Background

 Generally speaking, provider payment can be used as a tool of health reforms to ‘align’ the incentives facing providers with the overall health system objectives (efficiency, quality, etc.). Providers (doctors or other service providers) are generally perceived to be unusually important in helping patients make decisions about treatment patterns. Consequently, the kinds of payment incentives faced by providers will help determine health system utilization and overall resource consumption (Eggleston and Hsieh, 2004). And, since incentives can be formed by payment policy, the specific objectives of reform (improved efficiency, reduction of unnecessary hospitalization, promotion of better patient outcomes, etc.) can be promoted through the tailoring of the incentives. Often, the government implements incentive payment schemes for its own facilities by changing the rules for budgeting or compensation in order to create incentives. But, sometimes the provider payment reforms are narrower, such as when they are part of a ‘private provider contracting scheme’, whereby the government arranges for some type of care to be delivered by private providers, or when a private insurer establishes arrangements for reimbursement of private providers. Another narrow application is when incentives are established to supplement salaries or budgets in the form of bonuses that may or may not be paid based upon pre-set performance goals. ‘Pay for performance’ payment policies (P4P) are often schemes based on bonuses, designed to supplement salaries or budget allocations.

Provider payment reforms are generally any form of “contingent compensation”, as an alternative to salaries (for health professionals) or line item budgeting (for institutions). Sometimes the contingencies are prospective, such as when a patient is admitted for a procedure, and the payment will be set at X. The ‘contingencies’ can also be set after a procedure in some P4P programs, and if a desirable result is achieved, then the bonus will be Y. The contingencies in payment rules tell the provider what will increase and decrease payment. Hence, incentives are designed to induce providers to do more of those things for which there is higher payment, and to do fewer things leading to lower payment.

Unlike traditional forms of ‘line item budgets’ and ‘salaries,’ all payment arrangements that involve a fee schedule or some other payment schedule (like capitation), there will be explicit efficiency incentives (downside risks), and also explicit possibilities for capturing increases in revenues (upside risks). Upside potential exists because the payment schedule pays a fixed amount per procedure, per day or per visit. Whatever the ‘unit of payment’ is, the provider can earn more revenue by doing more of it. In this sense, incentive payment approaches emulate competitive market mechanisms by ‘paying more to providers who do more’. If consumers have free choice of providers, then this incentive on volumes may also create incentives to upgrade ‘service quality’ as providers actually compete for patients in order to expand service volumes. As has been the experience in many countries (Langenbrunner and Wiley, 2002) these volume incentives of fee schedules can be so strong as to promote rapid growth in health expenditures, leading to further reforms to control volumes through budgeting the service package.

Prospective fee schedules may also create provider incentives to reduce costs in ways that prove detrimental to patient care. This consequence of incentives always poses a risk, and may result from deliberate choices, or be the result of poor management within the facility. In all cases, payers need to monitor providers for the possible risk of under service in incentive systems. This type of vigilance often requires data and analytical skills unavailable in poor countries, making it difficult to consider provider payment reforms. This issue is reconsidered in the last section of the paper.

The scope of the incentives to change provider practice patterns is related to the scope of the services included in the fixed payment amount: the larger the bundle of services paid by the payment rate, the more the potential for economies and the greater the tendencies to under serve the patients. To understand the scope of incentives we refer to the risk factors whose variation can contribute to overall cost of health care:

• Efficiency: resource cost per unit of service. Contributing factors include prices

paid for inputs, as well as number and mix of inputs per unit of service.

• Intensity: ancillary service utilization including testing rates, special service

utilization (ICU, CCU). Length of stay.

• Case-mix: care needs of the different kinds of patients usually associated with

age, diagnosis, and level of function.

• Volume: according to the unit of payment and the number of patients, visits, and

days.

• Referrals: cost associated with sending the patient to another provider.
• Epidemiological risks: factors that affect disease occurrence rates in a population

like random factors as well as epidemics, flu outbreaks, etc.

Salaries and line item budgets do not give any authority over resource allocation to the provider. Here there are still strong incentives: for replacing work with leisure, for supplementing income with moonlighting or informal fees, and for under-investing in management skills. Fee schedule reforms, on the other hand, cause the provider and the payer to share the (constant sum of) risk. The more services that are bundled into the single unit of payment, the more risk is imposed on the provider, and the less is borne by the payer. Under a per diem approach for hospitals, for example, the provider is at risk for all matters that affect the cost per diem, and the payer is at risk for all factors that influence the number of days of care. Under a capitation arrangement, providers who are paid a fixed per person rate are essentially at full risk. The trend in policy regarding provider payment policy appears to be to shift more and more risk to providers, thereby encouraging stronger efficiency incentives (Quinn, 2003). Table 1.2 shows the risk sharing agreements between payer and provider for various payment approaches.

 

Table 1.2: Provider Payment Methods and Risk Sharing by Payer and Provider

Payment Policy	Contracted Provider at risk for:	Payer at risk for:
Hospitals
Line item budget	Efficiency, intensity, casemix,	Volumes, epidemiological
Fee schedule Per diem   Per case (DRG)	Efficiency, some aspects of intensity   Efficiency, intensity	Some aspects of intensity, casemix, volume and referrals, Casemix, volume and referrals, epidemiological
Global Budget (with volume and casemix      ‘ adjustments subject to a limit)	Efficiency, intensity, and the portion of casemix and volume changes above the limit	Casemix and volume up to a limit. and referral care (to other organizations), epidemiological
Physicians
Salary	Nothing	Efficiency, intensity, referrals, volume, epidemiological
Fee Schedule (FFS)	Efficiency	Intensity, case mix, referrals, volume, epidemiological
Bundled Fee Schedule	Efficiency, some aspects of intensity	Some aspects of intensity, case mix, referrals, volume, epidemiological
Fund-holding By primary physician/clinic     Full Capitation— payment per person per year	Efficiency, intensity, casemix, some aspects of referrals, volumes per enrollee   Efficiency, intensity, casemix, volume, referrals, epidemiological	Some aspects of referrals, epidemiological     Nothing

The strength of incentives (to do things that will generate more profit for the provider) also depends, in part, on the dynamic aspect of the payment formula: e.g. how does the rate (or budget) set in one year relate to the rate or budget in the subsequent year? If the new rate is set based on how much profit was made by the provider, then the incentives will be weak. If the new rate is blind to the amount of profit made by the provider in the prior year, then the incentives will be sharper. To illustrate, consider the case of the provider that, in the prior year, successfully reduces costs and earns a nice profit. If the rate formula sets the new rate lower, in consideration of the now lower costs, then the provider (and others) will be discouraged from becoming more efficient, since the next year’s rate will punish them for becoming more efficient. The alternative dynamic policy is to set the new rate based on the prior year’s rate, blind to how efficient the provider has (or has not) become. This sort of policy will continue to encourage efficiency by allowing the provider to keep any surpluses they earn.

The volume and practice pattern incentives facing providers can be especially strong when providers are “inducing” consumers to follow their expert advice about needed services. This self-interested provider behavior may be stronger for some services than others. (Robinson, 2005). This possibility is important for policy since care seeking (demand) can be managed, in part, from the supply-side of the marketplace. (Ellis and McGuire, 1993). Using payment incentives, physicians can be put at full or partial financial risk for the resource consumption of patients (e.g. the practice pattern). These payment policies will influence higher or lower utilization of services depending on the nature of the incentives. This supply-side policy can complement or substitute for demand-side interventions aimed at influencing access and utilization. For example, it is theoretically possible to control the excessive utilization stemming from moral hazard that accompanies health insurance by supply side cost sharing by physicians rather than imposing large user fees (which can have adverse equity consequences) (Ellis and McGuire, 1993). It is also possible to stimulate access for merit goods (vaccinations, other preventative services) by paying providers for these services in a way that has strong service volume incentives, such as fee for service. Mixed provider payment schemes (capitation for primary care, augmented by FFS incentives for important merit services) are frequently used in the U.K. (LeGrande, 1999) and in other places such as Romania (Vladescu and Radulescu, 2002).

Provider payment technologies to stimulate particular incentives were largely developed in the United States and Western Europe in the 1970s and 80s. In the U.S., the Medicare program initiated payment reforms for hospitals in order to contain program spending by enlisting new provider incentives to promote efficiency. Specifically, the idea was to shift some of the risks of untoward events (high prices paid for inputs, high ancillary testing rates, unnecessary length of stay) to the provider, motivating them to develop ways to control those risks. The adoption of the DRG (diagnosis related groups) payment methodology for compensating hospitals for inpatient care paid a fixed price per admission, and caused hospitals to reduce length of stay, eliminate unnecessary tests and reduce other aspects of intensity of care. Prior to this change in 1983, the insurer (Medicare) was at full risk for the consequences of unexpected changes in admission rates, efficiency, intensity, and length of stay. After the changed payment policy, hospitals were at risk for these aspects of cost. The effect of the new fee schedules were dramatic, with striking reductions in the rate of growth of hospital spending, in the share of health spending in hospitals, in length of stay, and in hospital days per capita (Wouters, et al, 1998). There was no evidence that these effects lead to reductions in quality of care or patient outcomes, as political opponents had warned would accompany incentives for cost containing behavior (Coulam and Gaumer, 1991). In the next two decades, Medicare implemented incentive payment policies (fee schedules) for nursing homes, doctors, and other providers.

Other western countries have adopted many of these payment and casemix technologies that were developed in the U.S., often using DRG casemix measures to improve the fairness of their global budgeting schemes. Among OECD countries, where governments own or purchase much of the care, many modern payment schemes for physicians have been developed and refined. Simple, capitation schemes for paying doctors have been replaced by more complex combinations of payment approaches. The incentives for under-serving patients associated with capitation have been modified to create essentially separate payment schemes for particular services in order to optimize the effects of the volume incentives. In the U.K., for example, the capitation arrangements for physicians are not applied to preventative and family planning services. For these services, a fee-for-service payment scheme is used to create incentives for promoting more preventative and family planning services (Liu and O’Dougherty, 2005). In Germany and Canada, where there are explicit limits on the volume of care provided by physicians (e.g. a cap on payments), these limits are waived for prevention services (Davis, 1998). In Japan, FFS prices are deliberately raised for preventative services (Campbell and Ikegami, 1998).

Provider payment reforms have been popular components of health system reform in LMICs. While some of this demand is attributed to donor push, the popularity of such reforms certainly reflects broad belief that payment incentives can be a powerful tool to promote efficiency, alter longstanding mal-distribution of health system resources (too many resources flowing to hospitals, not enough resources flowing to prevention and primary care), to promote better practice patterns, and other objectives. Patterns of payment reforms generally follow a progression, starting with the replacement of line item budgets and salaries with various forms of fee schedules. And, as strong incentives for ‘increasing volumes of care’ have followed, there have been further reforms to stem these volume problems by putting the provider at risk for volumes of care (some global budget designs and capitation Cuimas and Vaidean 2008).

 The purpose here is to do a review of the studies that attempt to estimate the impact of payment reforms. The vast majority of the literature is for hospitals and doctors. Our approach will be to organize the literature along the following lines:

Hospitals and other Institutions

Global budgets

Fee Schedules

Quality/Performance Schemes

 

Physician Services

Fee Schedules

Quality/Performance Schemes

Fund-holding/Capitation

In the review that follows we focus on the impacts of provider payment reforms. Deliberately broad, we are trying to understand how wide-ranging the policy intentions have been for provider payment reform, as well as how effective reforms have been in changing provider behavior, particularly in LMICs. We are also interested in evidence bearing on the under service or quality risks posed by these reforms.

There are a number of challenges in this literature. The results of almost all studies are not generalizable elsewhere because the choice of intervention in almost all studies stemmed from a policy selection process, not a research selection process (policies were selected when conditions suggest to decision-makers that they would work). The problem of generalizability is a particular concern when reporting on possible side-effects of incentive payment programs. These include under-serving patients, discharging patients too early, failing to admit the most difficult cases, cutting corners on quality, and more. Many of the metrics for evaluating side effects like ‘quality of care’, ‘access’ and ‘cream skimming’ require more sophisticated data than are routinely available in these countries. By and large the literature rarely reports such findings, focusing instead on metrics like volumes of care and expenditures for care. We would be more concerned about these side effects in developing countries than in the West (where there has been little such evidence) because (1) monitoring of side effects is likely less intense in LMICs, (2) ‘slack’ or inefficiency or excess capacity at the onset of the program may not be as prevalent as in the West (there may not be as much provider cushion to help buffer financial problems) and (3) financial incentives may be quite strong in LMICs since the provider may have only one payer.

The validity of impacts of provider payment reforms is also questionable in many instances because these interventions are often confounded by other reforms including management interventions, decentralization and other policies that grant more operating and financial autonomy to providers. Many of the studies of provider payment, particularly in the area of physician capitation and P4P, tend to couple programs of “contracting” with private providers or NGOs with approaches to incentive payment. Measured impacts cannot be attributed to reforms of provider payment or to contracting (to promote competition between providers, and to replace “management” mechanisms with controls via contract terms). And, in the case of physicians, sometimes government salaried physicians are allowed or even encouraged to supplement income by private sector moonlighting (where FFS rates can be charged patients) (Bir and Eggleston 2003). As a result of combining payment reforms with other closely related changes in provider autonomy, more exposure to competitive pressures, privatization, contracting, dual practice arrangements and other factors, it is not possible to know whether results stem from payment policies or other policy changes.

A similar confounding issue arises about coterminous changes in demand incentives, which can confound the apparent influence of provider payment changes. The use of vouchers in India for free obstetric care, for example, created such strong demand incentives as to overwhelm the effects of capitation used to pay the providers (Bhat, 2006). Attempts to intervene and improve quality of services in Cameroon (Livack and Bodart, 1993) and Nigeria (Akin, 1995) confounded (or neutralized) the impacts of user fees on demand volume, where higher user fees were found to be associated with increases in the volumes of services demanded. A UNICEF report argued that “the trend toward decreased demand for services can be reversed when efforts are made to improve the services before a system of payment is introduced”(UNICEF,1990).

 

Evidence from the Literature

The literature on the impacts of payment reforms in LMICs is large, but methodologically relatively weak. It reflects global interest in payment reforms following successes in the U.S. and other OECD countries in the 1980s and 90s. A general assessment of findings is as follows:

• Impacts of payment reforms in LMICs on efficiency and other outcomes is rarely

studied, and when reported, tends mainly to be subjective assessments of trends

(and uncontrolled pre-post methods).

• There is significant global interest in using provider payment technologies in both

organized health systems and in contracting activities and these are trends over time favoring use of capitation methods for primary care, global budgets for hospitals, and P4P bonuses (for physicians and clinics) for achieving target service volumes.

• The reported impacts of provider payment reforms tend to be weaker and less consistent than we would expect from the experience in developed countries.
• Authors attribute weak and inconsistent pattern of impacts to critical barriers (constraints) to their being an effective policy instrument in LMICs.

 

Hospitals and other Institutions

 

Fee Schedules. Per visit, per diem and per case payment systems are used to pay for hospital services in many countries. These fee schedules are generally set by payers, and providers are paid fixed amounts with the promise of retaining any surplus that may result from economies they make. These payment policies offer substantial incentives to increase volumes of care (in order to increase provider revenues) and contain incentives to under-serve patients as managers try to economize the costs which they bear in full.

 

Some of the earliest lessons in structuring incentives for hospitals occurred in the former Soviet Union. During the 1990s, when privatization (of hospitals, health insurance and other sectors) was rampant, much was learned about using fee-for-service payment for hospitals. In the Czech Republic, for example, hospitals were billing several dozen private insurers using about 5000 codes for services. Between 1993 and 1998 volumes of care rose due to the incentives and per capita billings grew by nearly double, and about a third of the insurers went bankrupt (Langenbrunner and Wiley, 2002).

 

A number of early per diem payment schemes were also used in Europe, Indonesia, Brazil and China (Langenbrunner and Liu, 2005). Significant evidence exists about the volume incentives under such systems, and they have generally given way to per case, global budgeting, or capitation schemes. In Brazil, for example, admissions tripled over a decade under per diem incentives (Rodrigues, 1989). In Germany when per diem payment was used, the lengths of stay increased (Schulenburg, 1992). Various per diem approaches have been used in Croatia, Slovak Republic, Estonia, Slovenia, and Latvia with evidence of length of stay growth and subsequent use of caps on inpatient spending, and migration to per case payment methods (Langenbrunner and Wiley, 2002).

 

The experience in the U.S. with per case (DRG) payment system implementation is by far the most studied of the fee schedule approaches to hospital payment. In spite of concerns about possible increases in admission rates, deteriorating quality, skimming and dumping patients, and sicker-but-quicker discharges, the U.S. implementation of DRGs showed reductions in the rate of increase in hospital spending, shortened stays, downsizing of the hospital industry and no real evidence of quality problems (Coulam and Gaumer, 1991). To be sure, there were many anecdotally reported instances of inept hospital managers who overreacted to incentives, and others who did not know how to manage under the changed circumstances of payment. Many hospitals closed because days of care were reduced, and undersized and inefficient facilities could not survive. In other hospitals, many ineffective executives and managers were replaced. But, in general, the implementation situation was characterized by very effective scrutiny of impacts and aggressive processes in hospitals to find solutions and change the behavior of physicians and other health workers.

 

In Eastern Europe and Central Asia there has been widespread adoption of new incentive payment systems (capped per diem systems, casemix systems, or global budgets) in nearly all of the 26 countries[1]. (Langenbrunner et al 2005). These new per case systems were largely implemented to stem the rapid increases in volumes (and expenditure) that were created by per diem systems following the market reforms in the late 1980s and early 1990s. To be sure, the per case reforms generated extensive descriptive evidence of large reductions in length of stay in the region. But, now, the rising numbers of admissions is again testing the appetites for growing volumes in these systems (Langenbrunner et al, 2005). Poland’s increase in admissions of 30% in one year after implementation of per case hospital payment is an example (Orosz, 2001) and there are also allegations of cream skimming. Structural changes, such as bed elimination and reorganization have not been seen when examined (in Kyrgyzstan for example, O’Dougherty, 1999) though have been seen in other places as in Hungary (Orosz and Hollo, 2001). All of this evidence is descriptive, and provides only weak support for the mixed impacts of per case payment systems in Eastern Europe and Central Asia.

 

Yip and Eggleston (2001) report on the per case payment system in China. They used a quasi experimental design to study impacts of per case payment reforms, finding some slowing of expenditure growth per admission, and slower growth of spending on expensive drugs and high technology when compared to the prior FFS system. There were reports that some hospitals were fined due to problems related to under-service, but there was no real evidence of patterns of adverse quality side effects (Eggleston and Hsieh 2004).

 

Not all hospital experience with administrative fee schedules is for inpatient services. There are reported applications of inclusive rates being used for day surgery (Lebanon) and for outpatient visits (China) (Langenbrunner and Liu, 2005). No evaluation evidence is available.

 

In general, the evidence is sparse on the performance of per diem and per case payment systems in hospitals. Almost nothing exists regarding impacts in LMICs, though there is some evidence of mixed findings of the incentives such as strong cost containment incentives, adverse volume effects, and some reports of poor quality, as noted above.

 

Global budgets. Global budgets set limits on spending, usually with ex post adjustments for volumes and casemix in order to not discourage admissions, particularly for costly treatments. This payment method has been defined as “an overall spending target or limit that constrains the price and quality of services provided” (Dredge, 2004). Global budgeting for hospitals is very popular in European and other countries with national health systems including Canada, UK., France, Australia, Italy, Spain, Ireland, Portugal, the Nordic countries, and the U.S. Veteran’s health care system[2]

 

Though varied in design, global budgeting is generally regarded as the strongest approach to encouraging hospital efficiency and for “capping” the overall rate of increase in hospital spending (Liu, 2003). Sometimes, as in most NHS applications, the global budget is a fixed budget allocation to the hospital, which is adjusted ex post based on casemix and overall volume of admissions. Other approaches (often used for non governmental hospitals) set a cap on the budget that a hospital may earn in revenue (possibly with an ex post adjustment also), and the hospital has to set fees and contract amounts in order to live within this cap. In either case, the incentives promote efficiency, tend to slow the adoption of new technologies, and encourage high capacity utilization. Some western studies have documented these cost containment successes in terms of overall health spending, spending per case and per diem, and shortened waiting times (Wolfe, 1993 and Duckett, 1995, and Redmon and Yakoboski, 1995), although Wolfe (1993) notes that there are only modest amounts of real evidence of effectiveness.

 

Global budgets for hospitals are popular in various countries in Eastern Europe and Central Asia. To stem the trends in rising volumes of care there is strong interest in global budget systems (or explicit global caps on fee schedule payments) in Albania, Croatia, Czech Republic, Georgia, Romania and Russia (with about seven other countries working on development of such systems).Evaluation efforts are limited for these systems, though there is fairly strong but descriptive evidence that by using such payment systems countries are able to shrink the fraction of health spending being spent on hospitals (Langenbrunner and Wiley, 2002). In pilot districts in Russia, early, very positive evaluation results were reported including fewer admissions, lower unit costs, less specialty referrals, and bed and staff reductions (Langenbrunner and Wiley, 2002).

 

There is no real literature on the implementation of Global Budgets in other LMIC countries in other regions, though there is mention of global budget schemes in Brazil and Chile (though the latter seems to be a historical budget process, with different incentives).

 

Evidence of counter-veiling side effects on quality is rare. Documented increases in discretionary admissions and deterioration in other quality indicators occurred in Taiwan (Chen et al, 2007), where the authors summarize the findings as “cost containment comes at the expense of health care quality” (Chang and Hung, 2008).

 

Beyond hospitals, global budgets have been used in countries with national health systems to control national or regional spending levels for physician services (Germany, Canada, U.K., Netherlands) and even for national limits on pharmaceutical spending (Belgium). Even broader applications of spending limits for health plans and geographic regions are extensions of the philosophy of global budgeting (Bishop and Wallack,1994). In many countries the use of capitation-type formulae are used to create fair allocations of budgets for health services across districts or regions. For example, in Zambia (districts) and South Africa (provinces) such formulae were used to change inequitable allocation policies (Gibson, 2000). But, these uses of budgeting reallocation are not truly provider payment policies.

 

Pay for Performance Schemes for Hospitals. Pay for performance schemes (P4P) in hospitals are probably too varied in design to attempt to generalize about their effectiveness, even if there was a substantial literature about their impacts. Almost nothing seems to be known about the impacts of schemes to change the behavior of hospitals using financial incentives in an ad hoc manner. Each application provides some financial incentive for achieving a goal (e.g. a quid pro quo).

There is very limited evidence of middle income country applications of hospital bonusing schemes. Two are known in Latin America, namely Costa Rica and Nicaragua. The Costa Rican quality payment scheme was implemented by the Social Security Institute with public hospitals in 2000 which rewarded hospitals for compliance with best practices in care (preventing hospital acquired infections, etc.). In Nicaragua, the Ministry of Health recently put in place a similar concept to create incentives for meeting performance targets in six pilot hospitals. No evaluations have been reported thus far.

 

The major and recent literature review of this, and all other published evidence on hospital P4P schemes concludes that literature on the effectiveness of this rapidly growing form of payment for meeting explicit quality targets is very small, drawn almost exclusively from the west, and is also very inconclusive about the effectiveness of P4P in achieving intended impacts (Christianson et al 2007).

 

Physicians and Primary Care

Fee Schedules (FFS and bundled per visit). There have been few efforts to implement physician fee schedules as matter of policy in LMICs, and we are not aware of any attempt to bundle services into a more aggregated fee schedule. Salaries are still popular for paying physicians, since most physicians in these countries tend to work for the government or for a hospital. While many of these countries are experiencing growth in the private physician marketplace, private insurance to pay for private sector medical services is uncommon. Payments made for ambulatory care delivered by private physicians tend to be dominated by fee for service (prices set by the physician) and paid in full out-of-pocket.

Generally, the incentives of prospective fee schedules would be to increase efficiency, and to cause volumes to increase as well. Only several studies are reported about the impacts of FFS impacts in LMICs. In what has become a notorious example, following independence the Czech Republic used FFS to pay doctors. An overall 40% increase in real health care spending in a two year period was observed following the introduction of FFS. In a related comparison in, private doctors (paid off of a FFS schedule) billed significantly more in every category of service than their salaried counterparts (Ciumas and Vaidean, 2008). After experimenting with various alternative payment schemes they have moved to capitation because of an inability to control the strong volume increases of fee schedules. Brazil implemented the Unified Health System in 1985, which included a set of national fees for physician payment. These rates have seldom been updated since, resulting in falling utilization for primary care (because of failure of physicians to accept these rates for patient care), and poor quality, especially for maternal and prenatal care (Wouters, et al 1998). Rwanda implemented a pilot fixed rate fee schedule program in 19 primary health centers in order to try to increase utilization. The intervention was successful, and they reported large increases in all volume indicators (Eichler, 2006).

Clearly, physicians in these LMICs are responding to the volume incentives of fee schedule payment in predictable ways, increasing the volume of care in order to augment income.

There have been successful attempts in the OECD countries to mobilize fee schedules for particular services which were underutilized (preventative services). In the NHS in U.K., for example, certain services are now paid separately outside the capitation rate (prenatal care, PAP smears, immunization, etc.) (Liu and O’Dougherty, 2005). In Japan, selected services and primary care is encouraged by using high fees relative to the fee schedules for other curative services (Campbell and Ikegami, 1998)

Fund-holding and Capitation. Capitation and fund-holding for primary care providers became popular following the Alma Ata resolution (which emphasized more and better primary care) and the poor performance of fee schedule incentives (which tended to be exploited by physicians to augment their own incomes). There were also earlier experiences with capitation and fund-holding in the U.K., including findings of reductions in hospitalization (le Grand 1999, Klein 1998) which captured the attention of many LMICs, since the hospital sector is usually perceived as overbuilt and over-funded. Because of concerns about volume incentives of fee schedules capitation has become a very popular form of payment for primary care services. When this ‘capitation fee’ includes all or a portion of the expected costs of referrals and hospitalization, then we would refer to the payment scheme as ‘fund-holding’ by the primary care physician. If a capitation (per patient per year) fee is set for primary care services only, then the incentives are strong for the primary physician to limit access, under serve, and over-refer. With some ‘fund-holding’ (and a commensurate required payment by the primary physician when a referral/hospitalization is made) then primary physicians have additional incentives to control expensive referrals and hospitalizations.

Many LMICs are now taking steps to implement some form of capitation for primary care services. Some have implemented capitation approaches to paying for primary care by contracting with individuals or groups of providers and other have used capitation to pay district authorities (as part of a decentralization scheme). These payment approaches have been increasingly relied upon in places as diverse as Egypt, Nigeria, Eastern Europe (Hungary, Albania, Kosovo, Macedonia, and Poland), and in Kyrgyzstan, Indonesia, Thailand (Mills, 2000), Chile (Cuimas and Vaidean, 2008), and most of Latin America (Langenbrunner and Liu, 2005).

Some limited evaluation results exist about capitation and fund-holding in LMICs. The primary care reforms begun in Kyrgyzstan in 1994 have been widely reported to be a successful implementation of privatized small group practices (composed of a pediatrician, an internist, an OBG, and a business manager) and an open enrollment process. The small group practices, which now number about 700 in the country (Hardison, et al 2007) are paid a capitation rate. There have been a number of evaluations of the overall program, which now includes a strong retraining program in family medicine, concluding increased patient satisfaction (than the old polyclinic system), lower referrals and hospitalizations, improved blood pressure control, and some improved health indicators for the population (Hardison, 2007).

Langenbrunner and Liu (2005) report that in eastern Europe privatization of primary care has resulted in contracting with private physicians using capitation for just their own services (e.g. no fund holding). One study of Croatia and Hungary noted that the capitation program resulted in higher rates of referral than for salaried physicians (Barnum et al, 1995). In a study of contracting within the Croatia capitated program (Hebrang et al. 2003), the authors report that indicators of access were much better for the contracted providers than for the public providers. In Romania, a privatization scheme for private doctors combined fee schedule and capitation program resulted in improved use of prevention, higher patient satisfaction, but no improvement to the access situation of the poor (Vladescu and Radulescu, 2002).

Partial capitation has been tried in India for a bundle of services surrounding the birth episode (Bhat, 2006). In a one year pilot study, poor families in 5 rural districts were provided a voucher that entitled them to free medical and institutional care for delivery and follow-up. The idea was to stimulate demand for inpatient births among persons who otherwise could not afford the care out of pocket, to enlist services from private providers, who dominate in these regions, and to lower high MMR and IMR. Obstetric providers were able to redeem the voucher for a partial capitation fee (fixed payment per episode), which included prenatal, postnatal, and obstetric care, as well as fund-holding for anesthesia and pediatric care for newborns. In an uncontrolled assessment, the program increased the number of institutional deliveries, reducing the C-section rate, and improved both MMR and IMR rates. This pattern of strong impacts on care seeking, suggest that the findings are probably more related to the strong demand incentives (vouchers) than to the provider payment incentives, though the reduction in C-Section rates is probably a consequence of the conditioning of provider behavior by the fixed payment.

In a related but different reform approach, NGOs are sometimes paid capitation rates by the government to deliver all formal services through a contracting arrangement. This approach is essentially a convenient and fast way to reform through delegation of management authority and financial risk to the NGO, which can organize and manage the district/regional health system resources through fee schedules or other means to contract with providers. These schemes can be used when providers are not well organized, or have insufficient management skills, to be contracted with directly.   In Guatemala (Danel and LaForgia, 2005), NGOs contractors were paid a capitation rate for delivering a basic package of primary care services to a rural population. Access measures and satisfaction were comparable to the public clinic experience, but the utilization measures were better under a contracting-in alternative (management strengthening). In a large but unstudied capitation program for basic services in Haiti, a capitation rate is also paid to NGOs. Over 50% of the population is covered in this program. NGOs are paid 95% of the capitation rate, with a potential bonus of up to 10% if performance targets are met (Chowdhury, 2001, Eichler et al, 2001). The use of capitation and contracting together makes it impossible to un-bundle the effects of payment incentives from effects of contracting-out. We attempt to separate the literature according to whether the P4P program is distinctive from the usual capitation program in these instances.

In a widely reported experimental study of two forms of contracting in Cambodia, NGO contractors for district-level maternal and child were paid a capitation rate with a contractually stipulated penalty for not fulfilling contractual obligations. Self reported immunization levels and attended delivery rates, among other utilization measures obtained from a household survey. These measures were better under capitation incentives than the traditional district delivery system, and better than an alternative where management strengthening was used rather than capitated contracting. The results were clouded by concerns about the comparability of baseline levels of resources in the three models of care being tested. There were some findings of poor effects on quality indicators. Several studies were done of this program, interpreting the impacts very positively (Bhushan et al 2002, Palmer et al, 2004).

In Costa Rica (Cercone et al 2005), a capitated contracting program for primary care services was done, including a penalty for not achieving 85% of contract performance targets. There were mixed indicators of utilization including fewer specialty visits, more general care, and about the same first time and emergency visits as the government clinics. Costs and mortality were no different than the controls.

These very limited results are encouraging, particularly in the case of using capitation payments for contracted providers. Clearly, the intent of policy is not only to implement capitation, but is also to build (work with) organizations large enough to assume capitation risk. There are also positive results for individual providers and small groups. The results are not uniformly good and most of the studies do not examine the risks of under-service. There certainly is broad global interest in using capitation incentives to provide primary care fund-holding, even in LMICs.

Quality/Performance Schemes. The literature that examines the impact of physician P4P on explicit quality objectives is small and inconclusive, even for preventative care where these kinds of quid pro quo bonuses are common. This literature is drawn mainly from the West. One literature review was conducted on the impacts of target payment on primary care physician behavior using the Cochrane Effective Practice Registry (mainly North American and Western Europe members). Only two studies meeting the criteria were found, including only 149 practices. There was some evidence of an increase in immunization rates in one study, but not the other (Guiffrida, et al 2008).

Though the numbers of reported ‘P4P projects’ in the West appears to be growing, there are few, relatively indecisive, impact studies.. One recent comprehensive literature review of financial incentives on physician practice and quality of care concluded that the evidence to date on impacts is very inconclusive, and more research is needed on this rapidly growing form of physician compensation (Christianson et al 2007).

A number of P4P schemes are reported for LMICs, with examples from many different kinds of ‘quality’ objectives. These include target payments for immunization and prevention services (Czech Republic), the payment of inflated prices for cost effective services (Brazil), and the compensation of DOT vaccine incentives(South Africa).Other examples include   the payment of bonuses to private doctors by the local health department upon the curing of patients in Pune, India (reported in Eichler, 2006). In Brazil, municipalities (who operate primary clinics) are given a financial payment when TB patients are cured. In the Congo, eight NGOs contracted to manage district health delivery systems implemented contracts with local hospitals and doctors that included performance payment schemes. In Bangladesh, NGO field staff was paid on the basis of their client’s knowledge of Oral Rehydration Therapy. In Cambodia a provider incentive program was aimed at increasing basic service utilization of the poor. And, in China doctors and community workers were paid when they referred patients (who tested positive) to a TB dispensary. No evaluation results are available for these programs.

There are, however, several LMIC P4P schemes for which evaluation results have been reported. In Bangladesh, a broad set of MCH and other treatments were included in a contract with NGOs as a supplement to other sources of care for an urban poor population. Performance bonuses in the contract featured access incentives for certain services (immunization, lab testing, and client satisfaction). The results showed that the NGO clinics did better than government clinics in promoting access for the urban poor, with no difference in costs, which could be attributed to the various aspects of the contracting, or the incentives. In Bolivia, contractors were paid incentives for delivering MCH services (Lavadenz, 2001). The incentives were based on process and outcome indicators. The results showed strong effects on utilization in the contracted provider sites including outpatient visits, and percent of institutional deliveries.

In Indonesia, bonus incentives were shown (by simulation) to be potentially cost effective in getting young doctors to relocate to rural areas. The size of the incentive needed to accomplish the change in practice location is reported to be less expensive than the current approach (to pay a “bonus” in the form of free specialty training) and less costly than compulsory approaches. (Chomitz et al 1998). In a study in Kyrgyzstan, clinics were paid more if they served poor patients. Though other results were not studied, a household survey found that the patients were more satisfied with this program (Kutzin, 2003).

In Nicaragua a program was aimed to deliver a range of basic services to poor families. It paid contract providers a per capita payment per household to deliver free services such as growth monitoring, developmental monitoring, vaccinations, anti parasitics, and vitamins all for kids. Later, some other maternal and child and Family Practice services were added. Of the per capita payment, 3% was fixed, and 97% contingent on results where the contingent payment is based on documentation that the services were provided. There was weak evidence of impact. Some positive evidence was shown that the program increased service volumes of growth monitoring, but there was not much effect on immunization. The time series comparisons that were used to evaluate these outcomes were confounded by some demand incentives for households that were also included in the program. But, when the demand incentives were terminated the observed effects of the supply incentives seemed to persist, implying that the provider incentives may have been more important that the demand incentives (Regalia and Castro, 2007).

In both Senegal and Madagascar, contracting programs with NGOs were implemented to deliver community based nutrition services in poor areas. Performance thresholds were set and, in some cases, contracts were terminated due to poor performance. Evidence suggests that utilization increased and malnutrition rates fell in study areas, and among enrolled children (Marek, et al 1999).In Haiti (Eichler, 2001), NGO contractors received as much as a 10% bonus based on performance achievement, increasing access to a certain degree. Immunization and contraceptive coverage improved, but the volume of prenatal visits, among other measures, did not. There are some studies about P4P activities in Cambodia and Haiti reporting that P4P incentives need to be accompanied by management training and support (Soeters, 2002).

Finally, a reported side effect of a P4P scheme was seen in China, where hospital based physicians were paid bonuses for seeing more patients in a target group. Not only did the hospital-based doctors successfully generate payments for themselves, but their behavior also contributed to higher hospital revenues (Liu and Mills, 2003).

P4P incentives are being widely used in developing countries. Reported evidence is generally positive that the incentives work as expected. There are exceptions, and it is often hard to say whether the incentives or the ‘contracting’ is responsible for effects. Like capitation for primary care, there is promise for these kinds of incentives for physicians, though no clear consensus about the kinds of impacts to be expected in a given situation.

An Endnote on Contracting. There is literature on the effectiveness of contracting-out with private primary care providers. But, unfortunately, it does not separate the effects of private contracting from the effects of “using incentive payment schemes, and instead tends to bundle these interventions together. So, we do not consider “contracting” a form of payment. There is also some literature in the form of case studies on inpatient outsourcing of dialysis and other hospital services (Nikolic, 2006).

The ‘contracting’ concept is predicated on the idea that contractors have better skills in management than the government resources they are replacing. In a review of this literature, some authors conclude that contracting reforms for service delivery has broad empirical support as a way of effectively reorganizing the delivery system to achieve fast improvements in health and utilization (Loevinsohn and Harding 2005). Other more recent reviewers of the literature are less convinced that the evidence is overwhelmingly favorable (Liu, Hotchkiss and Bose 2008), though there is no question that results of contracting to stimulate access are strong and convincing. But, it is unclear whether the results of contracting are due to the use of “private providers” or due to the volume “incentives” that are used to pay the private providers. They conclude that the effects of contracting with private providers for primary care services is a quick way to stimulate access to such services, though the effects on quality, equity and other objectives have not been frequently studied. Generally, they find that the results of the particular contracting program depend on whether the program of contracting was the sole provider (in some cases replacing a government program) and the kind of payment incentives used. Contracting per se for primary care is not found to be intrinsically successful or unsuccessful.

Beyond primary care, health sector contracting with the private sector is widely done. Loevinsohn and Harding (2005), in a review of the evidence, believe that “based on the successes thus far, there should be a significant increase in the amount of contracting undertaken in developing countries as a means of rapidly improving service delivery and achieving MDGs” (p 208). They do not comment on the role of incentive payment programs for contractors in this recommendation, though they do urge autonomy of providers (and exclude line item budgeting as a payment approach for this reason).

In general, the literature on capitation and P4P for Physicians is confounded by the use of contracting so as to make it impossible to say whether the improvements in access and other outcomes is the result of one intervention or the other.

 

(**) This piece taken from

What is Known About Demand and Other Related Interventions to Improve Access and Strengthen Health Care Systems in Low and Middle Income Countries?, Razavi,M. Gaumer, G. Wallack, S. and others, Brandeis University, Schneider Institutes, April 2009

[1] All of the 26 countries use such methods except Azerbaijan, Tajikistan, Moldova, and Turkey where line item budget systems are still used.

[2] The Clinton health reforms would also have used global budgeting for hospitals, had those reforms been implemented in the U.S.