Assessment of the quality of data on maternity care in the Demographic and Health Surveys 1986-1989.

Data quality was evaluated for prenatal care of current and prior pregnancies within 5 years of the Demographic and Health Survey for 27 countries in 4 major regions. A brief review and statistical chart of data availability was provided of the type of information received from the questionnaires on prenatal care tetanus vaccination and service provision. Results were presented for internal and external consistency checks. There appeared to be reliable estimates of the number of pregnancies at the time of the survey based on a comparison with births in the 2 years prior to the survey. Examination of pregnancy by month confirmed the expected underreporting in early pregnancy. Higher ratios in the second trimester may have been due to inaccurate recall of the menstrual period. The expected trend for increased tetanus toxoid (TT) immunization with increased pregnancy duration was confirmed. Missing values for TT during pregnancy was <2% and was complete in 45% of surveys. The check on quality of prenatal care (PNC) data showed the appropriate proportion reported receiving care from "no one" in early pregnancy and decreasing "no one" responses with pregnancy duration. 58% of surveys had no missing data on the question of who provided PNC during the pregnancy. The reliability check for pregnancies 5 years prior to the survey was to compare PNC and TT immunization. The results showed coverage was the same or higher for previous births than for current pregnancies with the exception of Egypt Guatemala and Senegal for TT and of Colombia Guatemala Sudan and Tunisia for PNC. Data on live births in previous pregnancies showed Trinidad and Tobago had the highest "dont know" (DK) responses (7.0%) for TT. For decreased children the highest DK responses were in Ecuador (10.9%) Mali (9.7%) Sri Lanka (11.6%) Thailand (12.1%) Trinidad and Tobago (13.6%) Togo (10%) and Zimbabwe (17.0%). High frequency responses of DK for PNC provider for decreased children occurred in Ecuador (10.9%) Mali (9.7%) Sri Lanka (11.6%) Thailand (12.1%) Trinidad and Tobago (13.6%) Togo (10%) and Zimbabwe (17%). Multivariate analysis confirmed the impact of dead children on accurate reporting. Problems were identified by country in chart form. 74% of the surveys had at least one problem.