kenya’s capacity to monitor children’s goals: a medium-term assessment

93
-A- KENYAS CAPACITY TO MONITOR CHILDRENS GOALS: A MEDIUM-TERM ASSESSMENT by John Thinguri Mukui Consultant Report Prepared for UNICEF, Kenya Country Office 14 July 1994

Upload: jtmukui2000

Post on 21-Nov-2015

21 views

Category:

Documents


9 download

DESCRIPTION

The World Summit for Children goals covering the last decade of the twentieth century were translated into the Kenya National Plan of Action (NPA) goals. The report established baseline data on World Summit for Children goals as translated into Kenya NPA goals; assessed the reliability of indicators available together with recommendations on possible improvements in definitions, data collection procedures, and analysis; and evaluated the institutional capacity to collect indicators on the World Summit for Children goals, with special emphasis on mid-decade goals.

TRANSCRIPT

  • -A-

    KENYAS CAPACITY TO MONITOR CHILDRENS GOALS:

    A MEDIUM-TERM ASSESSMENT

    by

    John Thinguri Mukui

    Consultant Report Prepared for UNICEF, Kenya Country Office

    14 July 1994

  • -i-

    TABLE OF CONTENTS

    ACKNOWLEDGEMENTS .................................................................................................................... iii

    ABBREVIATIONS AND ACRONYMS ............................................................................................... iv

    INTRODUCTION .................................................................................................................................. 1

    POVERTY .............................................................................................................................................. 3

    INTRODUCTION ............................................................................................................................................. 3

    MEASURES OF POVERTY .............................................................................................................................. 4

    THE EXTENT, DEPTH AND SEVERITY OF RURAL POVERTY, 1992 ...................................................... 5

    GROWTH IMPLICATIONS OF THE NATIONAL PROGRAM OF ACTION (NPA) GOALS .................... 6

    COMMENTS ON THE NATIONAL HOUSEHOLD WELFARE MONITORING SURVEY, 1992 .............. 6

    THE NATIONAL HOUSEHOLD WELFARE MONITORING AND EVALUATION SYSTEM ................... 7

    NUTRITION ......................................................................................................................................... 10

    MALNUTRITION ........................................................................................................................................... 10

    MICRONUTRIENT DEFICIENCIES ............................................................................................................. 13

    BREASTFEEDING AND THE BABY-FRIENDLY HOSPITALS INITIATIVE ......................................... 14

    CHILD GROWTH MONITORING ............................................................................................................... 16

    HEALTH ............................................................................................................................................... 19

    INFANT AND UNDER-FIVE MORTALITY ................................................................................................ 19

    MATERNAL MORTALITY ............................................................................................................................ 20

    IMMUNIZATION COVERAGE .................................................................................................................... 21

    CONTROL OF DIARRHOEAL MORBIDITY AND MORTALITY ............................................................. 27

    GUINEA WORM DISEASE ........................................................................................................................... 31

    EDUCATION ....................................................................................................................................... 33

    EARLY CHILDHOOD EDUCATION ............................................................................................................ 33

    PRIMARY SCHOOL ENROLMENT AND RETENTION ............................................................................. 35

    LITERACY RATES ......................................................................................................................................... 39

    WATER AND SANITATION.............................................................................................................. 42

    ACCESS TO SAFE DRINKING WATER ....................................................................................................... 42

    ACCESS TO SANITARY MEANS OF EXCRETA DISPOSAL ...................................................................... 43

    ONGOING ACTIVITIES ON WATER AND SANITATION INDICATORS ............................................... 45

    CHILD PROTECTION ........................................................................................................................ 47

    INSTITUTIONS ENGAGED IN PRODUCING PERFORMANCE INDICATORS ........................... 49

    CENTRAL BUREAU OF STATISTICS .......................................................................................................... 49

    HEALTH INFORMATION SYSTEM ............................................................................................................. 53

  • -ii-

    KENYA EXPANDED PROGRAMME ON IMMUNIZATION ..................................................................... 55

    ANNEX 1: TERMS OF REFERENCE .................................................................................................. 58

    ANNEX 2: POLICY MATRIX OF THE WORLD SUMMIT FOR CHILDREN GOALS .................. 60

    ANNEX 3: STATISTICAL ANNEX ..................................................................................................... 65

    ANNEX 4: REFERENCES .................................................................................................................... 80

    ANNEX 5: LIST OF PERSONS INTERVIEWED ............................................................................... 87

    TEXT TABLES

    Text Table 1: Provincial Status of Rural Poverty, 1992 (%)

    Text Table 2: Population Parameters in the KEPI CEIS Computer Program

    Text Table 3: Primary School Gross Enrolment Rates, 1992 (%)

    Text Table 4: Primary School Net Enrolment Rates, 1992 (%)

    Text Table 5: Primary School Age/Grade Mismatch, 1992 (%)

    Text Table 6: Literacy Rates by Region, 1992 (%)

    Text Table 7: Literacy Rates by Region, 1980/81, 1988 and 1989 (%)

    STATISTICAL ANNEX TABLES

    Table 1: Nutritional Status by District, 1987

    Table 2: Nutritional Status by Demographic and Background Characteristics, 1993 (%)

    Table 3: Percentage of Living Children by Breastfeeding Status, 1993

    Table 4: Breastfeeding and Supplementation by Age, 1993 (%)

    Table 5: Median Duration and Frequency of Breastfeeding, 1993 (%)

    Table 6: Baby-Friendly Hospitals Initiative: Status as of June 1994

    Table 7: Infant and Under-Five Mortality Rates per 1,000 Live Births

    Table 8: Percentage of Children 12-23 Months Who Received Specified Vaccines, 1989

    Table 9: National Immunization Coverage for Children Aged 12-23 Months, 1992 (%)

    Table 10: Percentage of Children 12-23 Months Who Received Specified Vaccines, 1993

    Table 11: Prevalence of Diarrhoea and Knowledge and Ever Use of ORS, 1993 (%)

    Table 12: Treatment of Diarrhoea, 1993 (%)

    Table 13: Primary School Retention Rates for 1984, 1985 and 1986 Standard 1 Entrants (%)

    Table 14: Households by Main Source of Water in Wet Season, 1992 (%)

    Table 15: Households by Main Source of Water in Dry Season, 1992 (%)

    Table 16: Households by Type of Toilet, 1992 (%)

    Table 17: Households by Access to Water and Sanitary Facility, 1989 (%)

    Table 18: Households by Access to Water and Sanitary Facility, 1993 (%)

    Table 19: Households by Source of Water and Type of Excreta Disposal for Selected Districts

  • -iii-

    ACKNOWLEDGEMENTS

    I thank all the individuals in Government, international organizations and

    nongovernmental organizations who were extremely helpful in providing ideas that

    are reflected in this report. Generous support and encouragement was received

    from Alfred Okinda (Monitoring and Evaluation Officer, UNICEF/Kenya Country

    Office), Francis Kamondo (Chief, Integrated Community-Based Programmes,

    UNICEF/Kenya Country Office), and Mahesh Patel (Monitoring and Evaluation

    Officer, UNICEF/East and Southern Africa Regional Office). The excellent research

    assistance provided by Rita Achieng Obura is much appreciated.

    John Thinguri Mukui

    Nairobi

    14 July 1994

  • -iv-

    ABBREVIATIONS AND ACRONYMS

    AMREF African Medical and Research Foundation

    ANPPCAN African Network for the Prevention and Protection against Child Abuse and Neglect

    CBR Crude Birth Rate

    CBS Central Bureau of Statistics

    CDC U.S. Centers for Disease Control and Prevention

    CDD Programme for the Control of Diarrhoea Diseases

    CEIS Computerized EPI Information System

    CHANIS Child Health and Nutrition Information System

    CRC Convention on the Rights of the Child

    DICECE District Centre for Early Childhood Education

    FEWS USAID Famine Early Warning System

    GOK Government of Kenya

    HIS Health Information System

    HRSSD Human Resources and Social Services Division, Ministry of Planning and National

    Development

    IDD Iodine Deficiency Disorders

    IGADD Inter-Governmental Authority on Drought and Desertification

    IMR Infant Mortality Rate

    KCPE Kenya Certificate of Primary Examination

    KDHS Kenya Demographic and Health Survey

    KEPI Kenya Expanded Programme on Immunization

    MIS Management Information System

    MMR Maternal Mortality Ratio

    MOH Ministry of Health

    MPND Ministry of Planning and National Development

    MTC Medical Training Centre

    NACECE National Centre for Early Childhood Education

    NASSEP National Sample Survey and Evaluation Programme

    NCHS U.S. National Centre for Health Statistics

    NCPD National Council for Population and Development

    NDVI Normalized Difference Vegetation Index

    NHWMES National Household Welfare Monitoring and Evaluation System

    NPA National Plan of Action

    ODA (British) Overseas Development Administration

    ORS Oral Rehydration Salts

    ORT Oral Rehydration Therapy

    PEM Protein-Energy Malnutrition

    PHC Primary Health Care

    PPA Participatory Poverty Assessment

    PSRI Population Studies and Research Institute, University of Nairobi

    VAD Vitamin A Deficiency

    WHO World Health Organization

    WMS Welfare Monitoring Survey

  • -1-

    CHAPTER ONE

    INTRODUCTION

    1. The World Summit for Children goals covering the last decade of the twentieth century

    were translated into the Kenya National Plan of Action (NPA) goals. Some of the Kenya NPA goals

    were less ambitious than the World Summit goals e.g. on malnutrition, reduction in measles incidence

    and mortality, primary school retention rate, infant and under-five mortality rates, access to antenatal

    care, and reduction in acute respiratory infection mortality rates. Kenya also brought forward the

    target date of achieving the mid-decade goals from 1995 to 1994, and intends to utilize 1995 to

    consolidate the achievements. Kenyas mid-decade goals include (a) adoption of the Dakar consensus

    based on the 1992 International Conference on Assistance to African Children convened by the

    Organisation of African Unity, and (b) enhanced Dakar consensus e.g. Kenya NPA has a higher target

    on measles immunization coverage (90%) than the Dakar consensus (80% coverage for the six antigens

    of the Expanded Programme on Immunization). The GOK/UNICEF 1994-98 programme of operations

    provides support to the World Summit for Children and mid-decade (1995) goals through advocacy

    and direct support. The GOK/UNICEF goals on child survival and development had a profound impact

    on the thrust and the contents of the National Development Plan 1994-96.

    2. The purpose of the activity as spelt out in the Terms of Reference is to carry out an

    assessment of the current availability of relevant data relating to the Child Summit goals in general but

    with special emphasis on the mid-decade goals. The Mission took its broad agenda to include:

    (a) Laying of baseline data on World Summit for Children goals as translated into Kenya

    NPA goals;

    (b) Reliability of the indicators available together with recommendations on possible

    improvements in definitions, data collection procedures, and analysis;

    (c) An evaluation of the institutional capacity to collect indicators on the World Summit

    for Children goals, with special emphasis on mid-decade goals.

    Two supporting World Summit goals under health were (a) give all couples access to family planning

    information and services to enable them plan their families, and (b) give all pregnant women access to

    antenatal care and to safe child birth. Due to time constraints, these two supporting goals were not

    included in the assessment. The interim report was also supposed to be presented in a

    UNICEF-sponsored sub-regional meeting involving three or four other countries in East and Southern

    Africa region, but the meeting did not take place.

    3. The Mission did not take the issue of Kenyas ability to achieve the goals as its main

    concern, but only the ability to monitor its achievements regardless of whether the achievements are

    expected to exceed or fall short of the NPA goals. However, in some cases, the difference between

    ability to monitor and ability to achieve a goal is pedagogical. For example, in the case of eradication of

    the guinea worm disease, the worm is identified and extracted during an active case search, and

    monitoring and achievement are basically a simultaneous process. In most cases, a program with a

    high degree of accomplishment of its objectives gives incentives to the implementing personnel to

  • -2-

    leave footprints of their activities, thereby laying a better factual basis for analyzing the program. In

    addition, program monitoring allows implementers to identify weaknesses and take remedial actions,

    thus improving on both reporting and performance.

    4. The report is divided into three main sections. Section I (chapters 2 to 7) discusses each

    indicator, its definitional issues, baseline data available or being collected, and recommendations on

    improvements. The second section (chapter 8) analyzes institutional capacity of the main agencies

    involved in collection of indicators pertaining or coincidental to the Kenya NPA goals (Central Bureau

    of Statistics, Health Information System in the Ministry of Health, and the Kenya Expanded

    Programme on Immunization) to ascertain their institutional strengths and weaknesses, and

    recommendations on possible restructuring and financing arrangements to enable them to collect the

    requisite indicators. The third section is the Statistical Annex showing the recent baseline data

    available for each indicator.

  • -3-

    CHAPTER TWO

    POVERTY

    INTRODUCTION

    5. The World Summit for Children underscored the need to alleviate poverty and revitalize

    economic growth as the foundation for achieving the goals on child survival and development. The

    Kenya Governments National Plan of Action (NPA) operationalized the concept by specifying targets

    to reduce the overall incidence of poverty by 50% of 1993 levels by the year 2000; and depth and

    severity of poverty by one-third.

    6. The Central Bureau of Statistics undertook the National Household Welfare Monitoring

    Survey in December 1992 and January 1993, funded by the World Bank under the Health

    Rehabilitation Project (World Bank, 1991b), to fill the lacuna in poverty statistics in Kenya.

    Consequently, the Government has produced (a) a basic report on the 1981-82 Rural Household

    Budget Survey (Barasa and Wakanyora, 1994); (b) a basic report on the 1992 National Welfare

    Monitoring Survey (Ayako, 1994); and (c) poverty profiles for rural 1981-82 and urban and rural

    Kenya 1992 (Mukui, 1994). Chapter 3 of the Economic Survey 1994 gave a synopsis of the poverty statistics derived using the 1992 National Household Welfare Monitoring Survey data. However, the

    1982-83 Urban Household Budget Survey (UHBS) database could not be used since the data provided

    was in aggregated form, and did not include key variables on household characteristics (e.g. household

    members age and educational characteristics) and expenditure on key food items, mainly maize and

    bread. The loss of the 1982-83 UHBS data underscores the need to refocus attention on the

    development of appropriate data archival systems in the Central Bureau of Statistics.

    7. The Human Resources and Social Services Division (HRSSD) in the Ministry of Planning

    and National Development (MPND) undertook a Participatory Poverty Assessment (PPA) during

    February-April 1994. The purpose of the PPA was to understand poverty as seen by the poor in order

    to complement quantitative studies of poverty. The study elicited rare donor coordination in an

    exercise, as it was sponsored by the British Overseas Development Administration (ODA), the

    principal technical coordinator was from the World Bank, and field coordination was undertaken by

    the African Medical and Research Foundation (AMREF). The PPA covered communities in seven

    poor rural districts (Bomet, Busia, Kisumu, Kitui, Kwale, Nyamira and Mandera) and Mathare Valley

    in urban Nairobi. The activity in Kisumu and Mandera districts was sponsored by UNICEF/Kenya

    Country Office. The analytical reports for each district and the report summarizing the main lessons

    and conclusions from the entire PPA survey were ready by June 1994.

    8. The first National Welfare Monitoring Survey (WMS1) was a priority survey whose main

    objectives were the identification of policy target groups and the production of key socioeconomic

    indicators describing the wellbeing of different groups. The primary purpose of the Welfare

    Monitoring Survey was to gauge the present and future net socioeconomic consequences of economic

    management and structural adjustment in Kenya. The design of the survey was to draw on the

    experience of the Kisumu Household Welfare Monitory and Evaluation Survey (Kenya, 1990d) as well

    as the World Banks Social Dimensions of Adjustment Priority Survey (Grootaert and Marchant, 1991;

    World Bank, 1991c).

  • -4-

    9. WMS1 collected data in 44 districts, excluding Turkana, Marsabit and Samburu. Data for

    North Eastern province was obtained from urban clusters only, and its results do not therefore

    represent rural areas of the province. The questionnaires were intended to capture information on

    household characteristics, household expenditures, household incomes, assets and amenities owned

    and available to the households, and land utilization.

    MEASURES OF POVERTY

    10. The degree of poverty depends on the incidence of poverty (numbers in the total population below the poverty line), the intensity of poverty (the extent to which the incomes of the poor lie below the poverty line), and the degree of inequality among the poor. In addition to reflecting the three dimensions, a poverty index should also be decomposable among sectors and socioeconomic

    groups. A summary measure which meets the four requirements is that of Foster, Greer and

    Thorbecke (1984). If real expenditures or income are ranked as follows:

    Y1 Y2 ... Yq < z < Yq+1 ... Yn

    where z is the poverty line, n is the total population, and q is the number of poor, the FGT measure is:

    P = (1/n)[(z - Yi)/z]; 0.

    The poverty measure takes the proportional shortfall of income for each poor person [(z - Yi)/z], raises

    it to a power () which reflects societies concern about the depth of poverty, takes the sum of these

    over all poor households, and normalizes by the population size.

    11. The parameter is a policy parameter that reflects concern for the poor; as increases

    greater weight is attached to the poverty gap of the poorest. The main measures in the poverty study

    were (a) the head-count index (=0), which measures the prevalence of poverty and is insensitive to

    how far below the poverty line each poor unit is; (b) the income-gap ratio (=1), the average of the

    poverty gaps expressed as a fraction of the poverty line; and (c) =2, which gives the severity of

    poverty. The head-count index (H) simply shows the proportion of people below the poverty line.

    However, the income-gap ratio (HI) takes into account both the incidence of poverty (H) and its

    intensity (I). The sum of the poverty gaps is the total income required to eliminate poverty.

    12. There is a conceptual problem in the definition of gender of household head that needs to

    be highlighted in the measurement of poverty, since Governments NPA goals includes reduction in

    poverty disparities between female-headed and male-headed households. As Clark (1985) points out,

    conjugal structures in Kenya evades easy data collection and analysis due to biased identification of

    households, heads of households and women heads of households. The standard definition of the

    household assumes that (a) the physical boundaries of the household define units of social and

    economic organization (thereby ignoring economic exchanges between households), and (b) the

    household is a basic decision-making unit behaving according to the rule of household utility (thereby

    ignoring intra-household inequality in resource allocation based on age and gender). It is assumed that

    head of household and the primary breadwinner is a male, while women rather than men are

    socially recognized as primary providers for their children through their efforts in subsistence

  • -5-

    agriculture. Frequently, woman-headed households are identified on the basis of the absence of a male

    spouse in the household. There is need to break down woman-headed households into (a) de facto female household heads defined by the temporary but long-term absence of a male spouse in the

    household; and (b) de jure female household heads identified by lack of adult male/spouse in the household. The aggregation of data into a single category of woman-headed households masks the

    levels of poverty faced by de jure woman-headed households (single, separated, divorced).

    THE EXTENT, DEPTH AND SEVERITY OF RURAL POVERTY, 1992

    13. The overall prevalence of rural absolute poverty was 46.3% by adult equivalents1 and

    41.5% by households. The depth of poverty was 18.4%, while the overall severity of poverty was 9.8%.

    The prevalence of poverty was highest in Western (54.8%), followed by Rift Valley (51.5%), Nyanza

    (47.4%), Coast (43.5%), Eastern (42.2%) and Central (35.9%). Some of the districts with the lowest

    prevalence of poverty were Kajiado/Narok (25.1%), Kiambu (32.7%), Meru/Tharaka (32.7%), Laikipia

    (34.4%), Nyeri (35.4%), Nyandarua (36.7%) and Muranga (37.3%). The prevalence of poverty was

    highest in Busia (67.7%), West Pokot (65.2%) and Kericho/Bomet (64.7%). The depth of poverty was

    highest in West Pokot (35.4%) and Busia (33.3%).

    Text Table 1: Provincial Status of Rural Poverty, 1992 (%)

    All Coast Eastern Central Rift Valley Nyanza Western

    Absolute Poverty Line

    Prevalence (ad eq) 46.3 43.5 42.2 35.9 51.5 47.4 54.8

    Prevalence (HHs) 41.5 37.9 38.1 31.2 44.5 43.4 53.5

    Depth (ad eq) 18.4 15.4 14.9 12.1 22.3 19.7 23.0

    Severity (ad eq) 9.8 7.6 7.4 5.4 12.7 10.6 12.6

    Absolute Hard Core Poverty Line

    Prevalence (ad eq) 37.4 32.8 32.2 28.1 42.9 39.1 45.4

    Prevalence (HHs) 32.8 27.4 29.1 24.2 36.2 34.8 42.9

    Depth (ad eq) 13.7 10.9 10.5 8.1 17.4 15.1 17.6

    Severity (ad eq) 7.0 5.2 5.1 3.4 9.5 7.6 9.2

    Prevalence of Absolute Poverty by Household Head (ad eq)

    Total 46.3

    Male 45.6

    Female 48.4

    Male-married 45.7

    Male-other 44.3

    Female-married 44.6

    Female-other 52.9

    Note: Ad eq means that the statistics are in adult equivalents, while HHs stands for households.

    1 Equivalence scales are deflators used to convert household real expenditures into money metric utility

    measures of individual welfare, mainly based on child costs and economies of scale in consumption (saving on

    consumption by living together versus living apart). This procedure derives directly from Engels (1895)

    pioneering work using a single good (food), although there is no reason why the model cannot be applied more

    generally to other goods e.g. adult goods (tobacco, alcohol and adult clothes) see Prais and Houthakker, 1955;

    Rothbarth, 1943; Working, 1943; Deaton, 1986; Deaton and Muellbauer, 1986; and Lewbel, 1989.

  • -6-

    14. Some of the variables used to delineate socioeconomic groups were sex of household head,

    education level of the head, household size, and age of household head. The sex of the head of the

    household did not appear to be a significant factor in the determination of the incidence of rural

    absolute poverty. However, female-married headed households had prevalence of poverty of 44.6%

    at the absolute poverty line, compared with 52.9% for female-other. The depth and severity of

    poverty were also lower in female-married compared with female-other households.

    15. Using education level of the household head, the lowest prevalence of absolute poverty was

    among heads of households with secondary education (26.7%), compared with primary education

    (45.5%) and no education (57.4%). Poverty consistently increases with household size. Poverty

    measures using size of land holding did not portray any clear trend, probably because of different

    agricultural potential of land holdings. Poverty increases with age of household head.

    GROWTH IMPLICATIONS OF THE NATIONAL PROGRAM OF ACTION (NPA) GOALS

    16. The issue of whether WMS1 poverty statistics will be used as the baseline will probably be

    settled when the next round of welfare monitoring survey data collected in June 1994 is analyzed. This

    is because the Government used the lessons of the first survey to improve the design of the

    questionnaire and enumerators reference manual for the second survey. However, given the overall

    rural poverty indicators for 1992, the NPA targets imply that the overall incidence of poverty would

    be 23.2% in the year 2000, while the overall depth would be 12.3%, and severity 6.5%. Since the

    poverty statistics are insensitive to how far above the poverty line a non-poor household is, the

    economic growth implications of the NPA targets are difficult to quantify. Based on the analysis of the

    1992 WMS1 data, the targets imply that the national rural depth and severity of poverty in the year

    2000 would be the same as those for Kiambu and Nyeri districts of Central province in 1992, while the

    targeted incidence of poverty (23.2%) would be far much below those of the latter districts in 1992

    (around 33%). The NPA targets for the poverty statistics are not feasible given the current economic

    stagnation fuelled by bad weather, standoff with the donor community, and the quality of economic

    management.

    COMMENTS ON THE NATIONAL HOUSEHOLD WELFARE MONITORING SURVEY, 1992

    17. In general, the enumerators reference manual was brief, and it is difficult to know whether

    the trainers clarified the issues to enumerators during training. There were also inadequacies in

    definitions of, say, main economic status that could permit generation of meaningful socioeconomic

    groups.

    18. In relation to crop income, the crops were not identified by name, and it was therefore not

    possible to compute total household consumption of, say, maize and its products, since information on

    maize purchases was available but itemized consumption of own-produce was not.

    19. The questionnaire put the analytical burden of the survey data on the respondents and

    enumerators. The enumerator or the respondent was left to determine what is an export crop, while

    export crops are also consumed locally, and the respondent may not know whether his/her cash crop is

    ultimately exported.

  • -7-

    20. The 1992 WMS1 was conducted only a few weeks prior to the Christmas festivities and the

    first multiparty elections since Independence. Christmas is normally associated with above-average

    expenditure on some items e.g. luxurious foods and items of clothing and footwear. The euphoria of

    the elections would also affect the responses, but the direction of the bias is indeterminate.

    21. A survey design which involves consecutive visits to the same household is said to be bounded if the recall is based on the period since my last visit. Under this definition, the reference periods

    (last week, last month, last year) used in the 1992 WMS1 were not bounded, which can lead to serious

    telescoping (mis-dating) errors. Telescoping errors are likely to increase with the length of the recall

    period.

    22. A problematic issue is the comparability of data on food crop consumption from own

    production with food purchases. The recall period for food crop own consumption was for long

    season and short season, and both components were added up in the analysis to derive total food

    crop consumption. The error from the unbounded recall periods described in terms of seasons is likely

    to be higher than for calendar-defined recall periods e.g. last week or last month.

    23. The 1992 WMS1 survey period was characterized by unstable and rising commodity prices,

    which implies that the prevailing prices last week and last month for the same commodity were

    different. In addition, price variations by regions during the survey period were high and atypical,

    mainly due to shortages of key commodities like sugar and maize. This factor complicates the

    interpretation of shares in consumption of items collected under different recall periods.

    24. The changes in district boundaries and the number of districts have necessitated updating of

    the national sample frame since districts are supposed to be treated as distinct strata. The creation of a

    new district entails transfer of some households from a stratum. If a dry area within a predominantly

    arable region was made an independent stratum, the original district might register a spurious

    improvement in household welfare due to removal of the poorer households. The creation of new

    districts will make district-specific inter-temporal poverty profiles less meaningful.

    25. Some of the data from the 1992 Welfare Monitoring Survey, especially on total income and by

    its components, could not be meaningfully used in the preparation of poverty profiles and in

    establishing socioeconomic groups. The survey was limited by the brevity of the enumerators

    reference manual, which was not particularly useful in clarifying concepts. In terms of survey

    organization, the initial steps of preparing an analysis plan for the survey, dummy tables of the most

    important data from the survey, and a specification of data needs for development of poverty profiles,

    do not appear to have been prepared before the survey was launched. The eventual authority on the

    quality of survey data was difficult to identify.

    THE NATIONAL HOUSEHOLD WELFARE MONITORING AND EVALUATION SYSTEM

    26. The National Household Welfare Monitoring and Evaluation System (NHWMES) is supported

    by the World Bank under the Health Rehabilitation Project and is coordinated by the Human

    Resources and Social Services Division in the Ministry of Planning and National Development. The

    NHWMES also included a component of review of available CBS field staff, and the recruitment

  • -8-

    necessary so as to include North Eastern province in the national sample frame.

    27. One of the problems encountered in the implementation of the project has been the inability

    to use the project funds due to the superimposition with Government procurement procedures. The

    inaccessibility of donor funds has also been aggravated by Governments expenditure ceilings. Donors

    and Government ought to explore ways of speeding up disbursement procedures, and for activities

    (e.g. technical assistance for data analysis and report preparation) to be funded directly or through

    trust funds deposited with nongovernmental organizations.

    28. The first round of the Welfare Monitoring Survey received comments from the analysts who

    prepared the Basic Report (Ayako, 1994) and the Poverty Profiles (Mukui, 1994). The second round of

    the Welfare Monitoring Survey strengthened the core questionnaire through clarification of concepts

    in the enumerators reference manual and introduced more questions to make the interpretation of

    data more meaningful.

    29. Some improvements in questionnaire design were the inclusion of proper definitions of types

    of farmers, and itemization of food own-consumption expenditure. The survey also includes (a) an

    anthropometry module; (b) fertility questions for all females aged 12 and above (including

    immunization for Tetanus Toxoid), and use of family planning methods for females aged 12-49; (c)

    child survival module (attendance in growth monitoring, immunization, breastfeeding, and the type

    of personnel who assisted in child delivery); and (d) maternal mortality as related to pregnancy and

    child birth. The anthropometry module will be the fifth rural nutrition survey and the first national

    nutrition survey undertaken by the Central Bureau of Statistics since independence. The core

    questionnaire and modules have common serial numbers of the household members, thereby making

    it possible to interlink all the data collected in the second round of the Welfare Monitoring Survey.

    30. The survey will also fill data gaps in water and sanitation. Some of the improvements in water

    and sanitation questions are (a) reliability of water source, and (b) time taken to fetch water so as to

    ascertain the opportunity cost of not providing water close to homesteads and as a rough guide to the

    corresponding female drudgery.

    31. The NHWMES will be the main source of poverty statistics. Given the number of modules

    tagged on to the WMS2, it will be necessary to prepare a detailed analysis plan and coordination in

    data processing. One of the constraints will be that some modules are not financed by the World Bank,

    but by other agencies e.g. UNICEF. There will therefore be a tendency by the funding agencies to

    exercise territorial rights to some modules at the initial stages.

    32. Steps taken to improve on WMS2, based on the lessons from WMS1, include (a) improved

    questionnaire and enumerators reference manual, (b) stricter supervision of fieldwork, and (c)

    preparation of data edit and consistency checks during data entry using a dedicated data-edit computer

    package (Integrated Microcomputer Processing System, IMPS, developed by the U.S. Bureau of the

    Census). Data entry and cleaning is expected to be done during July-August, 1994. The issue of

    whether the baseline poverty statistics will be based on WMS1 or WMS2 need therefore to be

    postponed until the WMS2 database is analyzed.

    33. Although CBS has sufficient number of computers for data entry, there is a shortage of

  • -9-

    computers for merging of data files after data entry is completed. CBS therefore needs at least two

    high-memory, 486-capacity personal computers for data merge and analysis. Arrangements are under

    way for UNICEF to provide two computers to CBS for that purpose. It will also be necessary for the

    National Household Welfare Monitoring and Evaluation project to build CBS human resource

    capacity through formal training and workshops, so as to provide them with new tools of analysis and

    to enable them supervise consultants handling various aspects of the analysis more effectively.

  • -10-

    CHAPTER THREE

    NUTRITION

    MALNUTRITION

    Introduction

    34. The mid-decade goal is to reduce the 1990 levels of severe and moderate malnutrition among

    under five years of age by one-fifth (20 percent) or more. Kenya accepted that goal. However, before

    evaluating Kenyas achievement on the goals, it is important to review the working definitions for

    measuring the goal and the scope of existing data to provide baseline information on the requisite

    goals.

    35. The indicators of Protein-Energy Malnutrition (PEM) are underweight (moderate and severe), underweight (severe), stunting (moderate and severe) and stunting (severe). Optional indicators are wasting (moderate and severe) and wasting (severe). All the definitions are based on the U.S. National Centre for Health Statistics (NCHS) and accepted by the U.S. Centers for Disease Control and

    Prevention (CDC), which uses Standard Deviation (SD) scores from the median of the reference

    population. The recent sources of indicators of PEM malnutrition that will be discussed are the Fourth Rural Child Nutrition Survey, 1987 (NS4), and Kenya Demographic and Health Survey, 1993.

    36. Underweight is measured by the proportion of children under five years of age falling below minus 2 standard deviations from the median weight-for-age of the reference population (moderate

    and severe) and below minus three standard deviations (severe). Stunting is measured by the proportion of children under five years of age falling below minus 2 standard deviations from the

    median height-for-age of the reference population (moderate and severe) and below minus three

    standard deviations (severe). Wasting is measured by the proportion of children under five years of age falling below minus 2 standard deviations from the median weight-for-height of the reference

    population (moderate and severe) and below minus three standard deviations (severe).

    37. However, statistical treatment of anthropometric indicators of malnutrition can be based on

    either Z-Scores or Percentages of the Reference Median. Z-scores are presented in terms of predefined

    standard deviations below the median of the reference population, while percentage scores refer to the

    proportion below a predefined percentage of the median score e.g. 80 percent of the median

    weight-for-height of the reference population. The two measures do not generate identical indices of

    malnutrition, even if the analyst adjusts the cutoff points.

    Nutrition Surveys

    38. The NS4 was based on the national sample frame and covered all districts except North Eastern

    province, and Isiolo, Marsabit, Samburu, Turkana, Lamu, Tana River and West Pokot districts. The

    NS4 indicators were based on the growth reference curves of American children from the NCHS/CDC,

    as studies have shown that children from well-to-do families in Kenya have growth patterns similar to

    those of American children (Alnwick, 1980). In all, 6,909 children aged 6-60 months were included in

    the analysis. The NS4 survey report cautions that (a) there was high non-response in some areas e.g.

  • -11-

    Kajiado and Narok districts due to nomadic nature of the population; and (b) supervision of

    enumerators slackened in some districts, which may have affected the quality of nutritional indicators

    particularly height-for-age index.

    39. The NS4 considered children with height-for-age median of less than 90 percent of the

    reference population to be stunted, and children with weight-for-height of less than 80 percent as

    wasted. District-level Z-scores are also presented for those below minus 2 standard deviations of the

    height-for-age (moderate and severe stunting) and below minus 2 standard deviations of

    weight-for-height (moderate and severe wasting). Analysis of anthropometric data to generate

    indicators of underweight (weight-for-age) was not done in the last two nutrition surveys, and recent

    national indicators of underweight based on the nutrition surveys do not therefore exist. Kenya

    presently uses only stunting and wasting in their reporting on nutritional status. Underweight has

    been recommended as the indicator for monitoring the World Summit goal for reduction in

    malnutrition (de Graft-Johnson, 1991)

    40. Comparison of data from different nutrition surveys conducted in Kenya is limited by a

    number of factors. First, the surveys covered different age groups. For example, the target population

    for the 1977 rural nutrition survey was 1-4 years, 6-60 months for the 1978-79 survey, and 3-60

    months for the 1982 survey. Second, all the surveys exclude North Eastern province and some districts

    of Eastern and Rift Valley provinces, which may have higher percentages of malnourished children.

    Third, the surveys do not cover a whole year and therefore do not allow for estimation of seasonal

    factors on malnutrition. This factor is compounded by the fact that the surveys were carried out at

    different times of the year and comparisons between them (especially on indicators of wasting since

    wasting uses data on weight) is distorted by seasonal factors. Fourth, the first two surveys used the

    Harvard growth curves, while the two latter surveys used the NCHS/CDC/WHO reference

    population. Fifth, the results were analyzed using percentage rather than standard deviation (Z)

    scores. The 1987 survey was analyzed using both methods although the text of the report is based on

    percentage scores. In the case of the 1987 survey, the height-for-age cutoff point of 90 percent is

    considered too high, as it includes children who may have been ill or temporarily malnourished but

    who may have later caught up. A lower cutoff point of, say, 80 percent would only reflect children

    who may suffer functional impairment due to malnutrition.

    41. The district-level indicators of nutritional indicators, namely, moderate and severe stunting

    (below minus 2 standard deviations height-for-weight) and moderate and severe wasting (below

    minus 2 standard deviations weight-for-height) are presented in Statistical Annex Table 1. The 1987

    rural national moderate and severe stunting was estimated at 32.2 percent, with Kilifi (51.7 percent),

    Kwale (56.1 percent) and Narok (59.7 percent) having over 50 percent stunted children. The 1987

    rural national estimate of moderate and severe wasting was estimated at 4.5 percent, the highest rates

    being in Siaya (11.7 percent), Kajiado (9.9 percent), Laikipia (8.2 percent) and South Nyanza (7.8

    percent) among the surveyed districts. Provincial estimates of stunting and wasting based on Z-scores

    were not published.

    Kenya Demographic and Health Survey, 1993

    42. The 1989 Kenya Demographic and Health Survey did not contain an anthropometric module. However, the 1993 KDHS obtained data on weight and height of all children born since January 1988

  • -12-

    (i.e. under five years of age since the survey was conducted during February-August 1993) and whose

    mothers were interviewed in the KDHS. The anthropometric data, combined with a childs age, was

    used to derive height-for-age (stunting), weight-for-height (wasting) and weight-for-age

    (underweight) indicators for children of 1-60 months of age. The survey used the international

    reference population defined by the U.S. NCHS/CDC and recommended by the World Health

    Organization (WHO).

    43. Overall, 32.7 percent of Kenyan children were classified as stunted and 12.2 percent as

    severely stunted. Stunting was highest among children of 12-23 months (40.3 percent). Stunting was

    more prevalent among rural children (34.2 percent) than urban children (21.5 percent). The

    proportion of stunted children was highest in Coast province (41.3 percent) and Eastern (39.4

    percent), and lowest in Nairobi (24.2 percent). The 1993 KDHS states that the high levels of stunting at

    the Coast have been observed in previous nutrition surveys, but the appearance of Eastern province in

    the league of high stunting regions may be associated with the recent drought conditions.

    44. In 1987, the proportion of rural children who were reported as moderately and severely

    stunted was 32.2 percent, which is roughly the same estimate for rural Kenya based on the 1993 KDHS

    (34.2 percent). However, region-specific estimates are not comparable since the 1987 NS4 reported

    Z-scores nutrition indicators by district while the 1993 KDHS reported by province. In comparing

    1987 and 1993 estimates, it should be noted that (a) NS4 sample excluded Lamu, Tana River, and West

    Pokot districts, which were included in the 1993 KDHS; and (b) the 1993 KDHS purposely

    oversampled 15 districts so as to produce reliable estimates for the oversampled districts without

    expanding the total sample to unmanageable levels. The 1987 indicators of stunting and wasting for

    the districts oversampled in the 1993 KDHS were not uniformly below or above the national averages,

    and the effect of oversampling on the national estimates of stunting and wasting in 1993 is therefore

    indeterminate.

    45. An estimated 5.9 percent of the Kenyan children were wasted, and 1.2 percent severely

    wasted. Variations in wasting by demographic characteristics show that it was highest for children of

    12-23 months of age (10.0 percent). The highest prevalence of wasting was recorded in Coast province

    (10.6 percent), followed by Rift Valley (7.9 percent) and Eastern (6.8 percent). The 1987 estimate of

    moderate and severe wasting was 4.5 percent in rural Kenya, compared with 6.0 percent in rural 1993.

    46. The 1993 KDHS reported that 22.3 percent of the Kenyan children under-five years of age

    were moderately and severely underweight for their age, and 5.7 percent severely underweight. As

    with other anthropometric indicators, underweight was highest among children of 12-23 months

    (31.6 percent), the period which is characterized by weaning - gradual termination of breastfeeding -

    and highest incidence of diarrhoea. The prevalence of underweight children was higher among

    children in rural areas (23.5 percent) than urban areas (12.8 percent). Children in Coast (31.7 percent)

    and Eastern (28.8 percent) are much more likely to be underweight than children from other

    provinces. Prevalence of underweight was not reported in the 1987 NS4 Basic Report, and no

    comparisons can therefore be made for reference periods 1987 and 1993. However, in interpreting

    1989 and 1993 KDHS statistics on various indicators included in this report, it is important to keep in

    mind that observations for urban centers, excluding Nairobi, are included in the respective provinces.

    In addition, the 1993 KDHS included children aged 1-60 months while the 1987 NS4 covered 6-60

    months, which affects comparability of the indicators based on the two surveys since children below

  • -13-

    six months are less likely to be stunted than older age cohorts.

    Baseline Indicators of Nutritional Status

    47. Since the indicators in the 1987 NS4 do not meet the full reporting requirements of UNICEF

    under the World Summit for Childrens goals, UNICEF could explore the possibility of requesting CBS

    to provide tabulations for underweight using the Z-scores, and provincial estimates of stunting and

    wasting. The 1993 KDHS probably presents the best baseline source of nutritional status for the 1990s.

    However, UNICEF should request the National Council for Population and Development to provide

    tabulations for 1993 district-level estimates of stunting, wasting and underweight, at least for the 15

    districts which were purposely oversampled to produce reliable district-level estimates. Such request

    could also include information on immunization, and other indicators of concern to UNICEF, but

    which do not appear in the 1993 KDHS analytical report.

    48. The second round of the National Household Welfare Monitoring Survey (June 1994) contains

    an anthropometry module financed by UNICEF. UNICEF therefore has proprietary access to the

    anthropometric data immediately it is collected and keyed in. The initial efforts will, of course, be to

    produce a basic report on anthropometric indicators to satisfy the immediate reporting needs under

    the World Summit for Children goals. However, since all the modules in the Welfare Monitoring

    Survey have common household identification information and household members serial numbers,

    interrelationship between various household socioeconomic characteristics can be generated. UNICEF

    should encourage coordinated analysis of the modules in the WMS2 database, to ensure that timely

    analysis and dissemination of information is made possible.

    MICRONUTRIENT DEFICIENCIES

    49. The World Summit goals in regard to micronutrient deficiencies are (a) to reduce iron

    deficiency anemia in women by one third of the 1990 level, (a) to virtually eliminate iodine deficiency

    disorders (IDD), and (c) to virtually eliminate Vitamin A deficiency (VAD) and its consequences,

    including blindness. On nutritional anemia, the Kenya NPA stated that, since the prevalence of

    nutritional anemia among women in Kenya is unknown, the Government will undertake or

    commission studies to determine the prevalence of nutritional anemia among women and children.

    On IDD and VAD, the Government was to undertake nationwide surveys to determine their

    prevalence.

    50. The Government has already taken a number of administrative and statutory steps with

    respect to reduction of IDD. In 1988, the Government amended The Food, Drugs and Chemical Substances Act (Cap. 254 of the Laws of Kenya) restricting the sale of table salt which does not contain the required potassium iodate. Meetings have already been conducted with salt manufacturers on the

    restriction of sale of iodized salt, and easier channels of manufacturers access to potassium iodate have

    been created. The Ministry of Healths public health officers and Kenya Bureau of Standards

    inspectors conduct spot-checks in salt manufacturing firms and on salt sold in the market. However,

    the supervision needs to be strengthened by sensitizing extension officers on the usefulness of doing

    more field and on-site (i.e. salt manufacturing factories) spot-checks, while the Ministry of Health also

    needs to supply more kits for such spot-checks.

  • -14-

    51. The public health education system has not been as successful on the need for iodine in the

    salt, and the systems of handling table salt in wholesale and retail outlets, during transportation, and in

    the households, to prevent loss of iodine in iodated salt. For example, partial loss of iodine occurs if the

    salt is put too early during cooking, and kept in open containers or in damp places. A costless activity

    would be to sensitize the mass media, so that the media can cover the issues of the consequences of

    iodine deficiency and handling of salt as part of their normal reporting and/or features.

    52. The Food, Drugs and Chemical Substances Act states that table salt or salt for general household use shall contain 168.5 mg. per kilogram of potassium iodate. However, the wording of the

    legislation is vague since it is supposed to refer to 168.5 mg. of potassium iodate per kilogram of salt

    rather than 168.5 mg. per kilogram of potassium iodate.

    53. The Nutrition section in the Division of Family Health, with financial support from UNICEF,

    started a micronutrient survey in February 1994. The main objectives of the survey are to (a)

    determine the prevalence of Vitamin A, iodine and iron deficiencies in Kenya, (b) determine the

    possible causes of these deficiencies where they occur, (c) determine the geographical distribution of

    these deficiencies, and (d) identify the groups at risk. The survey is being carried out in 49 districts in

    the republic. The survey entails clinical assessments, and interviewing for food intake frequency and

    available food resources. The survey results are expected to be ready by August 1994.

    BREASTFEEDING AND THE BABY-FRIENDLY HOSPITALS INITIATIVE

    Situation Analysis

    54. The Kenya Fertility Survey, 1977-78, conducted under the World Fertility Survey Programme, solicited information on full breastfeeding and breastfeeding (not necessarily exclusive breastfeeding) for the penultimate pregnancy prior to the survey. The national mean period of full

    breastfeeding was estimated at 3.5 months. However, the statistics have not been widely used as

    baseline information on Kenyas infant feeding practices since they included the latest offsprings

    regardless of the offsprings age at the time of the survey. Indeed, the period between the penultimate

    pregnancy and interview was 37 months and over for 65 percent of the index offspring. In addition, a

    quarter of the mothers interviewed were over 40 years of age at the time of the survey. The

    breastfeeding statistics in the Kenya Fertility Survey, 1977-78 can not therefore be associated with any specific time period.

    55. In 1982, the Central Bureau of Statistics undertook a survey of Infant Feeding Practices on 980 low and middle income Nairobi women who had given birth in the previous 18 months. Sampling was

    done using the CBS national sample frame based on the 1979 population census. The survey showed a

    common pattern of almost universal successful and prolonged breastfeeding overlaid with widespread

    supplementation with infant formula in the first six months of life (Kenya, 1984a). The report noted

    that negative results of this unnecessary use of breast milk substitutes include a drain on family

    income, shorter intervals between births due to lactational amenorrhea, and increased child

    morbidity. The correlation between breastfeeding and fertility was considered important because only

    18 percent of the women surveyed reported using any form of birth control since the birth of the

    index child.

  • -15-

    56. The survey included questions on influences on feeding practices. While 77 percent of the

    women surveyed had given birth in a health facility, only 14 percent recalled receiving any

    information on infant feeding at that time, and half of this 14 percent reported being wrongly told that

    exclusive infant formula feeding was best for the child.

    57. Other surveys which have collected household-based information on breastfeeding includes

    the four nutrition surveys (1977, 1978-79, 1982 and 1987) and the 1989 and 1993 Kenya Demographic and Health Surveys. The focus of our analysis will only be based on the Fourth Rural Child Nutrition Survey 1987 and the 1989 and 1993 Kenya Demographic and Health Surveys.

    58. The fourth rural nutrition survey (1987) solicited responses on (a) whether the child was

    breastfeeding at the time of the survey, (b) number of months breast-fed, and (c) whether the child

    had breast-fed in the two weeks preceding the survey. The survey did not solicit responses on the use

    of breast-milk substitutes and infant formula. The published survey results only reported on district-

    and national-level estimates of (a) still breastfeeding without giving the age cohorts, and (b) average

    length of any breastfeeding. The mean breastfeeding period was 16.1 months, with the lowest

    recorded in Narok (13.2 months) and the highest in Meru (18.7 months) and Laikipia (18.8 months).

    59. The 1989 KDHS solicited responses on (a) whether the child was breastfeeding at the time of

    the survey, (b) whether the child was exclusively breastfeeding, (c) age of child when mother stopped

    breastfeeding, and (d) the use of itemized breast-milk substitutes and infant formula. The published

    results refer to the number of months of breastfeeding (not necessarily exclusively). The mean number

    of months of any breastfeeding was 19.5 in the rural areas (compared with 16.1 months obtained from

    the 1987 rural nutrition survey) and 18.8 months in urban areas. The Nairobi mean was 19.9 months

    in 1993.

    60. The 1993 KDHS questions on breastfeeding were largely similar to those in the 1989 KDHS,

    but solicited additional information on the age of child when breast-milk substitutes and infant

    formula, as well as other types of food were introduced on a regular basis. The survey showed that the

    mean duration of breastfeeding, not necessarily exclusive, was 21.1 months, 19.6 months in urban

    areas and 21.5 months in rural areas. The mean length of exclusive breastfeeding was 0.5 months and

    0.7 months for exclusive breastfeeding and plain water only.

    61. The 1993 KDHS asked mothers about breastfeeding status of all last-born children under five

    and, if the child was being breast-fed, and whether various types of liquids or solid food had been

    given to the child yesterday or last night. Children who are exclusively breast-fed receive breast milk only, while those who are fully breast-fed include those who are exclusively breast-fed and those who receive plain water in addition to breast-milk. The results are shown in Statistical Annex Tables

    3, 4 and 5.

    62. In the first month of life, only 26.8 percent of the children were exclusively breast-fed during

    the 1993 survey period, while an additional 17.7 percent breast-fed in addition to plain water only.

    The overall duration of exclusive breastfeeding in Kenya was 0.5 months, while those mothers with no

    education registered a mean of 0.6 months of exclusive breastfeeding. The mean number of months of

    fully breastfeeding (breastfeeding and plain water only) was 0.7 months, and was highest in Central

    (1.4 months) and Western (1.0 months) among the provinces and among mothers with no education

  • -16-

    (1.0 months) when tabulated against mothers highest grade reached. The type of personnel who

    assisted in child delivery did not appear to influence subsequent breastfeeding practices.

    Assessment of Baby-Friendliness

    63. In the mid-decade goals, Kenyas aim is that by the end of 1994, all the circa 300 facilities

    providing maternity services will be practicing the 14 steps of the National Policy on Infant Feeding

    Practices. According to the Situation Analysis (1992), Kenyas Minister for Health officially launched the Baby Friendly Hospitals Initiative in September 1991, making Kenya one of the first 12 countries

    attempting to achieve Baby Friendliness by December 1992. By the end of 1994, it is anticipated that

    all the estimated 300 maternity facilities will be declared Baby Friendly on a global rating.

    64. The Baby-Friendly Hospitals initiative aims to discourage the use of commercial baby foods; to

    mobilize hospital personnel and health professionals (especially those involved in maternity hospitals

    and wards) to support breastfeeding; and to create among women a demand for hospitals which are

    optimally supportive of mothers wishing to breastfeed. By March 1992, two of Kenyas leading health

    institutions, Kenyatta National Hospital and Pumwani Maternity Hospital, were formally declared

    Baby Friendly on a global rating.

    65. The Division of Family Health in the Ministry of Health, with support of UNICEF, carries out

    random checks on facilities providing maternity services. A detailed questionnaire is filled using

    responses from the hospital administration, healthcare providers and mothers in the maternity wards.

    The data is analyzed in the HIS section of the Ministry of Health using SPSS (Statistical Package for the

    Social Sciences) software to ascertain the baby-friendliness of the maternity facility. As of June 1994,

    17 maternity facilities had been declared baby-friendly by the Division of Family Health. Some health

    institutions have not been reported as baby-friendly since they have not been inspected by the

    Division of Family Health. An assessment of baby-friendliness of maternity facilities was being

    conducted in July 1994. However, since the number of maternity cases per health facility is not

    reported, the success of the baby-friendly hospitals initiative may be understated since the initial

    impetus has been on health facilities with relatively higher than average number of child births. There

    is therefore need to collect data on children born in each maternal facility for a common base year,

    say, 1993, so as to measure the success of baby-friendliness by both the number of facilities and

    maternity cases handled on average.

    66. The main risk in the continuous assessment of baby-friendliness is lack of funds to enable the

    Division of Family Health to make frequent visits to the maternity facilities. There is therefore need to

    focus on budgetary issues, both from the Government budget lines and donors, to prevent a relapse

    due to infrequent supervision. In addition, although the promotion of breastfeeding is primarily the

    responsibility of nurses, the doctors strike in Government hospitals may have led to a slight relapse in

    the baby-friendly hospitals initiative, due to the increased responsibilities on the nurses.

    CHILD GROWTH MONITORING

    67. The Health Information System (HIS) has until recently been processing data on child growth

    monitoring through the Child Health and Nutrition Information System (CHANIS), but this

    responsibility has now been transferred to the Division of Family Health. Health facilities fill daily

  • -17-

    tally sheets for all children-weighing, broken down by age (0-11, 12-35, 36-59 months) and whether

    the child is normal weight or underweight based on the standard child growth reference curve.

    Monthly facility totals are filled in Chanis 3 forms, which are forwarded to the districts. The districts

    compile monthly district totals in Chanis 5 forms, which are then forwarded to the Division of

    Family Health.

    68. The Chanis system has been affected by delayed reporting from health facilities. At the facility

    level, the data captures child growth monitoring based on the US National Centre for Health Statistics

    (NCHS) standard child growth reference curve. Those children whose weight-for-age fall below the

    third percentile of the distribution of the standard reference population are classified as underweight

    while those above the third percentile are classified as normal. However, the summary returns

    capture only cross-sectional nutrition status rather than growth monitoring since the data is not by

    individual child in the growth process. In addition, since the data does not distinguish between first

    visit and revisits, the summarized data refer to children-weighing rather than children-weighed. Over

    a period of, say, one year, different children will have been weighed different number of times. It is

    therefore difficult to set up quantitative targets based on population parameters or to formulate

    appropriate indicators of coverage of growth monitoring for the target age groups.

    69. Conventional child growth monitoring as an indicator of nutritional status is usually defended

    on the grounds that, a childs revisit for reweighing is a fresh case since its weight is read against a

    different age on the standard growth reference curve. However, severely malnourished cases are more

    likely to be referred for reweighing; thereby leading to a downward bias in the reported overall

    nutritional status derived using child growth monitoring data.

    70. The data from Chanis was reportedly used as a food deficit early warning system, and the

    information is therefore shared with other ministries (e.g. agriculture) and the regional

    Inter-Governmental Authority on Drought and Desertification (IGADD). IGADD also gives financial

    and technical support to the Division of Family Healths child growth monitoring programme. Chanis

    data is graphed with the Normalized Difference Vegetation Index (NDVI) to study the linkages

    between the two datasets. The NDVI analyzes remote sensing measurements and assesses whether the

    target being observed contains live green vegetation, and has been used to estimate crop yields, pasture

    performance, and rangeland carrying capacities, among others. There was a proposal to link the

    Chanis database with the USAID Famine Early Warning System project (FEWS) geographical

    referencing system. No substantial progress was made by either USAID/FEWS or the Division of

    Family Health due to (a) the fact that Chanis database was at district level rather than clinic-based

    while NDVI are generated at more disaggregated levels; and (b) weight-for-age is more of an outcome

    of poverty/food deficit rather than a food deficit early warning indicator. A link between the Chanis

    database (on weight-for-age and the proposed sentinel-based weight-for-height indicators) and

    USAID/FEWS geographical mapping system could be of great benefit to both the Division of Family

    Health and USAID/FEWS.

    71. The Chanis information is used at health facilities to follow up on the progress of the

    individual child especially for severely malnourished cases. Underweight (weight-for-age) as an

    anthropometric indicator is affected more by seasonal factors e.g. harvests, food shortages and

    temporary child sickness, compared with, say, stunting (height-for-age). However, data on incidence

    of kwashiorkor and marasmus, which is submitted on the same Chanis facility and district returns, are

  • -18-

    more commonly used since they reflect severe malnutrition that can lead to functional impairment of

    the distressed child.

    72. Chanis has developed new forms to gather data on wasting (weight-for-height) in some pilot

    districts, namely, Busia, Nyamira, Siaya, Kitui, Garissa, Laikipia, Kajiado, Turkana, Embu, Muranga,

    Kilifi and Kwale districts. The data will be collected from four sentinel sites in each district, i.e. health

    facilities with good reporting rates and a high turnover of children. However, there is a shortage of

    equipment, especially wall charts and length boards. Data from such sentinel sites will be more easily

    mapped against NDVI, compared with district-level summaries of nutritional status (weight-for-age)

    currently entered in the Chanis information system.

    73. The Chanis was moved from HIS in the Ministry of Health headquarters to the Division of

    Family Health in early 1994. The section has only one member of staff (a medical records officer) who

    has previously received three-year training at the Medical Training Centre (MTC). The computing

    facilities are one desktop PC, two laptop computers and an Epson dot-matrix printer, which are also

    shared with secretaries for word processing. The desktop was an old computer which was not

    functioning at the time of the Missions visit, and data entry was being done in one of the laptops. It is

    necessary to provide a high-memory computer to handle data on child-growth monitoring and other

    aspects of data analysis in the nutrition section in the Division of Family Health e.g. questionnaires on

    baby-friendly hospitals initiative. The Chanis section did not have any data archival systems other

    than diskettes. Loss of data in case of computer systems failure is therefore a real danger.

  • -19-

    CHAPTER FOUR

    HEALTH

    INFANT AND UNDER-FIVE MORTALITY

    74. Infant mortality is the probability of dying before the first birthday, and under-five mortality

    is the probability of dying between birth and the fifth birthday. Both rates are expressed per 1,000 live

    births. The World Summit Goal in respect of infant mortality is to reduce the rate by one third or to 50

    per 1,000 live births, whichever is less; and to reduce under-five mortality by one third or to 70 per

    1,000 live births, whichever is less. The Kenya NPA accepts the World Summit goals on infant and

    under-five mortality, but cautions that achieving these reductions, however, could be adversely

    affected by the additional mortalities due to AIDS.

    75. The sources of baseline information on the two indicators are the 1977 Kenya Fertility Survey, the 1969 and 1979 Population Censuses, and the 1989 and 1993 Kenya Demographic and Health Surveys. The estimates of under-five mortality rates derived from the 1979 population census are based on mothers aged 15-49 years at the time of the census, and do not therefore reflect a defined

    time period. The Central Bureau of Statistics has not completed analysis of infant and under-five

    mortality rates based on the 1989 Population and Housing Census.

    76. The 1989 KDHS collected under-five mortality data based on retrospective birth history, in

    which data was collected from respondents aged 15-49 years. The sources of data collection errors

    would include underreporting of events, misreporting of age at death, and misreporting of date of

    birth. The national infant mortality rate for the period 1979-89 was 58.6; 56.8 for urban areas and 58.9

    for rural Kenya. Coast (107.3), Nyanza (94.2) and Western (74.6) had infant mortality rates above the

    national average, while Rift Valley (34.6), Central (37.4), Eastern (43.1) and Nairobi (46.3) had lower

    infant mortality rates than the national average.

    77. The regional under-five mortality differentials for the period 1979-89 based on the 1989

    KDHS followed the same pattern as the infant mortality rates. The national under-five mortality rate

    was estimated at 90.9, with Coast (156.0), Nyanza (148.5) and Western (132.8) being above, and

    Central (47.0), Rift Valley (50.9), Eastern (64.3) and Nairobi (80.4) being below the national average.

    This means that, for the reference period, almost 10 percent of live births did not live to see their fifth

    birthday.

    78. The 1993 KDHS solicited information on under-five mortality. The results show an overall

    increase in infant mortality rate from 58.6 for the period 1979-89 (based on the 1989 KDHS) to 62.5 for

    1983-93 reference period (based on the 1993 KDHS). Although the reference periods that the

    estimates refer to overlap, analysis for the 1988-93 reference period also depicts an increase in infant

    mortality in Kenya. Infant mortality rates for Nyanza (127.9), Coast (68.3) and Western (63.5) were

    above the national average, while Central (30.9), Nairobi (44.4), Rift Valley (44.8) and Eastern (47.4)

    were below the national average.

    79. Under-five mortality rates for the reference period 1983-93 based on the 1993 KDHS also

    depict similar regional differentials as infant mortality. While the national average under-five

  • -20-

    mortality rate for the reference period 1983-93 was 93.2, Nyanza (186.8), Western (109.6) and Coast

    (108.7) were above the national average, while Central (41.3), Rift Valley (60.7), Eastern (65.9) and

    Nairobi (82.1) were below the national average. The trend for Nyanza province is disturbing as it

    shows that about 19 percent of live births do not live to see their fifth birthday, and the rate is almost

    twice that of the second highest (Coast province, 10.9 percent) and four and a half times that of the

    lowest rate (Central, 4.1 percent). The regional differentials of both infant and under-five mortality

    followed the same pattern in both the 1989 and 1993 KDHS, with Coast, Nyanza and Western

    provinces being above, and Eastern, Central, Rift Valley and Nairobi being below the respective

    national averages.

    MATERNAL MORTALITY

    80. Maternal mortality is defined as the death of a woman while pregnant or within 42 days (6

    weeks) of a termination of pregnancy, irrespective of the duration and site of the pregnancy, from any

    cause related to or aggravated by the pregnancy or its management but not from accidental or

    incidental causes. Maternal deaths are essentially the product of two factors: the risk of mortality

    associated with a single pregnancy or a single live birth, and the number of pregnancies or births that

    are experienced by women of reproductive age (fertility). Maternal mortality ratio (MMR) is defined

    as the number of maternal deaths in a population divided by the number of live births. The World

    Summit goal is to reduce maternal mortality ratio by one half between 1990 and the year 2000. The

    Kenya National Plan of Action states that, since no reliable estimates of maternal mortality are

    currently available, the immediate plan was to determine the current rate of maternal mortality.

    81. There are only a few, facility-based, estimates of maternal mortality in Kenya (Boerma, 1987a;

    Boerma, 1987b; Boerma and Baya, 1990; Boerma and Mati, 1989; Ewbank, Henin and Kekovole, 1986;

    Makokha, 1980; Makokha, 1991; Ngoka and Bansal, 1987; Kenya and UNICEF, 1992a). A 1980 study of

    maternal deaths at Kenyatta National Hospital yielded an MMR of 224 per 100,000 live births at that

    hospital between 1972 and 1977 and 320 during 1978-1987 (Makokha, 1980; Makokha, 1991); while

    MMR at Pumwani Maternity Hospital was much lower at 67.2 deaths per 100,000 live births over the

    period 1975-84 (Ngoka and Bansal, 1987). A third study in Kwale district of coastal Kenya led to an

    estimate of maternal mortality of 254 per 100,000 live births. The much higher MMR rate for Kenyatta

    National Hospital compared with Pumwani Maternity Hospital may be due to the fact that Kenyatta

    National Hospital is a referral hospital and would, on average, see more difficult cases.

    82. The Population Studies and Research Institute (PSRI), with financial support from UNICEF, is

    conducting a Kenya Maternal Mortality Baseline Survey (KMMBS) during June-July 1994. The survey

    covers all districts except seven -- the three districts in North Eastern Province and four other

    northern districts (Isiolo and Marsabit in upper Eastern province and Samburu and Turkana in Rift

    Valley province). The excluded areas account for only about 5 percent of Kenyas population. The

    survey utilized the National Sample Survey and Evaluation Programme (NASSEP) master frame

    maintained by the Central Bureau of Statistics. To obtain reliable district-level estimates, ten districts

    were oversampled: Homa-Bay, Nyamira, Kisumu, Busia, Baringo, Nyeri, Embu, Kitui, Kwale and Taita

    Taveta. The baseline survey has been designed to (a) collect information on maternal mortality, (b)

    estimate levels of maternal mortality at the national level and for the ten oversampled districts, (c)

    establish regional maternal mortality differentials, (d) determine socioeconomic, socio-cultural and

    demographic factors that influence maternal mortality, and (e) determine the effect of maternal

  • -21-

    mortality on child survival.

    83. The survey will consist of three modules. The estimation of maternal mortality will be

    through the sisterhood method, i.e. asking questions about respondents sisters experience (whether

    alive or dead at the time of the survey), such that each interviewee becomes a respondent for several

    sisters. At the end of the survey, the survey organizers expect to have interviewed 40,000 respondents.

    The information being collected includes mortality experience of all his/her sisters who ever reached

    menarche (the time in a girls life when menstruation first begins), deaths due to pregnancy/child

    birth, and survival and health status of the index child in case of maternal deaths.

    84. The second source of information is the study of maternal admissions in all the district

    hospitals during the period January-December 1993, while the third source is selected participants for

    Focus Group Discussions from local communities. For the Focus Group Discussions, a total of 5 groups,

    consisting of medical personnel, local women of reproductive age, traditional birth attendants and

    other community-based supervisors will be involved in each of the three districts: Homa Bay, Embu

    and Kwale. The Focus Group Discussions will focus on socio-cultural beliefs and practices relating to

    pregnancy/child birth, marriage and nutrition as they relate to maternal health/mortality.

    85. Although the survey is being conducted alongside the CBS National Household Welfare

    Monitoring Survey, data collection is the responsibility of enumerators selected and supervised by

    PSRI; and data entry, analysis and report writing will also be done by PSRI. This is probably the first

    national survey using the CBS sample frame where data collection has been undertaken by a research

    institute. The quality of data collected in the ongoing maternal mortality survey will therefore have

    ramifications on the future distribution of responsibilities for national survey data collection activities

    in Kenya. The report of the survey findings is expected to be ready by November 1994.

    IMMUNIZATION COVERAGE

    Kenyan Goals on Immunization Coverage

    86. The World Summit Goals relating to immunization include (a) eradication of polio by the year

    2000; (b) elimination of neonatal tetanus by the year 1995; (c) reduction in measles deaths and measles

    cases for children under five years of age by 95 percent and 90 percent, respectively, compared to

    pre-immunization levels, by 1995; and (d) maintain at least 90 percent immunization coverage against

    DPT (diphtheria, pertussis, tetanus), measles, polio and tuberculosis among under one year of age and

    against neonatal tetanus.

    87. The Kenya NPA goals on immunization are: (a) achieve virtual eradication of polio by the year

    2000, (b) elimination of neonatal tetanus by the late 1990s, and (c) achieve national immunization

    coverage rate of 90 percent by the year 2000. The mid-decade goal expected to be achieved by the end

    of 1994 is to raise immunization coverage against measles, polio, DPT and TB to 90 percent.

    88. The focus of analysis is on data sources for statistics on immunization coverage, mainly routine

    records from immunization sites, and occasional surveys conducted by KEPI and the National Council

    for Population Development/Macro International Inc (Kenya Demographic and Health Surveys).

  • -22-

    KEPI Routine Reporting from Immunization Sites

    89. The information from the districts for the six antigens in the Expanded Programme on

    Immunization (EPI) is entered in the Computerized EPI Information System (CEIS) program provided

    by the World Health Organization (WHO). The KEPI Management Information System (MIS) moved

    from Version 4 to Version 5 of the CEIS program in 1993. However, the system was not operating

    properly for a number of reasons:

    (a) From the dates of the program files in the system, the Version 5 of the program was not

    installed but copied on Version 4, thereby leaving some batch (*.bat) files of Version 4.

    (b) The specification of the minimum number of files in the CONFIG.SYS was far below the 55

    required in the CEIS Version 5 manual.

    90. The program is installed in a 286, 104 Megabytes, IBM-compatible computer, whereas the

    WHO recommends a minimum of a 386 computer. The program for printing graphs is Harvard

    Graphics Version 2.3. There is therefore need to provide a 386 computer as required by the CEIS

    programme developers, install Version 5 batch files in the program, and install a more up-to-date

    version of Harvard Graphics. There is also no data archival system in KEPI other than floppy diskettes,

    and the data in the computer is not available in software form outside of the room where the computer

    is located.

    91. Information on immunization for the six antigens provides the numerators in the computation

    of immunization coverage. The denominators are calculated from baseline information fed into the

    computer on population by district for a baseline date, crude birth rate per 1,000 (CBR), population

    growth rate, and infant mortality rate per 1,000 live births. The region-specific population parameters

    are based on the 1979 Population Census, rather than the 1989 Population Census. However, the delay

    to use the more recent census results was attributed to increase in the number of districts due to

    subdivision of existing districts. Since the 1989 Population Census, the first group of districts created

    was Machakos/Makueni, Kakamega/Vihiga, Homa-Bay/Migori, Kericho/Bomet, Kisii/Nyamira and

    Meru/Tharaka-Nithi in 1992. The more recent divisions comprised Meru/Nyambene, Kitui/Mwingi,

    Bungoma/Mt. Elgon and Migori/Kuria in 1993. This is a problem that has wreaked havoc to data

    systems which assume population parameters (e.g. water and sanitation coverage) and morbidity and

    mortality statistics from administrative records as percentage of total target population.

    92. Text Table 2 below shows the 1979 population, population growth rate, crude birth rate, and

    infant mortality rate parameters fed into the CEIS program. Also included in Text Table 2 are the 1989

    population projections based on the above population parameters, and data from the 1989 Population

    Census. In the case of the 1979 baseline data, one problem with use of district-specific data is that the

    total census figure of 15.327 million was adjusted to 16.141 million to compensate for

    under-enumeration in Nyanza province, while the KEPI program uses the unadjusted Nyanza

    province population. In addition, the assumption on the annual population growth rate was

    unreasonably low for Turkana district (-0.20 percent), compared with an actual compound growth

    rate of 2.56 percent for the period 1979-89 based on the 1979 and 1989 censuses.

    93. The use of inappropriate population denominators imply that the reported immunization

    coverage is low if the population figure used is more than actual. In the case of the new districts,

  • -23-

    population estimates of the mother districts have not been amended, while reporting on

    immunization is per district. The mother district therefore shows lower immunization coverage than

    actual, while the new district does not have population parameters and no immunization coverage

    statistics are therefore computed. The Central Bureau of Statistics has not provided user ministries

    with information on population for the new districts, probably because the new districts were hived

    off after the 1989 Population Census. The Central Bureau of Statistics has also not published

    information on crude birth rate and infant mortality rate by region based on the 1989 Population

    Census.

    94. One of the factors behind Kenyas miracle in experiencing an accelerated demographic

    transition may be the validity of the population adjustment in 1979 in respect of Nyanza province. For

    example, the compound growth rate in Kenyas population during 1979-89 was 3.415 percent using

    the unadjusted total population, compared with 2.881 percent using the adjusted total. The Central

    Bureau of Statistics should clarify the baseline information that should be used in interpreting the

    recent demographic transition, and in revising the data series for Nyanza province statistics on, say,

    immunization coverage from administrative records. The published population data based on the 1989

    Population Census also seem high for some districts, and could therefore cause a downward bias on

    immunization coverage statistics for the respective districts. For example, the 1979-89 intercensal

    population growth rates were over 5 percent for Isiolo (5.01 percent), Kajiado (5.67 percent), Narok

    (6.61 percent) and Baringo (5.49 percent) districts. Since the abovementioned districts are not urban

    settlements, such phenomenal growth in population is unlikely to have resulted from net

    in-migration.

    95. The denominator problem has been aggravated by displacement of people in some parts of

    the country, mainly Rift Valley province. Information on immunization coverage based on data from

    immunization sites might give a false impression of deterioration in the areas where the people have

    been displaced (due to reduction in the actual denominator), and an improvement in their

    destination districts. Although this is likely to be a more serious source of error for data from

    immunization sites, the data based on the NASSEP frame might also be biased since NASSEP is based

    on the 1989 Population Census. The denominator problem will also be aggravated by the AIDS

    scourge before the next population census is undertaken and analyzed.

    96. There may also be an inherent numerator problem in apportioning facility-based

    immunization data by districts, as the data analysis is based on catchment area rather than

    catchment population. An immunization facility in, say, Thika municipality may serve two

    contiguous districts (Muranga and Kiambu) - the catchment population - while its immunization

    coverage data is posted to Kiambu district - the official catchment area - since Thika municipality is

    physically located in Kiambu district.

  • -24-

    Text Table 2: Population Parameters in the KEPI CEIS Computer Program

    KEPI PROGRAM 1989

    population

    projection

    (mn)

    1989

    population

    Actual (mn)

    1989

    projection/

    actual (ratio)

    Annual

    growth rate

    1979-89 (%) 1979

    population

    (mn)

    Growth

    rate (%)

    CBR IMR

    Nairobi 0.828 5.17 34 92 1.370 1.325 1.03 4.81

    Kiambu 0.686 4.00 56 92 1.016 0.914 1.11 2.92

    Kirinyaga 0.291 3.65 56 92 0.417 0.392 1.07 3.01

    Muranga 0.648 4.05 56 92 0.964 0.858 1.12 2.85

    Nyandarua 0.233 3.55 56 92 0.331 0.345 0.96 4.02

    Nyeri 0.486 3.65 56 92 0.696 0.607 1.15 2.25

    Kilifi 0.431 3.92 41 92 0.633 0.592 1.07 3.22

    Kwale 0.288 3.94 41 92 0.424 0.383 1.11 2.89

    Lamu 0