Can paying for results help to achieve the Millennium Development Goals? Overview of the effectiveness of results-based financing

Authors


Correspondence
Dr Andy Oxman, Norwegian Knowledge Centre for the Health Services, PO Box 7004, St. Olavs plass, N-0130 Oslo, Norway. Tel: 47-4825-4924 Fax: 47-2325-5030 Email:oxman@online.no

Abstract

Objective Results-based financing and pay-for-performance refer to the transfer of money or material goods conditional on taking a measurable action or achieving a predetermined performance target. Results-based financing is widely advocated for achieving health goals, including the Millennium Development Goals.

Methods We undertook an overview of systematic reviews of the effectiveness of RBF. We searched the Cochrane Library, EMBASE, and MEDLINE (up to August 2007). We also searched for related articles in PubMed, checked the reference lists of retrieved articles, and contacted key informants. We included reviews with a methods section that addressed the effects of any results-based financing in the health sector targeted at patients, providers, organizations, or governments. We summarized the characteristics and findings of each review using a structured format.

Results We found 12 systematic reviews that met our inclusion criteria. Based on the findings of these reviews, financial incentives targeting recipients of health care and individual healthcare professionals are effective in the short run for simple and distinct, well-defined behavioral goals. There is less evidence that financial incentives can sustain long-term changes. Conditional cash transfers to poor and disadvantaged groups in Latin America are effective at increasing the uptake of some preventive services. There is otherwise very limited evidence of the effects of results-based financing in low- or middle-income countries. Results-based financing can have undesirable effects, including motivating unintended behaviors, distortions (ignoring important tasks that are not rewarded with incentives), gaming (improving or cheating on reporting rather than improving performance), widening the resource gap between rich and poor, and dependency on financial incentives.

Conclusion There is limited evidence of the effectiveness of results-based financing and almost no evidence of the cost-effectiveness of results-based financing. Based on the available evidence and likely mechanisms through which financial incentives work, they are more likely to influence discrete individual behaviors in the short run and less likely to create sustained changes.

inline image

Introduction

Results-based financing (RBF) is used by the Global Alliance for Vaccines and Immunisation (GAVI) (1), the Global Fund (2), the World Bank (3), UK Department for International Development (DFID), USAID, and other donor agencies (4). It is advocated by the Global Campaign for the Health, Millennium Development Goals (MDGs), and other development aid groups (4–6) to motivate patients (7–9), and improve the performance of healthcare providers (10–12), organizations, and governments (13).

RBF is also referred to as pay-for-performance, performance-based funding, and output-based aid. All of these terms refer to the transfer of money or material goods conditional on taking a measurable action or achieving a predetermined performance target (13). In this overview, we have synthesized evidence of the effects of RBF in healthcare from systematic reviews with the aim of informing decisions about when to use RBF and how to design, implement, and evaluate RBF schemes in low- or middle-income countries (LMIC). Although our focus is on LMIC, we have included systematic reviews on the effects of RBF, regardless of whether included studies were conducted in LMIC.

Methods

We undertook an overview of systematic reviews. We included any review with a methods section that addressed the effects of any type of RBF. We searched for relevant systematic reviews on RBF in the Cochrane Effective Practice and Organization of Care (EPOC) register of systematic reviews, a database of over 1000 systematic reviews of the effects of health systems interventions. These were identified through electronic searches (up to August 2007) of MEDLINE, the Cochrane Database of Systematic Reviews (CDSR), the Database of Abstracts of Reviews of Effectiveness (DARE) and EMBASE (up to October 2006). We searched for the following terms in the title or abstract: cash, conditional, contract*, finance*, pay*, performance-based, results-based, subsid* (where * is a wildcard). In addition, we reviewed the list of EPOC reviews in progress, contacted key informants (see Acknowledgments), and checked the reference lists of articles that were retrieved. Finally, we searched for related articles in PubMed restricted to systematic reviews (using systematic [sb]) for references (7–10, 14–20).

One of us (ADO) screened citations identified using the above methods. For each included systematic review we summarized the characteristics of the review and the main findings. We excluded reviews of the effects of user fees, payment methods to remunerate health workers (fee-for-service, capitation, salary, and mixed systems), private sector strategies, broad overviews of interventions to improve professional practice or patient adherence, and reviews that focused on financial incentives for specific behaviors, such as smoking cessation (Appendix A includes references for excluded reviews). Quality assessment was done informally, taking into consideration widely used criteria (21). No systematic reviews were excluded based on our judgments of quality.

Results

We identified 12 systematic reviews that met our inclusion criteria (Table 1) (7–9, 12, 16–20, 22–26). The reviews had overlapping scopes and some studies were included in more than one review. Most of the reviews had only minor methodological limitations, but the searches for two reviews were completed over 10 years ago, and, for another three, more than five years ago (Table 1). Four of the reviews were restricted to English language publications (8, 12, 16, 20).

Table 1.  Systematic reviews of results-based financing (RBF)
InterventionsParticipants and settingsTargeted behaviorsStudy designs
What was searched forWhat was foundWhat was searched forWhat was foundWhat was searched forWhat was found
  1. CABG, coronary artery bypass graft; LMIC, low- or middle-income countries; RBF, results-based financing.

RBF targeted at recipients of care
Conditional cash transfers (7,26) (Most recent search, April 2006)
Direct monetary transfers made to households conditioned on a particular behavior or action (eg visit to a health facility for regular checkups)Cash provided that children attended school and appointments for preventive health care with or without incentives for mothers to attend education courses and prenatal care (five studies). Financial incentives for collecting HIV test results (one study)Populations who would potentially access health services in low and middle-income countriesPoor and disadvantaged groups in Latin America (Mexico, Brazil, Nicaragua, Colombia, Honduras), mostly infants and children, and pregnant and lactating women (five studies), people tested for HIV in Malawi (one study)Health care use or access to health care (also household health expenditure, or health or anthropometric outcomes)Attendance for preventive care, school and health education (five studies), collecting HIV test results (one study)Randomized trials (4), quasi-randomized trial (1), controlled before–after study (1)
Effects of economic incentives on preventive behavior (8) (Articles published between 1966 and 2002 included)
Economic incentives including cash, gifts, lotteries, and other free or reduced-price goods and services for the benefit of the specific consumerThe following incentives were offered: 15 coupons for free or reduced-price goods or non-medical services, 11 cash incentives, 10 lotteries, 10 incentives involving negative reinforcement or the opportunity to avoid punishment (eg losing access to services or benefits), 7 gifts, and 6 free or reduced-price medical services. Several studies included additional intervention components, particularly social pressures, which potentially confound the impact of the incentiveConsumers who were healthy or physically at risk but not yet labeled with a diagnosis in clinical and non-clinical settingsSixteen of 24 studies of simple preventive care (eg immunization, cancer screening) included vulnerable populations of low socioeconomic status. Nineteen of 23 studies of more complex health promotion (eg lifestyle change) included generally healthy, middle-class populations, recruited from work sites or the general population. The studies were conducted in the US (41), Australia (1), Denmark (1), New Zealand (2), and UK (2)Adoption of preventive health behaviors. Mental health, substance abuse, health protection concerns such as injury prevention, and oral health were excludedSeventy-eight per cent of incentives targeted a specific behavior (eg going for a preventive service). The remainder required the participant to attain a particular outcomeRandomized trials (39), quasi-experimental (8)
Financial incentives to enhance patient compliance (9) (Most recent search, April 1997)
Financial incentives (money, cash, or vouchers) versus ‘free’ treatment (no financial charge to the patient)The incentives were mostly small ($5–10), but ranged from $5 to nearly $1000 (for a treatment program for cocaine dependency). Lotteries were used in three studies and vouchers or gifts in fourAny patients and settingThe participants were low-income or disadvantaged populations in six studies. All 11 studies were conducted in the USA between 1976 and 1996Compliance with medication, medical advice, or medical appointmentsAttendance + completion – anti-TB treatment (2 studies), attendance – dental care for children (2), postpartum attendance (2), child immunization (1), attendance – children with behavioral difficulties (1), cocaine free (1), antihypertensive treatment attendance + target BP (1), weight loss (1)Randomized trials (11)
RBF targeted at individual or groups of health care professionals
Effects of pay for performance on quality of care (12) (Most recent search, November 2005)
Explicit financial incentives designed to improve health care qualityPhysician level financial incentives (6 studies), provider group-level financial incentives (9), payment system level financial incentives (2)Any participants in any settingNot reportedQuantitative measure of health care qualityProcess of care (mostly preventive services, 13 studies), access to care (3 studies)Randomized trials (9), controlled before–after studies (4), cross-sectional studies (4)
Quality-based purchasing (22) (Most recent search, December 2003)
Payment or reputational strategies aimed at providers that individual employers, employer coalitions, or government programs could plausibly adoptFinancial incentives (fees or bonuses) (8 studies), reputational incentives (1 study) from a fee of $0.80 per vaccination to a bonus of $10,000 per clinic per yearNot specifiedIndividual providers who received incentives (4 studies), provider groups or an individual provider (4 studies), all in the USAQuality (as measured by processes or outcomes) or costPreventive care (7 studies), chronic care (1 study)Randomized trials (9)
Quality-based payment systems (23) (Most recent search, late 2003)
Explicit financial awards for good performancePerformance feedback and financial bonuses (3 studies) and targeted financial incentives (3 studies) from $0.80 per vaccination to a maximum bonus of $43,750 for medical groups (with between 16 and 500 physicians)Not specifiedPractices or individual physicians serving Medicaid populations (3 studies), Medicare populations (1 study), and outpatient clinics (2 studies) in the USANot specifiedImmunization delivery (3 studies), smoking cessation (2 studies), cancer screening (1 study)Randomized trials (4), non-randomized trial (1), before–after study (1)
Economic incentives for delivery of preventive care (20) (Articles published between 1966 and 2002)
Explicit economic incentives for preventive care targeted at specific individual providers, including direct payments or bonuses to the provider or his/her group. Multi-component interventions were excludedBonuses for reaching a target (5 studies), per input bonuses for immunization (2). Potential payments ranged from $50 to a bonus of $4682, where such data were reportedAny participants in any settingVulnerable populations (Medicaid enrollees) (6 of 8 findings). All incentives were aimed at primary care physicians in the USA (6 studies)Primary or secondary preventive care or health promotionImmunizations (4), cancer screening (2), assorted preventive services (1)Randomized trials (6)
Target payments in primary care (17,18) (Most recent search, October 1997)
Target payments, in which a lump sum payment is made if, and only if, the a predetermined quantity or target level of care is reachedAn additional 10% ($0.80) or 20% payment to the standard fee of $8 for each influenza immunization for each influenza immunization over the target rate of 70% and 85% respectively (1 study), a lump sum payment if > 70% or 90% of childhood immunizationsPhysicians in primary careGeneral practitioners in the US (1 study), Scotland (1 study)AnyImmunization ratesRandomized trial (1), interrupted time series analysis (1)
Effects of financial incentives for prescribers (19) (Most recent search, October 2005)
Policies that intend to affect prescribing by means of financial incentives for prescribersGeneral practice fund holding (10 studies), drug budgets for physicians in private practice (2), indicative drug target savings scheme (1)Health care consumers and providers within a large jurisdiction or system of care in any settingGeneral practitioners in the UK (10 studies), Germany (2), and Ireland (1)PrescribingPrescribingInterrupted time series analyses (6), controlled before–after studies (10)
Effects of pay for performance and public reporting on racial disparities (16) (Most recent search, March 2006)
Programs that explicitly link rewards and sanctions to performance on measures of specific healthcare processes and/or outcomes. Incentives could be either monetary (eg bonus payments or higher per member per month reimbursements) or reputation based (eg public report cards)Publicly reporting risk-adjusted CABG mortality rates (1 study)Any participants in the USAHospitals providing and patients eligible for CABG in New York stateAnyCABG ratesControlled before–after study (1)
Effects of financial incentives for service in rural and underserved areas (24) (Articles published between 1966 and 2002)
Any form of financial support in exchange for restrictions on practice location. Studies not applicable to the Canadian health system were excludedFive return-of-service programs, four multidimensional programs, one no intervention (a survey of career intentions)Physicians in any countryPhysicians in the US (6 studies), Canada (3), and New Zaland (1)Practice in rural and underserved areasRecruitment (3), retention (2)Prospective cohort (1), retrospective cohort (1), cross-sectional surveys (5), descriptive studies (3)
RBF targeted at organizations
Contracting between government and non-state providers (25,26) (Most recent search, April 2006)
A formal contractual relationship between the government and a non-state provider had to have been definedContracts with private not-for-profit providers. Only one of the contracts (in Cambodia) appears to have used RBF (13)Populations that would potentially access health services in LMICPrimary health services (2), maternal health services (1) in Cambodia, Pakistan, and BoliviaHealth services access and utilisationImmunization coverage (1 study), household health expenditure (1), bed occupancy rates (1), number of deliveries (1), number of primary care visits (1)Randomized trial (1), interrupted time series analysis (1), controlled before–after study (1)

RBF targeted at recipients of health care

One systematic review addressed the effects of conditional cash transfers (CCTs) on the uptake of health interventions in LMIC. CCT programs offer money to households conditioned on a particular behavior or action, such as attending appointments for preventive care. Overall, randomized trials from Latin America show that CCT programs are effective in increasing the use of preventive services for children and women, and sometimes improving health status (7). The observed increase (risk difference) in the proportion of children visiting health centres ranged from 11% to 33% (across studies and outcome measures). The impact appears to vary, with disadvantaged populations having greater benefits. The impact on immunization coverage (four programs) was less robust and appears smaller (from 0 to 7%). Positive findings were reported (two programs) for the impact on mothers’ reports of health outcomes of their children (reported ill, diarrhea, or respiratory disease), whereas the impact on objectively measured health outcomes (anemia or hemoglobin values) was mixed (three programs).

Only one study in Malawi evaluated the effect of different amounts (from $1 to $3). The overall effect was a near doubling in the proportion of people returning for their HIV test results (72% of people who received incentives compared to 39% without the incentive). A positive association was found between the size of the incentive and the effect: a 9% absolute increase per extra dollar.

The flows of money required for CCT programs may be significant with an average cost of between $60 and $560 per family, and the actual transfer budget accounting for only 4–28% of the total budget of a CCT program. The cost–effectiveness of CCT programs compared with classic supply-side interventions (improving quantity and quality of infrastructure and services) has not been examined, as most CCT programs have so far been implemented in settings with relatively adequate (health) infrastructures. Unanticipated perverse effects can occur. For instance with one program, the fertility rate increased unexpectedly after implementation of CCT, possibly because only pregnant women were eligible for the subsidy.

A systematic review of economic incentives (including reduced price goods and services, lotteries, cash incentives, and gifts) found 39 randomized and eight quasi-randomized trials (8). Overall, preventive behaviors were increased 73% of the time. All of the simple preventive care studies used a discrete, readily measurable outcome. Complex preventive care studies used physical measures, as well as self-reporting in some instances. For simple behaviors, the proportion of studies with positive findings ranged from 40% (for lotteries and gifts) to 100% (for cash and punishment); for complex behaviors, it ranged from 50% (for cash and free medical services) to 100% (for gifts). Incentives in the form of rewards for participating in and adhering to goals, whether for simple or complex prevention, were generally effective inducements for behavior change.

Most studies matched a short-term incentive with a short-term behavioral change or outcome. The technique of rewarding the achievement of specific outcomes was reserved for more complex preventive behaviors, such as weight loss. Behavioral changes were generally not sustained past the short-term. While many of the studies that rewarded specific outcomes showed positive effects in the short run, of the four studies that checked for long-term results, all of the significantly improved measures had returned to their original levels.

Cash incentives had the expected rank ordering: the higher the cash incentive, the higher the response to the incentive. Coupons, more convenient and flexible, were preferred to gifts. Both studies that pitted a coupon incentive against a gift incentive found the coupon more effective. Whereas coupon incentives were effective, with 12 of 15 incentives showing positive results, only four of seven gift incentives had positive results, and two of the positive results were potentially confounded by additional lottery or competition intervention components.

In five of seven cost-effectiveness analyses reported, an intervention that consisted of a similar intervention without the economic incentive was reported to be a more cost-effective approach. For example, in one study it was estimated that the cost per prevented influenza-related death was $3,990 for those who received an invitation letter reminding the patient of the upcoming flu season, versus $17,860 for those who received the letter plus free flu shots. No study attempted to estimate the cost-effectiveness of the economic incentive for impacts on population morbidity or mortality over time.

A systematic review of financial incentives (money, cash, or vouchers) to improve patient compliance versus ‘free’ treatment (no financial charge to the patient) found 11 randomized trials (9). Improvements in compliance ranged from −1% (for compliance with clinic appointments by parents of children with behavioral difficulties offered a lottery for $10 vouchers for toys, meals, or bus tokens) to 37% (for compliance with appointments for prevention by mainly immigrants with tuberculosis offered a mixture of cash, tokens, and vouchers worth $5 to $10 per appointment). The median improvement in compliance was 17% (risk difference). For 5 of 13 comparisons there was an improvement of less than 10% and the results for 11 of the 13 comparisons were not statistically significant (a P value of 0.05 was used as a cut-off to determine statistical significance).

No evidence of cost-effectiveness ratios was provided, but the review authors note that “financial incentives are likely to be cost effective if substantial treatment benefits accrue not only to the patient but to society at large – in economic parlance, if there are positive externalities to treatment. Treating or preventing tuberculosis is an example of this. If patients comply badly with treatment this not only leads to more expensive treatment for the individual patient but increases the possibility of the development of drug resistant strains of the disease and the infection of other people.”

RBF targeted at individuals or groups of health professionals

Three systematic reviews with similar inclusion criteria addressed the effects of paying for performance based on quality, and reached similar conclusions (12, 22, 23). The most recent of these reviews found nine randomized trials, four controlled before–after studies and four cross-sectional studies (12). Five of six studies found partial or positive effects of incentives directed at individual physicians. Nine studies evaluated the use of financial incentives directed at provider groups. Of these, seven found partial or positive effects of financial incentives on measures of quality. Most of the effect sizes were small. In two studies the improvement in the measure of quality of care was statistically significant. In the five other studies there was a partial effect. For example, one found a small improvement in rates of cervical cancer screening between the intervention and comparison groups after the quality incentive program (difference = 4%, P = 0.02). Improvements in mammography screening rates and hemoglobin A1c testing were not statistically significant. In two randomized trials, the group-level incentives for preventive health services were ineffective.

Two studies evaluated financial incentives provided at the payment system level. One evaluated the effectiveness of an incentive to improve access to healthcare for nursing home patients with debilitating, acute, and chronic conditions. The program included incentives to admit severely dependent patients, incentives for attainment of health status goals, and an incentive to discharge clinically appropriate patients. The intervention sites admitted statistically significantly more severely ill patients than nursing homes in the control group. The other (‘negative’) study of incentives at the payment system level was on performance-based contracting for substance abuse. One potentially important finding was an unintended effect of ‘adverse selection’. There was a significant decrease in the likelihood of the most severely ill group receiving treatment from providers that received financial incentives for achieving predetermined quality measures.

Several other studies identified the potential to ‘game the system’. For example, “there was an incentive for nursing homes to claim that they were admitting extremely disabled patients who then ‘miraculously’ recovered over a short period” (12). Another study examined the effect of bonus payments for both identifying smokers and for providing tobacco cessation advice. The incentive was associated with an increased documentation of tobacco use status, but not in the provision of advice to quit smoking.

A systematic review of economic incentives to improve the delivery of preventive services found six randomized trials of incentives targeted at primary care physicians providing care to vulnerable populations (Medicaid enrollees) in the USA (20). The trial reports were not clear on whether the financial incentives were paid to the physician or the practice, and if the payment was made to the practice, how the practice financial incentives were transmitted to individual physicians. Only one of the eight results found that increasing financial incentives translated into a statistically significant increase in the provision of preventive care. This study used fee-for-service (FFS) payments to physicians for providing immunizations. The remaining studies were roughly evenly split between using bonuses and increased FFS payments. One study found that most of the increase in measured immunization rates due to the financial incentives was a consequence of better documentation, not the result of physicians providing more immunizations (20).

Several studies examined the impact of formal physician performance feedback without economic incentives. For example, in one study, the ‘feedback only’ cohort increased their mammography screening referrals, but their mean behavior was not significantly different from the ‘feedback with a token bonus’ ($50). In another study, the ‘feedback only’ group was also not significantly different from either the ‘feedback plus financial incentive’ group or the control group.

Since most interventions were assessed as not being effective, cost-effectiveness analyses were not undertaken. In the one study with a positive finding, revenue increased by an average of $82 for physicians in the incentive group. That amount of incentive translated into an increase in immunization rates of 7%, which corresponds to a cost of $3 per additional influenza immunization. Influenza vaccines have been shown to save $117 in direct medical expenditures in the elderly. Thus, in the one case where economic incentives were shown to be effective, they were also cost saving.

A systematic review of the effects of target payments, in which a lump sum payment is made if, and only if, a predetermined quantity or target level of care is reached, found only one randomized trial and one interrupted time-series analysis (17, 18). The use of target payments in the remuneration of primary care physicians was associated with improvements in immunization rates. However, the increase was small for the overall influenza vaccination rate (risk difference = 7%, P = 0.03) in the randomized trial, and the authors found no evidence that the overall linear trend for childhood immunization rates changed as a result of the introduction of target payments in the second study. A systematic review of the effects of financial incentives on prescribing found only very low-quality evidence of the effects of fund holding or drug budgets in the UK, Germany, and Ireland (19). Another review of the impact of pay for performance on racial disparities in the USA did not find any studies (16). Finally, a systematic review of the effects of any form of financial support in exchange for restrictions on practice location found only very low-quality evidence from North America, suggesting that return-of-service programs may be successful with respect to short-term recruitment, but may not be successful with respect to long-term retention (22).

RBF targeted at organizations

A systematic review of the effects of formal contracts between government and non-state providers on health services access and use in LMIC found one randomized trial, one interrupted time-series analysis, and one controlled before–after study (25, 26). All three studies had methodological limitations. These studies suggest that contracting out services to non-state providers can increase access and use of health services.

The three evaluations did not present evidence on whether this approach was more effective than making a similar investment in the public sector, as there was not an exact control available in any of the settings. In addition, the introduction of non-state providers into some settings and not others also brings many potentially confounding variables such as the presence of additional management expertise or expatriate doctors, which may improve drug supply or increase use. In one study in Cambodia, performance was measured, and a resulting poor performance could result in sanctions and no renewal of the contract. Performance targets were identified for maternal and child health. Otherwise, none of the three contracting schemes appear to have used performance-based payments targeted at organizations. Although some positive impacts were reported for contracting, none of these effects can be attributed to RBF directly.

Discussion

The primary strength of this report is that it systematically and transparently summarizes a breadth of evidence of the effects of RBF from systematic reviews. The primary weakness of the report is that we did not undertake a systematic review ourselves. By capitalizing on available systematic reviews we were able to prepare this report quickly and to synthesize the findings of 12 systematic reviews published between 1997 and 2008 that are directly or indirectly relevant to the use of RBF in LMIC. However, only one review focused specifically on the use of one type of RBF (CCT) in LMIC. One other review that focused on contracting out in LMIC did not focus directly on RBF. The evidence summarized in the other reviews came largely from high income countries; the reviews addressed overlapping questions, and not all of the reviews were up-to-date. Nonetheless, the primary conclusion of this overview is robust in light of the sparse evidence of the effects of RBF cited in any of the reviews and background documents, including another recent overview (27), or by key informants, as described in a companion article (28). There are few rigorous studies of RBF, and overall, the evidence of its effects is weak.

Does RBF work?

Conditional cash transfer (CCT) programs have been found to be effective in increasing the uptake of some preventive services that were already free. This indicates that properly designed incentives may address barriers to access. Their implementation in a context where services are not free remains to be tested. Further, even with important incentives, some programs did not succeed in improving vaccination coverage, perhaps because they were implemented where rates were already high.

Despite the success of CCT programs, several questions remain regarding their feasibility in lower income settings. Indeed, they involve relatively complex mechanisms for targeting and logistics for the delivery of transfers. Additionally, they require good co-ordination with service providers in health and education for the tasks of monitoring and supervision. The success of CCT depends on the existence of effective primary health services and local infrastructures. With more complex programs, it also depends on effective systems for identifying and making payments to low-income families.

More generally, there is some evidence from randomized trials of positive effects of financial incentives on patient compliance and preventive health behaviors. However, most of those trials have been carried out in the USA; the results may not translate directly to other countries with different socioeconomic and cultural contexts. Well-designed randomized trials are needed, particularly in low-income settings.

Although financial incentives are considered to be an important element of strategies to change professional practice, there are relatively few well-designed studies and overall the evidence is weak. Most of the literature is descriptive rather than evaluative. There may also be a publication bias because those responsible for RBF schemes may have some disincentive to publish negative or ambiguous findings. A small number of more rigorous evaluations have examined relatively simple preventive interventions, such as the impact on rates of immunizations and screening, as opposed to more complex interventions. The success of a financial incentive is likely to be inversely related to the complexity of the tasks it seeks to motivate. Few rigorous evaluations have evaluated more complex interventions and outcomes, such as contracting.

Is the evidence relevant to LMIC?

It might be anticipated that RBF is potentially more effective in LMIC than in high-income countries, particularly for patients and community health workers, since small financial incentives may represent a larger proportion of their income and there may be fewer competing economic incentives. In addition, although the impacts of RBF schemes on long-term development goals, including building institutional capacity and human capital, are uncertain, it could be argued that short-term outcomes, such as improved access to and use of health services, represent development goals.

On the other hand, there are a number of uncertainties about the use of RBF in LMIC. These include challenges in sufficiently specifying contracts or arrangements for RBF and monitoring and measuring the attainment of targets. The more remote the point of service delivery, or the more complex the service to be delivered, the more likely it appears that contracts or agreements will be governed by informal means. Government capacity to manage RBF may also be challenging. The broader the services, the harder it will be to precisely define targets and negotiate contracts or arrangements. Feasibility of adequately monitoring service delivery in remote areas is also a key implementation issue.

There are also a number of questions regarding the long term desirability of RBF, and particularly contracting out, as an option in LMIC. While contracting out appears effective as a means to scale up service delivery in small areas rapidly, there are potential constraints that these schemes face in the longer term. It is unclear, for example, whether the capacity exists among non-state providers to scale up their service delivery efforts. It is also not clear whether incentives will hold their effects over the long term, or whether undesirable effects, such as sophisticated gaming, will emerge over time. Finally, a focus on contracting may encourage donors to bypass failing or fragile states, thereby overlooking the important role of helping to build the institutional capacity of the Ministry of Health as either a steward or a service delivery organization.

Does RBF have undesirable effects?

There is a danger that unanticipated undesired effects may occur with all types of financial incentives. Undesirable effects that have been identified include:

  • • Unintended behaviors

For example, with CCT some mothers kept their child malnourished in order to retain eligibility; CCT may have increased fertility by 2% to 4%, because pregnant women only were eligible for a subsidy; and an unexpected small negative impact of CCT on children's weight gain may be explained by a misinterpretation of eligibility rules. Beneficiaries may have mistakenly thought that having at least one malnourished child was necessary for continued membership of the program (7).

  • • Distortions

Financial incentives may cause recipients to ignore other important tasks.

  • • Gaming

Financial incentives can result in gaming (changes in reporting rather than desired changes in practice).

  • • Corruption

Financial incentives may be stolen or misused if not adequately managed.

  • • Cherry-picking

Performance incentives for providers can influence whether healthcare is accessible to patients by altering how willing healthcare workers or organizations are to care for sicker patients, more disadvantaged populations, or more difficult patients. For example, with RBF schemes that are based on scoring systems that are sensitive to small changes, eliminating a small number of ‘difficult’ patients with greater co-morbidities, more disability, or lower health literacy may improve a provider's score dramatically. RBF schemes may cause providers to cherry-pick patients, either by selecting those who may help them score well or by avoiding those who may cause them to score poorly. While there is more than one way to protect against this behavior, programs that ‘risk-adjust’ or stratify quality scores on the basis of health or socioeconomic status may reduce the incentives for healthcare organizations to cherry-pick patients. On the other hand, programs that adjust for risk can also perpetuate disparities if they ‘excuse’ providers from reaching equitable standards of care for disadvantaged populations. RBF schemes can also be targeted specifically at disadvantaged populations.

  • • Widening the resource gap between rich and poor

Performance incentives for providers may widen the resource gap that exists between organizations that serve disadvantaged patients and those that do not, and between countries. RBF schemes that inadequately level the playing field may reward organizations for meeting standards that are much less attainable with disadvantaged populations. This problem can be mitigated by RBF schemes that reward improvement, in addition to absolute achievement, as well as RBF schemes that are specifically targeted at disadvantaged populations.

  • • Dependency on financial incentives

Relying on incentives may foster dependency on them. If provider behaviors are not ingrained, they may disappear when the incentives end or new incentives are introduced.

  • • Demoralization

Financial incentives can result in feelings of injustice and demoralization. For example, this occurs if short-term professionals receive more financial incentives than those who have established long-term practices, or if there are perceptions of favoritism.

  • • Bureaucracy

RBF schemes may have substantial administrative costs associated with monitoring performance and managing disbursement of the financial incentives.

  • • Dilution of professionals’ intrinsic motivation

It is generally accepted that professionals are motivated by the satisfaction of doing their jobs well (intrinsic motivation). Indeed, it is doubtful whether some valued dimensions of quality that are difficult to measure (such as empathy or listening in the medical encounter) would be provided at all if physicians were solely interested in income. Professionals and others have both non-monetary and monetary incentives, all of which affect effort or behavior. It is possible that financial incentives may dilute professionals’ or patients’ intrinsic motivation.

What are the ethical implications of RBF?

  • • Nonmaleficence (do no harm)

To reduce the risk of doing harm, RBF schemes should be designed to minimize perverse effects, and potential unintended effects should be monitored. RBF schemes that threaten to withhold public benefits as an incentive to induce desired behaviors, such as parents taking their children to be immunized, may also create financial penalties for people already at high risk for economic deprivation. Even simply requiring people to return more frequently to collect benefits imposes additional costs. Incentives may further fracture care for vulnerable populations. For example, mothers with low socioeconomic status may not wish to take their children to special clinics for free immunizations, because they prefer to see their regular providers for shots. The ethical imperative to ‘do no harm’ requires that people be treated with dignity, respect, and fair procedural justice.

  • • Justice

If non-compliance is associated with low income, financial incentives for recipients of health care might also improve equity. All things being equal, financial incentives would be expected to have a greater effect among low-income populations. On the other hand, financial incentives may have negative consequences and may affect other aspects of behavior.

The success of CCT programs for improving nutritional and health status among beneficiary children contributed to reduce the burden of disease. However, a proportion of eligible households, such as those belonging to disadvantaged groups, did not enroll (7).

Due consideration must be given to the ethical implications of incentives that can change the behavior of low-income or vulnerable groups. In particular, the ethics of some programs using monetary incentives to promote irreversible contraceptive methods have been criticized (7).

Performance incentives for healthcare providers have the potential to narrow, widen, or maintain disparities in healthcare. The nature of their effect is likely to depend on how well programs can promote inclusive approaches to diverse populations, and how able they are to avoid creating cherry-picking opportunities and greater gaps in resources between ‘rich’ and ‘poor’ organizations. While there are many ways to address inequities, performance incentive programs that promote disparity reduction or help identify disadvantaged groups in need of more tailored programs are more likely to reduce disparities at the same time as they improve overall quality.

  • • Autonomy

The use of RBF may create an environment that does not promote personal responsibility and autonomy, particularly when it is targeted at recipients of care.

Is RBF worth it?

There are few rigorous evaluations of the effects of RBF schemes and even fewer cost-effectiveness studies. Indeed, cost-effectiveness studies are of limited value when the effects of an intervention are not known. There is also limited information available regarding the administrative costs of RBF schemes.

Is RBF sustainable?

The financial sustainability of RBF schemes beyond pilot programs in low-income settings is uncertain. The flow of money that they require is often large, which leads back to the question of whether they are worth it compared to other interventions.

When should RBF be used?

For RBF to be cost-effective it must be targeted at cost-effective interventions or behaviors for which there is low compliance. If there is low compliance, it is important to ask why. This will help to identify the best type of intervention to improve compliance or uptake, and to determine whether financial incentives are likely to be helpful. Incentives intended to cause a temporary change in behavior may be more effective than as an inducement to make a permanent change. Patients and health professionals are often under stress, with many competing demands for their time. Incentives may gain a temporary priority, but sustained change may require an investment of energy to address the underlying mechanisms that can reinforce the desired behaviors in a more permanent way. In many cases, simple changes may be needed, such as prompting actions or delegating authority to support staff. In those cases, resources used for incentives could be put to more efficient use elsewhere.

If they are large enough, financial incentives targeted at individual patients, providers, or small groups can motivate discrete behaviors in the short run. The benefits of such incentives may be magnified if they are co-ordinated with each other and with system-level incentives, although this potential synergy remains largely untested. When provider incentives do work, they may not provide a sustained behavior change. There is then a danger that they will be displaced by a new set of incentives targeted at a new problem.

The mechanisms through which financial incentives given to governments or organizations can improve performance are less clear, and the evidence is sparse. Investing in system changes that support long-term improvements in practice may be a better choice than relying on incentives alone.

Conflicts of interest

ADO is an editor for the Cochrane Effective Practice and Organisation of Care (EPOC) Group, which has published four of the systematic reviews included in this report, and is a co-author of one of those reviews. AF heads the unit at the Norwegian Knowledge Centre for the Health Services which hosts a Norwegian satellite of the EPOC group.

Appendix

Appendix A Excluded systematic reviews

User fees

Austvoll-Dahlgren A, Aaserud M, Vist G, Ramsay C, Oxman AD, Sturm H, Kosters JP, Vernby, Å. Pharmaceutical policies: effects of cap and co-payment policies on rational drug use. (Cochrane Review). Cochrane Database of Systematic Reviews 2008, Issue 1.

Simkhada B, van Teijlingen ER, Porter M, Simkhada P. Factors affecting the utilization of antenatal care in developing countries: systematic review of the literature. Journal of Advanced Nursing 2008; 61: 244–60.

Lexchin J, Grootendorst P. Effects of prescription drug user fees on drug and health services use and on health status in vulnerable populations: a systematic review of the evidence. International Journal of Health Services 2004; 34: 101–22.

Lagarde M, Palmer N. The impact of user fees on access to health services in low and middle income countries. Cochrane Database of Systematic Reviews 2008, in press.

Payment systems to remunerate health workers

Gosden T, Forland F, Kristiansen IS, Sutton M, Leese B, Giuffrida A, Sergison M, Pedersen L. Capitation, salary, fee-for-service and mixed systems of payment: effects on the behavior of primary care physicians. Cochrane Database of Systematic Reviews 2000, Issue 3.

Examples of reviews of private sector strategies

Patouillard E, Goodman CA, Hanson KG, Mills AJ. Can working with the private for-profit sector improve utilization of quality health services by the poor? A systematic review of the literature. International Journal for Equity in Health 2007; 6: 17.

Peters DH, Mirchandani GG, Hansen PM. Strategies for engaging the private sector in sexual and reproductive health: how effective are they? Health Policy and Planning 2004; Suppl 1:i5–i21.

Waters H, Hatt L, Peters D. Working with the private sector for child health. Health Policy and Planning 2003; 18: 127–37.

Examples of broad overviews of interventions to improve professional or patient adherence

Bosch-Capblanch X, Abba K, Prictor M, Garner P. Contracts between patients and healthcare practitioners for improving patients’ adherence to treatment, prevention and health promotion activities. Cochrane Database of Systematic Reviews 2007, Issue 2.

Eight of 30 included trials (all from high income countries) used economic incentives (five used deposits by the patients that were totally or partially reimbursed and three used tokens or goods, such as cash credits). The included studies were mostly small and all failed to meet three or more of 10 quality criteria. The authors concluded that there was not enough evidence to recommend the widespread introduction of patient contracts into health services.

Chaix-Couturier, C, Durand-Zaleski, I, Jolly, D, Durieux, P. Effects of financial incentives on medical practice: results from a systematic review of the literature and methodological issues. International Journal for Quality in Health Care 2000; 12: 133–42.

Eighty-nine included studies of which eight reported results from randomized trials and the others did not describe actual studies. A wide range of financial incentives for healthcare professionals were identified, including different remuneration systems, drug formularies, payments for referrals, shareholding in diagnostic facilities, kickbacks for preferential drug prescriptions or use of supplies, ceilings for annual patient revenue, sanctions or bonuses, and the risk of medical malpractice suits.

Grimshaw JM, Thomas RE, MacLennan G, Fraser C, Ramsay C, Vale L et al. (2004) Effectiveness and efficiency of guideline dissemination and implementation strategies. Health Technology Assessment 8(6).

The review did not find any evaluations of the use of financial incentives to implement guidelines that met the inclusion criteria (randomized trials, interrupted time-series analyses, or controlled before–after studies).

Examples of focused reviews of financial incentives for specific behaviors

Achat H, McIntyre P, Burgress M. Health care incentives in immunization. Australian and New Zealand Journal of Public Health 1999; 23: 285–88.

Hey K, Perera R. Competitions and incentives for smoking cessation. Cochrane Database of Systematic Reviews 2005, Issue 2.

Lussier JP, Heil SH, Mongeon JA, Badger GJ, Higgins ST. A meta-analysis of voucher-based reinforcement therapy for substance use disorders. Addiction 2006; 101: 192–203.

Paul-Ebhohimhen V, Avenell A. Systematic review of the use of financial incentives in treatments for obesity and overweight. Obesity Reviews 2007; 9: 355–67.

Insufficient information

Armour BS, Pitts MM, Maclean R, Cangialose C, Kishel M, Imai H, Etchason J. The effect of explicit financial incentives on physician behavior. Archives of Internal Medicine 2001; 161: 1261–66.

Methods not described in the published article and unable to obtain description from the authors.

Acknowledgements

This report was commissioned by the Norwegian Agency for Development Cooperation (Norad). ADO and AF are employed by the Norwegian Knowledge Centre for the Health Services. The views expressed herein are those of the authors and do not necessarily represent the views of the project's funder or the Norwegian Knowledge Centre for the Health Services.

The authors thank the following individuals who provided background information for this report or commented on an earlier version: Jennie Barugh, DFID; Amie Batson, World Bank; Abdallah Bchir, GAVI Alliance; Sara Bennett, Alliance for Health Policy and Systems Research; Logan Brenzel, World Bank; Tessa Tan-Torres Edejer, WHO; Rena Eichler, CGD Working Group on Performance-Based Incentives; Timothy Evans, WHO; Matt Gordon, DFID; Daniel Low-Beer, Global Fund; Ingvar Olsen, Norad; Jean Perrot, WHO; Don de Savigny, Swiss Tropical Institute; Susan Stout, World Bank.

Ancillary