Pay for performance can be modernized

November 1, 2007

Today, hospitals and other facilities are implementing forms of pay for performance in previously unimaginable ways. Such programs typically use established ratings methods and indicators to measure levels of quality, then offer incentives or compensation to entities or providers who are rated highly based on these indicators.

Today, hospitals and other facilities are implementing forms of pay for performance in previously unimaginable ways. Such programs typically use established ratings methods and indicators to measure levels of quality, then offer incentives or compensation to entities or providers who are rated highly based on these indicators. And although the pay-for-performance concept has been developing over time, it’s likely that we are still in the early stages of the phenomenon and have yet to see how, and to what degree, it will take hold.

The challenge of pay for performance lies in finding a balance between the cost of the programs and the quality they elicit. As with many healthcare management strategies, it will take time to see how the newest generation of pay-for-performance programs will stack up in terms of generating consistent, authentic quality.

In the modern era, HMOs began utilizing capitation fees and gain-sharing in the 1980s. These forms of pay for performance promoted the sparest, fewest, or lowest-costing treatments. Physicians received compensation dollars for all patients treated in the system, no matter what doctoring services were provided. These programs essentially rewarded physicians for any healthcare dollars saved in order to promote efficiency.

Programs from this time period reached for the easiest measurements of quality and efficiency, and stopped there. Quality indicators were simpler than those of today.

The real groundswell of U.S. pay-for-performance programs occurred about seven years ago, when the Institute of Medicine issued two reports that pointed to the distressing rate of error in the U.S. healthcare system. At that time, general low-quality indicators were often seen. All of this led to a drive for some type of control and reform.

Today, pay-for-performance programs use and build upon concepts from 20 years ago but look beyond present-moment efficiency to focus on clinical performance and patient satisfaction in various healthcare settings. Also, Medicare and Medicaid programs have begun to implement pay for performance to rate quality; at least five such programs are already under way. According to the National Pay for Performance Survey, there was a 24% increase in the number of commercial pay-for-performance programs between 2004 and 2005.

If a recent national survey has indicated that pay-for-performance programs are on the rise, that also suggests, by some accounts, that quality indicators for American healthcare are rising year after year. However, a slippery issue with measuring the quality of healthcare procedures is that they, like many scientific measures, can be influenced by the measurement itself. When one healthcare indicator is measured for quality, we know quality tends to improve; but it is unclear whether quality is rising beyond those items being measured. While claim data is helpful in calculating process measurement (e.g., was a retinal exam performed on a diabetic?), it is difficult to use claim data to calculate outcomes measures (e.g., what percentage of patients have blood pressure below the target level?).

However, a 2006 report in the Annals of Internal Medicine reported positive results in quality indicators from pay-for-performance incentives. Another report, conducted by Medicare in the second year of its quality improvement organization program, reported improvements in 34 of 41 measures studied.

Measuring quality takes time and costs money. That is just one potential roadblock that administrators may face in deciding whether or not to implement pay for performance. Programs should be designed intelligently to minimize or avoid excess costs. Administrators must decide: Should existing quality measures be used, or should some be added and at what cost?

In the context of offering "carrots" to physicians to help them perform, what steps will administrators take in regard to the underachievers, those physicians in the system who may simply not be interested in improving performance, or who consistently score low on indicators? In a related way, we might ask if pay for performance ultimately creates a two-class system, ranking certain physicians into higher tiers, and if this higher ranking will make it difficult for patients to see those physicians. How might we redesign programs to eliminate any unfair advantages?

Contemporary programs tend to emphasize rewards for providers, and at times, rely on the fact that team members can thrive inside the spirit of incentive, participation, and scoring-motivated group effort. It's also fair to say that current programs seek healthier lives and a more careful administration of healthcare as their goals, placing performance and satisfaction in the foreground. Ideally, future pay-for-performance programs will balance clinical quality, efficiency, and patient satisfaction.

Still, the costs of implementing quality measurement and physician and clinic incentives remain high, despite efforts to be economical. Pay-for-performance advocates say the costs of incentives are nominal; for example, at 1% to 2% of a physician's total compensation. Given the already high cost of healthcare, these added percentage points beg an important question: Will pay-for-performance programs cause premiums to rise, with the unintended effect of excluding some of the very individuals we wish to cover? At present, the answer is unknown.

Proponents of pay for performance state that quality measurements help keep healthcare entities alert to quality in a more global way, more so than if there were no programs in place in that entity. Proponents also suggest that the costs of pay for performance represent an investment in the future. Improved quality today, they say, will function in the long term as a ballast for each entity and for the system as a whole, creating stability and preventing future problems. Although we do not yet know if these programs save money in the long run, the consensus is that they improve quality where measured. In a perfect world, today's current pay-for-performance trend may contribute to a wholesale change in healthcare administration and a more quality-driven U.S. healthcare system overall.

Catherine Murphy-Barron is a consulting actuary in the New York City office of Milliman.