We have traditionally depended on household survey estimates of the coverage of contacts with health facilities to understand access, for example institutional delivery.   And used facility survey approaches separately to track quality indicators in facilities.   But the need to track the coverage of high quality care requires standardised methods for bringing these two data platforms together.

The manuscript Linking data sources for measurement of effective coverage in maternal and newborn health: what do we learn from individual- vs ecological-linking methods? published by the IDEAS researcher Barbara Willey et. al. reports on methods for linking household and facility survey data to generate an effective coverage measure of high quality skilled attendance at birth in Uganda.  It links household survey data collected at the same time and place as data from a census of health facilities.  A gold-standard effective coverage measure was generated by combining women’s reports in the household survey about the specific health facility they delivered in with information from the facility survey about the quality of care available in that specific facility (individual linking).  This was then compared to an alternative and more practical method of linking, that relied on aggregate measures of facility quality rather than specific facility information (ecological linking).

From the household survey, we learned that 55% of women with a recent birth reported an institutional delivery with a skilled birth attendant.  Using the individually linked gold-standard method that adjusted for inputs available in the individual facilities, this crude coverage measure revealed that only 10% (95% CI 3-17) of women received quality skilled attendance at birth.   The ecological method of linking household and facility data that adjusted for the level of health facility generated a very close estimate to this gold-standard: 11% (95% CI 4-18).  When linking data sources from household surveys and facilities without adjustment for specific facility levels, results did not come close to the gold standard. This is mainly due to the fact that different types of facilities provide very different levels of quality in their care and also do not have a comparable volume of patients to attend to.

The magnitude of the quality gap between crude and effective coverage measures is shocking but not new.  But in this manuscript the new evidence on methods for linking household and facility datasets demonstrates the potential – and importance – of standardised effective coverage methods to better understand progress towards universal life saving care.   For maternal and newborn care this means linking methods that take into account the characteristics of different facility types are preferable.