Bandafassi HDSS INDEPTH Core Dataset 1970 - 2014 (Release 2017)

INITAL CENSUS - The initial census was followed by several surveys designed to improve the information of the census and collect other data needed for subsequent studies. These included an age survey to estimate ages of adults and children or improve the unreliable data collected on these during the census. It also included a genealogical survey to collect genealogies, going up to known ascendants and down to living collateral relatives. One use of the genealogies in the project is to get detailed information on the relationships between members of a compound and in particular the relationship of each one to the head of the compound (Pison 1985). Finally, a union- and birth-histories survey was conducted for adult men and women.
At the census, a person was considered a member of the compound if the head of the compound declared it to be so. This definition was broad and resulted in a de jure population under study. Thereafter, a criterion was used to decide whether and when a person was to be excluded or included in the population.

A person was considered to exit from the study population through either death or emigration. Part of the population of Mlomp engages in seasonal migration, with seasonal migrants sometimes remaining 1 or 2 years outside the area before returning. A person who is absent for two successive yearly rounds, without returning in between, is regarded as having emigrated and no longer resident in the study population at the date of the second round. This definition results in the inclusion of some vital events that occur outside the study area. Some births, for example, occur to women classified in the study population but physically absent at the time of delivery, and these births are registered and included in the calculation of rates, although information on them is less accurate. Special exit criteria apply to babies born outside the study area: they are considered emigrants on the same date as their mother.
A new person enters the study population either through birth to a woman of the study population or through immigration. Information on immigrants is collected when the list of compounds of a village is checked ("Are there new compounds or new families who settled since the last visit?") or when the list of members of a compound is checked ("Are there new persons in the compound since the last visit?"). Some immigrants are villagers who left the area several years before and were excluded from the study population. Information is collected to determine in which compound they were previously registered, to match the new and old information.

Information is routinely collected on movements from one compound to another within the study area. Some categories of the population, such as older widows or orphans, frequently move for short periods of time and live in between several compounds, and they may be considered members of these compounds or of none. As a consequence, their movements are not always declared.

REGULAR UPDATE ROUNDS - The Mlomp DSS is a multiround demographic surveillance, with annual rounds. Once each year, in February and March, all compounds are visited, and information on events occurring since the last visit is collected. This is done in three steps. First, the list of people present in each compound at the preceding visit is checked, and information is obtained on new births, marriages, migrations, deaths, and current pregnancies. Information is provided by the head of the compound or key informants in the village or hamlet. The information on events is recorded directly on the nominative list.

CONTINUOUS SURVEILLANCE - Information provided by local registers is matched with that collected independently during the surveillance. Information from registers with fair quality is used to systematically correct errors and complete the information collected at the yearly rounds. These are maternity-clinic registers (for prenatal visits and deliveries), civil and parish registers (for births), and dispensary or hospital registers (for death, growth monitoring, and vaccinations). The local dispensary collaborates with the research project, and one completes several registers, in particular a death register. Although the local registers rarely cover the entire population and are sometimes subject to errors, using them improves the quality and the precision of data. Verbal autopsies (VAs) have been performed for all deaths since the beginning of the study. For each death identified in the first step of the annual surveillance, information on its cause is obtained from a close relative of the dead person, usually the mother in the case of a child's death, using a VA questionnaire.

Data management and analysis
Information collected during the baseline and follow-up surveys has been coded and stored in databases designed in 1980s, with some adaptations since then. The information collected during each annual surveillance is processed in two steps: in the villages, it is entered into laptops, with state-of-the-art software, during the surveillance; thereafter, the information is verified and added to the database, using PostgreSQL software.