Cover Image
close this book05. The Role of Traditional Birth Attendants in the Reduction of Maternal Mortality
View the document(introduction...)
View the documentSummary
View the documentIntroduction
View the documentThe Rationale for TBA Training
View the documentThe Role of Traditional Birth Attendants
View the documentTBA Training
View the documentEvaluation of TBA Training Programmes
View the documentTheoretical Considerations in Measuring the Outcome of Training TBAs
View the documentEvidence of Maternal Mortality Reduction from Programmes of TBA Training
View the documentAdditional Health Benefits from TBA Training Programmes
View the documentThe Role of TBAs in Referral to Essential Obstetric Care Facilities
View the documentThe Costs of TBA Training
View the documentThe Debate Continues
View the documentConclusion
View the documentReferences

Theoretical Considerations in Measuring the Outcome of Training TBAs

This review was asked to focus on maternal mortality as an outcome. However, studies of this nature are few. The main reason for this is that it is difficult and very expensive to prove significant reductions in maternal mortality in the absence of accurate vital events registration, which is absent in most developing countries. Maternal death is a comparatively rare event with an incidence of at most 0.5-1.5%, which implies a wide confidence interval. Huge sample sizes are required to prove any change in maternal mortality. For example, a household survey in Addis Ababa in 1984 estimated a maternal mortality ratio of 566 maternal deaths per 100,000 livebirths. The study was based on 45 maternal deaths identified in a survey of 32,000 households. Even such a huge undertaking has a wide 95% confidence interval, in this study calculated to range from 374 to 758 deaths per 100,000 live births (Kwast et al . 1986). It follows that it is impossible to state that a significant reduction has been observed unless the sample size, and the number of deaths, are presented. This is further illustrated in a study of maternal deaths in Kerian District, Malaysia which reported a maternal mortality decline of 41.8% based on a reduction in MMR from 189 maternal deaths per 100,000 livebirths in 1976 to 110 / 100,000 in 1980 (Yadav 1982). However, since the number of maternal deaths in the entire study period was only 35 with a total sample size of 22,977 deliveries, we can see immediately that even such a conspicuous percentage-wise decline hardly reaches statistical significance. This kind of statistical analysis is important for the interpretation of reports in which an alleged “reduction” of maternal mortality has taken place due to a programme effect, but is rarely performed.

Maternal morbidity is both an outcome in its own right and a pre-cursor for maternal death. However, the relationship between maternal morbidity and maternal death is not clear cut (Campbell & Graham 1990). Although some morbidity, e.g. infection and vaginal fistula, may be determined by service delivery practices, there is evidence that some acute maternal morbidity is physiologically inevitable. In these circumstances service availability prevents death but not the morbidity itself. There are also a number of methodological difficulties in measuring morbidity, particularly in community-based studies, which means that it can be difficult to use maternal morbidity generally to measure success of TBA training programmes (Graham et al . 1995).

Perinatal deaths are more frequent than maternal deaths. Smaller sample sizes are needed to detect change and this indicator is more often used in programme evaluations than measures of maternal mortality. However, perinatal outcomes are influenced by a number of factors in addition to obstetric care and the link between perinatal death and maternal death is not always clear cut (Akalin et al. 1997). In addition, it can be difficult to measure perinatal deaths in communities where stillbirths and early neonatal deaths are not recorded or reported.