Calculation of adult mortality from family reconstitution data is problematic because of the difficulty of determining with any precision, for those individuals for whom a death date is unknown, the period during which they are present in observation and hence at risk of dying. In the special case of maternal mortality, provided it is defined as the risk of dying during or shortly after confinement and is expressed as a rate relative to the number of confinements, this problem is essentially absent. The beginning of the period of risk is clearly defined by the birth of a child and ends, according to different definitions, within a few weeks or months following confinement. Since it is unlikely that many women migrate out of a village shortly after giving birth, and that those few who do are unlikely to have died in some other village within the specified period, women for whom no death date is known can be safely assumed to have survived the critical period after confinement.

Issues of measurement and definition

In measuring maternal mortality from reconstituted family histories, there are special problems that are essentially inherent to the parish register sources on which these histories are typically based. The most important one is the fact that maternal mortality may be associated with miscarriages and undelivered pregnancies, which are often not recorded at all, or with stillbirths, which are less than fully recorded in the registers (see Appendix B).