Radiology Income Distribution: From Salary to Independent Practice to Megapractice
A CPA offers his perspective on the past and future of radiologist compensation, based on 30 years on the business side of radiology Hospital-based physicians (radiologists, anesthesiologists, and pathologists) were either compensated as salaried employees or were members of unincorporated entities paid by the hospital well into the 1960s. Historically, the role of radiologists in the delivery of care included departmental administrative oversight (Medicare Part A hospital services) and direct patient care (Part B physician services). Hospitals required completion of time reports that distinguished between the two roles in order to receive compensation for Part A and Part B costs that came out of different insurance-trust or risk pools. 1960s and 1970s Two events hastened the transition to completely independent practice. First, legislation was passed that allowed licensed professionals to incorporate, thus availing themselves of the significant tax benefits for employees of these entities. Second, the Health Care Financing Administration (HCFA) approved separation of fees for the technical and professional components of radiography for radiologists at Massachusetts General Hospital, Boston. The first transition changed mindsets. Radiologists began to negotiate formula-based compensation arrangements, similar to a separate fee, in which monthly payments were made to their corporate entities. They determined their own compensation/benefit structures and began to view themselves as independent of the hospitals, much as office-based primary care and specialist physicians were. They still depended on the hospital as the income source, however. This changed with the advent of separate billing, in which a fee per procedure was established for direct patient care The initial responses from hospitals were mixed. Some immediately required the physicians to transition to collecting a separate fee because they believed that it was possible to gain payor income, through arbitrage, that exceeded the fees paid to radiologists. Most continued to pay an administrative stipend, believing that the payors had carefully structured the separate fees to cover only direct patient care. Those hospitals expected the physicians to help govern the departments, and felt that they needed to pay for this time commitment. The remaining hospitals initially resisted the change because they felt that a hospital-based income system gave them control over the physicians’ coverage responsibilities. The fee-for-service system was so profitable, however, that groups tied to the hospital system could not compete in recruiting new candidates. Part A payments disappeared when hospitals determined that there was no earmarking of independent fees for the hospital-based systems. HCFA originally set the reading fee at 40% of global payments to freestanding imaging centers. In all cases, this was well above the aggregate amounts paid under the hospital formulae, and this triggered dramatic growth in the individual incomes of radiologists nationwide. By the end of the 1970s, the average compensation package of a senior practice member was in the top 15% of all specialists. It took about two years for payors to realize that these fees caused significant increases in cost for interpretations. They made numerous modifications in both the method and extent of payments; however, during this transition stage, new technologies also increased caseloads. Payors did not understand the caseload that a full-time radiologist could handle. There is more than one generalization to be made about how groups structured their compensation plans after the transition phase ended in the mid-1970s. First, groups covering health care systems with relatively unfavorable payor mixes tended to have higher caseloads per FTE, but lower compensation packages (wages plus benefits). These practices still experienced gains in relative income, but not enough to change staffing patterns. Second, groups covering systems with favorable payor mixes experienced very large increases in income. Initially, they kept staffing levels the same. This was followed by the decision to increase time off; radiologists have the most generous vacation packages in health care. This made productivity measurement difficult because the definition of FTE was a moving target. Third, income distribution was predominantly equal among senior members, regardless of the payor mix. 1980s, 1990s, and Beyond The alarming rise in health-insurance premiums prompted examination of a national health care system, a divisive issue within the industry. The US market’s alternative was competition between managed care programs, in which physicians would be organized into loosely or formally structured staff models working within health care delivery systems such as PPOs, PHOs, and HMOs. This is an arena that radiologists hoped to avoid because ordering physicians’ perceptions of the economic value of radiology services were vastly different from those of insurance companies. The physicians who referred patients perceived radiologists as sophisticated support, not necessarily as peers. The purchasers of health care services sought to fix (capitate) the cost of services, leaving it up to the managed care plans to determine how income was distributed. This was a battle that radiologists could not win. By the end of the 1990s, most of the payment systems reverted back to fee-for-service payment because of general rejection of capitation by physicians and payors. Running parallel to managed care initiatives (and, some speculate, having them as a cause) was the ownership, by referring physicians, of large-scale imaging centers. Busy imaging centers owned by referring physicians were extremely profitable, but examination volumes had much to do with self-referral. The radiologists were generally compensated by the imaging centers at rates less favorable than those seen in payor-based markets. The proliferation of these centers also depleted traditional outpatient caseloads. Eventually, they were legislated out of existence by the Stark self-referral laws, resulting in the sale of existing centers to hospitals, radiologists, or joint ventures involving both parties. Many of these ventures strengthened both the radiologists’ incomes and their relationships with hospitals. The Stark laws, however, had safe-harbor provisions that enabled practices to image their own patients within their office campuses. Practically speaking, the economic factors involved required practices to be large in order to operate cost effectively. Many of the first such practices focused on CT and MRI, because of high profit margins, and then on ultrasound, because equipment costs were low. Later generations of on-site imaging centers involved resource leases. Radiologists, reluctantly, entered these arrangements to keep their caseloads. Hospital-based practices saw their volumes declining because of these office-based initiatives. Income problems were further compounded by the selective office imaging of those patients with the best insurance coverage; the self-paying or low-paying patients were sent to the hospital. The other impact of these initiatives was the coverage inefficiency of having multiple sites. Those radiology practices most affected by office ventures set up reading facilities under their control to limit or eliminate the need for a physical presence. This mirrored the efforts of the off-site reading services that had been operating since teleradiology systems first came on the market. Federal Program Changes HCFA embarked on a series of cost-cutting measures, the first being making payments to hospital systems based on DRGs. This profoundly changed the way hospitals operated by linking payments to a patient’s condition, thus forcing hospitals to match nationwide guidelines for addressing patients’ medical issues and to meet discharge timetables. This also affected inpatient testing because the DRG level dictated the extent to which the hospital could afford to test a patient. The result was the shifting of examinations to the outpatient arena whenever possible. The initiative on the physician side began in the mid-1980s with a massive research project to quantify practice-expense patterns that would be mapped to reimbursement; ultimately, this would be known as the RBRVS. The ACR moved quickly to help develop RVUs for radiology because of the threat that DRG-based reimbursement would contain payments attributed to the work of radiologists, anesthesiologists, and pathologists (because of their unique connection to the services of their hospital departments).1 The ACR used survey input from its members to estimate work magnitude for 45 procedures (probably chosen based on volumes and organ systems) and to gather volume and charge information for approximately 740 procedures.1 Both work-magnitude and charge information produced median statistics for each CPT® code. It was critical for work-magnitude baselines to be established because it was unlikely that national pricing was market efficient. Insurance companies caused this situation by using usual-and-customary methodology to determine payment rates. Fees were generally set at the 75th percentile of the prior year’s regional fee-schedule population for each CPT code. This system fostered fee increases for those procedures done with the greatest frequency, rather than pricing examinations based on the relative intensity of clinical effort. The numerical values were driven by the benchmark procedure code, 74400 (intravenous pyelography).1 The 1989 implementation of RBRVS for radiology was delayed until April 1, with a separate conversion factor, because the RVUs had not been mapped to the primary/specialty-care resource database. Ultimately, HCFA used the accepted relativity of the ACR units, merging them with the entire system at 27% (the single-view chest radiograph became 0.27). Practices began to experiment with use of work units as a productivity measure. Statistically, the total and work RVUs will offer similar data because the three components have the same proportions for each CPT code (rounding causes modest differences). The work unit can be used to assign a time value per CPT code by converting the single-view chest radiograph’s value to 1, then making an assumption concerning the average time that it takes a radiologist to review, dictate, and sign off on this examination. It not a perfect measure, but does offer a baseline indication of the weekly/monthly time needed to handle a given caseload. Appropriateness Guidelines The ACR’s managed care initiative was a combined development of appropriateness guidelines and voluntary participation in a program to catalog case findings according to clinical relevance to the patient’s condition. These were intended to emphasize the specialty’s ability to help control utilization and create an ordering-pattern report card on referring physicians. The ACR had conflicts with implementation, however. All practices derived income from examinations that should not have been ordered; it was speculated that this ranged from 15% to 40% of income. Few, if any, practices introduced programs at their coverage sites, and efforts to catalog findings proved difficult, administratively. There was genuine concern that questioning their ordering patterns would alienate the referring doctors, causing them to send their patients elsewhere. The insurance companies picked up on the concept of appropriateness, and it evolved into an industry. Initially, the insurers sought to introduce gatekeeping, but had mixed results. Numerous radiology benefit management organizations evolved, contracting with insurance companies to manage imaging referrals. They either use ACR guidelines or develop their own algorithms, which are designed to limit inappropriate ordering and measure the performance of the referring physicians. Much of the focus is on high-end procedures because of their cost. Medical schools began to develop electronic order-entry systems that enabled all referring physicians in a network to order examinations within the RIS. These systems include guideline information to assist ordering physicians with their decisions. As they advanced, the degree of control exercised by the programs gained flexibility. This ranged from simply executing the order to preventing it because the requested examinations were incompatible with appropriateness guidelines. The key factor that limits their market-wide use is fee-for-service reimbursement, since organizations have an incentive to limit overutilization when payments for care are not driven by procedural volume. These systems are now commercially available as front-end RIS modules. PACS and Digital Imaging The most important change in the imaging marketplace has been the evolution of the digital record. PACS has gone from being a high-tech luxury, in the early 1990s, to being a necessity available to all imaging facilities. There are no technological or financial impediments to complete portability of all images. Health care delivery systems are moving toward immediate access to images, across the entire enterprise, for ordering physicians. Those specialists heavily reliant on imaging have become more proficient in interpreting cases because of ready access. Off-site reading services began to evolve very early in the PACS product cycle and have expanded to meet the needs of health care systems of all sizes. Small practices have benefited from reading services that enable them to comply with 24/7 contract obligations. Many of these groups, however, have been put at a financial disadvantage due to the cost of the service. This has resulted in negotiated coverage fees paid by hospitals so that the groups can compete with the income systems of the larger groups. The days of small and intermediate-sized groups are waning, though, as groups consolidate into subspecialized megagroups able to meet the needs of any health care delivery system with both on-site presence and off-site reading services. At the close of the 20th century, in summary, revenue systems for radiologists continued to place them in the highest income quartile, usually with generous time-off packages. Group sizes have expanded, combined with subspecialization supported by the ability to triage cases regardless of the members’ reading sites. Initiatives to lower reading fees were countered by dramatic increases in caseloads for MRI and CT, paid at more favorable rates. The more forward-thinking groups sought to expand into freestanding imaging once the Stark amendment forced referring physicians into divestiture. There is a direct correlation between the incomes of radiologists and the size of their nonhospital caseloads. The Future of Radiology What follow are some speculations on the shape of future clinical information delivery and income systems for radiologists. Many of the components of this scenario are already in place, on smaller scales. The trigger mechanism that would push the market in another direction would be aggressive reduction of fees by insurance companies, which would force practices to alter the traditional method of delivering services. The current fee levels of dominant commercial carriers offer little incentive for operational change, but what if all payors reimbursed at the Medicare level—or below it? It is likely that payors would be willing to compensate entrepreneurial physicians highly if the net effect were an overall reduction in cost. They prefer to have fewer radiologists make exceptional incomes from high-volume operations that also improve clinical information.The entire concept underlying the reading center is to maximize productivity by strategically organizing caseloads by specialty and by removing all distractions. The predicted annual volumes shown in Table 2 might seem unrealistic, but there is anecdotal evidence that these volume targets can be met, even in a film-based environment, where physician extenders have been used to organize cases by difficulty and to prehang the films.Reading centers will need to consider an incentive-oriented system for compensating readers based on work RVU volume (Table 3). We only have current RVU guidelines to use; it must be strongly emphasized that current RVU values have flaws as productivity tools. This is a hypothetical caseload distribution for 20 million examinations interpreted by 364 radiologists. It is, at best, a speculation on how a homogeneous population of readers would respond to this environment. About 10 million work RVUs are attributable to 20 million cases, so 55,000 examinations per year equals 27,500 work RVUs per reader (column B). The top of the model suggests an overall budget of $600,000 per reader for salary and benefits, with $100,000 as fixed compensation for all members. The remaining pool of $182 million is distributed based on work RVUs. Column F shows the resulting compensation package. Traditional practices’ income models avoid incentive-based components because they affect the chemistry of the group. This new model, however, will create a circumstance in which all readers will have the largest caseloads that they can handle, from the moment they begin their workdays. Refinement of the work RVU to reflect both time and clinical significance better will also eliminate the tendency to favor one modality over another. In a perfect world (and, perhaps, not an achievable one), it will make no difference whether the caseload consists of plain films or MRI; both would produce the same daily volumes of work RVUs. This model has both fixed and variable distribution components. This means that the gap in compensation between the highest-paid and lowest-paid radiologists is lower than the gap in work RVUs. Elimination of the base compensation might be suitable in this environment, whereas it would not be suitable in the traditional private practice. Nonetheless, this model includes base compensation because of uncertainty about whether work RVUs are a perfect representation of the differences in clinical effort. Table 3’s footnote pertains to the compensation of consulting radiologists and the CEO. Setting the consulting radiologists’ incomes at the median of the readers’ incomes will put their incentives in line with organizational goals. It will be the role of the consulting radiologists to be the face of the organization to the ordering physicians. Their quality-control responsibilities will prevent readers from carelessly focusing on high volumes at the expense of accurate clinical findings. The more normalized the work RVU volumes, the tighter the income range and, in theory, the better the clinical quality will be. The CEO’s income package should be tied to the success of the readers, not the organization. The CEO will have income-oriented incentives pertaining to securing reading contracts with as many centers as possible at the best price. Simultaneously, it will be critical for the organization to foster a relatively stress-free environment in which radiologists can maintain high clinical standards and volumes. If the CEO succeeds in expanding business and keeping the number of readers at optimum levels, then the readers’ budget will go up, and so will the income of the CEO. Is this business model realistic? Its components may require fine-tuning. Table 4 summarizes financial planning for one version of this model. The income level is the lowest considered feasible, at 60% of Medicare rates. The entire physician budget is based on average salary and benefits of $600,000, readjusted for productivity differences. The compensation packages of the consulting radiologists (one for every 10 readers) and the CEO are driven by the reader-income model. This leaves $77 million per year for space and office equipment, triage, IT, general and administrative costs, and supplies. It is beyond the scope of this article to quantify the remaining components that are critical to the ability to deliver reading services at this price. The compensation package in this model is one that considers dramatic differences in regional incomes. Some private practices achieve more favorable results, but there are many others with salaries and benefits lower than those shown here. Summary On a smaller scale, off-site reading services have existed for years. The feasibility of doing this on a very large scale has increased with the capacity to transmit vast amounts of data over long distances and the ability to store terabytes of information cost effectively. The traditional private-practice model has served the needs of radiologists and referring physicians for more than 35 years. It has always been inherently expensive, however, and is now unnecessarily so, in light of what can be done to channel cases to the best subspecialists. This concept lets radiologists become invisible to patients and referring physicians. It takes radiologists out of the departments they have identified with since their training and places them in a virtual world of clinical information. Some will prefer the consulting role because they prefer interaction with the ordering physicians, but there are also many radiologists who prefer to read their cases with as little interruption as possible. The implication of a transition to reading centers is that there would be too many radiologists, since fewer of them could meet the demand for their services. Those who did find positions would continue to be compensated at rates in the highest quartile of specialists. Why not consider exporting the reading service to any imaging facility in the world? There are no technological impediments to this, and it might even expand the demand for radiologists within the US market. Additional Reading - What’s the Deal? A Look at Current Recruiting Packages in Radiology
Table 3. Caseload Distribution for 20 Million Examinations Interpreted by 364 Radiologists*The portability of examinations and the increasing referring-physician access to images will contribute to changes in the income model of the private radiology practice. I predict that the model of the future will involve either virtual or actual reading centers responsible for interpretation of tens of millions of cases per year. These centers will be electronically linked to all imaging facilities.
Table 1. Gross Income to Reading CentersThese entities will not bill patients or insurance companies for interpretation. Rather, they will invoice imaging facilities using a fee schedule deeply discounted from that used by the Medicare program. This is because the average administrative cost (for the provider and the payor) of billing and processing claims in the United States is approximately $7.50 to $10 per examination. There is no room for this in a low-fee environment. Organization Components Five divisions will provide the key operating units of the new model’s reading center. First, the triage division will separate all cases into two groupings: those examinations with clinical findings so obvious that they will require only a brief review and those examinations with clinical findings that are not so obvious. A key efficiency component will be the isolation of cases that can be rapidly interpreted, perhaps with bar coding of standardized report language to reduce the dictation time used on voice-recognition systems. This will leave the complex cases to the most experienced specialists. The time savings will be substantial, leading to at least a tripling of the cases that can be handled by the average radiologist. Second, the consulting-radiologist division will have multiple roles. It will support computerized physician order-entry systems whenever a conflict exists between the ordering physician and the system’s appropriateness guidelines. It will be the contact point for discussing a case with an ordering physician. This division will provide checks and balances for the reading physician, who will be prudent in committing time to a case, knowing that some cases will be reviewed and discussed by other radiologists. The division will randomly interpret a percentage of cases independently, with no knowledge of the designated reader; the two readings will then be matched for anomalies. The consulting radiologists would naturally be in a reading rotation to maintain their proficiency and credibility, but their caseloads would be far smaller than those of the readers. This division could also be responsible for teaching residents and fellows. Because speed and volume will be emphasized, a vigorous internal review, with consulting radiologists providing a blinded second read for quality assurance, will be needed.
Table 2. Estimated Need for Reading RadiologistsThird, the reading division is where all formal dictations will be performed. It will be organized by organ system. The income system will be heavily oriented toward incentives, using work RVUs that will support a variable income component in conjunction with a fixed component. The units will need to be modified to correct for the likelihood that a reader would gain a financial advantage for reading cases of one type versus another. Fourth, the IT division will deal with all data retrieval, distribution, and retention, as well as the privacy of patient information. Interpretations will need to be returned to the imaging site in a timely manner, and consultations with referring physicians will involve online access to the images. Internally, the cases will have to be organized and distributed to the readers based on specialization and clinical difficulty to minimize nonproductive time. There will be extensive (and expensive) hardware and software required to make this large technological enterprise work seamlessly. Fifth, the administrative infrastructure will be consistent with company revenues reaching hundreds of millions per year. The CEO is likely to be a radiologist with advanced business degrees; this will not be a necessity, but it has often been shown that physicians and nonphysician executives have conflicts. The largest budget item for the organization will be physicians’ salaries and benefits. Income System The 2008 Medicare conversion factor is $38.087 per RVU. The weighted-average work RVU for a population, excluding surgical/interventional cases, is 0.5. The work RVU median is 72.6% of the total professional-component RVU. Therefore, 0.5 divided by 0.726 equals 0.69 total RVUs per examination, and the average Medicare fee is 0.69 times $38.087, which equals $26.30. Medicare fees are generally among the lowest in most geographic markets, although some commercial payors have schedules below this level. Universally, state Medicaid fees are lower than Medicare fees. The key to getting payors to allow reading centers to bill the imaging facilities will be to offer rates at or below Medicare levels. The tables examine a possible scenario for a reading center driven by this payment structure.
Table 4. Abbreviated Budget SimulationTable 1 shows gross income for 5 to 20 million examinations per year for average charges per examination at three levels, with the highest level representing Medicare rates. Imaging centers will be likely to consider taking the risk of global billing if they will be paying reading fees at or below Medicare levels; the lowest fee shown is 60% of the 2008 Medicare average fee for nonsurgical procedures.