| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorial |
1 From the Department of Radiology, Children’s Hospital, 300 Longwood Ave, Boston, MA 02115. Received July 18, 2001; accepted July 23. Address correspondence to the author (e-mail: thorne.griscom@tch.harvard.edu).
Index terms: Diagnostic radiology, observer performance • Images, interpretation • Radiography, production of images • Radiology, production of images and interpretation • Radiology and radiologists
The history helps (1–4). Interpreting images while unaware of the patient’s problem is foolish. If you aren’t sure what the question is, it’s tough to give the right answer.
Sometimes, however, the history misleads (5). About 15% of reported bone ages are shifted into the normal range when a child’s age is known, as compared with the bone ages determined by readers blinded to the actual age (6,7); that 15% must represent bias, introduced by knowing the history ahead of time. If you are aware that an infant has cough and fever and that you are in the midst of a bronchiolitis epidemic, it is easy to think you see hyperinflation and inflammatory thickening of the bronchial walls (ie, bronchiolitis) (8), no matter what the images, whether on hard copies or monitors, actually show. In a test situation, incorrect observations were much more frequent when the fictitious history falsely suggested their presence (1,9). If you know before looking at the chest radiographs that the infant has a murmur and congenital heart disease is suspected, the likelihood of suggesting minor cardiomegaly is increased, even though without that prompt you would not have mentioned the heart. Although most of these examples are from pediatric radiology, the problem is inevitably present in any branch of imaging. We are all suggestible.
So, the history is valuable but sometimes misleading. What can be done?
This problem has a partial solution: Review the images first, before you read the history; make your observations and come to a preliminary opinion; then read the history; then revise or abandon your opinion as seems best in light of the history and reach a final conclusion. If you read the history first, learn that the child has a murmur, and then look at the images and suggest mild cardiomegaly, your opinion is suspect; neither you nor anyone else can say whether you actually saw cardiac enlargement or merely inferred its presence from the history. If, on the other hand, you think you see cardiomegaly and only then discover that heart disease is a real possibility, your opinion carries much more weight. If you find a lucent line in the navicular bone that might be a subtle undisplaced fracture but might be nothing, the patient being unavailable for physical examination, and then you learn that the injury was to a terminal phalanx, you are probably justified in renouncing your suspicion of wrist fracture without further ado. But if, on the next child’s radiographs, you find a questionable buckling of the metaphyseal cortex of a proximal phalanx and only then learn that the injury and pain are at precisely that spot, you should be much more confident of fracture and can congratulate yourself on your perspicacity. It helps, when following this plan, not to put too much emotional investment in your preliminary impression, for it may soon need drastic revision.
This technique has many advantages. It enhances the credibility of any observations made before the diagnosis-directing history is known. It tends to reduce the number of false-positive findings without increasing the number of false-negative findings. It is particularly useful with such findings as bronchial wall thickening in infancy, highly subjective and almost impossible to verify. It forces one to scan the entire image or images, since the specific area of interest is not yet known. It promotes unexpected observations. It makes one identify and explain shadows that would otherwise be ignored. It keeps the radiologist engaged in the process—it converts it to an intellectual game, turning a chore into fun—and reminds him or her to consider rarities. It may sustain diagnostic sharpness decades after ennui would otherwise have set in; after all, each case is a test, and the history, when ascertained, often amounts to immediate feedback. If you fail to note a finding and then realize, after reading the history, that it is present and you should have seen it, this is an obvious learning opportunity.
However, the approach also has problems. It requires independent thought and intellectual vigor, which are often in short supply at the end of the day. Self-control is needed not to let one’s glance slip to the provided history. It takes more effort than the routine disposition of a case, the radiologist concurring with whatever the history suggests. If five more cases are urgently awaiting your attention, you may not have the time to follow this plan. The technique is much easier with simple stereotyped examinations—a set of chest radiographs, for example—than with complicated multiimage examinations requiring the radiologist to be part of the planning process. If you are actually performing the examination, the approach will work only in part, but even then it is critical to keep an open mind and not marry yourself to the anticipated diagnosis too soon. If the observer knows what illnesses are likely—if today is hip dysplasia clinic, or if a skeletal survey in infancy usually implies concern about the battered-child syndrome—then the system may break down. If one is reluctant to entertain exotic diagnoses or is temperamentally unable to think seriously about the unlikely, there will be difficulty. If you follow this approach in public, thinking out loud, you must reconcile yourself to sometimes looking foolish—but sometimes looking clairvoyant—when the history is finally revealed. Finally, the approach may be unsettling for those not yet confident of their observational skills.
Thus, looking at the images before you get the history is only a partial cure for the bias the history brings. A partial cure, however, is better than nothing.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  C. T. Loy and L. Irwig Accuracy of Diagnostic Tests Read With and Without Clinical Information: A Systematic Review JAMA, October 6, 2004; 292(13): 1602 - 1609. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Dhingsa, A. Qayyum, F. V. Coakley, Y. Lu, K. D. Jones, M. G. Swanson, P. R. Carroll, H. Hricak, and J. Kurhanewicz Prostate Cancer Localization with Endorectal MR Imaging and MR Spectroscopic Imaging: Effect of Clinical Data on Reader Accuracy Radiology, January 1, 2004; 230(1): 215 - 220. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Spieth and N. T. Griscom Interpretation of Images before Knowledge of Patient History * Dr Griscom responds: Radiology, December 1, 2002; 225(3): 922 - 923. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
