| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Mater Misericordiae Hospital, Eccles Street, Dublin D7, Ireland. e-mail: bruzzij@hotmail.com
Editor:
In the October 2003 Radiology article by Dr Taylor and colleagues (1) on the effectiveness of intravenous hyoscine butylbromide (Buscopan; Boehringer Ingelheim, Bracknell, England) as a smooth muscle relaxant in computed tomographic (CT) colonography, the authors reported significantly improved distention in the right side of the colon and in the transverse and descending colon. Results from our own study (2) support their observations in the transverse colon and descending colon, but we found that such an effect was evident only in patients with diverticular disease, and we postulated that hyoscine butylbromide may have its greatest effect on the proximal colon in patients who resisted optimal air insufflation because of spasm of the sigmoid colon. The improvement in colonic distention with hyoscine butylbromide was not statistically significant when patients without diverticulosis were considered. It would be interesting to know to what degree the prevalence of diverticulosis in the study population influenced the findings of Dr Taylor and colleagues.
The conclusion from our own study that the routine use of intravenous hyoscine butylbromide as a muscle relaxant in CT colonography is not justified is based on two observations: First, we did not find that the improvement in colonic distention with hyoscine butylbromide was sufficient to render inadequate distention adequate. Second, we found no improvement in overall polyp detection with intravenous hyoscine butylbromide. Although, like Dr Taylor and colleagues, we did not assess the effect of intravenous hyoscine butylbromide on patient discomfort during CT colonography, the administration of this drug necessitates intravenous injection, a prospect not welcomed by patients undergoing CT colonography. Any potential beneficial effect of intravenous hyoscine butylbromide on colonic distention must be balanced against a possible reduction in patient compliance (which would be deleterious to the effectiveness of CT colonography as a screening tool [3]) and the increased cost of the procedure. Although hyoscine butylbromide may have a role in patients known to have diverticular disease, it is our opinion based on the current evidence that it should not be used routinely for CT colonography.
REFERENCES
Intestinal Imaging Centre, St Mark’s Hospital, Level 4V, Watford Road, Northwick Park, Harrow, London HA1 3UJ, England. e-mail: s.halligan@imperial.ac.uk
We wish to thank Drs Bruzzi and Fenlon for their interest in our article (1). It is reassuring that in their own study (2), they gave results broadly similar to ours: Hyoscine butlybromide improves colonic distention during CT colonography. However, Dr Bruzzi and colleagues did not find that administration significantly increased the number of diagnostic studies, whereas we found clear benefit: Overall, the odds of a patient having at least one inadequately distended segment were approximately six times higher if a spasmolytic was not administered (and possibly up to 16 times according to the upper limit of our 95% CIs). Since the methodology adopted in both studies was similar, the source of discrepancy might rest with differences in statistical power and analysis.
Our a priori hypothesis was that use of a spasmolytic improved distention by approximately 25%, and we powered our study at a conventional 80% to detect this. We recruited 136 subjects, which provided 1,632 observations (six segments per colon for each prone and supine study), which ostensibly gave us 98% power at a 5% significance level to detect a difference in adequately distended segments after accounting for the partitioning as a result of randomization. However, this naively assumes that all individual segmental observations are completely independent of each other, which is clearly not the case, since each patient contributes 12 segments. Rather, it is highly likely that the 12 segmental scores obtained from the same subject are more similar than those obtained from other subjects. We used robust standard errors to counter this lack of independence in the data, which gave us an adjusted power of 80%. The assumption that individual observations are independent of each other is required for many statistical tests, including the Fisher exact test used by Dr Bruzzi and colleagues (2), which additionally makes no account for the ordering between groups, also diminishing power. By our calculation, Dr Bruzzi and colleagues (2) had power possibly as low as 46% to detect a difference of 25% in proportions at a 5% significance level if we assume that all 12 segments in each subject are related.
Drs Bruzzi and Fenlon also base their opinion on the fact that use of a spasmolytic did not significantly improve polyp detection in their study, but lack of statistical power may also substantially weaken this conclusion. For example, there were only seven polyps that were 9 mm or larger, two in the group that received spasmolytics and five in the control group (2). By our calculation, their study had 12% power to find a 10% difference in detection between groups at the 5% level, even assuming complete independence of segments. Again, by assuming complete segmental independence (which is highly unlikely for the reasons described), we calculate that they would have actually needed 233 subjects per group (ie, 466 overall) to detect a 10% difference in proportions at the 5% significance level with 80% power. Clearly, small or even moderate differences in polyp detection rates with use of spasmolytics were well beyond the resolution of the study. The thorny issue of detection of flat adenomas also remains, and in our experience, such lesions are best detected in well-distended segments. In support of this, investigators in a recent study (3) found poor colonic distention to be a major cause of false-negative findings for computer-assisted detection.
Whether the benefits of hyoscine butylbromide are limited to those with diverticular disease is an interesting point. The median age of our cohort was 63 years, and the prevalence of diverticulosis was around 50% (typical of this age in a Western population). Although we did not perform a subset analysis, in our experience, the advantage of hyoscine butlybromide also extends to those without diverticular disease. The routine use of hyoscine butlybromide is widely advocated prior to barium enema, irrespective of the presence of diverticular disease (4,5), and we are unsure how a policy of selective administration based on resistance to gas insufflation would fare in day-to-day practice.
With regard to use of spasmolytics in screening, it seems to us unlikely that the additional cost of administration would be important within the overall context of a national program (one dose costs around 12 cents, although the drug remains unlicensed in the United States), especially since we believe that our study demonstrates unequivocal benefit. Reduction in recall rates due to poor distention and their associated expense would rapidly accrue in a national program, even if the benefit were smaller than we have demonstrated. Furthermore, we have recently shown that the injection itself does not impair patient tolerance of CT colonography (6).
In summary, we feel the evidence from our study strongly supports administration of hyoscine butylbromide wherever the drug is available.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
