food availability or disease prevalence. A low height-for-age reflects reduced skeleton growth resulting from repeated or chronic malnutrition and is referred to as stunting. This dependence on past malnutrition means that stunting is accumulating in the sense that repeated incidences of undernutrition add to the degree of stunting. The long-term effects of stunting include lower energy intake, a lower immune response, and poorer mental and physical capabilities (Grantham-McGregor et al., 1999a). For children under 2 years of age the long-term consequences are still reversible, but become permanent disabilities thereafter. This permanence of stunting explains the targeting of the CGPC at children between 0 and 2 years of age, since actions to preempt and reverse stunting have to be taken in this age bracket. 17 The third anthropometric index weight-for-age is a composite of the two other indices. A low weight-for-age (usually referred to as underweight) can derive both from an insufficient height-for-age of the child or from a low weight-for-height of the child, and therefore reflects both chronic malnutrition and present (acute) malnutrition. BOX 1: WHO definitions of nutrition indicators The three most commonly used anthropometric measures to assess the growth status of children are weight-for- height, height-for-age and weight-for-age. These anthropometric indices can be interpreted as follows: Low weight-for-height: Wasting or thinness indicates in most cases a recent and severe process of weight loss, which is often associated with acute starvation and/or severe disease. However, wasting may also be the result of a chronic unfavourable condition. Provided there is no severe food shortage, the prevalence of wasting is usually below 5%, even in poor countries. The Indian subcontinent, where higher prevalences are found, is an important exception. A prevalence exceeding 5% is alarming given a parallel increase in mortality that soon becomes apparent (Toole and Malkki, 1992). On the severity index, prevalences between 10-14% are regarded as serious, and above or equal 15% as critical. Typically, the prevalence of low weight-for-height shows a peak in the second year of life. Lack of evidence of wasting in a population does not imply the absence of current nutritional problems: stunting and other deficits may be present (Victora, 1992). Low height-for-age: Stunted growth reflects a process of failure to reach linear growth potential as a result of suboptimal health and/or nutritional conditions. On a population basis, high levels of stunting are associated with poor socioeconomic conditions and increased risk of frequent and early exposure to adverse conditions such as illness and/or inappropriate feeding practices. Similarly, a decrease in the national stunting rate is usually indicative of improvements in overall socioeconomic conditions of a country. The worldwide variation of the prevalence of low height-for-age is considerable, ranging from 5% to 65% among the less developed countries (de Onis et al., 1993). In many such settings, prevalence starts to rise at the age of about three months; the process of stunting slows down at around three years of age, after which mean heights run parallel to the reference. Therefore, the age of the child modifies the interpretation of the findings: for children in the age group below 2-3 years, low height-for-age probably reflects a continuing process of "failing to grow" or "stunting"; for older children, it reflects a state of "having failed to grow" or "being stunted". It is important to distinguish between the two related terms, length and stature: length refers to the measurement in recumbent position, the recommended way to measure children below 2 years of age or less than 85 cm; whereas stature refers to standing height measurement. For simplification, the term height is used all throughout the database to cover both measurements. Low weight-for-age: Weight-for-age reflects body mass relative to chronological age. It is influenced by both the height of the child (height-for-age) and his or her weight (weight-for-height), and its composite nature makes interpretation complex. For example, weight-for-age fails to distinguish between short children of adequate body weight and tall, thin children. However, in the absence of significant wasting in a community, similar information is provided by weight-for-age and height-for-age, in that both reflect the long-term health and nutritional experience of the individual or population. Short-term change, especially reduction in weight-for-age, reveals change in weight-for-height. In general terms, the worldwide variation of low weight-forage and its age distribution are similar to those of low height-for-age. Source: WHO (1997, p.46/47) Z-SCORES For the comparison of the anthropometry of children across population groups and countries the weight-for-height, height-for-age and weight-for-age indices are usually interpreted using Z-scores. A Z-score is a statistical concept that describes an outcome in 18 terms of its number of standard deviations (SD) or Z-scores from the median of its distribution (WHO, 1997). Z-scores thereby become independent of the measurement unit of the specific measure. To compare the nutritional outcomes, they are related to a reference population of children. In this report the most commonly used reference standards developed by the US National Center for Health Statistics (NCHS), and recommended by the World Health Organization5 are applied. A Z-score is written as [observed value ­ median of reference population] Z-score = [standard deviation of reference population] The following discussion of the nutritional status of children in the CGPC baseline survey will use Z-scores, or measures derived from these. Critical values for malnutrition expressed in Z-scores are ­2 SD (standard deviations) and ­3 SD. Percentages of children wasted, stunted or underweight describe the percentage of children with the respective Z- scores below ­2 SD. DATA HANDLING The assessment of nutritional status requires a number of composite background data. Body stature is captured by length, weight and Mid-Upper-Arm Circumference (MUAC). Length is recorded instead of height, since children below 2 years of age are the target group of the CGPC and therefore of the evaluation baseline survey. For any child, the recumbent length measurement is always greater than the standing height measurement. Children below 2 years of age cannot always stand well, and to ensure consistency length was measured for all children. This corresponds to the general practice for the anthropometric measurement of children (Cogill, 2003; Dibley et al., 1987; WHO, 1997). In a first stage of data cleaning these anthropometric variables are cleaned for outliers by applying the bounds detailed in table 3. This looses a minor number of observations on length and MUAC. Table 3: Range control for anthropometric variables Measure Range Comment Total number of Set to recorded obs. missing Length Between 37cm -6/+6 standard deviations from the 5405 21 (0.39%) and 105cm NCHS/WHO reference tables. Weight Between 0.3kg -6/+6 standard deviations from the 5666 0 and 23kg NCHS/WHO reference tables. MUAC 0cm and 25cm Maximum measure on UNICEF MUAC- 5548 37 (0.67%) tape for children 5See http://www.who.int/nutgrowthdb/reference/en/ and WHO (1997) for further information on the reference standard data. 19 The exact age of the child is crucial for two of the nutritional indicators. The age of the surveyed children is calculated after transforming the interview date and the recorded birth date from the Ethiopian calendar6 to the Gregorian (European, US) calendar. 320 (5.5%) observations on the age of the children are missing (due to invalid interview or birth date) and 160 (2.8%) are outside the 0 to 24 month range. Beyond the cleaning of the underlying variables, the anthropometric Z-scores exceeding +/-6 standard deviations are set to missing. Table 4 provides a full overview over the sources of missing anthropometric Z-scores. The full cleaning of the anthropometric data looses approximately 9% of the observations for each of the three anthropometric measures, and leaves approximately 4900 valid observations for each of the three anthropometric indices. Table 4: Sources for missing anthropometric Z-scores Weight-for-Height Height-for-Age Weight-for-Age Freq. Percent Freq. Percent Freq. Percent Valid entrya 4,892 91.23 4,781 89.16 4,911 91.59 Weight is missing 13 0.24 - - 38 0.71 Length is missing 157 2.93 172 3.21 - - Age is missing - - 175 3.26 179 3.34 Both var.s are missing 35 0.65 17 0.32 10 0.19 Z-score outside rangeb 265 4.94 217 4.05 224 4.18 Totalc 5,362 100 5,362 100 5,362 100 b if both underlying variables are not missing and in the valid range, see table 3. a The range is set to +/-6. c The total number of observations on the anthropometric data is lower than the total number of households sampled, since the combination of several data modules incurs losses through unclear identifications. NUTRITIONAL STATUS The malnutrition indicators for the full survey are reported in the first row of table 5. 19% of the children between 0 and 2 years are found to be wasted, i.e. have Z-scores of the height-for-age index below ­2. This share of wasted children is higher than generally observed, and the Ethiopian Demographic and Health Survey 2005 (reproduced in the last row of table 5) found 13.7% of the children between 0 and 24 month to be wasted. The timing of the CGP baseline survey in July can be contributory to the divergence, as the time period before the main rains is in general considered a lean period in Ethiopian agriculture. The CGPC survey found 33% of the children to be chronically undernourished, or stunted, while 34.2% were underweight, i.e. fell below ­2SD for the weight-for-age index. The stunting and underweight indices include the accumulating effect of earlier malnutrition spells, and are therefore less influenced by current or acute malnutrition. The CGPC baseline survey results on these indicators are similar to the EDHS 2005 findings. 6The Ethiopian calendar consists of 13 months where the first 12 months have 30 days each, and the last (thirteenth) month has 5 days (6 days in a leap year). The Ethiopian New Year falls on September 11 (September 12 in the leap year). 20 The region between the 7 wor with the highest ra reflecting longer-t malnutrition for stu wasting among its Table 5: Aggregated Woreda Table 6: Malnutrition Treatment group (T) Control group (C) Total Test of difference bw T H_o: T = C H_1: T > C child malnutrition 1-2 percentage poi Table 7: Malnutrition Gender & age group Wast (month) Percent Male 0 - 3 25 3 - 6 20 6 - 9 14 circumference me standardize and th (Conolly, 2005). T as the true measur tool for malnutritio 12.5cm as the cut- 2SD as the cut-off) A Pearson through the wastin the 0.1%-level). H column in table 9 A measure Body-Mass-Index as and is considered adults the BMI is BMI contributes t Table 11 c control woredas ha and the t-test of H0 and the correspond Table 11: BMI comp Group Obs Fathers 2417 Mothers 4084 leading to a lower age groups had had In the com difference. Approxima ailment. Almost th facility, while app incidence of visits high proportion th households who Table 12: Illness and treatment Pneumonia/ % Diarrhea Malaria Cold Total (1) (2) (3) (4) Total 16 16 13 36 By gender Male 17 17 14 37 Female 16 15 13 35 By age group 0 - 3 9 8 11 22 3 - 6 16 9 17 34 6 - 9 19 15 16 41 SK 9 - 12 18 16 14 38 12 - 15 18 18 13 37 WEE 15 - 18 15 24 14 39 2 T 18 - 21 16 18 11 37 LAS 21 - 24 16 18 12 38 LLI Treatment/Control Treatment 18 16 13 35 LD HI Control 15 15 14 37 C Seek treatment if sick? 60 59 59 Where treatment? Traditional healer 12 8 8 Family member 1 1 2 Drug shop 4 7 7 Health post 33 40 31 Health center 39 29 37 Hospital 1 1 1 Private Clinic 6 10 6 Other 1 0 1 Missing obs. 4 4 7 Kind of treatment ORT/ORS 57 5 13 Medication 30 85 68 Medicine/herbal drugs 6 5 8 Other 2 1 4 Missing obs. 5 4 7 Why no treatment? Don't know where to go 5 3 4 Not permitted to go 3 1 4 T No money for treatment 69 73 63 EN No health facility nearby 15 17 15 TM No transport 2 2 3 REA Other 2 1 3 T Missing obs. 4 4 8 28 ILLNESS AND MALNUTRITION Illness contributes to malnutrition, but malnourished children are also more likely to fall sick. For both diarrhea and pneumonia we can detect a (two-way) correlation between malnutrition and illness in the CGPC baseline survey. The upper left quadrant of table 13 shows the tabulation of wasting against diarrhea. A slightly higher share of the wasted children had had diarrhea over the previous 2 weeks than the not acutely malnourished children, and the correlation is significant at the 2%-level. We can however not conclude on the direction, since diarrhea can cause the loss of weight, possibly resulting in wasting, but the weakness of the wasted body also can increase the probability of incurring diarrhea. We find a stronger correlation if we in lower left quadrant tabulate diarrhea against stunting, where there is a difference of 4%-points between the malnourished and not malnourished children. From the correlation of stunting with diarrhea we can conclude that stunting increases the incidence of diarrhea, since stunting is the result of long-term, cumulative malnutrition and a spell of diarrhea cannot directly affect the rate of stunting of the child. For pneumonia in the rightmost quadrants of table 13, we find similar correlations with malnutrition, albeit with a lower correlation coefficient and lower level of significance for stunting. Table 13: Malnutrition and illness Diarrhea Pneumonia/Cold % No Yes No Yes No 68 12 70 10 Wasted Yes 16 4 16 3 0.033 (0.020) 0.035 (0.014) No 57 10 58 9 Stunted Yes 27 6 29 5 0.056 (0.000) 0.025 (0.080) Note: Numbers denote cell percentages. Correlation coefficients below subtables (p-value in parenthesis). HEALTH SEEKING BEHAVIOR Illnesses and nutrient deficiencies can be prevented through vaccinations, medicines and the intake of supplementary nutrients. We will here describe the coverage for the 0 to 2 year old children for various vaccines, deworming medicines and supplementary vitamin A. A vaccination against tuberculosis through the injection of BCG typically causes a visible scar in the shoulder, and 83% of the surveyed children had received such a vaccination, see column 1 of table 15. If the children hadn't received any BCG-vaccination, the caregiver mainly gave the young age as the reason (47%), while 21% didn't know about the vaccination (see table 14). However, in 4% of the non-vaccination cases the caregiver 29 didn't assign the v or the drug not ava Table 14: Reason for Reason stated Don't know vaccinatio Vaccination does not h Too young Too far to go Drug not available than 6 month. 70% of all surveyed households in the CGPC baseline had given their child(ren) older than 6 month vitamin A during the last 6 months (see column 5 of table 15). This percentage is quite stable across the gender of the child and the 7 woredas, where the fraction varies between 54 and 70%. 88% of these households had received their vitamin A capsules from public clinics or institutions or an NGO, while a further 10% had received it through the EOS campaign. A comparison of the treatment of the children across the treatment and the control group in the last block of table 15 shows only minor differences in coverage levels. Table 15: Medical treatments Percentage of children with treatment BCG Polio Measles Deworming Vitamin A (1) (2) (3) (4) (5) Gender Male 83 94 77 56 71 Female 83 93 75 54 69 Woredas Boloso Sore 79 95 84 46 69 Damot Woyde 81 92 73 63 72 Kucha 85 94 70 57 71 Konso 90 95 83 49 68 Kedida Gamela 85 93 83 72 72 Uba-Debre Tsehay 79 92 62 51 73 Burji 83 93 74 49 64 Age group (month) 0 - 3 55 69 3 - 6 75 90 6 - 9 88 97 54 9 - 12 88 98 72 71 12 - 15 87 97 77 57 73 15 - 18 90 98 77 58 71 18 - 21 88 97 76 50 67 21 - 24 90 98 82 55 75 Treatment/Control group Treatment 84 94 78 53 70 Control 82 93 73 57 70 TOTAL 83 94 76 55 70 6. ATTITUDES AND PRACTICES Sufficient and correct knowledge on nutrition, health and care of children is a prerequisite for proper action by the caregivers. A number of questions in the CGPC baseline survey therefore examine the attitudes and practices of the caregivers with respect to childcare, health and nutrition. In 93% of the cases the mother of the youngest child in the household answered these questions. 31 Already th acutely malnourish higher probability weight generally h et al., 1989, c.f. G during pregnancy t securing this is to CGPC baseline sur Advice fro CHILD FEEDING PRACTICES The assessment for child feeding practices of the 0 to 2 year old children of the CGPC baseline survey includes breastfeeding, bottle use, the frequency of feeding solids or semi- solids, and the food group frequency over the last 7 days. This section discussed these issues one by one. Breastfeeding Breastfeeding has been found to be one of the most effective strategies to improve nutrition and prevent and reduce illnesses in children, see Huffman and Steel (1995) for a discussion. The benefits derive both from exclusive breastfeeding over the first months of life and from continued breastfeeding with supplementary foods later on. The reduction in fertility from breastfeeding and the resulting longer birth intervals also ensure better nutrition and childcare. In an overview over a number of studies examining the effect of breastfeeding on the mental development of children 24 month and younger, Grantham-McGregor et al. (1999b) conclude that breastfed infants appear to have a small but consistent advantage over non-breastfed infants at all time points up to 24 month. In the CGPC baseline survey 88% of all children covered have been breastfed sometime since their birth12. If they were not breastfed, over half of the cases were related to the mother in terms of illness, nipple problems, insufficient milk or work, see table 18. Table 18: Why was child never breastfed? Reason Percent Mother ill/weak 21.8 Child ill/weak 8.6 Child died 0.8 Nipple/breast problem 3.7 Insufficient milk 21.8 Mother working 4.1 Child refused 12.8 Other 26.3 Total 100 An immediate start of breastfeeding after birth is crucial. However, only 53% of the children in the CGPC baseline survey had been given the breast during the first hour after birth. The first breast milk/colostrum provides unique nutrients and protection against infectious disease, but only 56% of the children received it, while in the remaining cases it was pumped out and thrown away. For the first 6 month the only dietary recommendation for children is to be exclusively breastfed, where children do not receive fluids, had been fed mashed, pureed or solid food during the last 24 hours, nor were bottle-fed. Figure 3 illustrates the percentage of exclusive breastfeeding for 3-month age groups for the CGPC baseline survey. We derive a lower and 12If only non-missing observations on breastfeeding are included, this percentage increases to 94%. 33 an upper bound for exclusive breastfeeding. For the lower bound missing observations for food, fluids or bottle-fed are treated as "Yes" (thus marking the child as non-exclusively breastfed), and as "No" for the upper bound. The missing observations can both confirm or refute exclusive breastfeeding, and the true percentage will therefore be between the two bounds. There is a decreasing trend in the percentage of exclusively breastfed children as the children grown older. For the 0 to 3 month olds between 17 and 48% only receive breast milk, while the rate falls to between 12 and 36% for the 3 to 6 month old children. Even some the 6- 12 month old infants, who should be receiving complementary foods by that age, are still exclusively breastfed. Figure 3: Exclusive breastfeeding, by age groups It is recommended to breastfeed the infants and children beyond the age of 6 month alongside other foods at least up to 2 years of age. Figure 4 graphs the percentage of children still breastfed by age groups, allowing for an upper and a lower bound, where missing observations have been interpreted as "Yes" and "No" respectively. The figure shows a declining but high rate of breastfeeding, where the rate of breastfeeding falls from around 90% for the 0-6 months old infants to approximately 70% for the 18 to 24 months old children. 34 Figure 4 Percentage o Frequency of feed It is advis breastfeeding. In t received any solid Figure 6 re (complementary) f observations in complementary-fo Table 19 shows summary statistics for the food groups. All groups have a mean significantly different from 0 at least at the 1%-level, but there are clear differences between the food groups. While milk, other liquids, and food made of grains are given comparably often, the other food groups are very rarely fed to the children. Milk and food made of grains had on average been fed to the children 3 of the previous 7 days, and more than 10% of the children had had food from these groups or from the other liquids groups on each of the last 7 days. For the other groups less than 10% had received the food more than 3 days (groups e, f and i), more than 2 days (group g), more than 1 day (group j) and less than 10% of the children had had any meat at all over the last 7 days. The comparison of the treatment and control group in the last columns shows a mixed picture. For all but group g there is a significant difference in the mean number of times food from these groups had been given to the children, but the difference goes either way. There does not seem to be any qualitative trend in the difference, since the children of treatment group for instance receive food rich in vitamin A more often, while the children in the control group receive foods with oil, fat or butter more often. This last comparison however fits into the earlier discussed trend that the control group areas are slightly more affluent than the treatment group areas. 37 Table 19: Food Group Frequencies All households Treatment Control T & C different? Type of food F-stat (sign.) Mean (Std. dev) Median q90 Mean (Std. Dev) Mean (Std. dev) of Ho: T=C H1 a Milk 3.17 (3.03) 3 7 3.52 (3.14) 2.82 (2.88) 8.713 (0.000) T > C b Fruit juice 0.28 (1.05) 0 0 0.3 (1.12) 0.25 (0.97) 1.864 (0.031) T > C c Other liquids 1.67 (2.42) 0 7 1.58 (2.38) 1.77 (2.45) 3.049 (0.001) C > T d Food made of grains 2.86 (2.99) 2 7 2.74 (2.97) 2.98 (3.01) 2.960 (0.002) C > T e Food rich in vitamin A 0.7 (1.48) 0 3 0.76 (1.55) 0.64 (1.41) 3.169 (0.001) T > C f Food made from roots/tubers 0.68 (1.55) 0 3 0.79 (1.7) 0.56 (1.37) 5.641 (0.000) T > C g Other fruits & vegetables 0.54 (1.17) 0 2 0.53 (1.11) 0.55 (1.23) 0.571 (n.s.) C > T h Meat 0.14 (0.64) 0 0 0.16 (0.74) 0.11 (0.52) 3.006 (0.001) T > C i Food made from legumes 0.8 (1.88) 0 3 0.71 (1.81) 0.89 (1.95) 3.445 (0.000) C > T j Food with oil, fat or butter 0.3 (1.08) 0 1 0.24 (1.01) 0.35 (1.15) 3.635 (0.000) C > T The hygien in the occurrence, Intestinal b the preparation of will kill off most their drinking wat resource costs in t firewood for cooki Clean sanit Figure 7: Care of chi malnutrition. The p the growth of thei sources for knowle A prerequi growth of the chil child growth moni malnutrition and im 22 reports that 58 28.1% think that th range of correct as The result also qui nationally represen correct assessment KNOW In section over the first six complementary fo semi-solid foods optimum is missed Table 23: Diverge in knowledge and practice for complementary foods Exclusively Not excl. breastfed/When best to start Age group breastfed (%) Given compl. foods (%) complementary food? Low High Low High Pct. reverse cum pct.a (1) (2a) (2b) (3a) (3b) (4a) (4b) 0 - 3 17.4 48.4 51.6 82.6 2.7 100.0 3 - 6 11.7 35.7 64.3 88.3 21.5 97.3 6 - 9 4.4 14.1 85.9 95.6 68.5 75.8 9 - 12 1.3 5.8 94.2 98.7 3.7 7.3 12 - 15 1.3 4.6 95.4 98.7 3.3 3.6 15 - 18 0.7 3.6 96.4 99.3 0.2 0.3 18 - 21 0.2 2.4 97.6 99.8 0.0 0.1 21 - 24 0.2 1.0 99.0 99.8 0.1 0.1 aPercentage who advice the introduction of complementary food at this age or later. Malaria and diarrhea are two typical illnesses affecting children and their growth and development negatively. A first step towards reducing the incidence of illness is to ensure a sufficient knowledge about the specific causes of malaria and diarrhea. Table 24 reports the causes stated by the caregivers in the CGPC baseline survey. Table 24: Stated causes of illnesses Percentage of caregivers who stated the following causes a: for diarrhea for malaria 1) Dirty water & food 58.1 1) Mosquito bites 67.6 2) Food left outside 15.3 2) Impure food or water 16.2 3) Dirty dishes for eating 10.9 3) Spiritual or evil eye 1.7 4) Evil eye 5.4 4) Teething 1.8 5) Teething 15.2 5) Lack of food 16.4 6) Lack of hygiene 17.3 6) Other 2.5 7) Exposure to sun light 1.7 7) Don't know 19.4 8) Lack of food 17.5 9) Other 0.6 10) Don't Know 24.1 aThe caregivers could state several causes. The majority of caregivers/households ­ 58% ­ do know that dirty water and food can contribute to the emergence and persistence of diarrhea. Smaller percentages also give other unhygienic or unclean practices as causes. However, the second largest share of households ­ 24% ­ do not know the causes of diarrhea at all, and considerable shares of households believe that the evil eye, teething, the lack of food or even the exposure to sunlight contribute to diarrhea. A similar picture emerges from the answers given on the cause of malaria. Two-third of the households do state mosquito bites as the cause of malaria, but 19% are ignorant, and 16% think that impure water or food or the lack of food are contributory to malaria incidences. 43 Caregivers could state several causes and the shares given in table 24 are therefore not unambiguous. We can instead count the number of right answers for the causes of diarrhea (causes 1-3 and 6) and wrong answers (causes 4, 5, 7 and 9), and find the percentages given in table 25. This shows us that 43.8% of the respondents were fully right in the sense that they only stated right causes for diarrhea, while only 7.4% of the respondents were fully wrong and only gave wrong causes. The remaining caregivers either do not know at all (23.5%) or give mixtures of right and wrong causes. Table 25: Right and wrong answers for diarrhea Number of wrong answers % 0 1 2 3 0 23.5 5.2 1.4 0.8 Number 1 27.9 7.8 5.9 0.0 of right 2 9.9 7.0 0.2 0.0 answers 3 5.8 0.4 0.1 0.0 4 0.2 1.4 0.1 0.0 Legend: Fully right Fully wrong The same analysis of right and wrong answers for the causes of malaria (where only cause 1 is right) shows that 45.5% of the caregivers are fully right, 8.6% are fully wrong, 22% mention mosquito bites as the cause for malaria alongside other (wrong) causes, and 19.4% do not know at all. This result is in so far encouraging as even people who state mosquitoes alongside other causes know the importance of mosquito nets, and therefore can be counted into the share of people aware to the cause of malaria. SOURCES OF KNOWLEDGE A first clue to sources of misleading information as well as an indication of ways to change false perceptions on illness causes is the examination of where the households and caregivers obtain their knowledge. General sources Over 70% of the caregivers gave older family members or other relatives as their most important source of information on child health and nutrition (table 26). The small 3%-share of media (newspaper, radio, TV) as a source of information is clearly a result of the low penetration of the rural areas by printed or electronic media. In the present survey 95% and 90% of the caregivers answered that they never read a newspaper or listen to the radio, respectively. This is however slightly at odds with the finding that 12% of the households state that they own a radio and 1.5% a TV (table 27), in particular since a wider audience than only the owner often listens to the radio or watches TV. The ownership data also show an increase in the frequency of these media receivers in the 3 years preceding the survey. 44 Table 26 : Most important source of knowledge on child nutrition Source Percent Older family members 56.39 Other relatives, friends and neighbors 15.71 Community bulletin board 1.09 Community or local newspaper 2.42 National newspaper 0.07 Radio 0.60 Television 0.05 Groups or associations 4.29 Community leaders 7.15 Health post 9.99 Clinic 1.94 Hospital 0.07 NGOs 0.24 Total 100 Table 27: Radio & TV ownership (% of hhd.s) Now 3 years ago Radio 12 8.8 TV 1.5 1.3 Knowledge sources for specific topics Along with the knowledge on the causes of malaria and diarrhea the caregivers were asked on the source(s) of their knowledge, and table 28 reports the results. About one-third of the caregivers had their knowledge from a health institution or a nurse or doctor. The next biggest source of information ­ one-quarter ­ is the previous generation, which can possibly reproduce outdated or false knowledge. Only 4% had received their knowledge from school, which both points at the low rate of schooling and at a low transmission of knowledge on child health and nutrition in school. Below we will show that only 21% of the mothers of the 0 to 2 year old children have completed any schooling grade, but even of these a high proportion did not consider or recall school as a source of knowledge. However, even though 21% of mothers went to school, an even smaller percentage made it beyond the first few grades. 45 Table 28: Sources of Pct. of caregivers wh Diarrhea Mother/grandmother School Health institution NGO/CBO Written information Experience Don't know The share of boys and girls is approximately equal in each of the age groups, and a 2- test cannot be rejected. In both the treatment and control areas the distribution of children in the age groups follows the overall overrepresentation of children in the 9-18 months range, but children in the treatment group are on average 22 days older15. Table 30: Age profile for children under 2 years of age Age group (months) All 0 ­ 3 3 - 6 6 - 9 9 - 12 12 - 15 15 - 18 18 - 21 21 - 24 Total % Total 100 10.7 10.8 10.3 14.1 20.9 13.8 9.8 9.6 100 Male 51.9 11.0 10.7 10.0 13.7 21.0 13.6 10.3 9.6 100 Female 48.1 10.2 10.8 10.7 14.5 20.8 14.0 9.2 9.7 100 Treatment 49.7 9.4 10.0 10.6 13.5 23.5 13.2 10.2 9.7 100 Control 50.3 11.9 11.5 10.0 14.7 18.5 14.4 9.5 9.6 100 SCHOOLING A basic transfer mechanism for knowledge and awareness is general schooling. However, of all household members older than 5 years of age, 62% had not completed any level of schooling, see table 31. After an increase in the percentage from grade 1 to grade 2 the percentage of children having completed subsequent grades falls steadily. Only 0.1% (or 13 persons) of all surveyed household members finished their schooling with a university degree. Small percentages have completed adult education, religious school or other education. Table 31: School achievements FORMAL SCHOOLING Percent Cum. percent OTHER SCHOOLING Percent No level completed 62.4 62.4 Adult education 0.9 Grade 1 4.5 66.9 Religious school 0.1 Grade 2 5.9 72.8 Other 0.04 Grade 3 5.3 78.0 Missing 1.1 Grade 4 4.6 82.6 Don't know 0.3 Grade 5 3.7 86.3 Grade 6 3.6 89.9 Grade 7 2.7 92.6 Grade 8 2.0 94.6 Grade 9 1.3 95.9 Grade 10 0.9 96.8 Grade 11 0.2 96.9 Grade 12 0.5 97.4 Certificate 0.1 97.5 College/University 0.1 97.6 15This age difference is significant different from 0 at the 0.1% level. 47 Basic schooling is expected to pass on both general abilities in terms of reading, writing and calculus and more specific knowledge on for instance child-related issues. For the transmissions of knowledge on child health and nutrition both kinds of transfers are relevant, for different reasons. The specific teaching on the one hand can have a direct impact on the knowledge on child nutrition if it includes topics on e.g. (child) health, hygiene and nutrition. The general abilities, in particular reading, on the other hand enable people and their households to absorb information at later points in time. Their usefulness is however conditional on the maintenance of the skills and the availability of written materials. The latter is an issue for policy programs and outside the scope of this study, but we can examine the maintenance of the general skills using specific questions in the CGPC baseline survey on the reading and writing skills of the household members. Table 32 presents the shares of household members (5 years of age or older) by educational level, who state that they can read a letter16. Overall are 30% of the household members able to read a letter, a slightly lower share than the 38% who have at least 1 year of schooling (table 31). This disparity seems to stem mainly from the household members who have only completed between 1 and 3 years of schooling. In this group only one-third has acquired reading skills in their (few) years of schooling. Table 32: Reading abilities Reading ability (%) Educational level Yes No Missing No education 1 82 17 Grades 1-3 63 34 4 Grades 4-6 96 3 2 Grades 7-10 97 2 1 Grades 11-14 99 0 1 Other education 12 84 5 Missing 5 17 78 Total 30 58 12 Distinguishing schooling by gender shows a clear difference among men and women, or boys and girls. Figure 9 shows the percentages of household members (over 5 years of age) that have achieved certain levels of schooling. 55% of the surveyed men/boys have no education, against 72% of the women/girls. The percentage of persons who have completed higher levels of schooling is correspondingly higher for men/boys than for women/girls. 16The ability is not fully self-reported, but stated by the respondent for the household, in most cases the household head. 48 Figure 9: Schooling by gender 80 70 60 e 50 agt 40 cen erP30 20 10 0 No education Grades 1-3 Grades 4-6 Grades 7-10 Grades 11-14 Other educ. Male Female When we look at the mothers of the surveyed 0 to 2 year old children, the picture is even gloomier. Of these 79% have not had any schooling and correspondingly fewer have made it to any (higher) grades. The school attendance rate is also determined by the distance and transport time to school. Table 33 reports the existence of a primary-secondary school in the kebele and the average distance in terms of kilometers and travel by usual mode of transport to the schools. Primary schools are often found nearby. 80% of all kebeles have a primary school, and the average distance is correspondingly small. Secondary schools on the other hand are more dispersed, with on average 20 km or 2.5 hours of transport from the kebeles to the school. Table 33: Distance to primary and secondary school Primary school in kebele km hours Secondary school in kebele km hours Boloso Sore 90% 0.6 0 Boloso Sore 0% 13.0 0.7 Damot Woyde 78% 2.1 0.3 Damot Woyde 0% 16.0 2.6 Kucha 70% 4.3 0.2 Kucha 0% 36.0 3.9 Konso 88% 0.3 0.2 Konso 13% 14.0 0.8 Kedida Gamela 80% 1.1 0.2 Kedida Gamela 0% 19.0 2.8 Uba-Debre Tsehay 100% 0 0 Uba-Debre Tsehay 22% 21.0 4.4 Burji 56% 1.4 0.3 Burji 11% 17.0 1.3 Total 80% 1.4 0.2 Total 7% 20.0 2.5 ASSET OWNERSHIP The income generation of agricultural households is ­ beyond the availability of land to be discussed in the next section ­ based on its movable assets. In this section we will therefore discuss the ownership of assets that are important of the households earning capacity. Table 34 lists the surveyed assets and the 2nd and 3rd columns report the percentage of households who owned the specific asset at the time of the survey and 3 years prior to the 49 survey. The general picture is that an increasing share of households owned the various assets. The share of oxen-owning households had for instance increased from 48% 3 years ago to 54% at the time of the survey17. Similar increases are observed for most other assets. A high- flyer among the assets are bed-nets, for which the share of owners among the households has increased fourfold. This is most likely the result of national malaria campaigns that have distributed mosquito nets or sold them cheaply. The comparison of asset ownership in the treatment and control groups in columns 4 and 5 reveals the same pattern as observed in other contexts; a larger share of the households in the control group own the larger and more valuable assets, such as cattle and other livestock. A difference that may turn out to be important for the level of information and knowledge is the far higher share of households in the control group who own a radio or a TV. 17The difference is significant at the 0.1%-level. 50 Table 34: Asset ownership now and before Pct. of households who own asset Asset At time survey of 3 years earlier Treatment group Control group at at time of survey time of survey (1) (2) (3) (4) (5) Oxen 54 48 46 61 Milk cow 56 48 54 58 Other cattle 43 31 43 43 Horses,donkeys,etc. 17 16 12 22 Sheep/goat 44 33 43 45 Chicken/poultry 38 33 33 43 Scythe 49 43 50 47 Sickle 79 71 74 85 Axe 73 64 75 71 Pick axe 36 33 42 31 Plough 49 43 44 53 Wheelbarrow 3.7 3.4 3.1 4.4 Sewing machine 1.4 1.4 0.7 2.1 Loom 2.2 2.1 2.2 2.2 Sprayer 1 0.9 0.4 1.6 Tractor 0.9 0.9 0.3 1.5 Pump 1.1 1 0.4 1.8 Radio 12 8.8 6.5 18 TV 1.5 1.3 0.8 2.1 Video 1.1 1 0.5 1.6 Refrigerator 1.2 1 0.5 1.9 Bed nets 6.8 1.6 0.6 13 Bicycle 1.5 0.9 0.6 2.4 Car/Truck 1 0.8 0.4 1.6 Motorcycle/Moped 1 0.8 0.4 1.7 Other (specify) 2.3 1.7 1.8 2.8 LAND OWNERSHIP 85% of the Ethiopian population is linked to the agricultural sector, and land access is thus a major determinant of their livelihood. The sample of the present survey is taken from rural areas, and the majority of households rely on farming for their income. Only 0.3% of the households state that they do not own any land18. Figure 10 illustrates the distribution of land holdings, where a high frequency of plot sizes between 0 and 1 ha is obvious. 62% of the households only own 0.5ha or less, and 78% own 1 ha or less19. 18Land ownership is here not in a literal sense, as the Ethiopia state owns all land, and only assigns user rights that in principle can be revoked on short notice. 19There are 7.6% missing observations on land ownership, which may also indicate that the respective households do not own land. 51 Figure 10: Land hold Table 36: Comparison of land ownership and cultivation Land ownership (ha) Pct. of hhd.s Mean Median Cultivated = owned 76 0.53 0.5 Cultivated < Now owned 3 0,76 0.5 Cultivated > Now owned 6 0.40 0.25 Missing 15 - - Total 100 0.53 0.5 A comparison of land holdings across the two evaluation groups (table 37) shows that the households in the control group on average own plots twice as large as the households in the treatment group. This confirms the earlier discussion of the differences between the two groups, where the treatment areas have a priori been selected from the food-insecure, i.e. poor, areas. A t-test on the equality of means is also soundly rejected in favor of larger land holdings in the control group. Table 37: Land ownership in treatment and control group Land ownership (ha) Evaluation group Mean Std.dev. Min Max Median Treatment 0.37 0.29 0 4.5 0.25 Control 0.65 0.46 0 3.75 0.5 Total 0.53 0.42 0 4.5 0.5 H_o: T = C t = 24.78 H_1: C > T P=0.000*** SHOCKS TO HOUSEHOLD CONSUMPTION The surveyed households are prone to shocks, both from their natural environment and illness and death, given the low level of health services in the remote areas. With a lack of insurance markets shocks to the farming or property of the household are easily translated into consumption downturns. And shocks can in line with their overall effect on the household have a high impact on child nutrition. Table 38 (column 2) shows that a high share of households has been hit by one or more of the shocks in the 3 years prior to the CGPC survey. Especially droughts and rainfall- related shocks have been frequent with respectively 95% and 67% of the household experiencing at least one drought or rainfall-related shock over the previous 3 years. And considerable shares of the households have also been hit by a major harvest loss, a loss of livestock, and (at least) one major illness in the household that has not resulted in death. The remaining columns of table 38 report on the spread of the shocks in the community, as stated by the households. The answers show some clear tendencies. The ecological shocks (drought, rainfall-related shocks, harvest losses) and price changes on the 53 one hand affect almost all households21. Theft, loss of livestock and not least illness and death on the other hand to a higher degree only hit the specific household. Table 38: Shocks in the previous 3 years Shock Pct. of hhd.s Affected households affected Only this hhd Few hhd.s Many hhd.s Almost all hhd.s (1) (2) (3) (4) (5) (6) Drought 95 2 3 17 79 Heavy rainfall, flooding, untimely rains 67 1 10 33 55 Unexpected decline in prices 12 4 8 21 67 Major harvest loss 53 3 11 29 57 Theft of household assets 4 30 9 27 34 Unemployment from paid job 2 10 8 22 59 Loss of livestock (death, theft, illness) 43 37 11 20 32 Loss of land (reallocation, transfer) 6 35 8 13 44 Substantial storage loss 2 13 18 43 27 Major illness not resulting in death 54 57 7 19 17 Death 14 86 9 5 1 The households were asked on how the shocks affected their consumption (see table 39), and across all shocks a high percentage of the household reports a very negative effect on their consumption level. Table 39: Effect of most recent shock on household consumption Effect on household consumption (row-percentage of households) Shock Very Somewhat Not Not at negatively negatively much all Drought 87 11 1 1 Heavy rainfall, flooding, untimely rains 69 27 4 1 Unexpected decline in prices 65 30 4 0 Major harvest loss 73 22 5 1 Theft of household assets 62 31 5 2 Unemployment from paid job 66 28 3 2 Loss of livestock (death, theft, illness) 76 21 3 0 Loss of land (reallocation, transfer) 73 25 1 1 Substantial storage loss 44 54 2 0 Major illness not resulting in death 68 28 3 0 Death 83 16 1 0 Total 77 20 3 1 21Unemployment also seems to affect many households, but we disregard it here, since it has only affected 2% of all households. 54 Each comm efforts. This includ The present sectio characteristics. Th malnutrition situat malnutrition are su Health inst and treat illnesses. Subtables A kilometers away, w The outlier here is track to reach an a during and after th Private me accessibility to so distances, for exam treatment. Howeve from the kebeles, s In lack of p Table 42: Access to utilities A. Bore hole for water in kebele hours a B. Public water tap in kebele hours a Boloso Sore 56% 0.4 Boloso Sore 50% 0.3 Damot Woyde 33% 1 Damot Woyde 38% 1.6 Kucha 67% 0.1 Kucha 11% 2.5 Konso 50% 0.0 Konso 25% . Kedida Gamela 30% 1.6 Kedida Gamela 10% 3.2 Uba-Debre Tsehay 90% 0 Uba-Debre Tsehay 10% 3 Burji 56% 0.1 Burji 11% 3.3 Total 55% 0.5 Total 21% 2.2 C. Village well in kebele hours a D. Electricity in kebele hours a Boloso Sore 78% 0 Boloso Sore 10% 0.5 Damot Woyde 50% 0.7 Damot Woyde 0% 3.1 Kucha 50% 0.2 Kucha 0% 5.9 Konso 75% 0.1 Konso 13% 0.3 Kedida Gamela 0% 3.5 Kedida Gamela 20% 1.8 Uba-Debre Tsehay 50% 3.1 Uba-Debre Tsehay 0% 4.6 Burji 44% 0.2 Burji 0% 3.6 Total 48% 1 Total 6% 3.1 aTravel distance by usual mode of transport. BUSINESS INSTITUTIONS Sustainable improvements in peoples' living standard require improvements of their basis of living. In the surveyed rural communities most if not all households are engaged in agriculture, and improvements therefore require the transmission of agricultural knowledge, better input and output markets and better access to capital. For the day-to-day agricultural business the households need to buy inputs for their farming such as seeds or fertilizer. Sales points where the households can obtain inputs are on average 13 km or 2 hours transport22 away from the kebeles (table 43A). This is considerably far, even though most households will use donkeys to transport e.g. fertilizer. New knowledge on agricultural practices is disseminated through a network of agricultural extension agents. These are in general quite accessible, as table 43B shows. If the farmers want to realize new ideas or simply extend their present farming, they will often need some capital to for instance buy additional livestock. Formal credit sources are on average 3 hours away, as table 43C shows. 22The majority (97%) accesses these business institutions by foot. 57 Table 43: Business in A. Sales point for agr Boloso Sore Damot Woyde Kucha Konso Kedida Gamela Uba-Debre Tsehay Burji Total C. Village bank / credit source for 2 or 3 years23 ( funding for schem The Food 150.000 birr per k households intervi need for credit in loans, as over half 32% (out of the investments. The k the households, an households, but als Table 44: Use of cred Use of credit Oxen Other livestock Tools Inputs (fertilizer, etc Petty trade ) Operate business Pay other loan Consumption Social activity Construction source of credit. households, where Table 45: Alternative Credit source Credit association Private lender Neighbor/Friend Cooperative Micro-finance instituti NGO/CBO Other many of the HWP provincial or regio 42% of th professionals, and mainly independen owning a farm an health promoters' the advocacy and c Table 46: Perso The HWP members in gener schooling, see tabl 10 years of school difference probabl work of the health Table 47: Formal edu Education answers are statisti 3% of the HW ( significantly differ Columns 3 and 58 days of trai at all, see columns topics. Table 48: Training o Training topic Table 49: Was the tra Training topic (A) Ante and postnata (B) Child food & nutr (C) Child growth mon (D) Child Care (E) Hygiene (F) Family planning an Table 50: Sufficiency Number of topics, f training was insufficient 0 1 2 3 4 5 6 7 Total A clear ma respectively ­ corr Figure 13. The go who gave 6 month section 7 above. Figure 13: Introducti One basic w before drinking, bu share than the 5% The HWP Between 83 and 87 health promoters ( (BCG vaccination) measles vaccinatio had had their youn of age or 82% of th higher rates of vac are implemented. additional purchas CGPC baseline sur Two-third o in public works o kebeles at least 18 dependent on the b and dry midland ar needed assistance the woredas of K therefore not fully in table 53, where The pair-w works is calculate remuneration for t 28% and 14%, th significant at the 5 Table 55: Free hando % Household Yes received free No food/cash Missin Table 57: Dependenc Total Mean 1.61 St. dev. 0.931 Min 0 Max 10 initiated the Child Security Project (F years of age throug an evaluation of th in July and August of the Southern Na Malnutritio in general in line comparison of measurements are children that are m Arimond, Mary, an Practices" A Nutrition Te Aylward. G.P., Pfei infants publ 658. life. Part 2: Nutrition Bu Huffman, Sandra L. The Dark S Per Pindstru Press, Ithaca Kostermas, Kees (19 Guidelines Washington Toole, M.J.; and Recommend 25. AP A1. SAMPLE SIZE The necessary sample the desired detectable 3) the variance of ma Given that the same c a certain degree of c group, any aggregate will not be perfect. W both the treatment an Cluster sampling is u cluster are likely to d clusters. To take acco is defined as DE = constitute the cluster assume an intra-kebe These selected house below 2 years of age selected households h Example: 1000 ho The kebele list of interviewed, but 117 number between 0 an first household to be to be visited, the 22nd ... , ... , the 996th hou WOREDA KEB Boloso Sore Bom Sha Ada Ham Adi Gos Ham Yeb Uba-Debre Gela Tsehay Zeq Yela She Uba She Qaw APPE Before child younger years ago, for To be filled later MODULE 2: HOUSEHOLD ROSTE 1 2 3 Name Sex (start with household head) 14 15 16 17 Name How long (copy fromhave you lived If moved into k module 2) in this kebele?did [NAME] liv (adults only) (adults only) Always......1 Where Part A: CHILD 1 Name of ch MODULE 4: H 1 PART A: ASSET no. Asset Item 1 Oxen 2 Milk cow 3 Other cattle 9 CROP Teff Maize Barley Wheat Sorghum Pulses (beans, fenugreek) Please translate am Used measurement unit MODULE 5: S MODULE 6: H 1 What is the (circle) 2 What is th (circle) 11 What do (circle) What do you use 12 (circle) 13 Is cooking done 9 Have you received year? What were the sou 10 (Circle one or more) 11 Has any household last 6 month? (circl 12 If so, did he/she rec Has the HH receiv during pregnancy w 12Did the mother of food during pregna 13Does [NAME] hav see it) (circle one) Is the growth card 14 the household or health worker/clini 15How many times in weight or height bee Has [NAME] had 16 tuberculosis, that you seek treatment dWhat treatment received? eIf no, why did y seek treatment? During the illne once a week or no If a child is sick, the household 33whether to take th for diagnosis treatment at the post, clinic, etc.? If you earn mo receive in kind p 34 or gift, who decides how it used? 13 Do you believe that On how many days D each of the following? 14a Milk, other than 14b Fruit juice? 14c Any other liquid or soup broth? 14d Any food made wheat, barely, tef MODULE 10: Ask about child cl 1a Child's name 1b Child's ID-nu Who is 2 caregiver of [NAME]? 3 How long wa house yesterd 14When do you food to childr What causes 15 (Do not probe given.) AP To be a for instance th school or clinic MODULE 2: C 1 How many P 2 How many H 3 How many years? (Circ How many h MODULE 3: I 1A. FACILITY no. Facility- MODULE 4: I 1 Is there a healt 2 When was the 3 Has the im (Health worke Are there any APPEN MODULE 1: I 5 Do you hav 6 If so, how 7 How old is 8 How old is What 9 activities currently (Note one o MODULE 3: K What substa 1 baby? (Do not probe When do y 2 complementa breastmilk)? do not read answ If a general ans specific answer, Africa Region Working P Series # Title ARWPS 1 Progre Manag World ARWPS 2 Towar Develo Repub Africa Region Working P Series # Title ARWPS 13 Confli ARWPS 14 Reform Role Africa ARWPS 15 The Africa Region Working P Series # Title ARWPS 26 What Traditi ARWPS 27 Free T Econo Africa Region Working P Series # Title Indust ARWPS 38 A M Povert With a ARWPS 39 The Im Reduc the Co Macro Africa Region Working P Series # Title ARWPS 49 Rural Ghana and Pe ARWPS 50 Microf Implic Develo Indust Africa Region Working P Series # Title AWPS 59 Rwand Socio- AWPS 60 Linkin Malian AWPS 61 Evolut Ghana Africa Region Working P Series # Title AWPS 69 Tanzan Challe AWPS 70 Tanzan and Enviro AWPS 71 An An 1998 a Policy Africa Region Working P Series # Title AWPS 82 The Frame The C SSA c AWPS 83 Rules for ch the Tra AWPS 84 Africa Region Working P Series # Title AWPS 92 Comm South AWPS 93 The R from S Expan