| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Schneider Children's Hospital New Hyde Park, NY 11040
I found the article by Jabra et al. [1] very interesting in that it raised several questions. First, for my generation of radiologists, who were trained before CT, there is the question of whether CT is needed for the diagnosis of small-bowel obstruction. I have relied heavily on radiographs over the years, and I believe that they are extremely valuable if scrutinized carefully and interpreted always in conjunction with the clinical findings. Today, of course, patients have CT scans for a variety of reasons—not always good. One may argue that small-bowel obstruction may be unsuspected—a situation hard to imagine, even with the minimum of clinical information, which is quite common these days. Even then, an abdominal radiograph should have been obtained before CT.
The authors studied a selected population, in that all their patients had CT scans. Obviously, not all those suspected of having small-bowel obstruction underwent CT. The reason for performing CT in their group, as opposed to not doing it, is not clear. Should everyone suspected of having small-bowel obstruction have undergone CT, the sensitivity and specificity of CT might have been different. Because this is a selected population, the predictive value of CT signs of small-bowel obstruction, which depends on the prevalence of the disease under study (or the pretest index of suspicion), cannot be defined. The group discharged with the diagnosis of ileus is very likely to have been inhomogeneous, and cases of transient incomplete obstruction may have been included as well.
The group of children younger than 2 years old raises more questions. The sample is too small for any conclusions. Statistically speaking, the sensitivity of 100% reported by the authors (i.e., seven successes in seven trials) has 95% confidence intervals from 0.56 to 1.0. This means that with many more cases, the sensitivity might have been much lower than 100%. Because all 10 patients younger than 2 years old were called positive cases, including the three who did not have small-bowel obstruction, the specificity was 0. Had those interpreting the scans decided to call them positive for small-bowel obstruction without looking at them, they would have had the same diagnostic accuracy.
The graph on page 433 is also wrong, in that it implies that sensitivity and specificity are directly proportional. The horizontal coordinate should plot the false-positive rate (1-specificity), not the specificity.
In practice, imaging procedures are not being interpreted in a vacuum. Intussusception and volvulus, for instance, present special clinical characteristics, and the authors' attempt to look into images and arrive at a precise diagnosis becomes artificial, because it does not reflect an actual clinical situation.
CT is a powerful diagnostic tool when used appropriately, but I think that the article by Jabra et al. [1] has serious methodologic limitations and loses its impact.
References
Associated Radiologist, Ltd. Mesa, Arizona 85204
Johns Hopkins Medical Institutions Baltimore, MD 21287
We thank Dr. Leonidas for his interest in our article [1]. The purpose of our article was not a comparison between CT and radiography. We do not propose that CT should replace radiography in diagnosing small-bowel obstruction in children. In fact, we acknowledge in the introduction and the discussion section that CT is usually not the first imaging examination used in evaluating for small-bowel obstruction. However, radiography can be inconclusive in diagnosing small-bowel obstruction: In a previous study, we found radiography to be inconclusive in 35% of cases [2]. Similar data are found in the literature regarding adults, where 40-50% of radiographic findings were inconclusive in the diagnosis of small-bowel obstruction [2,3,4].
As is true in all clinical research studies, our results can only be generalized to populations of patients who are similar to our study population. For this reason, we have attempted to define our study population as clearly as possible, leaving it to the reader to decide whether our study can be generalized to the reader's patient population. The decision as to which patients underwent CT imaging was made by the clinical services when additional diagnostic information was sought. We did not state the positive predictive value of CT in the diagnosis of small-bowel obstruction because it depends on the disease prevalence, which could not be defined in our study population. In our closing remarks, we mentioned the need for a prospective study that would give more accurate results and define the predictive values of CT.
Ileus has many causes; our study did not focus on any one in particular. If some of the patients we diagnosed as having ileus actually were cases of obstruction, the diagnostic accuracy that we reported would be an underestimate of the actual value. We addressed this issue under the "false-positive interpretations" portion of the discussion section.
Our statistical analysis comparing the sensitivity and specificity between age groups does account for the small number of patients involved. Even with this small population, we observed a statistically significant difference in specificity. We do not believe that robust conclusions should be based on so few patients. However, such results may from the basis for future investigation.
We agree with Leonidas that the graph on page 433 has a labeling error. The horizontal axis should read "1-specificity of CT for small-bowel obstruction" (Fig. 1).
|
We agree that imaging procedures are never interpreted in ignorance of the clinical setting. However, the accuracy of diagnostic tests, when expressed in terms such as sensitivity and specificity, relate to the information contributed by the diagnostic test independent of other available clinical information. Thus in the research setting, diagnostic tests must be evaluated with blinding to other clinical information.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
