Profiling and Pay-for-Performance: Is It Quality or Cost That Matters?

Managed care has a new angle to contain physician costs: physician profiling to select or exclude clinicians from tiered networks. It was no surprise to me last month when I received 5 insurance spreadsheets detailing my practice profile. In essence, I was given a breakdown of treatment costs for my patients (compared with those of my peers) in the hope that I would change my management and referral habits.Of course, I had known for some time that claims data were being mined. But what came as a shock was that most of the expenses associated with the episode of treatment group (ETG) methodology applied to my practice were attributed to charges that were not under my control. For example, an MRI scan of the brain-which cost the insurer an average of $391.11 per study on my patients-was 186.78% more expensive than the average $136.38 per study for patients going to other Metro New York-area neurologists. Data like this are now being studied by managed care companies to determine which providers are least cost-effective and to discourage patients from going to them. In some cases, the doctors to whom they are redirected are dubbed as providers of "higher quality of care." As this strategy is tested in various parts of the country, many physicians are wondering just how valid is this information, and how such ratings will affect their patients.By developing tiered networks-insurance products that select out "efficient" providers-health care insurers are hoping to create panels of physicians who will help them control costs. It's an idea, certainly, that worked for pharmacy benefits. Differential copayments steered beneficiaries toward lower-cost drugs. Insurers quickly applied the strategy to hospitals, with less clear results. Most recently, they have set their eyes on the physician market, with new programs poised to spread across the nation. For some payers, profiling is part of a concerted shift in health care toward pay-for-performance (P4P).Tiering of any kind, however, is built on erroneous assumptions-that all products are the same and that differences in costs are attributable to differences in efficiency. For example, in Aetna's Aexcel Specialist Network, which includes neurologists and neurosurgeons, physicians are stratified based on database information that identifies episode costs. Aetna decides who can join the network by measuring a minimum of 20 "episodes of care" over a 2-year period. An episode of care includes medical treatment for outpatient and inpatient care as well as related costs. Aexcel, which expanded into 20 US markets as of the beginning of this year, focuses on specialists because it has determined that specialty care is the greatest cost driver.Average Cost Per EpisodeIn my case, as a neurologist with a high proportion of referrals from orthopedic surgeons, the largest of my "treatment pattern comparisons" groups was for "joint degeneration localized, without surgery." For this category, largely composed of neck and back pain patients, Aetna determined that the average cost per episode for my patients was $1485.09, compared with $826.54 for all designated providers in my urban market. In other words, my "efficiency index" for this category was much worse: 1.42 compared with 0.76.This was attributed to increased use of inpatient visits (even though I do only outpatient work) and increased use of ambulatory facilities, other physicians, radiology, and laboratories. Where was all this coming from? Under this new math, if I refer a patient for imaging and those charges per service exceed the average, I am held responsible. If my patients' physical therapists bill for more "therapeutic exercise" codes per episode (in my case, 6 per episode) and do so at a higher cost per service (in my case, 120.55% higher), I am doubly "inefficient." It turns out that the biggest "total dollar impact" to Aetna was in this line item, with MRI facility bills close behind.None of Aetna's data were wrong. So why was the resulting profile so inaccurate and misleading? Consider, for starters, the size of the numbers that composed the snapshot. The highest number of episodes in any of my ETGs was 29 compared with 828 aggregate episodes (from an unspecified total universe of ETGs) in Aetna's pool of designated providers, and 9 of my 14 ETGs had only 1 or 2 episodes each. My total efficiency score for general neurologic conditions was 0.70-much better than the mean-but included only 60 episodes of care. Yet, it made my total average-1.01-look average. Imagine if other health care decisions-say, evaluating the effectiveness of the latest drug for stroke-were to ride on such sketchy confidence figures.And what of the many hidden issues that do not get factored into a profile but most assuredly affect ranking? Providers who care for the sickest patients, for example, are not going to profile well because those patients require care at "inefficient" levels. Patients suffering from many concurrent illnesses also will skew results for the physicians who care for them. Finally, while Aetna says it considers quality, there was no evidence that it incorporated clinical guidelines or outcome measures in the data that I reviewed.Neurosurgeon James R. Bean, MD, is concerned that most payers are focusing on tiering without a sound measurement of quality. The American Association of Neurological Surgeons treasurer is also president of a 350-member multispecialty Independent Practice Association. His concern as a neurosurgeon is that such profiles miss the most important parameter of all: how well a patient does after surgery.He added, "The system fails to differentiate between patients who have already had conservative care and have been pre-screened for surgery and those presenting at onset of their conditions." For example, he pointed out, a back surgeon who is referred patients who have already failed physical therapy or epidural steroid injections will be penalized for having a higher relative surgical rate than his or her peers. Bean summed up, "Profiling is an artificial administrative measure that is misleading or, frankly, exclusionary."Methods QuestionedWilliam Thomas, PhD, has been studying quality and economic profiling for some time. A health policy analyst working on a study about tiered networks with funding from the Robert Wood Johnson Foundation, Thomas is concerned about the dearth of research on important methodologic issues that can influence the accuracy of physicians' economic profiles. He's also uneasy about some of the methods being used to judge physicians' quality-of-care performance. "Reasonably strong evidence has been around for a good deal of time that the popular risk adjusted outcome measures-mortality, readmissions, and complications-are often not valid indicators of quality performance. Yet few employers and few health plans know that they're invalid," he said."This issue and others, including attribution methodology," explained Thomas, "need to be researched at greater depth." In some cases, an ETG is assigned to a physician responsible for the majority of professional fees associated with an episode and not necessarily the majority of prescribing or ancillary costs. Under this system, a surgeon may be given the attribution for an ETG even though a primary care doctor is managing the case, including referrals. An even bigger problem, said Thomas, is that conclusions are being drawn based on small sample sizes.Health plans involved in profiling do not know what sample size should be used because researchers haven't yet given them any guidance, Thomas explained. "As a result," he added, "they choose a number and hope it works." The problem they run into, he said, is that choosing too low a minimum number, like 10 (which incidentally, Aetna said it used with me), leads to unreliable tier assignments. He added, "Large percentages of physicians are shifted from one tier to another each year, and that's not just a problem for physicians, it's a problem for health plan members, too, and therefore for the plan itself."Aetna, of course, is not the only insurer experimenting with such plans. In the past few years, PacifiCare, Premera Blue Cross, and UnitedHealthcare all have rolled out tiered networks. UnitedHealth Premium's Web site tells members "UnitedHealthcare is providing you with important information on physicians, so that you are empowered to make more informed choices on where to receive care-you can rest easy knowing that you are receiving care from a physician that has met UnitedHealth Premium's strict quality standards."Negative IncentiveTo further push consumers into choosing physicians in tiered networks, insurance companies hit those who go to physicians outside the network with higher co-pays and deductibles. Aetna's Web site boasts that compliant members "benefit from expected lower costs in medical care as well as the value of having information to make better health care decisions." Since when, however, has lower-cost care proved to be better care? Enlightened consumers might perceive, instead, that a provider has incentives to spend less money in finding out what is wrong with the patient. Misdiagnoses lead to less effective treatments-and that's hardly better health care.Physician groups such as the Medical Group Management Association, the AMA, and the American Academy of Family Physicians (AAFP) have expressed strong doubts about current insurer performance measures. AMA trustee John H. Armstrong, MD, said, "The AMA does not regard tiered networks as quality improvement programs, and we certainly don't see how they help patients. These networks merely represent that latest insurer effort to economically credential physicians and cut costs."The AMA is very concerned, said Armstrong, that such initiatives put a barrier between patients and their doctors and restrict access to care. "How can 20% of physicians care for 100% of all patients?" He believes that there is no correlation between the tiered network efficiency ratings and clinical outcomes. "Any ratings must focus on clinically relevant performance measures that measure what matters to improve patient care," he noted. Once such performance measures are defined, the data must be collected and analyzed. "When it comes to data collection, the source of that data matters," he emphasized. "Claims data, while accessible, do not reliably capture the clinically relevant elements necessary to measure performance. Further, to collect this volume of data and facilitate data analysis, we need a robust information technology infrastructure, which is simply not yet in place."Some believe that physicians need to help create a common set of measures. Bruce Bagley, MD, medical director for quality improvement at the AAFP, conceded that while the programs are cost-motivated, a few, like United's, take an important step toward assessing and possibly motivating better performance. "Some see these plans as a ploy to steer patients to doctors who spend less money on patients," Bagley reflected, "but we need to move from judging physicians based on pedigree to making decisions based on actual clinical performance." Once such measures become available to primary care physicians who are making referrals, Bagley predicts that specialists will be chosen on a whole new set of criteria.Those with the most potential influence to choose are patients themselves. Gail Wilder, MD, who has been a medical director for 8 years, including 6 years with Aetna, said, "While conceptually, tiered networks are attractive to managed care organizations, they are very hard to implement because patients want to go to their own physicians." Wilder believes that it is useful to incorporate quality standards in physician profiles, but inappropriate and misleading to use rankings based on cost-efficiency alone.She pointed out that certain sought-after physicians who have a level of expertise and experience in their field above and beyond the ordinary may already have negotiated special fee contracts with managed care companies as a condition for agreeing to join the network. Said Wilder, "It would be ironic if these same physicians are later tagged as inefficient because of higher cost and, as a consequence, deselected out of a special tier." Moreover, what are the ethical implications of offering care to people based solely on cost and not quality? She reflected, "Persons may be unable to access the most proficient providers because they are the most costly."What's To Come Of It?For now, specialists feel it's a rigged game as to who is selected for referral and why. Daniel B. Hier, MD, MBA, a University of Illinois neurologist, said he was not offered an opportunity to participate in his United Healthcare profile. "I doubt that any quality data was reviewed," said Hier, who questions the methodology. "It is very hard to compare neurologists unless you analyze performance at the level of a specific diagnosis with access to detailed clinical data-and even then it is quite difficult."In P4P programs, insurers review claims to see whether doctors meet certain criteria such as prescribing medications that have been shown to reduce risk of illness or ordering screening tests such as mammograms. Right now, physicians may decline to participate in P4P.AMA sources said, however, that in another pilot by United Healthcare, physicians who have been selected for both quality and cost efficiency in Atlanta, Cleveland, and Chicago will be offered higher fee schedules. This essentially converts their current tiered program into a P4P program. And if profiling-in which physicians' participation is involuntary-is any indication of the future of managed care, P4P is not likely to remain a choice much longer. Then, we may be looking at an environment that will encourage practitioners to prescribe care regimens based not on what is best for patients, but on what will boost a tier ranking.In that scenario, only the payer wins.ORLY AVITZUR, MD, MBA, is a practicing neurologist in Tarrytown, NY.Part I: Tiered NetworksIn Part II, Bruce Sigsbee, MD, will discuss current pay-for-performance initiatives and their role in the future of medical practice.