The Third Rail: Measuring and Managing Physician Productivity
Richard Duszak, MD, bears the authority of experience when he cautions those who dare attempt to manage physician productivity: “If you don’t think there is a problem in your practice, you don’t want to go down this road and touch the third rail on the Metro—because you will get electrocuted, and you may die.” Chair of the ACR Coding and Nomenclature Committee and a diagnostic and interventional radiologist with Mid-South Imaging and Therapeutics, Memphis, Tenn, Duszak did go down this road at a previous practice, and he shared specific attributes of that model as well as his thoughts on metrics and benchmarks with attendees of the Managing a Radiology Business From the Top: Physicians & Administrators meeting, held on February 23, 2008, in San Francisco and sponsored by the RBMA. Metrics used in such programs can be based on the number of examinations read or the amount of revenue produced, but Duszak’s measure of choice is the physician work RVU (PWRVU). While some practices use total RVUs, Duszak believes that breaking the practice-expense and malpractice components out of the total number offers the best reflection of physician work. “[RVUs are] more complex to understand and to measure because you have to import all of the RBRVS data into your files,” he says, “but I believe that the physician work component of RVU is a good proxy of the time and complexity of physician services. It is inherently an imperfect system, but it is the best available measure of clinical work.” What the RBRVS does not measure are administrative and academic activities, including publishing, teaching, and grant support, and Duszak directed attendees to a paper that creates what the authors refers to as an academic RVU.1 “How much you weigh each of these components may vary depending on the priorities of your practice, but I think it does show that there are ways to measure to address productivity for some of these things that are much more difficult to measure, or certainly are not measured by the RBRVS system,” Duszak says. Productivity changes, by year, for 11 radiologists; the red bar indicates the acceptable performance level and the initials under the bars designate individual radiologists.
No Place Like Home Published benchmarks for radiologist productivity do exist,2-4 but most are based on data several years old, and a marked trend in increasing radiologist productivity suggests that using these benchmarks may result in setting the bar too low. Duszak cited an ACR survey5 of 1,924 radiologists, published in 2005, that revealed an average of 9,100 PWRVUs and 13,900 examinations logged annually. “One of the things this survey showed—no surprise to any of us—is that we are all working harder,” he says. Duszak strongly recommends working from custom benchmarks based on the individual group’s productivity. “Some groups are workaholic groups, some groups like to take a lot of time off, some groups have a more efficient PACS, some groups have a terrible voicerecognition system; there are many reasons why these benchmarks may not apply to your practice,” he notes. “The best benchmark is . . . to normalize yourself within your own group; use your own historical data . . . to determine what a reasonable benchmark is or should be for your practice.” For those practices intent on pressing ahead, Duszak emphasized the following pitfalls. RBRVS metrics measure only RVUs. There are many other activities that a radiologist can or should be providing: administrative work, subspecialization (tough cases that need to be presented to the tumor board), and practice and relationship development Practices with a high level of subspecialization need to address this in their metrics. “A neuro MR person is going to have a better opportunity to generate more RVUs than a semiretired radiologist who does plain films and fluoro—all valuable services that somebody needs to be providing,” he advises. There are ongoing corrections to the RBRVS, and they will need to be incorporated in the metrics. A Case Study In sharing the model used by a former practice, Duszak cautions against unilaterally applying any such model without adapting it to the individual practice. “If you are going to go down this route, you will need to implement a model in your individual practice that works for your individual needs,” he says. The program was implemented in 2001 in a practice of 24 radiologists, two radiologist assistants, and two physician assistants, with an exclusive affiliation with a community teaching hospital of more than 700 beds. Initially, the group released blinded data, but the recalcitrant people continued to be low producers. The group then released the data with everyone’s names attached, still with no effect on the low performers. “Ultimately, there ended up being enough outrage and morale issues that enough higher performers in the group actually swayed the group to do something about it,” Duszak shares. Ultimately, charges were used as a proxy of total RVUs because of logistics. “While PWRVUs were the best data, the most contemporaneous data we could get very, very quickly back from the billing system were charges, which were established as proportional to total RVUs and which were reasonably proportional to physician RVUs,” Duszak explains. Corrections were made for administrative time, vacation, and other time away, so that the department chair was considered a 50% FTE; the practice president, who was negotiating a new hospital contract, was considered a 70% FTE. Singledigit offsets were made for other activities, such as those of the scheduler, PACS liaison, interventional-radiology chief, mammography chief, and other chiefs. All partners and partnership-track physicians were tracked, though the nonpartners were simply tracked without penalties. “They knew where they stood, but the recognition was it takes people a couple of years to get up to speed, and we didn’t want to disenfranchise those individuals,” Duszak notes. “We established a minimal expectation of measurable work, and if you did anything above that, that’s great, but we didn’t want you falling below that,” he explains. The penalty for not meeting that expectation was the loss of two weeks’ vacation, with an additional two days lost for every percentage point below the minimum threshold. To establish a minimal expectation, the practice went back over 5 to 10 years’ data and found that, historically, almost everybody performed within one standard deviation of the norm, 13%. That deviation was liberalized to 15%, and everyone was expected to produce at 85% of the mean. They established an arbitrary fixed number, and set it as a threshold. Although administrative work was measured as noted above, practice and relationship building and the subspecialization effect were not measured because it was determined that the wide latitude in acceptable productivity corrected for these variations. Each month, a spreadsheet was issued with each radiologist’s name on the left side and the percentage listed on the right. The results were dramatic. “What we saw was an overall increase in productivity across the practice: The most productive people had the smallest increases, and the least productive people in the group had the largest increases,” Duszak says (see figure). Overall, during the course of a few years, the group actually created 2.4 more FTEs of work, with more than half coming from three physicians. The group lost one valuable and conscientious radiologist, however, due in part to the stress of barely making his threshold for several years. “In summary, my take on this is measuring productivity is an important issue, but there’s no one-size-fits-all solution,” Duszak notes. “It sometimes can be more disruptive than helpful, so you want to be very, very careful of what you ask for, because you may not get the results you expect.”