1 From the Department of Radiology, Thomas Jefferson University Hospital, 111 S 11th St, Philadelphia, PA 19107 (D.C.L., L.P., C.I.); and the American College of Radiology, Reston, Va (J.H.S.). From the 2000 RSNA scientific assembly. Received February 12, 2001; revision requested March 5; revision received April 30; accepted June 20.
PURPOSE: To evaluate cardiac nuclear medicine practice patterns
in different physician specialty groups to better understand
a recent rapid increase in utilization of radionuclide myocardial
perfusion imaging (MPI) and certain supplementary examinations.
MATERIALS AND METHODS: National Medicare Part B databases from 1996 and 1998 were used to evaluate utilization of four primary procedure codes for radionuclide MPI and two supplementary codes (add-on left ventricular wall motion or left ventricular ejection fraction). Utilization rates were calculated for cardiologists, radiologists, and other physicians. Other cardiac imaging for which radionuclide imaging might be substituted was similarly studied.
RESULTS: Overall utilization rate of radionuclide MPI per 100,000 Medicare beneficiaries increased 19.1%, from 4,046 in 1996 to 4,820 in 1998 (P < .001). However, for cardiologists the rate increased from 1,771 to 2,413 (36.3%), whereas for radiologists it increased from 1,958 to 2,031 (3.7%) (P < .001 for both changes). Overall utilization rate of add-on codes increased 264% from 1,006 to 3,657 (P < .001). By 1998, the ratio of these add-on examinations to primary MPI was 0.94 among cardiologists compared with 0.53 among radiologists (relative risk, 1.77; 95% CI: 1.76, 1.78). Cardiologist-performed stress echocardiography and cardiac catheterization and coronary angiography increased by 24.2% and 8.7%, respectively.
CONCLUSION: Growth in utilization of radionuclide MPI between 1996 and 1998 was almost 10 times higher among cardiologists than radiologists. Utilization of the two add-on codes increased even more dramatically. The greater use of MPI is not a substitute for other cardiac imaging.
Index terms: Economics, medical • Myocardium, ischemia, 511.1939 • Myocardium, radionuclide studies, 511.12171 • Radiology and radiologists, socioeconomic issues
In recent years, radionuclide myocardial perfusion imaging (MPI)
has become the principal method of noninvasively imaging suspected
coronary artery disease. This technique provides greater sensitivity
and specificity than does exercise electrocardiographic stress
testing alone (
1,
2). The addition of electrocardiographic gating
and technetium 99m–labeled radioisotopes, such as
99mTc
sestamibi and
99mTc tetrofosmin, have brought further improvements.
An important advantage of
99mTc–labeled compounds, aside
from providing better counting statistics for MPI, is that they
also allow determination of regional and global left ventricular
wall motion (WM) and left ventricular ejection fraction (EF)
(
2). In 1992, largely as a result of this development, two new
codes were incorporated into the nuclear medicine section of
the Current Procedural Terminology, 4th Edition (CPT-4) coding
manual (
3). Codes 78478 and 78480 for left ventricular WM and
left ventricular EF, respectively, were specifically designated
as "add-on" codes. That is, users of the manual were instructed
that these two codes were to be used only in conjunction with
one of the four primary codes (78460, 78461, 78464, or 78465)
for radionuclide MPI.
Although there is little doubt about the utility of assessing myocardial perfusion and left ventricular WM and EF by using radionuclide imaging techniques, concern has been raised about overutilization. The fiscal year 2000 work plan of the Office of Inspector General of the Department of Health and Human Services identified MPI as a medical service undergoing unusually rapid expansion in utilization, with a 23% increase in billing to the Health Care Financing Administration (HCFA), the administrator of the Medicare program, in just 1 year (4). Among the many thousands of physician services offered to patients, it was the only one specifically targeted by the Office of Inspector General for assessment for medical appropriateness.
The goal of this study was to evaluate cardiac nuclear medicine practice patterns among different physician specialty groups to better understand the rapid increase in utilization of these examinations.
Our data sources were the HCFA Physician/Supplier Procedure
Summary Master Files for 1996 and 1998. These files contain
all Medicare Part B services performed nationwide by physicians
for beneficiaries enrolled in the traditional fee-for-service
Medicare program. In 1996 there were 38.1 million Medicare beneficiaries
in the United States—33.2 million in traditional fee-for-service
Medicare and another 4.9 million enrolled in Medicare health
maintenance organizations, or HMOs. In 1998 there were 38.5
million Medicare beneficiaries—31.9 million in traditional
fee-for-service and 6.6 million others in Medicare HMOs. Because
services to Medicare HMO patients are generally capitated and
not handled directly by Medicare fiscal intermediaries, their
records are not included in these files and were therefore not
included in this study.
In the files, each physician service is classified in a number of ways. The first is by type of service by using the CPT-4 codes. A second classification is by the location where the service is performed by using one of 27 HCFA location codes. A third classification is by specialty of the physician provider by using one of 107 HCFA specialty codes. For the purposes of this study, physicians were categorized as cardiologists, radiologists (including nuclear medicine physicians), or other physicians.
Table 1 lists the CPT-4 codes that were analyzed and brief descriptors from the coding manual. The first four (78460, 78461, 78464, 78465) are the primary codes used for radionuclide MPI. The next two (78478 and 78480) are the add-on codes for determination of left ventricular WM or EF when used in conjunction with a primary MPI examination. The last four codes (78472, 78473, 78481, and 78483) are "freestanding" codes for WM and EF determination when these examinations are performed separately and not in conjunction with an MPI. These four codes are used less frequently, usually in patients with some form of heart disease other than coronary disease; aside from determining the total number of these examinations performed, we did not analyze these codes further.
fig.ommitted |
TABLE 1. Cardiac Radionuclide Imaging Codes in 1998
| |
For each of the four primary MPI CPT-4 codes and the two add-on
WM and EF codes, we first compared utilization rates during
1996 and 1998 among radiologists, cardiologists, and all other
physicians. The difference in proportions for 1996 rates versus
1998 rates was calculated by using the
z test. Since the rates
are complete counts of the entire Medicare population rather
than a sample, it might be argued that no inferential statistics
are required. However, the particular counts obtained in 1996
and 1998 can be considered theoretically as samples of a superpopulation
of samples influenced by various random factors and traditional
sampling statistics, such as the
z test, and can be calculated.
Of course, population parameters change systematically from
year to year in ways that may be associated with increased utilization—such
as the aging of the Medicare population. While it would have
been desirable to adjust for age differences, the data set utilized
does not contain demographic information, and no adjustment
was possible. Because the points are close in time, changes
in such parameters are not great, and it is reasonable to treat
these years as samples of a superpopulation. Since this confounder
could not be eliminated, we caution that our inferential statistics
should be considered descriptive rather than true tests of significance.
We also calculated the percentages of the examinations performed
by each of the three physician groups. We further analyzed the
physician utilization rates according to location of the examinations.
For this, we used the location codes for
(a) hospital inpatient
settings,
(b) hospital outpatient settings,
(c) private offices,
and
(d) a final group encompassing all other locations. Utilization
rates were calculated as the number of examinations per 100,000
Medicare fee-for-service beneficiaries that year. We then calculated
the ratios of the add-on WM and EF studies to the primary MPI
studies according to physician specialty and location to determine
if these variables influenced the utilization of WM and EF studies.
The ratios were measures of the risk that a patient undergoing
MPI would have a WM and/or EF study added. Relative risks (one
ratio divided by another) and CIs were calculated separately
for 1996 and 1998 utilization of add-on WM and/or EF studies
in all places of service for cardiologists and other physicians
versus radiologists.
Because increases in utilization of diagnostic studies like cardiac radionuclide imaging might be offset by decreases in utilization of other imaging tests that provide comparable or supplementary information, we also assessed stress echocardiography and cardiac catheterization. Cardiologists perform the majority of these procedures. We therefore compared 1996 and 1998 utilization rates among cardiologists for stress echocardiography (code 93350) and the seven codes encompassing adult cardiac catheterization and coronary angiographic procedures (codes 93510, 93511, 93526, 93539, 93540, 93543, and 93545).
HCFA uses eight "specialty" codes in which it is not actually possible to determine the medical specialty of the physician who provides the service—multispecialty clinic or group practice, ambulatory surgical center, portable x-ray supplier, clinical laboratory, independent physiological laboratory, skilled nursing facility, intermediate care nursing facility, and other nursing facility. We excluded claims filed under these specialty codes; they accounted for only 4% of all Medicare fee-for-service claims in 1998.
Data are presented in the Tables.
Table 2 demonstrates 1996
and 1998 utilization rates per 100,000 Medicare beneficiaries
among cardiologists, radiologists, and other physicians. Total
utilization per 100,000 of the four MPI codes increased 19.1%
from 4,046 in 1996 to 4,820 in 1998. However, the utilization
rate increased 36.3% among cardiologists compared with only
3.7% among radiologists. Utilization of these codes by other
physicians was considerably lower but increased 18.6% during
the 2-year interval. The total utilization rate of the two add-on
WM and EF codes increased 264% from 1,006 in 1996 to 3,657 in
1998. The growth in utilization of the latter two codes during
the 2 years was high for all three physician groups—277%
among cardiologists, 227% among radiologists, and 314% among
other physicians. Differences in utilization rates between 1996
and 1998 reported in
Table 2 all show probabilities of less
than .001 by using the
z-test. As we noted in the Materials
and Methods section, these probabilities are to be interpreted
descriptively rather than as customary significance tests.
fig.ommitted |
TABLE 2. Changes in Utilization Rates of Cardiac Radionuclide Imaging between 1996 and 1998 among Cardiologists, Radiologists, and Other Physicians in All Places of Service
| |
Table 3 is derived from
Table 2 and shows the percentages of
MPI and add-on WM and/or EF examinations performed by radiologists,
cardiologists, and other physicians during 1996 and 1998. During
1996, radiologists performed 48.4% of MPI examinations, while
cardiologists performed 43.8%. By 1998, the cardiologists’
share had increased to 50.1% while radiologists’ share
had decreased to 42.1%. However, during the 2-year interval,
the utilization rate among radiologists increased (from 1,958
to 2,031). The shift to the greater utilization proportion by
cardiologists thus appears to be due to a much more rapid increase
in their utilization (from 1,771 to 2,413), rather than to a
shift in procedure volume from radiologists to cardiologists.
fig.ommitted |
TABLE 3. Cardiac Radionuclide Imaging Performed by Radiologists, Cardiologists, and Other Physicians during 1996 and 1998 in All Places of Service
| |
Table 4 further demonstrates overall physician utilization by
categorizing it according to the place where the service was
performed. The three principal places of service where imaging
is performed are hospital inpatient settings, hospital outpatient
settings, and private offices. All other locations were grouped
together as a fourth category, but the table shows that utilization
in this category was much less than in the three principal locations.
The numeric columns in
Table 4 show utilization rates per 100,000
beneficiaries for both 1996 and 1998, as well as the percentage
change between them. For hospital inpatients, the utilization
rate of MPI increased 21.8% between 1996 and 1998 among cardiologists
(from 252 to 307) compared with 6.0% among radiologists (from
581 to 616). In hospital outpatient settings, where the utilization
of MPI was considerably higher, the rate increased 18.2% between
1996 and 1998 among cardiologists (from 396 to 468) compared
with 2.2% among radiologists (from 1,109 to 1,133). In private
offices, cardiologist utilization increased 45.8% (from 1,115
to 1,626) during the period, whereas radiologist utilization
increased 8.1% (from 223 to 241). The utilization of the add-on
WM and/or EF codes between 1996 and 1998 increased proportionately
among cardiologists and radiologists in the hospital inpatient
and outpatient settings. However, in private offices, utilization
of these codes increased 282% among cardiologists compared with
117% among radiologists.
fig.ommitted |
TABLE 4. Changes in Rates of Utilization of Cardiac Radionuclide Imaging between 1996 and 1998 by Physician Category and Place of Service
| |
Table 5 shows the 1996 and 1998 ratios of add-on WM and EF studies
to primary MPI studies. This ratio indicates the proportion
of MPI examinations to which a WM or EF examination is appended.
Since the physician performing the examination can elect to
add both WM and EF studies to a basic MPI study, the ratio can
range from 0 to 2.0. Ratios are shown for cardiologists, radiologists,
and other physicians in each of the four place of service categories.
The ratios in this table are derived from
Table 4. For example,
Table 4 shows that in 1996, the total utilization rate of WM
or EF studies was 1,006 per 100,000 Medicare beneficiaries,
while the total utilization rate of MPI that year was 4,046.
The ratio is 1,006/4,046, or 0.25. Because the WM or EF codes
can be used only in conjunction with MPI, this indicates that
approximately 25% of all MPI studies were accompanied by a WM
or EF determination in 1996. In 1998, this ratio was 3,657/4,820,
or 0.76, indicating that by then more than three-fourths of
all MPI studies were accompanied by a WM or EF determination.
In 1996, the ratio among cardiologists was 0.34 versus 0.17
among radiologists. By 1998, the ratio among cardiologists was
0.94 compared with 0.53 among radiologists. Analysis by location
shows that the highest ratios were generally found in private
offices. By 1998, the ratios among cardiologists and other physicians
in private offices exceeded 1.0.
Table 5 shows that for both
1996 and 1998, the relative risk of a patient undergoing WM
and/or EF studies is higher for cardiologists and other physicians
compared with radiologists.
fig.ommitted |
TABLE 5. Changes in Ratios of Add-on WM and/or EF Studies to Primary MPI Studies between 1996 and 1998
| |
We noted a different utilization pattern for the four freestanding
WM and EF codes (78472, 78473, 78481, and 78483) than for the
add-on codes. Claims under the freestanding codes were much
less frequent than claims for the WM and EF studies. In 1996
there were 194,585 claims for the four freestanding codes and
333,820 for the two add-on codes; in 1998 there were 178,738
claims for the former and 1,166,114 for the latter. Thus, WM
or EF determinations were much more commonly performed along
with MPI as part of the evaluation of suspected coronary disease.
The utilization rate for stress echocardiography among cardiologists increased 24.2%, from 727 per 100,000 Medicare beneficiaries in 1996 to 903 in 1998. For the seven cardiac catheterization and/or coronary angiographic codes, the utilization rate among cardiologists in 1996 was 7,318 per 100,000 beneficiaries. By 1998, this rate had increased 8.7% to 7,958. Cardiologists performed 85.3% of all stress echocardiograms and 91.7% of all cardiac catheterization/coronary angiographic procedures in 1998.
Our data provide interesting insight into the concerns expressed
about MPI in the Medicare program. Between 1996 and 1998 there
was a substantial increase (19.1%) in the overall utilization
rate of MPI. However, there was a striking difference between
the practice patterns of radiologists and cardiologists. The
utilization rate increased 3.7% among radiologists during the
2-year interval compared with 36.3% among cardiologists. As
shown in
Table 4, the most dramatic MPI increase among cardiologists
occurred in private offices, with a 45.8% increase in 2 years.
In hospital settings, the utilization increase among cardiologists
was more modest (21.8% for inpatients and 18.2% for outpatients).
It is thus apparent that a major contributing factor in the
increase in Medicare billing for radionuclide myocardial perfusion
codes was the rapid increase in utilization of MPI by cardiologists.
Overall utilization of the add-on WM and EF codes increased far more rapidly (264%) than MPI between 1996 and 1998. This is perhaps not surprising, since these studies rely on the use of radioisotopes, nuclear camera improvements, and billing codes that have been developed relatively recently. As shown in the listing of relative value units in Table 1, these studies are considerably less costly than the primary MPI studies. Increases in rates of the WM and/or EF studies during the 2-year period were 277% among cardiologists, 227% among radiologists, and 314% among other physicians. The ratios shown in Table 5 represent a more direct measure of the tendency to utilize these supplementary procedures. This table shows that the ratios for cardiologists were considerably higher than for radiologists in both 1996 and 1998, in all locations. By 1998, the overall ratio for cardiologists was 0.94 compared with 0.53 for radiologists. The relative risk that patients undergoing an MPI examination performed by a cardiologist would also undergo an add-on WM and/or EF exam was 1.77 compared with the risk if the patient was referred for an MPI examination to a radiologist.
The rapid increase in use of cardiac radionuclide imaging might be justified if it was being substituted for other examinations for coronary artery disease. However, at the same time the increases in utilization of cardiac radionuclide imaging were occurring, cardiologists’ use of stress echocardiography increased by 24.2%, and their use of cardiac catheterization and coronary angiography increased by 8.7%. Thus there was no evidence that the growth in utilization of radionuclide examinations resulted in lower utilization of these other related diagnostic studies.
MPI and the associated add-on WM and EF studies performed by cardiologists are often self-referred. The opportunity for physicians to self refer has been shown to be a potent stimulus to increased utilization of imaging studies. Hillman et al (5,6) demonstrated that self-referring physicians who operated their own imaging equipment used 2–8 times as many imaging studies as did physicians who referred their patients to radiologists. Findings of a large-scale General Accounting Office study (7) of the Medicare population in Florida showed substantially the same results. These findings have been confirmed by other study findings as well (8–11). It is not clear whether the increased utilization of imaging among self-referring physicians is due to a belief that their patients are sicker than the norm, to an enthusiasm for technology, to a desire to maximize income, or to some other motivation, but the net effect is increased cost to the health care system.
Some limitations of our study should be noted. First, although it is possible that the MPI utilization increase among cardiologists may be due to self referral within a single practice or group, our database does not allow precise determination of the degree of self referral. Second, we cannot determine whether the rapid growth resulted from higher utilization among a small group of cardiologists, or whether a larger number of cardiologists acquired nuclear cameras and began performing the examinations. Third, the data do not allow us to assess the appropriateness of the imaging examinations. However, there is no reason to assume that the populations of patients studied by radiologists, cardiologists, or other physicians are inherently different or that the latter two populations have greater need for cardiac nuclear imaging examinations. It would be difficult to ascertain whether the increased utilization detected in this study was medically necessary or not. Fourth, this study was conducted among the Medicare population only and may not exactly reflect events occurring in other health insurance databases. Fifth, there are small year-to-year changes in the underlying Medicare population demographics, which may contribute to small changes in utilization and which we are unable to adjust for. Consequently, as noted earlier, probability levels reported should be interpreted as descriptive rather than as traditional significance tests. Finally, the 107 HCFA physician specialty codes are self designated by physician providers and this may lead to minor inaccuracies. For example, in a given hospital, a cardiologist may work in the nuclear medicine section of the department of radiology, and his billings to HCFA might be classified as being from a "radiologist."
In summary, this study has provided insight into the concerns expressed in the Office of Inspector General work plan for 2000 (4). There was sharp growth between 1996 and 1998 in the utilization rate of MPI; this growth was almost entirely due to increased utilization by cardiologists, particularly in the office setting. There was an even more striking increase in the use of add-on WM and/or EF codes; however, this can be at least partially explained by the fact that these were still relatively new codes, which had been available only for 4 years in 1996. Although the increase in utilization of the add-on WM and/or EF codes was high among all physicians, by 1998 the probability that a patient would undergo one of these examinations was substantially higher if the primary MPI examination was performed by a cardiologist than if it was referred to a radiologist. The recent higher utilization seen in cardiac radionuclide imaging is not being offset by declines in use of other related imaging studies.
- Gianrossi R, Detrano R, Mulvihill D, et al. Exercise-induced ST depression in the diagnosis of coronary artery disease: a meta-analysis. Circulation 1989; 80:87-98.
- Beller GA, Zaret BL. Contributions of nuclear cardiology to diagnosis and prognosis of patients with coronary artery disease. Circulation 2000; 101:1465-1478.
- Physicians’ current procedural terminology 4th ed. Chicago, Ill: American Medical Association, 1992; 423-424.
- Health Care Financing Administration. HHS/OIG fiscal year 2000 work plan, Office of Inspector General Projects; 13 .
- Hillman BJ, Joseph CA, Mabry MR, Sunshine JH, Kennedy SD, Noether M. Frequency and costs of diagnostic imaging in office practice: a comparison of self-referring and radiologist-referring physicians. N Engl J Med 1990; 323:1604-1608.
- Hillman BJ, Olson GT, Griffith PE, et al. Physicians’ utilization and charges for outpatient diagnostic imaging in a Medicare population. JAMA 1992; 268:2050-2054.
- Referrals to physician-owned imaging facilities warrant HCFA’s scrutiny: report to the Chairman Subcommittee on Health Committee on Ways and Means House of Representatives. GAO/HEHS-95-2 Washington, DC: U.S. General Accounting Office, 1994; 1-61.
- Hemenway D, Killen A, Cashman SB, Parks CL, Bicknell WJ. Physicians’ responses to financial incentives: evidence from a for-profit ambulatory care center. N Engl J Med 1990; 322:1059-1063.
- Childs AW, Hunter ED. Non-medical factors influencing use of diagnostic x-ray by physicians. Med Care 1972; 10:323-335.
- Strasser RP, Bass MJ, Brennan M. The effect of an on-site radiology facility on radiologic utilization in family practice. J Fam Pract 1987; 24:619-623.
- Levin DC. Merrill C. Sosman Lecture: the practice of radiology by nonradiologists—cost, quality, and utilization issues. AJR Am J Roentgenol 1994; 162:513-518.
作者:
David C. Levin MD Laurence Parker PhD Charles 2007-5-14