Premerus, a new teleradiology company focusing on subspecialized reads, is furthering the debate over whether radiologists should become more subspecialized, suggesting that subspecialists can lower health care costs by decreasing error rates.The ACR and other experts on diagnostic errors respond.

The trend in subspecialization—and teleradiology with subspecialization—is growing and includes established subspecialized teleradiology companies.

Joining the subspecialization teleradiology marketplace is Franklin, Tenn-based Premerus, which recently conducted a study involving 160 radiology exams that shows a 44% difference in the interpretations of general radiologists and subspecialists. While the company noted that reviewing clinicians in their study said that the subspecialist interpretations would not have significantly changed the patients’ management, Premerus’ analysis suggests that subspecialist interpretations can reduce diagnostic error rates and lead to significant health care savings by curtailing unnecessary exams. The ACR has seen only a summary of the study, but is cautious about drawing any conclusions.

Although Premerus’ data is preliminary, its initial findings are furthering the debate about whether radiologists should niche themselves into one or two areas of expertise.

Radiology in the New Millennium

The subspecialization debate is not new. A 2000 editorial in the American Journal of Roentgenology by Philip O. Alderson, MD, of the Department of Radiology at Columbia University, suggested that radiologists find ways to keep up with the ever expanding sophisticated technology without losing their versatility. In the same editorial, he not only foretold the expansion of cutting-edge imaging techniques, but also saw the potential of teleradiology turning radiology into a commodity and isolating radiologists from fully communicating with referring physicians.1

Eight years later, the expansion of the Internet, PACS, teleradiology, and new techniques and modalities have all come to pass and are now putting pressure on radiologists to focus on one thing and to do it well.

More recently, Scott W. Atlas, MD, professor of radiology and chief of neuroradiology at Stanford University Medical School, made a passionate plea for subspecialty-trained radiologists to be reading exams that are ordered by subspecialist-referring doctors. “To continue having non-subspecialty-trained radiologists interpreting sophisticated, complex imaging studies on patients with diseases that are virtually always cared for by subspecialist-referring doctors is unacceptable patient care.”2

Naturally, all physicians want to provide quality, cost-effective, and accurate medicine, and administrators and radiologists are spending time and resources to make that happen. Premerus’ theory is that becoming more subspecialized should be a part of that effort.

Is Subspecialization an Answer to Reducing Errors?

Is radiology subspecialization an answer to reducing errors? For some areas of radiology, there is at least some evidence to support that theory.

Premerus executives point to several peer-reviewed papers, but most notably to a 2002 study published in Radiology, which showed subspecialists’ cancer detection rate was approximately 76% higher with screening mammograms than the general radiologists. The same study showed that sub-specialists’ early cancer detection rate was approximately 77% higher with screening mammograms than the general radiologists.3

The company also cites a more recent 2007 article in Clinical Radiology that compared the neuroradiology second opinion reports for CT and MRI imaging procedures to the original general radiologists’ interpretations. The authors found a 34% discrepancy rate and concluded, “There is a significant major discrepancy rate between specialist neuroradiology second opinion and general radiologists. The benefit of a formal specialist second opinion service is clearly demonstrated; however, it is time-consuming.”4

Norman A. Scarborough, MD, a radiologist and medical director of Premerus, said that it was these and other studies that prompted the company to do its own research in comparing general radiologists to radiology subspecialists in areas outside of mammography.

“We tried to find those kinds of studies that looked at [comparing general radiologists and subspecialists], and there weren’t that many out there,” he said. “That’s what prompted us to perform our own study and see what that would demonstrate.”

Premerus’ Study

Robert M. Wachter, MD, professor and associate chairman, Department of Medicine, University of California, San Francisco, has written several books on errors in medicine. Though he has not seen the details of the Premerus study, he supports their exploring whether subspecialty reads would be more reliable and cost-effective.

“What Premerus is doing, which is an interesting concept, is asking the question: Are some errors in radiology made because, for example, in a small community hospital, you have one radiologist reading all sorts of different kinds of radiographs? One at least would wonder whether an expert who reads only head MRIs or mammograms might do a more reliable job than the generalist who’s doing everything,” Wachter said. “Up until 5 years ago, that was an uninteresting question because you couldn’t have all of those experts sitting in a reading room at a 150-bed community hospital. But now with teleradiology technology, it’s a very interesting question.”

The other question that Premerus set out to answer was, if more accurate, would the subspecialty interpretations also have reduced health care costs.

Premerus outlined its methods and preliminary findings for Axis Imaging News. For space considerations, we further summarized this information below. Readers should keep in mind that Premerus stated that its results are preliminary, that they have so far not submitted their study for peer-reviewed publication, and that the data is based on a small sample (160 MRI and CT cases). However, they did say that they may perform a larger study in the future.

In an e-mail, the company also stated that it used “Thomson Reuters clinicians and health services researchers to review the clinical findings and costing methods employed by Premerus. Thom-son Reuters reviewed a random sample of the cases and agreed with the Premerus staff estimate of unnecessary cost for each of the sample cases, indicating that the methodology used was reasonable.”

Premerus’ Methods

  • Premerus took 160 CT and MRI procedures from a single health plan in two states that were initially reviewed by general radiologists. These cases were then distributed and reinterpreted by subspecialists according to their area of expertise. The subspecialists were provided with the original clinical history and images for each case, but were not given the original interpretation results.
  • The subspecialist readings were compared to the original general radiologists’ reports. Reports that were not in agreement were labeled as nonconcordant and examined by clinical specialists in the appropriate fields of medicine (neurology, orthopedic surgery, family practice, general and cardiothoracic surgery) to assess potential differences in patient management based on the two reports.
  • As much as possible, the available clinical information regarding the patients’ care after the initial imaging was also collected. This included outpatient progress notes, subsequent imaging reports, and biopsy results.

Preliminary Results

  • 44% of the reports had disagreements between the original general radiologist’s reading and the subspecialist’s interpretation. These nonconcordant cases were categorized as false positive, false negative, or equivocal.
  • Costs for additional imaging and laboratory tests were compared.
  • Overall, the clinician reviewers found the subspecialist reports to be more definitive and less likely to recommend follow-up procedures.
  • There were three cases in which the original interpretations were found to be more accurate than the subsequent subspecialty interpretations. One case was a true negative and two were true positives.
  • It should be noted that the reviewing clinicians did not feel that the subspecialists’ reports would have resulted in a significant change in the patients’ management.

Fiscal Analysis

  • An examination of the insurance claims data for nonconcordant cases was performed. Procedures that appeared to be the result of these nonconcordant interpretations were identified with their associated costs.
  • “The actual cost found to have been unnecessary within the case population was $52,407.18, which, divided by the case total, yields $327.54 of unnecessary cost per scan reviewed. This was trended to late 2007 dollars to $371.85.
  • “The results were applied to the insurance population from which the study’s cases were generated and showed a potential savings by the health plan of $4.45 per member per month. There were 575,000 members in the health plans for these two states. The potential annual savings would have been more than $27,000,000.

These early results were presented at a May 2008 meeting of the American Medical Informatics Association. Premerus also said that it may add data and/or conduct similar studies to confirm its conclusions before submitting any material to a peer-reviewed publication.

Skeptical of Results

James H. Thrall, MD, chairman of the Board of Chancellors of the American College of Radiology (ACR), Reston, Va, has seen only a summary of Premerus’ study, but is skeptical of its results.

“I think it would be risky to draw many conclusions from it,” Thrall said. “For example, it is well known from literature going back 50 years and more that substantial discrepancies arise in the interpretation of imaging examinations. They have chosen to compare subspecialists with generalists, but how about comparing subspecialists with each other? What is the discrepancy there? And without that information, how can you say that the discrepancy between subspecialists and generalists is any greater or lesser than the discrepancy between specialists?”

Thrall also cast doubt on Premerus’ 44% discrepancy rate between generalists and subspecialists. “That is a higher discrepancy rate than I have seen in the literature, and it sounds like a very small sample. So, if Premerus or anyone else thinks that subspecialization is a good thing, why frame it around such limited and potentially faulty data? Why go out on such a slender limb if they think their case is so solid? Why not bring us much more substantial data if they think that data is available?”

Premerus recognizes that the sample was small. Gregg P. Allen, MD, chief medical officer of Premerus, said that another study was under way and they may combine the results. Allen also encourages others to conduct similar and larger studies.

“I would strongly encourage that,” Allen said. “I would be very happy if that were to happen. It’s work that has not been done in radiology as a whole, with someone looking how to improve these things.”

Premerus is just launching its subspecialized teleradiology service in a pilot program and says that it will be implementing rigorous peer-review and quality control programs, as well as giving their radiologists both clinical and cost outcome data about their patients.

The Debate Continues

When Thrall was asked if it is not intuitive that subspecialists would be less prone to errors in a particular area of expertise, he said, “I think there are many nuances to looking at that. You have to put it in the context of the practice.” Thrall added that he would not cast the issue as a generalist versus subspecialist situation.

“I would cast it as, are there situations where you might want a second opinion? The answer is yes. Are there very complicated cases coming out of tertiary medical centers where you’d want a subspecialist? Yes.”

Wachter has not performed any studies comparing generalists with subspecialists, but thinks Premerus’ premise is logical. “One assumes that someone who does this [exam] a lot, who’s seen 100 plump adrenal glands, not three, and who’s seen the results of what happens when you work them up, can say to me, ‘Yes, the adrenal gland looks a little big on the right, but it’s not anything you need to worry about because I’ve seen this 100 times and it never turns out to be anything.’ That’s really helpful,” Wachter said. “It’s not a misdiagnosis that the other person might have made, but it’s a more accurate reading of the radiograph and it helps me to know how best to manage the patient and not waste additional resources tracking things down.”

As to the question of whether radiologists should be or need to be more specialized in the age of teleradiology and unique imaging technology and techniques, Thrall sees continuing education—for all radiologists—as being most important for keeping up with new technology and techniques.

At the same time, Thrall recognizes the growing tendency of radiologists to focus on one area of radiology and that more group practices are consolidating and organizing themselves so that members can focus in one area. “It’s not clear to me how many people are practicing as ‘true general radiologists’ anymore,” he said. “I actually think from the data that I’ve seen that a very substantial percentage of radiologists have one or two special areas of expertise where they spend a lot of their time.” He also estimates that 75% to 80% of new radiologists are receiving additional fellowship training.

There is also the question of the definition of a “subspecialist.” Thrall noted, “There are some fellowships that are associated with certificates of added qualifications, provided by the American Board of Radiology (ABR), but then there are other fellowships like musculoskeletal where there is no board certification related to that fellowship. So, do you even recognize these different classes of training the same way, because one has a certification and the other doesn’t? It’s really nuanced with lots of shades of gray.”

Premerus said that it is addressing this issue by requiring its radiologists to have 3 to 5 years of experience in a particular field and have at least 10,000 studies in their declared subspecialty. They also prefer fellowship training, but do not require it.

Finally, Thrall emphasizes that the current crop of general radiologists have been examined in 11 different subspecialty areas in an oral exam, and required to pass each one in order to become board certified. “They can’t pass 80% of them, but have to pass every subspecialty oral examination in order to get board certified,” he said.

On the other hand, Premerus noted that starting in 2010, the ABR’s final board examination given to radiologists will be tailored to reflect the training emphasis, experience, and corresponding planned practice emphasis of the individual resident, who will have the option to specify up to four areas of concentration.

Of course, the resident also can opt to be examined only with general radiology modules instead.


Tor Valenza is a staff writer for Axis Imaging News. For more information, contact .

References

  1. Alderson PO. A balanced subspecialization strategy for radiology in the new millennium. AJR Am J Roentgenol. 2000;175(1):7-8.
  2. Atlas SW. Embracing subspecialization: the key to the survival of radiology. J Am Coll Radiol. 2007;4(11:752-3
  3. Sickles EA, Wolverton DE, Dee KE. Performance parameters for screening and diagnostic mammography: specialist and general radiologists. Radiology. 2002;224:861-869.
  4. Briggs GM, Flynn PA, Worthington M, Rennie I, McKinstry CS. The role of specialist neuroradiology second opinion reporting: is there added value?. Clin Radiol. 2008;63(7):791-5. Epub 2008 Apr 22.