James A. Kieffer, MBA

Productivity is an important management tool in any business, including that of the radiology practice. Relative value units (RVUs) are frequently pressed into service to measure radiologist productivity, but there are some inherent shortcomings in the RVU system in that some examinations are undervalued. The goal of this article is to suggest how examination information can be compiled and shared nationally to provide credible benchmarking information for all radiologists. Because the information portrayed in the tables of this article is derived from a small private practice, all practices cannot necessarily assume that their numbers should be similar.

The simulation model at the end of the article is based on an incorrect assumption, but the mechanics of the model could be a valid tool for recompiling work RVUs. The assumption is that work RVUs are based strictly on the relative average time to perform the procedure in relation to others. Time certainly is a direct by-product of the complexity of an examination, but it appears that a “clinical significance factor” is also included. The model would help test the time/clinical significance component in each RVU and also identify procedures that are undervalued.

Table 1.(Click the image for a larger version.)

The radiology practice profiled uses a scheduling program that provided useful coverage information that facilitates compilation of work RVU per hour, something not commonly available. This article will explore the linkage of coverage hours and a simulation of caseload hours using work RVUs. It is an experiment that will hopefully trigger constructive feedback from the readership.

This practice, out of economic necessity, is a busy one. This does not eliminate the administrative duties not reflected in work RVU. Whenever possible, the administrative duties are limited to “off-hours” when the department caseload is lightest. The scheduling program does codify these time commitments and, hopefully, they were accurately factored in the time simulations.

There is another factor that is difficult to quantify when comparing productivity statistics for this practice versus others. The major coverage location is still film-based in many areas, but is moving forward with the installation of a full PACS. It is likely that the work RVU per FTE could be higher in a fully digital environment. There is little or no information published on this issue because it is difficult to quantify, plus there is a behavioral component. If a digital environment does make a radiologist more productive, then practices will either restructure themselves to handle caseload with fewer members, thus enhancing income, or choose to take more time off because productivity per hour/day/week is higher.

PAYOR MIX

The group portrayed in this case study services a health care delivery system with a payor mix that likely impacts its ability to generate a competitive salary/benefit package. Table 1 puts this in context.

Table 2.(Click the image for a larger version.)

Column A represents a generalized payor mix; column B shows the distribution of practice charges. Column C might be considered the payor mix of the Average Practice, although the percentage ranges within each payor group can be broad, regionally and nationally. The group’s payor configuration has significant amounts of low to medium payors and self-pay population. The self-pay/indigent populations of all practices account for most of bad debt, where the collection ratio ranges between 5% and 15%.

Column D is a relative measure of income per payor group, using Medicare as a baseline. Many commercial and managed care payors establish fees that are multiples of the Medicare rates since HCFA (now CMS) established its relative value system in the late 80s. The multiplication of the two payor mixes by these income factors establishes a weighted difference in overall income, all other factors being constant (more on this point later). This implies that the average practice in the sample practice’s region could produce almost 20% more income from its cases.

If two practices covered different hospital/health care delivery systems with approximately equal caseload demands, and one generated 20% more income from its caseload, the number of radiologists in that group might be greater, choosing to reduce coverage hours per member rather than receive higher income.

Recommendation: When national organizations seek out work RVU statistics from their membership, they should ask for a payor breakdown as a data subset, to establish a correlation between payor mix and productivity.

COVERAGE SCHEDULES

Table 3.(Click the image for a larger version.)

The practice’s scheduling program had extensive coding detail to define coverage responsibility. Table 2 is an abbreviated distribution of all calendar days per member, distinguishing call, weekday and weekend coverage, and time off.

There were seven full-time and one part-time members in 2004, the compilation year for the examination statistics. Three of the full-time members handled the interventional caseload and are assigned more backup call. As a practical matter, they were potentially on backup call one third of the time to handle any off-hour interventional demands that, fortunately, are infrequent. The trade-off was fewer weekend call days. The “CALL-NOON” and “CALL-10AM” designations imply that the individual assigned that responsibility arrived at either noon or 10 AM, remaining in the department until 10 PM. Any diagnostic cases after 10 PM were handled as a wet reading by a nighthawk service, with the formal dictation performed by one of the eight radiologists the following day. The four columns at far right start the simulation of coverage hours by assignment of assumed physical time in the departments. The backup call time is a non-scientific attempt to simulate annual hours actually in the department while on backup call. Table 3 portrays the on-site hours.

The physical hours on-site are an important component of the differences in productivity between groups. It is not likely that any surveys have asked respondents to provide both work RVUs and coverage hours. If, as suggested, material differences in the income systems affect decisions about coverage hours, we would likely see the more lucrative practices showing fewer hours per full-time-equivalent (FTE).

PRODUCTIVITY SUMMARY

Table 4.(Click the image for a larger version.)

Column A in Table 4 reproduces the simulated coverage hours from Table 3. Column B is an organization of examinations by physician for calendar 2004 derived from the practice receivable system. The data was provided by CPT code, facilitating compilation of work RVU. Subsequent tables will offer more detail to illustrate the extent in which work RVUs can be organized. Column D averages the work RVU per procedure. Those with the higher average are the three members who handle the interventional caseloads. The practice average of .58 is consistent with a complex caseload universe. This is an important point given the size of this group. The case mix will be illustrated shortly.

Columns E and F represent caseload handled by the practice, not billed through the receivable system. It consists primarily of plain films with no CPT code detail and was arbitrarily distributed equally among the full-time diagnostic members, using the average RVU/procedure for plain films. Columns G and H are the combined totals, and column I represents the average work RVU per coverage hour.

The full-time member statistics are segmented at bottom as to totals, mean, and standard deviation. The standard deviation, as a percent of the mean, suggests the value of using the work RVU as a productivity measure. Procedure volume has a predictably large variance because of case mix differences, yet the work RVU and resulting RVU/hour are tightly bunched.

CASE MIX DISTRIBUTION

Table 5.(Click the image for a larger version.)

This practice has a complex case mix that demands, because of the volume in relation to group size, that each full-time member handle all modalities. Concessions are made to the three members who handle the interventional workload, yet they all do cover other areas to some degree. This explains why the total work RVUs per member are tightly bunched.

There is a theory that specialists who concentrate on only a few modalities can produce a great number of work RVUs because of disproportionate weighting of the values. Compiling the work RVU distribution this way might help answer questions about wider disparity among members of other practices that are larger and more subspecialized. The final model of this article also is a useful tool to test if some examination groupings are over/under weighted.

The section above on payor mix discussed the theory that practices with more lucrative income systems would likely favor quality of life over enhanced income. Table 5 also hints at another reason why the work RVUs for this practice might be higher than average, again influenced by income-related issues. This group has a disproportionate amount of surgical cases for its size. The payor fees for surgical codes are less favorable than for diagnostic ones. This implies that other groups with a procedure volume of 127,000 might have not only a better payor mix but also fewer surgical codes.

SIMULATION: CASELOAD USING RVUS

Table 6a.(Click the image for a larger version – PDF.)

Table 6a and Table 6b represents a hypothetical simulation of annual caseload time using work RVUs. The statistics for FTD #4 were chosen because the examination mix was the most broadly based, including interventional procedures, and this individual was not asked to interpret the outside plain films.

Column A contains an abbreviated description of the sections in the CPT manual where the imaging and surgical procedures are grouped. The number of codes in each group may range from as few as three to as many as 25. Column B is the imaging modality. The best way to describe the interrelationship between columns A and B is to reference “Head & Neck.” There are groups of codes covering CT, MRI, and Plain Films within the “Head & Neck” section of the manual (Ultrasound has its own “Head & Neck” section).

Column C is the 2004 examination volume for the grouping and modality configurations, and column D is work RVU resulting from the examinations. Column E shows average work RVU/procedure for the grouping/modality configuration. Column F is the normalization of the column E values, using the Chest/Plain Film as “1.00”; all other groupings are restructured in relation to this benchmark.

Column G is the simulation of average procedure times for each grouping based on the assumption made about the average reading time for Chest cases. Column H is the multiplication of the time per procedure by the annual volume. The total is then divided by 60 to estimate annual hours handling caseload. Now we need to critique this model:

  1. The 1,241 hour simulation was based on an assumption that caseload time has to be less than coverage time. It is not reasonable to expect any physician to handle clinical workload 100% of the hours worked. No data exists that suggests maximum workloads before stress interferes with clinical efficiency. An arbitrary assumption was made that 6 reading hours out of 8 coverage hours is a maximum limit.
  2. FTD#4 simulated average of 6.04 work RVU per coverage hour would translate to a different figure per reading hour. If 6 hours per day is a limit (75%), then the work RVU per reading hour is 6.04/.75 = 8.05. The chest film(s) average RVU is .21, implying: 8.05/.21 = 38+ Chest examinations per hour. That translates to 1.58 minutes per examination (top of column G).
  3. The surgical codes likely have a work RVU that does not fit well with the diagnostic examinations; this is just a theory based on the history of how the original RVU values were assigned. Each specialty had direct input about its own coding mix, and the performance of cases involving the surgical codes in the 1980s was impacted by turf battles. Also, interventional cases involve multiple surgical codes, implying that simulation of the time for an interventional case requires addition of two or more surgical and S&I codes.
  4. The challenge to the reader is to determine if the column G averages per procedure are realistic. Changing the baseline time to a higher number will result in simulated hours at or greater than the coverage hours. This model could just as easily be expanded to include each CPT code in this radiologist’s 2004 universe where we used the single view Chest as the baseline.
  5. If the work RVU includes both a time and “clinical significance” component, would there be a value in splitting them up? Is it reasonable to assume that large groups that might subspecialize should use the work RVU to distribute part of their income pool because some specialists might be handling more “clinically significant” cases?

CONCLUSIONS

Table 6b.(Click the image for a larger version – PDF.)

This article suggests that the work RVU per member for this practice is higher than average out of economic necessity. If the reader is a member of a practice with a more favorable payor mix, and the proportion of surgical codes is lower, then the theory put forth by this article is that the reader’s annual work RVU will be lower.

An important caveat in using this data is to consider how efficient the radiologists are because of department infrastructure. Could the work RVU per FTE for this group be higher (thus fewer members) if their major coverage sites were more digital? This suggests that practices with more favorable payor and case mix might produce greater work RVU per FTE because they are more efficiently organized. This gives them the double benefit of greater income in fewer reading hours.

It would not be difficult for the American College of Radiology, if their members wanted it, to produce a work RVU based strictly on time commitments per CPT code. It would be more straightforward than the subjective “clinical significance” component.

The practice profiled democratically distributes the caseload out of logistical necessity. It is too small to subspecialize. This eliminates any inappropriate weighting of CPT code groups where an hour spent on a single modality or organ system artificially generates more work RVU than is appropriate. The only way to test if the assigned work RVU is wrong is to speculate on how many cases can be done in an hour for a procedure grouping and then decide if the resulting total work RVU difference is justified because one type of caseload is more clinically important than the other.

James A. Kieffer, MBA, is president, Proforma Financial Group Inc, Nashua, NH; www.Proformafg.com .