Rethinking WHO guidance on misoprostol use in the prevention of postpartum heamorrhage: authors’ response

Abstract

Pathfinder takes exception to our methods but provides a false account of our analysis nor does it say how or why the use of the PRISM instrument would have come to a different conclusion. Pathfinder notes that the WHO decision was based on 21 and not 4 studies as argued in our paper. This is true for overall safety and efficacy profile of misoprostol. However, the sections directly relevant to community settings are based on 4 RCTs. WHO unedited report states that with respect to the prevention of PPH ‘The application provided the results from 4 RCTs (278, 279, 280, 281) to support the use of misoprostol for the prevention of PPH in settings where injectable uterotonics are not available or feasible to use’.2 Pathfinder criticizes us for not conducting a meta-analysis however the heterogenous nature of the studies meant this was not possible. Pathfinder might note that the Cochrane systematic review could not conduct a meta-analysis of community trials and comes to the same conclusion ‘all four recent trials have design and setting differences that make the summing up of their results difficult.’3 We identified randomized controlled trials of misoprostol in community and home setting from academic databases and systematic reviews and meta-analyses; our methods critically appraised study design, intervention and outcomes of these studies (see method section). The recently published Cochrane Systematic Review ‘Prostaglandins for preventing postpartum haemorrhage’3 looked at a total of 72 trials (52,678 women) conducted in low-, middle- and high-income countries included 68 trials conducted in hospitals and performed by skilled caregivers and four community studies. Our search of academic databases identified the same 4 RCTs, the same four studies discussed in the WHO unedited report. According to the Cochrane review of the four studies, the Guinea-Bissau trial showed that almost half the women in the trial experienced higher than usual blood loss while the Gambia trial was inadequately powered. Derman's own study, the India 2006c trial, was deemed more applicable to community settings where the ‘expectant’ management of the third stage of labour is the norm. Although the results of this study were significant (RR 0.20; 95% CI 0.04 to 0.91, 2/812 versus 10/808), Cochrane reviewers do not discuss the weakening of the association by the observation of temporal trends in the intervention and control arms (suggesting effects from factors other than misoprostol) nor do they mention the lack of generalisability because the RCT excluded women with or at high risk of complications. None of the responses to our paper engage with the science and significantly all overlook the absence of generalizability of findings. They ignore the fact that, even where tests are significant, if the study excludes all or most women with pregnancy complications or trials include antenatal screening and other services that are not available to women outside the trial, the applicability and generalizability of the study findings are seriously limited. In this respect, Derman's response to our review is curious not least given his earlier acknowledgement of the lack of generalizability of the findings of his own study4 and the existence of temporal trends5, two criteria which he lists as important for causal associations. As Kerr and Potts note conducting trials in community setting is difficult but the same standards of evidence apply and reviews of evidence and recommendations based on these must not use difficulties in study design to skate over the shortcomings and a lack of strong evidence base. Our review is concerned with science and evidence underpinning the WHO decision to add misoprostol to the Essential Medicines List. As such it provides an important new contribution to the science in that it provides a separate detailed appraisal of randomized control studies evaluating misoprostol for PPH prevention in community or home settings of low and middle income countries, neither of which has been done in the same detail by the recent Cochrane Review3 or by the WHO.2 In sum, our study shows that current available evidence does not support misoprostol use for pregnant women in community settings in the absence of skilled birth attendants, antenatal screening and good referral systems. Misoprostol is already being distributed in the community in many countries, e.g. in Uganda and Nepal and Bangladesh. On the basis of current evidence we have written to the WHO to ask it to review and rescind its recent decision to add misoprostol to the EML and to review the networks involved and potential conflicts of interest behind the promotion of non evidenced based therapy for women.