A pilot program that provided vouchers to pregnant women in Bangladesh was effective in increasing use of maternal health services, a recent analysis suggests.1 After initiation of the program, which provided free access to antenatal care, institutional delivery and other services, women in the voucher areas who had recently given birth were more likely than their counterparts in comparison areas to have had three antenatal visits (55% vs. 34%), a qualified provider at their delivery (64% vs. 27%), an institutional delivery (38% vs. 19%) and at least one postnatal care visit with a qualified provider (30% vs. 15%). In multivariate analyses that controlled for baseline differences between groups, women in the intervention areas were 35 percentage points more likely than those in comparison areas to have had a delivery by a qualified provider and 18 percentage points more likely to have had an institutional delivery.

Financial barriers are among the most important obstacles to use of maternal health services in developing countries, including Bangladesh, where most women give birth at home without the aid of a trained provider. To encourage use of antenatal, delivery and postnatal care, Bangladesh initiated a voucher program in 2004. Originally launched in just two of Bangladesh’s 489 subdistricts, the pilot program remains limited in scale; in 2011, it was available in only 46 subdistricts, and to about 10 million people.

The program provides pregnant women with vouchers that allow them to obtain, at no charge, the following services: three antenatal care visits; safe delivery at a facility, or at home, with a trained provider; emergency obstetric care, if necessary; and one postnatal care visit. In addition, participants receive cash to help cover transportation costs, as well as a gift box and another cash payment following delivery by a qualified provider. Women are eligible for the program only if they are having their first or second delivery; if it is their second, they must have practiced contraception between pregnancies. Most participating subdistricts offer the program only to poorer women, though some provide vouchers regardless of household income and assets.

Because rigorous evidence of the program’s success was lacking, researchers conducted a large household-based survey in 2009. The survey was conducted in 32 subdistricts, half of which had implemented the voucher program two years earlier; the remaining districts, which were chosen because they were near an intervention subdistrict that had a similar literacy rate and number of hospital beds, served as a comparison group. In each subdistrict, the researchers chose several villages that had a large number of births and attempted to interview all women who had had a singleton delivery in the past six months. The 2,208 respondents provided information about their social and demographic characteristics; they also answered questions regarding their use of health services related to each of their deliveries in the past five years and their out-of-pocket expenditures for their latest birth.

Using cross-sectional, multivariate and panel analyses, the investigators compared use of maternal health services in the intervention areas with that in the comparison areas, controlling, where appropriate, for respondent, household and birth characteristics, as well as for preintervention levels of service use and subdistrict service capacity (number of community skilled birth attendants and availability of emergency obstetric care). The primary outcomes of interest were use and number of antenatal care visits, institutional delivery, delivery by a qualified provider (doctor, nurse, midwife, paramedic, family welfare visitor or community skilled birth attendant), cesarean delivery, postnatal care, and total out-of-pocket expenditures for pregnancy- and delivery-related services.

The final analytic sample consisted of 2,861 deliveries—the 2,208 most recent births, 141 earlier births that had occurred after the voucher program’s initiation and 512 deliveries that predated the program. On average, women in the intervention areas were slightly younger than those in the comparison areas (24 vs. 25 years); in both groups, about half of women had no more than a primary education and nine-tenths were Muslim. Intervention areas had a slightly higher density of community skilled birth attendants than did comparison areas (eight vs. seven per 10,000 population) and were much more likely to have emergency obstetric care available (81% vs. 22%).

For their most recent births, women in the voucher areas were more likely than those in comparison areas to have had at least one antenatal visit (92% vs. 76%), three antenatal visits (55% vs. 34%), a qualified provider at the delivery (64% vs. 27%) and at least one postnatal care visit with a qualified provider (30% vs. 15%); they were also more likely to have delivered in an institution (38% vs. 19%). The proportion of deliveries done by cesarean section did not differ between groups (9–10%), a “notable” finding, the authors say, given concerns that the voucher program might encourage providers to perform unnecessary cesareans because such deliveries are reimbursed at higher rates than vaginal deliveries. Although the proportion of women who incurred out-of-pocket expenses was similar in the two groups (87–90%), women in the voucher group paid 34% less, on average, than did women in the comparison group.

Additional analyses suggest that the differences in service use between the voucher and comparison areas increased over time. For example, the proportion of deliveries attended by a qualified provider was slightly higher in the intervention areas than in the comparison areas before program implementation (19% vs. 14%), but the difference was much larger for births during the first part of the voucher period (33% vs. 15%) and greater still for women’s most recent births (64% vs. 28%). Similarly, the proportion of deliveries that took place in institutions rose slowly at first in intervention areas but then increased dramatically (from 15% at baseline to 20% and then 38%), while the increases in comparison areas were consistently small (from 11% to 13% to 18%). Subanalyses revealed that increases in service use among women in the poorest quintile were at least as large as, and in one case larger than, those among wealthier women.

Three different types of multivariate analysis provided further evidence of the program’s benefits. For example, difference-in-difference regression analyses that controlled for baseline differences in service provision and other variables found that the proportion of deliveries attended by qualified providers was 35 percentage points higher in the intervention areas than in the comparison areas; the corresponding difference for institutional delivery was 18 percentage points. Other regression approaches (cross-sectional multivariate regression and fixed-effects regression) yielded similar results. None of the analyses found differences in rates of cesarean delivery.

Overall, the findings suggest that the voucher program had “large and positive effects on the utilization of maternal health services,” according to the authors. Such increases in service use, they note, could have a “substantial” impact on maternal and infant health. However, they caution that if the voucher program is expanded, the benefits in other regions may be smaller than those found here, in part because the density of reproductive health facilities and providers in the intervention areas was greater than is typical in Bangladesh.—P. Doskoch


1. Nguyen HTH et al., Encouraging maternal health service utilization: an evaluation of the Bangladesh voucher program, Social Science & Medicine, 2012, 74(7):989–996.