Private providers outnumber public providers in nearly all low- and middle-income countries (LMICs), delivering more than 50% of all healthcare services in Africa and Asia and more than 70% of all healthcare in the most populous countries in these regions: Nigeria, India, Pakistan, Bangladesh, and Indonesia [1]. Given private health facilities' large share of national health markets, LMIC national health systems are increasingly challenged to ensure these private providers sufficiently advance public health goals and adhere to national standards of care.

Social franchising is an innovative way of leveraging existing private sector infrastructure in LMICs to serve these national public health goals. To do so, social franchising promotes and improves clinical services among existing private providers through technical assistance and the application of social marketing techniques, a health intervention that has proven successful in fostering widespread access to and use of public health commodities in LMIC countries [2, 3]. As the fastest growing mechanism for engaging private practitioners in national public health initiatives, this social franchising approach is supported by a highly collaborative community of practice and a growing body of evidence that demonstrates social franchising's ability to improve service quality, increase access to essential services, and serve the poor [4, 5].

Social franchises engage existing for-profit, private clinics in a contractual exchange. The franchise program, nearly always run by a non-governmental organization (NGO), builds providers' capacity in new clinical methods, clinic management, business skills, and marketing and advertising techniques. Doing so builds demand for the health services offered. The program also organizes ongoing education and technical support as well as access to subsidized commodities and medicines which are often unavailable. Service quality is further supported through franchise-mediated linkages to private and national referral systems. In exchange, providers who enroll as franchisees commit to providing the health services that the implementing organization prioritizes, often low-margin services such as family planning or antenatal care, or socially unattractive services such as treatment for tuberculosis (TB) or HIV/AIDS. Franchise providers also are required to adhere to clearly defined clinical, reporting, and pricing practices. Increased client load, the resulting increase in profits, training opportunities, and reputation enhancement are the most frequently cited motivations for providers to join a social franchise [6, 7].

Begun in the early 1990s in Pakistan and Nepal, social franchising programs have steadily expanded throughout the developing world; as of June 2012, more than 50 programs operate in 35 countries [8] (see map, Figure 1). Initially, participating clinics and providers offered only family planning services, but many franchise service portfolios have expanded in recent years to include pediatric and maternal care as well as infectious disease testing and treatment, among others. More than half of the world's social franchise programs now offer a range of non-family planning services, and the number of program areas addressed by each franchise is growing each year [8]. While social franchises are often linked to large-scale social marketing programs which promote and distribute non-clinical health commodities (e.g., condoms or mosquito bednets) through retail outlets, the franchises themselves are, by definition, focused on clinical service delivery. Currently, two large international NGOs, Population Services International (PSI) and Marie Stopes International (MSI), operate or support the majority of global social franchise programs, 25 and 10 programs respectively [9, 10].

The growth in the number of programs has led to the development of a global community of practice among social franchisors. Coordinated by the University of California, San Francisco (UCSF)'s Global Health Group, this network of social franchising implementors facilitates information exchanges and supports almost all of the known social franchising programs, compiling data annually about its members in a Compendium of Clinical Social Franchising Programs [8, 11]. The Social Franchising Compendia are comprised of both large and small franchise programs. Large programs are defined as reporting an operating budget of more than $500,000 per year and including more than 50 member clinics. From 2010-2011, the Compendia gathered data on 58 social franchise networks, spanning Asia, Africa, and Latin America.

Increased confidence in this service delivery model on the part of donor agencies and NGOs accounts, in part, for the rapid expansion of social franchising. These stakeholders believe that franchises increase access to important clinical services by extending the geographic reach of the government healthcare system. Related clinical service models, such as NGO delivery or one-time training programs, face challenges in providing widely distributed services. It can be difficult for these models to meet high quality standards at outlets while offering only a low volume of services to a small catchment area and patient base. In contrast, by leveraging existing infrastructure and in situ skilled providers, social franchise networks are theoretically able to assure quality at comparatively low cost, even in low volume settings [2].

As with any health intervention, understanding the health impact of social franchising programs is essential, to inform decision making and to demonstrate the viability and value of this innovative approach. Currently, programs track a number of service delivery measures: the number of patient visits, not tracked by individual patients, for any cause or service (called "patient volume" hereafter); the number of clinics operating (called "number of outlets" hereafter); and (rarely) the number of prevention or treatment interventions received by patients (called "services provided" hereafter). Family planning-focused programs frequently track and report CYPs, a measure of projected impact that can be applied to all family planning methods based on weights to approximate the number of years a method safeguards against pregnancy and the number of products or services provided. For example, 120 condoms are considered to protect the couple from pregnancy for one year, and are thus, equal to one CYP, as are 13 cycles of monthly contraceptive pills, or .26 of a 5-year implant. (Note that the average implant is estimated to provide only 3.8 CYPs, not five, because some 5-year implants are removed early; hence .26 implants leads to one CYP.) [12].

These process measures are widely acknowledged as crude and poorly correlated with the true health impact that these programs likely achieve [13]. Patient volume and number of outlets have inherent problems: the former includes no information on severity of disease, while the latter does not reflect information on the patients served or the types of illnesses and health conditions treated [14]. Although the CYP metric is a good aggregate measure of family planning program projected impact, it is a metric of protection, not health impact, and it is limited to one program area only. As social franchises increasingly expand their portfolio of services beyond family planning, CYPs will capture only a small portion of all care offered.

Program managers and donors currently seek more robust measures of program impact, both to benchmark program performance against appropriate comparators, and to track year-to-year improvement [15–18]. A standardized metric for projected health impact would enable these program comparisons and allow for more accurate assessments of social franchising program variations, after accounting for changes in population health and service delivery. Such a method would also provide a basis for more accurate analysis of the cost-effectiveness of complex delivery systems and programs.

An existing metric, DALYs averted, offers promise as an impact measure of social franchising programs. This metric is based on the DALY, the standard unit representing disease burden in a population, originally developed for the World Bank in 1990 and adopted by the World Health Organization (WHO) in 2000 as part of the Global Burden of Disease Study [19, 20]. Disease burden can be expressed as the number of DALYs lost due to a health condition, either from premature death (mortality) or disability (morbidity), as compared to an ideal life expectancy [20]. The DALYs averted measure, then, denotes the disability-adjusted life years that are not lost - or, are averted - as a result of a health intervention. The attraction of using a metric based on the DALY is that the DALY incorporates both mortality and morbidity, it is widely used by global development agencies, it enables comparison across countries on an standardized scale, and it can be aggregated or disaggregated by disease [21, 22].

The DALYs averted metric is intended conceptually to be the inverse of DALYs in burden and thus benefits from a growing understanding of what a DALY, as a unit, is. Such attributes allow decision makers to use DALYs for comparing the overall burden of disease due to TB, malaria, diabetes, or other illnesses. Conversely, decision makers can also compare projected DALYs averted for understanding the potential to reduce burden with specific interventions, and to allocate resources based on these data.

Despite its established importance in global burden of disease assessment, the DALY has only recently been adapted into a measure of the impact of healthcare interventions by service providers. Beginning in 2007, PSI initiated the use of an impact metric based on the DALY. In each of the 58 countries where PSI operated, researchers calculated a country- and disease-specific DALY coefficient for each prevention or treatment intervention [23]. These models project the impact of specific products and services across a wide range of program areas, such as male circumcision, malaria rapid diagnostic test kits, intrauterine device (IUD) insertions, and pneumonia treatment [24]. Note that PSI bases the DALYs averted metric on models of health impact which are necessarily different from the WHO and World Bank burden models of DALYs. PSI has also never used age weighting in DALY calculations.

PSI DALYs averted models are tailored to different diseases and interventions. As a generalized description, DALYs averted are calculated based on reduced risk in a population and the corresponding years of healthy life preserved. Using country-specific population and health data, the reduced risk of death is calculated by applying the demonstrated effectiveness of prevention or treatment to baseline disease burden. To calculate impact in DALYs averted, the reduction in mortality is multiplied by the number of years between average age at death for the targeted disease and ideal life expectancy. The ratio of years lost from death to years lost from disability is used to calculate the additional years that would be lost to disability for the same disease. The result of this calculation is a DALYs averted coefficient, the impact of a single product or service in years of healthy life preserved. Other publications describe the PSI methodology in more detail [24].

Each of PSI's DALYs averted models projects health impact from the use of one service or product delivered (e.g., a safe child delivery or the sale of a packet of oral rehydration salts (ORS) for pediatric diarrhea), producing a coefficient for this unit. Besides being product- or service-specific, this coefficient is also country-specific, as the number of DALYs averted by any treatment or service varies according to the national burden of disease. For example, in a country with a low malaria burden, the sale of a long-lasting, insecticide-treated net (LLIN) will have a small coefficient whereas this coefficient will be much larger in a country with higher incidence and widespread malaria. Therefore, the DALYs averted coefficient for an LLIN in Nicaragua will differ sharply from that used for an LLIN in Benin. To generate estimates of the number of DALYs averted by a specific service or product intervention, the coefficient is then applied to the total number of services or products distributed within the country during the past year. For example, the LLIN coefficient for Benin is multiplied by the number of LLINs distributed, resulting in the total number of DALYs averted by the LLIN intervention in Benin.

The final model outputs for each intervention, DALYs averted, can serve as a program management tool as these outputs allow comparisons of impact between countries and between different interventions offered within each country program. Organizations can use the DALYs averted metric to benchmark program impact and determine program emphasis and design, as described elsewhere [25]. Thus far, both PSI and MSI have adopted DALYs averted for internal program management use; other organizations are considering doing the same.

For managing social franchising programs, the DALYs averted metric would be useful for projecting program impact, by applying the coefficients to reported service delivery data, on a monthly, quarterly, or annual basis. DALYs averted can be considered for specific interventions within a health program area (e.g., HIV or family planning) for a social franchise network, country, or region. To understand the overall health impact of a social franchise program, the DALYs averted across all individual products or services can be summed. The resulting aggregate will show an estimate of the health impact of family planning services as well as the social franchising program's impact on malaria, diarrhea, and other targeted health conditions. With this aggregate, comparisons of overall effect between social franchising programs managed by different implementing organizations and operating in different national contexts are possible because DALYs averted adjusts for the burden of disease in each country.

Data for this analysis comes from the 2010 and 2011 self-reported programmatic and services data collected by the Social Franchising Compendia for each year [8, 11]. All social franchising programs known to UCSF's Global Health Group at the end of each year are included in the Social Franchising Compendia. To maximize the likelihood that all active social franchises are captured, those programs tracked by the Social Franchising Compendia are compared on a quarterly basis with other lists of social franchises (e.g., those compiled by the Center for Health Market Innovation and the International Centre for Social Franchising), in addition to conducting a thorough search of programs referenced on the internet, and canvassing donors and implementers.

Nearly all Social Franchising Compendia programs voluntarily report service delivery data annually between February and April, using either internet-based survey forms or Adobe Portable Document Format (PDF) documents filled in by hand. Data are checked for empty response cells and for consistency with program descriptions. Programs report the total number of patient visits (patient volume) and the treatments, procedures, and products provided (services provided) for a wide range of conditions and diseases. In addition, programs report a range of aggregate program data, including the number of clinics included in the social franchising network, the number and type of providers per clinic, and the size of annual operating budgets. Additional file 1 provides an overview of the social franchising programs that are Compendia members.

The key output measures emphasized in this study - services provided, patient volume, and number of outlets - were directly reported by the social franchising networks to the Social Franchising Compendium. These annual service delivery data formed the basis for calculating both DALYs averted and CYPs. CYPs were calculated using the 2011 USAID conversion factors specific to each family planning method, called CYP coefficients in this paper [12].

To apply DALYs averted to franchised program services, we adopted the most widely used tools for estimating the number of DALYs averted from service delivery data that we are aware of, the health impact estimation models used by PSI. PSI's DALYs averted models are appealing for this analysis because of their breadth: these models provide an estimate of per-unit impact of 44 different service and product interventions, many of which are offered by social franchisors and used in the LMICs where most social franchising programs are currently located. Therefore, the country- and service-specific DALYs averted coefficients produced by these models can be easily applied to multiple social franchises, regardless of their program affiliation.

To estimate DALYs averted, we multiplied the appropriate PSI DALYs averted country- and service-specific coefficients to the service delivery data reported by the social franchising program of interest. For example, the Bangladesh DALYs averted coefficients were used for determining the number of DALYs averted by each of the services offered by the Blue Star Bangladesh social franchise. Once DALYs averted were calculated for each service reported, we totalled all of them to produce an aggregated estimate of the number of DALYs averted for that social franchising program in either 2010 or 2011.

Among the full dataset reported to the Social Franchising Compendium, certain services could not be paired with coefficients for estimating DALYs averted because of one of the following reasons: PSI did not have a DALYs averted model for the intervention; the PSI model did not include country data for the country where the intervention was reported; or the social franchise program reported insufficient information, prohibiting the pairing of the service with a DALYs averted coefficient. Additional file 2 presents the full set of interventions reported by social franchising networks. Note that most, but not all, of these interventions could be matched to a country- and intervention-specific coefficient to estimate DALYs averted. All of the family planning methods offered by social franchisors could be converted to CYPs.

A total of 58 social franchising networks reported data to the Social Franchising Compendium for the years 2010 and 2011, with 50 programs reporting data in 2010 [11] and 51 programs doing so in 2011 [8]. Spanning three continents and 36 countries, these social franchising networks vary in age and size, with Asian networks being the oldest and largest. The oldest network is Sangini Franchising, launched in Nepal in 1994, while the youngest five networks were launched in 2011 in Laos, Mozambique, Sudan, Nicaragua, and Somaliland. Sub-Saharan Africa has the largest number of social franchising programs whereas Latin America has the fewest.

Fifty-one programs provided sufficient service delivery data to enable the calculation of DALYs averted or CYPs provided for 2010 or 2011. For the statistical analysis, we included only those programs from this sample which provided data for both years (N = 32). Half of these (16 programs) are controlled by or affiliated with PSI. The remaining half are either affiliated with MSI (6 programs), the US-based NGO, DKT International (1 program), or are independent (9 programs), meaning they are unaffiliated with an international NGO but may be linked to a local NGO or private organization within the country. While most of the non-PSI programs deliver healthcare in the same countries where PSI works, social franchising programs are also included from five countries where PSI is not present: Bangladesh, Indonesia, Peru, the Philippines, and Sierra Leone.

This evaluation sought to demonstrate the application of using DALYs averted as a standard health impact metric for social franchising. Our results show it is possible to apply a metric based on the DALY to service delivery data from multiple social franchising programs, whether DALYs averted is a new metric for these programs or not. As calculation of the DALYs averted metric just requires social franchise service delivery data as well as sufficient data on disease burden for a country, the PSI DALYs averted models could be applied to non-PSI social franchise data. In some cases, we could not estimate DALYs averted because we did not have sufficient information reported on a franchise program's interventions, such as the number of services provided or what specific disease was targeted. In other cases, we lacked country-specific data, such as disease prevalence, and also could not estimate DALYs averted. However, most social franchises in the Social Franchising Compendia appear to report sufficient detail in service delivery data, and operate in countries in which disease burden data are available. Therefore, as demonstrated, the conversion of this service delivery data into DALYs averted is feasible for global social franchising programs.

Our findings also support the theoretical basis for adopting DALYs averted as a summary measure of program health impact: this new metric can be applied to project health impact across a range of products and services. While patient volume and number of outlets are important process metrics for tracking program progress, they do not indicate health impact. The CYP metric, while commonly understood as a comprehensive measure of program impact unlike these process measures, is actually based on intervention protection and is not context-specific in terms of pregnancy risk in a population. Moreover, the CYP's measurement is limited to family planning services. As demonstrated in this study's analysis of DALYs averted by service type and program area, many social franchising programs offer a diverse mix of services that cannot be measured by CYPs. Certainly, large numbers of DALYs averted from 2010-11 can be attributed to family planning (likely reflecting the reproductive health origins of many social franchising programs). However, treatment and services for health conditions not incorporated in CYPs - but with high importance to current burden of disease [18] and offered by a growing number of social franchise programs - are important drivers of DALYs averted. DALYs averted, therefore, enables social franchisors to capture a diverse array of health services when estimating health impact.

On account of this new metric's wide applicability, the importance of DALYs averted as a summary measure of health impact may be most notable for social franchising programs in which non-family planning services contribute significantly to overall treatment numbers. For programs where family planning is the focus of all or nearly all of the services provided, a shift to DALYs averted is likely to offer only limited improvement in measurement of impact, resulting from the country context adjustments in the family planning DALYs averted model. However, given the current trend in LMICs to broaden clinic offerings beyond specialized, family planning-only services to more integrated and comprehensive care, the benefit of using DALYs averted as the standard health impact metric for social franchising is likely to grow in the future.

In addition to capturing the impact of clinics offering diversified services, the adoption of DALYs averted offers benefits for social franchisors and their stakeholders in programmatic decision making. When tracking ongoing progress and the overall impact of service delivery, social franchisors can calculate, and then compare, DALYs averted at different levels of the network structure. For example, they can compare impact between individual health services, or sum up the DALYs averted from all services to compare the performance of individual clinics within the network. The aggregate health impact estimate also enables benchmarking of an entire franchise network's performance between programs, countries, and regions, comparisons that are currently only possible in social franchising for family planning impact. As the DALYs averted metric incorporates disease burden - and thus, population need - into its impact estimates, social franchisors can use DALYs averted data to help determine which services to emphasize in a particular catchment area, country, or region. While service delivery data can show demand, the potential number of DALYs averted is a better indicator of health needs, and ultimately, health impact.

The DALYs averted metric can also be interpreted alongside other data to reflect program aims and values. The cost per DALY averted is a common indicator of cost-effectiveness. With cost data from a franchising network or for outlets within a franchise, franchise managers can drive toward increased efficiency. A franchise-specific net cost per DALY averted can help social franchisors and stakeholders better consider the opportunities and trade-offs of specific service delivery interventions, leading to greater optimization of care.

Despite more than a decade of the global health community using DALYs as a measurement of disease burden, the application of this metric for projecting health program impact is still in its nascent stages. For the rapidly emerging social franchise service delivery model, our analysis finds a strong theoretical basis for adopting DALYs averted as its program impact metric. DALYs averted is likely to be the most inclusive and comprehensive measurement available, enabling inter- and intra-program comparisons for tracking program progress and guiding strategic planning and program design. In theory as well as in practice, DALYs averted captures the impact of a wide range of health care services, not limited to the delivery of family planning methods, and thus best represents the breadth of care and treatment offered by a majority of social franchise programs. Our analysis supports this conclusion, even while we found that the application of the DALYs averted remains problematic given current challenges in data reporting and the burden of expanding model development.

The social franchising community of practice offers a useful forum to pilot the DALYs averted metric as a standardized measure for service delivery programs managed by multiple implementing organizations. First, the diversity of health services offered by social franchisors enables opportunities for learning how to best measure program impact in different program areas. Second, the established platform for collaboration and transparent data sharing provides a forum, such as that found in the Social Franchising Compendia, for improving and streamlining data management systems. Certainly, considerable work still needs to be done to assure that service delivery data collected by programs are based on shared parameters and are not subject to the vagaries of interpretation during data entry as appears to have occurred in 2010 and 2011. Motivated by shared interests and the attraction of collaborative work, members of the social franchising community are currently collaborating on developing standards to address these reporting issues. Once this process is accomplished, we expect the social franchising community of practice to be well-positioned to begin using, and eventually standardizing, DALYs averted as its metric for estimating programmatic health impact in all service areas.