Estimating maternal mortality: what have we learned from 16 years of surveys in Afghanistan?

Maternal mortality is notoriously difficult to measure because even in high-mortality areas maternal deaths are relatively rare. These difficulties are further compounded in low-income and conflictaffected settings such as Afghanistan. Between 2002 and 2018, maternal mortality estimates for Afghanistan have been provided by six surveys, using a variety of costly methodologies yet yielding contradictory and sometimes implausible results. The 'failure' of household surveys to provide robust data on maternal mortality in Afghanistan is a call to reconsider the value of maternal mortality measurements to assess safe motherhood interventions in low-income and conflict-affected settings. We encourage stakeholders involved in the commissioning and use of maternal mortality estimates in Afghanistan and similar contexts to take stock of experiences so far and carefully consider how to best make use of existing resources. In the short-term efforts to measure improvements in maternal health should be redirected towards measurements of access, use and quality of services for pregnant women and women giving birth, with longer-term investments towards civil registration as a source of robust maternal mortality data.