Dhaka Education And Health Foundation

Survey Methodology

1. Study design

The sample for the National Hygiene Survey 2018 was designed to provide estimates for the large number of indicators of Hygiene and Sanitation situation of the household, school, food vendors & restaurants and the health facility providers. This was a cross sectional survey. The survey comprised of four broad components (1) Household-level hygiene component including Menstrual Hygiene Management (MHM) among girls and women in the age group 10 – 49 years, (2) School hygiene including MHM, (3) Food hygiene in restaurants and among street food vendors, and (4) Health facility hygiene (see Table A).

2. Sample size and sampling units

The household survey used two-stage stratified cluster sampling. First 176 Enumeration Areas (EAs) were selected from the total of 293,570 EAs in Bangladesh using Probability Proportional to Size (PPS) Sampling. These EAs formed the Primary Sampling Units (PSUs or clusters), 176 PSUs were covered in the survey throughout the country. The Bangladesh Population and Housing Census 2011 was used as the sampling frame with modifications as some rural areas had been declared urban since the 2011 census (BBS 2012). The mean PSU size was 120 households.

In case of school, restaurant and food vendors and health facility component, the field team also listed the available primary and secondary level schools, restaurants, street food vendors, and all available health facilities (government, non-government and private) around the household clusters. Later on, from the list of primary and secondary school, four schools were selected with the ratio of primary and secondary education is 2:3. This means for every 10 schools, 4 schools were primary and other six schools were secondary level. In the restaurants or street food vendors’ component, two restaurants and four street food vendors were sampled and in the health facility components, five health facilities were selected from the list at each cluster. Now to have an estimate for WASH situation in this study population precision of 3.6 to 4.5 with design effect of 4.5 has been considered. The sampling technique of schools, restaurants, street food vendors and health facilities was aimed to maximize study efficiency; the alternative of using separate sampling frames for each population would make the study very costly. This strategy also allows linkages between households and the schools, restaurants/vendors, and health facility providers them service.

3. Sampling units, selection criteria and data collection methods

3.1 Household Component

As for primary sampling unit 176 Enumeration Areas were selected; thereafter listed all the Enumeration Areas and then 30 households were selected by using systematic sampling technique at each Enumeration Areas. If any of the eligible respondents of the household was not available or refused to participate, then the next eligible household from the list was surveyed. However, the replacement was taken after having multiple visits (up to three) at households for those respondents was not instantly available. Sampling unit for MHM - If the household had more than one adolescent female then randomly selected one girl for the interview. If there was no adolescent girl in the household, the team administered the menstrual hygiene module to the female caregiver if she was in the reproductive age range. If the female caregiver was not in the reproductive age range, the team interviewed any women of that household in the reproductive age range. Informed consent and participant eligibility criteria: Households were considered eligible if the following criteria were met. • Household head or available primary caregiver of the children in the household provided informed consent for this survey and spot checks. • For MHM survey, if the household had an adolescent girl then the female data collectors took informed consent from her and her guardian and performed the interview. An adolescent girl (10-19 years) if available and a female of age group 20 to 49 years. Method of data collection and data collection tools – Household level data collection was done by face-to-face interviews with the eligible respondents, conducted spot checks for sanitation facilities and hand hygiene practices, and by conducting handwashing demonstrations of households caregivers, and children under-five years of age. The primary target respondent for hand cleanliness spot-checks and handwashing demonstrations were the youngest child <5 years of age, and the primary caregiver of the children (male or female) since s/he has the closest contact with the children.

For the menstrual hygiene data, our female data collectors conducted interviewed face to face to adolescent females (10 to 19 years old) and similarly for women in age group (20 to 49 years old).

3.2 School Component

Around the household cluster communities from which the study households were sampled, the field team listed eight nearby primary and secondary level schools. From the list, four nearby schools were selected for the survey and hence the sample size stands at 704 schools in the 176 randomly selected clusters. If any of the school refused to participate, the next nearby school from the list was included for survey. In consultation and permission from the headmasters in the sampled schools, the team selected 4 students for face to face interviews at each school with equal proportion of boys and girls. For primary schools, girls were selected for MHM from Class V only and for secondary schools, the girls were selected from Class VI-X. Four girls who menstruated before the survey were selected for the interview by the head masters at each school. School survey eligibility criteria and consent taking: Schools were considered eligible for the survey if the following criteria were met. • Primary or high school. • Headmaster or designated school teacher provided informed consent for the survey. • The field team excluded Madrasahs (Islamic teaching institutions) and English medium schools (following the British curriculum) as they used different curricula and are controlled by a different school board. If any of the authorities from the selected school refused to participate in the study, the team replaced it with the next school from the list. Method of data collection and data collection tools - In the eligible schools, interview was conducted with headmasters/designated teachers and conducted spot checks to water, sanitation, handwashing facilities and MHM facilities at schools. Also, Four sampled students were interviewed face to face using structured questionnaires. The team observed handwashing skills of students by conducting handwashing demonstrations. Finally, the female team members conducted MHM face to face interviews with sampled girls in secondary schools.

3.3 Restaurant Component

Across all clusters where the household survey was conducted, the field team listed all restaurants around the household survey clusters by conducting transact walks and talking to the key informants in the communities. From the list of restaurants two restaurants were randomly selected at each cluster or Enumeration Areas. If the sampled restaurant was refused to participate, the field team replaced the restaurant by the next available restaurant from the list Eligibility criteria: Restaurants were considered eligible for the survey if the following criteria were met: • Cooked food at least one time in a day inside the restaurant and sold food at least 5 hours time in a day • Head/manager of the restaurant was available to give informed consent for observations, and conducting surveys and spot checks Method of data collection and data collection tools - In the eligible restaurants, the survey team initially conducted face to face interviews and spot-checks of available facilities (handwashing locations, covering of foods in the kitchen and other locations, sanitation facilities and water points). For the face to face interviews and spot-checks, the field team used structure questionnaires. The team conducted 3 face-to-face interviews to determine knowledge and practices of handwashing and sanitation behaviors with the manager/ owner of the restaurant (1), cook/food maker (1) and service boy (1). In order to check the safe drinking water serving practices of service staff at restaurants, the interviewers requested the service a glass of drinking water and then recorded the critical handwashing behaviour before serving drinking water. The team observed handwashing skills of service staff by conducting handwashing demonstrations. At the end, the team members conducted 90-minute structured observations of the handwashing behaviors of the restaurant staff and their customers. Structured list of questionnaire was used for recording the handwashing behaviors. 6 National Hygiene Survey 2018

3.4 Street food vendor Component

As described for restaurants, the field team listed all available street food vendors in public points in or nearby to the household survey clusters where the people of survey communities mostly travel and have food from the street food vendors. The field team identified street food vendors by conducting transact walks and in discussion with the key informant in the communities. Thereafter four street food vendors were selected from the list for each cluster by using the simple random sampling technique. Since street vendors are highly mobile, many of the sampled vendors from the list were not available during the time of the survey. In these cases, food vendors available in the list at the time of the survey were interviewed instead. Eligibility criteria: Street food vendors were considered eligible for the survey if they met the following criteria: • Reported as a source of readymade food by household members during the survey • Available to give informed consent and spot-check • Sold at least one food item which was made involving his/her own hand contact Method of data collection and data collection tools - In the eligible food vendors, the survey team initially conducted face to face interviews and spot-checks of available facilities (handwashing locations, covering of foods, sanitation facilities and water points). For the face to face interviews and spot-checks, the field team used structured questionnaires. The team conducted face-to-face interviews to determine knowledge and practices of handwashing and sanitation behaviors with food vendors. In order to check the safe drinking water serving practices of food vendors, the interviewers requested the service a glass of drinking water and then recorded the critical handwashing behaviour before serving drinking water. The team observed handwashing skills of service staff by conducting handwashing demonstrations. At the end, the team conducted 90-minute structured observations of the handwashing behaviors of the food vendors and its customers. Handwashing events before food contacted events and after fecal contacted events were recorded. The urination events were confirmed by the observers if there was no symptom of defecation in latrine/toilet. Structured set of questionnaire was used for recording the handwashing behaviors.

3.5 Health facility Component

Initially the field team listed up to 10 health facilities (tertiary level hospitals or private/non-government health facilities that provide overnight inpatient healthcare facilities) within the upazila in which the household cluster was sampled. This listing was done in consultations with key informants in the communities, and upazila (sub-district) level health offices. Thereafter five health facilities were selected randomly for conducting face to face survey, spot-checks and structured observations. If any of the sampled facility authorities refused to participate in the study, the field team replaced it with the nearest facility from the list. In this way 880 sampled health facilities data were collected i.e; 5 from each of the 176 clusters/PSUs. In case there were insufficient numbers of health facilities in the upazila, the full district was considered instead of the upazila, and the health facility closest to the PSU was listed. In the sampled health facilities 4 wards (1 male ward, 1 female ward, 1 pediatric ward and 1 common ward) were chosen for spot checks related to sanitation and hygiene. At each cluster, out of the sampled five facilities, one health facility was sampled for conducting 5-hour structured observations of handwashing practices of hospital staff and patients/caregivers available in the observation wards. The health facility at each PSU/cluster for structured observations was selected based on the facility that had maximum number of patients admitted. The maximum number of patients in a facility was determined from the data of face to face interview and spot checks section. Usually structured observations were conducted in the pediatric ward in the sampled facility. In case the pediatric ward not available, the ward that was attended by the maximum number of patients was selected for the structured observations. Eligibility criteria: Health facilities were considered eligible for the survey if the following criteria were met: • tertiary level hospitals or private/ non-government health facilities that provided overnight inpatient healthcare service inside the health facilities National Hygiene Survey 2018 7 • At least one patient admitted to health facility on the day of interview • Facility head was available to give informed consent for the survey, spot check and 5-hour structured observation Method of data collection and data collection tools - The team conducted 5-hour structured observations in the pediatric ward at in the sampled health facilities using structured set of checklist of handwashing behaviors. The observations included handwashing before touching patients, before conducting clean/aseptic procedures, after body fluid exposure or toileting, after touching patients or wounds, after touching patient surroundings in addition to other key handwashing events such as before feeding, before eating, after general cleaning, after sneezing/coughing and before preparing/serving food or water. The team observed handwashing skills of service staff by conducting handwashing demonstrations. The urination events were confirmed by the observers if there was no symptom of defecation in latrine/toilet. Once the team started recording a handwashing event to observe, the observation of that event continued till the event ends, and then started to observe another handwashing event. The field team conducted interviews with doctors/administrators of the facilities, nurse and ward boy/ayas. The spot-checks were conducted to the available facilities including toilets, water points, handwashing locations and handwashing agents at each of the wards (male, female, pediatric and common wards) and corridors in the facilities. The team also recorded the functionality of each and other basic information including number of beds, average number of admitted patients per day, number of doctors and nurses. Finally, the team conducted environmental cleanliness and general waste disposal systems in hospital compounds and spot checks in clinic disposal systems in pathological labs (if available).

4. Instrument designing

The data collection instruments were designed by Demography and Health Wing Team of BBS. The BBS team arranged several consultations workshops together with the experts from BBS, WaterAid, Institute of Statistical Research and Training (ISRT) of Dhaka University, and the International Center for Diarrheal Diseases Research, Bangladesh (icddrb).

5. Recruitment of field team, training and data collection

For data collection, 25 teams of skilled field professionals were recruited. Each team comprised 1 supervisor, 1 male enumerator and 3 female enumerators. The teams attended seven days in-house training and one day field practice in Dhaka prior to data collection. Training was conducted together and sequentially for all study components (household, school, restaurant and food hygiene and health facilities) one after another. Immediately after the in-house training, day-long field test was conducted based on the integrated set of tools. Data collection started from 23 March 2018 and continued till 14 May 2018.

6. Data quality control

Data were captured using paper based data collection technique. The supervising team had multiple levels of supervision, and ongoing quantitative evaluation of the amount of work achieved by each field team and its quality. The study team developed standard operating procedures for data collectors, supervisors and followed those standard operating procedures to ensure quality of data. The supervising team monitored the field work by assessing the number of household and school visits completed each week by each field team. The field level team supervisor reviewed all collected data daily to ensure that it was in the proper format. Collected data was checked regularly for completeness and consistency. Data processing team was responsible to entry of the survey data using CS Pro Software. The data entry team maintained strong liaison with the data collection team in case any error was noted in the paper-based questionnaire.

7. Data analysis

Data were analyzed using STATA©version 14.2. For determining standard estimates representing each cluster population, weighted proportions and means were calculated by adjusting the inverse probability weighting (listed numbers/sampled numbers) for national estimates. Data are mostly presented in figures and tables. For preparation of the wealth index variable, principal component analysis (PCA) was done and a proxy variable of wealth scores using variables of household ownership of assets, utilities and services was created. The wealth scores are grouped into quintiles called – poorest, 2nd, 3rd, 4th and richest quintiles. Numerous studies have used such technique of poverty analysis (Vyas and Kumaranayake 2006).

8. Ethical considerations

Each member of the field team received formal training on how to protect the rights of the participants prior to conducting our survey, including obtaining written informed consent. As part of the consent process the team made clear the amount of time they were asking prospective participants to give. They explained that there were no individual benefits or compensation for participating, that there would be questions about use of water or toilet facilities and handwashing, and they noted that these topics could be sensitive, and/ or that it could be uncomfortable to have a stranger interviewing them and conducting spot check in their household or in the school. During the consent process the field workers specified that participation was voluntary. They explained to the beneficiaries that they could withdraw their consent at any time. Study supervisors made unannounced visits to field teams to ensure that the enrolment and consent process were followed.

9. Limitations

Household survey followed the population based random sampling technique for example BDHS or MICS in Bangladesh and so it is a nationally representative hygiene survey. However, samples for school, restaurant, food vendor and health facilities were selected from a list based on the availability of National Hygiene Survey 2018 9 the respected components in the household cluster or nearby. Since the list is not exhausted, the results may not be represented nationally. Sampling strategy followed the replacement criteria to maintain the sample size and therefore, this may have deviated from the standard sampling strategy of ‘no replacement’. The study followed paper based data collection technique. Therefore, for such a huge volume of data to enter may have had some level of errors. To minimize data entry error, the data entry supervisors rechecked at least 5% of the entered data and analyzed the level of errors were done, shared the errors with data entry members. After data entry and editing, if any further inconsistency was observed, the analysis team cross-checked and corrected the errors. When conducting structured observations at health facilities, restaurants and food vendors it was not possible to maintain a uniform time slot due to travelling distance, lack of transportation facilities and consent taking process. However, this probably had minimal impact on sampling error because the observations were uniformly conducted during the official hours at health facilities and peak business hours at restaurants and food vendors.

Chapter 3

Indicators at a glance

Household Component