Objectives To identify where vouchers have been used for reproductive health (RH) services, to what extent RH voucher programmes have been evaluated, and whether the programmes have been effective.
Methods A systematic search of the peer review and grey literature was conducted to identify RH voucher programmes and evaluation findings. Experts were consulted to verify RH voucher programme information and identify further programmes and studies not found in the literature search. Studies were examined for outcomes regarding targeting, costs, knowledge, utilization, quality, and population health impact. Included studies used cross-sectional, before-and-after and quasi-experimental designs.
Results Thirteen RH voucher programmes fitting established criteria were identified. RH voucher programmes were located in Bangladesh, Cambodia, China, Kenya (2), Korea, India, Indonesia, Nicaragua (3), Taiwan, and Uganda. Among RH voucher programmes, 7 were quantitatively evaluated in 15 studies. All evaluations reported some positive findings, indicating that RH voucher programmes increased utilization of RH services, improved quality of care, and improved population health outcomes.
Conclusions The potential for RH voucher programmes appears positive; however, more research is needed to examine programme effectiveness using strong study designs. In particular, it is important to see stronger evidence on cost-effectiveness and population health impacts, where the findings can best direct governments and external funders.
Objectifs: Identifier là où des bons ont été utilisés pour des services de santé reproductive (SSR), dans quelle mesure les programmes de bons pour des SSR ont étéévalués et si les programmes ont été efficaces.
Méthodes: Une recherche systématique des revues publiées et de la littérature grise a été menée afin d’identifier les programmes de bons pour SSR et les résultats trouvés. Des experts ont été consultés afin de vérifier l’information sur les programmes de bons pour SSR et d’identifier des programmes et études ne se trouvant pas dans la recherche documentaire. Les études ont été examinées pour les résultats concernant le ciblage, les coûts, les connaissances, l’utilisation, la qualité et l’impact sur la santé de la population. Les études retenues ont utilisé l’étude transversale, pré et post intervention et des modèles quasi expérimentaux.
Résultats: 13 programmes de bons pour SSR répondant aux critères établis ont été identifiés. Les programmes de bons pour SSR analysés étaient situés au Bangladesh, au Cambodge, en Chine, au Kenya (2), en Corée, en Inde, en Indonésie, au Nicaragua (3), en Taiwan et en Ouganda. Parmi les programmes de bons pour SSR, 7 ont étéévalués quantitativement dans 15 études. Toutes les évaluations ont rapporté quelques résultats positifs, ce qui indique que les programmes de bons pour SSR ont augmenté l’utilisation des SSR, l’amélioration de la qualité des soins et une meilleure santé de la population.
Conclusions: Le potentiel des programmes de bons pour SSR semble positif, mais des recherches supplémentaires sont nécessaires pour examiner l’efficacité du programme en utilisant de puissants concepts d’étude. En particulier, il est important de voir des preuves plus solides sur le rapport coût-efficacité et les impacts sur la santé de la population, où les résultats peuvent guider les gouvernements les bailleurs de fonds externes.
Objetivos: Identificar donde se han utilizado cupones en los servicios de salud reproductiva (SR), hasta que punto el programa de cupones ha sido evaluado, y si los programas han sido efectivos o no.
Métodos: Se realizó una búsqueda sistemática de literatura de “revisión por pares” y de literatura gris, para identificar programas de SR con cupones y sus evaluaciones. Se consultaron expertos para verificar la información sobre los programas de cupones en SR e identificar estudios no encontrados durante la revisión de la literatura. Se examinaron los estudios en busca de resultados con respecto a los objetivos, costes, conocimiento, utilización, calidad, e impacto sobre la salud de la población. Los estudios incluidos utilizaron diseños croseccionales, antes y después, y cuasi-experimentales.
Resultados: Se identificaron 13 programas de SR con cupones que cumplían con los criterios establecidos. Los programas revisados estaban localizados en Bangladesh, Camboya, China, Kenia (2), Corea, India, Indonesia, Nicaragua (3), Taiwán, y Uganda. Entre los programas de SR con cupones, 7 fueron evaluados de forma cuantitativa en 15 estudios. Todas las evaluaciones reportaban algún hallazgo positivo, indicando que los programas de SR con cupones aumentaron la utilización de los servicios de SR, mejoró la calidad de los cuidados, y mejoró los resultados sanitarios de la población.
Conclusiones: El potencial de los programas de SR con cupones parece ser positivo; sin embargo, son necesarios más estudios bien diseñados para examinar la efectividad de los programas. En particular, es importante ver una mayor evidencia sobre costo efectividad e impacto sobre la salud de la población, donde los hallazgos pueden ser especialmente útiles para guiar a los gobiernos y financiadores externos.
Providing accessible and quality reproductive health services to the poor is critical for countries to make substantial progress towards achieving Millennium Development Goal 5. The increased use of maternity services such as antenatal care, attended deliveries, and post-natal care yield better maternal and infant outcomes (Lawn et al. 2009); increased family planning utilization allows for healthier birth spacing (Yeakey et al. 2009); and timely testing and treatment of sexually transmitted infections (STIs) reduce morbidity and mortality associated with STIs and HIV (Aral et al. 2006).
One strategy for increasing the use of reproductive health services in developing countries is the establishment of voucher programmes where vouchers are distributed for free or highly subsidized reproductive health services and providers are reimbursed for seeing voucher-bearing patients. Much of the literature on reproductive health (RH) voucher programmes has focused on defining the terminology around vouchers and related concepts, describing how voucher programmes work, and detailing the advantages and limitations of voucher programmes.
Typically, voucher programmes are described as a part of consumer-led or demand-side financing, where donor or government funds are used to stimulate demand for services by directly connecting the benefit to the intended beneficiary (Sandiford et al. 2005). Voucher programme monies are linked to outputs rather than inputs and therefore are often referred to as output-based aid programmes (Bhatia & Gorter 2007). These terms refer to monies tied to specific goals, such as the number of services provided, quality indicators, or completion of a specific task (Oxman & Fretheim 2009). With voucher programmes, vouchers are distributed to targeted beneficiaries and subsequently exchanged at contracted providers for RH services. Providers are then reimbursed based on the number of services provided (Sandiford et al. 2005).
There are several advantages to using voucher programmes. First, voucher programmes allow for the targeting of low-income or high-risk individuals in specific geographic areas or according to a means test (Ensor 2004). By focusing on those most in need of services, voucher programmes have the potential to reach individuals who would not otherwise receive reproductive health services.
Second, voucher programmes typically engage the private sector and therefore can introduce greater competition in reproductive health services by increasing supply and improve consumers’ choice (Bhatia & Gorter 2007). Another advantage is that minimum quality standards can be used to accredit facilities and encourage providers who do not qualify to make improvements to become eligible (Sandiford et al. 2005). These improvements can have ‘spillover’ effects improving the quality of care for non-voucher patients too.
Lastly, voucher programmes can facilitate greater transparency through the review of administrative data that track voucher distribution, receipt of services, and performance measures. If information systems are set up properly when designing a voucher programme, providers can be monitored and financially rewarded for providing quality care (Sandiford et al. 2005).
The possibility for fraud at the provider level is one limitation to voucher programmes; however, fraud detection can also be built into a voucher programme, where voucher sales and services are monitored for unexpected spikes or patterns consistent with fraud and patients can be followed up to confirm they received services. Another limitation of voucher programmes is the high overhead costs in the initial stages in order to achieve the advantages detailed above (Bhatia & Gorter 2007).
RH voucher programmes can vary substantially, not only in the services they offer but also in how they distribute vouchers, the conditions under which providers are contracted, and the level of monitoring conducted. At present is that there is a lack of knowledge on the effectiveness of voucher programmes overall and under what conditions they are most appropriate (Bhatia & Gorter 2007; Oxman & Fretheim 2009).
This systematic review examines the evidence on past and present RH voucher programmes. Three questions are addressed: (i) Where and when have vouchers been used for reproductive health services in developing countries? (ii) How have RH programmes been evaluated? (iii) What have we learned regarding RH voucher programmes?
Searching the literature for voucher programmes
A seven-step process was used to conduct the literature search. First, in March 2010, a keyword search of the PubMed (MEDLINE), Popline, and Cochrane Library and a title search of Google Scholar databases were used to identify peer review and grey literature publications on RH voucher programmes. Figure 1 details the search terms used for the review. No timeframe was placed on the searches and reviewed publications were limited to those reported in the English language. Both programme-specific references and more general documents were examined to identify relevant RH voucher programmes. For programme-specific publications, the voucher programme had to be located in a developing country as defined by the Human Development Index at the time the programme was operating (UNDP 2009).
Second, the reference lists of the originally identified publications were reviewed to identify more potential qualifying RH voucher programmes. Subsequently obtained publication reference lists were also examined to identify any further potentially qualifying reproductive health service voucher programme.
Step 3 consisted of searching additional websites of voucher programme funders, administrators, or catalogers of voucher programmes that were identified in the peer review/grey literature searches. These websites comprised worldbank.org, usaid.gov, dfid.gov.uk, http://kfw.de, ps4h.org, rhvouchers.org, gpoba.org, psi.org, and msi.org.
In step 4, from the identified list of programmes, inclusion in the study was based on three criteria: (i) the programme provided reproductive health services including antenatal care, assisted deliveries, other maternity-related services, family planning, treatment and testing for STI, and other ailments related to the reproductive system; (ii) the programme used vouchers to distribute services to the targeted population; and (iii) vouchers were used to administer payments to providers.
Voucher programmes were excluded from the review if: vouchers were used as an incentive to be in a research study or to complete a treatment cycle; conditional cash transfers took place where vouchers did not reimburse providers; vouchers were only for goods, such as bed net vouchers directed at pregnant women and vouchers only for oral contraceptives or condoms; programmes did not use a physical voucher to reach the targeted population; other demand-side financing techniques were in place to provide general health insurance as opposed to specific goods/services related to reproductive health.
Step 5 involved a search for programme-specific information that may not have been obtained in the original database searches. Searches were conducted via google.com using key identifiers such as programme name, location, and type of service.
The 6th step involved extracting programme and evaluation information from the most relevant references. Extracted information includes programme dates, place of service, targeted population, external funders, and voucher distribution and reimbursement (Table 1) and relevant study designs and outcomes (Table 2). In the case of conflicting information (i.e. differently reported programme dates), a hierarchy of information was used with peer reviewed references, followed by governmental reports, non-governmental reports, and other grey literature.
Table 1. Reproductive health care services voucher programmes
Distribution of vouchers
Low income pregnant women
World Bank Sector-Wide Approaches
Distributed by skilled birth attendants at antenatal checks.
Statistically significant increases in qualified attended deliveries, facility deliveries, antenatal care and post-natal care visits compared to controls. 92% of pregnant women in voucher areas had at least one ANC visit compared to 76% in control areas; 58% delivered with qualified provider compared to 27% controls; 44% facility delivery compared to 19% controls; 31% PNC visit compared to 20% controls.
Before and after intervention, with control group. Administrative data analysis
Facility deliveries increased more in treatment areas (vouchers plus other interventions) compared to controls but no reported test of statistical significance. In voucher areas, facility deliveries increased by 28.6% over 2 years compared to non-voucher areas (14.5% and 8.6% increases). Voucher-payment of total facility deliveries increased from 2.4% in 2007 to 7% in 2008.
Nicaragua/Reproductive health care for adolescents
Cross-sectional community sample of voucher receivers and non-voucher receivers post-intervention
Voucher receivers had significantly higher utilization of reproductive health care (OR 3.1). Among the subgroup of sexually active, non-pregnant girls who do not want a baby in the next year, compared to non-voucher receivers, voucher receivers had significantly higher use of condoms at last sexual contact (OR 1.8) but not statistically significant higher rates of modern family planning methods.
Population surveys before and after intervention, with control group
Non-significant increase in utilization of STI services after programme was initiated (15%); however, significant increase (48%) in utilization of STI services for respondents located near contracted facilities (within 10K).
Nicaragua/Reproductive health care for adolescents
Simulated patients before – during – after intervention, no controls
Some aspects of service quality improved from before to after intervention –increase in patients leaving with a contraceptive method/receipt (50% to 80%), increase in patient involvement in decision making (31% to 73%), increase promotion of condoms (53% to 93%). Most measures did not find a statistically significant difference. Several quality measures increased during intervention but dropped after intervention concluded.
Nicaragua/Reproductive health care for adolescents
Cross-sectional survey of randomly sampled voucher users and non-users three to 15 months after voucher distribution
Overall user satisfaction was significantly higher among voucher users compared to non-voucher users (OR 2.2). Clarity of doctors explanations were not statistically significantly different between the two groups
Population surveys before and after intervention, with control group
Significant reduction in syphilis prevalence in year 2 for treatment group compared to controls. A 57% reduction for persons living within 10K of contracted facilities.
Next, in step 7, the information extracted for each programme was sent to a contact person for each programme. In most cases, the contact person was a corresponding author on a publication. Contacts were asked to review and comment on the validity of the extracted information and were asked for any further references to evaluate the programme. Additionally, contacts were given the list of programmes being analysed and asked whether they knew of further RH voucher programmes that would fit the inclusion criteria.
Detailing quantitative evaluation outcomes
In examining the quantitative evaluations of voucher programmes, 6 types of outcomes were reviewed: (i) targeting – the extent that vouchers reached the intended population, (ii) costs – voucher programme costs compared to non-voucher programmes delivering similar services, (iii) knowledge – patient or general population knowledge on issues related to reproductive health, (iv) utilization – change in use of services over time, (v) quality – assessment of provider quality or patient satisfaction, and (vi) health impact – disease measures, mortality, or fertility impact. Evaluation studies were limited to those with an observable contrast of time and/or control group.
Reproductive health voucher programmes
The flow diagram in Figure 2 details the results of the review. 13 RH voucher programmes were identified that fulfilled all our criteria for inclusion, shown in Table 1. The majority of the RH voucher programmes were located in Asia (Bangladesh, Cambodia, China, Korea, India, Indonesia, and Taiwan). Three additional RH voucher programmes were located in Africa (two in Kenya, one in Uganda) and 3 more were in Central America (all in Nicaragua).
Of the 13 RH voucher programmes, 7 were identified via the PubMed search (Bangladesh, Cambodia, Kenya, Korea, and the three Nicaragua programmes). The Popline and Google Scholar searches each identified 1 additional programme (India and Uganda, respectively). Three RH voucher programmes (China, Indonesia and Taiwan) were identified through reviewing general overview documentation on voucher programmes or reviewing references of obtained articles and reports. A second Kenyan voucher programme was identified by a content expert.
The oldest RH voucher programmes were family planning services in Taiwan and Korea which started in 1964 and 1965. Five RH voucher programmes began in the 1990s and six RH voucher programmes began in the 2000s. Most of the voucher programmes offer more than one type of RH service (i.e., maternity services and family planning). Two programmes (Bangladesh and Cambodia) focused exclusively on maternity services, 2 delivered only family planning services (Korea and Taiwan), and 1 programme (Nicaragua) provided screening for cervical cancer.
Most RH voucher programmes (9) contracted with both public and private providers (including for-profit and not-for-profit providers), while 2 contracted only with public providers and 2 contracted only with private providers. External funders of RH voucher programmes included the World Bank and bilateral aid agencies from Belgium, Germany, the Netherlands, the United Kingdom, and the United States.
Quantitative evaluation information was identified in 7 of the 13 RH voucher programmes (Bangladesh, Cambodia, all three Nicaragua programmes, Taiwan, and Uganda). Outcomes addressing targeting, cost-effectiveness, knowledge, utilization, quality, and health impacts are described in Table 2. Quantitative evaluation designs varied, including cross-sectional analyses, before and after intervention analyses with and without control data, simulated patients (individuals acting as patients for the purpose of evaluation) before and after intervention, and a cost analysis of administrative data. Four quantitative studies had strong study designs that would qualify them for inclusion in a Cochrane review (randomized controlled trials, non-randomized cluster controlled trials, controlled before and after studies, interrupted time series) (EPOC 2007).
Two studies examined the success of targeting their services to low-income and high-risk populations. In Taiwan, Chow (1968) found that voucher acceptors had initially higher fertility rates, higher rates of use of abortion services, and higher rates of not wanting additional children compared to non-voucher acceptors. Nicaragua’s cervical cancer screening programme exceeded established screening benchmarks in terms of reaching and treating high-risk women (Howe et al. 2005).
Two studies examined costs. In Nicaragua, the voucher programme for STI care among high-risk groups had higher costs per STI patient treated but lower costs per STI effectively cured compared to costs in public sector facilities (Borghi et al. 2005). In Bangladesh, women delivering in voucher areas reported lower out-of-pocket costs than women in control areas (Hatt et al. 2010).
In Uganda, a quasi-experimental study found that knowledge of the programme among the sampled population increased significantly in the first 16 months of the programme, as did knowledge about STI symptoms in voucher areas (Bellows 2009). In Nicaragua, a cross-sectional study found that 3–15 months after the intervention began, individuals who received a voucher had statistically significantly higher levels of knowledge about modern contraceptives and STIs than those who did not receive a voucher (Meuwissen et al. 2006e).
Four studies reported utilization outcomes. In Bangladesh and Cambodia, facility-based deliveries had a greater increase in voucher areas compared to control areas (Hatt et al. 2010; Ir et al. 2010). Additionally, Bangladesh saw significant increases in antenatal and post-natal care visits compared to controls (Hatt et al. 2010). In Nicaragua, use of contraceptives doubled among sexually active non-pregnant voucher redeemers and voucher receivers had significantly higher utilization rates of reproductive health care and condoms compared to non-voucher receivers (Meuwissen et al. 2006c; e). In Uganda there was a non-significant increase in utilization of STI services after the programme was initiated in the general population; however, a significant increase was found among the poor located within 10 km of contracted health facilities (Bellows 2009).
In Bangladesh, one quality concern was that the increased payment amount for providing a surgical delivery would result in unnecessary surgical deliveries; however, there was no statistical difference in the proportion of surgical deliveries between voucher and control areas (Schmidt et al. 2010). In Nicaragua, user satisfaction was significantly higher among voucher users than controls and some aspects of service quality improved over time (Meuwissen et al. 2006a; b). Additionally, among the contracted facilities, physicians’ knowledge of contraceptives and recognition of STI symptoms increased during the programme, which is expected to yield better quality care (Meuwissen et al. 2006d).
Studies of 3 RH voucher programmes reported population health outcomes. Using matched controls, the Taiwan IUD programme calculated that for the first 3 years the programme prevented 99 live births a year for the first 1,000 IUD insertions (Chang et al. 1969). In Nicaragua, a time series analysis found significant reductions in the prevalence of three STIs (trichomonas, syphilis, and gonorrhea) after each round of voucher distribution, where shorter time periods between treatment rounds resulted in more substantial decreases in prevalence (McKay et al. 2006). In Uganda, a significant reduction in syphilis prevalence was found in treatment areas compared to controls between baseline and follow-up surveys 16 months later, with a 57% reduction in prevalence among persons located within 10 km of a contracted facility (Bellows 2009).
Voucher programmes are increasingly being used to address disparities in reproductive health as donors and national governments are keen to target demand subsidies to the most in need within transparent healthcare delivery systems. Five of the thirteen RH voucher programmes were launched since 2005 and several proposed RH voucher programmes were identified while preparing this analysis, expected to be implemented in the near future. As such, it is important to review the effectiveness of past and present RH voucher programmes.
Seven of the RH voucher programmes were evaluated using quantitative data in 15 studies. Some programmes were evaluated extensively, such as the Nicaragua STI programmes, which had several studies evaluating the programmes, finding mostly positive results (Borghi et al. 2005; McKay et al. 2006; Meuwissen et al. 2006a–e).
All of the examined studies found positive outcomes or non-significant change. The majority of studies; however, were cross-sectional or before-and-after without controls and thus not considered particularly strong study designs. Conducting evaluation research for RH voucher programmes in developing countries is difficult. The provision of RH services can be controversial and sensitive with regard to confidentiality. Furthermore, implementing robust study designs can be particularly complicated in developing countries where standard sampling techniques in developed countries are not feasible. Still, there is an important need to better understand the causal relationships between RH voucher programmes and outcomes. Although they are challenging, cluster controlled trials (randomized and quasi-random), controlled before and after studies, and interrupted time series are viable designs for future evaluations that can be used to better inform policymakers on the effectiveness of voucher programmes.
The most commonly examined outcome was increased utilization of RH services. Three of the five utilization studies were cross-sectional reviews with positive results – voucher users/receivers having higher utilization than non-voucher users/receivers or voucher areas having greater utilization than control areas. While these results indicate potential success, they are contemporaneous comparisons of proportions and it is not possible to say whether the observed differences were due to the voucher programme. Two other studies of utilization had more robust designs. Ir et al. (2010) showed a greater increase in facility deliveries in treatment areas compared to controls between survey periods; however, the study did not report a test of statistical significance. In Uganda, the increase in STI services was not statistically significantly higher in voucher areas than controls, although in subgroup analyses there was a significant increase among poor individuals located within 10 km of contracted facilities (Bellows 2008). Based on these studies, it does appear that RH voucher programmes are associated with an increase in utilization, but there are potential spatial and economic limits to RH voucher programmes, effects on utilization that should be further explored.
Four studies examining quality also found primarily positive or non-significant results. The most robust study looking at quality – an interrupted time series with controls – found that unnecessary surgical deliveries were not occurring in Bangladesh in spite of the greater financial return when providers performed surgical deliveries (Schmidt et al. 2010). Measures of patient/user satisfaction and provider knowledge found evidence that voucher programmes have likely enhanced some aspects of quality in the Nicaragua voucher programmes (Meuwissen et al. 2006a,b,d). More research is needed on how vouchers can improve quality of care for other programmes.
While only 3 studies examined population-level health outcomes, these 3 had relatively strong study designs (before and after with controls and interrupted time series). Because of the stronger designs, the evidence on health impacts shows the most promise on the potential for RH voucher programmes. More studies focusing on health impacts can lend further evidence on the effectiveness of RH voucher programmes.
The findings on targeting, costs, and knowledge were mostly positive; however, with only two studies in each of these categories and none with strong study designs according to the Cochrane group (EPOC 2007) it is not possible to come to a definitive conclusion on the effectiveness of voucher programmes according to these outcomes.
As always, it is necessary to acknowledge the limitations of this analysis. First, this review was limited to voucher programmes for RH services that specifically distributed physical vouchers in order to reach targeted populations and used vouchers for provider reimbursement. Other demand-side financing programmes that operate similarly to voucher programmes were excluded because they did not use a physical voucher. One example of an excluded programme is a maternal services programme in Gujarat India called Chiranjivee Yojana (CY), which shares features with voucher programmes such as contracting with providers and having set prices for services; however, CY does not use a physical voucher, instead relying on a below poverty line (BPL) card that the government uses to identify individuals as low-income and qualify them for a variety of government services (Bhat et al. 2007). Future research on RH voucher programmes should evaluate whether ‘virtual’ or ‘paperless’ voucher-like programmes have the same impact as physical voucher programmes.
Additionally, this analysis did not include programmes or evaluations of vouchers for reproductive health goods such as condoms, oral contraceptives, and insecticide treated bed nets for pregnant women. These types of RH commodity voucher programmes have been implemented in various developing countries (e.g. China, Vietnam, Iran, Malaysia, Phillipines, Costa Rica, Tanzania, Mali and Zambia) (Korten et al. 1973; Iran Ministry of Health 1975; Mushi et al. 2003; Sidibe 2004; Grabowsky et al. 2005; Hammett et al. 2006). The rationale for excluding programmes that distribute vouchers for RH commodities is that their management is markedly different from voucher programmes for RH services. There is less need for extensive provider accreditation and contracting, quality control is focused on manufacturing and distribution, and reimbursement after product sales is concerned more with unit volumes and less with service delivery quality. Future reviews should evaluate whether vouchers for RH goods have been successful in increasing utilization, reducing costs, and improving health status for the targeted populations.
Another limitation is that it is possible that not all information on RH voucher programmes was obtained in this analysis. In spite of a thorough peer review and grey literature search strategy, a number of sources were obtained through expert contacts that were not identified through the search process and were not available online. It is possible that other relevant reports exist but are not readily identifiable. Additionally, it is more likely that reports of no findings or negative findings are those that are difficult to find, either because they are unflattering to programme managers or because they are not seen as interesting to the development community. Either way, there is a likely publication bias that favours a positive outcome for RH voucher programmes.
In spite of these limitations, this study has important policy implications. At present, the potential for RH voucher programmes appears positive; however, more research is needed to identify the conditions under which vouchers function optimally. This review sets up a framework to not only examine past programmes, but to guide future programmes on areas for evaluation. More robust evaluations of RH voucher programmes are needed. In particular, evaluations examining cost-effectiveness and population health impacts can best direct governments and external funders. As new RH voucher programmes move forward, it is vital to find the most effective way to improve reproductive health outcomes and achieve the Millennium Development goals.