| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Health Policy and Practice |
1 From the Research Department, American College of Radiology, 1891 Preston White Dr, Reston, VA 20191 (J.H.S., J.H.B.); and the Society of Cardiovascular and Interventional Radiology, Fairfax, Va (M.R.M.). Received April 25, 2000; revision requested June 9; revision received July 7; accepted July 25. Address correspondence to J.H.S. (e-mail: jonathans@acr.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
MATERIALS AND METHODS: In 1996 and 1997, practices were surveyed, and data on costs and other operational characteristics were obtained from approximately 170 practices. Several components of practice costs (eg, physician-related costs, administrative and business costs) were calculated separately for different group types (eg, academic, private hospital-only), each on four bases: per full-time equivalent (FTE) radiologist, per relevant procedure, per relevant relative value unit (RVU), and as a percentage of revenue.
RESULTS: Median total practice costs per FTE radiologist ranged from approximately $90,000 to $190,000, depending on group type. Per procedure, the median ranged from $9 to $21; and as a percentage of revenue, it ranged from 27% to 41%. Median technical costs were approximately $36 per technical RVU in private hospital-and-office groups. Within any category of group, for every cost category, there was substantial variation among groups.
CONCLUSION: The sizable variation implies that means or medians should not be regarded as norms. Nonetheless, the data on 75th and 25th percentile costs can show a radiology group where savings and inadequate resources, respectively, are relatively likely to be found. Physician-related costs are best measured per FTE. Technical costs and administrative and business costs are best measured per RVU or for categories of groups defined by having similar percentages of nonhospital services.
Index terms: Economics, medical • Radiology and radiologists • Radiology and radiologists, socioeconomic issues
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
Whatever the past reality, practice costs currently clearly do matter greatly. A practice should be sure to make the expenditures necessary to ensure high-quality patient care. However, to a large extent, Medicare and other payers now decide how much they will pay, usually on the basis of a fee schedule, giving little regard to physicians’ billed fees. Thus, any excess in expenditures tends to produce a corresponding decrease in net practice income and, possibly, to constrain physicians’ ability to provide services.
In this environment, it is important for radiologists to know what levels of practice costs are reasonable and whether components of their practice’s costs are out of line with those of other similar practices. To meet this information need, and as part of its mission of assisting radiologists in socioeconomic matters, the American College of Radiology performed a survey of practices’ costs. The purpose of this article is to report the results of that survey.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
|
A substantial percentage of responding practices did not supply data on their workload or on their revenue. To keep the overall response rate up, we had placed the sole question about a practice’s revenue on a separate sheet at the end of the survey and said we realized revenue is a sensitive issue and practices might not wish to report their revenue. Hence, the number of groups for which we obtained expense data per procedure, per relative value unit (RVU), or as a percentage of revenue was substantially lower than the number for which we obtained data per full-time equivalent (FTE) radiologist.
Data Cleaning and Inflation Adjustment
We conducted extensive data cleaning. Values covering time periods less than a year were annualized to represent a full year’s data. Column sums were verified for accuracy. Through this process and other reviews of survey forms, we identified the many values that had been left blank but that really were zeros, and we recoded them from "missing" to zero in our data file. We conducted logic checks. We calculated financial ratios, such as total practice costs per FTE radiologist, and we investigated outliers (highest and lowest values far different from any other values), checking that data had been entered correctly. For any questionable items that remained, we telephoned the responding practice. Seven practices were deleted from the data set because of data problems that could not be resolved; some of these were practices that did not return repeated phone calls seeking information regarding apparent problems.
To analyze survey responses covering different years on a common basis, we increased all reported dollar amounts, both expenses and revenue, by the percentage increase in the consumer price index, or CPI, that occurred between the time period the data covered and 1998.
Definitions and Measurements
On the basis of the services practices reported offering, we categorized responding practices as diagnostic-radiology-only practices, combined diagnostic radiology and radiation oncology practices, or radiation-oncology-only practices. This article includes information solely from diagnostic-radiology-only practices, which were approximately three-fourths of all respondents. We plan a subsequent article on radiation oncology practice costs. Further information regarding definitions and measurements is found in the Appendix.
Presentation of Data
We tabulated data separately for the 12 categories of practices described in the Appendix. In the interest of reliability, however, we report data for a category only when there were at least six practices in the category for which we had data, and we report the 25th and 75th percentiles only when we had data for at least 10 practices. Some of the 12 categories had so few responding practices that there were no data to report. This, for example, was true for government employee practices and for nonhospital-only practices.
Because the percentage of services delivered at nonhospital sites was expected to affect technical expenses, and possibly other expenses, we divided private nonacademic radiology groups that function in both hospital and nonhospital sites, by far the largest category of practice, into three subcategories on the basis of this percentage. We report data for each subcategory and for the category as a whole. Some practices did not report the percentage of their services that were performed in nonhospital sites and so could not be placed into subcategories. Thus, the number of responding practices in the three subcategories, added together, is less than the number of responding practices in the category as a whole.
Practice costs are reported in commonly used categories—namely, physician-related practice costs, technical costs (ie, medical support expenses), and administrative and business costs—to facilitate practices comparing their own experience with the information we report. Within each of these three categories of practice costs, we also present data on major subcategories. Readers should note that the types of expenses we include in practice costs, as detailed in Table 1, are not identical to the "practice cost relative value unit" concept used in Medicare’s resource-based relative value scale (RBRVS). The latter, for example, excludes malpractice expenses.
To provide a variety of perspectives, we present cost data on four bases: (a) per FTE physician, (b) per relevant procedure, (c) per relevant RVU, and (d) as a percentage of net practice revenue. The Appendix and Table 1 provide more details.
In the tables, for each combination of practice category and cost category presented, we report (a) the number of practices from which the data came; (b) the mean cost; (c) the SD, which is the statistic most commonly used to measure variation; (d) the standard error of the mean (SEM), which is the usual measure of the uncertainty of the mean; and (e) the 25th percentile, median (50th percentile), and 75th percentile. The usual interpretation of the SEM is that, assuming no response bias, there is a 95% probability the actual mean for all practices in a category is within approximately two SEMs for the practices that responded to the survey.
In the text, we generally refer to medians and the 25th and 75th percentiles. These statistics not only give information about typical practices and variation among practices, but also, unlike the mean and SD, they are influenced relatively little by anomalous outliers and erroneous survey responses. We often note the difference between the 25th and 75th percentiles. This interquartile range is a frequently used measure of variation.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
|
Per procedure.—The median of total physician-related practice costs per procedure was approximately $7 in academic groups that functioned in both hospital and nonhospital sites and was $5 in private nonacademic radiology groups regardless of whether these nonacademic groups were hospital-only ones or functioned in both types of sites (Table 3). Among the nonacademic groups, the ratio of the 75th to the 25th percentile groups’ expenditures was approximately 1.5:1.0; for the academic groups this measure of variability exceeded 2:1. Table 3 presents further details, including information on individual components of total physician-related practice costs.
|
|
|
Per relevant RVU.—The median of total technical costs per technical-component RVU was approximately $36 in private nonacademic radiology groups that functioned in both hospital and nonhospital settings (Table 6). There were no statistically significant differences related to the percentage of a group’s services that were provided in a nonhospital setting. However, there was much variability, with expenditures of the 75th percentile group more than twice those of the 25th percentile group. Technical personnel costs and technical nonpersonnel costs were broadly similar to each other in magnitude.
|
Per relevant procedure.—The median of total nonphysician-related practice costs per procedure was approximately $12 in academic groups that functioned in both hospital and nonhospital sites, $4 in nonacademic private hospital-only radiology groups, and $9 in nonacademic private radiology groups that functioned in both types of sites (Table 7). Variability was high, with a ratio of greater than 3:1 for 75th percentile to 25th percentile costs, except in the nonacademic private hospital-only radiology group category. Table 7 provides further details, including the breakdown between technical costs on the one hand and administrative and business costs on the other and, within each, the breakdown between personnel and nonpersonnel costs.
|
|
Total technical costs showed a similar pattern in private nonacademic radiology groups that functioned in both settings. The median total technical cost was $22,000 per FTE radiologist in groups with less than 20% of services outside the hospital, compared with $158,000 in groups with more than 30% to 50% nonhospital services. In contrast, the median of total administrative and business expenses was in the range of approximately $50,000–$70,000 per FTE radiologist in all categories of groups with two exceptions: The median was $112,000 for private nonacademic radiology groups with more than 30% to 50% nonhospital services, and it was $19,000 for multispecialty groups operating in both locations. Table 8 presents further details.
As a percentage of net practice revenue.—The median of total nonphysician-related practice costs was 13% of revenue in nonacademic private hospital-only radiology groups, and among these groups variability was relatively small (Table 9). The median was 24% of revenue in academic groups that functioned in both hospital and nonhospital sites and, in contrast, the 75th percentile cost was more than 2.5 times the 25th percentile cost.
|
Total Practice Costs
By definition (Table 1), total practice costs are the sum of physician-related and nonphysician-related costs.
Per FTE radiologist.—Depending on the category of group, median total practice costs per FTE radiologist ranged from approximately $90,000–$100,000 (their level in hospital-only academic groups and private nonacademic multispecialty groups that functioned in both hospital and nonhospital settings) to approximately $190,000 (their level in private nonacademic radiology groups that functioned in both settings) (Table 10). Among the private nonacademic radiology groups that functioned in both settings, the median ranged from $150,000 for groups with less than 20% of services in nonhospital settings to more than $250,000 for groups with more than 30% to 50% nonhospital services. Variability in every category and subcategory of group was substantial: Costs of the 75th percentile group typically, but not for every category, were approximately twice those of the 25th percentile group or even higher.
|
As a percentage of net practice revenue.—Depending on practice type, median total practice costs ranged from 27% of revenue in private nonacademic hospital-only radiology practices to 41% of revenue in academic practices that functioned in both hospital and nonhospital settings. In most categories and subcategories of practices, variability, although substantial, was less than 2:1 as measured by the ratio of the 75th percentile group’s costs to the costs of the 25th percentile group.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
The response rate was disappointing, although far better than the barely double-digit response achieved by the Health Care Financing Administration, or HCFA (the federal agency that manages Medicare), in a similar survey. We know that responding practices that supplied workload data included few very small practices and performed fewer procedures per FTE per year than is typical of all practices (3). How, and even whether, such response biases affect practice costs is not obvious.
Practices clearly had difficulties in providing even the relatively simple information on costs that we requested, or they were unwilling to do so. With the same effort devoted to follow-up, most of our surveys of practices obtained a response rate of approximately twice the 34% achieved in the 1996 portion of this survey.
Although we used standard cost categories, some of the distinctions we made—for example, between fringe benefits of physicians and fringe benefit costs for all other personnel—are probably not part of the routine record keeping of many practices. These distinctions may have led to inaccuracies in reporting costs.
Comparability of our data across practices may be limited because types of expenses that are paid by some radiology practices—and, therefore, are reported in our survey—are instead paid by individual physicians, hospitals, or an entire multispecialty group in other practices and therefore are not included in the survey. This limitation is likely to be particularly important for academic practices and for multispecialty groups. The low professional liability expenses reported by academic practices and the low technical costs reported by academic and multispecialty groups may well reflect reporting problems of this type.
An expense that is a personnel expense in one practice may be a nonpersonnel expense in another—for example, one practice may do its billing with its own staff, whereas another may contract with an outside billing service. Hence, totals for technical costs and for administrative and business costs are likely to be more comparable across practices than are the separate personnel or nonpersonnel components of each.
Information on costs per RVU has special limitations. Some are mentioned in the Appendix. Also, RVU values in the RBRVS are being revised annually in the period after 1998, which makes our results regarding RVUs of lesser use to readers using an RBRVS dated after 1998.
Some data problems and errors probably have not been fully eliminated by our data cleaning process. Also, information in this study is somewhat dated. Although we made an inflation adjustment and readers can adjust further to the year in which they are using the data, such adjustments do not take account of specific categories of costs changing at rates different from the general inflation rate or changes in staffing patterns, equipment used, and other factors that affect costs.
Comparison with Other Information Sources
In its Socioeconomic Monitoring System, or SMS, survey, the American Medical Association annually contacts a random sample of all physicians in the United States. The overall survey response rate is usually more than 60%, and more than two-thirds of those asked for financial information supply it. Pooling 1994–1996 data, the American Medical Association reported financial information from 174 diagnostic radiologists whose responses met relatively strict criteria for completeness (4). The Socioeconomic Monitoring System’s "total expenses excluding professional liability insurance and physician payroll" corresponds approximately to our category of total nonphysician-related costs and was reported by the American Medical Association to average $148,000 per FTE physician. The 25th percentile was $29,000; the median, $64,000; and the 75th percentile, $159,000.
We calculated comparable figures from our survey by pooling data from all private nonacademic radiology-only practices and adjusting prices to 1995. Thus calculated, the mean from our survey was $149,000; the 25th percentile, $45,000; the median, $82,000; and the 75th percentile, $187,000. The Socioeconomic Monitoring System statistics and ours are similar, especially in their means. More important, the approximately 2:1 disparity between the mean and median in both the American Medical Association data and ours and the more than 4:1 disparity between the 75th and the 25th percentiles in both sets of data—a far larger disparity than appears for any broad cost measure as we have presented in our data—result from pooling hospital-only practices with hospital-and-nonhospital practices. These disparities show the desirability of treating these two types of practices as separate categories in reporting costs, which our study does.
The Medical Group Management Association (MGMA), an organization primarily of relatively large, relatively professionally managed groups, annually surveys its members in detail about their finances. The 1996 survey (of 1995 finances) obtained data for 28 of 210 member radiology groups, a 13% response rate (5). The MGMA’s "total operating cost" corresponds in content almost exactly to our total nonphysician-related cost plus professional liability insurance. The MGMA reported that 1995 median total operating cost was $148,000 per FTE and 27% of revenue. Comparable statistics from our survey, calculated as described in the preceding paragraph, are $86,000 and 22%. The results do not agree well. The MGMA’s "total physician benefit cost" is essentially identical to our total physician-related cost minus professional liability insurance. The MGMA found the 1995 median of total physician benefit cost was $47,000 per FTE and 9% of revenue; comparable statistics from our survey are $45,000 and 9%. MGMA data are not representative of all practices because of the nature of its membership and, perhaps, its low survey response rate.
The Health Care Financing Administration gathered information from insurers, covering all states and most insurers, on professional liability insurance premiums (6). It found the mean 1995 premium was $11,400. Our comparable statistic is less: $8,000. The Health Care Financing Administration priced a standard coverage—a $1 million–$3 million mature claims-made policy. Our data reflect the actual coverage each practice had; this difference may affect the means. Differences in coverage limits may also explain some of the differences we found among different types of groups in professional liability insurance costs.
Interpretation of Findings
All measures of cost show substantial variability. Variability is typically larger when less aggregated cost categories are considered. Variability is substantial even after controlling for practice type and when costs are measured from different perspectives (eg, per FTE radiologist, per procedure).
The inescapable presence of substantial variability implies that central tendencies—such as the median (50th percentile) or mean—should not be regarded as performance targets for every practice. Nonetheless, practices with a given cost item at or above the 75th percentile should probably regard that item as one relatively likely (but not sure) to yield savings, if it is explored carefully. Conversely, expenditures at or below the 25th percentile can be regarded as probable evidence of parsimony—or, possibly, skimping. The lesser variability of relatively aggregate cost categories means that comparisons of relatively aggregate categories are more likely to be useful for identifying inappropriate expenditure levels than are comparisons of component categories.
We examined costs from four perspectives (per FTE radiologist, per relevant procedure, per relevant RVU, and as a percentage of revenue) in hopes of finding a best measure. We used two criteria for evaluating measures. One was that variability within a category of practices should be small, which would make means or medians more useful as indicators of targets. (As an example of a poor measure by this criterion, consider measuring costs in dollars per practice; this measure obviously would vary tremendously by practice size, so we did not use it.) The second criterion was that means or medians should differ little according to the percentage of a practice’s services that took place in a nonhospital setting. We found, as noted, that all four ways of measuring cost were limited according to the first criterion, so we chose among them mostly according to the second.
We found that median total physician-related costs per FTE radiologist do not differ much by percentage of services rendered in the nonhospital setting, a finding which is not obvious a priori. Measured per procedure or per RVU, median total physician-related costs are also little affected by percentage of services rendered in the nonhospital setting, but the per-FTE measure is probably easier for practices to compute and so is most convenient to use. The larger the percentage of a practice’s services rendered in the nonhospital setting, the smaller the percentage of revenue constituted by median total physician-related costs, because in the nonhospital setting, practices’ revenues are augmented by receipt of technical component payments.
As expected, median total technical costs are larger per FTE radiologist and as a percentage of revenue, the larger the percentage of a practice’s services rendered in the nonhospital setting. In contrast, measured per technical RVU, they are relatively constant, and this is the preferred measure to use.
Per nonhospital procedure (our measure of relevant procedures), the findings regarding median total technical costs were somewhat unexpected: Uniformity might have been expected, but reported median total technical costs per nonhospital procedure clearly were higher in practices with 20%–30% nonhospital services than in practices with less than 20% nonhospital services. Data analysis performed to find the explanation showed that mean technical component RVUs per nonhospital procedure typically were approximately one in the latter practices compared with approximately two in the former. This suggests the nonhospital procedures of the latter practices were predominantly conventional radiographic studies, which typically have a technical component relative value of about one, whereas the former practices were performing more varied nonhospital procedures.
Also unexpected in terms of total technical costs was the very low 25th percentile value for private nonacademic radiology groups with 20%–30% nonhospital services. Most likely, for a sizable minority of these practices, an entity other than the practice owns all or most of the nonhospital practice settings and, therefore, incurs the technical expenses.
Median total administrative and business costs per FTE radiologist were higher in practices that render a relatively high percentage of their services in the nonhospital setting than in others. This finding is somewhat unexpected and shows that the additional costs a practice encounters in providing services in the nonhospital setting affect not only its direct medical support costs (ie, its technical costs) but also its administrative and business costs. We found that median total administrative and business costs per relevant RVU vary little by percentage of services rendered in the nonhospital setting. Thus, the per-RVU measure is the best one to use.
Although measuring costs on a per-RVU basis would, other things being equal, thus be our recommended approach for technical costs and for administrative and business costs, the data systems of many practices do not permit this and, as noted, RVUs have special limitations. A useful and simpler alternative is for practices to measure costs according to one of the other three bases (eg, per FTE radiologist) and then to be sure to compare themselves with practices that have a similar percentage of services performed in the nonhospital setting.
Properly used, data from this study can be a valuable guide to practices seeking to evaluate the appropriateness of their costs, even though the substantial variability of all cost measures we studied implies that means or medians should not be taken as hard and fast targets.
|
APPENDIX: DEFINITIONS AND MEASUREMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
Practice Costs and Revenue
The practice cost and revenue categories about which the survey inquired are shown in Table 1, together with the definitions it gave. These categories were based on recommended categories of the Radiology Business Managers Association and the MGMA, and so should have been relatively familiar to practices.
Procedures and RVUs
Because technical costs normally occur only for nonhospital procedures, in counting procedures to calculate technical costs per procedure, only nonhospital procedures were counted. For all other cost-per-procedure statistics, the count of procedures used to calculate costs per procedure was the total number of procedures a group performed in all locations. The number of procedures counted for each radiologic service was the number of CPT codes billed for that service. Thus, what is usually thought of as one interventional procedure was normally counted as more than one procedure because interventional procedures usually are billed with one or more CPT codes for the imaging; other codes for any interventions performed, such as angioplasty or stent placement; and codes for insertion and positioning of the catheters.
To calculate costs per RVU, RVUs were measured by using the 1998 Medicare RBRVS (2) as follows: To calculate technical costs per RVU, we included only RVUs from nonhospital procedures and assigned to each such procedure a number of RVUs equal to its technical component relative value. To calculate physician-related costs per RVU, we included RVUs from all procedures performed in a practice and assigned to each procedure a number of RVUs equal to the procedure’s professional component relative value, regardless of where it was performed. To calculate administrative and business costs per RVU, we included RVUs from all procedures performed by a practice; procedures performed at a hospital were assigned RVUs equal to their professional component relative value, and nonhospital procedures were assigned RVUs equal to their global relative value.
Because there is no measure of RVUs that would be appropriate, we did not calculate total nonphysician-related costs per RVU or total costs per RVU. Because the RBRVS does not assign RVUs to screening mammography, we assigned this examination 80% of the RVUs of bilateral diagnostic mammography, which is what the American College of Radiology had recommended when Medicare first introduced coverage of screening mammography.
Although defined technical and professional component relative values generally exist for radiologic services (essentially, procedures with CPT codes numbered between 70,000 and 79,999), they often do not exist for other services, such as the CPT codes used for catheter insertion and placement during interventional procedures. In general, only a single relative value exists for these codes. When only a single relative value existed, we used this value whenever a relative value was needed—that is, for technical costs per RVU, for physician-related costs per RVU, and for administrative and business costs per RVU.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, J.H.S.; study concepts, J.H.S., M.R.M.; study design, J.H.S., J.H.B., M.R.M.; definition of intellectual content, J.H.S., J.H.B., M.R.M.; literature research, M.R.M.; data acquisition, J.H.B., M.R.M.; data analysis, J.H.B.; statistical analysis, J.H.B.; manuscript preparation, J.H.S.; manuscript editing, J.H.S., J.H.B.; manuscript review, J.H.B., M.R.M.; manuscript final version approval, J.H.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION APPENDIX: DEFINITIONS AND...REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. T. Gray, W. Hollingworth, C. C. Blackmore, M. A. Alotis, B. I. Martin, S. D. Sullivan, R. A. Deyo, and J. G. Jarvik Conventional Radiography, Rapid MR Imaging, and Conventional MR Imaging for Low Back Pain: Activity-based Costs and Reimbursement Radiology, June 1, 2003; 227(3): 669 - 680. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
