| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Harvard Medical School Massachusetts General Hospital Avon Foundation Comprehensive Breast Evaluation Center Wang Ambulatory Care Center Boston, MA 02114
It is with a great deal of angst that I write in support of anecdote over science. Unfortunately, the study by Sumkin et al. [1] purporting to evaluate which study was the best to use for comparison with the present screening mammogram requires a response, lest its conclusions be used to guide practice. Clearly, a large amount of work went into their article, but the results cannot actually be used to determine which prior study is the optimal reference mammogram. The primary, and perhaps only, true value for mammography is in screening to detect breast cancers earlier (sensitivity).
There are a number of cancers that are not detectable on mammography, so even though its specificity is high, mammography is not used to exclude the presence of a cancer. Because specificity is a secondary concern, the finding that specificity was higher with the 1-year comparison is interesting, but not critical. The real question is which comparison film (1 or 2 years preceding the present examination) maximizes the likelihood of detecting a small cancer. The article suggests that the sensitivity for detecting cancers is the same, regardless of the comparison year used. However, the cases used in this study were highly selected, and it is impossible to know what biases might have been introduced.
Anecdotally, it has been my experience that subtle changes indicative of a cancer may not be evident from one year to the next but become apparent when compared with studies that precede the present by 2 or more years. I have seen this on multiple occasions, although it is not common. Because that happens only occasionally, it would be unlikely to occur in the small, highly selected series that was studied by Sumkin et al. [1]. Conversely, I am unaware of any cancers having been missed because a 2-year-old or older study was the comparison study. Thus, we are left with a situation in which it would be difficult to prove scientifically that comparison with the 2-year or older study provides greater sensitivity, but there is every reason to believe that it would. The only advantage of using a more recent study might be the improved specificity. This, however, does not preclude using the older study as the primary reference because, if there appears to be a change from the 2-year or older study, the more recent studies are easily reviewed for the small number of cases in which the question is raised.
In the absence of science, anecdotal experience must be our guide, and it makes more sense to use an older study than the most recent annual examination as the comparison mammogram.
References
University of Pittsburgh Pittsburgh, PA 15213-3180
Our study [1] was a small one that included selected cases. However, these cases were not selected with any specific outcome in mind. Actually, we were somewhat surprised by the results. Although we cannot (and should not) discard anecdotal findings, we do not believe we should discard this study either. Indeed, at our institution we have been using, when available, the 2-year prior examination for comparison during the initial review. Our preliminary study was carried out because we were uncertain whether this is an optimal practice, and we were not aware of any published experimental data to indicate that it is. We do not advocate that we rush to change practices that use 2-year-old mammograms for comparison. What we suggest is that this may be an important issue that has not been adequately investigated. It should be assessed in a rigorous manner using a large and diverse dataset before the anecdotal route is taken as a given. Contrary to Dr. Kopans' personal observation, even in this small set, three of the cancers (of a total of 30) were missed by at least three of 12 radiologists in the mode that used 2-year-old mammograms for comparison, and these cancers were detected in the mode that used the 1-year-old ones. Because overall sensitivity did not change in our study, obviously there were a comparable number of other cases that went the other way around. We can think of several scenarios in which the longer period between the examinations being compared could yield different results. These include patients who have undergone a significant weight loss, those with multiple and rapidly changing cysts, and those near menopause who experience significant changes in breast tissue composition or appearance. It is also possible that relatively small areas of increasing density might be considered to be within normal range because of a longer elapsed time between the examinations being compared.
We agree with Kopans that sensitivity is important in screening mammography, but we disagree that sensitivity is the only important issue here. Performance parameters have to include both sensitivity- and specificity-related measures in today's environment, in which medicine is being judged by society on both efficacy and cost-effectiveness.
In summary, our study was preliminary in nature, was small in sample size, and included a retrospective review of selected cases. At the same time, it was an initial attempt to address in a quantitative manner a practice-related question, the answer to which has been largely addressed by anecdotal observations. It may or may not prove to be important, but it could have implications on both sensitivity and specificity in the clinical environment. Hence, it should be evaluated more carefully than it has been to date.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  L. A. Hardesty, A. H. Klym, B. E. Shindel, D. M. Chough, J. H. Sumkin, and D. Gur Is Maximum Positive Predictive Value a Good Indicator of an Optimal Screening Mammography Practice? Am. J. Roentgenol., May 1, 2005; 184(5): 1505 - 1507. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
