Incentives to Select a High Value Health Plan, Provider Network, or Provider
Consumer Financial Incentives: A Decision Guide for Purchasers
Tiered health plans offer provider lists sorted into tiers based on quality of care, cost, or some combination of these. Patients in these plans are given financial incentives in the form of lower out-of-pocket costs—copayments or premiums—to use providers in the preferred tier. Other providers may be used, but the patient must pay more to use them.
The first approaches that health plans took to arranging providers in tiers in the 1990s and in the early 2000s were based wholly or primarily on the providers' agreement to discounted fee schedules (or, less often, on measures of total annual cost). More recently, however, the trend has been for plans to incorporate quality into the equation, although the relative weight given to quality versus cost varies among programs. Another major change is the increasing use of measures of efficiency, such as the average expenditures required to treat an episode of bronchitis, rather than agreement to discounted fees or prices as the measure of "cost."
There is substantial information from the Medicare program and other sources that quality of care is not correlated with cost or with efficiency.44 This implies that, in most parts of the country, providers can be arrayed along two axes—quality and efficiency (go to Figure 1)—and that some providers will rate favorably on both dimensions whereas others will not. As a value-enhancing strategy, health plans could, for example, reduce out-of-pocket costs to patients who choose providers with both higher quality and lower costs (lower left hand corner of Figure 1).
The application of tiering can occur either at the point of care—when a patient has a clinical need—or annually during the open enrollment season (this is referred to as a premium-tiered plan). An example of the point-of-care approach is the plan offered by Boeing to its employees. When Boeing beneficiaries need hospital care, they must choose to use either a hospital compliant with Leapfrog measures or a non-compliant hospital.(f) If the patient uses a Leapfrog-compliant hospital, all costs are covered after the deductible. If a patient chooses a non-compliant hospital, the patient must pay 5 percent of the bill.45
United Healthcare (UHC) has a similar program, the UnitedHealth designation program.46 In this program, UHC designates physicians from 21 specialties as either high quality (one-star physicians) or both high quality and low cost (two-star physicians). Enrollees pay the lowest copayments for using designated two-star providers, the highest copayments for providers with no stars, and medium copayments for one-star providers. An innovative feature of this program, which may boost credibility with consumers, is that physicians cannot be designated as being low cost without also being high quality—that is, there is no "one-star" designation for being only a low cost physician.
The Buyers Health Care Action Group (BHCAG) is a Minnesota-based coalition of employers that began purchasing health care together in 1993. In 1997, BHCAG introduced the Patient Choice program, the Nation's first premium-tiered health plan.47 In Patient Choice, local provider networks are placed into performance tiers based on both quality of care and risk-adjusted total cost of care. BHCAG members' employees can choose among three tiers of networks, with decreasing performance ratings associated with increasing monthly premiums.
In 2006, patients in Tier 1 networks had the lowest premium; patients selecting Tier 2 networks paid the lowest premium plus 16 percent of total costs; and patients selecting Tier 3 networks paid the lowest premium plus 38 percent of total costs. This tiering strategy is accompanied by extensive resources for consumers to understand each network's performance rating, including reports of quality measures, such as which physicians have received Bridges to Excellence awards and the hospitals' performance ratings on Leapfrog and other measures.
Efforts to establish tiering approaches have been hampered by a paucity of adequate and reliable data on quality, creating some concern that tiering will be based primarily on cost or efficiency. For this reason, some tiering programs include only conditions for which data on quality are available; that is, there may be tiers among hospitals for patients seeking maternity services if data are available about the quality of maternity care, but no tiers for neurosurgical services in the same hospitals if there are no data on the quality of those services.
Another important limitation of some tiering programs is that costs are sometimes assessed in terms of expenditures per component of care, such as per hospital day or fee per visit, rather than an overall measure of the costs of treating a particular disorder, disease, or condition.
Question 5. How do tiering and other benefit design options fit into the framework of consumer financial incentives?
Premium-tiered health plans ask the consumer to choose a network of providers during the annual enrollment period—which for most consumers is not a time of immediate clinical need. Other approaches create incentives that are present whenever the consumer faces a specific clinical decision. Health plans with a high deductible in combination with an HSA create an incentive for patients to seek conservative options, such as a trial of bed rest before having magnetic resonance imaging (MRI) for low-back pain.
There are strengths and weaknesses in each approach. On the one hand, some might argue that it is more feasible to get the consumer's attention during an open enrollment period, when consumers expect to receive information about health benefits and prices. Patients with a chronic illness may be particularly motivated to seek information about providers' performance ratings. On the other hand, healthy consumers usually are not anticipating care utilization, and they may not take the necessary time to review data about quality and cost during the open enrollment period.
In addition, most plans and providers have good results in some clinical areas but need improvement in others. Incentives that target decisions during annual open enrollment, to some extent, assume that there are providers or networks that are preferable or less preferable for the entire spectrum of clinical medicine, which may not be the case.
In contrast, point-of-care approaches, whether involving differential copayments for higher performance providers or incentives to choose specific treatment options, are just the opposite. They highlight the link between clinical and financial outcomes and allow the consumer to choose their provider after they have become ill, which could permit them to better tailor their decision to the specific circumstances and needs they have. However, in some instances—especially when the consumer has insufficient information, education, or time—these approaches may be burdensome to the consumer.
Unfortunately, there is no evidence yet on the relative quality and cost implications of annual enrollment period incentive strategies versus point-of-care approaches.
Do incentives to reduce drug costs work?
One of the areas in which consumer financial incentives have been applied and studied most closely is the use of tiered copayments for drug coverage. The goal of tiering drug coverage is to direct patients to choose the lower cost option from among a group of drugs that insurers consider potentially equally effective.
Tiered drug benefits that offer lower copayments for drugs in lower tiers (e.g., generic drugs) generally have been successful in decreasing the use of higher cost drugs for many drug classes.48-52 However, tiering programs that focus solely on cost can have the effect of increasing out-of-pocket drug expenditures for a subset of patients, increasing the risk that they will stop buying the medications rather than switching tiers. Therefore, although consumer incentives designed to discourage overspending on drugs have reduced total drug costs, they sometimes keep patients from using drugs they need to prevent complications later.
In 2005, 74 percent of employer-sponsored drug coverage had three or four tiers, compared with 27 percent in 2000. In 2006, the average copayment for generic drug tiers was $10, as opposed to $22 for preferred brand-name tiers and $35 for non-preferred brand-name tiers.2 Increasing consumers' out-of-pocket costs for drugs is potentially a concern because, although there are few studies of the health impact of implementing tiered benefits or adding additional tiers,53-54 there are substantial data showing that cost-sharing applied with a broader brush—for instance, increasing all prescription copayments or dropping brand-name coverage—leads to underuse of drugs that are actually important to a patient's care.55-63
This sensitivity to medication cost occurs even in high-risk groups. Among seniors with a history of coronary artery disease or myocardial infarction, the use of statins—important cholesterol-lowering drugs—was 27.4 percent among those with drug coverage provided by their former employers, but only 4 percent among those without drug coverage.59 Additionally, worse health outcomes,64 such as uncontrolled hypertension, worsening heart disease,65 increases in emergency room visits,66 hospitalizations,65 nursing home admissions,66 serious adverse events,66 and declining self-reported health status,65,67 have been associated with greater cost-sharing.
An example in which consumer financial incentives have been applied more selectively is "reference pricing." This concept involves coverage of a preferred drug among a class of drugs—for example, statins—at low cost, and requires patients who choose other drugs within the class to pay the difference in price out of their pocket. The distinction between reference pricing and tiering generally is that, in most tiered formularies, if there is no generic drug within a class, then all the drugs are in a higher copayment tier.
A reference pricing strategy recognizes that, for patients who need a drug from a brand-name-only class, leaving those patients with no alternative but high copayments will reduce adherence to the drug regimen. Therefore, a preferred brand-name drug in that class should be chosen—usually after negotiating for a discount from the manufacturer—and offered at a low copayment as the reference priced drug.
In theory, reference pricing—and tiering generally—is most applicable in classes where several drugs are equally effective but vary in cost, especially if the health plan or government purchaser is able to negotiate a good price for one of the drugs in the class. A reference pricing strategy has been adopted in parts of Canada for selected drug classes without increasing hospitalization or decreasing medication use.53-54
In summary, simply increasing copayments for all drug classes or dropping brand-name drug coverage altogether will move some patients to lower-cost drugs. However, this will also increase the rate at which patients with chronic diseases stop taking important medications, which can lead to worse health and increased long-term costs. More selective financial incentives—such as reference pricing within specific classes of drugs in which there is an effective, low-cost option—appear to be more effective in shifting medication use without having a negative health impact.
Table 2 lists criteria sometimes considered in evaluating candidate performance measures. These criteria include the potential impact of improving performance on the measure, based on how common or severe a condition is or how much variation in performance there is among providers. Recently, it also has been proposed that the potential impact on disparities should be considered when choosing performance measures.68 However, practical considerations can also be important, such as whether a measure can be calculated in a transparent way from a reliable data source at a reasonable cost and whether it can be used in quality improvement initiatives or for consumer choice of provider.
Because a major goal of the use of incentives may be to reduce total costs, some employers and government programs use cost or efficiency measures—ideally in combination with quality measures—as a basis for designating preferred providers. The Agency for Healthcare Research and Quality recently released a summary of the state-of-the-science on efficiency measurement (Go to the box "Measuring Efficiency," below for more information).
Measuring Efficiency: The First Step in Incorporating Efficiency into Consumer Incentive Strategies
Efficiency measurements may have a place in a consumer incentive strategy, but the use of efficiency measurements without companion quality indicators may not benefit either the purchaser or the consumer.
McGlynn70 suggests that we measure efficiency as the resources required to create a health care product. Efficiency can be measured either for a specific service, like providing chemotherapy for a patient with colon cancer, or for specific outcomes, like preventing death from colon cancer. In the current environment, efficiency is more often assessed relative to services than to outcomes. The focus on cost per service raises the issue that services might not be comparable—for example, that chemotherapy is delivered at Clinic A in such a way that fewer complications ensue than at Clinic B, or that the doctors and nurses are more empathetic and supportive at Clinic A than at Clinic B.
The calculation of resources used can address either physical inputs—for example, the number of doctor visits and nursing hours spent over the course of chemotherapy at Clinic A—or the dollar value of the input. The approach focusing on physical inputs is called "technical efficiency," while the approach focusing on dollar value is referred to as "productive efficiency." Most existing commercial efficiency measurement software, known as "episode groupers," is used to measure productive efficiency for services—for example, the dollars required to complete the chemotherapy course at Clinic A.
Most efficiency measurement systems are proprietary and are available from a small number of vendors specializing in this area. They generally report observed-to-expected cost ratios, or similar ratios, after some adjustment for severity of illness or case mix, either for episodes of care or for a population of patients over a period of time. Assessing the cost of chemotherapy for colon cancer at Clinic A is an example of the episode-based approach. A population approach involves assessing the total expenditures on a group of patients, usually over the course of a year, correcting for the illnesses they experience during that time period. While the population approach is in some ways analogous to summing episode expenditures over a year, it also incorporates the frequency of episodes in the population, correcting for the diseases that its members have.
There are some important caveats to the use of these commercial software packages to assess efficiency. First, none of the measurements generated by these packages have been carefully validated in the way that drugs, for instance, are evaluated before they are approved for use. In general, existing commercial systems are not explicitly designed to account for quality of care or outcomes. In fact, the Ambulatory Care Quality Alliance (AQA), among others, suggests that we use the term "cost" to refer to all measures that lack a quality component.
Either definitional framework for cost and efficiency leads to a similar recommendation: purchasers should consider presenting efficiency measurements together with quality data—otherwise, consumers may conclude that the program addresses only cost control on behalf of purchasers, rather than value and other factors that reflect the consumers' own interests. In addition, if providers offer low cost but poor quality, they may score well on efficiency measurements in the short term but cause greater long-term costs.
For all of these reasons, the use of efficiency measurements without simultaneous measurement and reporting of quality of care in the same clinical areas may not actually benefit purchasers. A better approach, for example, would be to measure the costs of chemotherapy for colon cancer at Clinics A and B while also calculating survival rates and surveying patients about their experiences with the respective clinics.
A second measurement issue related to tiering is that providers who have high quality and/or efficiency on one service may perform poorly on another. In most existing tiered plans, the tiering method puts providers in the same tier for all categories of care; that is, Hospital A is simply given "preferred" status rather than being identified as "preferred" for orthopedic surgery but not for cardiac care. With this approach, quality criteria should cover as broad a spectrum of care as possible. Alternatively, some tiered plans target only the categories of care for which measurements of clinical quality are available.
Question 7. What do consumers want to know about the quality and cost measures used to create tiers?
There are no studies that define what consumers want to know about, or even whether consumers understand, the basic economic concept behind tiering or any other incentive program. It is clear, however, that consumers respond to price differences among providers or health plans. Financial incentives most affect those people who are in poor health,71 use more medications,57 have more chronic health problems,57-58,72-74 and have a lower income.54,74-76
There are limits on how much complexity consumers can understand in an incentive plan. For instance, tiered pharmaceutical formularies are one of the simplest and most common forms of tiering, yet only 29 percent of patients in such plans know their usual copayments, and only 43 percent know whether there are limits on the total number of medications or the total medication costs that their plan covers.77 This low degree of comprehension is consistent with findings reported earlier, which confirmed the difficulty that patients have with understanding complex benefit designs such as those that have been offered by some Medicare HMOs.78-80
Among commercially insured patients, knowledge about benefit coverage—outpatient, inpatient, mental health, emergency room, and out-of-area services—is also poor.81 While many enrollees in commercial plans know about coverage for hospitalizations (90 percent) and doctor visits (80 percent), fewer know whether their insurance covers mental health (54%) or treatment for alcohol abuse (43 percent).82 Those who are least likely to be knowledgeable about their plans are those who are less educated,77,83-84 have a lower income,83 are not already enrolled in the program,80 are older,85 and are members of a minority group.84-86
The implications of these findings vary depending on the approach to consumer incentives used. Consumers' lack of knowledge is of particular concern, for instance, when point-of-care tiering or tiered formularies are used to influence decisions that occur frequently and with time constraints, such as choosing a medication to be prescribed during a doctor's office visit. To make an optimal choice of medication often requires the patient and the provider not only to have access to accurate information quickly, but also to be able to process complex information about the benefits and costs of various treatment options within a short timeframe.
Many studies show that patients84,86 and providers alike currently lack timely access to these data.87 For example, 54 percent of patients who self-reported being in a tiered plan said they are never or only sometimes aware of their out-of-pocket drug costs at the time a drug is prescribed for them.86 In contrast, there may be better opportunities for information gathering and for careful consideration when decisions must be made less frequently, such as decisions about selecting a doctor or the annual choice of a health plan, or when the decision-making process—while it may have a deadline—can be conducted over weeks or months.
Many of the more vulnerable patients, such as those who are elderly or very sick, may not be the primary decisionmaker in choosing among their health care options. Among Medicare beneficiaries, only 68 percent made their own health insurance decisions; 24 percent received help, and 9 percent had a proxy.78 There is some indication that this lesser degree of involvement may be because health decisions are increasingly complicated, requiring help even when consumers may be capable of making decisions by themselves in other areas of life.88
There are studies in the Medicare literature showing that people want and use many different sources of information about health care programs,83 and they use information in multiple formats before they enroll in a plan. In one study, seniors eligible for Medicare wanted different sources of information, including addresses, phone numbers, a Web site, and postcards about seminars; and they wanted this information at different sites—Social Security offices, senior centers, doctor's offices, pharmacies, libraries, churches, and grocery stores.89
In practice, consumers want and actually use information from multiple sources. Medicare beneficiaries used, on average, three sources of information: advertisements (55 percent); newspaper or magazine stories (47 percent); friends or family (44 percent); experiences with HMOs (34 percent); and television (33 percent).83 With respect to their decision about enrolling in Medicare Part D and choosing a plan, 27 percent of beneficiaries said they talked to a pharmacist, 26 percent to a physician, 17 percent to an insurance agent, and 17 percent to a counselor.90
It is important not only to measure a person's knowledge about information sources but also to assess what information is actually used. With respect to Medicare Part D, 64 percent of beneficiaries knew about the toll-free number, and 62 percent knew about the Web site, but only 12 percent had called the hotline, and only 9 percent had looked for information on the Web site.90 There is less research about how seniors use these information sources after enrollment—for example, research that would help determine how to give them the information they need about drug benefits or keep them up to date with changes.
The source of information is also important. Medicare beneficiaries have different levels of trust, depending on the information source. For example, the Social Security Administration is a source most seniors would trust.78 Among those who are 64 years old and younger, 43 percent say they have limited knowledge about where to find information.89
Question 8. In the special case of incentives for choosing among treatment options, what information or decisionmaking tools, if any, should be offered to consumers as accompaniments to financial incentives?
As most patients initially lack relevant medical knowledge, they cannot be expected to make good medical decisions without help. The introduction of incentives, in itself, will not obviate the need to help patients facing difficult clinical decisions understand their situation and their options.
In those areas where good decisionmaking aids are available for patients, they have been very beneficial. However, for most clinical conditions, we do not yet have such decisionmaking aids.
Patients who have access to decisionmaking aids have greater knowledge about their options, have more realistic expectations about the impact of treatment, and are more active participants in making decisions about their care. Interestingly, informed patients are more conservative than their doctors and are about 20-25 percent less likely than their physicians to choose the most aggressive surgical option for a particular condition.91 Among the clinical areas in which decisionmaking aids are available are prenatal screening, screening for breast and prostate cancer, management of actual symptoms from gynecologic and prostate disease, management of back pain, and treatment options for coronary artery disease and several cancers.
Decision aids are available from a variety of medical specialty societies and commercial vendors, although not all have been studied equally well or are of equivalent quality. Decision aids vary in the content they include, but generally they explain the underlying condition and its prognosis, describe the treatment options available, and discuss the effectiveness of each option. Many different media are used, including written materials, audio or videotapes, interactive computer programs or Web sites, and counseling or educational sessions—but there is little research to demonstrate the relative strengths of one medium versus another.
In general, information from a variety of media and repetition are preferable because the topics are usually complex, and patients may not be prepared to hear a message clearly the first time it is presented. Narratives about the results of different options, especially when presented by people whose background is similar to the patient's, may be particularly effective in helping patients understand the differences among treatment options.92
Decision aids are best when used in the larger context of the patient's involvement in medical decisionmaking. Five elements are essential in enabling patients to realize this participation:93
- The patient is aware of and understands all the treatment options available to him or her.
- The patient understands the risks and benefits of the available options.
- The patient realizes that he or she has a right to play an active role in decisionmaking.
- The physician encourages the patient's participation in choosing treatments.
- The patient is given time to consider the decision.
These elements emphasize two factors: first, getting the patient and the physician to recognize that a variety of options are available—patients initially may not know that they do have options, and physicians may tend to focus on the option that they believe the patient should choose; and second, acknowledging that the patient has a role in the decisionmaking process and needs time to make a decision.
This discussion of decisionmaking aids focuses on situations in which there is a major decision to be made at a specific point in time—for example, "Should I have surgery or radiation therapy for my prostate cancer?" A conceptually related issue, but one with a different timeline, is involvement of the patient in the management of his or her chronic disease. In that situation, given the right support, patients can take over decisionmaking about managing their condition from their health care providers—for example, a patient with heart failure could adjust her medication doses based on instructions from her physician without necessarily having to have an office visit.
Some patients' self-management programs have produced higher quality outcomes at lower costs than conventional models of care.94 Several elements are essential to enabling patients to manage their participation in such a program (go to Table 3). These include a self-management toolkit that teaches the patient how to recognize symptoms that suggest a change in treatment is needed, how to make the change and assess its effect, and then how to make adjustments as needed, even without consulting the physician. This approach also presumes that patients are taught simple testing procedures, such as checking their blood sugar levels, and already have the prescriptions needed to change their medicines or dosages.94
Asthma offers a classic example of the opportunities for patients' self-management. With the appropriate instructions and prescriptions, a patient with asthma could learn to recognize increasing shortness of breath as a potential reason to change his inhaler regimen. He could then perform a simple test with a reusable, hand-held peak flow meter costing about $20 to see if his airflow is reduced. If it is, he could take extra doses of one of his medications and test again later to see if his peak flow has improved. If not, he could add an inhaler to reduce lung inflammation. This self-management can be done safely and often without involving his physician for every episode of increased shortness of breath.
As with decision aids, self-management tools are available from a variety of medical specialty societies and commercial vendors. Vendors of disease management systems and products often incorporate a self-management program into the overall disease management method. Again, not all of these programs—whether for self-management or disease management—are of uniform quality or have been tested thoroughly, so it is important to incorporate a plan for evaluating the impact of the program into the initial arrangements for implementing them in the patient's care regimen.
In general, there has been less use of both decision aids and self-management tools than would be optimal. Although there is no research addressing the impact of incentive programs on use of decision aids or self-management, it is logical to anticipate that incentives may focus patients' and providers' attention on the fact that there are important choices to be made, so that patients and their providers may be more willing to use decision aids and self-management tools.
f The Leapfrog Group is a voluntary employer association aimed at mobilizing employer purchasing power to recognize and reward improvements in health care safety, quality, and customer value. The Group's membership includes representatives from many of the Nation's largest corporations and public agencies that buy health benefits on behalf of their employees, dependents, and retirees. For more information on the Leapfrog Group and Leapfrog measures, go to http://www.leapfroggroup.org. Accessed on October 10, 2007.