| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
St. John's Hospital Eisenberg-Keefer Breast Center Santa Monica, CA 90404-2032
The study by Sumkin et al. [1] wisely addresses an issue that has been the subject of anecdotal experience and attempts to provide objective data to support not only an impact on sensitivity and specificity in the comparison of prior studies, but also a suggested protocol. It is valuable, as the authors note, to arrive at a reasonable time interval for comparison images so that off-line film loading on multiviewer stations can be facilitated.
The statistical significance of comparison versus noncomparison is on far firmer ground than their recommendation for comparing the prior year instead of older studies. The impact of the results is weakened by relatively small numbers of cases examined, selection criteria that are not fully explained (so that selection bias cannot be excluded), and, more importantly, the small number of cancers reported.
When too few slow-growing cancers are included in a study such as this, the difference—if any—between comparing 1- and 2-year previous examinations cannot be adequately assessed. Thus, both the sensitivity and specificity results cannot be adequately measured, especially the former. Even the latter showed border-line significance (p = 0.06), which may simply be considered not significant, given the small number of cases [2].
Many mammographers will continue to advocate a longer interval for comparison; I use 3 years, if the films are available. However, this anecdotal approach, which emphasizes stability, especially for areas of asymmetry, lacks the objective validation that Sumkin et al. [1] have attempted. Their report appropriately and validly stresses the need for comparison. Even if their statistics and methodology are unpersuasive in supporting a specific comparison interval, it is laudible that they have pointed out that the question needs further study.
References
University of Pittsburgh Pittsburgh, PA 15213-3180
We thank Dr. Brenner for his constructive comments. We fully agree our study [1] was preliminary, and the topic of an optimal comparison during the initial review is an important issue that needs to undergo further study, preferably in a prospective clinical setting rather than in a laboratory setting. We cannot and should not ignore old habits and anecdotal experiences because they are often important (hence our surprise with the results in that our reviewers subjectively indicated a preference for a 2-year comparison). At the same time, we should be willing to challenge those habits that have not been validated. Our study was small and clearly statistically underpowered to show a possible small improvement in sensitivity when using the 2-year-old mammograms for comparison. At the same time, our case selection could have been somewhat biased toward the latter, because many of these cases had been originally reviewed with the 2-year-old mammogram for the initial comparison and, as a result were recalled for further follow-up.
That we all feel strongly about this issue and that it has not been validated to date suggest that this should be done, and we thank Brenner for recognizing and accepting the need to do so. As we consider the change to digital mammography, the issue will become a practical matter as well, because the combination of film and electronic review (mixed mode) is not likely to be as efficient as a totally film-based or totally electronic-based practice. At least during the first 2 years after the transition, there will be no electronic 2-year-old mammograms available for comparison. It may be important to know if the subjectively expected benefit of the 2-year mode outweighs the additional inconvenience that would result from a mixed-mode review.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
