Most analysts of provider payment programs have remarked on inconsistent or weak performance of provider payment reforms in LMICs? Lundberg and Wang (2006) offer the suggestion that there may be systematic barriers (or key assumptions) that are simply different, but offer no list of what these factors might be. McPake and Banda (1994) suggest that “such advantages (to reforms like contracting out in ways that emulate competition) may not always be realized.” Some of the earliest writers on this topic warn of the barriers: “Low-income countries should avoid complex payment systems requiring higher levels of institutional development” (Barnam et al 1995).

From this commentary we conclude that weak, inconsistent or even counterproductive patterns of impacts may be due to some critical barriers (constraints) or unmet assumptions that prohibit provider payment reforms from being an effective policy instrument in LMICs. Previously, Hanson et al (2003), in path-breaking work, had enumerated a hierarchy of constraints in LMICs that would limit access to new services and other technologies. Jutting (2004) also attempts to enumerate the issues associated with a related and often complementary reform of decentralization, arguing that performance of decentralization interventions have been found to be highly sensitive to organizational and institutional capacity. Poor countries, particularly ones with poor performing management and governance structures in the health system, simply do not allow decentralization reforms to flourish. Among the factors that these authors found to be correlated with successful decentralization in poor countries are:

• Sufficient and stable local finances
• Sufficient local management capacity
• Political commitment at the national level
• Donor support
• Free flow of information
• Accountability
• Policy coherence, particularly between donors and national government”

Langenbrunner and Liu (2005) summarize the reasons for failures of provider payment programs which experience “diluted or neutralized” results:

• Fragmented public sector pooling and purchasing
• Low operational autonomy of providers
• Lack of timely information and routine information systems
• Poor complementarities of payment systems across settings
• Institutional impediments
• Technical capacity and management skills
• Monitoring and quality assurance systems

What follows is an attempt to catalogue the types of constraints that may be preventing provider payment reforms from being effective in LMICs. There seem to be four types of barriers or limitations to success of provider payment programs:

• Health System Limitations (primarily conflicting coordination between benefit plans, civil service incentives, and payment incentives)
• Capacity Limitations (primarily information and management limitations)
• Resource Limitations (making provider payments on time, and level of generosity)
• Tempering Incentives on Underserving Patients

We discuss and illustrate these constraints in the following sections.

Capacity Limitations

Data and Analytic Support. Many countries do not have sufficient patient and provider data resources to set administrative fee schedules or capitation rates, or to monitor their performance. Provider payment reforms, particularly ones that require monitoring medical necessity, or the risks of under-service, and those that measure casemix in some fashion require considerable payer resources to implement and operate. Very little is reported in the literature about administrative costs of implementation of various types of payment incentives. Wouters et al (1998) report that administrative expenses generally follow the continuum of provider risk, i.e., the more provider risk, the greater the cost of administration. Line item budgets are the least costly, and capitation is the most costly.

Data is needed to set fair base rates, to regularly recalibrate payment rates to keep up with changing practice patterns and technology, to monitor results, and to help provider managers do their job. In the absence of insurance systems (with paid claim records) or computerized patient records, there is no easy way to operate a casemix system or to monitor patient care results. Almost all LMICs have this problem. In Egypt, reforms in primary care were centered on family medicine clinics where provider teams were paid bonuses to achieve objectives relating to service quality (prevention services, waiting time, etc.) The pilot data for measuring and monitoring the program came from computerized patient records for every visit. Though donors funded the pool for paying these bonuses, the rollout to hundreds of new clinics in the pilot scale-up neglected to implement the computer system, so the bonus part of the payment system was not deployed (Gaumer et al, 2008).

Analytic resources are also needed to work with these data and refine and update payment rates and methods. If these investments are not made, impacts can be distorted (uncontrolled volume incentives, obsolete rates, under-service and deteriorating quality) as was the case in Brazil’s physician fee schedule reforms (Wouters et al, 1998). One observer remarks: “Case based reimbursement such as DRGs is, from a technical perspective, an improvement on FFS systems, because it pays for outputs rather than inputs. Such systems require sophisticated and expensive methods to monitor and update payment rates, and therefore are probably not feasible in poor countries” (Kutzin, 1995).

Provider Management Capabilities. Provider payment incentives presume that beyond the obvious ‘dire’ consequences for taking action in response to payment reforms, it is also assumed that managers see the need for change, that the can set the new course, and that they can be effective in getting it done. Deficiencies in these aspects of management effectiveness are common deficits in LMICs. Without these competencies, response to incentives may range from ignoring the need for change, to poor implementation process. These can stem from poor management skills to inadequate data. As discussed early US experiences with per case payment and capitation were accompanied with anecdotal reports of inaction, over-reaction and failure to find a good balance between efficiency and quality. Facilities closed (dire consequences) and managers were replaced with more effective people. Considerable investments continue to be made in information systems by these institutions to support the managerial need for information.

Weaknesses in management buffer the intended incentive effects of payment reforms. One aspect of management that is particularly critical for payment reform to create impacts is management autonomy. This does not mean that it is necessary to have private governance of hospitals or clinics, rather that in responding to incentives, managers must have control over staffing level, staff selection, staff performance expectations, capital strategy, and other spending decisions. This flexibility is needed to respond to incentives and facilitate improved performance of the organization. Without this autonomy there is limited response to the incentives of performance based payment. Examples of contracting in Bolivia and Cambodia and other hospital autonomy reforms suggest that “in instances where management is given only limited autonomy, performance has improved very little” (Harding and Preker (2003).

There are many other examples of attempts to implement incentive payment for providers where the central or regional government retains autonomy over staffing and salary levels, capital and technology spending, purchasing of pharmaceutical supplies, and other matters. In Kosovo, for example, decentralization reforms of primary care with a capitation grant to the municipalities was done to encourage better performance and more accountability than the previous centralized system under the MOH (Gaumer, 2007). But here, facility managers have essentially no authority over their staff and their compensation because of civil service law and other policies. Once the budget is set, the authority to make timely deviations to deal with changing circumstances is not given to the managers. Changing the formula for the block grant to the municipalities (to some form of performance based incentive budget) would not have any effect on clinic behavior. It would be wrong to conclude that there is ‘autonomy’ in any conventional sense of the term. There have been recent discussions in Kosovo about trying to improve performance of government hospitals by means of a per case payment system. But, as is clear from the situation in municipal clinics, hospital directors would need to be able to shift resources, resize facilities and the workforce, and change the internal culture to effect change in response to the incentives of a per case system. None of this is possible now within the policies of the MOH which provide no management autonomy for facility directors.

To support better management actions in response to incentives, data and information systems are also critical. But, sophisticated measurement and monitoring systems will not be valuable until autonomy is available and decision support is needed. Continuing the example, there is no real evidence of demand for more information at any level in the Kosovo system (and in other countries where managers do not have to ‘manage’ because they have no autonomy). No patient feedback information is sought or collected by facilities. Facility managers who have special computer systems and staff, do not ever request special reports to facilitate ‘management’. At all levels, performance measurement is simply not a priority for managers. Holding facility directors accountable for facility performance would change this, but making them accountable for aspects of performance they cannot control would be futile. Demand for performance information should ultimately follow policies that provide more autonomy for managers to allocate resources and manage staff.

 

Resource Limitations

A second barrier to effective response is lack of sufficient financing and delayed payments by the payer (usually the government). One important source of inadequacy in paying providers is policy that ties provider payments to the annual government budget, which creates vulnerability in tight budget situations. These ‘tight’ budgets can slow payments and dull the financial incentives of the incentive payment scheme. This happens because of loss of ‘trust’ in the payment rule, and because of the loss of liquidity. Without timely payments based on the payment rule, many payers fall into debt to providers. Subsequently, in the absence of adequate capital market instruments to borrow, providers often ‘borrow’ by slowing down payments to staff and suppliers. In some areas of southeastern Europe this is commonly referred to as a “Balkan financing scheme” due to it’s prevalence in the region. This “debt” situation (however financed) can buffer the effects of payment incentives by eliminating necessary liquidity for investments needed for organizational effectiveness, and by creating dependencies on staff and suppliers who are owed money by the facility. Institutionally, this is resolvable by allowing facilities access to capital markets, or better yet, by creating a health financing fund (for paying providers) that is managed separately from the government budget.

Chronically low budgets may also limit the generosity of payment rates. While the direction of the marginal incentives of the payment policy are not altered by this, the providers may seek other better paying sources of patients. This eventually will buffer the incentives attached to the payment policy, and possibly cause providers to limit access to patients whose services command inadequate payments. This happened in Brazil (Wouters, 1998).

Health System Limitations

Weak or conflicting incentives. In some instances, payment schemes are designed with flaws like weak or conflicting incentives. Conflicting incentives were a serious problem in Croatia, where there was a combination of capitation for primary care physicians, coupled with FFS payment for specialists and hospitals. This encouraged “dumping” or excessive referrals from primary care to higher levels of care creating increases in total spending and the share of spending going to hospitals (Langenbrunner, et al 2005). In other instances hospital payment incentives conflict with incentives for the managers themselves, where they are often salaried at a level that is commensurate with bed-size, creating conflict between the decision to close beds and to be paid more). This is seen in many countries.

There are also situations where the incentives of provider payment are simply too weak to create large impacts. The P4P program in Haiti, where NGOs are paid 95% of the capitation rate, supplemented by a bonus payment of up to 10% is a fairly modest incentive to meet the performance targets. In the Nicaragua case, to contrast, the NGOs are paid only 3% as a base, with 97% contingent compensation based on performance.

Conflicting Benefit Plan and Provider Incentive Policies. The impacts of the incentives of provider payment depend, in large part, on the consequences to the organization for failing to offer a ‘quality’ product within the parameters of available financing. What happens to providers who fail to do this? What happens to providers who face fee schedules and must compete for consumers who can “vote with feet,” and who fail to attract enough business? In the case of a free market, the answers are simple: they fail and close their doors. This consequence provides very strong incentives for managers to be efficient and offer a quality product with what is available. Less mature markets may buffer those incentives. For example, a provider payment reform for hospitals in isolated circumstances (district hospitals in rural areas, for example) may not work well because the government may not be able to let facilities close if they fail to keep costs in line with revenue.

Hanson et al (2003) offers a good description of many of the problems facing households and communities that also undermine the market forces upon which payment incentives are based. Consumers who lack awareness of options and quality differentials, cannot afford the price at the point of service, are uncertain about the effectiveness of formal care, and face distance/transportation barriers, can mute the operation of competitive incentives. So, if one provider is careful, thorough and offers a good product, then it is possible under these circumstances that they may not be successful in attracting more business from consumers than their competitor. In this kind of household- and community-constrained environment, provider incentives that rely on survival and related market outcomes may not work very well. Provider incentives are muted because being responsive to incentives may bring little reward.

Coordinating provider payment with demand incentives in the benefit package/pooling design is important. Out of pocket payment (or informal payments) in poor countries reduces overall demand for care, and may buffer incentives of the provider payment scheme by reducing the payoff to providers for improving value of their services to the marketplace. For example, if providers are paid more for attracting more patients (a competitive incentive) then anything that reduces the demand for care will detract from (buffer) the impact of the payment incentives. Pressure to have the patients pay more may also accompany payment reforms that put financial pressure on providers. In Eastern Europe, for example, there has been a growing reliance on out of pocket payments in the wake of provider payment reforms (Langenbrunner and Wiley, 2002). When provider payment was established, the providers were, in some instances, given the flexibility to charge user fees. Though the impacts of reforms on hospital efficiency may be positive, the financing system “shift” to out-of- pocket may be an important negative consequence. And, higher price at the point of service may mute the competitive incentives in as much as consumers may lower their demand response to providers who are more effective in responding to the payment incentives.

Any type of coordinated change in the benefit program that has the effect of stimulating demand will increase the apparent effectiveness of provider payment incentives. When options for care exist, efforts to increase the purchasing power of consumers (demand policies) will stimulate utilization and increase the return to search. This complementary stimulation of demand would invigorate the incentives facing providers and increase impact of the payment reform. Insurance schemes, for example, should increase the effectiveness of provider payment reforms by strengthening demand as the point of service price falls. Concurring and coordinated, payment policies and benefit programs should improve (Kutzin, 2003, Falkingham, 2001).

Even in instances where there is no presumption of competition (such as many of the capitation situations, or some of the global budgeting situations) the absence of dire market consequences may buffer payment incentives. This would follow from failing to punish providers for not doing an adequate job of balancing efficiency and quality. Of course, governments in most LMICs do not want to do things to providers that might discourage access or equity—these systems objectives are (rightly) more important than efficiency (and possibly quality as well). One author writing about New Zealand states: “The effect of introducing market-like incentives into a health system depends upon the particular institutional arrangements that are in place. As long as governments place high priority on ensuring access to services for those in need, incentives for efficiency will inevitably be blunted (Ashton, 2002, p103).”

There may also be higher level institutional conflicts that prohibit effective operation of provider payment systems. The most obvious of these is civil service, which often ties the hands of facility managers, as in Kosovo. This is also mentioned by Langenbrunner, et al (2005). The manners in which civil service can impede or conflict with provider payment incentives are numerous. An obvious situation is created when hospital managers are compensated according to hospital bed-size, which may conflict with incentives to downsize the facility in response to payment incentives.

 

Tempering the Underservice Incentives of Provider Payment

The use of provider payment as an incentive device for improving health system performance in LMICs is problematic because of the underservice incentives and the weak control mechanisms in many countries. This may be responsible for the inconsistencies in impacts.

Any form of prospective bundled payment involves economic incentives to underserve patients. Whether a DRG system, or a per diem payment, or capitation, all involve incentives for providers to underserve. And, the variability in impacts across LMICs we see may be the result of variation across providers in the self imposed limits on underservice.

In the OECD countries there are typically 3 major vehicles for controlling or tempering the incentives for underservice. Absent these vehicles, it may be risky (high levels of variability) to implement such payment systems without providing for some way to temper or monitor the impacts of the incentives.

 

Availability of Monitoring Data. There are less likely to be industry wide data systems that can be used to study the extent of underservice. The insurance systems of many western countries produce administrative data on every patient that can be used to examine service usage, LOS, and other measures of intensity.

 

Legal System. Underservice brings a real risk to providers in terms of lawsuits. A healthy supply of lawyers, and knowledge of legal remedies, poses a significant “defensive” barrier to underservice (whether systematic or occasional). This somewhat annoying tradition in the West, has be a source of excessive spending, but also a form of protection against underservice. In LMICs, this protection may not function in a way to limit underservice risks.

 

Free Press. Working with the Legal System, the press provides protection by occasionally publicizing care patterns that reflect underservice. The threat of such adverse publicity is a very real concern of most institutions, and helps prevent gross underservice. This mechanism may not function at all in many LMICs.