Much of the newest issue of Health Affairs is dedicated to putting hard numbers to the rapid proliferation and over-use of diagnostic imaging technology like computed topography (CT) and magnetic resonance imaging (MRI) scans. This research warrants special attention:  quantifying the “imaging boom” provides an important contribution to understanding  America’s health care woes.

The Scope of Growth

The lead article from Health Affairs comes from Laurence Baker and Scott Atlas at Stanford and Christopher Afendulis at Harvard. The research team notes the explosion of imaging machines in recent years, estimating that “the number of CT units [in the United States] grew more than 50 percent between 1995 and 2004” and that “the estimated number of MRI units more than doubled.” As this technology has become more widely available, it’s been used more often: the number of MRI procedures per 1,000 Medicare beneficiaries increased from 0.3 in 1985 to 173 in 2004. Use of CT scans more than doubled from 235 per 1,000 in 1995 to 547 per 1,000 in 2005. Baker et al. crunch the numbers to find that, over the years, each new MRI unit on the market led to 733 additional MRI procedures, adding $550,000 to Medicare spending annually. Each new CT unit on the market prompted 2,224 additional CT scans and tacked on $685,000 to the yearly Medicare bill.

These are striking numbers, and the shock persists when you put diagnostic imaging in the context of other medical services. In another Health Affairs study, Ariel Winter and Nancy Ray from the Medicare Payment Advisory Commission (MedPAC) note that between 2000 and 2005, the volume of services per Medicare beneficiary grew by 31 percent (in other words, the average Medicare patient in 2005 received 31 percent more care than she did in 2000). In contrast, the volume of diagnostic imaging (including MRI and CT scans, x-rays, and ultrasound) grew by 61 percent—twice as fast as broader physician services. More services means more payments, and this increase has been coupled with a doubling of Medicare spending on imaging services, from $6.4 billion in 2000 to $12.3 billion in 2006.

It’s not just Medicare, either. In a third Health Affairs study, a research team from the University of California-San Francisco, the University of Washington and the Group Health Cooperative (GHC)—a non-profit insurance company that provides care to 10 percent of Washington state residents—dissect the imaging boom within the GHC system. Looking at 377,048 patients, twenty-eight primary care/family medical centers, five specialty medical centers, two hospitals, and forty-five affiliated medical institutions, the authors found that utilization of CT imaging doubled and MRI imaging tripled between 1997 and 2006. Year to year, CT use grew by 14 percent, and MRI utilization grew by 26 percent annually. None of these increases were due to changes in the distributions of disease over time, meaning that the health plan wasn’t performing more scans because patients needed more scans. Yet year after year, patients have been getting more scans. For example, between 1997 and 2006, the median number of cross-sectional abdomen imaging tests among patients who underwent the procedure increased by 30 percent. Unsurprisingly, GHC has found that its average annual imaging cost per plan enrollee has steadily risen every year and has more than doubled between 1997 and 2006, from $229 per enrollee to $463 per enrollee.

Reasons for Growth

Clearly, diagnostic imaging services are on the rise, and have been for a while now. The reasons for this boom are pretty simple: there are more machines available to physicians and the financial incentives to over-use them are great. The connection between availability and use may seem self-evident, but it’s important to examine—and it’s real. Indeed, in their study, Baker, Atlas, and Afendulis mapped out the number of scanners and procedures for CT and MRI across U.S. metropolitan areas between 1995 and 2004. They found that, for both machines, “the places that had the largest growth in the number of units also had the largest changes in the number of procedures billed to Medicare.” When a facility has more MRIs, it uses them more often.

But why? If they don’t need more MRIs—because the incidence of disease hasn’t really increased—then why use them more often? One possibility is that the sensitivity of our diagnostic imaging has improved to the point that we can now use the machines more often, to detect more problems. Maybe diagnostic imaging has simply gotten more useful?

Probably not. The Health Affairs GHS study breaks down CT and MRI scans by anatomical region and finds steady increases in the imaging of areas like the spine, for which it’s  “hard to document clear evidence of improved accuracy or outcomes that would have prompted…a dramatic increase in imaging.” Further, in other areas where diagnostic imaging is routinely used—such as in the detection of breast cancer—its benefits are questionable. "We have yet to see any evidence that MRI improves outcomes when used routinely to evaluate breast cancer, and yet more and more women are getting these scans with almost no discernable pattern," said Dr. Richard J. Bleicher, a specialist in breast cancer surgery at Fox Chase Cancer Center in Philadelphia, in September.

Another concern for breast cancer patients is that MRI scans are too sensitive: the machines are really good at detecting matter inside the breast, but not so good at discerning what that matter actually is. If you have cancer, the scans will detect it; but if you have some other, harmless growth in your breast, the machine can’t discern that it’s not cancer. Thus the chance of false positives is high; but patients, eager to play it safe, increasingly opt for mastectomies if the scans pick up something. “We're concerned that the well-documented false-positive rate with MRIs may be leading – or misleading – women into choosing mastectomies,” said Bleicher. In other words, more sensitive scans don’t necessarily mean better medicine. Indeed, across the board, as much as 50 percent of high-tech diagnostic imaging fails to provide information that improves patient diagnosis and treatment.

Another possible explanation is the ubiquitous bogeyman of defensive medicine—the idea that doctors are scanning more and more patients to cover their tails and avoid the lawsuits that could spring up if they miss something. No doubt this happens sometimes. A 2005 survey of 844 physicians found that 92 percent reported “ordering tests, performing diagnostic procedures, and referring patients for consultation” for the sake of assurance; 33 percent reported using imaging technology in clinically unnecessary circumstances.

But this doesn’t necessarily mean that, at the national level, defensive medicine is the culprit. In the above survey, 88 percent of the physician respondents had previously been sued, and 48 percent had been sued in the previous 3 years– over-representing doctors who have suffered through malpractice scares (The doctors surveyed were all in Pennsylvania, a state notorious for a flurry of legal and political activity surrounding medical malpractice). At the national level, only 25 percent of doctors are sued annually and just 50-65 percent of physicians are sued over the course of an entire career. Defensive medicine probably contributes to the imaging boom, but how often—and how much it costs us—isn’t at all clear.  

What is clear is that diagnostic imaging machines are extraordinarily expensive: a state-of-the-art MRI or CT scanner costs a medical practice $3 million or more. This is important. If you pay millions of dollars for a piece of equipment, you better use it. Indeed, a major factor in the imaging boom is that physicians who procure imaging devices need to recoup their investments, so they use them more often. It’s probably not a coincidence that MRI and CT scans—extremely expensive scans which cost patients thousands of dollars–—are the fastest growing form of imaging. By contrast, ultrasound and x-rays, each of which costs patients only a few hundred dollars, have grown much more slowly. Within the GHC plan, for example, CT use grew by 14 percent, and MRI utilization grew by 26 percent, every year, but ultrasound use grew by only 5 percent annually.

It’s hard to overstate the role that the cost of diagnostic imaging equipment has in determining its use. For example, Winter and Ray report that equipment costs account for nearly 90 percent of the total direct cost of an MRI brain scan; by way of contrast, equipment accounts for only one-half of the direct cost of a chest x-ray.

“Direct cost” refers to the basic costs of actually performing a procedure (non-physician clinical staff, medical equipment, and medical supplies). Direct costs comprise the “practice-expense” component of Medicare reimbursement, one of the major determinants of how much doctors get paid for specific procedures. (The other considerations are the time, effort, skill and stress of a procedure and liability insurance expenses). In other words, when the direct costs are more—i.e. when a device costs more for a physician to buy, maintain, and use—doctors get paid more for using it. Literally, we’re talking about just using it, regardless of whether its necessa
ry or useful. Medicare pays providers separately for performing an imaging study (called the “technical component”) and for interpreting the results/writing a report (the “professional component”). Reimbursement for procedures is generally much larger than it is for evaluation. Winter and Ray offer an example: if the same doctor both performs a brain MRI and evaluates it, thus receiving payment for both the technical and professional aspects of the procedure, the technical payment would account for 87 percent of his total payment, and the evaluation would comprise just 13 percent of his total pull. The evaluation, of course, is where we find out if the scan was useful or necessary; but that’s undervalued in the reimbursement schedule.

Meanwhile, diagnostic imaging is drastically overvalued. Recall that, in the example above, 90 percent of the costs to the doctor are equipment costs, related to buying or leasing and maintaining the imaging device. Medicare knows this and tries to craft reimbursement rates that reflect the capital necessary to procure and maintain a machine. In designing reimbursement for equipment-heavy procedures like diagnostic imaging, Medicare looks at how much a machine costs and matches that amount to how often it is expected to be used. The assumption for all medical equipment—including imaging machines—is that it operates 50 percent of the time that practices are open. But this is just an assumption, and if it’s not accurate, Medicare can end up paying too much for specific procedures—including diagnostic imaging.

Winter and Ray report that a “MedPAC-sponsored survey of imaging providers in six markets [Boston, Miami, Greenville, SC, Minneapolis, and Orange County, CA]  found that MRI scanners are used 91 percent of the time [that a practice is open], on average and CT machines are operated 73 percent of the time, on average.” This matters because if a practice performs scans more frequently, then it has more opportunities to recoup the fixed cost of having a machine than Medicare assumes; thus Medicare is overpaying the providers for the service. Thus physicians are, all things considered, getting a pretty sweet deal with regards to reimbursement for diagnostic imaging.

Physician Who Are Not Radiologists Contribute to Boom

This explains why doctors who are not radiologists have been investing in scanning equipment, and then referring their own patients for imaging.  The Center for Medicare and Medicaid Services (CMS) has thought seriously about clamping down on this type of self-referral. In July, CMS proposed that physician groups that wanted to invest in the equipment must enroll as independent diagnostic testing facilities (IDTFs) and be subject to the quality standards that govern IDTFs . Self-referring physicians would no longer be able to operate outside of the IDTF framework.

As AuntMinnie.com explained earlier this month: “The IDTF rules include a number of provisions designed to limit on unnecessary utilization of imaging services. One such rule, the antimarkup provision, sought to take the profit out of both the professional and technical components of diagnostic tests that are billed by one entity and provided by another. The rule mandates that an IDTF must have a supervising physician onsite, and that the supervising physician must have ‘proficiency’ in the test that's being performed. Most observers interpret the rule as requiring the supervising physician to be a radiologist. “

But when CMS issued its final rules at the end of October, “the agency punted,” Aunt Minnie reports, “and the new rules for both payment systems actually may make it easier for nonradiologist physicians to perform imaging procedures in their offices. CMS said it would not implement the provision to require physician groups to enroll as IDTFs. What's more, CMS relaxed its requirements on how it defines who can be part of a practice, and thus who qualifies as a "supervising physician." Critics say that the rules make it easy for a practice to nominate one of their own physicians as the supervising physician, thereby continuing to self-refer.”

Radiologists are not happy with this decision, nor are they entirely pleased with the recent changes to the payments they receive. Congress has already been working to reduce reimbursement for diagnostic imaging through the Deficit Reduction Acts of 2005 and 2007, which cut payments for certain aspects of imaging by as much as one-third. In 2009, radiologists can expect a 2.7 percent increase in the professional component of their reimbursement; the technical component, will be cut by 5.3 percent.

In the future, physicians—as well as those who manufacture diagnostic imaging equipment–will resist all and any cuts.   But if our health care system is going to be sustainable, changes must be made in equipment costs are calculated, and the reimbursement for the technical component of  imaging must be revised further.

We have other options as well. In countries like Canada, the public sector actively regulates the proliferation of cutting-edge medical technologies, helping to reduce the risk of excess capacity leading to overtreatment. Here in the U.S., we have certificate of need (CON) laws in many states that require a facility to prove to a state board that a new MRI or CT machine is medically necessary; but the problem is that medical necessity is defined as using the machine a minimum amount of times in a year. This creates yet another to use it as much as possible. If CON laws were revised to define medical necessity as a question of how many people already needed an MRI scan and had not been able to receive one in a given community—instead of prospectively requiring scans—then there’d be one less incentive to over-use the machines.

These reforms are profoundly important. Excessive medical technology is one of, if not the, major factor driving health care costs in the U.S. In a recent post, Maggie noted that Peter Orszag, head of the Congressional Budget Office (and soon-to-be Director of the Office of Management and Budget in the Obama Administration) estimates that the proliferation and utilization of new medical technology accounts for somewhere between 38 and 65 percent of the growth in health care spending from 1940 to 1990. That’s huge—and the trend can’t continue. We’re already spending $2.3 trillion on health care every year.

Ultimately, we need to either make costly diagnostic imaging services less ubiquitous or reduce the incentives for their over-utilization. As Don Berwick, of the Institute for HealthCare Improvement and Harvard University has said, “much of the waste and delay in health care comes from mismatches between supply and demand.” Diagnostic imaging is a textbook case of this mismatch, and it’s something we need to rectify soon—otherwise the imaging boom will become a key contributor to America’s health care bust.