| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Health Policy and Practice |
1 From Direct Research, Vienna, Va (C.H.); and Department of Research, American College of Radiology, 1891 Preston White Dr, Reston, VA 20191-4397 and Department of Diagnostic Radiology, Yale University, New Haven, Conn (J.H.S.). Received March 20, 2003; revision requested May 28; revision received August 5; accepted August 22. Address correspondence to J.H.S. (e-mail: jonathans@acr.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: In 1999, the American College of Radiology surveyed radiology practices by mail. The final response rate was 66%. Weighting was used to make responses representative of all radiology practices in the United States. Self-reported financial ratios (payments, charges, accounts receivable turnover) were analyzed; 449 responses had usable data on these ratios. Comparison with results of a similar 1992 survey and combined analysis with Medicare data on billed charges provided information on trends.
RESULTS: All measures of payment collections declined sharply from 1992 to 1999, with the gross collections rate (revenues as percentage of billed charges) decreasing from 71% to 55%. Average payment for a typical radiology service decreased approximately 4% in dollar terms or approximately 19% in inflation-adjusted terms. In 1999, nonmetropolitan practices appeared to fare better than others. Among insurers, Medicaid stood out as a low and slow payer, but neither managed care nor Medicare had a consistent effect on financial ratios. The gross collections rate varied substantially across geographic areas, as did, in an inverse pattern, the level of billed charges. One-quarter of practices had accounts receivable equal to 90 or more days of billings.
CONCLUSION: The opposing geographic pattern of billed charges and gross collection rate suggests that geographic variation in the latter is driven more by variation in billed charges than by variation in payment levels. Radiologists saw a substantial decrease in the real (inflation-adjusted) value of payment per service during the 1990s. The large fraction of practices with accounts receivable of 90 or more days of billings—a level considered potentially imprudent by financial management advisors—suggests that many practices should improve financial management and that state prompt-payment laws have not had a substantial positive effect.
© RSNA, 2004
Index terms: Economics, medical • Radiology and radiologists, departmental management • Radiology and radiologists, socioeconomic issues
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
Monitoring financial performance need not be difficult. A few financial ratios are typically used to gauge financial performance, and these ratios quantify common-sense questions about the practice: How do amounts collected compare to amounts billed? What fraction of the amounts due does the practice actually collect? How long must the practice wait, on average, between submitting a bill and getting paid?
The purpose of our study was to evaluate variation in these financial measurements for radiology practices nationwide and to examine trends in these measurements and in payments.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
After out-of-scope responses and responses with important incompleteness were eliminated by the lead author of reference 3 and C.H., the overall, nonduplicated survey response rate was 66%—close to the 69% response rate for the previous (1992) ACR Survey of Practices (4,5).
Each responding practice was classified into one of 28 strata—seven practice size categories in each of four census regions—and responses were then weighted by the lead author of reference 3. To make the weighted data representative of all U.S. radiology practices, responses in each stratum were weighted by ENPUS/NRS, where ENPUS is the estimated number of practices in the stratum in the United States and NRS is the number of responses in the stratum (1–3).
After weighting was performed, several types of practices were excluded from all or part of the analysis in this study. First, the analysis was limited to diagnostic radiology–only practices; we excluded 43 responses from practices that provided some radiation oncology services. Second, data from 12 responding practices at government facilities were eliminated because much or all of the financial ratio information from these practices would be irrelevant to this study, given that such practices generally do not bill for their services. Third, data from solo practices were included only in portions of the tables and were excluded from the regression analyses. Because the number of responses from solo practices was small (only 38), the weighting method would have given very large weights to the solo practices. Commingling these heavily weighted observations with those for the multiradiologist practices might have distorted the results and definitely would have degraded the statistical precision of the estimates. Fourth, more than 100 practices provided missing, incomplete, or inconsistent information on the sites at which they practiced, and these responses were deleted.
In total, 449 responses from solo and multiradiologist practices were included in some part of the analysis. Nonresponse or inconsistent response to individual survey items further reduced the number of respondents included in the regression analyses. No weighting adjustments were made to account for item nonresponse.
Definition of Study Measures
For the purposes of this study, the gross collection rate was defined as net revenue divided by gross charges. Net revenue is the money the practice actually collects for provision of medical care—that is, total payments from all sources (both insurers and patients), minus refunds. Gross charges are the total amounts billed. The gross collections rate reflects the fraction of billed charges the practice collects from all payment sources (both insurers and patients). If insurers paid practices’ billed charges and patients paid all coinsurance and deductible amounts due, the ratio would be 1.0. In reality, because few insurers actually pay charges and because practices incur bad debt, this number is usually substantially less than 1.0.
The net collection rate was defined as net revenue divided by net charges. Net revenues are defined, as above, as the amounts the practice actually collects. Net charges, by contrast, are the amounts the practice ought to collect, absent bad debt, unpaid claims, and similar factors. Technically, net charges are billed charges minus any contractually agreed or legally mandated adjustments. A practice that succeeds in collecting all payments due will have a net collection rate of 1.0. The rate falls below 1.0 due to bad debt, deliberate noncollection of patient coinsurance amounts, non- or partial payment by insurers, and/or other factors preventing collection of all amounts owed to the practice.
The net-to-gross charge ratio is the ratio of net charges to gross charges. This reflects the gap between the practices’ billed charges and the payers’ allowed amounts, fee schedules, or contractually agreed-upon discounted fees. A low ratio could reflect either a situation in which insurers have low payment rates or a situation in which practices have high billed charges.
Accounts receivable turnover (in days) was defined as total accounts receivable (ie, charges on bills submitted but not yet paid) divided by a practice’s average daily charges. A low number indicates that a practice is being paid rapidly, while a high number indicates that a substantial average lag time exists between a practice’s submission of a bill and its being paid.
Practices were categorized by several characteristics, including type of practice, size, region, urban-rural location, and type of setting served (3).
Three measures of managed care (6) were used. First, practice members were asked to report their percentage of practice in each of nine payer categories. "Percentage managed care" was calculated as the sum of the reported percentage in health maintenance organizations (HMOs) and the reported percentage in preferred provider organizations. This percentage-of-practice variable was edited for internal consistency as described below.
Second, members of practices were asked about the extent to which managed care had changed their practice. The five possible responses (from "none" to "greatly") were coded linearly from 0% to 100%. This was a subjective, broad question. This perceived impact of managed care might encompass both financial aspects (low payment) and nonfinancial aspects (care management, paperwork burden, or "other").
Third, the HMO penetration rate in the practices’ local area—obtained from 1998 data calculated by Wholey and colleagues at the University of Minnesota (7)—was included. The HMO penetration rate was defined as the proportion of the area population enrolled in HMOs. The data of Wholey et al (7) were ultimately based on data collected and made available annually by InterStudy (Minneapolis, Minn), a private firm that specializes in collecting information on the managed care industry. The HMO penetration rate was calculated for the metropolitan statistical area for practices in metropolitan statistical areas and for the county for practices outside such areas. We used three managed care variables rather than one because these three variables are quite distinct measures and are only slightly correlated (6).
Supplemental Data Sources
Two sources of information on physicians’ billed charges were used to supplement the survey data. For one, data in the publicly available Medicare Physician-Supplier Procedure Summary File data files (formerly known as Part B Medicare Annual Data, or BMAD, and purchasable from the federal agency that operates Medicare, the Centers for Medicare and Medicaid Services in Baltimore, Md) were used to measure changes in radiologists’ billed charges between 1992 and 1999. These files summarize all fee-for-service bills submitted to Medicare for services provided during these years. Billed charges on Medicare bills should be a reasonable proxy for billed charges for all patients because the price physicians bill for a service (not the amount they actually collect) is normally the same for all patients.
Second, publicly available 2001 Medicare claims data for a 5% sample of fee-for-service beneficiaries (purchasable from the Centers for Medicare and Medicaid Services) were used to investigate geographic variation in billed charges. Unlike the summary file used to track trends, this file identifies states and counties and can be used to assess urban-rural and other geographic differences in billed charges. This claims-level file contains charge data for about 4 million services performed by Medicare-participating diagnostic radiologists. A county-level price index was constructed from the billed charge data on the basis of the state and county of the patient’s residence. The resulting index data were aggregated to yield case-mix–adjusted estimates of typical submitted charges by region and urban-rural location, which we present in the Results section.
For two reasons, these Medicare billed charge data are not an exact match with the survey data. First, although the survey results are practice weighted (reflecting the behavior of the typical radiology practice), Medicare claims reflect the volume of care provided. Large practices have a greater influence on the claims data. Second, patient border crossing confounds the urban-rural contrast in the Medicare data. The county recorded on the claim is the county of patient residence. Border crossing by patients tends to blur differences in charge levels when patients travel for care—for example, when rural patients receive care from urban radiologists. The Medicare data should nevertheless provide a reasonable indication of the direction and size of differences in billed charges across geographic regions.
Data Analysis and Statistical Methods
Key variables were edited to remove responses that were incomplete or logically inconsistent as follows: Respondents were asked to report the percentage of their practice that occurred in each of a number of settings and the percentage of practice from each type on a list of payer types. Where the sum of these percentages was below 80% or above 125%, the reported response was removed, and the response was instead recorded as "missing." Otherwise, the individual percentages were divided by (the sum/100%) to make the adjusted percentages add up to 100%. This data editing and all data analyses were performed by using SAS System software release 8.01 (SAS Institute, Cary, NC).
For the financial ratios, extreme or illogical values were removed by C.H., with the assumption that these probably reflected misinterpretation of the questions or clerical errors. For example, accounts receivable amounting to 1 day’s charges were also changed from the actual response value and instead recorded as missing. Net collections in excess of 100% (ie, collections exceeding the amount actually owed) were also recorded as missing values. Responses indicating a gross collections rate that was higher than the net collections rate (ie, a gross collections rate reflecting no contractual discounts from insurers but instead apparent contractual bonuses) were recorded as missing. For any given financial variable, approximately 10%–15% of otherwise eligible responses were set to missing in this fashion.
For all four financial ratios, we present descriptive statistics (means and percentiles of the distribution) and results of multivariate statistical analyses. Data from sample surveys are subject to sampling variability (8). The usual measure of sampling variability is the standard error, which we present—for means—in many of the tables of descriptive statistics. In general, the 95% CI extends from approximately 2 standard errors below the reported statistic to 2 standard errors above it.
The multivariate analyses enabled us to sort out the independent importance of each of several factors, with all other factors being studied held constant. For example, practice size, urban location, and percentage of practice in managed care are all correlated. The simple tabulation of financial ratios according to these individual characteristics would not reveal the extent to which each factor independently contributes to financial condition, while multivariate regression analysis does enable the independent attribution of the effects of each factor. For this reason, discussion of the correlates of financial condition is based on the multivariate regression analysis results, not the tables of descriptive statistics.
Ordinary least-squares regression analysis, rather than other regression procedures, was used because the regression coefficients it yields are far more readily understood and because we wanted to compare the amount of variation in each of the four financial measures that is explained by the factors we studied. Such comparison is appropriate only if the same model is used for all the dependent variables being compared, and a departure from ordinary least-squares regression would call for different models for different dependent variables. The regression analyses included variables that are not included in the tables of descriptive statistics: the managed care variables, the practice’s mix of payers, and practice ownership. Results were considered statistically significant if they met the P < .05 criterion given the assumption of normality—that is, if there was less than a one-in-twenty chance of observing such an effect purely by chance if the residuals in the regression analysis were normally distributed.
We compared previously published 1992 gross and net collection rates (5) with those reported in the 1999 survey. We present estimates of changes in payment per service for a fixed mix of radiologists’ services derived from information on the gross collections rate and Medicare billed charges. The estimate was derived as follows: By the definition of the gross collections rate, the total payment is the product of billed charges times the gross collections rate. Because the billed charges and payments refer to the same services, we can divide the previous equality by the number of services to obtain a second equality: Payment per service equals gross collections rate times billed charge per service.
From our 1992 and 1999 survey data on the gross collections rate, we know what the change in this rate was. The 1992 and 1999 Medicare data on charges per service, converted into an index of billed charges per service, show how billed charges per service changed. Combining the information on changes in the gross collections rate with the information on changes in billed charges per service and applying the second equality reveals aggregate trends in per-service payment for radiologists.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
|
|
|
|
|
The net-to-gross charge ratio had univariate variations similar to those of the gross charge rate (Table 6). The average of 65% means that the amount practices ought to collect (ie, net charges, which are billed charges net of agreed-to adjustments and discounts) was about two-thirds the level of billed (gross) charges. The net-to-gross charge ratio was higher for solo practices and nonhospital-only practices. The ratio was lowest in the Northeast and California and was above average in the West excluding California.
|
The accounts receivable turnover averaged 69 days across all practices (Table 7). That is, radiology practices, on average, had bills awaiting payment that amounted to somewhat over 2 months worth of charges. Payment was slower, with an 80-day average turnover, for practices in the Midwest.
|
The regression analysis revealed some statistically significant predictors of accounts receivable turnover (Table 2). A higher percentage of practice in Medicaid predicted higher accounts receivable turnover (ie, slower average time to payment). Nonmetropolitan location and location in the Northeast predicted lower accounts receivable turnover (ie, faster payment). Practices classified in type as "other" (a very small, miscellaneous category with fewer than five responding practices) had markedly lower accounts receivable turnover. Larger practices had lower accounts receivable turnover; however, the larger the practice, the smaller this effect of size.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
Not only do our data indicate that radiology practices’ gross collections rate (revenue as a percentage of billed charges) decreased by a large amount—from 71% to 55%—between 1992 and 1999, they also show that payment per service (for a fixed mix of services) decreased approximately 4% in dollar terms and 19% in inflation-adjusted terms. These estimates of payments per service are based on data from two surveys and administrative data and therefore may not be particularly precise. But they are accurate enough to show that any changes in payment levels (in dollars) were, on average, quite small during this period and were dwarfed by the overall increase in the general price level from 1992 to 1999. We are therefore reasonably confident that the inflation-adjusted (ie, "real") payment level for radiologists’ services declined substantially from 1992 to 1999—by approximately 19%.
The observed reduction in payment rates for radiologists is consistent with payment changes recently measured for all U.S. physicians in aggregate. The Medicare Payment Advisory Commission, or MedPAC, estimated that average private insurers’ payment levels for all physicians decreased slightly in dollar terms from 1994 to 2001 (9). MedPAC attributed the decline primarily to the shift of private enrollment out of indemnity plans and into various managed care plans. Thus, the trend in payments to radiologists was similar to the trend in fees paid to physicians generally.
A decrease in net collection rate—from an average of 86% to an average of 78%—contributed in a major way to this decline in payments. Not only did insurers reduce the amounts they purportedly were willing to pay, but collecting amounts nominally due appears to have become more difficult as well. Increased bad debt burden (patient nonpayment) is probably not the issue, because economic growth was high and unemployment low throughout this period. This suggests instead that changes in insurer behavior, such as partial payment, nonpayment, claims denial, and requirement of increasing levels of documentation, have contributed to the reduced net collections rate. This is certainly consistent with the ongoing widespread discussion of the "hassle factor" of managed care and with the resulting passage of prompt-payment laws in 47 states (10). Findings of what, to our knowledge, is the only detailed published study of the effect of these laws—a study which, however, is limited to one state (New Jersey)— show that insurers have not complied with prompt-payment laws and that the problem has remained severe despite the passage of these laws. Our finding is that about 40% of the overall decline in the gross collections rate can be attributed to the reduction in net collections—that is, to a reduced ability to collect money due for services. The remainder is attributable to reductions in insurers’ allowed amounts or fee schedules relative to amounts billed.
It is our impression that radiologists do not appear to regard their income as having decreased by as much as we found payment levels to have decreased. Radiologists are correct in this regard because workload has increased, offsetting the decline in payment rates. Our previously published data (11,12) indicate that the average radiologist’s workload (measured in relative value units) increased by approximately 28% from 1992 to 1999. Thus, inflation-adjusted income has been maintained—indeed, it has probably increased by a small amount. However, this has been achieved at the cost of a much-increased workload that has left radiologists feeling overworked. In a 2000 ACR survey, fully 51% of radiologists, recognizing that income depends largely on work done, responded that they had too much work, while only 5% said they had too little work (2).
Geographic Variations in Financial Ratios
Given the general impression that rural practices are particularly difficulty plagued, in part because of the fact that rural areas have a disproportionate share of poverty, we were somewhat surprised to find that nonmetropolitan practices fared much better than others in three of the four financial measures. Even accounting for their size, region, payer mix, and other characteristics, we found that nonmetropolitan radiology practices were paid a substantially higher fraction of their billed charges than other radiology practices and were paid more quickly than other practices. Their higher gross collection rate mainly reflected lower charges, not higher payments by insurers.
In general, practices in the Northeast reported low gross collection rates but fairly rapid payment (low accounts receivable turnover). Practices in the Midwest and West excluding California reported relatively high gross collection rates. Again, these variations in gross collection rate appeared to be due more to variation in billed charges than to variation in insurers’ payments.
In general, geographic variations in the gross collection rate may reflect either differences in the level of payments (the numerator of the gross collection rate) or differences in billed charges (the denominator of the rate). The pattern found in billed charges—namely, lower charges in rural areas and in the West excluding California, and higher billed charges in the Northeast and California—closely parallels (but is the inverse of) the pattern of the gross collection rate. Thus, variation in billed charges appears to account for most of the geographic variation in the gross collection rate.
Given that relatively high billed charges generally result in an offsettingly lower gross collections rate, the opportunity for increasing revenue by increasing charges appears quite limited. This finding of our study accords with current institutional realities in that most payers pay according to a fee schedule, not according to billed charges. However, the opportunity to increase revenue by increasing fees is not entirely absent, because some patients "go outside network" and thus are liable for the full amount of billed charges. Also, if billed charges for any procedure are less than any fee schedule amounts, revenue is lost thereby.
Accounts Receivable
Physician practice management advisors often use a 90-day threshold to flag practices with excess accounts receivable (13). That is, they believe that practices for which accounts receivable equal or exceed 90 days worth of billings should be examined for possible problems in claims processing or in the general financial management of the practice.
We found that one-quarter of practices were at or above the 90-day threshold for accounts receivable. Some of this is almost certainly due to insurers’ behavior, not practice management. For example, for each additional percentage point of a practice’s patient group that consisted of Medicaid patients, the accounts receivable balance was larger by approximately half a day. This notwithstanding, the high proportion of practices at or exceeding the 90-day threshold suggests the potential for many practices to improve their financial position by paying greater attention to billing practices and financial management. Concentrating on dealings with payers may both reduce accounts receivable and bring actual collections closer to what is due—that is, increase the net collection rate.
On the assumption—which seems reasonable and not overly demanding—that good financial management can achieve what the best 25% of practices achieve, practice members might set themselves targets of 50 days or a bit longer for accounts receivable turnover (in contrast to the observed mean of 69 days) and of 92% for the net collections rate (in contrast to the observed mean of 78%).
There were some limitations to this study. It was based on survey data and is subject to the uncertainty that that implies. We addressed differences in response rate associated with practice size and region through weighting, but we were unable to measure the extent of nonresponse bias any further because no information other than address was available for the approximately one-third of surveyed practices that did not respond.
The results reported here are self-reported measures of practice financial performance that are based on respondents’ best recall of relevant financial data. For each of the financial ratio questions, missing, out-of-bounds, or otherwise unusable responses accounted for approximately 10%–15% of survey responses. This is a high rate of item response error and suggests the need for some caution in assessing the apparent accuracy of the results. Nonetheless, our data appear adequate for capturing substantial trends and cross-sectional differences in financial data.
Confidence in the soundness of our results for these purposes is enhanced by two findings. First, there is reasonably good agreement between our data on the 1992–1999 trends in actual payments to radiologists and comparable Medicare Payment Advisory Commission data on payments to all physicians. Second, there is similarly good broad agreement between our data on the trend in radiologists’ billed charges and the consumer price index for all physicians’ charges, which had a 30% increase during that period (compared with the 24.5% increase we calculated for radiologists). Nonetheless, our data probably are somewhat less accurate than the standard errors alone imply.
Finally, the world has changed since 1999, particularly regarding the impact of managed care. The payment decreases outlined in this report reflect the heyday of the managed care revolution and the growth of HMOs and other forms of heavily managed care. But the percentage of elderly persons in HMOs began decreasing in 1998, and the percentage of employed persons in HMOs declined in both 2001 and 2002 (14). By all accounts, managed care plans have become less restrictive in their utilization management and contracting practices. Thus, the declines observed through 1999 may give an overly pessimistic picture of current payments for radiologists’ services. On the other hand, with total U.S. health spending again increasing rapidly and with employers, who pay most health costs, having seen profits decrease in the economic downturn, insurers may soon resume intensive downward pressure on the prices they pay physicians.
We believe that the relatively large number of practices surveyed, the relatively high response rate achieved, the elimination of incorrect and implausible responses, and the weighting by practice size category in each geographic region make our data reasonably representative of all radiology practices in the United States.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, C.H.; study concepts and design, J.H.S., C.H.; literature research, J.H.S., C.H.; data acquisition, J.H.S., C.H.; data analysis/interpretation, C.H.; statistical analysis, C.H.; manuscript preparation, C.H.; manuscript definition of intellectual content, revision/review, and final version approval, J.H.S., C.H.; manuscript editing, J.H.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSION REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. P. Krestin, J. C. Miller, S. J. Golding, G. G. Frija, G. M. Glazer, H. G. Ringertz, and J. H. Thrall Reinventing Radiology in a Digital and Molecular Age: Summary of Proceedings of the Sixth Biannual Symposium of the International Society for Strategic Studies in Radiology (IS3R), August 25 27, 2005 Radiology, September 1, 2007; 244(3): 633 - 638. [Full Text] [PDF]
|
|
|
|
 |
|
 R. S. Lewis, J. H. Sunshine,, and M. Bhargavan A portrait of breast imaging specialists and of the interpretation of mammography in the United States. Am. J. Roentgenol., November 1, 2006; 187(5): W456 - W468. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Merewitz and J. H. Sunshine A Portrait of Pediatric Radiologists in the United States Am. J. Roentgenol., January 1, 2006; 186(1): 12 - 22. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Sunshine, R. S. Lewis, and M. Bhargavan A Portrait of Interventional Radiologists in the United States Am. J. Roentgenol., November 1, 2005; 185(5): 1103 - 1112. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Ondategui-Parra, J. G. Bhagwat, K. H. Zou, E. Nathanson, I. E. Gill, and P. R. Ros Use of Productivity and Financial Indicators for Monitoring Performance in Academic Radiology Departments: U.S. Nationwide Survey Radiology, July 1, 2005; 236(1): 214 - 219. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Sunshine and Y. S. Cypel Types of Procedures Performed by Diagnostic Radiology Practices: Past Patterns and Future Directions Am. J. Roentgenol., November 1, 2004; 183(5): 1193 - 1201. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
