Abstract

The 2017-18 Bangladesh Demographic and Health Survey (2017-18 BDHS) is a nationwide survey with a nationally representative sample of approximately 20,250 selected households. All ever-married women age 15-49 who are usual members of the selected households or who spent the night before the survey in the selected households were eligible for individual interviews. The survey was designed to produce reliable estimates for key indicators at the national level as well as for urban and rural areas and each of the country’s eight divisions: Barishal, Chattogram, Dhaka, Khulna, Mymensingh, Rajshahi, Rangpur, and Sylhet.

The main objective of the 2017-18 BDHS is to provide up-to-date information on fertility and fertility preferences; childhood mortality levels and causes of death; awareness, approval, and use of family planning methods; maternal and child health, including breastfeeding practices and nutritional status; newborn care; women’s empowerment; selected noncommunicable diseases (NCDS); and availability and accessibility of health and family planning services at the community level.

This information is intended to assist policymakers and program managers in monitoring and evaluating the 4th Health, Population and Nutrition Sector Program (4th HPNSP) 2017-2022 of the Ministry of Health and Family Welfare (MOHFW) and to provide estimates for 14 major indicators of the HPNSP Results Framework (MOHFW 2017).

Geographic coverage

National coverage

Analysis unit

• Household
• Individual
• Children age 0-5
• Woman age 15-49
• Community

Universe

The survey covered all de jure household members (usual residents), all women aged 15-49 and all children aged 0-5 resident in the household.

Kind of data

Sample survey data [ssd]

Sampling procedure

The sample for the 2017-18 BDHS is nationally representative and covers the entire population residing in non-institutional dwelling units in the country. The survey used a list of enumeration areas (EAs) from the 2011 Population and Housing Census of the People’s Republic of Bangladesh, provided by the Bangladesh Bureau of Statistics (BBS), as a sampling frame (BBS 2011). The primary sampling unit (PSU) of the survey is an EA with an average of about 120 households.

Bangladesh consists of eight administrative divisions: Barishal, Chattogram, Dhaka, Khulna, Mymensingh, Rajshahi, Rangpur, and Sylhet. Each division is divided into zilas and each zila into upazilas. Each urban area in an upazila is divided into wards, which are further subdivided into mohallas. A rural area in an upazila is divided into union parishads (UPs) and, within UPs, into mouzas. These divisions allow the country as a whole to be separated into rural and urban areas.

The survey is based on a two-stage stratified sample of households. In the first stage, 675 EAs (250 in urban areas and 425 in rural areas) were selected with probability proportional to EA size. The sample in that stage was drawn by BBS, following the specifications provided by ICF that include cluster allocation and instructions on sample selection. A complete household listing operation was then carried out in all selected EAs to provide a sampling frame for the second-stage selection of households. In the second stage of sampling, a systematic sample of an average of 30 households per EA was selected to provide

statistically reliable estimates of key demographic and health variables for the country as a whole, for urban and rural areas separately, and for each of the eight divisions. Based on this design, 20,250 residential households were selected. Completed interviews were expected from about 20,100 ever-married women age 15-49. In addition, in a subsample of one-fourth of the households (about 7-8 households per EA), all ever-married women age 50 and older, never-married women age 18 and older, and men age 18 and older were weighed and had their height measured. In the same households, blood pressure and blood glucose testing were conducted for all adult men and women age 18 and older.

The survey was successfully carried out in 672 clusters after elimination of three clusters (one urban and two rural) that were completely eroded by floodwater. These clusters were in Dhaka (one urban cluster), Rajshahi (one rural cluster), and Rangpur (one rural cluster). A total of 20,160 households were selected for the survey.

For further details on sample selection, see Appendix A of the final report.

Mode of data collection

Computer Assisted Personal Interview [capi]

Research instrument

The 2017-18 BDHS used six types of questionnaires: (1) the Household Questionnaire, (2) the Woman’s Questionnaire (completed by ever-married women age 15-49), (3) the Biomarker Questionnaire, (4) two verbal autopsy questionnaires to collect data on causes of death among children under age 5, (5) the Community Questionnaire, and the Fieldworker Questionnaire. The first three questionnaires were based on the model questionnaires developed for the DHS-7 Program, adapted to the situation and needs in Bangladesh and taking into account the content of the instruments employed in prior BDHS surveys. The verbal autopsy module was replicated from the questionnaires used in the 2011 BDHS, as the objectives of the 2011 BDHS and the 2017-18 BDHS were the same. The module was adapted from the standardized WHO 2016 verbal autopsy module. The Community Questionnaire was adapted from the version used in the 2014 BDHS. The adaptation process for the 2017-18 BDHS involved a series of meetings with a technical working group. Additionally, draft questionnaires were circulated to other interested groups and were reviewed by the TWG and SAC. The questionnaires were developed in English and then translated into and printed in Bangla. Back translations were conducted by people not involved with the Bangla translations.

Cleaning operations

Completed BDHS questionnaires were returned to Dhaka every 2 weeks for data processing at Mitra and Associates offices. Data processing began shortly after fieldwork commenced and consisted of office editing, coding of open-ended questions, data entry, and editing of inconsistencies found by the computer program. The field teams were alerted regarding any inconsistencies or errors found during data processing. Eight data entry operators and two data entry supervisors performed the work, which commenced on November 17, 2017, and ended on March 27, 2018. Data processing was accomplished using Census and Survey Processing System (CSPro) software, jointly developed by the United States Census Bureau, ICF, and Serpro S.A.

Response rate

Among the 20,160 households selected, 19,584 were occupied. Interviews were successfully completed in 19,457 (99%) of the occupied households. Among the 20,376 ever-married women age 15-49 eligible for interviews, 20,127 were interviewed, yielding a response rate of 99%. The principal reason for non-response among women was their absence from home despite repeated visits. Response rates did not vary notably by urbanrural residence.

Sampling error estimates

The estimates from a sample survey are affected by two types of errors: nonsampling errors and sampling errors. Nonsampling errors are the results of mistakes made in implementing data collection and data processing, such as failure to locate and interview the correct household, misunderstanding of the questions on the part of either the interviewer or the respondent, and data entry errors. Although numerous efforts were made during the implementation of the 2017-18 Bangladesh Demographic and Health Survey (BDHS) to minimize this type of error, nonsampling errors are impossible to avoid and difficult to evaluate statistically.

Sampling errors, on the other hand, can be evaluated statistically. The sample of respondents selected in the 2017-18 BDHS is only one of many samples that could have been selected from the same population, using the same design and expected size. Each of these samples would yield results that differ somewhat from the results of the actual sample selected. Sampling errors are a measure of the variability among all possible samples. Although the degree of variability is not known exactly, it can be estimated from the survey results.

Sampling error is usually measured in terms of the standard error for a particular statistic (mean, percentage, etc.), which is the square root of the variance. The standard error can be used to calculate confidence intervals within which the true value for the population can reasonably be assumed to fall. For example, for any given statistic calculated from a sample survey, the value of that statistic will fall within a range of plus or minus two times the standard error of that statistic in 95% of all possible samples of identical size and design.

If the sample of respondents had been selected as a simple random sample, it would have been possible to use straightforward formulas for calculating sampling errors. However, the 2017-18 BDHS sample is the result of a multi-stage stratified design, and, consequently, it was necessary to use more complex formulas. Sampling errors are computed in SAS, using programs developed by ICF. These programs use the Taylor linearization method to estimate variances for survey estimates that are means, proportions, or ratios. The Jackknife repeated replication method is used for variance estimation of more complex statistics such as fertility and mortality rates.

Note: A more detailed description of estimates of sampling errors are presented in APPENDIX B of the survey report.

Data