WSP Global Scaling Up Rural Sanitation Access (TSSM) Impact Evaluation, Madhya Pradesh State, Baseline and Endline Surveys 2009-2011 - India

Abstract
---------------------------

In India, WSP's Global Scaling Up Rural Sanitation Program is supporting the Government of India's Total Sanitation Campaign (TSC) in two states: Himachal Pradesh and Madhya Pradesh. TSC is an ambitious countrywide, scaled-up rural sanitation program launched in 1999, which seeks to attain an Open-Defecation Free (ODF) India by 2012. In contrast to earlier, hardware-centric supply approaches to rural sanitation, TSC aims to generate demand for and adoption of improved sanitation at the community level.

This impact evaluation aimed at better understanding what health and welfare impacts can be expected from rural sanitation improvements. Researchers hypothesized that promotion of rural sanitation through community-led total sanitation (CLTS) and social marketing campaigns will improve the health of the population especially children under five years old, a population that is vulnerable to unsafe disposal of feces in the environment and fecal-oral contamination.

This impact evaluation consisted of baseline and endline surveys. In collaboration with the government of Madhya Pradesh, two districts - Dhar and Khargone - were selected. In each district, 80 Gram Panchayats were chosen and randomized into two groups: 1) treatment group (to participate in Total Sanitation Campaign immediately following the baseline survey) and 2) control group (to receive TSC after follow-up data collection).

The baseline survey collected information from a representative sample of the population targeted by the intervention. This baseline survey was administered to approximately 2,000 households between June and July 2009. The survey provided information on the characteristics of household members, access to sanitation facilities, self-reported open defecation, prevalence of child diseases such as diarrhea and respiratory infection, and child growth and development.

The endline survey was carried out in February-March 2011. It followed the same households sampled in baseline, as well as additional children to increase statistical power.

Geographic coverage
---------------------------

The surveys covered Dhar and Khargone districts in Madhya Pradesh state.

Analysis unit
---------------------------

- Household
- Person
- Caregiver
- Child (under 5 and under 2)

Kind of data
---------------------------

Sample survey data [ssd]

Sampling procedure
---------------------------

The selection of the sample in Madhya Pradesh was completed in several stages. First, at the design stage of the project, MP was selected a priori as one of two states to participate in the IE. Second, two districts in MP - Dhar and Khargone were selected by WSP in collaboration with the state government. Third, within each of these districts, a total of 80 Gram Panchayats (GP) were selected as candidates for Total Sanitation Campaign (TSC) implementation. In the fourth stage, one village from each candidate GP was identified by the GP as a community that is suitable for implementing TSC yielding a list of 80 villages in each of the four districts. Within each district, 40 of the candidate GPs (and their appointed village) were randomly assigned to the treatment group, and the remaining 40 were assigned to the control group.

Approximately, 1,000 households were sampled in each district to achieve a total sample size of 2,000 households. The final selection of households to participate in the IE survey was carried out by the survey firm contracted to conduct the IE baseline data collection. A household listing of all participating villages was conducted and from this list, 25 households with children under two years old were randomly selected for participation. When 25 eligible households were not available in the listed village, a neighboring village was listed and sampled to achieve the desired number of households in the GP.

Mode of data collection
---------------------------

Face-to-face [f2f]

Research instrument
---------------------------

Household Questionnaire: The household questionnaire collected information about household membership and demographics, income, assets, dwelling characteristics, access to water and sanitation, sanitation- and hygiene-related behaviors, maternal depression, mortality, exposure to health interventions, and other outcomes. Enumerators also conducted standardized observations of dwellings and child cleanliness and of sanitation and handwashing facilities at the time of the HH interviews.

Health Questionnaire: The health questionnaire collected information about children’s diarrhea prevalence, acute lower respiratory infection (ALRI), other health symptoms, and child development and growth. As part of this questionnaire, hemoglobin concentrations were measured in children younger than two years of age at the household level using the HemoCueTM Hb201 photometer, a portable device that allows for immediate and reliable quantitative results. Anthropometric (child growth) measures were made according to standardized protocols using portable stadiometers, scales, and measuring tape (Habicht 1974).

Community Questionnaire: The community questionnaire was administered at the GP-level to collect information about GP and district-level characteristics that could influence the intervention or the outcomes of interest (e.g., ongoing health and sanitation programs, connectivity to district headquarters, and other factors).

Water Samples: Water samples were collected from sources at the GP-level and at the household level for a subset of the households (n = X GP-level source samples; n = 354 HH samples). All of the water samples were analyzed by an accredited lab in Indore to determine presence of E. coli and other types of coliforms. The samples were collected within the household, inoculated using the Colilert reactive, and transported to a lab. At the lab, samples were incubated at 35 degrees Celsius for 24 hours, and results were read using an ultraviolet lamp. This procedure precluded sampling in areas where a cold chain could not be maintained.

Stool Samples: Stool samples were collected from children to examine the prevalence of parasites. These were collected from a subset of sampled households (n=216). The same lab in Indore analyzed these samples.

Cleaning operations
---------------------------

Baseline: The baseline survey was processed using the assistance of Sistemas Integrales in Chile.

Endline: Kimetrica International was contracted to design the data reduction system to be used during the endline. The data entry system was designed in CSPro (Version 4.1) using the DHS file management system as a standard for file management. Details of the system can be found in the attached manual entitled: Data Entry Manual for the Endline Survey.

The data entry system was based on a full double data entry (independent verification) of the various questionnaires. CSPro supports both dependent and independent verification (double keying) to ensure the accuracy of the data entry operation. Using independent verification, operators can key data into separate data files and use CSPro utilities to compare them and produce a report that indicates discrepancies in data entry.

The DHS system uses a fully integrated tracking system to follow the stages in the data entry process. This includes the checking in of questionnaires; the programming of logic in what is known as a system controlled environment. System controlled applications generally place more restrictions on the data entry operator. This is typically used for complex survey applications. The behavior of these applications at data entry time has the following characteristics:

- Some special data entry keys are not active during data entry.
- CSEntry will keep track of the path.
- 'Not applicable' or blanks values will not be allowed. Missing values have to be coded.
- More appropriate to the heads up methodology of data capture.
- Logic in the application is strictly enforced; operator cannot bypass or override.

Files were processed using the unique cluster number and then concatenated after a final stage of editing and output to both SPSS and STATA.

Furthermore, attempts were made to respect the values and the naming conventions as provided in the baseline. This required using non-conventional values for "missing" such as -99. In most cases the same value sets were applied or during the questionnaire review process the WSP was alerted to such discrepancies.

Sampling error estimates
---------------------------

Not applicable

Data appraisal
---------------------------

Although there was no formal or independent appraisal of the data, an appraisal was undertaken when the data files for Peru, India and Vietnam were prepared for a WSP presentation in Mexico. These data were presented in a public forum and scrutinized by various analysts. There was a process of feeding back information which helped correct or format or revise the data.

摘要
---------------------------
在我国，世界卫生组织（WSP）的全球农村卫生扩展项目支持印度政府在两个州——喜马偕尔邦和中央邦——实施全面卫生运动（TSC）。TSC是一项于1999年启动的雄心勃勃的国家级农村卫生扩展项目，旨在到2012年实现无露天排便的印度（ODF）。与之前以硬件为中心的农村卫生供应方法相比，TSC旨在在社区层面激发对改善卫生条件的需求并促进其采用。
本次影响评估旨在深入了解农村卫生改善可能带来的健康和福利影响。研究人员假设，通过社区主导的全面卫生（CLTS）和社会营销活动来推广农村卫生，将改善人口的健康状况，特别是五岁以下儿童的健康状况，这一群体易受环境粪便不当处理和粪口传播的影响。
本次影响评估包括基线和终线调查。在中央邦政府的合作下，选择了两个区——达拉和卡尔戈内。在每个区，选择了80个村议会，并将其随机分为两组：1）处理组（在基线调查后立即参与全面卫生运动）和2）对照组（在后续数据收集后接受TSC）。
基线调查在2009年6月至7月期间对大约2,000户家庭进行了信息收集。该调查提供了关于家庭成员特征、卫生设施的可获得性、自我报告的露天排便、儿童疾病（如腹泻和呼吸道感染）的流行情况以及儿童的生长和发育的信息。
终线调查于2011年2月至3月进行。它遵循了基线调查中抽样的相同家庭，以及额外的儿童以提高统计效力。
地理覆盖范围
---------------------------
调查涵盖了中央邦的达拉和卡尔戈内区。
分析单位
---------------------------
- 家庭
- 个人
- 护理者
- 儿童（5岁以下和2岁以下）
数据类型
---------------------------
样本调查数据 [ssd]
抽样程序
---------------------------
在马德拉斯邦的样本选择分为几个阶段。首先，在项目设计阶段，MP被事先选为两个参与IE的州之一。其次，WSP与州政府合作在MP中选择了两个区——达拉和卡尔戈内。第三，在每个这些区中，总共选择了80个村议会（GP）作为实施全面卫生运动（TSC）的候选者。在第四阶段，每个候选GP中的一个村庄由GP确定为适合实施TSC的社区，从而在每个四个区中产生了一个80个村庄的名单。在每个区中，40个候选GP（及其指定的村庄）被随机分配到处理组，其余的40个被分配到对照组。大约在每个区抽取了1,000户家庭，以达到2,000户家庭的总体样本量。最终选择参与IE调查的家庭由负责进行IE基线数据收集的调研公司执行。对所有参与村庄的家庭进行了清单编制，并从该清单中随机选择了25户有2岁以下儿童的家庭。当列出的村庄中没有25户符合条件的家庭时，列出了相邻的村庄并进行抽样，以达到GP中所需的户数。
数据收集方式
---------------------------
面对面 [f2f]
研究工具
---------------------------
家庭问卷：家庭问卷收集了关于家庭成员和人口统计信息、收入、资产、住宅特征、水和卫生设施的可获得性、卫生和卫生相关行为、母性抑郁、死亡率、暴露于健康干预措施以及其他结果的信息。普查员还进行了住宅、儿童清洁以及卫生和洗手设施在HH访谈时的标准化观察。
健康问卷：健康问卷收集了关于儿童腹泻流行率、急性下呼吸道感染（ALRI）、其他健康症状以及儿童发展和生长的信息。作为该问卷的一部分，使用HemoCueTM Hb201光电光度计在家庭层面测量了2岁以下儿童的血红蛋白浓度，这是一种便携式设备，可提供即时可靠的定量结果。根据标准化协议使用便携式身高计、秤和卷尺（Habicht 1974）进行人体测量（儿童生长）测量。
社区问卷：社区问卷在村议会级别进行，以收集可能影响干预措施或感兴趣的结果的村议会和区级特征信息（例如，持续进行的健康和卫生计划、与区总部连接以及其他因素）。
水样本：从村议会级别和家庭级别的来源收集了水样本，用于对部分家庭（n = X GP级别来源样本；n = 354 HH样本）进行测试。所有水样本均由印多尔的一家认可实验室分析，以确定大肠杆菌和其他肠杆菌的存在。样本在家庭内部收集，使用Colilert反应性接种，并运送到实验室。在实验室，样本在35摄氏度下培养24小时，并使用紫外线灯读取结果。此程序排除了无法维持冷链的地区抽样。
粪便样本：从儿童中收集粪便样本以检查寄生虫的流行情况。这些样本来自部分抽样家庭（n=216）。这些样本由同一家位于印多尔的实验室分析。
清洁操作
---------------------------
基线：基线调查在智利的Sistemas Integrales的帮助下进行处理。
终线：Kimetrica International被委托设计用于终线的数据缩减系统。数据输入系统在CSPro（版本4.1）中使用DHS文件管理系统作为文件管理标准进行设计。有关该系统的详细信息可在附带的名为《终线调查数据输入手册》的手册中找到。
数据输入系统基于各种问卷的全双倍数据输入（独立验证）。CSPro支持依赖性和独立性验证（双键输入），以确保数据输入操作的准确性。使用独立性验证，操作员可以将数据输入到单独的数据文件中，并使用CSPro实用程序进行比较，生成报告，表明数据输入中的差异。
DHS系统使用完全集成的跟踪系统来跟踪数据输入过程的各个阶段。这包括问卷的登记；在所谓的系统控制环境中编程逻辑。系统控制应用程序通常对数据输入操作员施加更多限制。这通常用于复杂的调查应用程序。这些应用程序在数据输入时的行为具有以下特征：
- 在数据输入期间，某些特殊数据输入键是不活跃的。
- CSEntry将跟踪路径。
- 不适用或空白值将不被允许。缺失值必须编码。
- 更适合数据捕获的头抬方法。
- 应用程序中的逻辑被严格执行；操作员不能绕过或覆盖。
使用唯一的集群编号处理文件，然后在编辑和输出到最后阶段后连接。
此外，还尝试尊重基线中提供的值和命名约定。这需要使用非传统值“缺失”如-99。在大多数情况下，应用了相同的值集，或在问卷审查过程中，WSP被通知了这些差异。
抽样误差估计
---------------------------
不适用
数据评估
---------------------------
尽管没有正式或独立的评估数据，但在为秘鲁、印度和越南的数据文件准备用于WSP在墨西哥的展示时进行了评估。这些数据在公开论坛上展出，并由各种分析师进行了审查。有一个反馈信息的过程，有助于纠正或格式化或修改数据。