The removal of user fees in Kenya, Ghana and Senegal resulted in 3.1 additional facility-based deliveries per 100 live births, representing a 5% increase, according to an analysis of Demographic and Health Survey (DHS) data from an array of Sub-Saharan African countries.¹ The estimated increase was smaller than that reported in prior analyses of individual countries, in part because those studies, unlike the current one, did not account for the increases that likely would have occurred even if user fees had still been required. Rates of caesarean delivery did not change following elimination of user fees; neonatal mortality declined by 9%, although the decrease was not statistically significant.

The new analysis used DHS data collected between 1997 and 2012 to examine whether levels of the three outcomes—delivery at a facility, caesarean delivery and neonatal mortality—changed when countries eliminated user fees for facility-based delivery services. Three countries in Sub-Saharan Africa eliminated some or all such fees during the study period and served as the intervention group: Ghana, which eliminated fees for deliveries at all public, private and faith-based facilities in 2003 (selected regions) and 2005 (elsewhere); Kenya, which removed fees for deliveries at public dispensaries and health centers, but not at hospitals, in 2007; and Senegal, which eliminated fees at public facilities in most of the country in 2005–2006.

The comparison group consisted of a pool of seven countries (Cameroon, Congo, Ethiopia, Gabon, Mozambique, Nigeria and Tanzania) that did not change their delivery fee policies. Although three of these countries had existing policies that exempted women from paying fees for maternity care, these exemptions were not widely known and were rarely enforced. In addition, because changes in fee policies in Ghana and Senegal were implemented in stages, women who gave birth during the transition period before fees were eliminated in their region of residence were included in the comparison group.

To assess the relationship between removal of user fees and the three outcomes, the researchers used difference-in-differences regressions, which examined whether outcomes changed to a greater extent in countries with policy changes than in those without. All analyses adjusted for women’s characteristics (age, area of residence, education, wealth and whether the woman had given birth previously), area fixed effects (which take into account regional characteristics) and year fixed effects (to account for time trends). In addition, to ensure appropriate comparisons, the investigators adjusted the make-up of the comparison groups for each outcome so that trends in the outcome were similar in the intervention and comparison groups prior to the removal of policy fees. For example, the comparison group for caesarean delivery excluded Nigeria, where the rate of such deliveries declined at an unusually high rate in 1995–2003; after that exclusion, the six remaining comparison countries and the three intervention countries had statistically similar rates of change in caesarean delivery levels prior to the fee removals. They also examined change levels by year to see if patterns (e.g., lags in improvements) were consistent with implementation of a new policy.

The difference-in-differences analysis of facility deliveries yielded an average marginal effect of 3.1, which can be interpreted as indicating that removal of fees was associated with an additional 3.1 deliveries per 100 births, after adjustment for covariates. This represented a 5% relative increase in the proportion of deliveries that occurred in facilities.

Fee exemptions were also associated with an estimated reduction in neonatal mortality of 2.9 deaths per 1,000 births (a 9% relative decrease), although the change fell short of statistical significance. No change was apparent in the level of caesarean deliveries. The associations between covariates and the three outcomes were consistent with findings from prior studies: First births, education level, urban residence and wealth were positively associated with facility deliveries and caesarian deliveries; first births were positively associated, and education and wealth were negatively associated, with neonatal mortality.

Because prepolicy trends in the three outcomes were more extreme in Senegal than in the other two countries that removed user fees, the researchers also conducted sensitivity analyses to examine whether data from Senegal were driving the results. One analysis revealed that removal of user fees in Ghana and Kenya resulted in 3.8 additional facility deliveries per 100 live births; another indicated that the policy change in Senegal reduced the rate of neonatal mortality by 4.3 deaths per 1,000 live births. In both analyses, the set of comparison countries was adjusted so that the parameters of the main analyses were met (i.e., prepolicy trends had to be similar in intervention and comparison countries); in each case, the result was statistically significant.

Finally, the analysis of potential lag effects found that the increase in facility deliveries was greater in the year after implementation than in the year of implementation, suggesting, according to the researchers, that "the policy change took some time to be fully implemented, which is a plausible scenario for a large national-scale program."

The investigators noted that the estimated increase in facility-based deliveries was smaller in their study than in prior studies (3.1 vs. 5–12 per 100). However, those studies, which focused on single countries, simply compared pre- and postpolicy levels and did not account for existing trends (and hence for increases in facility deliveries that likely would have occurred even without the elimination of user fees). If the current study had used that approach, it would have found statistically significant changes in all three outcomes—8.6 additional facility deliveries per 100 births, 1.3 additional caesarean deliveries per 100 births and 9.5 fewer neonatal deaths per 1,000 live births.

Limitations of the study, according to the authors, include its reliance on self-reported data and the lack of consistency in fee removal policies among countries. Nonetheless, the findings suggest that eliminating user fees "led to substantial increases in facility-based deliveries" and were consistent "with a meaningful reduction in neonatal mortality." Even after fees were eliminated, however, the proportion of births that took place in facilities was still less than 60%, indicating that other barriers—such as geographical access, transportation costs and cultural barriers—"need to be simultaneously addressed in efforts to reduce maternal and neonatal mortality."—P. Doskoch