WATER AND SANITATION PROGRAM: TECHNICAL PAPER 58007 Global Scaling Up Handwashing Project Scaling Up Handwashing Behavior: Findings from the Impact Evaluation Baseline Survey in Vietnam Claire Chase and Quy-Toan Do November 2010 The Water and Sanitation Program is a multi-donor partnership administered by the World Bank to support poor people in obtaining affordable, safe, and sustainable access to water and sanitation services. Claire Chase Water and Sanitation Program Quy-Toan Do The World Bank Global Scaling Up Handwashing is a Water and Sanitation (WSP) project focused on applying innovative behavior change approaches to improve handwashing with soap behavior among women of reproductive age (ages 15­49) and primary school-age children (ages 5­9). It is being implemented by local and national governments with technical support from WSP in four countries: Peru, Senegal, Tanzania, and Vietnam. For more information, please visit www.wsp.org/ scalinguphandwashing. This Technical Paper is one in a series of knowledge products designed to showcase project findings, assessments, and lessons learned in the Global Scaling Up Handwashing Project. This paper is conceived as a work in progress to encourage the exchange of ideas about development issues. For more information, please email Claire Chase at wsp@worldbank.org or visit www.wsp.org. WSP is a multi-donor partnership created in 1978 and administered by the World Bank to support poor people in obtaining affordable, safe, and sustainable access to water and sanitation services. WSP's donors include Australia, Austria, Canada, Denmark, Finland, France, the Bill & Melinda Gates Foundation, Ireland, Luxembourg, Netherlands, Norway, Sweden, Switzerland, United Kingdom, United States, and the World Bank. WSP reports are published to communicate the results of WSP's work to the development community. Some sources cited may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed herein are entirely those of the author and should not be attributed to the World Bank or its affiliated organizations, or to members of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The map was produced by the Map Design Unit of the World Bank. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank Group concerning the legal status of any territory, or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to wsp@worldbank.org. WSP encourages the dissemination of its work and will normally grant permission promptly. For more information, please visit www.wsp.org. © 2011 Water and Sanitation Program Global Scaling Up Handwashing Project Scaling Up Handwashing Behavior: Findings from the Impact Evaluation Baseline Survey in Vietnam Claire Chase and Quy-Toan Do November 2010 Acknowledgements An integral component of the Water and Sanitation The Vietnam impact evaluation also benefits from continuous Program's Global Scaling Up Handwashing Project, a cross- support from Eduardo Perez, the global task team leader for country impact evaluation (IE) study is being conducted in the handwashing project; Nga Kim Nguyen, country task Peru, Senegal, Tanzania, and Vietnam. The World Bank's manager for the handwashing project in Vietnam; Minh Thi Water and Sanitation Program (WSP) Global Impact Hien Nguyen, country monitoring & evaluation officer; and Evaluation Team in Washington, DC, leads the study, with the global technical team comprised of Hnin Hnin Pyne, the contribution of WSP teams and consultants in each of Jacqueline Devine, Nathaniel Paynter, and the Water and the participating countries. The baseline data collection for Sanitation Program support staff. all countries was conducted during 2008 and 2009, and the The baseline survey was conducted by the National reports have undergone several peer review processes. Institute of Hygiene and Epidemiology in Hanoi with The handwashing project's Global Impact Evaluation Team management oversight from Dr. Tham Chi Dung, acting oversees the impact evaluation design, methodology, and chief, under the overall direction of Dr. Nguyen Tran Hien, country teams. It is led by Bertha Briceno (in its early stages director. A cadre of survey enumerators at the provincial, the Global IE was led by Jack Molyneaux), together with district, and commune administrative levels provided Alexandra Orsola-Vidal and Claire Chase. Professor Paul support. Photographs courtesy of WSP, Claire Chase, and Gertler has provided guidance and advice throughout the Tham Chi Dung. project. Global IE experts also include Sebastian Galiani, Finally, we wish to express our sincere gratitude to all the Jack Colford, Ben Arnold, Pavani Ram, Lia Fernald, survey respondents for their generous donation of time and Patricia Kariger, Paul Wassenich, Mark Sobsey, and participation in this study. Christine Stauber. At the country level, the Vietnam Impact Evaluation Team, led by principal investigator Claire Chase with advisory assistance of Quy-Toan Do, manages the in- country design, field activities, and data analysis. Executive Summary Background the intervention and to track changes in key outcomes that In December 2006, in response to the preventable threats can be causally attributed to the intervention. posed by poor sanitation and hygiene, the Water and Sani- tation Program (WSP) launched Global Scaling Up Hand- Vietnam Intervention washing and Global Scaling Up Rural Sanitation1 to In Vietnam, the handwashing project is carried out in 540 improve the health and welfare outcomes for millions of communes across 56 districts in 10 provinces. Underway poor people. Local and national governments implement since 2006, Phase 1 of the intervention has reached a total these large-scale projects with technical support from WSP. of 1.8 million people. Phase 2 of the intervention aims to reach an additional 30 million people through interpersonal Handwashing with soap at critical times--such as after communication (IPC), community marketing events, and contact with feces and before handling food--has been mass media, and is being evaluated through a randomized- shown to substantially reduce the incidence of diarrhea. It controlled impact evaluation. reduces health risks even when families do not have access to basic sanitation and water supply. Despite this benefit, This technical paper describes the baseline findings from rates of handwashing with soap at critical times are very low Vietnam, and is part of a series of technical reports sum- throughout the developing world. marizing baseline findings from similar surveys conducted in each of the Scaling Up project countries. Global Scaling Up Handwashing aims to test whether hand- washing with soap behavior can be generated and sustained Methodology and Design among the poor and vulnerable using innovative promo- The Vietnam Scaling Up Handwashing IE baseline survey tional approaches. The goal of Global Scaling Up Handwash- collected information from a representative sample of the ing is to reduce the risk of diarrhea and therefore increase population targeted by the intervention. The survey was household productivity by stimulating and sustaining the be- conducted between September and November 2009 in a havior of handwashing with soap at critical times in the lives total of 3,150 households containing 3,751 children of 5.4 million people in Peru, Senegal, Tanzania, and Viet- under the age of five. The survey results provide informa- nam, where the project has been implemented to date. tion on the characteristics of household members, access to handwashing facilities, handwashing behavior, preva- In an effort to induce improved handwashing behavior, the lence of child diseases such as diarrhea and respiratory in- intervention borrows from both commercial and social fection, and child growth and development. In addition, marketing fields. This entails the design of communications community questionnaires were conducted with key in- campaigns and messages likely to bring about desired be- formants at the village level in all sample locations to havior changes and delivering them strategically so that the gather information on community access to transporta- target audiences are "surrounded" by handwashing promo- tion; commerce; health and education facilities, and other tion via multiple channels. relevant infrastructure; contemporaneous health and de- velopment interventions; and environmental and health One of the handwashing project's global objectives is to shocks. learn about and document the long-term health and welfare impacts of the project intervention. To measure magnitude Summary of Findings of these impacts, the project is implementing a random- Handwashing behavior ized-controlled impact evaluation (IE) in each of the four The baseline findings in Vietnam in regards to handwashing countries to establish causal linkages between the interven- behavior suggest that there is still a need to improve hand- tion and key outcomes. The IE uses household surveys to washing with soap practices in the target population, par- gather data on characteristics of the population exposed to ticularly among the poorest. Some of the key times during which handwashing should take place are not at the top of 1 For more information on Global Scaling Up Rural Sanitation, see www.wsp.org/ the mind for caretakers of young children, since less than scalingupsanitation. one-third reported handwashing with soap after cleaning a www.wsp.org v Findings from the Impact Evaluation Baseline Survey in Vietnam Executive Summary child's bottom and before cooking or preparing food, and among children under five of 11.0 percent and 6.8 percent just around one-third before feeding children. While a little respectively, the findings in relation to caregiver-reported over 80 percent of households have a place for handwashing diarrhea for this sample of children under five is around with soap and water present, the poorest households are 1.0 percent. Similarly, caregiver-reported ALRI prevalence 23 percent less likely to have access to a place for handwash- is just 0.7 percent. Importantly, these caregiver-reported ing. Moreover, the place for handwashing is more often lo- illness symptoms are internally consistent with the child cated inside the toilet facility or food preparation area in growth measures and anemia prevalence found in the sam- wealthier households (55.1 percent) as opposed to the poor- ple population, both of which provide more objective mea- est (10.0 percent). The handwashing place was observed to sures of child health than caregiver-reported diarrhea and be more than three meters from the toilet or food prepara- respiratory illness. tion area in 31.6 percent of the poorest households. Despite these positive findings, there are still key differences Water and soap were generally available in the households found in child health outcomes by household wealth status, sampled, creating a suitable environment for improved with the poor being consistently worse off. Nearly one-fifth handwashing behavior. In 98.0 percent of households, of the children under two in the sample are stunted in the water was observed at the place used for washing hands poorest households, and over 10 percent are malnourished after going to the toilet, and at least one type of soap was in the two lowest wealth quintiles. Moreover, children from present at the place for washing hands in close to 94 percent households in the lowest wealth quintile exhibit lower of households. The type of soap most commonly found in weight-for-age (­0.90 SDs lower than median) and length- the household regardless of wealth was powdered soap, such for-age (­0.96 SDs lower than median) on average. Finally, as laundry soap or detergent, and an average of 61.9 percent presence of anemia as measured by hemoglobin concentra- of households had this type of soap present at the place in- tion is 31.7 percent in all children sampled, while it is dicated for washing hands. slightly higher at 35.5 percent in the lowest wealth quintile, suggesting an inverse association between anemia and Child health and development household wealth. Over the past decade Vietnam has made significant strides in poverty reduction and is on track to achieve nearly all of The structure of this report proceeds as follows: In Chap- the Millennium Development Goals (MDGs) by 2015, in ter 1 we provide an overview of the Global Scaling Up particular those relating to child undernutrition.2 This Handwashing and Global Scaling Up Rural Sanitation progress is reflected in the baseline findings presented here, projects, as well as background on the handwashing proj- where indicators of child health are largely positive and ect in Vietnam. Chapter 2 details the methodology that indicative of an overall healthy child population. Whereas underlies the impact evaluation, and provides details on estimates from the 2002 Vietnam Demographic and the sampling design, sample selection, and field work pro- Health Survey and third round of the 2006 Multiple Indi- tocols. The baseline findings for general household charac- cator Cluster Survey reported prevalence of diarrhea teristics, handwashing behavior, child health, and child growth are presented in depth in Chapter 3. In Chapter 4 2 United Nations Development Program. 2010. Achieving the Millennium Development Goals in an Era of Global Uncertainty: Asia-Pacific Regional Report 2009/10. Bangkok, we conclude with a summary of the next steps of the im- Thailand: United Nations. pact evaluation study. vi Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Abbreviations and Acronyms Abbreviations and Acronyms ALRI Acute Lower Respiratory Infection C Control DCC Direct Consumer Contact Hb Hemoglobin HH(s) Household(s) HW Handwashing HWWS Handwashing with Soap IE Impact Evaluation IPC Interpersonal Communication IV Intravenous Fluid Injection M&E Monitoring and Evaluation MICS Multiple Indicator Cluster Survey MDG Millennium Development Goals NGO Nongovernmental Organization NIHE National Institute of Hygiene and Epidemiology ORS Oral Rehydration Solution PCA Principal Components Analysis T1 Treatment 1 T2 Treatment 2 USD United States Dollars VND Vietnamese Dong VNDHS Vietnam Demographic and Health Survey VWU Vietnam Women's Union WHO World Health Organization WSP Water and Sanitation Program www.wsp.org vii Contents Executive Summary................................................................... v Abbreviations and Acronyms ................................................... vii I. Overview .................................................................................... 1 1.1 Introduction ....................................................................... 1 1.2 Project Background .......................................................... 2 1.3 Project Components ......................................................... 2 1.4 Objectives of the Study ..................................................... 4 II. Methodology .............................................................................. 5 2.1 Randomization .................................................................. 5 2.2 Study Design..................................................................... 5 2.3 Sampling Strategy and Sample Size .................................. 6 2.4 Variables for Data Analysis................................................. 8 2.5 Instruments for Data Collection ......................................... 8 2.6 Field Protocols ................................................................ 10 III. Findings ................................................................................... 11 3.1 General Household Characteristics ................................. 11 3.2 Handwashing Behavior ................................................... 18 3.3 Diarrhea, Acute Lower Respiratory Infection, and Anemia Prevalence ................................................... 24 3.4 Child Growth Measures ................................................... 28 IV. Conclusion ............................................................................... 35 References ............................................................................... 36 Annexes Annex 1: Communes Selected for Handwashing Project IE Sample .................................................................................. 37 Annex 2: Baseline Comparison of Means Tests for Balance ....... 43 Annex 3: Comparison between WSP IE Baseline Survey and VNDHS Survey ................................................................... 51 Figures 1: Vietnam Impact Evaluation Sample Selection ......................... 8 2: Histogram of Child Growth Measures (Z-Scores) for Children <2 .......................................................................... 30 3A: Arm and Head Circumference Z-Scores by Sex and Months of Age (Children <2) ......................................... 32 3B: Weight-for-Age and Length-for-Age Z-Scores by Sex and Months of Age (Children <2) ......................................... 33 viii Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Contents 3C: BMI-for-Age and Length-for-Height Z-Scores by Sex and Months of Age (Children <2) ......................................... 33 4: Distribution of Wealth Scores for the WSP Survey and VNDHS ......................................................................... 52 Tables 1: Summary Statistics .............................................................. 11 2: Socio-Demographic Characteristics of the Household ......... 13 3: Educational Attainment of Household Members ................... 14 4: Percent Distribution of Household Assets and Non-Labor Income ................................................................................ 15 5: Employment Characteristics of Household Members ........... 16 6A: Self-Reported Handwashing with Soap Behavior by Wealth Quintile (Previous 24 Hours) ................................ 19 6B: Self-Reported Handwashing with Soap Behavior by Province (Previous 24 Hours)............................................... 19 7: Observation of Place for Washing Hands by Wealth Quintile and Province ........................................... 20 8A: Observation of a Place for Washing Hands After Going to Toilet ............................................................................... 22 8B: Observation of a Place for Washing Hands When Preparing Food or Feeding a Child ...................................... 23 9: Observation of Caregiver's Hands by Wealth Quintile ........... 24 10: Diarrhea, ALRI, and Anemia Prevalence by Poverty Status and Access to Place for Washing Hands (Children <5) ......... 25 11: Diarrhea and ALRI Prevalence by Province (Children <5) ........................................................................ 26 12: Diarrhea Prevalence and Treatment by Wealth Quintile (Children <5) ........................................................................ 27 13: ALRI Prevalence and Treatment by Wealth Quintile (Children <5) ........................................................................ 27 14: Care-Seeking Behavior for Child Illness by Wealth Quintile..................................................................... 27 15: Households with Lost Hours Due to Child Illness by Wealth Quintile and Province........................................... 28 16: Anemia Prevalence by Wealth Quintile and Province (Children <2) ........................................................................ 28 17: Prevalence of Malnutrition, Stunting, and Wasting by Wealth Quintile and Province (Children <2) ...................... 29 18A: Child Growth Measures (Z-Scores) by Wealth Quintile (Children <2) ....................................................................... 31 18B: Child Growth Measures (Z-Scores) by Province (Children <2) ........................................................................ 31 www.wsp.org ix Findings from the Impact Evaluation Baseline Survey in Vietnam Contents 19: Child Growth Measures (Z-Scores) by Poverty Status and Access to Place for Washing Hands (Children <2) ......... 31 20A: Communes Selected to Receive Treatment 1 (IPC + Mass Media).............................................................. 37 20B: Communes Selected to Receive Treatment 2 (IPC + DCC + Mass Media) .................................................. 39 20C: Communes Selected to Serve as Control (Mass Media) ....................................................................... 41 21A: Comparison of Means Tests for Household Demographics ..................................................................... 44 21B: Comparison of Means Tests for Household Primary Work, Labor Income, and Non-Labor Income ................................ 45 21C: Comparison of Means Tests for Household Assets .............. 46 21D: Comparison of Means Tests for Handwashing Behavior....... 47 21E: Comparison of Means Tests for Handwashing Facilities ....... 48 21F: Comparison of Means Tests for Acute Lower Respiratory Infection and Diarrhea Symptoms Prevalence (% Children <5) .................................................................... 50 21G: Comparison of Means Tests for Child Growth Measures (Z-Scores)............................................................................ 50 22: Demographic Characteristics of Household Respondents in WSP Survey and VNDHS ................................................. 51 23: Educational Attainment of Household Population in WSP Survey and VNDHS ............................................................. 53 Boxes 1: Health and Welfare Impacts ................................................... 9 2: Handwashing Behavior and Determinants.............................. 9 Map 1: Geographic Representation of Communes Selected for Handwashing Project Impact Evaluation ................................ 7 x Global Scaling Up Handwashing I. Overview 1.1 Introduction In response to the preventable threats posed by poor sanitation The handwashing project's global activities test innovative and hygiene, in December 2006 the Water and Sanitation Pro- approaches at scale, with the following four main objectives: gram (WSP) launched two large-scale projects, Global Scaling · Design and support the implementation of innovative, Up Handwashing and Global Scaling Up Rural Sanitation, to large-scale, sustainable handwashing programs in four di- improve the health and welfare outcomes for millions of poor verse countries (Peru, Senegal, Tanzania, and Vietnam). people. Local and national governments are implementing · Document and learn about the impact and sustainabil- these projects with technical support from WSP. The goal of ity of innovative large-scale handwashing programs. the Global Scaling Up Handwashing project is to reduce · Learn about the most effective and sustainable ap- the risk of diarrhea and therefore increase household pro- proaches to triggering, scaling up, and sustaining ductivity by stimulating and sustaining the behavior of handwashing with soap behaviors. handwashing with soap at critical times in 5.4 million peo- · Promote and enable the adoption of effective hand- ple in Peru, Senegal, Tanzania, and Vietnam. On average, washing programs in other countries and--through the project will improve the handwashing behavior of over the translation of results and lessons learned--position one million people per country. handwashing as a global public health priority through effective advocacy and applied knowledge and commu- Handwashing with soap at critical times (such as after con- nications products. tact with feces and before handling food) has been shown to substantially reduce the incidence of diarrhea. It reduces The handwashing project also aims to complement and im- health risks even when families do not have access to basic prove on existing hygiene behavior change and handwashing sanitation and water supply service. Despite this known approaches, and to enhance them with novel approaches-- benefit, rates of handwashing with soap at critical times are including commercial marketing--to deliver handwashing very low throughout the developing world. with soap messages, along with broad and inclusive govern- ment partnerships of government, private commercial mar- The project aims to test whether improved handwashing keting channels, and concerned consumer groups and behavior at critical times can be generated among the poor nongovernmental organizations (NGOs). These innovative and vulnerable using innovative promotional approaches. methods will be combined with proven community-level in- In addition, it will undertake a structured learning and dis- terpersonal communication and outreach activities, with a semination process to develop the evidence, practical focus on sustainability. In addition, the project incorporates knowledge, and tools needed to effectively replicate and a rigorous impact evaluation component to support thought- scale up future handwashing programs. ful and analytical learning, combined with effective knowl- edge dissemination and global advocacy strategies. WSP's vision of success is that the project will have dem- onstrated that handwashing with soap, at scale, is one of As reflected above, the process of learning, which is sup- the most successful and cost-effective interventions to im- ported in the project's monitoring and evaluation compo- prove and protect the health of poor rural and urban fami- nents, is considered critical to the project's success. As part lies, especially children under the age of five. Moreover, of these efforts, the project will document the magnitude of the project seeks to develop the evidence, practical knowl- health impacts and relevant project costs of the interven- edge, and tools for effective replication and scaling up of tions. To measure impact, the project is implementing a future handwashing programs, potentially reaching more randomized-controlled trial impact evaluation (IE) of the than 250 million people in more than 20 countries by handwashing project in the four countries, using household 2020. surveys to measure the levels of key outcome indicators. www.wsp.org 1 Findings from the Impact Evaluation Baseline Survey in Vietnam Overview This report is part of a series presenting the analysis of base- line data collection conducted in the implementation coun- tries during 2008 and 2009. Global Scaling Up Project Impact Evaluation Rationale and Aims The overall purpose of the IE is to provide decision makers with a body of rigorous evidence on the effects of the hand- washing and sanitation projects at scale in reference to a set of relevant outcomes. It also aims to generate robust evi- dence on a cross-country basis, understanding how effects vary according to each country's programmatic and geo- graphic contexts and generating knowledge of relevant im- pacts such as child growth and development, child illness, and productivity of mother's time, among others. A young Vietnamese child The studies will provide a better understanding of at-scale sanitation and hygiene interventions. The improved evi- dence will support development of policies and programs, and will inform donors and policy makers on the effective- handwashing promotion. Some key elements of the inter- ness and potential of the Global Scaling Up projects as vention include: large-scale interventions to meet global needs. · Key behavioral concepts or triggers for each target audience 1.2 Project Background · Persuasive arguments stating why and how a given In Vietnam, the handwashing project targets mothers and concept or trigger will lead to behavior change, and caregivers of children under five years old, and is aimed at · Communications ideas to convey the concepts improving handwashing with soap practices. Children through many integrated activities and communica- under five represent the age group most susceptible to diar- tion channels. rheal disease and acute lower respiratory infections, which are two major causes of childhood morbidity and mortality 1.3 Project Components in less developed countries. These infections, usually trans- The overall objective of the project is to improve the health ferred from dirty hands to food or water sources, or by di- of populations at risk for diarrhea and acute lower respira- rect contact with the mouth, can be prevented if mothers tory infections, especially children under five years old, and caregivers wash their hands with soap at critical times through a strategic communications campaign aimed at in- (such as before feeding a child, cooking, or eating, and after creasing handwashing with soap behavior at the critical using a toilet or contact with a child's feces). times. In an effort to induce improved handwashing behavior, In Vietnam, the handwashing project has been underway the intervention borrows from both commercial and so- since 2006 in a total of 540 communes across 56 districts cial marketing fields. This entails formative research on in 10 provinces. Phase 1 of the handwashing project, barriers to handwashing with soap, the design of commu- which was funded by the Danish Embassy and had an nications campaigns and messages likely to bring about estimated reach of 17 million through mass media, di- the desired behavior changes, and the strategic delivery of rect consumer contact, and interpersonal communica- messages so that the target audience is "surrounded" by tion, ended in September 2008. Phase 2 of the project, 2 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Overview funded by the Bill and Melinda Gates Foundation, took place between May 2008 and June 2009 and has reached an estimated 650,000 through mass media and interper- sonal communication activities. The third and final phase (Phase 3) of the handwashing project, with contin- ued funding from the Bill and Melinda Gates Founda- tion, aims to reach an additional 17 million through interpersonal communication, mass media, and direct consumer contact. This phase is being evaluated using a randomized-controlled trial impact evaluation. The handwashing project in Vietnam uses a behavior change approach to address barriers to effective handwashing among the target population. Communications activities focus on the importance of handwashing with soap by caretakers for Vietnam Women's Union members teach women to wash the health and development of young children; the need to hands with soap in the market wash hands with soap immediately before cooking or eating, before feeding a child, and after using the toilet; and the need to make soap available at a water source. The target popula- tion for the intervention is mothers and other caretakers age · Component 2--Direct Consumer Contact (DCC) 15 to 49, and children from 6 to 12 years of age.3 Activities: Rooted in the communications objec- tives of the handwashing project, this component The IE seeks to evaluate two distinct combinations of the fol- reinforces the IPC components of the implementa- lowing three components of Phase 3 of the program: tion by integrating commercial marketing events, · Component 1--Interpersonal Communication or DCC, and social marketing of handwashing (IPC) Activities: with technical support from the with soap. The DCC events use education and en- WSP, the Vietnam Women's Union (VWU) is im- tertainment as the primary means of communicat- plementing an extensive training program for vil- ing handwashing with soap messages through skits, lage health workers, teachers, and Women's Union songs, dances, and question and answer sessions to members in how to promote group and household reinforce the messages delivered through the IPC level IPC activities that reinforce handwashing with activities and mass media. These events also pro- soap behavior in the target population. In total, over vide an opportunity for the campaign to distribute 14,000 front-line workers have been trained as hand- physical reminders (including promotional flyers, washing motivators to carry out the IPC activities soap samples, and handwashing campaign branded in their communities. These IPC activities include hand clappers and hats) to participants to wash group meetings with mothers and other caretakers hands with soap. of children under five, group meetings with women · Component 3--Mass Media Campaign: The WSP, ages 18­49, group meetings with grandparents, in collaboration with various national and provincial household visits, market meetings, Women's Union television stations is launching several mass media club meetings, and handwashing with soap festivals, campaigns throughout the life of the project, in- among others. cluding a large scale campaign scheduled to roll out from March 2010 to January 2011. The mass media 3 A school-based handwashing campaign carried out by the project targeting children campaign features television spots carried out on a 6 to 10 years of age is not part of the impact evaluation. national scale across ten channels. The frequency of www.wsp.org 3 Findings from the Impact Evaluation Baseline Survey in Vietnam Overview the spots will vary over time in an effort to reach the information on the extent to which improved handwashing target audience as often as possible. behavior contributes to child health and welfare. One experimental arm of the IE will evaluate the impact of The primary hypothesis of the study is that improved hand- IPC and mass media (components 1 and 3), while the other washing behavior leads to reductions in disease incidence, experimental arm will evaluate the combination of IPC, and results in direct and indirect health, developmental, DCC, and mass media (components 1, 2, and 3). Both ex- and economic benefits by breaking the fecal-oral transmis- perimental arms will be measured against a control arm that sion route. The IE aims to address the following research will benefit from handwashing messages via national mass questions and associated hypotheses: media, but that will not be exposed to either IPC or DCC activities promoting handwashing with soap. 1. What is the effect of handwashing promotion on handwashing behavior? 1.4 Objectives of the Study 2. What is the effect of improved handwashing behav- The objective of the IE is to assess the effects of the hand- ior on health and welfare? washing project on individual-level handwashing behavior 3. Which promotion strategies are more cost-effective and practices of caregivers. By introducing exogenous varia- in achieving desired outcomes? tion in handwashing promotion (through randomized ex- posure to the project), the IE will also address important The purpose of this report is to provide baseline descriptive issues related to the effect of intended behavioral change on information on the selected indicators included in the child development outcomes. In particular, it will provide survey. 4 Global Scaling Up Handwashing II. Methodology 2.1 Randomization To address the proposed research questions, a proper IE Random assignment of treatment helps to prevent addi- methodology is needed to establish the causal linkages be- tional problems that affect our certainty that the observed tween the handwashing project and the outcomes of inter- changes in outcomes are due to the intervention. In many est. In order to estimate the causal relationship between the cases, communities chosen for programs such as the hand- handwashing project (treatment) and the outcomes of in- washing project are selected precisely due to the high like- terest, a counterfactual is required--in other words, a com- lihood of their success due to favorable local conditions parison group that shows what would have happened to the (strong leadership, existing water and sanitation infra- target group in the absence of the intervention. structure, highly educated population, etc.), and are likely to be systematically different from areas that are less desir- Random assignment of treatment, whereby a statistically able for implementation. If random assignment is not random selection of communities receives the treatment used, a comparison of treated and untreated areas would and the remaining serve as controls, generates a robust confuse the program impact with pre-existing differences counterfactual to measure the causal effect of the interven- between communities, such as different hygiene habits, tion. The randomization process ensures that on average the lower motivation, or other factors that are difficult to ob- treatment and comparison groups are equal in both ob- serve. This is known as selection bias in economics and con- served and unobserved characteristics,4 and that an appro- founding bias in the health sciences.5 Random assignment priate counterfactual can be measured. A randomized of treatment avoids these difficulties, by ensuring that the experimental evaluation with such a comparison group is communities selected to receive the intervention are no valuable because it reduces the possibility that observed different on average than those that are not. A detailed changes in outcomes in the intervention group are due to comparison of means between the treatment and control factors external to the intervention. groups on an exhaustive list of covariates is provided in Annex 2. In the context of this evaluation, where implementation spans nine months, it is possible that factors such as 2.2 Study Design weather, macro-economic shocks, disease outbreaks, or To assess the impact of each component of the handwash- other new and ongoing public health, nutrition, sanita- ing project on the health of children under five, the evalua- tion, and hygiene campaigns, for example, could influence tion will have two treatment arms. Treatment 1 (T1) the same set of outcomes that are targeted by the hand- comprises the IPC and mass media campaign components, washing project (e.g., diarrhea prevalence in young chil- and Treatment 2 (T2) comprises the IPC, DCC and mass dren, health, and welfare). If no control group is media campaign components. As mentioned previously, in maintained and a simple pre- to post-assessment is con- order to measure the health and developmental impact of ducted of the handwashing project, the observed changes each component, a counterfactual to T1 and T2 is needed, in outcomes cannot be causally attributed to the which we will refer to as the Control (C). The design allows intervention. us to investigate the impact of both T1 and T2 (relative to the control). Each group, T1, T2, and C, comprises a rep- resentative sample of the population of households with at least one child under the age of two at baseline. 4 Technically, this is only true with infinite sample sizes, which is unaffordable and unnecessary. Instead, this study seeks to minimize the risk that the means of the treatment and comparison groups differ significantly. For details of mean comparison tests across treatment and control groups, please see Annex 2: Baseline 5 Balance Comparison of Means Tests. Hernan 2004. www.wsp.org 5 Findings from the Impact Evaluation Baseline Survey in Vietnam Methodology 2.3 Sampling Strategy and Sample Size 0 Districts that have not participated in large hy- The primary objective of the handwashing project is to im- giene programs, particularly in handwashing, prove the health and welfare of young children. Thus, a suf- over the past five years, and ficient sample size was calculated to capture a minimum 0 Districts with the willingness, commitment, and effect size of 20 percent on the key outcome indicator of capacity of VWU staff to carry out the planned diarrhea prevalence among children under two years old at activities. the time of the baseline. By focusing on households with From the list of 18 eligible districts provided by the children under two, the evaluation aims to capture changes VWU, a total of 15 were selected to participate in in outcomes for the age range during which children are the experimental phase of the handwashing project. most sensitive to changes in hygiene in the environment. These included five districts from the province of Power calculations indicated that approximately 1,050 Hung Yen, four districts from Thanh Hoa, and six households per treatment arm would need to be surveyed in districts from Tien Giang. order to capture a 20 percent reduction in diarrhea preva- · Stage 2: Commune Selection lence, and in order to account for the possibility of house- Within the 15 selected districts a total of 315 com- hold attrition during the project study phase. Therefore, munes were used as the sampling frame. The sample since the evaluation consists of two treatment groups and was first stratified by province to account for regional one control group, the total sample incorporates 3,150 variation between the provinces. Within each prov- households, each of which has at least one child under two ince, communes were matched into groups of three years of age at the time of the survey. so as to minimize the statistical distance between the so as to minimize statistical distance between Rather than using simple random sampling, which is much the three communes based on covariates of popu- more costly, the study randomly sampled households in lation size, number of households, and geographic clusters at the commune administrative level. Households location (coastal, flat, or mountainous area). A total were randomly selected from a sampling frame of 210 com- of 70 groups of three were then randomly selected munes randomly selected from 15 districts in three prov- into the study (Hung Yen = 24; Thanh Hoa = 20; inces. Data were collected using structured questionnaires Tien Giang = 26). Finally, the communes in each in all 3,150 households and in each of the 210 communes group of three were randomly assigned to one of (one per commune). Further details on the selected list of the three treatment groups, T1, T2, or C. A total of districts and communes can be found in Annex 1. 70 communes were assigned to T1, 70 to T2, and 70 to control. In total, 401 communes across 18 districts in the three proj- · Stage 3: Household Selection ect provinces were listed by the VWU as eligible to partici- Approximately one month prior to fieldwork a list pate in the project. From this list a total of 210 communes6 was obtained from the commune health station. It across 15 districts in the three provinces were selected for contained all households with a child younger than the study (as shown in Map 1) using the following three- the age of two. A random sample of 15 households stage design: was drawn at the time of the survey in each com- · Stage 1: District Selection mune. Each household contained at least one child District selection was not randomized, but was in- between the age of 0 and 24 months at the time of stead discussed and agreed upon with VWU at cen- listing. An additional 10 replacement households ter and provincial administrative levels. The criteria were randomly selected at the time of the survey to for district selection were: accommodate households that refused to participate 0 Districts with a large population in the survey. Households in which specially trained community motivators lived were excluded from the sample, since these volunteers would later play a role 6 The remaining 191 communes were not part of the evaluation sample and will not receive the IPC or DCC handwashing project interventions, but will be exposed to in delivering handwashing project messages to the handwashing messages via national-level mass media. community. 6 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Methodology MAP 1: GEOGRAPHIC REPRESENTATION OF COMMUNES SELECTED FOR HANDWASHING PROJECT IMPACT EVALUATION Treatment Group 1: IPC and Mass Media CHINA Treatment Group 2: IPC and DCC and Mass Media Control Group Not included in IE sample À NÔI HÀ NÔI LAO PEOPLE'S DEMOCRATIC REPUBLIC THAILAND CAMBODIA 0 50 100 150 200 Kilometers 0 50 100 150 Miles IBRD 38036 AUGUST 2010 www.wsp.org 7 Findings from the Impact Evaluation Baseline Survey in Vietnam Methodology FIGURE 1: VIETNAM IMPACT EVALUATION SAMPLE SELECTION Vietnam Impact Evaluation Sample Selection Tien Giang Hung Yen Thanh Hoa 104 84 127 Clusters Clusters Clusters Cai Lay Yen My T1 = 26 T1 = 24 T1 = 20 Quang Xuong T2 = 26 Chau Thanh T2 = 24 T2 = 20 Kim Dong C = 26 C = 24 C = 20 Tin Gia Cho Gao Tien Lu Trieu Son Go Cong Tay 78 72 60 Clusters Clusters An Thi Clusters Thach Thanh Tan Phuoc Tan Phu Dong Phu Cu This sample selection process is illustrated in Figure 1. Fur- The above variables are collected through three different sur- ther details on the selected list of districts and communes veys: the baseline survey, collected before the intervention and can be found in Annex 1. reported on here; a longitudinal survey, collected a total of three times prior to the intervention; a mid-term monitoring survey, 2.4 Variables for Data Analysis collected three to six months after the intervention began; and The IE aims to assess both the effect of project on hand- a post-intervention survey, to be collected after the intervention washing behavior and the effect of handwashing on child is complete. health and welfare. In order to measure potential impacts of the intervention, the study will collect data on child Box 1 and Box 2 summarize the variables measured and illness, nutrition, child growth and development, anemia, how measurements were performed. productivity, education, environmental contamination,7 and handwashing behavior and its determinants. 2.5 Instruments for Data Collection The baseline survey was conducted from September to December 2009 and included the following instruments: 7 · Household questionnaire: The household ques- Environmental contamination as measured by water samples will be collected during the post-intervention follow-up survey. tionnaire was conducted in all 3,150 households to 8 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Methodology BOX 1: HEALTH AND WELFARE IMPACTS What Does the Evaluation Measure? How Is It Being Measured? Measuring Instrument Diarrhea prevalence Caregiver reported symptoms col- Household questionnaire lected in a 14-day health calendar Productivity of mother's time Time lost to own and child's illness Household questionnaire Education benefits School enrollment and attendance Household questionnaire 8 Child growth Anthropometric measures: In-household collection of - Weight anthropometric (child growth) - Height measures - Arm and head circumference Anemia Hemoglobin concentration In-household collection and (< 110g/L per international analysis of capillary blood using standards)9 the HemoCue photometer BOX 2: HANDWASHING BEHAVIOR AND DETERMINANTS What Does the Evaluation Measure? How Is It Being Measured? Measuring Instrument Handwashing with soap behavior Direct observation of place for Household questionnaire handwashing stocked with soap and water Self-reported handwashing with Household questionnaire soap behavior Determinants to handwashing with Opportunity, ability, and motivation Household questionnaire soap behavior10 determinants collect data on household composition, education, · Health questionnaire: The health questionnaire was labor, income, assets, spot-check observation of conducted in all 3,150 households, to collect data handwashing facilities, handwashing behavior, and on children's diarrhea prevalence, acute lower respi- handwashing determinants. ratory infection (ALRI) and other health symptoms, child development, child growth, and anemia. 8 Habicht 1974. · Community questionnaire: The community 9 Stoltzfus and Dreyfus 1999. 10 questionnaire was conducted in 210 communes, The analysis of the determinants of handwashing with soap behavior is not included in this report. to collect data on socio-demographics of the www.wsp.org 9 Findings from the Impact Evaluation Baseline Survey in Vietnam Methodology Hemoglobin concentrations were measured in children under two years of age at the household level using the HemoCue Hb201 photometer, a portable device that al- lows for immediate and reliable quantitative results. Using sterile and disposable lancets (pricking needle), a drop of capillary blood was obtained from the child's second or third finger and collected in a cuvette, and then intro- duced into the HemoCue machine. Hemoglobin concen- tration appeared in the display screen of the device in about one minute, and results were transferred to the questionnaire. Anthropometric measures were made ac- cording to standardized protocols using portable infant- ometers, scales and measuring tape.11 Enumerators cross a bridge in Tien Giang province for a household interview 2.6 Field Protocols The National Institute of Hygiene and Epidemiology (NIHE) was contracted to conduct the field work for the baseline survey. With support from the principal community, accessibility and connectivity, edu- investigator and the global IE team, NIHE researchers cation and health facilities, water and sanitation trained field supervisors and enumerators on all data related facilities and programs, and government collection protocols and instruments and were in charge assistance or programs related to health, educa- of standardization of anthropometric and anemia tion, cooperatives, agriculture, water, and other measures. development schemes. Each field survey team consisted of one province level A total of three pre-intervention longitudinal surveys and staff, two district level staff, and one to two commune- one mid-term monitoring survey will be conducted during level staff. There were a total of 15 survey teams, one per the study. The post-intervention follow-up survey will be district. Province-level staff served as supervisors and conducted from November 2010 to January 2011 and will oversaw quality control of the interviews. District-level collect data on all the indicators collected during the base- staff included one health staff in charge of interviewing line survey, plus dwelling characteristics, water sources, the household, and one laboratory staff in charge of child drinking water, sanitation, exposure to health interven- anthropometric and hemoglobin concentration mea- tions, and mortality. surements, as well as backstopping the primary inter- viewer. One to two commune-level health staff/nurses The survey instrument was drafted by the WSP global impact were recruited from each commune to assist in anthro- evaluation team, a group of experts from different disciplines. pometric measurements and to receive training on the The complete instrument, which included a set of household, child health calendar for administration of the longitudi- community and longitudinal questionnaires, was translated nal survey. Three field managers from NIHE oversaw the into Vietnamese, underwent back-translation into English, work in each province. and the final version was pre-tested prior to use in the baseline survey. Questionnaires were administered to respondents in Vietnamese by native speakers. 11 Habicht 1974. 10 Global Scaling Up Handwashing III. Findings In this section, we present summary descriptive statistics for which varies highly across household heads (3.51 million key demographic, socioeconomic, hygiene, health, and VND). Other household members are, on average, much child development variables. Findings are cross tabulated by younger (19.1 years old) and slightly smaller percentages household wealth quintile and province, and for outcomes have completed primary school education (81%). Three- of interest such as child growth measures, diarrhea, and quarters of the other members of the household are em- ALRI in relation to access to a place for handwashing. The ployed and earning an average monthly income of 670,000 cross tabulations are valuable for understanding relation- VND (US$36), but this income is highly variable among ships between study outcomes and socioeconomic, geo- households. Household income per capita is slightly lower graphic, and environmental characteristics of the household, than the average income of the household head, at 1.02 and can help generate hypotheses regarding important fac- million VND. tors to child health and development. The following tables provide a more detailed analysis of the 3.1 General Household Characteristics socio-demographic and socioeconomic characteristics of Table 1 shows a brief summary of basic household socioeco- the household by wealth quintile. Table 2 presents the age nomic characteristics. We find that the average household distribution of household members and household size by (HH) comprises 4.6 individuals and that a male heads wealth quintile. Little difference is found across wealth 86.7% of households. The head of household is 42 years of quintiles at the younger ages; however, households in the age on average, with the proportion completing primary higher wealth quintiles contain a higher proportion of indi- school 83.3%. The household head is employed in 85.8% viduals over 45, and most noticeably over 50 (16.2% in the of households with an average monthly income of 1.06 mil- lion Vietnamese dong (VND), equivalent to US$5712), TABLE 1: SUMMARY STATISTICS Standard Mean Deviation HH size 4.6 1.2 HH Head: HH head is male (% HH heads) 86.7% -- Age 42.2 15.1 HH head completed primary school education (% HH heads) 83.3% -- HH head is employed (% HH heads) 85.8% -- Labor income in VND (millions) 1.06 3.51 Other HH Members: Age 19.2 18.2 Other HH member completed pri- mary school education (% other HH A household interview takes place in Tien Giang province members) 81.0% -- Other HH member is employed (% other HH members) 75.4% -- Labor income in VND (millions) 0.67 2.19 12 The US dollar-Vietnamese dong exchange rate of 18,544 VND per US$1 was provided by the Vietnam Central Bank as of April 23, 2010. HH per capita income (in VND) 1.02 4.59 www.wsp.org 11 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings wealthiest quintile, compared to 8.3% in the poorest quintile). Older individuals may contribute to higher human capital in the household, leading to more wealth attainment, measured by the asset index in this study. On average, poorer house- holds contain a larger proportion of younger members. More specifically, there is approximately a five percentage-point difference between the poorest and wealth- iest quintile in terms of the number of children younger than five. This is further demonstrated by the higher than average number of children younger than five per household in the lowest quintile, 1.24, compared with the overall average of 1.19. Both household heads and other members of the household are younger on average in these poorer households. Table 3 presents the percent distribution of education for individuals age five years and older. Education is an important socioeconomic indicator, closely associated with household income, child health status, and in the case of the handwashing intervention, may be related to the receptiveness to the communications messages of improved handwashing behavior. Educational attainment is high in Vietnam, achieving around 100% gross primary enrollment13 in 2008, according to the 13 The ratio of primary school enrollment to the number of primary school-age children (usually children ages 6­11). This figure can be greater than 100% if enrolled children are older or younger than the corresponding age group. Households with children under age two were included in the survey 12 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 2: SOCIO-DEMOGRAPHIC CHARACTERISTICS OF THE HOUSEHOLD Wealth Quintile 1st 2nd 3rd 4th 5th Total Age: 0­4 29.2% 26.2% 25.1% 24.7% 24.2% 25.8% 5­9 7.7% 7.1% 7.0% 6.6% 6.0% 6.8% 10­14 4.4% 4.6% 3.6% 3.3% 3.4% 3.9% 15­19 2.7% 3.1% 3.6% 2.7% 2.8% 3.0% 20­24 9.7% 9.7% 9.6% 10.0% 7.5% 9.3% 25­29 15.5% 16.4% 16.6% 15.2% 15.8% 15.9% 30­34 11.6% 10.3% 9.8% 11.7% 11.1% 10.9% 35­39 6.4% 6.3% 5.1% 5.0% 6.4% 5.8% 40­44 3.0% 3.2% 2.8% 2.6% 2.9% 2.9% 45­49 1.4% 2.0% 2.9% 3.1% 3.6% 2.6% 50+ 8.3% 11.1% 14.0% 15.1% 16.2% 13.1% Age of HH head (average) 37.9 40.2 43.3 43.8 45.8 42.2 Age of other HH members (average) 16.2 18.2 19.5 20.3 21.1 19.2 Male head of household (% HH) 86.6% 87.9% 87.7% 85.6% 85.8% 86.7% HH size: 2 1.1% 0.0% 0.2% 0.0% 0.0% 0.3% 3 26.6% 22.7% 15.3% 14.1% 10.7% 17.9% 4 36.0% 31.8% 31.3% 32.1% 29.4% 32.1% 5 24.2% 28.1% 30.5% 30.4% 31.3% 28.9% 6 8.8% 12.9% 16.3% 15.7% 19.6% 14.7% 7 2.1% 2.6% 4.0% 5.8% 4.6% 3.8% 8 0.8% 1.3% 1.4% 1.1% 2.1% 1.3% 9 0.2% 0.5% 0.6% 0.8% 1.1% 0.6% 10 0.2% 0.2% 0.3% 0.2% 1.0% 0.4% 13 0.0% 0.0% 0.0% 0.0% 0.2% 0.0% HH size (average) 4.2 4.5 4.7 4.7 4.9 4.6 Total Number of Children Under Five Years of Age: 1 76.9% 83.1% 82.7% 83.4% 81.3% 81.5% 2 22.3% 16.6% 16.8% 16.1% 17.7% 17.9% 3 0.8% 0.3% 0.5% 0.5% 0.8% 0.6% 4 0.0% 0.0% 0.0% 0.0% 0.2% 0.0% Number of children under five years of age (average) 1.24 1.17 1.18 1.17 1.20 1.19 www.wsp.org 13 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 3: EDUCATIONAL ATTAINMENT OF HOUSEHOLD MEMBERS Wealth Quintile 1st 2nd 3rd 4th 5th Total Number of HH heads completed primary school (% HH heads) 78.0% 82.6% 82.7% 87.2% 88.5% 83.8% Educational Attainment of HH Head: No education 4.3% 1.8% 1.1% 1.5% 0.5% 1.8% Incomplete primary 18.6% 16.2% 16.6% 11.6% 11.2% 14.8% Complete primary 43.8% 42.6% 42.9% 42.6% 31.9% 40.7% Incomplete secondary 24.7% 27.2% 24.3% 24.1% 23.0% 24.7% Complete secondary 7.5% 10.7% 12.6% 13.9% 18.7% 12.7% Higher 1.2% 1.5% 2.5% 6.2% 14.7% 5.2% Female HH members (>5 years old) attended or currently attending school (% HH members) 95.8% 98.3% 98.4% 98.5% 99.4% 98.1% Educational Attainment of Female HH Members: No education 2.7% 1.3% 1.3% 1.0% 0.5% 1.3% Incomplete primary 23.9% 20.7% 18.1% 17.3% 15.0% 18.7% Complete primary 39.7% 40.5% 36.4% 35.4% 28.5% 35.7% Incomplete secondary 24.1% 25.8% 27.0% 24.8% 22.7% 24.9% Complete secondary 9.0% 10.6% 13.8% 16.0% 17.7% 13.8% Higher 0.5% 1.0% 3.4% 5.4% 15.7% 5.7% Other HH members (>5 years old) attended or currently attending school (% HH members) 97.4% 98.7% 98.8% 99.1% 99.5% 98.8% Educational Attainment of Other HH Members: No education 4.4% 1.8% 1.6% 1.6% 0.7% 1.9% Incomplete primary 21.6% 21.2% 19.8% 18.0% 15.2% 19.0% Complete primary 41.0% 40.6% 37.0% 36.4% 31.0% 37.0% Incomplete secondary 24.7% 25.0% 26.5% 24.6% 22.1% 24.5% Complete secondary 8.0% 10.4% 12.2% 14.1% 15.8% 12.3% Higher 0.3% 1.0% 2.9% 5.3% 15.2% 5.3% World Bank. Among household heads there is a reasonably Table 4 presents a complete summary of household assets small difference between primary school completion between by wealth quintile as well as non-labor income, such as gov- the poorest and wealthiest households, however the disparity ernment transfers and cash remittances. In the households in post-secondary educational attainment between the poor- sampled, televisions, bicycles, motorbikes, telephones (in- est and wealthiest is more pronounced. cluding mobile), and electric fans are common household assets, owned by over three-quarters of households. Since Female and other household member school attendance is the household assets shown in Table 4 make up the wealth high, at over 98% of household members, and female index, differences are expected in asset ownership by quin- household members in particular, attending or having at- tile. For instance, only 27% of the poorest households own tended school. a telephone, including a mobile phone, whereas 97% of the 14 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 4: PERCENT DISTRIBUTION OF HOUSEHOLD ASSETS AND NON-LABOR INCOME Wealth Quintile 1st 2nd 3rd 4th 5th Total Average HHs non-labor income in VND (millions) 0.96 1.39 1.93 2.26 8.98 3.05 HH Assets: Radio, CD, cassette 2.7% 5.9% 9.4% 9.3% 20.4% 9.5% TV 65.4% 95.8% 99.2% 99.5% 100.0% 92.0% VCR 13.9% 47.9% 69.8% 81.2% 91.7% 60.9% Computer 1.1% 0.2% 1.1% 5.6% 23.3% 6.3% Bicycle 69.2% 74.8% 75.7% 80.8% 77.0% 75.5% Motorcycle 34.9% 77.5% 89.8% 93.0% 97.6% 78.5% Automobile or truck 0.0% 0.2% 2.1% 2.7% 8.3% 2.7% Refrigerator 0.0% 3.4% 6.5% 32.7% 86.4% 25.8% Gas stove 3.5% 11.2% 39.0% 74.1% 94.1% 44.4% Blender 2.7% 8.1% 16.8% 39.3% 80.0% 29.4% Microwave 0.0% 0.0% 0.0% 0.5% 6.1% 1.3% Washing machine 0.0% 0.2% 0.3% 2.9% 35.0% 7.7% Water boiler 2.6% 3.7% 14.4% 28.3% 51.1% 20.0% Machinery, equipment for household business 0.2% 1.1% 1.6% 3.8% 5.1% 2.4% Boat 1.0% 1.8% 2.6% 3.0% 4.6% 2.6% Telephone (including mobile) 27.1% 74.0% 88.8% 93.6% 97.1% 76.1% Air conditioner 0.2% 0.0% 0.5% 0.5% 3.8% 1.0% Electric fan 81.8% 97.3% 98.1% 99.8% 99.8% 95.4% HH owns other piece of land 8.1% 11.8% 19.5% 24.9% 26.0% 18.1% HH owns farm equipment 11.6% 13.6% 22.7% 28.3% 23.5% 19.9% HH has animals 38.8% 39.1% 29.2% 34.7% 41.4% 36.6% Number of livestock owned per HH (average) 1.11 1.07 1.23 1.16 0.94 1.10 richest households own a phone. Ownership of a motor- government transfers, household production of products, bike is another asset owned largely by the wealthier house- and agricultural activity income not mentioned as primary or holds. Automobiles are still quite rare in rural Vietnam, secondary work earnings. The average household non-labor with just 2.7% of households in this sample owning a car or income, considering only positive values, is approximately truck. Computers are likewise absent in rural Vietnamese 3.05 million VND per household. Non-labor income is households. Around 45% of households own a gas stove, highly positively associated with wealth quintile, with house- but just 3.5% of the poorest households have this type of holds in the top quintile reporting more than nine times the cook stove. Ownership of animals is quite consistent across non-labor income of the poorest households. wealth quintiles, averaging 36.6% of households overall. Table 5 presents details on the principal economic activity Overall, 75.6% of the households declared having income for household respondents over 15 years of age. Overall, sources not classified as labor income, such as remittances, engagement in economic activity is high in the sample. www.wsp.org 15 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 5: EMPLOYMENT CHARACTERISTICS OF HOUSEHOLD MEMBERS Wealth Quintile 1st 2nd 3rd 4th 5th Total HH head is employed (% HH heads) 87.1% 88.2% 87.2% 84.6% 81.8% 85.8% Other HH member is employed (% other HH members) 76.0% 74.1% 75.8% 76.1% 75.1% 75.4% Last Week Activity--HH Head is Unemployed: Looking for work 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Studying 0.0% 1.4% 1.3% 1.0% 0.9% 0.9% Looking after the home 43.8% 39.7% 42.5% 37.5% 30.7% 38.1% Rent earner 2.5% 6.8% 3.8% 5.2% 1.8% 3.8% Not working and not looking for job 53.8% 52.0% 52.6% 56.3% 66.7% 57.1% Last Week Activity--Other HH Member is Unemployed: Looking for work 1.8% 0.7% 1.0% 1.0% 3.2% 1.6% Studying 16.0% 16.0% 21.3% 13.4% 19.5% 17.4% Looking after the home 64.8% 67.4% 61.8% 65.6% 51.5% 61.7% Rent earner 1.4% 1.8% 1.0% 2.6% 2.0% 1.8% Not working and not looking for job 16.0% 14.2% 15.0% 17.3% 23.8% 17.5% Primary Employment Status (% All Employed): Self-employed 4.0% 6.3% 7.5% 7.1% 8.7% 6.9% Employee 15.5% 19.7% 21.8% 27.0% 35.4% 24.4% Employer or boss 0.0% 0.1% 0.1% 0.5% 1.3% 0.4% Worker without remuneration 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% Day laborer 9.8% 7.2% 3.7% 3.3% 2.3% 5.0% Working in household production, trade or business 70.2% 65.9% 66.4% 61.5% 51.5% 62.7% Other 0.4% 0.6% 0.5% 0.5% 0.8% 0.6% Monthly Salary in VND (millions): Self-employed 1.31 1.39 1.51 1.94 2.53 1.85 Employee 1.52 1.77 2.21 2.23 2.79 2.26 Employer or boss14 -- 18.00 2.00 4.50 4.61 5.26 Day laborer 1.14 1.34 1.33 1.35 3.01 1.44 Working in household production, trade or business15 -- -- -- -- -- -- Other 0.21 0.73 0.92 2.56 2.30 1.58 Total 1.36 1.65 1.95 2.11 2.77 2.09 Hours Worked per Day: Self-employed 7.4 7.4 7.3 7.9 7.8 7.6 Employee 8.3 8.4 8.4 8.3 8.3 8.3 (Continued ) 14 There were no employers/bosses in 1st wealth quintile. The 18 million VND figure is the result of just one individual reporting income of 900,000 VND (approx. US$50) per day, which on a monthly basis is equivalent to 18 million VND. 15 Labor income from household production, trade, or business is reported under "Module 4: Household Income" in the household survey. 16 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 5: (Continued) Wealth Quintile 1st 2nd 3rd 4th 5th Total Employer or boss -- 9.0 9.0 8.6 8.7 8.7 Worker without remuneration 17.0 -- -- -- -- 17.0 Day laborer 7.9 8.0 7.8 8.1 7.5 7.9 Working in household production, trade or business 7.3 7.0 7.0 7.3 7.4 7.2 Other 8.0 6.3 10.5 6.9 7.2 7.4 Total 7.5 7.4 7.4 7.6 7.7 7.5 Days Worked per Month: Self-employed 20.2 22.1 20.7 23.2 23.2 22.1 Employee 21.8 23.6 23.9 24.4 23.9 23.7 Employer or boss -- 27.0 19.5 22.3 26.5 25.2 Worker without remuneration 28.0 -- -- -- -- 28.0 Day laborer 18.5 19.7 20.3 19.1 21.1 19.4 Working in household production, trade or business 19.3 17.8 18.0 19.0 20.5 18.9 Other 24.0 22.5 19.3 18.8 21.8 21.3 Total 19.7 19.5 19.6 20.8 22.0 20.3 Months Worked in Last 12 Months: Self-employed 10.2 12.0 12.0 10.1 12.0 11.4 Employee 9.6 10.6 10.9 11.3 11.4 11.0 Employer or boss -- 12.0 11.0 10.9 11.6 11.4 Worker without remuneration 2.0 -- -- -- -- 2.0 Day laborer 9.6 10.6 10.4 10.2 10.3 10.2 Working in household production, trade or business 8.1 8.1 8.0 8.2 8.6 8.2 Other 10.2 12.0 12.0 10.1 12.0 11.4 Total 8.6 9.0 9.0 9.3 10.0 9.2 Just under 86% of household heads were employed in the For household members, including household heads, who week prior to the interview, and 75% of other household were employed the week prior to the survey, 62.7% classified members older than 15 years were employed. Interest- their primary work over the past 12 months as work in ingly, the figures are higher for the poorest households household production or services in planting, breeding, for- (87.1% and 76.0% for HH heads and other HH mem- estry or aquaculture, or work in trade or business for the bers, respectively). The week before the interview, unem- household. Another quarter of the employed household ployed HH heads were either both not working and not members classified themselves as employees. This figure is looking for a job (57.1%), or were looking after their highest for the wealthiest households, while the poorest homes (38.1%). The majority of other HH members who households are more likely to work in household production were unemployed the previous week were looking after or trade. Very few households classified their primary work as the home (61.7%). self-employment, because of the fact that although they work www.wsp.org 17 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings for themselves in household production, services, or trade, times, that is after defecation or contact with a child's feces, they do not earn wages or salary in return for this work. and before cooking or preparing food and feeding a child, and through spot-check observations of whether the house- The average monthly salary for primary work is 2.09 mil- hold has a designated place for handwashing with both soap lion VND (US$113), but this varies from 1.07 million and water available. An additional measure assesses the VND for household production or services to 5.26 million cleanliness of the caretaker's hands through direct observa- VND for employers. As expected, there are large differences tion. These measures serve as proxy indicators of handwash- between the poorest and wealthiest quintiles in average ing with soap behavior in this study, since the actual behavior monthly salaries, with self-employed and employees in the and when it takes place is not observed in the context of the wealthiest quintile earning on average twice the monthly household survey. salary of those in the poorest. Working hours and days are roughly consistent across job type and wealth quintile, with As shown in Tables 6A and 6B, nearly all caregivers, despite an overall average working day of 7.5 hours and working their socioeconomic status, reported washing their hands days per month of 20.3. Those working in household pro- with soap at least once during the past 24 hours when duction or services worked the fewest number of months in prompted. However, self-reported frequency of handwash- the previous year, an average of 8.2 months. ing at particular critical times is lower. When prompted for the occasions over the past 24 hours during which they 3.2 Handwashing Behavior washed their hands with soap, an average of 47.1% reported The Scaling up Handwashing project seeks to achieve health to have washed hands with soap after using the toilet. This and non-health impacts by promoting handwashing with was followed by those who reported washing hands with soap at critical times. Objectively measuring handwashing soap before feeding a child (33.2%) and after cleaning a behavior is therefore critical to the assessment of impacts of child's bottom (32.1%). Of the four critical times, washing the intervention. Handwashing behavior is measured at hands with soap before cooking or preparing food was the baseline in two ways: self-reported handwashing at critical least frequently mentioned (31.0%). Self-reported hand- washing after using the toilet was lower on average in the lowest three wealth quintiles than in the wealthier quintiles. However, those in the bottom two quintiles were more likely to report washing hands with soap after cleaning a child's bottom. On average 78.4% of caretakers mentioned at least one of the four critical times, but the wealthiest were much more likely (86.4%) than the poorest (73.3%) to mention a critical time. There are some large differences evident between the three provinces as shown in Table 6B. Self-reported handwashing is lowest in Tien Giang province (90.3%), as is the percent- age who reported washing hands with soap on at least one critical time (68.0%). While self-reported handwashing is highest in Hung Yen province (98.9%), only 81.0% of care- takers in Hung Yen mention a critical time. Other occa- sions for handwashing that were commonly mentioned were doing laundry (45.6% of caretakers) and because they The Vietnam Women's Union demonstrates proper hand- look or feel dirty (47.7% of caretakers). The findings show washing technique that some critical times are not at the top of the mind for caretakers of young children, as less than one-third reported 18 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 6A: SELF-REPORTED HANDWASHING WITH SOAP BEHAVIOR BY WEALTH QUINTILE (PREVIOUS 24 HOURS) Wealth Quintile 1st 2nd 3rd 4th 5th Total Washed hands with soap at least once in previous 24 hours (% caregivers) 93.3% 90.4% 93.6% 96.7% 96.7% 94.1% Washed Hands with Soap At Least Once in Previous 24 Hours During the Following Events (% Caregivers): Using the toilet (% caregivers) 42.0% 40.9% 44.1% 52.1% 56.6% 47.1% Cleaning child's bottom (% caregivers) 37.7% 33.9% 31.9% 28.2% 28.7% 32.1% Cooking or preparing food (% caregivers) 34.3% 26.5% 30.5% 30.3% 33.1% 31.0% Feeding children (% caregivers) 33.5% 29.7% 33.5% 32.6% 36.6% 33.2% Washed hands with soap during at least one critical time (% caregivers) 73.3% 73.6% 77.0% 81.7% 86.4% 78.4% TABLE 6B: SELF-REPORTED HANDWASHING WITH SOAP BEHAVIOR BY PROVINCE (PREVIOUS 24 HOURS) Province Hung Yen Thanh Hoa Tien Giang Total Washed hands with soap at least once in previous 24 hours (% caregivers) 98.9% 93.5% 90.3% 94.2% Washed Hands with Soap at Least Once in Previous 24 Hours During the Following Events (% Caregivers): Using the toilet (% caregivers) 50.2% 57.5% 36.3% 47.1% Cleaning child's bottom (% caregivers) 37.2% 54.7% 9.7% 31.9% Cooking or preparing food (% caregivers) 24.1% 47.3% 25.0% 31.1% Feeding children (% caregivers) 22.5% 47.6% 32.3% 33.3% Washed hands with soap during at least one critical time (% caregivers) 81.0% 89.2% 68.0% 78.5% handwashing with soap after cleaning a child's bottom and Table 7 presents findings with regards to access to a place before cooking or preparing food, and just around one- for washing hands with water and soap present anywhere in third before feeding children. the home or yard. On average a place for washing hands with both soap and water present was observed in 80.8% of It is worth noting the limitations of this proxy measure for households. Less common, however, was access to a place handwashing behavior, since not all critical times can be ex- for washing hands with soap and water in the poorest pected to take place during the period 24 hours prior to the households (70.2%). This finding points to a clear positive survey. However, the differences noted by province and by association between wealth and presence of a place for wealth quintile are instructive since particular critical times washing hands, with the proportion of households with a would not be expected to be systematically associated with place to wash hands steadily increasing as households move either geographical location or household wealth status. up the wealth index. Furthermore, it underscores the www.wsp.org 19 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 7: OBSERVATION OF PLACE FOR WASHING HANDS BY WEALTH QUINTILE AND PROVINCE Observed Place for Washing Hands with Soap and Water (% HHs) Inside Toilet or Between More than Pond or Stream Anywhere Food Preparation Within 1 Meter of 1 and 3 Meters 3 Meters from Located Elsewhere in the Home Facility Toilet Facility of Toilet Facility Toilet Facility in the Yard Wealth Quintile 1st 70.2% 10.0% 9.3% 11.3% 31.6% 10.2% 2nd 74.4% 13.8% 13.5% 8.9% 28.1% 10.4% 3rd 80.8% 19.2% 15.6% 11.3% 26.3% 7.3% 4th 87.1% 32.1% 17.2% 10.1% 22.0% 3.7% 5th 91.4% 55.1% 11.9% 6.8% 15.6% 1.8% Province Hung Yen 83.2% 26.5% 14.4% 16.7% 23.7% 1.3% Thanh Hoa 87.2% 12.2% 13.8% 7.1% 43.8% 13.4% Tien Giang 73.5% 36.8% 12.8% 4.9% 11.0% 6.4% Total 80.8% 26.3% 13.6% 9.6% 24.6% 6.6% importance of targeting the handwashing project to the poor in order to achieve the greatest impacts. The findings by province are likewise instructive, where access to a place for handwashing is lowest in Tien Giang province (73.5%), and highest in Thanh Hoa (87.2%). The proximity of a place for washing hands to the latrine or place of food preparation is hypothesized to be a key deter- minant of handwashing behavior, since the farther an indi- vidual must walk to wash her hands after defecation or before preparing food, the more likely she is to be distracted by another activity. In the households sampled, a place for handwashing that has both soap and water present was most commonly found either inside the toilet or food prep- aration facility (26.3%), or in the yard more than three me- ters from the toilet facility (24.6%). However, there are large differences observed by socioeconomic status. The wealthiest households are most likely to have a place for washing hands in the toilet or food preparation facility (55.1%), while this is much less common for the poorest households (10.0%). Conversely, the poorest households are most likely to have the a place for washing hands located in the yard more than three meters from the toilet facility (31.6%), which is much less common in the wealthiest households (15.6%). In a little over 10% of households in A typical place for washing hands with soap in rural Vietnam the 1st and 2nd wealth quintile, the place for washing hands is observed to be a pond or stream located somewhere in the 20 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings handwashing device, the type of handwashing device, whether water was available at the time of observation, the type of soap present, and whether ash or mud was observed at the place for washing hands. These observations were made separately for places used to wash hands after going to the toilet, and those used before preparing food, eating, or feeding a child. Table 8A summarizes findings for the principal place used by the household members to wash hands after going to the toilet. A simple homemade water tap or dispenser (sometimes called a "tippy tap") that tips over to release a small amount of water, is the most common type of hand- washing device with 43.5% of households having this type. Another 27.4% of households have a water tap or At a community meeting members discuss the critical faucet for handwashing. This device is most common in times for handwashing the wealthiest households (45%) as opposed to the poor- est (14.3%). The basin or bucket is more common in poorer households (30.7%) than in wealthier households (8.1%). In 98.0% of households, water was observed at yard. What is evident from these findings is that the poorer the place used for washing hands after going to the toilet. the household, the farther they must travel to wash their The presence of soap was also common; at least one type hands with soap and water after using the toilet and before of soap was present at the place for washing hands in close preparing food and/or eating. If the location of the place for to 94% of households. Liquid soap was the least common handwashing is indeed a determinant of handwashing be- type of soap observed (17.6%), and bar soap was much havior, and the presence of soap and water at this place more common in the wealthier households (71.1%) than serves as an environmental cue to wash the hands, the in the poorest (28.8%). Interestingly, powdered soap, poorer households in this sample population may be less such as laundry soap or detergent, was the most common likely to wash their hands with soap and water at the critical type of soap regardless of household wealth. On average, times. 61.9% of households had this type of soap present at the place used to wash hands. Ash and mud, which are sub- Location of the place for washing hands by province helps stances often used for handwashing in poor communities to elucidate some of the findings by wealth quintile above. of South Asia, do not appear to be commonly used cleans- In Thanh Hoa we find a much higher than average propor- ing agents in Vietnam. On average, just 3.6% of house- tion of households has a place for washing hands that is lo- holds were observed to have mud for handwashing at or cated farther than three meters from the toilet facility near the handwashing device, 1.0% had ash, and 2.6% (43.8%), but this is much less common in Tien Giang had both ash and mud. These cleansing agents are slightly (11.0%), where the majority of households have a place for more common among the 1st, 2nd, and 3rd wealth quin- washing hands inside the toilet or food preparation facility tiles. On average, the complete absence of a cleansing (36.8%). It appears from the cross tabulation that house- agent was observed in just 6.0% of households, confirm- holds in Thanh Hoa province account for the sizeable per- ing formative research findings that availability of soap is centage of households where the place for washing hands is generally not a constraint to handwashing.16 located in a pond or stream. Further information was collected from all households on the place for washing hands about the location of the 16 Curtis 2009. www.wsp.org 21 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 8A: OBSERVATION OF A PLACE FOR WASHING HANDS AFTER GOING TO TOILET Wealth Quintile 1st 2nd 3rd 4th 5th Total Location of Handwashing Device (% HHs): Inside toilet facility 9.3% 11.8% 16.8% 27.8% 49.8% 23.3% Inside food preparation facility 2.0% 3.3% 3.9% 6.3% 6.4% 4.4% Less than 1 meter from toilet facility 11.3% 20.7% 19.2% 18.2% 12.1% 16.3% Between 1 and 3 meters from toilet facility 12.7% 10.0% 12.3% 10.5% 7.4% 10.5% More than 3 meters from toilet facility 37.9% 32.1% 29.0% 24.1% 17.7% 28.0% No specific place 14.2% 10.0% 9.8% 9.2% 4.3% 9.5% Type of Handwashing Device (% HHs): Tap, faucet 14.3% 20.9% 23.4% 30.3% 45.0% 27.4% Homemade water tap 38.2% 40.9% 48.3% 48.2% 41.5% 43.5% Basin, bucket 30.7% 28.0% 18.7% 12.9% 8.1% 19.2% Other 16.8% 10.1% 9.5% 8.5% 5.4% 9.8% Water is available at place for washing hands (% HHs) 96.1% 97.6% 97.3% 99.3% 99.3% 98.0% Soaps Available at Place for Washing Hands (% HHs): Bar soap 28.8% 34.4% 42.4% 51.3% 71.1% 45.8% Liquid/dishwashing soap 10.3% 16.1% 18.4% 17.4% 25.4% 17.6% Powder/laundry soap/detergent 61.3% 61.0% 62.9% 61.1% 63.0% 61.9% No soap observed 7.6% 11.7% 6.3% 4.2% 2.4% 6.4% Ash, Mud at Place for Washing Hands (% HHs): Ash 1.4% 0.9% 1.3% 0.9% 0.5% 1.0% Mud 4.8% 6.0% 4.0% 1.9% 1.6% 3.6% Ash and Mud 4.4% 3.2% 3.2% 1.7% 0.9% 2.6% Neither ash nor mud 89.4% 89.9% 91.5% 95.5% 96.9% 92.7% No cleansing agents at place for HW (no soap, nor ash, nor mud observed) (% HHs) 6.8% 10.9% 5.3% 4.3% 3.1% 6.0% Table 8B presents the findings for the same set of vari- inside the food preparation facility. In 44.3% of house- ables in regards to the place used for handwashing before holds the handwashing device is a homemade water tap, preparing food, eating, or feeding children. A total of and in 27.1% it is a tap or faucet. However, in the 37.1% of households reported that family members usu- wealthiest households a tap or faucet is the most com- ally use a different place for washing hands at these times mon device (44.8%). Again, in nearly all households than that used after going to the toilet. If the respondent water was observed at the place reported to be used for indicated the same place for washing hands at all critical washing hands before preparing food or feeding a child times, the results from Table 8A are reported. The find- (98.0%), and in 98.2% soap was observed. Powder soap ings show that 15.6% of the devices used for handwash- or detergent was again the most commonly observed ing when preparing food or feeding a child are located handwashing agent (67.8%), but bar soap was likewise 22 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 8B: OBSERVATION OF A PLACE FOR WASHING HANDS WHEN PREPARING FOOD OR FEEDING A CHILD Wealth Quintile 1st 2nd 3rd 4th 5th Total Location of Handwashing Device (% HHs): Inside food preparation facility 6.5% 9.9% 9.8% 18.2% 33.0% 15.6% No specific place 15.4% 10.1% 10.3% 9.5% 4.4% 9.9% Type of Handwashing Device (% HHs): Tap, faucet 15.5% 20.5% 22.1% 29.0% 44.8% 27.1% Homemade water tap 39.2% 41.7% 49.2% 49.1% 41.9% 44.3% Basin, bucket 31.1% 28.9% 20.0% 13.8% 8.2% 19.8% Other 14.2% 8.9% 8.7% 8.1% 5.1% 8.8% Water is available at place for washing hands (% HHs) 97.2% 98.3% 98.1% 99.4% 98.8% 98.4% Soaps Available at Place for Washing Hands (% HHs): Bar soap 30.1% 34.6% 42.9% 50.8% 64.9% 45.3% Liquid/dishwashing soap 11.8% 16.5% 19.9% 20.0% 21.1% 18.0% Powder/ laundry soap/detergent 69.1% 67.5% 71.6% 67.3% 64.2% 67.8% No soap observed 7.7% 12.9% 6.9% 4.4% 4.6% 7.2% Ash, Mud at Place for Washing Hands (% HHs): Ash 0.9% 0.8% 0.6% 0.0% 0.7% 0.6% Mud 4.3% 4.9% 3.4% 2.1% 1.6% 3.2% Ash and mud 3.5% 2.0% 1.2% 1.4% 1.1% 1.8% Neither ash nor mud 91.3% 92.4% 94.9% 96.5% 96.5% 94.4% No cleansing agents at place for HW (no soap, nor ash, nor mud observed) (% HHs) 6.5% 12.7% 7.4% 5.1% 6.1% 7.6% common and observed in 45.3% of households, followed the fingernails, palms, and fingerpads of the caretaker and by liquid soap in 18.0% of households. Finally, in 94.4% recorded their appearance on a scale of visibly dirty, un- of the households the interviewer observed neither ash clean appearance, and clean appearance. Both palms and nor mud at the place for washing hands, in 3.2% of the fingerpads were observed to be clean for 78.7% and 78.2% households only mud was observed, and in 1.8% of the of caretakers respectively, and fingernails were less clean households both ash and mud was observed. Again, the looking (63.4%). Around 20% of palms and fingerpads proportion of households with no cleansing agent avail- appeared unclean, as did nearly one-third of caretaker's able at the observed place for handwashing is very low fingernails. The observed cleanliness of hands does appear (7.6%). to be associated with socioeconomic status, most notably the appearance of fingernails, which were observed to have An additional objective indicator of caretaker hygiene was an unclean appearance in 37.0% of caretakers in the low- the observation of the caretaker's hands. During this por- est wealth quintile, compared with 23.6% of those in the tion of the survey the interviewer asked to look at highest quintile. The results are shown in Table 9. www.wsp.org 23 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 9: OBSERVATION OF CAREGIVER'S HANDS BY WEALTH QUINTILE Wealth Quintile 1st 2nd 3rd 4th 5th Total Caregiver's Fingernails Appear to Have: Visible dirt 7.8% 5.9% 6.3% 4.5% 1.9% 5.3% Unclean appearance 37.0% 34.3% 32.8% 27.9% 23.6% 31.1% Clean appearance 54.7% 59.6% 60.8% 67.5% 74.5% 63.4% Caregiver's Palms Appear to Have: Visible dirt 3.5% 1.9% 1.9% 1.3% 1.0% 1.9% Unclean appearance 20.5% 20.7% 21.5% 18.0% 15.4% 19.2% Clean appearance 75.5% 77.2% 76.4% 80.6% 83.6% 78.7% Caregiver's Fingerpads Appear to Have: Visible dirt 2.9% 2.1% 1.6% 1.0% 1.0% 1.7% Unclean appearance 22.2% 21.0% 22.5% 17.5% 16.1% 19.9% Clean appearance 74.4% 76.7% 75.7% 81.4% 83.0% 78.2% 3.3 Diarrhea, Acute Lower Respiratory Infection, and Anemia Prevalence Recent health histories were obtained from caretakers for all children younger than five in the household. Symptoms that were prompted included fever, cough, congestion, di- arrhea related symptoms, nausea, vomiting, stomach pain or cramps, and refusal to eat. The findings presented below focus on the prevalence of diarrhea and acute lower respira- tory infection in the under five population of the sample. Diarrhea was defined as the reported presence of three or more loose or watery stools over a 24-hour period, or one or more stools with blood and/or mucus present in the stool (Baqui et al. 1991) using the symptom data obtained from the child health histories. Acute lower respiratory infection (ALRI) was defined using the clinical case definition of the World Health Organization (WHO 2005), which diagno- ses a child as having an ALRI when he/she presents the fol- lowing symptoms: constant cough or difficulty breathing, and raised respiratory rate (>60 breaths per minute in chil- dren less than 60 days of age, >50 breaths per minute for children between 60 ­ 364 days of age, >40 per minute for children between 1­5 years of age). A summary of diarrhea, ALRI, and anemia prevalence in A child is tested for anemia the sampled population of children under five is shown 24 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 10: DIARRHEA, ALRI, AND ANEMIA PREVALENCE BY POVERTY STATUS AND ACCESS TO PLACE FOR WASHING HANDS (CHILDREN <5) Access to Place for Washing Hands Poor with Soap and Water (% HHs) Total Yes No Yes No Child had diarrhea symptoms in previous 48 hours (% children) 0.6% 0.8% 0.7% 0.7% 0.7% Child had diarrhea symptoms in previous week (% children) 1.2% 1.1% 1.2% 1.1% 1.2% Child had diarrhea symptoms in past 14 days (% children) 1.2% 1.3% 1.2% 1.3% 1.3% Child had ALRI symptoms in previous 48 hours (% children) 0.6% 0.4% 0.5% 0.3% 0.5% Child had ALRI symptoms in previous three days (% children) 0.9% 0.6% 0.8% 0.3% 0.7% Anemia (Hb <110 g/L) 34.6% 28.7% 31.8% 31.1% 31.6% in Table 10. Caregiver reported intestinal symptoms for between caregiver reported diarrhea and ALRI and the 6.0% of children; however, diarrhea prevalence as de- objective health measures is high. fined is less than 1% among children under five during the 48 hours prior to the survey, and just over 1% for These findings are cross tabulated by both poverty status both seven and 14 days prior to the survey. Similarly low and access to an observed place for handwashing with soap prevalence rates of ALRI were found. Although 21.6% and water. While some of the findings may appear counter- of children in the sample had caregiver reported respira- intuitive, such as the slightly higher two-day and 14-day tory symptoms in the two weeks prior to the survey, the prevalence of diarrhea in the non-poor households, scien- prevalence of clinically defined ALRI in the sample is tifically these findings are no different. Access to a place for low: just 0.5% of children had symptoms consistent with washing hands likewise does not appear associated with ALRI in the previous 48 hours and a three-day preva- prevalence of diarrhea or ALRI symptoms. However, we do lence of 0.7%. Contrary to estimates based on the find the anemia prevalence of 34.6% among children from VNDHS 2002 and MICS3 2006 data,17 the findings in poor households is significantly higher (t=3.46) than those relation to caregiver reported diarrhea and ALRI preva- from non-poor households (28.7%). lence for this sample of children under five is low. It is important to note, however, that relative to more objec- Diarrhea and ALRI prevalence by province are shown in tive health measures collected as part of the survey, such Table 11. We find that reported diarrhea prevalence is below as child anthropometrics and anemia, the findings are average in Tien Giang (two-day 0.4%, seven-day 0.7%, and internally consistent. Moreover, they are consistent 14-day 0.8%), whereas children in Hung Yen have the high- across the Scaling Up countries, where the correlation est reported seven-day (1.7%) and 14-day (1.8%) diarrhea prevalence. In Thanh Hoa children have higher than average 17 The nationally representative VNDHS 2002 survey reported two-week diarrhea ALRI prevalence for both 48 hour (1.0%) and three-day prevalence of 11%, whereas the nationally representative MICS 2006 (third round) survey reported two-week diarrhea prevalence of 6.8%. (1.1%) caregiver reported prevalence. www.wsp.org 25 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings Diarrhea prevalence and treatment by wealth quintile is quintile. Treatment with a pill or syrup for respiratory shown in Table 12. On average, 54.1% of caregivers symptoms19 was very common, with 90.2% of caretakers with children presenting intestinal symptoms 18 in the using this method of treatment. Just 3.4% opted not to two weeks prior to the survey treated the child with a treat, and another 11.1% used another type of treatment pill or syrup and 8.2% used an oral rehydration solu- such as an IV or traditional remedy. The findings are tion (ORS). Another 6.9% used another treatment consistent across wealth quintiles. such as an intravenous fluid injection (IV), traditional remedies, or a homemade sugar or salt water solution, As part of the child health history, caregivers were asked and 16.5% did not seek treatment for the symptoms. whether they sought medical advice for their child during These figures varied only slightly by wealth quintile, the past two weeks for diarrhea or respiratory symptoms. with those households in the 2nd and 3rd quintiles The findings are shown in Table 14. Although reported most likely to report treating intestinal symptoms with prevalence of diarrhea and ALRI is very low in the sample, a pill or syrup. ORS was more commonly given as a a high percentage of caregivers sought medical advice treatment in the higher wealth quintiles, while treat- (46.7%), with the majority of treatment sought from pri- ment with another method was higher than average vate providers (50.4%). This is followed by 26.1% of house- (11.8%) in the lowest quintile. holds who sought treatment from a pharmacist, and 10.0% who took the child for an overnight stay at a hospital or Table 13 shows ALRI prevalence and treatment by wealth clinic. Caregivers from the poorest households reported quintile. Children from households classified as poorest taking their child for an overnight stay at a hospital or clinic in the study sample show higher than average reported (16.0%) due to illness more than the average for the entire prevalence of ALRI (1.1% and 1.2% respectively for sample, while they took their child for a day visit to the two-day and seven-day prevalence). However, there is doctor less than average (38.2%). For all wealth quintiles higher than average reported prevalence in the 4th wealth medical advice was more often sought from private TABLE 11: DIARRHEA AND ALRI PREVALENCE BY PROVINCE (CHILDREN <5) Province Hung Yen Thanh Hoa Tien Giang Total Child had diarrhea symptoms in previous 48 hours (% children) 0.80% 0.90% 0.40% 0.70% Child had diarrhea symptoms in previous week (% children) 1.70% 1.20% 0.70% 1.20% Child had diarrhea symptoms in past 14 days (% children) 1.80% 1.30% 0.80% 1.30% Child had ALRI symptoms in previous 48 hours (% children) 0.40% 1.00% 0.30% 0.50% Child had ALRI symptoms in previous 72 hours (% children) 0.70% 1.10% 0.40% 0.70% 18 Intestinal symptoms include: stomach pain or cramps, nausea, vomiting, three or 19 more bowel movements in one day and one night, water or soft stool, mucus or Respiratory symptoms include: cough, congestion, panting/wheezing, or difficulty blood in stool, or refusal to eat. breathing. 26 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 12: DIARRHEA PREVALENCE AND TREATMENT BY WEALTH QUINTILE (CHILDREN <5) Wealth Quintile 1st 2nd 3rd 4th 5th Total Child had diarrhea symptoms in previous 48 hours (% children) 0.8% 0.3% 0.6% 0.9% 0.8% 0.7% Child had diarrhea symptoms in previous week (% children) 1.2% 0.8% 1.3% 1.4% 1.1% 1.2% Child had diarrhea symptoms in previous 14 days (% children) 1.2% 0.8% 1.3% 1.7% 1.3% 1.3% Treatment Sought for Intestinal Symptoms: No treatment 17.6% 8.2% 15.9% 20.8% 19.6% 16.5% Pill or syrup 50.0% 63.3% 56.8% 50.0% 50.0% 54.1% Oral rehydration solution 5.9% 8.2% 9.1% 8.3% 8.9% 8.2% Other 11.8% 6.1% 4.5% 8.3% 5.4% 6.9% TABLE 13: ALRI PREVALENCE AND TREATMENT BY WEALTH QUINTILE (CHILDREN <5) Wealth Quintile 1st 2nd 3rd 4th 5th Total Child had acute lower respiratory infection symptoms in previous 48 hours (% children) 1.1% 0.2% 0.3% 0.8% 0.2% 0.5% Child had acute lower respiratory infection symptoms in previous 72 hours (% children) 1.2% 0.8% 0.5% 0.9% 0.2% 0.7% Treatment Sought for Respiratory Symptoms: No treatment 2.3% 4.5% 2.5% 4.9% 2.6% 3.4% Pill or syrup 88.6% 90.9% 95.1% 90.3% 86.3% 90.2% Other 12.6% 11.7% 8.0% 8.1% 15.7% 11.1% TABLE 14: CARE-SEEKING BEHAVIOR FOR CHILD ILLNESS BY WEALTH QUINTILE Wealth Quintile 1st 2nd 3rd 4th 5th Total Caregiver Sought Medical Advice of (% Caregivers): Did not seek 5.3% 4.2% 6.3% 6.7% 5.4% 5.6% Day visit to doctor 38.2% 47.9% 46.8% 46.4% 54.0% 46.7% Overnight stay at hospital or clinic 16.0% 10.7% 6.3% 10.9% 6.3% 10.0% Pharmacist 26.2% 27.0% 28.3% 24.7% 24.1% 26.1% Herbalist 0.0% 1.4% 0.4% 0.0% 0.0% 0.4% Care sought from public provider (% caregivers) 37.6% 38.3% 35.6% 38.1% 35.2% 37.0% Care sought from private provider (% caregivers) 46.9% 51.5% 50.9% 49.3% 53.5% 50.4% www.wsp.org 27 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 15: HOUSEHOLDS WITH LOST HOURS DUE TO lowest wealth quintile had higher than average presence of ane- CHILD ILLNESS BY WEALTH QUINTILE AND PROVINCE mia (35.5%), measured by hemoglobin concentration, sug- HH Lost Hours Number of Hours gesting that anemia is inversely associated with household Due to Child Lost Due to Child wealth. The findings by province indicate a higher than average Illness (% HHs) Illness (average) prevalence of anemia in Thanh Hoa. While around one-third Wealth Quintile of samples taken from children in the sample indicate presence 1st 15.9% 4.6 of anemia, in Thanh Hoa province this figure is 47.9%. On 2nd 15.2% 4.7 average anemia was present in 31.7% of the samples taken. 3rd 17.3% 5.0 4th 18.1% 5.0 3.4 Child Growth Measures 5th 19.0% 5.2 The survey included baseline child growth measures of chil- Province dren under the age of two, including head and arm circum- Hung Yen 26.6% 4.9 ference, length, and weight. To analyze the child growth Thanh Hoa 3.7% 4.3 findings, anthropometric Z-scores were assigned by compar- Tien Giang 18.5% 5.0 ing children in the sample to the WHO reference population Total 17.0% 4.9 median and standard deviation for each of the aforemen- tioned variables (WHO 2006, 2007). The reference popula- tion is designed to be internationally applicable regardless of ethnicity, socioeconomic status, or feeding practices. providers than public providers. Overall in the sample, care seeking behavior is quite high: only 5.6% of caregivers The Z-score for each measure indicates the number of stan- chose not to seek medical advice when their child was ill dard deviation units from the median of the reference popu- during the two weeks prior to the survey. lation. The WHO guidelines for child growth and malnutrition use a Z-score cutoff of less than ­2 standard Finally, caregivers were asked whether they had lost working deviations (SD) below the median of the reference popula- hours in the previous 14 days due to their child's reported symp- tion for low weight-for-age, a measure of malnutrition, and toms. The findings, reported in Table 15, reveal that in an aver- less than ­3 SDs from the median indicating that a child is age of 17.1% of households, one or more primary caretakers lost severely malnourished. Low height-for-age, a measure of lin- time due to the illness of a child over the past 14 days. This is a ear growth, of ­2 SDs below the median indicates that a strikingly high percentage given that the prevalence of diarrhea and ALRI in the population is low. The figure is higher than TABLE 16: ANEMIA PREVALENCE BY WEALTH QUINTILE average at the higher wealth quintiles, which may be due to the AND PROVINCE (CHILDREN <2) perception that time off from unpaid or informal work (more % Children with Anemia typical of poorer households) to care for a sick child is not lost (Hb <110 g/L) time. On average, primary caretakers reported 4.9 hours of lost Wealth Quintile time. There is little variation in the number of hours lost by 1st 35.5% wealth quintile. However, we find large differences between 2nd 34.8% provinces in time lost to care for a sick child. Just 3.7% of house- 3rd 31.3% holds in Thanh Hoa reported lost time, while 26.6% of house- 4th 31.0% holds in Hung Yen reported lost time. Little variation is found, 5th 25.9% however, in the number of lost hours across provinces. Province Hung Yen 23.8% Hemoglobin concentrations were obtained from children be- Thanh Hoa 47.9% tween six months and two years of age in order to estimate the Tien Giang 26.2% percentage suffering from anemia. These results are reported in Table 16. Samples taken from children in households in the Overall 31.7% 28 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings poorest households, and over 10% are malnourished in the two lowest wealth quintiles. Stunting of children appears to be highest in both Thanh Hoa and Hung Yen provinces, while Tien Giang fares better on all three indicators. The histograms of the Z-scores for each child growth measure displayed in Figure 2 provide an additional illustration of the prevalence of inadequate child growth. Children outside of the normal range of healthy growth are plotted below the ­2 SD and above the +2 SD cutoff points on the graph. Children who are malnourished are represented between the ­5 and ­2 SD cutoff point on the weight-for-age Z-score histogram, while those who are stunted, and those who are wasted are represented An anthropometrician prepares to measure a child's arm between the ­6 and ­2 SD cutoff points in the length/height- circumference during a household interview for-age Z-score and weight-for-length/height histograms respec- tively.20 All measures besides arm circumference were found to be lower on average than the WHO reference population me- dian, as indicated by a red vertical line on the graph. child is short for his or her age and is moderately or severely stunted. Stunting is an indication of chronic malnutrition. Table 18A presents average Z-scores for the six child-growth Finally, a low weight-for-height of ­2 SDs below the refer- measures disaggregated by wealth quintile. All average ence median indicates wasting, which indicates a recent nu- Z-scores are within 1 SD of the reference population median, tritional deficiency rather than chronic malnutrition. indicating that on average the children in the sample exhibit healthy growth, although average Z-scores for all measures As shown in Table 17 there is a sizeable proportion of chil- except arm-circumference for age are below the reference dren under two in the sample that are stunted, malnourished, and/or wasted. This is particularly notable when the findings 20 Calculated Z-scores below ­5 and above 5 for weight-for-age and Z-scores below ­6 and above 6 for height-for-age and weight-for-height are considered to be are disaggregated by wealth and province. Nearly one-fifth of implausible and therefore are not included in the prevalence statistics presented in the children under two in the sample are stunted in the Table 18. TABLE 17: PREVALENCE OF MALNUTRITION, STUNTING, AND WASTING BY WEALTH QUINTILE AND PROVINCE (CHILDREN <2) Malnourished Stunted Wasted (% Children ­2 SDs (% Children ­2 SDs (% Children ­2 SDs Weight-for-Age Z-Score) Height-for-Age Z-Score) Weight-for-Height Z-Score) Wealth Quintile 1st 11.4% 19.4% 7.3% 2nd 11.5% 15.9% 5.6% 3rd 7.1% 13.2% 5.6% 4th 7.4% 11.0% 6.5% 5th 5.2% 10.7% 3.5% Province Hung Yen 10.3% 15.6% 6.5% Thanh Hoa 8.2% 16.3% 5.7% Tien Giang 7.1% 11.0% 4.9% Total 8.5% 14.0% 5.7% www.wsp.org 29 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings FIGURE 2: HISTOGRAM OF CHILD GROWTH MEASURES (Z-SCORES) FOR CHILDREN <2 Arm circumference-for-age z-score Weight-for-age z-score 15 15 10 10 Percent Percent 5 5 0 0 ­5sd 2sd 0 2sd 5sd ­5sd 2sd 0 2sd 5sd Z-score Z-score Length/height-for-age z-score BMI-for-age z-score 15 15 10 10 Percent Percent 5 5 0 0 ­5sd 2sd 0 2sd 5sd ­5sd 2sd 0 2sd 5sd Z-score Z-score Weight-for-length z-score BMI-for-age z-score 15 15 10 10 Percent Percent 5 5 0 0 ­5sd 2sd 0 2sd 5sd ­5sd 2sd 0 2sd 5sd Z-score Z-score 30 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings TABLE 18A: CHILD GROWTH MEASURES (Z-SCORES) BY WEALTH QUINTILE (CHILDREN <2) Wealth Quintile 1st 2nd 3rd 4th 5th Total Average arm circumference-for-age Z-score ­0.01 0.18 0.34 0.17 0.60 0.26 Average weight-for-age Z-score ­0.90 ­0.64 ­0.05 ­0.45 ­0.31 ­0.47 Average length-for-age Z-score ­0.96 ­0.73 ­0.67 ­0.53 ­0.42 ­0.66 Average body mass index-for-age Z-score ­0.41 ­0.19 0.55 ­0.17 ­0.07 ­0.06 Average weight-for-length/height Z-score ­0.56 ­0.38 0.47 ­0.23 ­0.12 ­0.16 Average head circumference-for-age Z-score 0.79 ­0.75 ­0.61 ­0.63 0.03 ­0.23 TABLE 18B: CHILD GROWTH MEASURES (Z-SCORES) BY PROVINCE (CHILDREN <2) Province Hung Yen Thanh Hoa Tien Giang Total Average arm circumference-for-age Z-score 0.33 0.12 0.29 0.26 Average weight-for-age Z-score ­0.52 ­0.29 ­0.57 ­0.47 Average length-for-age Z-score ­0.73 ­0.81 ­0.48 ­0.66 Average body mass index-for-age Z-score ­0.05 0.33 ­0.36 ­0.06 Average weight-for-length/height Z-score ­0.16 0.18 ­0.43 ­0.16 Average head circumference-for-age Z-score ­0.34 0.08 ­0.38 ­0.23 TABLE 19: CHILD GROWTH MEASURES (Z-SCORES) BY POVERTY STATUS AND ACCESS TO PLACE FOR WASHING HANDS (CHILDREN <2) Access to Place for Washing Hands with Soap Poor and Water (% HHs) Total Yes No Yes No Average arm circumference-for-age Z-score 0.18 0.33 0.32 0.00 0.26 Average weight-for-age Z-score ­0.51 ­0.43 ­0.41 ­0.72 ­0.47 Average length/height-for-age Z-score ­0.80 ­0.52 ­0.61 ­0.89 ­0.66 Average body mass index-for-age Z-score 0.02 ­0.13 ­0.01 ­0.24 ­0.06 Average weight-for-length/height Z-score ­0.13 ­0.20 ­0.12 ­0.34 ­0.16 Average head circumference-for-age Z-score ­0.10 ­0.36 ­0.39 0.42 ­0.23 population. While still within the healthy range for children mass index-for-age (+0.33), weight-for-length (+0.18), and under two, those in the lowest wealth quintile exhibit lower head-circumference-for-age (+0.08). weight-for-age (­0.90 SDs lower than median) and length- for-age (­0.96 SDs lower than median). Table 19 presents these same child growth measures disag- gregated by poverty status and access to a place for washing There are few evident differences in child-growth measures hands. There are some intuitive findings regarding the rela- by province, shown in Table 18B. However, the children in tionship between poverty and nutritional status of children. the Thanh Hoa sample do appear to be shorter on average Children from poor households are found to have lower (length-for-age of ­0.81 SD), heavier on average (weight- weight-for-age (­0.51 SD) and length-for-age (­0.80 SD) for-age of ­0.29 SD) and have higher than average body than children from non-poor households. Children from www.wsp.org 31 Findings from the Impact Evaluation Baseline Survey in Vietnam Findings poor households also have higher body mass index-for-age population mean, we find a negative relationship between (+0.02 SD) and weight-for-length (­0.13 SD) than non- Z-score and age in months for the remainder of the child poor households. An important association for the study, we growth measures for both males and females. The findings sug- find that all Z-scores are higher for households with a place gest the gap between the sample mean and the reference popu- for washing hands with soap and water than for those with- lation median widens as children age from 0­24 months, out, except for head circumference-for-age. indicating that the nutritional status of children in the sample deteriorates over time. This growth pattern is typical among Figures 3A­3C present scatterplots of the average Z-score for children under two in developing countries.21 each growth measure disaggregated by age in months and sex. Locally weighted polynomial regression (lowess) estimates are Another notable finding is the absence of a gap between overlaid on the scatterplot to capture the shape of the relation- male and female child growth, implying that the physio- ship between age in months and Z-score for male and female logical needs of young children in the sample are not met children separately. While the survey is a cross section of house- differentially as a result of the child's gender. However, it holds, and we cannot observe the evolution of child growth is not evident whether this trend will continue. In the ab- measures over time for the children sampled, we can approxi- sence of panel data on each child in the sample we cannot mate the trend in early child development for the sample pop- know whether the downward trend shown for arm- ulation by analyzing the average Z-scores for children under circumference-for-age, weight-for-age and height-for-age two years at each age. With the exception of average body mass will continue as females reach age two years and beyond. index-for-age and weight-for-length Z-scores, which appear to level off after around five months of age and hover around the 21 Victora et al. 2010. FIGURE 3A: ARM AND HEAD CIRCUMFERENCE Z-SCORES BY SEX AND MONTHS OF AGE (CHILDREN <2) Arm circumference-for-age z-score Head circumference-for-age z-score 3 3 2 2 Z-score Z-score 1 1 0 0 -1 -1 0 5 10 15 20 25 0 5 10 15 20 25 Age in months Age in months Male Lowess Male Male Lowess Male Female Lowess Female Female Lowess Female 32 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Findings FIGURE 3B: WEIGHT-FOR-AGE AND LENGTH-FOR-AGE Z-SCORES BY SEX AND MONTHS OF AGE (CHILDREN <2) Weight-for-age z-score Length/height-for-age z-score 3 3 2 2 Z-score Z-score 1 1 0 0 -1 -1 0 5 10 15 20 25 0 5 10 15 20 25 Age in months Age in months Male Lowess Male Male Lowess Male Female Lowess Female Female Lowess Female FIGURE 3C: BMI-FOR-AGE AND LENGTH-FOR-HEIGHT Z-SCORES BY SEX AND MONTHS OF AGE (CHILDREN <2) BMI-for-age z-score Weight-for-length/height z-score 2 2 1 1 0 0 Z-score Z-score ­1 ­1 ­2 ­2 -3 -3 0 5 10 15 20 25 0 5 10 15 20 25 Age in months Age in months Male Lowess Male Male Lowess Male Female Lowess Female Female Lowess Female www.wsp.org 33 IV. Conclusion The findings presented in this report provide a snapshot of figures on the prevalence of diarrhea and ALRI are good baseline characteristics of the target population in regards news for the Vietnamese population, they are likely too low to household demographics, socioeconomic situation, to enable detection of an impact of the intervention on di- mother's and other caretaker's handwashing behavior, and arrhea outcomes. Still, the evaluation study hopes to mea- key child health and development indicators. Limited base- sure and learn about the impact of the intervention on line knowledge of the critical times for washing hands indi- handwashing behavior change that will be used to guide cates that there is scope for improving handwashing future projects and policy both in Vietnam and globally. behavior in the target population, particularly among the poorest. Moreover, while baseline diarrhea and ALRI preva- As outlined in the methodology section, the impact evalua- lence are both low in relation to the other Global Scaling tion study utilizes a series of household and community Up Handwashing project countries, the poorest are still at a surveys. These include the baseline, four waves of longitudi- disadvantage, especially with regard to child growth and nal monitoring, and post-intervention follow-up question- development. naires. At the time of this report's publication, longitudinal data collection is completed, and post-intervention data In addition to providing useful information for the design collection is scheduled to begin by the end of 2010. Data of the intervention, the data presented here will be used to analysis and impact assessments will be conducted soon evaluate the impact of the Vietnam handwashing project on after, and a full impact evaluation report of the handwash- child health and caretaker productivity, and to track changes ing project will be published by the end of 2011. in handwashing with soap behavior. While the baseline www.wsp.org 35 Findings from the Impact Evaluation Baseline Survey in Vietnam References References Baqui, A. H., R. E. Black, M. Yunus, A. R. Hoque, H. R. Anemia: A Report of the International Nutritional Ane- Chowdhury, R. B. Sack. 1991. "Methodological Issues mia Consultative Group (INACG). Washington, DC: in Diarrhoeal Diseases Epidemiology: Definition of Di- The Nutrition Foundation. arrhoeal Episodes." Int J Epidemiol 20 (4): 1057­63. United Nations Development Program. 2010. Achieving Black, R. E., L. H. Allen, Z. A. Bhutta, et al. 2008. "Ma- the Millennium Development Goals in an Era of Global ternal and Child Undernutrition: Global and Regional Uncertainty. Asia-Pacific Regional Report 2009/10. Bang- Exposures and Health Consequences." The Lancet 371 kok, Thailand: United Nations. (9608): 243­260. Victora, C. G., L. Adair, C. Fall, P. C. Hallal, R. Martorell, Curtis, V. A., L. O. Danquah, R. V. Aunger. 2009. L. M. Richter. 2008. Maternal and Child Undernutri- "Planned, Motivated, and Habitual Hygiene Behavior: tion: Consequences for Adult Health and Human Capi- An Eleven Country Review." Health Education Research tal. The Lancet 371 (9609): 340­357. 24 (4): 655­73. Victora, C. G., M. de Onis, P. C. Hallal, M. Blössner, Engle, P. L., M. M. Black, J. R. Behrman, et al. 2007. R. Shrimpton. 2010. "Worldwide Timing of Growth "Strategies to Avoid the Loss of Developmental Poten- Faltering: Revisiting Implications for Interventions." tial in More Than 200 Million Children in the Devel- Pediatrics 125: e473­e480. oping World." The Lancet 369 (9557): 229­242. Walker, S. P., T. D. Wachs, J. Meeks Gardner, et al. Filmer, D. L., l. Pritchett. 2001. "Estimating Wealth Ef- 2007. "Child Development: Risk Factors for Adverse fects without Expenditure Data--or Tears: An Applica- Outcomes in Developing Countries." The Lancet 369 tion to Educational Enrollments in States of India." (9556): 145­157. Demography 38 (1): 115­132. WHO/UNICEF Joint Monitoring Programme for Water Grantham-McGregor, S., Y. B. Cheung, S. Cueto, P. Supply and Sanitation Website. Accessed June 20, Glewwe, L. Richter, B. Strupp. 2007. "Developmental 2010. http://www.wssinfo.org/definitions/infrastructure Potential in the First 5 Years for Children in Develop- .html. ing Countries." The Lancet 369 (9555): 60­70. World Health Organization. 2005. Pocket Book of Hospital Habicht, J. P. 1974. "Estandarización de Métodos Epide- Care for Children: Guidelines for the Management of Com- miológicos Cuantitativos sobre el Terreno" ["Standard- mon Illnesses with Limited Resources. Geneva: WHO Press. ization of Quantitative Epidemiological Methods in the World Health Organization. 2006. WHO Child Growth Field"]. Bol Oficina Sanit Panam 76 (5): 375­384. Standards: Length/Height-for-Age, Weight-for-Age, Hernan, M. A., S. Hernandez-Diaz, J. M. Robins. 2004. Weight-for-Length, Weight-for-Height and Body Mass "A Structural Approach to Selection Bias." Epidemiology Index-for-Age: Methods and Development. Geneva: 15: 615­625. WHO Press. Nestel, P. and INACG Steering Committee. 2002. "Adjust- World Health Organization. 2007. WHO Child Growth ing Hemoglobin Values in Program Surveys." Washing- Standards: Head Circumference-for-Age, Arm Circum- ton, DC: International Life Sciences Institute. ference-for-Age, Triceps Skinfold-for-Age and Subscapular Stoltzfus, R. J., M. L. Dreyfus. 1999. Guidelines for the Use Skinfold-for-Age: Methods and Development. Geneva: of Iron Supplements to Prevent and Treat Iron Deficiency WHO Press. 36 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20A: COMMUNES SELECTED TO RECEIVE TREATMENT 1 (IPC + MASS MEDIA) No. Province District Commune Population 1 Hung Yen An Thi Bac Son 7,590 2 Hung Yen An Thi Bai Say 6,300 3 Hung Yen An Thi Thi Tran An Thi 8,310 4 Hung Yen Kim Dong Chinh Nghia 6,730 5 Hung Yen Kim Dong Ngoc Thanh 6,680 6 Hung Yen Kim Dong Thi Tran Luong Bang 9,675 7 Hung Yen Kim Dong Tho Vinh 6,825 8 Hung Yen Kim Dong Vinh Xa 6,931 9 Hung Yen Kim Dong Vu Xa 5,510 10 Hung Yen Phu Cu Nhat Quang 5,139 11 Hung Yen Phu Cu Phan Sao Nam 5,470 12 Hung Yen Phu Cu Thi Tran Tran Cao 5,882 13 Hung Yen Phu Cu Tong Tran 7,172 14 Hung Yen Tien Lu Hoang Hanh 5,310 15 Hung Yen Tien Lu Nhat Tan 8,050 16 Hung Yen Tien Lu Tan Hung 5,350 17 Hung Yen Tien Lu Thien Phien 7,205 18 Hung Yen Tien Lu Trung Dung 6,003 19 Hung Yen Yen My Nghia Hiep 6,250 20 Hung Yen Yen My Ngoc Long 5,441 21 Hung Yen Yen My Tan Viet 8,440 22 Hung Yen Yen My Thanh Long 8,416 23 Hung Yen Yen My Thi Tran Yen My 13,184 24 Hung Yen Yen My Trung Hung 7,240 25 Thanh Hoa Quang Xuong Quang Hop 6,342 26 Thanh Hoa Quang Xuong Quang Linh 4,213 27 Thanh Hoa Quang Xuong Quang Loc 7,215 28 Thanh Hoa Quang Xuong Quang Loi 6,783 29 Thanh Hoa Quang Xuong Quang Minh 4,577 30 Thanh Hoa Quang Xuong Quang Nhan 6,900 31 Thanh Hoa Quang Xuong Quang Van 6,192 32 Thanh Hoa Thach Thanh Thach Son 6,757 33 Thanh Hoa Thach Thanh Thanh Van 6,418 (Continued ) www.wsp.org 37 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20A: (Continued) No. Province District Commune Population 34 Thanh Hoa Thach Thanh Thanh Vinh 6,064 35 Thanh Hoa Tinh Gia Binh Minh 6,048 36 Thanh Hoa Tinh Gia Hai An 5,783 37 Thanh Hoa Tinh Gia Hai Chau 10,000 38 Thanh Hoa Tinh Gia Hung Son 4,380 39 Thanh Hoa Tinh Gia Tan Dan 5,880 40 Thanh Hoa Tinh Gia Truc Lam 6,125 41 Thanh Hoa Trieu Son Hop Tien 4,081 42 Thanh Hoa Trieu Son Thi Tran Trieu Son 7,741 43 Thanh Hoa Trieu Son Tho The 5,022 44 Thanh Hoa Trieu Son Xuan Thinh 5,339 45 Tien Giang Cai Lay Hoi Xuan 953 46 Tien Giang Cai Lay My Hanh Dong 9,134 47 Tien Giang Cai Lay My Long 9,549 48 Tien Giang Cai Lay My Phuoc Tay 13,318 49 Tien Giang Cai Lay My Thanh Nam 13,316 50 Tien Giang Cai Lay Phu Nhuan 9,658 51 Tien Giang Cai Lay Thi Tran Cai Lay 27,898 52 Tien Giang Chau Thanh Diem Hy 10,014 53 Tien Giang Chau Thanh Kim Son 10,919 54 Tien Giang Chau Thanh Long Dinh 15,768 55 Tien Giang Chau Thanh Tan Ly Dong 13,456 56 Tien Giang Chau Thanh Thoi Son22 6,128 57 Tien Giang Chau Thanh Vinh Kim 10,908 58 Tien Giang Cho Gao Dang Hung Phuoc 11,499 59 Tien Giang Cho Gao Thi Tran Cho Gao 8,938 60 Tien Giang Go Cong Tay Dong Thanh 11,683 61 Tien Giang Go Cong Tay Thanh Nhut 13,392 62 Tien Giang Go Cong Tay Thanh Tri 10,609 63 Tien Giang Go Cong Tay Yen Luong 6,889 64 Tien Giang Tan Phu Dong Tan Thanh 4,894 65 Tien Giang Tan Phuoc Hung Thanh 6,141 66 Tien Giang Tan Phuoc Phu My 8,255 67 Tien Giang Tan Phuoc Phuoc Lap 890 68 Tien Giang Tan Phuoc Tan Lap 1 5,279 69 Tien Giang Tan Phuoc Tan Lap 2 2,095 70 Tien Giang Tan Phuoc Thanh Hoa 903 Total 533,449 22 Baseline data were collected in Thoi Son commune; however, due to a change in administrative boundaries this commune will not be part of the handwashing project intervention (T1). 38 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20B: COMMUNES SELECTED TO RECEIVE TREATMENT 2 (IPC + DCC + MASS MEDIA) No. Province District Commune Population 1 Hung Yen An Thi Hong Quang 6,475 2 Hung Yen An Thi Nguyen Trai 6,075 3 Hung Yen An Thi Phu Ung 8,190 4 Hung Yen An Thi Quang Vinh 1,598 5 Hung Yen Kim Dong Hung Cuong 4,060 6 Hung Yen Kim Dong Mai Dong 5,615 7 Hung Yen Kim Dong Nghia Dan 6,442 8 Hung Yen Kim Dong Pham Ngu Lao 7,420 9 Hung Yen Kim Dong Toan Thang 10,200 10 Hung Yen Phu Cu Doan Dao 9,844 11 Hung Yen Phu Cu Minh Hoang 5,154 12 Hung Yen Phu Cu Nguyen Hoa 5,012 13 Hung Yen Tien Lu An Vien 8,021 14 Hung Yen Tien Lu Cuong Chinh 8,215 15 Hung Yen Tien Lu Di Che 6,871 16 Hung Yen Tien Lu Hai Trieu 5,350 17 Hung Yen Tien Lu Minh Phuong 3,510 18 Hung Yen Tien Lu Thi Tran Vuong 4,932 19 Hung Yen Yen My Lieu Xa 8,679 20 Hung Yen Yen My Ly Thuong Kiet 6,166 21 Hung Yen Yen My Trung Hoa 11,347 22 Hung Yen Yen My Viet Cuong 3,842 23 Hung Yen Yen My Yen Hoa 5,596 24 Hung Yen Yen My Yen Phu 12,046 25 Thanh Hoa Quang Xuong Quang Dai 5,690 26 Thanh Hoa Quang Xuong Quang Dinh 5,439 27 Thanh Hoa Quang Xuong Quang Dong 4,952 28 Thanh Hoa Quang Xuong Quang Giao 4,822 29 Thanh Hoa Quang Xuong Quang Hai 9,450 30 Thanh Hoa Quang Xuong Quang Khe 7,169 31 Thanh Hoa Quang Xuong Quang Long 6,067 32 Thanh Hoa Quang Xuong Quang Vong 5,781 33 Thanh Hoa Thach Thanh Thanh Tho 5,435 34 Thanh Hoa Tinh Gia Hai Nhan 9,435 35 Thanh Hoa Tinh Gia Hai Yen 4,020 36 Thanh Hoa Tinh Gia Mai Lam 6,120 37 Thanh Hoa Tinh Gia Tinh Hai 6,282 38 Thanh Hoa Trieu Son An Nong 6,375 (Continued ) www.wsp.org 39 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20B: (Continued) No. Province District Commune Population 39 Thanh Hoa Trieu Son Tan Ninh 10,505 40 Thanh Hoa Trieu Son Tho Ngoc 7,284 41 Thanh Hoa Trieu Son Tho Tan 5,022 42 Thanh Hoa Trieu Son Tho Tien 5,562 43 Thanh Hoa Trieu Son Tho Vuc 4,915 44 Thanh Hoa Trieu Son Van Son 7,137 45 Tien Giang Cai Lay Binh Phu 17,284 46 Tien Giang Cai Lay Cam Son 8,892 47 Tien Giang Cai Lay Long Khanh 13,375 48 Tien Giang Cai Lay Long Tien 12,328 49 Tien Giang Cai Lay Long Trung 12,983 50 Tien Giang Cai Lay My Hanh Trung 7,763 51 Tien Giang Cai Lay Tan Hoi 12,256 52 Tien Giang Cai Lay Tan Phong 13,928 53 Tien Giang Cai Lay Thanh Hoa 5,794 54 Tien Giang Chau Thanh Ban Long 893 55 Tien Giang Chau Thanh Binh Trung 8,947 56 Tien Giang Chau Thanh Long An 13,104 23 57 Tien Giang Chau Thanh Phuoc Thanh 9,689 58 Tien Giang Chau Thanh Song Thuan 5,908 59 Tien Giang Chau Thanh Thi Tran Tan Hiep 5,939 60 Tien Giang Cho Gao An Thanh Thuy 13,443 61 Tien Giang Cho Gao Hoa Tinh 5,806 62 Tien Giang Cho Gao My Tinh An 9,206 63 Tien Giang Cho Gao Phu Kiet 10,721 64 Tien Giang Cho Gao Tan Binh Thanh 7,705 65 Tien Giang Cho Gao Tan Thuan Binh 10,416 66 Tien Giang Go Cong Tay Dong Son 10,178 67 Tien Giang Go Cong Tay Long Binh 13,457 68 Tien Giang Go Cong Tay Thanh Cong 4,690 69 Tien Giang Tan Phu Dong Phu Tan 3,643 70 Tien Giang Tan Phuoc My Phuoc 2,653 Total 529,123 23 Baseline data were collected in Phuoc Thanh commune; however, due to a change in administrative boundaries this commune will not be part of the handwashing project intervention (T2). 40 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20C: COMMUNES SELECTED TO SERVE AS CONTROL (MASS MEDIA) No. Province District Commune Population 1 Hung Yen An Thi Cam Ninh 4,650 2 Hung Yen An Thi Ha Le 5,090 3 Hung Yen An Thi Ho Tung Mau 7,690 4 Hung Yen An Thi Hoang Hoa Tham 6,220 5 Hung Yen An Thi Hong Van 5,310 6 Hung Yen Kim Dong Dong Thanh 6,059 7 Hung Yen Kim Dong Duc Hop 7,969 8 Hung Yen Kim Dong Hiep Cuong 7,835 9 Hung Yen Kim Dong Nhan La 3,856 10 Hung Yen Kim Dong Phu Thinh 6,450 11 Hung Yen Kim Dong Song Mai 6,750 12 Hung Yen Phu Cu Dinh Cao 11,361 13 Hung Yen Phu Cu Minh Tan 5,648 14 Hung Yen Phu Cu Minh Tien 5,537 15 Hung Yen Phu Cu Tam Da 5,399 16 Hung Yen Phu Cu Tien Tien 3,741 17 Hung Yen Tien Lu Hung Dao 8,135 18 Hung Yen Tien Lu Le Xa 6,653 19 Hung Yen Tien Lu Ngo Quyen 5,907 20 Hung Yen Tien Lu Phuong Chieu 3,805 21 Hung Yen Tien Lu Thu Sy 9,205 22 Hung Yen Tien Lu Thuy Loi 7,815 23 Hung Yen Yen My Dong Than 9,588 24 Hung Yen Yen My Mihn Chau 4,460 25 Thanh Hoa Quang Xuong Quang Chau 8,092 26 Thanh Hoa Quang Xuong Quang Chinh 7,960 27 Thanh Hoa Quang Xuong Quang Duc 6,499 28 Thanh Hoa Quang Xuong Quang Ninh 6,518 29 Thanh Hoa Quang Xuong Quang Phong 7,175 30 Thanh Hoa Quang Xuong Quang Phu 7,070 31 Thanh Hoa Quang Xuong Quang Tan 9,430 32 Thanh Hoa Quang Xuong Quang Yen 6,808 33 Thanh Hoa Thach Thanh Thach Binh 7,205 34 Thanh Hoa Thach Thanh Thach Dong 5,267 35 Thanh Hoa Thach Thanh Thanh Truc 5,915 36 Thanh Hoa Tinh Gia Anh Son 5,133 37 Thanh Hoa Tinh Gia Ngoc Linh 6,002 38 Thanh Hoa Tinh Gia Nguyen Binh 9,725 (Continued ) www.wsp.org 41 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 1: Communes Selected for Handwashing Project IE Sample TABLE 20C: (Continued) No. Province District Commune Population 39 Thanh Hoa Tinh Gia Tan Truong 7,600 40 Thanh Hoa Trieu Son Dan Ly 8,709 41 Thanh Hoa Trieu Son Tho Binh 8,125 42 Thanh Hoa Trieu Son Tho Cuong 5,049 43 Thanh Hoa Trieu Son Tho Phu 4,581 44 Thanh Hoa Trieu Son Trieu Thanh 5,844 45 Tien Giang Cai Lay My Thanh Bac 8,212 46 Tien Giang Cai Lay Ngu Hiep 15,898 47 Tien Giang Cai Lay Nhi My 6,848 48 Tien Giang Cai Lay Nhi Quy 11,538 49 Tien Giang Cai Lay Phu Cuong 1,328 50 Tien Giang Cai Lay Phu Quy 671 51 Tien Giang Cai Lay Tan Phu 5,995 52 Tien Giang Cai Lay Thanh Loc 12,274 53 Tien Giang Chau Thanh Binh Duc 13,968 54 Tien Giang Chau Thanh Dong Hoa 9,158 55 Tien Giang Chau Thanh Nhi Binh 17,362 56 Tien Giang Chau Thanh Tan Hiep 14,587 57 Tien Giang Chau Thanh Thanh Phu 7,924 58 Tien Giang Chau Thanh Tna Huong 15,791 59 Tien Giang Cho Gao Binh Ninh 11,430 60 Tien Giang Cho Gao Long Binh Dien 12,320 61 Tien Giang Cho Gao Song Binh 9,413 62 Tien Giang Cho Gao Trugn Hoa 6,432 63 Tien Giang Cho Gao Xuan Dong 9,311 64 Tien Giang Go Cong Tay Binh Nhi 12,654 65 Tien Giang Go Cong Tay Binh Tan 11,025 66 Tien Giang Go Cong Tay Vinh Binh Town 14,068 67 Tien Giang Tan Phuoc My Phuoc Town 2,762 68 Tien Giang Tan Phuoc Tan Hoa Tay 409 69 Tien Giang Tan Phuoc Thanh My 2,040 70 Tien Giang Tan Phuoc Thanh Tan 1,351 528,609 42 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance Annex 2: Baseline Comparison of Means Tests for Balance As mentioned in Section II: Methodology, a critical require- in 9.5 percent of the tests on key characteristics (12 out of ment of the IE methodology is that a robust counterfactual 127 tests) for Treatment 1 vs. Control. A key difference to for the treatment group can be approximated. The house- note is that in the Treatment 1 group, households were sig- holds surveyed possess many characteristics that are either nificantly less likely to report washing their hands with soap unobservable, or for which data were not collected, and during the last 24 hours when compared with the Control thus balance between the groups on these unobservable group (p = 0.087). For the comparison between Treat- characteristics cannot be tested. However, if a sufficiently ment 2 vs. Control the null hypothesis of equality of means large number of observed characteristics are found to be was rejected at the 10% level in 7.1 percent of the tests on balanced across the treatment and control groups, then we key characteristics (9 out of 127 tests). One of the notable can be reasonably confident that the unobserved character- differences is the significantly lower length/height-for-age istics are balanced as well. Z-score found in the Treatment 2 group when compared with the Control group. There are significant differences Shown below are a series of tables presenting the mean between Treatment 2 and Control groups on both two-day comparison tests24 across treatment and control groups for and seven-day diarrhea prevalence, however since the preva- key variables included in the baseline survey. The null hy- lence figures are so low across all households sampled these pothesis of equality of means was rejected at the 10% level findings do no generate concern for the study balance. 24 The standard errors used in the comparison of means tests were clustered at the district level, allowing the possibility of intra-district correlation. www.wsp.org 43 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21A: COMPARISON OF MEANS TESTS FOR HOUSEHOLD DEMOGRAPHICS Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE HH size 1050 4.630 0.060 0.795 1050 4.630 0.059 0.793 1050 4.609 0.059 Number children under five years age (per HH) 1050 1.173 0.015 0.229 1050 1.199 0.015 0.966 1050 1.200 0.017 Age of HH head 1050 41.457 0.723 0.316 1050 42.688 0.713 0.849 1050 42.492 0.737 Age of other HH members 3806 19.313 0.358 0.499 3808 19.347 0.321 0.426 3788 18.997 0.300 HH head is male 1050 0.880 0.016 0.331 1050 0.862 0.014 0.886 1050 0.859 0.014 Other HH members are male 3831 0.370 0.008 0.965 3839 0.376 0.007 0.637 3818 0.371 0.008 HH head ever attended school 1041 0.981 0.005 0.466 1034 0.980 0.006 0.412 1032 0.985 0.004 Other HH members ever attended school 2503 0.990 0.003 0.140 2461 0.986 0.003 0.657 2481 0.984 0.003 Educational Attainment of HH Head: Incomplete primary 1012 0.144 0.016 0.759 999 0.158 0.017 0.756 993 0.151 0.015 Complete primary 1012 0.430 0.020 0.316 999 0.410 0.020 0.781 993 0.403 0.018 Incomplete secondary 1012 0.243 0.017 0.170 999 0.235 0.017 0.089 993 0.279 0.020 Complete secondary 1012 0.111 0.012 0.301 999 0.147 0.017 0.381 993 0.129 0.012 Higher 1012 0.072 0.012 0.015 999 0.049 0.010 0.353 993 0.038 0.007 Educational Attainment of Other HH Members: Incomplete primary 2437 0.211 0.010 0.193 2376 0.201 0.009 0.538 2380 0.192 0.010 Complete primary 2437 0.358 0.012 0.954 2376 0.351 0.014 0.780 2380 0.357 0.015 Incomplete secondary 2437 0.239 0.011 0.261 2376 0.249 0.013 0.636 2380 0.258 0.013 Complete secondary 2437 0.132 0.009 0.644 2376 0.142 0.011 0.834 2380 0.139 0.011 Higher 2437 0.060 0.008 0.596 2376 0.057 0.008 0.800 2380 0.054 0.007 44 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21B: COMPARISON OF MEANS TESTS FOR HOUSEHOLD PRIMARY WORK, LABOR INCOME, AND NON-LABOR INCOME Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE HH head is employed 1045 0.869 0.013 0.577 1039 0.846 0.015 0.539 1046 0.859 0.013 Others in HH are employed 1967 0.747 0.016 0.813 1998 0.762 0.016 0.709 1981 0.753 0.019 Females in HH are employed 1593 0.702 0.021 0.963 1606 0.699 0.022 0.939 1602 0.701 0.022 Last Week Activity--Unemployed HH Head: Studying 136 0.000 0.000 0.322 160 0.019 0.011 0.355 147 0.007 0.007 Taking care of home 136 0.316 0.047 0.274 160 0.431 0.061 0.567 147 0.388 0.045 Rent earner 136 0.044 0.025 0.809 160 0.019 0.013 0.336 147 0.054 0.035 Permanently unable to work 136 0.125 0.045 0.041 160 0.100 0.040 0.089 147 0.027 0.016 Retired 136 0.199 0.038 0.567 160 0.213 0.041 0.752 147 0.231 0.043 Not working 136 0.316 0.054 0.759 160 0.219 0.056 0.347 147 0.293 0.055 Last Week Activity--Unemployed Other HH Members: Looking for work 495 0.014 0.006 0.638 469 0.015 0.007 0.722 487 0.018 0.007 Studying 495 0.164 0.023 0.526 469 0.173 0.025 0.728 487 0.185 0.024 Taking care of home 495 0.630 0.039 0.418 469 0.635 0.039 0.367 487 0.585 0.039 Rent earner 495 0.010 0.005 0.230 469 0.013 0.009 0.333 487 0.031 0.016 Permanently unable to work 495 0.016 0.005 0.812 469 0.030 0.010 0.377 487 0.018 0.008 Retired 495 0.034 0.012 0.441 469 0.032 0.011 0.348 487 0.047 0.012 Not working 495 0.131 0.028 0.668 469 0.102 0.028 0.739 487 0.115 0.025 Primary Employment Status (% All Employed Individuals): Self-employed 2474 0.067 0.012 0.777 2502 0.068 0.009 0.791 2505 0.072 0.013 Employee 2474 0.248 0.016 0.820 2502 0.240 0.017 0.932 2505 0.242 0.018 Employer or boss 2474 0.004 0.002 0.987 2502 0.004 0.002 0.899 2505 0.004 0.003 Worker with no remuneration 2474 0.000 0.000 0.316 2502 0.000 0.000 0.316 2505 0.000 0.000 Day laborer 2474 0.052 0.009 0.890 2502 0.049 0.010 0.934 2505 0.050 0.010 Working in household activities or production 2474 0.624 0.022 0.985 2502 0.632 0.025 0.830 2505 0.624 0.027 Other 2474 0.005 0.002 0.687 2502 0.006 0.002 0.889 2505 0.006 0.002 Monthly salary (in VND millions) 928 2.229 0.188 0.650 903 1.954 0.094 0.585 930 2.092 0.234 Months worked per year 2485 9.365 0.169 0.731 2506 8.967 0.184 0.218 2498 9.281 0.177 Days worked per month 2485 21.095 0.320 0.398 2505 19.357 0.436 0.026 2497 20.666 0.396 Hours worked per day 2485 7.582 0.101 0.998 2499 7.445 0.111 0.357 2490 7.583 0.100 HH has non-labor income 1050 0.790 0.026 0.170 1050 0.747 0.030 0.736 1050 0.731 0.034 Total HH non-labor income (in VND millions) 802 2.5 0.6 0.419 766 3.0 0.6 0.664 743 3.6 1.3 www.wsp.org 45 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21C: COMPARISON OF MEANS TESTS FOR HOUSEHOLD ASSETS Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE Radio, CD, cassette 1049 0.086 0.013 0.262 1050 0.096 0.014 0.559 1049 0.108 0.014 Television 1050 0.914 0.012 0.859 1050 0.935 0.009 0.100 1050 0.911 0.011 Videocassette, VCR, DVD player 1050 0.594 0.027 0.715 1050 0.630 0.025 0.537 1050 0.608 0.025 Computer 1050 0.071 0.012 0.662 1050 0.054 0.008 0.407 1050 0.065 0.009 Bicycle 1050 0.733 0.032 0.409 1050 0.767 0.025 0.960 1050 0.769 0.029 Motorcycle 1050 0.804 0.017 0.114 1050 0.787 0.017 0.399 1050 0.767 0.017 Automobile or truck 1050 0.023 0.005 0.506 1050 0.030 0.006 0.716 1049 0.028 0.005 Refrigerator 1050 0.278 0.023 0.282 1050 0.253 0.020 0.829 1050 0.248 0.017 Gas stove 1050 0.451 0.028 0.508 1050 0.459 0.028 0.379 1050 0.427 0.024 Blender 1050 0.310 0.025 0.575 1050 0.280 0.024 0.761 1050 0.290 0.025 Microwave 1050 0.017 0.006 0.082 1050 0.015 0.005 0.117 1050 0.007 0.002 Washing machine 1050 0.089 0.016 0.283 1050 0.075 0.012 0.666 1050 0.069 0.010 Water boiler, hot water heater 1050 0.254 0.033 0.003 1050 0.213 0.032 0.047 1050 0.136 0.022 Machinery, equipment for household business 1050 0.023 0.007 0.761 1050 0.022 0.008 0.701 1050 0.026 0.006 Boat 1050 0.042 0.014 0.122 1050 0.017 0.006 0.809 1050 0.019 0.006 Telephone (including mobile) 1050 0.758 0.026 0.737 1050 0.782 0.021 0.282 1050 0.746 0.026 Air conditioner 1050 0.006 0.003 0.348 1050 0.015 0.005 0.297 1050 0.010 0.003 Electric fan 1050 0.956 0.008 0.536 1050 0.957 0.010 0.533 1050 0.949 0.009 HH owns other piece of land 1050 0.190 0.035 0.622 1050 0.185 0.033 0.705 1050 0.168 0.030 HH owns farm equipment 1050 0.190 0.027 0.409 1050 0.187 0.026 0.366 1050 0.223 0.030 HH has animals 1050 0.609 0.037 0.312 1050 0.635 0.035 0.595 1050 0.663 0.039 Number of different kinds of livestock owned per HH 1050 1.036 0.078 0.401 1050 1.050 0.074 0.465 1050 1.129 0.077 46 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21D: COMPARISON OF MEANS TESTS FOR HANDWASHING BEHAVIOR Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE Washed hands with soap during the last 24 hours 1051 0.927 0.019 0.087 1051 0.935 0.020 0.200 1054 0.963 0.009 Washed Hands with Soap During the Last 24 Hours in the Following Instances: After using the toilet 1051 0.500 0.036 0.345 1051 0.461 0.039 0.856 1054 0.452 0.038 After cleaning child's bottom 1051 0.264 0.037 0.066 1051 0.328 0.041 0.519 1054 0.366 0.042 Before preparing food or cooking 1051 0.333 0.039 0.862 1051 0.275 0.036 0.358 1054 0.324 0.038 Before feeding children 1051 0.349 0.031 0.892 1051 0.294 0.035 0.230 1054 0.356 0.038 HWWS all critical times 1051 0.067 0.024 0.220 1051 0.048 0.018 0.067 1054 0.118 0.034 Because they look or feel dirty 1051 0.461 0.043 0.666 1051 0.477 0.047 0.874 1054 0.487 0.043 After or while doing laundry 1051 0.441 0.049 0.345 1051 0.410 0.048 0.155 1054 0.506 0.047 During at least one critical time 1051 0.788 0.026 0.964 1051 0.777 0.029 0.736 1054 0.789 0.021 www.wsp.org 47 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21E: COMPARISON OF MEANS TESTS FOR HANDWASHING FACILITIES Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE Place to HW in HH with S&W present 1050 0.794 0.028 0.556 1050 0.811 0.029 0.884 1050 0.817 0.027 Place to HW before food preparation area with S&W present 1050 0.290 0.033 0.983 1050 0.296 0.028 0.868 1050 0.290 0.029 Place to HW after using toilet with S&W present 1050 0.775 0.029 0.531 1050 0.794 0.030 0.887 1050 0.800 0.027 HH handwashing after using toilet 1046 0.982 0.006 0.719 1047 0.984 0.005 0.569 1047 0.978 0.009 Location of Handwashing Device, Toilet: Inside toilet facility 1027 0.242 0.030 0.638 1028 0.236 0.027 0.734 1024 0.224 0.026 Inside food preparation area 1027 0.057 0.018 0.133 1028 0.046 0.016 0.331 1024 0.028 0.008 Wash basin in yard, less than 1 meter from toilet facility 1027 0.171 0.037 0.962 1028 0.153 0.034 0.741 1024 0.169 0.035 Pond or stream in yard, less than 1 meter from toilet facility 1027 0.004 0.002 0.702 1028 0.003 0.002 0.996 1024 0.003 0.002 Wash basin in yard, between 1 and 3 meters from toilet facility 1027 0.094 0.023 0.469 1028 0.103 0.019 0.618 1024 0.117 0.021 Pond or stream in yard, between 1 and 3 meters from toilet facility 1027 0.009 0.004 0.743 1028 0.006 0.004 0.868 1024 0.007 0.004 Type of Handwashing Device, Toilet: Tap, faucet 926 0.253 0.038 0.332 933 0.257 0.040 0.381 924 0.310 0.044 Homemade water tap 926 0.414 0.044 0.806 933 0.449 0.049 0.776 924 0.430 0.048 Basin, bucket 926 0.215 0.033 0.191 933 0.195 0.030 0.377 924 0.160 0.026 Other 926 0.015 0.006 0.144 933 0.005 0.004 0.992 924 0.005 0.004 Water is available at the place for washing hands, toilet 911 0.981 0.006 0.789 927 0.980 0.009 0.966 907 0.979 0.006 Soaps Available at the Place for Washing Hands, Toilet: Bar soap 929 0.479 0.031 0.164 934 0.511 0.031 0.554 925 0.534 0.024 Liquid, dishwashing liquid soap 929 0.157 0.025 0.145 934 0.208 0.036 0.816 925 0.219 0.035 Powder soap, laundry detergent 929 0.635 0.041 0.106 934 0.694 0.038 0.583 925 0.722 0.035 No soap observed 929 0.087 0.021 0.117 934 0.077 0.023 0.279 925 0.048 0.014 (Continued ) 48 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21E: (Continued) Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE Ash, Mud Available at the Place for Washing Hands, Toilet: Ash 885 0.003 0.002 0.055 899 0.012 0.005 0.743 887 0.015 0.006 Mud 885 0.051 0.017 0.128 899 0.036 0.012 0.336 887 0.021 0.008 Ash and Mud 885 0.025 0.012 0.815 899 0.032 0.011 0.440 887 0.021 0.008 Neither ash nor mud observed 885 0.921 0.024 0.452 899 0.920 0.020 0.370 887 0.943 0.016 HH handwashing before/ after cooking or feeding a child 1045 0.965 0.010 0.710 1044 0.966 0.009 0.802 1047 0.969 0.008 Location of Handwashing Device, Food Preparation: Inside toilet facility 1008 0.033 0.008 0.678 1007 0.030 0.009 0.524 1015 0.037 0.008 Inside food preparation area 1008 0.184 0.026 0.092 1007 0.161 0.028 0.331 1015 0.127 0.021 Type of Handwashing Device, Food Preparation: Tap, faucet 345 0.446 0.058 0.675 336 0.452 0.059 0.730 330 0.482 0.062 Container from which water is poured 345 0.130 0.032 0.008 336 0.057 0.019 0.439 330 0.039 0.012 Other 345 0.423 0.061 0.536 336 0.491 0.060 0.890 330 0.479 0.065 Water is available at the place for washing hands, food preparation 346 0.983 0.010 0.948 336 0.976 0.013 0.691 331 0.982 0.007 Soaps Available at the Place for Washing Hands, Food Preparation: Bar soap 346 0.480 0.055 0.495 336 0.607 0.045 0.274 331 0.532 0.053 Liquid, dishwashing liquid soap 346 0.246 0.055 0.109 336 0.307 0.065 0.407 331 0.384 0.067 Powder soap, laundry detergent 346 0.572 0.053 0.422 336 0.646 0.051 0.917 331 0.637 0.062 No soap observed 346 0.075 0.021 0.495 336 0.045 0.017 0.726 331 0.054 0.022 Ash, Mud Available at the Place for Washing Hands, Toilet: Ash 331 0.003 0.003 0.541 320 0.003 0.003 0.557 316 0.006 0.004 Mud 331 0.076 0.037 0.066 320 0.041 0.021 0.117 316 0.006 0.004 Ash and Mud 331 0.027 0.008 0.023 320 0.028 0.016 0.187 316 0.006 0.004 Neither ash nor mud observed 331 0.894 0.041 0.040 320 0.928 0.025 0.047 316 0.981 0.009 www.wsp.org 49 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 2: Baseline Comparison of Means Tests for Balance TABLE 21F: COMPARISON OF MEANS TESTS FOR ACUTE LOWER RESPIRATORY INFECTION AND DIARRHEA SYMPTOMS PREVALENCE (% CHILDREN < 5) Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE ALRI in previous 48 hours 1061 0.006 0.003 0.797 1076 0.003 0.002 0.262 1057 0.007 0.003 ALRI in previous three days 1061 0.006 0.003 0.314 1076 0.006 0.003 0.321 1057 0.010 0.004 Diarrhea in previous 48 hrs 1061 0.008 0.003 0.358 1076 0.002 0.001 0.007 1057 0.011 0.003 Diarrhea in previous week 1061 0.013 0.003 0.493 1076 0.005 0.002 0.011 1057 0.017 0.004 Anemic: Hb level < 11 g/dl 984 0.312 0.021 0.828 1011 0.331 0.021 0.391 979 0.305 0.022 TABLE 21G: COMPARISON OF MEANS TESTS FOR CHILD GROWTH MEASURES (Z-SCORES) Treatment 1 Treatment 2 Control N Avg. SE p-value N Avg. SE p-value N Avg. SE BMI-for-age Z-score 1035 ­0.222 0.055 0.345 1056 ­0.253 0.045 0.557 1030 ­0.295 0.055 Head circumference-for- age Z-score 1054 ­0.650 0.049 0.316 1066 ­0.614 0.053 0.635 1050 ­0.579 0.051 Length/height-for-age Z-score 1043 ­0.724 0.064 0.139 1061 ­0.747 0.063 0.083 1043 ­0.584 0.070 Arm circumference-for- age Z-score 1050 0.124 0.048 0.491 1060 0.120 0.054 0.480 1051 0.172 0.050 Weight-for-length/height Z-score 1051 ­0.310 0.051 0.429 1067 ­0.335 0.042 0.641 1043 ­0.365 0.049 Weight-for-age Z-score 1046 ­0.632 0.043 0.276 1064 ­0.632 0.043 0.275 1045 ­0.565 0.045 50 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 3: Comparison between WSP IE Baseline Survey and VNDHS Survey Annex 3: Comparison between WSP IE Baseline Survey and VNDHS Survey The experimental group for the handwashing project impact TABLE 22: DEMOGRAPHIC CHARACTERISTICS OF evaluation was designed with the primary intention of produc- HOUSEHOLD RESPONDENTS IN WSP SURVEY AND VNDHS ing internally valid estimates of program impacts under the WSP Survey VNDHS unique constraints of the handwashing project, and is not in- Age: tended to be suitable for computing country, province, or 0­4 25.8% 7.3% commune level population statistics without additional as- 5­9 6.8% 10.0% sumptions. The experimental group is not a representative 10­14 3.8% 12.5% sample of the Vietnamese population for several reasons. First, 15­19 3.0% 11.2% the experimental group includes only three out of a total of 61 20­24 9.3% 8.0% provinces, and 15 out of a total of approximately 600 districts 25­29 15.9% 7.5% in Vietnam. These 15 districts were conveniently and purpo- sively selected due to their suitability for the intervention and 30­34 10.9% 7.6% willingness to participate in the study. Furthermore, within the 35­39 5.8% 7.6% districts chosen, only those communes with an active Vietnam 40­44 2.9% 7.3% Women's Union were eligible to participate in the study. Fi- 45­49 2.7% 5.5% nally, the experimental group comprises only those households 50+ 13.1% 15.6% with a child under the age of two at the time of the survey. Average age 24.2 28.8 These factors imply that causal inferences of the treatment on outcomes are limited to the experimental group. Age Distribution of Children Under Five (% per HH): Under 12 months 26.5% 17.6% Here we present a comparison of basic characteristics of the 12­23 mo 44.8% 21.6% Vietnamese population using the 2002 Vietnam Demographic 24­35 mo 18.2% 20.5% Health Survey (VNDHS)25 with characteristics of the individ- uals included in the WSP IE survey subsample. We concen- 36­47 mo 5.0% 19.1% trate on three groups of variables: demographics, educational 48­59 mo 5.4% 21.3% attainment, and household wealth measured by an asset index. Average age of under five 2.18 2.05 Table 22 presents the basic demographics for the two sample populations. The large proportion of children between 0 and Total Number of Children Under Five (% HHs): 4 years and household members from 25 to 35 are evidence 0 0.0% 62.0% of the WSP study sample selection restriction to mothers/ 1 81.6% 28.6% caretakers of children under five years old. On average, the 2 17.8% 7.9% individuals interviewed in the WSP survey are 24.2 years old, 3 0.6% 1.2% whereas the average age of the VNDHS sample is 28.8 years. 4 0.0% 0.3% While the average number of children under the age of five 5 0.0% 0.0% per household is 0.49 in the VNDHS, this figure is 1.19 in Average number of the WSP survey, again a likely factor of the unique sample children under five in HH 1.19 0.49 selection of the WSP survey. 25 The population considered in the VNDHS was selected following the restriction of age imposed by the WSP IE survey for each group of questions. www.wsp.org 51 Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 3: Comparison between WSP IE Baseline Survey and VNDHS Survey Table 23 summarizes the educational attainment of the an asset index was created using ownership of durable goods household population by age group. The Vietnamese formal common to both samples. The procedure uses principal com- educational system is divided into three tiers, known as the ponents analysis (PCA)26 to assign weights to each asset indi- five-four-three system. Individuals complete five years of pri- cator variable, which are then applied to the separate samples mary, four years of lower secondary, and three years of higher to estimate the wealth of each household.27 The durable secondary education. Graduates of higher secondary school goods included in the index are radio, television, refrigerator, may then pursue higher education through university, col- bicycle, motorcycle/motor scooter, car, telephone, washing leges, or technical schools. Overall there are no major differ- machine, boat, and plowing machine. ences in educational attainment between the two samples, although the WSP survey contains a lower proportion of un- The distribution of wealth scores for the WSP survey and the educated individuals (1.5%), compared with the VNDHS VNDHS samples are shown in Figure 4. The wealth score average of 8.3%, shown in the lower half of the table. places the household along a continuum of wealth from poorest to wealthiest. As illustrated in the leftmost graph, A final comparison between the WSP IE survey and the wealth scores in the WSP survey are approximately normally VNDHS is made on the socioeconomic makeup of the distributed with a mean of 0.84, while in the VNDHS sam- samples using an asset-based index of household wealth. An ple the distribution is skewed to the left with a mean of ­0.38, asset-based wealth index was chosen as the key socioeco- indicating households in the sample are poorer on average. nomic indicator over an income or expenditure based mea- There are several potential explanations for this pattern. First, sure since household asset ownership is a more stable the VNDHS survey was administered in 2002 and since this measure of household wealth and is less susceptible to time Vietnam has experienced rapid economic growth and short-term shocks. poverty reduction. Second, while the VNDHS is a nationally representative sample, the WSP survey is not a representative In order to make a valid comparison between socioeconomic sample of the Vietnamese population, but rather the target status across different surveys the data sets were pooled, and population of the handwashing project. FIGURE 4: DISTRIBUTION OF WEALTH SCORES FOR THE WSP SURVEY AND VNDHS WSP IE VNDHS 1400 1400 Frequency: Number of Households Frequency: Number of Households 1200 1200 1000 1000 800 800 600 600 400 400 200 200 0 0 ­3 ­2 ­1 0 1 2 3 4 5 ­3 ­2 ­1 0 1 2 3 4 5 Wealth score Wealth score 26 Filmer and Pritchett 2001. 27 The WSP IE survey wealth index used elsewhere in this report is constructed using only the WSP IE survey sample. It contains household ownership of durable goods, land and agricultural equipment, and livestock. 52 Global Scaling Up Handwashing Findings from the Impact Evaluation Baseline Survey in Vietnam Annex 3: Comparison between WSP IE Baseline Survey and VNDHS Survey TABLE 23: EDUCATIONAL ATTAINMENT OF HOUSEHOLD POPULATION IN WSP SURVEY AND VNDHS No Incomplete Complete Incomplete Complete Age Group Education Primary Primary Secondary Secondary Higher Total WSP Survey: 5­9 2.8% 93.0% 3.9% 0.1% 0.2% 0.0% 100.0% 10­14 0.6% 7.4% 79.6% 12.5% 0.0% 0.0% 100.0% 15­19 0.5% 1.9% 14.4% 58.4% 20.8% 4.0% 100.0% 20­24 0.5% 4.3% 29.0% 36.0% 23.8% 6.4% 100.0% 25­29 0.4% 5.0% 30.9% 31.8% 22.5% 9.4% 100.0% 30­34 0.7% 8.3% 41.5% 27.2% 12.9% 9.4% 100.0% 35­39 1.2% 8.2% 40.8% 27.5% 16.8% 5.4% 100.0% 40­44 1.5% 13.6% 42.1% 26.2% 13.6% 3.1% 100.0% 45­49 1.6% 16.7% 54.9% 18.5% 6.5% 1.8% 100.0% 50+ 4.3% 31.7% 47.5% 11.7% 2.4% 2.5% 100.0% Total 1.5% 18.3% 36.8% 24.6% 13.3% 5.5% 100.0% VNDHS: 5­9 16.4% 83.5% 0.1% 0.0% 0.0% 0.0% 100.0% 10­14 1.6% 29.2% 21.8% 47.4% 0.0% 0.0% 100.0% 15­19 2.9% 8.3% 6.3% 71.2% 11.3% 0.0% 100.0% 20­24 5.2% 14.2% 10.8% 41.4% 23.4% 5.1% 100.0% 25­29 5.7% 13.9% 8.6% 51.3% 12.8% 7.8% 100.0% 30­34 5.5% 13.7% 8.2% 50.5% 17.7% 4.4% 100.0% 35­39 4.2% 14.8% 7.7% 52.9% 16.8% 3.6% 100.0% 40­44 5.1% 15.0% 9.2% 50.8% 14.7% 5.2% 100.0% 45­49 6.9% 17.7% 8.1% 48.4% 12.7% 6.2% 100.0% 50+ 20.8% 30.9% 11.9% 24.3% 8.0% 4.1% 100.0% Total 8.3% 26.1% 9.8% 42.1% 10.5% 3.2% 100.0% www.wsp.org 53