A recent report by Iglehart [1] indicated that, between 2000 and 2007, imaging was the most rapidly growing of all physician services in the Medicare population. This growth has obviously been a major cost driver for the Centers for Medicare and Medicaid Services and has focused the spotlight on imaging among policymakers and payers [2–6]. A consequence of this situation has been a variety of efforts in the last few years to rein in the cost of imaging, including such steps as the Deficit Reduction Act, an increase in the assumed utilization level used to calculate technical component reimbursement, further reduction in payments for imaging of contiguous body parts, and a decrease in the calculated practice expense per hour for imaging. More cuts may be forthcoming.

Anecdotal reports have suggested that radiology departments and groups around the country have been experiencing lower growth in demand for imaging and even decreases in recent years. We attempted to determine whether a national database showed evidence of changes in utilization trends in noninvasive diagnostic imaging.

The nationwide Medicare Part B Physician/Supplier Procedure Summary Master Files for 1998 through 2008 were examined. These files cover the more than 35 million beneficiaries in Medicare fee-for-service, but do not include those who are enrolled in Medicare Advantage plans (approximately 10.3 million in 2008). All Current Procedural Terminology, Version 4 (CPT-4) codes [7] for discretionary noninvasive diagnostic imaging in the 70,000 series were selected, as were those in the 90,000 series pertaining to echocardiography and vascular ultrasound. By “discretionary,” we mean all imaging tests that might be chosen by a referring physician as part of a workup for a patient presenting with a clinical problem. It does not include imaging tests that are mandated by the patient's medical condition. Thus, we excluded all surgical procedure codes and the supervision and interpretation codes that accompany them. Codes for imaging studies done for radiation therapy planning were excluded, as were codes for MRI and CT image postprocessing. Codes for all radionuclide imaging scans were included, but those for radionuclide-based nonimaging tests of physiologic processes and radioimmunoassays were excluded. Screening examinations such as mammography and bone densitometry studies were included.

For each CPT-4 code, the Medicare Part B Physician/Supplier Procedure Summary Master Files provide information on annual volume, specialty of the physician providers, locations where the examinations are performed, reimbursement amounts approved, and other administrative information. Physician specialties are determined by self-designated codes used by all providers in filing claims. Locations (also referred to as places of service) of the procedures are also determined by codes; the vast majority of imaging studies are performed in one of four settings—hospital inpatients, hospital outpatient facilities, physician offices, or emergency departments. For purposes of tabulating volume, we counted global and professional component claims. Technical component claims were excluded because that would have led to double counting of examinations. Utilization rates per 1,000 beneficiaries were calculated for each code by dividing procedure volume by the number of thousands of Medicare fee-for-service beneficiaries each year.

Overall trends for noninvasive diagnostic imaging were studied during the entire 1998–2008 period. We also separately studied the trends in the various advanced imaging techniques—CT, MRI, nuclear medicine (including PET), echocardiography, and other noncardiac ultrasound—because these are the ones receiving the most scrutiny. Finally, we compared utilization rate trends among radiologists and all nonradiologist physicians as a group. The latter group did not include a small number of claims that were filed by entities in which the provider specialty could not be determined, such as independent diagnostic testing facilities and multispecialty groups.

There has been a dramatic change in the noninvasive diagnostic imaging growth trends in the Medicare fee-for-service population in recent years. Before 2005, there was steady and fairly rapid growth in imaging overall, especially in the advanced imaging techniques. However, after 2005, growth trend lines flattened in every aspect we studied. The compound annual growth rate in overall utilization of noninvasive diagnostic imaging (all techniques and all places of service), which had been 4.1% before 2005, decreased to 1.4% from 2005 to 2008 (Fig. 1). Among the two highest paid techniques (MRI and nuclear medicine), Figure 2 shows there was virtually no growth after 2005. Some flattening in CT growth was also observed after 2005, with the compound annual growth rate decreasing by half. With regard to nuclear medicine, the largest single component is myocardial perfusion imaging, most of which is self-referred in cardiologists' offices. Because of the potential conflict of interest this creates, cardiologists were criticized for fomenting such rapid growth [8]. However, at this point, it would appear they deserve credit for having been able to limit the rapid growth that characterized nuclear cardiac imaging before 2005. Echocardiography and other noncardiac ultrasound examinations also experienced slackening of growth after 2005 (Fig. 3), although here again the changes were not as striking as with MRI and nuclear medicine.

The slowdown in growth of outpatient advanced imaging, as seen in Table 1, is especially important, because the ambulatory settings (both private office and hospital outpatient facilities) are where most of the concern has been focused. The data reveal that, between 1998 and 2005, there was growth in all five modalities in both private offices and hospital outpatient facilities, but the growth was substantially greater in the former setting than the latter. After 2005, there was distinct curtailment of growth in both settings for all five modalities. For four of the five modalities, the compound annual growth rates after 2005, though much lower than before, remained greater in private offices than in hospital outpatient facilities. MRI was the only exception, and there the compound annual growth rates after 2005 were approximately equal in the two ambulatory settings (1.8% and 2.0%).

The data shown in Table 2 are also of considerable interest. Between 1998 and 2005, the utilization rate of discretionary noninvasive diagnostic imaging grew approximately twice as rapidly among nonradiologist physicians (who are generally in a position to self-refer) as it did among radiologists (who are not). After 2005, utilization growth slowed in both groups, but nonradiologist physicians' growth remained approximately twice that of radiologists. By 2008, the noninvasive diagnostic imaging utilization rate per 1,000 Medicare fee-for-service beneficiaries was 1,515 among nonradiologist physicians and 2,740 among radiologists. Nonradiologists thus play a substantial role that is growing more rapidly than that of radiologists. These data refute the claims of some in other specialties [9] that radiologists control imaging and are therefore responsible for the inappropriate growth in utilization.

There are several possible causes for the trends we have described. First, one might be tempted to ascribe the flattening of the growth curves solely to the Deficit Reduction Act of 2005, which went into effect in January 2007. The Deficit Reduction Act reduced the Medicare technical component reimbursements for private office advanced imaging from the levels of the Medicare Physician Fee Schedule down to those of the Hospital Outpatient Prospective Payment System for the many procedures for which the latter level was lower. The cuts were substantial, especially for MRI and CT. These cuts very likely discouraged entrepreneurs from opening new imaging offices. Some existing private offices undoubtedly closed and others scaled back operations, although there are no precise data on this trend. However, although the Deficit Reduction Act likely had an effect, it was not the sole cause of the changes. If it had been, we would have seen reductions in the growth rates in private offices, but not in hospital outpatient facilities, because reimbursements to hospitals were not significantly affected by the Deficit Reduction Act. Instead, growth slackened as dramatically in hospital outpatient facilities as it did in private offices. Other factors are clearly at work.

Second, the recession has likely been another factor. Many patients lost jobs and insurance coverage and were no longer able to afford expensive elective imaging tests, but the slowdown somewhat predates the beginning of the recession. Third, in the last several years, a number of articles have appeared in the medical literature expressing concern about the increasing exposure of the population to radiation [10–13]. This has no doubt influenced the thinking of many physicians, but of course MRI, echocardiography, and other noncardiac ultrasound techniques do not produce ionizing radiation and their growth rates have abated as well. Fourth, both the American College of Radiology and the American College of Cardiology have developed appropriateness criteria for imaging [14–16]. As physicians around the country have become more aware of the need to limit the costs of health care, it is quite possible that many of them are exercising more caution in their ordering of imaging tests and are applying appropriateness criteria more carefully. Fifth, commercial payers have, in some cases, become more restrictive about who they will reimburse for advanced imaging tests [17]. These restrictions have generally been directed at nonradiologist physicians, who are in a position to self-refer.

Finally, in recent years, radiology benefits management companies have instituted preauthorization programs, which have been widely adopted by commercial payers. Preauthorization programs make it somewhat more difficult and inconvenient for physicians to order advanced imaging tests. Although these programs have not been applied to the Medicare fee-for-service population, they have very likely made physicians think more carefully about which imaging studies to order for various medical conditions in all their patients, or whether to order imaging at all.

It seems apparent that no single factor can account for the recent slowdown in the growth of noninvasive diagnostic imaging. The most likely explanation is that all the factors just listed have played some role in the changes we have described. Whatever the explanation, the important reality is that imaging growth has decreased dramatically in recent years and that this is a favorable development for our health care system. A discussion of what should be done to further limit growth is beyond the scope of this article. We hope that by calling attention to these trends the concerns of the payers and policymakers may be somewhat allayed and, as a consequence, there will be less downward pressure on imaging reimbursements in the future.

Address correspondence to D. C. Levin ([email protected]).

This study was supported in part by a grant from the American College of Radiology. D. C. Levin is a consultant to HealthHelp.

This is a Web exclusive article.