| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the Department of Radiology, University of Washington, Box 357115, 1959 NE Pacific, Seattle, WA 98195. Received October 2, 2001; accepted October 8. Address correspondence to the author (e-mail: jarvikj@u.washington.edu).
Index terms: Editorials • Radiology and radiologists, research • Radiology and radiologists, socioeconomic issues
We live in a time of such rapid change and growth of knowledge that only he who is in a fundamental sense a scholar—that is, a person who continues to learn and inquire—can hope to keep pace, let alone play the role of guide.
Nathan M. Pusey (1)
The world of the radiologist is truly changing rapidly, and it is often confusing. New technologies seem to arrive almost daily, and all of us struggle to keep pace. In the current issue of Radiology, Hunink and Krestin (2) offer an approach to the rational evaluation of innovation "on the fly" and guide us toward adopting new methods of practicing medicine in a sensible way. Theirs is a well-written and thoughtful discussion of a problem that researchers face on a continuing basis. Hunink and Krestin propose not so much new techniques for conducting research but rather a new paradigm for the practice of academic medicine that integrates random allocation of patients and systematic data collection into daily practice while allowing the monitoring of trends in decision making and outcomes over time.
While new diagnostic tools stream into the medical marketplace, the actual benefits are frequently difficult to establish. Hunink and Krestin cite the work of Wright and Weinstein (3), who emphasize that changing from an old to a new diagnostic strategy generally results only in small benefits to the patient. If this is the case, what is driving this frantic rush for newer diagnostic techniques? Part of the answer undoubtedly lies in the minds of referring clinicians. Although new diagnostic tests may not make patients feel better, they almost certainly make physicians feel better. We all fear uncertainty, especially if the health and well-being of patients depends on crucial decisions that we make in the face of such uncertainty. Having a new and seemingly better test to confirm a diagnosis likely reassures physicians that they indeed have made the right decision regarding diagnosis and treatment. This reassurance may even be transmitted to the patient, although the effect that such reassurance has on currently used quality-of-life scales may be unmeasurable. There also continues to be pressure for clinicians, as well as radiologists, to remain au courant, which often may translate into use of the latest technology. Marketing by manufacturers also plays a role in the adoption of new technology, although the magnitude of the effect has been poorly quantified in the clinical literature. Finally, let us not forget that it is just plain fun to be the first on the block with a new toy.
Why do the Hunink and Krestin believe that a new research paradigm is needed? Radiology research has tended to focus on the diagnostic accuracy of new tests, but none of the aforementioned factors that lead to use of new diagnostic tests are reflected in measures such as sensitivity, specificity, or receiver operating characteristic curves. It should not be surprising that research "business as usual" has resulted in the dissemination of new technologies before they have been adequately evaluated.
The authors (2) spend the first portion of their article discussing the assumptions underlying current methods of assessing diagnostic technologies. They point out the lack of good reference standards, the biased assembly of cohorts, the limitations of sensitivity and specificity as metrics, the limitations of decision analysis, and the difficulty of performing large randomized controlled trials for assessing new diagnostic technologies. They also point out that information about a new test does not necessarily lead to optimal use. Pressure from referring clinicians and patients more often than not results in a new diagnostic test being performed in addition to, rather than instead of, the existing test. But perhaps the most compelling argument for a new research approach is that the extremely rapid pace of change in technology results in standard assessments frequently being out of date by the time they are disseminated.
The authors propose dispensing with the evaluation of diagnostic accuracy and instead focusing simultaneously on clinician decision making and patient outcomes. Their study design of choice is the randomized trial. In the 19th century, Alphonse Karr, a French journalist, remarked, "Plus ça change, plus c’est la même chose" (the more things change, the more they remain the same) (4). As it turns out, the paradigm suggested by Hunink and Krestin was used in nearly the form that they advocated more than a decade ago. In 1988, a British neurosurgeon, Graham Teasdale and colleagues (5) published a study in the British Medical Journal in which they compared computed tomography (CT) with magnetic resonance (MR) imaging for suspected lesions of the posterior fossa. They recruited new patients with posterior fossa signs and symptoms and then randomly assigned them to undergo either CT or MR imaging. They measured subsequent physician decision making and even reported trends over time for the ordering of the alternative test. Not surprisingly, requests for CT decreased and those for MR imaging increased over time. They stopped short of measuring patient outcomes, but this study nonetheless provides a classic example as to how the "new" technology assessment advocated by Hunink and Krestin can be performed.
Perhaps the two most important questions raised by Hunink and Krestin are, Who would pay for such an approach? and, How practical is it in a busy clinical practice (2)? The authors state that they have successfully implemented such a trial in their department, and for that they are to be commended. Having headed a multicenter randomized trial, I can say from firsthand experience that organizing the recruitment of subjects, overseeing randomization, and collecting the type of data that Hunink and Krestin suggest are time-consuming and labor-intensive tasks that require dedicated and well-trained personnel. The greatest expense in such trials is usually not the technologic intervention but rather the personnel needed to carry out the study. The authors argue that these research expenses would be incurred in any trial, irrespective of design. While this is true, the costs of nonrandomized cohort studies are frequently less than those of randomized trials. One reason is that recruitment for nonrandomized trials is generally much easier than it is for randomized trials, since patients (as well as the rest of us) hesitate to relinquish freedom of choice. Because recruitment rates are smaller, more subjects must be screened, which increases personnel time and thus increases the expense of the study.
The authors also argue that the trial may reduce the expense of the diagnostic work-up since, as illustrated in their example of intraarterial digital subtraction angiography versus CT angiography, fewer patients would undergo intraarterial angiography if a trial was ongoing (2). Of course, if the new diagnostic test is more expensive than the existing one, this argument vanishes. Even if the trial lowers the cost of the diagnostic work-up, downstream costs may overwhelm these savings. Frequently, new diagnostic tests are more sensitive, and the increased detection of abnormalities results in increased subsequent costs. It is often impossible to determine prior to a trial whether the costs of the new strategy will be lower or higher than those of the old strategy. If we could accurately predict these costs, there would be less need for a trial. In any case, some entity must be ready to bear the economic burden posed by this research paradigm.
Hunink and Krestin (2) suggest that agencies, presumably governmental, interested in technology assessment should be the funding source. The primary agency in the United States with such an interest is the Agency for Healthcare Research and Quality (AHRQ). Their budget for fiscal year 2001 is nearly $270 million (6). While this represents a 32.5% increase from the fiscal year 2000 budget, it is only about 1.3% of the estimated $20.3 billion fiscal year 2001 budget for the National Institutes of Health (7). The newly established National Institute of Biomedical Imaging and Bioengineering (NIBIB) is a potential source of money for this type of research, but the magnitude of the funding for NIBIB remains uncertain (8).
If the methods suggested by Hunink and Krestin (2) are to be adopted on a wide scale, novel sources of funding are probably needed. Device manufacturers and drug companies are an important source of research dollars, but much of their research is focused on the early development of a technology prior to its reaching the marketplace. Once reimbursement for a new device or drug is achieved, the impetus for further industry-funded research diminishes. One proposal, which has been suggested for expensive therapies that have no proven benefit beyond that of standard therapy, is for insurers to pay for new therapies only in the setting of a randomized trial. This strategy could also be applied to diagnostic technologies and would nicely complement the empirical research methods suggested in this article.
Researchers constantly try to balance the competing demands of scientific rigor with the equally important mission of maintaining the applicability of their research results. The thought-provoking article by Hunink and Krestin (2) provides some insight into how this tension between rigor and relevance can be effectively addressed by changing not only the way that academic medical centers conduct research but also how they practice medicine. As Hunink and Krestin say near the end of their article, their proposal is not a panacea. It is, however, certainly a good start.
FOOTNOTES
See also the article by Hunink and Krestin in this issue.
REFERENCES
Related Article
This article has been cited by other articles:
|
|
 |
|
  M. C. J. M. Kock, M. E. A. P. M. Adriaensen, P. M. T. Pattynama, M. R. H. M. van Sambeek, H. van Urk, T. Stijnen, and M. G. M. Hunink DSA versus Multi-Detector Row CT Angiography in Peripheral Arterial Disease: Randomized Controlled Trial Radiology, November 1, 2005; 237(2): 727 - 737. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B. Patwardhan, D. C. McCrory, D. B. Matchar, G. P. Samsa, and O. T. Rutschmann Alzheimer Disease: Operating Characteristics of PET-- A Meta-Analysis Radiology, April 1, 2004; 231(1): 73 - 80. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. C. Black and M. N. Brant-Zawadzki Invited Commentary * Author's Response RadioGraphics, November 1, 2002; 22(6): 1536 - 1539. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
