Estimating abortion safety: advancements and challenges

In The Lancet, Bela Ganatra and colleagues present an innovative and important analysis of global abortion safety, in which they attempt to move beyond the binary understanding (safe or unsafe) of abortion safety. As the availability of misoprostol increases, and abortion telemedicine services reach more women worldwide, fewer women are undergoing abortions with invasive or outdated methods and more women are having abortions outside of formal health-care systems. These changes prompt a need for rethinking how we view and measure abortion safety. Therefore, the study by Ganatra and colleagues is very timely. The approach used in the study, although it had limitations, offers a more nuanced gradation of safety: abortions were classified as safe or unsafe, and unsafe abortions were further divided into less-safe and least-safe categories. This three-tiered classification focused on two technical aspects of the abortion process (abortion provider and method) and is said by the authors to be aligned with the conceptual definition of unsafe abortion used by WHO. However, the safety classification did not consider abortion outcomes, as was recommended by Sedgh and colleages. Outcomes were instead considered by examining the association between abortion safety and case fatality rates. An editorial on how to operationalise and interpret the WHO definition of unsafe abortion states that, rather than a binary measure, abortion safety should be characterised along a risk continuum, which is affected by contextual factors, such as abortion laws and presence of stigma. The model and analysis used by Ganatra and colleagues did, to some extent, take the social and legal context into account. The authors divided factors affecting abortion safety into five conceptual domains: abortion service delivery environment; legal context of abortion; financial access to services; abortion stigma; and development. However, the model predictors used cannot be said to completely represent these domains because of the unavailability of predictor data. For example, although gender inequality might be the best available proxy for stigma, measures of gender inequality and data on the abortion process (provider type and abortion method) cannot capture all scenarios. Abortion stigma has a substantial impact on access to both safe abortion and post-abortion care. Young women who seek abortion (including from trained providers using evidence-based methods) sometimes turn to unsafe methods to manage bleeding and delay seeking care for complications because of costs and fear of stigma, exposure, and legal repercussions. Not accounting for these types of outcomes results in an incomplete picture. Efforts are needed to measure and quantify abortion stigma to understand its implications for safety and quality of care. Another issue, recognised by the authors, is the poor association between drug registration and availability. Registration of mifepristone and misoprostol provides no assurance of the drugs being available or of high quality. The emergence of telemedicine services and mifepristone and misoprostol sold on the black market have further confounded the issue. Additionally, information about availability of health-care providers who are willing and able to provide abortion care could be considered for future estimates because these providers are crucial for access to health care. It is unclear whether the proportion of abortions estimated as safe by the authors (54·9%, 90% uncertainty interval 49·9–59·4) would remain the same if data on availability of mifepristone, misoprostol, and health-care providers; stigma; and abortion outcomes had been obtainable and taken into account. It is remarkable that, despite being preventable, 25·1 million unsafe abortions were estimated to have Published Online September 27, 2017 http://dx.doi.org/10.1016/ S0140-6736(17)32135-9