Ateev Mehrotra, MD

		Telemedicine offers the potential for better care, lower costs, and increased convenience.

		Expanding existing fee-for-service payment models to include telemedicine may lead to its overuse. Already there are many skeptics among payers and providers regarding this method of payment.

		The greatest potential for effective and efficient use of telemedicine services lies in the use of emerging payment models.

Telemedicine has the potential to reduce healthcare costs, increase access to healthcare, and improve health outcomes.

 However, despite these prospective advantages, telemedicine has been adopted in only limited circumstances, such as for reading radiology reports and remote phone or video visits for minor health issues.

 To promote adoption, some have advocated for telemedicine services to be reimbursed by health plans and many states have issued "parity laws" that force payers to reimburse some forms of telemedicine services at a level equal to in-person care. In this perspective, we explore why health plans are reluctant to cover most telemedicine services as part of current payment arrangements and explain why newer payment models offer greater potential for its expanded and more effective use.

Telemedicine is the delivery of health-related services via telecommunication technologies, such as phones and the Internet. Although some services involve direct interaction between providers and patients.

 Others involve passively collecting patients’ data from their homes and sending the information to healthcare professionals for monitoring.

A central barrier to the adoption of telemedicine is fee-for-service (FFS), the predominant payment method in the United States. In FFS, providers submit a claim to be reimbursed for each clinical service they perform. Providers are not paid for any service not included in the payer’s fee schedule. In many cases, virtual “visits” via phone or the Internet may be as effective in delivering care as in-person visits and require fewer resources to provide.

 However, most healthcare providers who adopt virtual visits will be financially penalized because many types of visits are not reimbursed or are reimbursed at lesser amounts. The few forms of telemedicine that have been widely adopted, such as remote reading of radiology reports, are exceptions. In those cases, the technology does not jeopardize FFS revenue for the provider adopting the technology.

Proponents argue that payers should add telemedicine as a reimbursable service. Some states have enacted parity laws and Congress is considering legislation to expand FFS reimbursement of telemedicine by Medicare. However, arguments in favor of this approach meet resistance from providers and payers alike. Providers are concerned that offering telemedicine (which, other than services covered under a parity law, is usually considered lower intensity compared with in-person visits and therefore reimbursed at a lower rate) will eat up much of their income because it will reduce the need for in-person visits. Payers have the opposite concern: although telemedicine reimbursement may increase use of actual service, inappropriate use will increase costs.

 This concern regarding inappropriate use is consistent with other coverage decisions. For example, when payers expanded coverage of implantable cardiac defibrillators in the 1990s and 2000s, inappropriate use of the devices grew dramatically.

 Telemedicine may be particularly prone to overuse because of its key advantages—convenience and increased access.

Non-FFS payment models offer alternative ways to pay for telemedicine. Under capitation, the strongest form of these new models, providers receive a fixed payment to provide care for each individual. Thus, any reduction in care costs that results from using a lower-cost technology, such as telemedicine, would accrue entirely to the provider organization.

Because pure capitation involves substantial risks for providers, it is not a common payment arrangement. However, emerging models, such as Accountable Care Organizations (ACOs), involve a form of “global payment” in which providers’ payments partially depend on the total health costs that their patients incur. In ACOs, providers must also meet quality-of-care standards, thereby favoring technologies that both improve quality of patient care and reduce costs.

Evidence from health maintenance organizations (HMOs) in the 1980s and 1990s suggests that the new payment models can stimulate adoption of cost-effective technology. HMOs are similar to newer payment models in that they also typically hold providers accountable for the costs of caring for their patients. One study of HMOs found that hospitals in areas with higher concentrations of HMO patients were more likely to use low-intensity heart attack treatment technologies (eg, medical management techniques) and less likely to use high-cost technologies (eg, cardiac catheterization, angioplasty, or bypass surgery), which evidence suggests are overused.

 The HMO experience suggests that new payment models will likely stimulate the adoption of telemedicine, which has advanced considerably in technical capability since the 1990s, and early evidence suggests that this is the case.

By placing incentives for reduced cost and improved quality in the hands of provider organizations rather than insurers, new payment approaches should help foster the efficient use of telemedicine. Under FFS, payers are often faced with blanket coverage/no coverage decisions for all providers in their network, regardless of whether FFS is being used in an appropriate or inappropriate clinical situation.

Under new payment models, adoption decisions are made by the providers, who have the flexibility to experiment with telemedicine and use the technology in a way that maximizes clinical outcomes and minimizes costs.  We provide 2 examples of how this may happen: videoconferencing for outpatient visits, and home telemonitoring.

Real-time video communication between patient and provider allows for many of the same benefits as an inperson visit, without requiring the patient or provider to travel. This form of telemedicine has been evaluated in research studies for decades, and several studies have shown positive effects on patient outcomes even when low-bandwidth, telephone-based Internet connections reduced the quality of the video image.

 Videoconferencing in the form of visits for new acute problems is often covered by health plans, but use of videoconferencing is uncommon, particularly for chronic illness care, where it may have the most benefit. However, provider systems that have non-FFS payment models are already making extensive use of videoconferencing. In the Veterans Health Administration (VHA), videoconferencing is used routinely to deliver mental health services.

Kaiser, a capitated system that is providing 2.5 million “phone visits”each year, is testing videoconferencing for more general use,

and has made more than 350 after-hours video visits to patients who claim they would have otherwise gone to the emergency department.

As more providers adopt new payment models, they will likely also begin to use videoconferencing. As an early example, the Alternative Quality Contract from Blue Cross Blue Shield uses global payments, and providers under that contract have begun piloting a telemedicine platform that enables videoconferencing visits.

 Because of global payments, the pilot practices will be motivated to prevent overuse, perhaps by triaging patient requests for videoconferences, using some form of co-pay, or making the virtual visit available only to patients for whom a physician believes it would be most beneficial.

Many technologies are available that measure health indicators of patients in their homes and transmit the data to an overseeing provider. The provider, who might be a physician, nurse, social worker, or even a non-clinical staff member, can filter patient questions and report to a clinical team when necessary. Several studies show the potential benefit. In a cardiac study at Partners Healthcare, 3000 congestive health failure patients received in-home monitoring of weight, blood pressure, heart rate, and pulse oximetry. Decision support software helped identify high-risk patients. As a result, readmissions dropped by 44%, saving the health plan $10 million in 6 years.

The VHA also implemented a home telemedicine program in which the more than 17,000 patient participants had a 25% reduction in the number of bed days of care and a 20% reduction in the number of hospital admissions compared with usual care.

As new payment models that reward cost savings spread, providers will have incentives to adopt these home monitoring technologies as part of routine care. Early reports suggest that providers are responding to these incentives: ACOs are planning to use home monitoring to prevent more costly hospital visits or emergency care.

As with virtual visits, the new payment models will encourage telemonitoring to be used only for patients who will likely benefit from the service.

Telemedicine technologies have the potential to benefit patients by increasing access to care, promoting convenience, enhancing quality of healthcare, and reducing costs.

 However, payers and providers are unlikely to support adding telemedicine as a reimbursed service because of its uncertain effect on provider payments and its potential for overuse. Direct reimbursement might still be needed for certain telemedicine services, but such payments will likely be the exception, not the rule. Emerging payment models offer the greatest hope that telemedicine will be widely adopted and used in a way that will make it worth the cost.

 From RAND Corporation, Boston, MA (RSR, DA, AM), and Santa Monica, CA (MZ).

Some of this work was funded by a generous gift from Teletracking, Inc.

The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the content of this paper.

 Concept and design (RSR, DA, AM); acquisition of data (MZ); analysis and interpretation of data (MZ); drafting of the manuscript (RSR, DA, MZ, AM); critical revision of the manuscript for important intellectual content (RSR, DA, MZ); obtaining funding (RSR, DA); administrative, technical, and logistic support (RSR, AM, MZ); and supervision (RSR).

 Robert S. Rudin, PhD, RAND Corporation, 20 Park Plaza, Suite 920, Boston, MA 02116. E-mail: 

1. Kvedar J, Coye MJ, Everett W. Connected health: a review of technologies and strategies to improve patient care with telemedicine and telehealth. Health Aff (Millwood). 2014;33(2):194-199.

2. Schwamm LH. Telehealth: seven strategies to successfully implement disruptive technology and transform health care. Health Aff (Millwood). 2014;33(2):200-206.

3. Lewis RS, Sunshine JH, Bhargavan M. Radiology practices’ use of external off-hours teleradiology services in 2007 and changes since 2003. Am J Roentgenol. 2009;193(5):1333-1339.

4. Courneya PT, Palattao KJ, Gallagher JM. HealthPartners’ online clinic for simple conditions delivers savings of $88 per episode and high patient approval. Health Aff (Millwood). 2013;32(2):385-392.

5. Al-Khatib SM, Sanders GD, Mark DB, et al. Implantable cardioverter defibrillators and cardiac resynchronization therapy in patients with left ventricular dysfunction: randomized trial evidence through 2004. Am Heart J. 2005;149(6):1020-1034.

6. Al-Khatib SM, Hellkamp A, Curtis J, et al. Non-evidence-based ICD implantations in the United States. JAMA. 2011;305(1):43-49.

7. Cutler DM, McClellan M. The determinants of technological change in heart attack treatment. National Bureau of Economic Research; 1996.

8. Skinner JS, Staiger DO, Fisher ES. Is technological change in medicine always worth it? the case of acute myocardial infarction. Health Affairs. 2006;25(2):w34-w47.

9. Monegain B. Two Blues go for telehealth: tap American Well for projects in Louisiana and Massachusetts. Healthcare IT News. June 21, 2013. http://www.healthcareitnews.com/news/two-blues-plans-gotelehealth. Accessed December 1, 2014.

10. Nelson B. Two Accountable Care Organizations (ACOs) Share their Strategies for Success. The Hospitalist. April 1, 2013. http://www.thehospitalist.org/details/article/4519521/Two_Accountable_Care_Organizations_ACOs_Share_Their_Strategies_for_Success.html. Accessed December 1, 2014.

11. Garber AM. Cost-effectiveness and evidence evaluation as criteria for coverage policy. Health Aff (Millwood). 2004;(Suppl Web Exclusives):W4-284-296.

12. Wade VA, Karnon J, Elshaug AG, Hiller JE. A systematic review of economic analyses of telehealth services using real time video communication. BMC Health Serv Res. 2010;10(1):233.

13. Deen TL, Godleski L, Fortney JC. A description of telemental health services provided by the Veterans Health Administration in 2006-2010. Psychiatr Serv. 2012;63(11):1131-1133.

14. Pearl R. Kaiser Permanente Northern California: Current experiences with internet, mobile, and video technologies. Health Aff (Millwood). 2014;33(2):251-257.

15. Kulshreshtha A, Kvedar JC, Goyal A, Halpern EF, Watson AJ. Use of remote monitoring to improve outcomes in patients with heart failure: a pilot trial. Int J Telemed Appl. 2010;2010:3.

16. Darkins A, Ryan P, Kobb R, et al. Care coordination/home telehealth: the systematic implementation of health informatics, home telehealth, and disease management to support the care of veteran patients with chronic conditions. Telemed J E Health. 2008;14(10):1118-1126.

Articles

With effective methods of promoting telemedicine yet to be established, new risk-based payment models offer the greatest potential for telemedicine's effective and efficient use.

Paying for Telemedicine

Patient financial incentives are being promoted as a mechanism to increase receipt of preventive care, encourage healthy behavior, and improve chronic disease management. However, few empirical evaluations have assessed such incentive programs.

In South Africa, a private health plan has introduced a voluntary incentive program which costs enrollees approximately $20 per month. In the program, enrollees earn points when they receive preventive care. These points translate into discounts on retail goods such as airline tickets, movie tickets, or cell phones.

We chose 8 preventive care services over the years 2005 to 2011 and compared the change between those who entered the incentive program and those that did not. We used multivariate regression models with individual random effects to try to address selection bias.

Of the 4,186,047 unique individuals enrolled in the health plan, 65.5% (2,742,268) voluntarily enrolled in the incentive program. Joining the incentive program was associated with statistically higher odds of receiving all 8 preventive care services. The odds ratio (and estimated percentage point increase) for receipt of cholesterol testing was 2.70 (8.9%); glucose testing 1.51 (4.7%); glaucoma screening 1.34 (3.9%); dental exam 1.64 (6.3%); HIV test 3.47 (2.6%); prostate specific antigen testing 1.39 (5.6%); Papanicolaou screening 2.17 (7.0%); and mammogram 1.90 (3.1%) (P <.001 for all 8 services). However, preventive care rates among those in the incentive program was still low.

Voluntary participation in a patient incentive program was associated with a significantly higher likelihood of receiving preventive care, though receipt of preventive care among those in the program was still lower than ideal.

		Discovery Health, a private South African health plan, has implemented a patient incentive program which rewards healthy behavior using discounts on retail goods and travel.

		Over two-thirds of the health plan’s enrollees have voluntarily chosen to pay approximately $15 per month to join the program.

		Among those within the incentive program, we found the estimated increase in receipt of preventive services to be 3% to 8%, or a relative 19% to 97% increase.

		Our results support the idea that patient incentive programs might be a mechanism for health plans to increase rates of preventive care

Despite widespread efforts to encourage prevention, rates of preventive care use fall well short of recommendations.

 Much of the focus on improving preventive care has been on decreasing financial barriers. For example, new laws in the United States have eliminated patient out-of-pocket costs for preventive health services.

 While removing out-of-pocket costs will increase the number of people who receive preventive care, the increase is likely to be modest.

Employers and health plans are exploring whether patient incentive programs can spur greater use of preventive care.

In a patient incentive program, a patient receives money or some other financial reward for healthy behavior.

 In theory, these programs address a fundamental problem with preventive care—when making the choice to receive preventive care, patients balance the inconvenience of receiving preventive care with distant and often intangible benefits. Humans generally discount such future benefits

 and therefore it may not be surprising that many patients do not seek preventive care. Incentive programs might help address this discrepancy between immediate inconvenience and future benefit by increasing the perceived immediate benefits of prevention.

There have been several randomized trials focusing on patient incentives to promote healthy behavior.

 For example, Volpp and colleagues found that a $750 incentive led to a three-fold increase in the number of people able to quit smoking.

 While important, this prior research has been limited to small clinical trials with a narrow focus, relatively short follow-up periods, and an incentive structure that may not be sustainable.

 In this paper, we study the impact of a patient incentive program operated by a private health plan in South Africa which has been in place for over a decade and includes almost 1.5 million enrollees. In this program, receipt of preventive care services “earns” points for enrollees, and points translate into rewards such as discounted travel or cell phone minutes. We assessed the impact of enrollment in this incentive program on receipt of preventive care services by comparing the receipt of preventive services among those who joined the program to those that did not join the program.

We analyzed the receipt of preventive care for members of the Discovery Health Plan in South Africa between 2005 and 2011. In South Africa, approximately 15% of the population, typically the most affluent, obtain private health insurance either independently or through their employer. Those with private insurance receive care from physicians and hospitals in a system entirely separate from the larger public healthcare system. In our 

), we demonstrate that those with private health plan insurance in South Africa are socioeconomically comparable to the general US population.

Our study population included both health plan members in the incentives program and those not in the incentives program. Our only exclusion criteria were those in a separate low-cost insurance product. These members were not eligible for the incentive program and because this product is targeted to the poor, the enrollee population is very different.

The health plan’s reward program focuses on encouraging both prevention and healthy behaviors. The incentive program is offered on a voluntary opt-in basis as South African law does not permit such programs to be made mandatory in a health plan product. Members must pay $17 per month for an individual or $21 per month for a family to enroll in the incentive program (approximately 5% of the cost of health plan membership). Enrollees can drop the incentive program at any time and on average, 7.5% drop out in a year.

Members of the reward program earn points for a number health behaviors such as receiving preventive care services, visiting a gym, smoking cessation, seeing a nutritionist, and buying healthier food at the grocery store. Our focus is just on receipt of preventive care services. In 

, we list the allocation of points for these preventive services. The points earned by receiving preventive care did not change during study period. Points translate into increasing status levels. The default status is Blue. For an individual member, status increases as follows: Bronze (15,000-34,999 points); Silver (35,000-44,999); Gold (≥45,000); and Diamond (3 consecutive years on Gold).

Higher status level translates into increasing discounts on a range of goods and services from approximately 25 commercial partners in South Africa. These include store purchases, movie tickets, local and international flights, car rentals, and hotel booking. For example, the discount for 1 hotel chain was 30% and 50% for Blue and Diamond status, respectively. In 2009, the average annual value of incentives to members of the reward program was approximately $275 across all members and greater than $1500 in the highest Diamond tier.

Enrollees can sign up for the incentive program in several ways: they can sign up when they first join the health plan, or members already with the health plan can sign up with a phone call or download and submit an application form that is available online. The process of accruing points is automated and does not require members to submit any forms. For instance, when a claim for a mammogram is received by the health plan, points are automatically allocated to the patient. Rewards (discount) are either available at the time of the purchase or are paid in a check at the end of the month. During the study period, enrollees could receive the points even if they received the service ahead of what might be recommended by different guidelines. For example, some guidelines recommend a mammogram every 2 years, but if a woman in the incentive program received a mammogram yearly, she would receive the associated points each year.

For each year, we classified whether the health plan enrollee was a member of the reward program as of January 1 of that year. We used an “intention-to-treat” analysis such that a patient who enrolls in a reward program was considered to be in the reward program in all subsequent years even if they unenrolled at some point.

For each enrollee, we tracked receipt of 10 preventive care services within each calendar year. Eight preventive care services were associated with financial incentives (cholesterol testing, fasting glucose testing, human immunodeficiency virus [HIV] test, mammogram, Papanicolaou [Pap] test, dental screening, glaucoma screening, and prostate specific antigen test). Two preventive care services were not associated with financial incentives (colon cancer screening and bone density scan for osteoporosis). We tracked the 2 nonincentivized services because there is concern that financial incentives may lead enrollees to neglect nonincentivized services.

Although it is associated with an incentive, we did not track receipt of the influenza vaccine, because many enrollees receive the vaccine in the workplace or pay out of pocket and therefore it is not accurately tracked in the health plan claims. We also did not track human papillomavirus and pneumococcal vaccinations, because incentives for these preventive services began in 2010. Lastly, we did not track receipt of childhood vaccinations. As described below, we examined the likelihood of receiving a preventive service before and after entry into the program. Such a model cannot be used for services only offered during a narrow age period. For each service, the health plan designates age and gender eligibility for incentive program members to claim points (Table 1). Preventive care services were identified by relevant billing and diagnosis codes in health plan claims.

Health plan members must first visit a general practitioner to receive 6 of the 10 preventive services. For 3 services (cholesterol screening, glucose testing, and HIV test), health plan members can also get these tests without a physician’s order at a pharmacy or on wellness days at work sites. For dental screening, patients go directly to a dentist. As of 2006, South Africa law mandates that the 8 preventive services, except for dental screening and glaucoma screening, be provided without any patient co-payment. In 2005, coverage for preventive services varied by type of health plan. We did a sensitivity analysis in which we eliminated 2005 data.

In our analyses, we examined the receipt of a preventive service within a single year. Many preventive services such as breast or cervical cancer screening are not indicated yearly. The percentages reported are therefore not indicative of what fraction of the population is up to date with preventive care.

Measuring Utilization of Health Services by Health Plan Members

We tracked the number of visits to a general practitioner and whether the enrollee had 1 of 20 chronic illnesses. The chronic illnesses were identified by a diagnosis on an outpatient or inpatient claim or whether the patient filled a prescription for a medication related to that illness (diagnoses listed in 

). Chronic illness data was only available from the health plan from 2008 through 2011.

We estimated the association between incentive program membership and receipt of preventive care services using an individual random-effects logistic regression. We used a separate regression for each preventive care service. The unit of analysis is each enrollee time year. The dependent variables are 10 dichotomous variables denoting the annual receipt of each preventive care service. The key explanatory variable is a dichotomous variable denoting whether a given enrollee was enrolled in the incentive program for that year. This allows us to compare those who entered the incentive program versus those that are not. The other explanatory variables are gender, age (in years), dummy variables for type of health plan product (eg, highdeductible plan), and dummy variables for each calendar year from 2005 to 2011. We included a random effect for each enrollee in the analysis. The analyses were limited to the relevant population eligible for the preventive service based on age and gender (Table 1). For example, the model for mammogram analyzes only included women aged over age 35 years in the model. We ran separate models for all enrollees, enrollees with and without a chronic illness, and enrollees younger than and older than 50 years.

Because this is an observational study, we had to address selection bias. Those who chose to enter the incentive program were different on both observed and unobserved patient traits. In our main analyses, we used random patient effects to try and address this bias. There was controversy about the use of fixed-effect versus random-effect models and therefore we use a fixed-effect model as a sensitivity analysis. While a fixed-effect model might be viewed as having greater internal validity, it might have less external generalizability. For dichotomous variables, identification using fixed-effect or conditional logistic models was restricted to the relatively small group of “switchers” whose outcome measure (ie, annual take-up of a preventive care service) changed between years.

In sensitivity analyses, we limited our study population to those continuously enrolled in the health plan over the 7-year period. The goal of this sensitivity analysis was to address the possibility that the changing composition of the population might bias the results because those who were not screened tended to drop out of the program at higher rates.

Standard errors are estimated using the Eicker-Huber- White sandwich estimator.

 All statistical analyses are conducted in STATA 12.0 (StataCorp, College Station, Texas).

Of the 4,186,047 unique individuals enrolled in the health plan between 2005 to 2011, 65.5% (2,742,268) of them enrolled in the incentive program at some point. Enrollment in the health plan and the incentive program steadily increased during the study period. In 2005 there were 1.83 and 1.26 million people in the health plan and incentive programs, respectively. Enrollees in the incentive program were younger and healthier (

). For example, among enrollees in the incentive program and those not in the incentive program, the fraction that was 71 years and older was 1.0% and 7.0%, respectively. Among enrollees in the incentive program and those not in the incentive program, the fraction with no chronic illness was 81.5% and 74.7%, respectively. The fraction of the eligible population that received a given preventive service was generally low in a given year (Table 1). For example, only 13.8% of women 35 years and older received a mammogram.

Becoming a member of the incentive program was associated with an increased likelihood of receiving all 10 of the preventive services e tracked (

). Among the 8 preventive care services associated with a financial incentive, the odds ratio (OR) for receipt of cholesterol testing was 2.70; glucose testing, 1.51; glaucoma screening, 1.34; dental check-up, 1.64; HIV testing, 3.47; PSA testing, 1.39; Pap test, 2.17; and mammogram, 1.90 (

 <.001 for all 8 services). Among the 2 preventive care services not associated with a financial incentive, the OR for receipt of colon cancer screening was 1.30 and bone density scan was 1.25 (both 

From our models, we estimated the marginal effect of the incentive program on receipt of the preventive services (

). The percentage-point improvement in receipt of preventive services for cholesterol testing was 8.9%; glucose testing, 4.7%; glaucoma screening, 3.9%; dental exam, 6.3%; HIV test, 2.6%; PSA test, 5.6%; Pap test, 7.0%; and mammogram, 3.1%. This translated into a relative percentage increase from 19.3% (PSA test) to 97.4% (HIV test) increase in preventive care use.

Patient financial incentives are a potential mechanism to improve health. In a South African health plan, a patient incentive program is associated with increased prevention.

Impact of a Patient Incentive Program on Receipt of Preventive Care

 To describe improvements in the design of pay-for-performance (P4P) programs that reflect the psychology of how people respond to incentives.

Investigation of the behavioral economics literature.

We describe 7 ways to improve P4P program design in terms of frequency and types of incentive payments. After discussing why P4P incentives can have unintended adverse consequences, we outline potential ways to mitigate these.

Although P4P incentives are increasingly popular, the healthcare literature shows that these have had minimal effect. Design improvements in P4P programs can enhance their effectiveness.

Lessons from behavioral economics may greatly enhance the design and effectiveness of P4P programs in healthcare, but future work is needed to demonstrate this empirically.

	Although pay-for-performance (P4P) incentives are widely used in healthcare, the published literature has shown that P4P incentives result at best in only modest improvements.

		The design of P4P programs generally does not reflect what is known about the psychology of how people respond to incentives, which may contribute to a lack of success. 

		This article discusses design improvements that can enhance the effectiveness of P4P programs and potentially mitigate the risk of adverse unintended consequences.

The use of pay-for-performance (P4P) incentives in healthcare is widespread. In the United States, P4P incentives are used by half of all commercial health maintenance organizations and are found in contracts with ambulatory physicians, hospitals, and nursing homes.

 Numerous state Medicaid programs also use P4P incentives, and the proposed Medicare hospital P4P program could include incentives totaling more than $3 billion annually.6 In the United Kingdom, almost 25% of family practitioner income is tied to P4P incentives.

 Despite the widespread application of P4P, much of the published literature

 on the effect of P4P has concluded that these incentives have resulted in small or no improvements.

There have been various interpretations of these results. Some investigators have raised concern that the premise underlying P4P is flawed.

 Others researchers believe that the magnitude of the incentives has been insufficient.

 Another potential reason for failure is that the current design of P4P programs does not reflect the psychology of how people respond to incentives. This is not surprising, as there has been scant literature

 on the effectiveness of specific design features of a P4P program to guide health plans or government sponsors. Program sponsors do what seems reasonable, and there is great variation in the design of programs.

 The behavioral economics literature reviewed herein can serve as a useful guide on how to structure provider incentives.

In this article, we discuss several design alternatives drawn from the behavioral economics literature that we believe could lead to greater provider response for the same amount of money devoted to a P4P program. We start by describing the importance of design features and a prototypical P4P program being used today. We then discuss several design features that could improve P4P programs. Last, we discuss some potential unintended consequences of P4P and how design changes could minimize these.

The Goal of P4P and Why Design Features Matter

The primary goal of most P4P programs is to improve healthcare quality, but incentives have been applied for other goals, including improving patient experience, implementing electronic prescribing, increasing patient safety, and decreasing utilization.

 In this article, we focus on an illustrative physician P4P program that focuses on healthcare quality, but we believe that our recommendations can extend to other care settings, types of providers, and their goals.

Most evaluations of P4P programs have measured change in performance based on quality metrics.

 However, in designing a P4P program, one has to consider the more proximal goal. To improve quality, a P4P program has to change the behavior of physicians and, more specifically, to increase the time and resources they allocate to quality improvement. The goal is to render a desired behavior (eg, taking the time to speak to patients about receiving a mammogram) front and center in the physician’s mind during a busy day.

We recognize that quality improvement, cost reduction, or any other goal of a P4P program often requires more than physician behavioral change, including system changes such as implementing an electronic medical record. Keeping physicians engaged is still critical, however. When P4P programs target physician groups, it is notable that the groups often create internal P4P programs for their individual physicians.

To make our recommendations more concrete, we begin by describing a prototypical physician P4P program that is designed to increase the number of women who receive a mammogram. After claims from a previous calendar year have been processed, an incentive is paid out in 2 steps. The health plan (1) determines the number of women who should have received a mammogram and how many did and (2) ascertains which physician is responsible for each patient’s care and calculates a physician-specific mammogram rate.

There is significant heterogeneity in how health plans structure some aspects of their P4P programs.

 Some health plans give incentives to physicians who meet a relative threshold (eg, the top 25% of physicians in terms of mammogram rate), and others use an absolute threshold (eg, physicians with a mammogram rate >75%). These top physicians can receive their incentive in various ways. They most commonly receive an increase in their reimbursement for each visit in the following year (eg, $106 vs $100) or a lump-sum incentive payment at the end of the year (eg, $1000).

We next discuss 7 potential design changes to these commonly used P4P programs (

). We recognize that some of the design changes conflict with each other. We see them as a menu of options to be considered and are not meant to be applied altogether.

Seven Design Features That Could Improve P4P Programs

A Series of Small Incentives Is Better Than 1 Large Incentive. 

Why do people go across town to save $10 on a clock radio but not to save $10 on a large-screen TV? In both cases, $10 is saved, but $10 is not always viewed the same. It is believed that an individual perceives the difference between $0 and $10 as being greater than the difference between $100 and $110.14 Similarly, 10 payments of $10 may be more motivating than a single $100 payment.

For P4P program design, it may be more psychologically motivating to provide a physician with smaller and more frequent incentive payments than a larger single lump-sum incentive payment. As an example, consider that a total of $1000 is available to give in incentives to the top physician performers. Applying this principle, a physician’s behavioral response is likely to be greater if the $1000 is divided into several payments (eg, 100 payments of $10 each) rather than paid as a single payment. Each $10 is perceived as a new $10 gain.

We recognize that a reward program with frequent payments is administratively more difficult. However, the more frequent incentive can be symbolic and still be effective.

 For example, every time a physician’s patient receives a mammogram, an e-mail could be sent: “Your patient Edith Jones received a mammogram on this date. We will credit you with $10 at the end of the quarter.” The combination of this frequent symbolic reward and a larger separate check at the end of the quarter might be doubly satisfying because the incentive is reinforced.

A Series of Tiered Absolute Thresholds Is Better Than 1 Absolute Threshold.

An individual’s motivation and effort when faced with a goal greatly depend on his or her baseline performance. Economists and psychologists have described this phenomenon as a “goal gradient.”

 If baseline performance is far away from goal performance, the individual exerts little effort because the goal is viewed as not immediately attainable. As baseline performance gets closer to goal performance, the individual exerts more effort to reach the goal (eg, 75% mammogram rate). However, the motivation to improve decreases significantly when the goal is achieved.

 A simple illustration of this phenomenon is a study

 of a coffee shop reward program in which the 10th coffee purchased was free. Participants in this experiment decreased the time between coffee purchases as they got closer to the free coffee.

Goal gradient theory has several applications in a physician P4P program. In aggregate, a greater behavioral response is likely if there was a series of quality performance thresholds to meet (eg, increasing amounts of money for achieving 50%, 60%, 70%, 80%, and 90% performance thresholds) rather than 1 (eg, a 75% performance threshold). In a single-threshold system, physicians who at baseline have low performance (eg, 25%) or high performance (eg, 80%) have little reason to devote more resources to attempting to improve quality.

Some researchers have proposed eliminating thresholds entirely and using a continuous gradient (eg, a physician receives $1000 × 76% performance = $760).

 Our opinion is that such a continuous gradient may be less effective than a series of thresholds because there is some benefit in having the clear bright-line goal of a threshold. However, this needs to be proven empirically.

Reducing the Lag Times Between Care and Receipt of Incentives Increases the Behavioral Response.

 Money received right away is perceived as different in value than money to be received in the future, even the near future.20 This steep initial discounting is much greater than would be expected by “rational” economic discounting and has been termed 

 In a typical P4P program, the time required to collect and validate the data, create physician scores, and make the payout often means that the incentive payment comes many months or even 1 or 2 years after the actual delivery of care. We believe that this long delay undermines the behavioral response of physicians. Ideally, there should be little or no lag time between the behavior being rewarded and the receipt of the incentive; otherwise, the competing incentives of a busy day might trump the P4P incentive. For example, a physician might have to make the following choice: “If I spend 5 more minutes with Mrs Jones discussing the advantages of a mammogram, I might receive an incentive next March versus if I skip this discussion, I will catch up on my schedule for the day.” The immediate incentive of not being behind in his or her schedule will likely trump the P4P incentive in the physician’s thinking. On the other hand, if the physician knows that the discussion might result in an immediate $10, the cost-benefit equation might change.

Although Withholds Have More of an Effect Than Bonuses, One Needs to Be Cognizant of the Negative Psychological Response.

 Previous research has found that individuals are more sensitive to incentives when they perceive that they are losing something as opposed to gaining something.

 Physicians in an experiment were asked to make a choice of treatment (surgery or radiation therapy) for a patient with cancer.

 In some cases, the choice was framed as a loss (probability of dying after surgery) or as a gain (probability of surviving after surgery). Physicians were more likely to choose the surgical option when the surgical risk was framed in terms of the probability of living rather than the probability of dying. The difference in the behavioral response for a choice framed as a loss rather than as a gain can be significant, almost 2-fold in magnitude.

This loss aversion has implications for structuring P4P incentives. Incentive payments can be structured as a withhold (a perceived loss in income) or as a bonus (a perceived gain in income). If the goal is to drive physicians to make changes that improve quality, withholding money (ie, framing the incentive as a possible loss) may lead to a greater behavioral response than framing the incentive as a “gain” in the form of a bonus, even if the same amount of money is at risk.

Although framing something as a loss rather than as a gain may result in a greater behavioral response, experiments have shown that doing so generally causes a significant negative psychological reaction and violates what the parties exposed to the incentive believe to be fair.

 Therefore, while the behavioral response is stronger with a withhold, this benefit is likely outweighed by the risk of angering physicians.

A possible way to take advantage of loss aversion without the negative reaction is through the use of a deposit contract. The P4P program would have 2 options. The first option would be for the physician to receive $1000 if his score increases by 10 percentage points from the previous year. However, the physician has a second option to enter a “deposit.” If the physician submits this $500 deposit and his or her mammogram score increases by 10 percentage points, the physician receives $2000 (instead of $1000). If the mammogram score is not increased by 10 percentage points, the physician loses the $500 deposit. Such a program has several advantages. First, it introduces loss aversion, as the physician will be motivated not to lose the $500. Second, it takes advantage of the fact that individuals are overly optimistic in predicting their success and that physicians electively enter the program. Third, it will make it easy for the health plan to identify physicians who are engaged in the program. The downside is that the health plan’s fraction of the incentive has increased from $1000 to $1500.

Reducing the Complexity of an Incentive Plan Increases the Behavioral Response. 

When given a choice of potential rewards, most people are risk averse; they will choose an option with absolute certainty over an option involving an uncertain but likely more valuable outcome. This principle of risk aversion is illustrated in a study

 whereby subjects were given a choice between a 1-week vacation that was certain or a 3-week vacation that they had a 50% chance of winning. Most subjects chose the 1-week vacation. Although the 50% chance of a 3-week vacation might be considered a more rational choice in strict economic terms because the expected return of such a choice is 1.5 weeks of vacation, most people will choose the sure thing because they perceive it to be a better choice than the possibility of getting nothing at all.

A related phenomenon is that individuals often cannot process complex decisions that are tied to a financial incentive. Current P4P incentive programs are complex for a physician. It is cognitively difficult to keep track of complex trade-offs such as the following: “If I spend 5 more minutes with Mrs Jones discussing the advantages of a mammogram, I could increase my overall mammogram rate to 

%, which might put me in the 75th percentile for my peer physicians and possibly lead to an incentive at the end of the year, versus spending 5 minutes with Mrs Jones might put me behind for my morning.”

 Because the P4P program decision is complex, while the concern of being late is clear and tangible, the physician is going to push off discussing the mammogram so that he or she is not late in the patient schedule.

The authors describe several simple changes that health plans can make in the design of pay-for-performance programs that may improve their effectiveness.

Ateev Mehrotra, MD

Articles

Paying for Telemedicine

Impact of a Patient Incentive Program on Receipt of Preventive Care

Using the Lessons of Behavioral Economics to Design More Effective Pay-for-Performance Programs