Review of: ‘Traditional birth attendant training effectiveness: a meta‐analysis’

The training of traditional birth attendants (TBAs) was a priority in the 1980s, but its effectiveness is now questioned by many. Yet, little scientific evidence was and is available to support the enthusiasm of the past and the criticisms of today [1]. The meta-analysis recently published by Sibley and Sipe [2] is thus very timely. The authors summarized the results of 63 studies of the effectiveness of TBAs training. Unfortunately, the studies included are not listed in the references, and the tables do not allow identification of the studies included in each pooled analysis. The meta-analysis method used by Sibley and Sipe is derived from the social sciences literature and is appropriate for continuous data. Epidemiologic meta-analysis methods for categorical data and generally used in the evidence-based medicine literature might have been more appropriate to summarize the impact of training on discrete events such as neonatal and maternal mortality [3]. The authors have only included studies with an experimental or quasi-experimental design, but the data presented do not allow isolation of randomized controlled trials from other studies. The section on ‘Sampling procedure and sample size’ provides information on the sampling of the populations included in the studies, but not on random allocation between intervention and control groups. The quality of the random allocation, if any, was thus not assessed. The readers might thus want to assume that none of the included studies was a randomized controlled trial. Despite its many limitations, this meta-analysis provides potentially useful information. The data suggest that TBA training is associated with substantial improvement in TBA ‘knowledge, attitude, behavior, and advice.’ There was also a small decrease in perinatal or neonatal mortality associated with training. Two studies included data on maternal mortality, and the meta-analysis concluded that there were 8% fewer deaths among women cared for or living in areas served by trained TBAs. These results appear to confirm that training TBAs could be a useful component of a safe motherhood program, but that the impact should not be expected to be large. The availability of reference to an effective health care system is a necessary condition for TBAs to be effective [1]. In our opinion, large randomized controlled trials should be conducted to evaluate the effectiveness of TBA training as part of integrated programs. The opinions on the potential role of TBAs are conflicting enough to make experts in the field reach the level of equipoise. This is an ideal timing for a large community randomized trial.