Correspondence
Occult Gastrointestinal Bleeding in Celiac Sprue
N Engl J Med 1996; 335:752-753September 5, 1996
- Article
To the Editor:
Dr. Fine (May 2 issue)1 concludes, on the basis of guaiac (Hemoccult) testing, that patients with celiac sprue have a high prevalence of occult gastrointestinal bleeding. A fact mentioned, but not discussed, is that the testing was performed with two different products: Hemoccult II for specimens analyzed between 1985 and 1990 and Hemoccult Sensa for specimens analyzed between 1991 and 1995. These two products have different sensitivities to peroxidase-containing materials.
In a recent large screening study comparing the performance of the two tests on the same stool samples from about 8000 subjects, 2.5 percent were positive with the Hemoccult II test (false positive rate, 1.9 percent), as compared with 13.6 percent with the Hemoccult Sensa test (false positive rate, 12.3 percent).2 In patients with celiac sprue, the false positive rate may be appreciably higher because of malabsorbed peroxidase-containing foods. Moreover, group differences attributable to false positive results may be amplified by the more sensitive Hemoccult Sensa test. In any case, it is critical to know the proportion of subjects in each group whose stool samples were tested with Hemoccult Sensa. Many of the control subjects had participated in an earlier study, so a disproportionate number of them may have been tested with the older, less sensitive test. If so, the results are biased toward higher rates of positivity in the patients with celiac sprue.
Finally, it is worth noting the subjective nature of Hemoccult testing, which depends on the observer's judgment of a change in color. When such end points are compared in study groups, a bias can be prevented only if the group designation is concealed from the observer. Dr. Fine does not state that this precaution was taken.
2 ReferencesWilliam C. Duane, M.D.
Minneapolis Veterans Affairs Medical Center, Minneapolis, MN 554171
Fine KD . The prevalence of occult gastrointestinal bleeding in celiac sprue. N Engl J Med 1996;334:1163-1167
Full Text | Web of Science | Medline2
Allison JE ,Tekawa IS ,Ransom LJ ,Adrain AL . A comparison of fecal occult-blood tests for colorectal-cancer screening. N Engl J Med 1996;334:155-159
Full Text | Web of Science | Medline
To the Editor:
We think the finding of a high frequency of positive Hemoccult tests in patients with celiac sprue should be interpreted cautiously. Dr. Fine fails to tell us when the stool samples were collected during the diagnostic evaluation. If the stool samples were collected soon after an intestinal biopsy was done, the prevalence of positive Hemoccult tests may well have been falsely elevated. We also should be told what percentage of the patients had positive Hemoccult tests at the time of the initial rectal examination. Before these findings can be accepted as definitive, more quantitative information about actual blood loss in patients with celiac sprue will be needed.
Michael A. Heneghan, M.B.
Ciaran F. McCarthy, M.D.
University College Hospital, Galway, IrelandAuthor/Editor Response
Dr. Fine replies:
To the Editor: If the switch from the Hemoccult II test to the more sensitive Hemoccult Sensa test had been a causal factor in the outcome of this study, there would have been a higher frequency of positive results among the patients with villous atrophy only with the Hemoccult Sensa test, and an overwhelming majority of the 17 patients with villous atrophy and Hemoccult-positive stools would have been tested with Hemoccult Sensa. With both tests, however, the percentage of positive results was significantly higher in the patients with villous atrophy than in the control subjects, and the positive results in patients with villous atrophy were equally distributed between the Hemoccult II and Hemoccult Sensa tests (nine and eight, respectively).
Although fecal peroxidase activity derived from food causes more false positive results with the Hemoccult Sensa test than with the Hemoccult II test, the 12.3 percent rate of false positivity for the Hemoccult Sensa test cited by Dr. Duane is not applicable to the present study, since it was derived from a study of colon-cancer screening.1 In this setting, a false positive Hemoccult test is defined as a positive test in the absence of a colonic cancer or a large polyp, regardless of the cause. Although some of these false positive results are caused by the ingestion of peroxidase-containing foods, others reflect the presence of hemoglobin from sources of occult bleeding other than a colonic cancer or polyp.
The possibility that malabsorbed foods in patients with villous atrophy could have resulted in false positive Hemoccult tests is the reason that patients with pancreatic insufficiency and malabsorption were included among the control subjects. The 4 percent rate of positivity among the patients with pancreatic insufficiency (and malabsorption of an average of 43 g of fat per day), as compared with the 47 percent rate of positivity among the patients with villous atrophy (and malabsorption of only 24 g of fat per day), argues strongly against the idea that malabsorption of peroxidase-containing foods was responsible for the findings of this study.
Although it is true that all 17 specimens from the patients with chronic idiopathic diarrhea were analyzed with the Hemoccult II test, all 18 specimens from the normal subjects were analyzed with the Hemoccult Sensa test. Overall, an equal number of control specimens was analyzed with each test.
With regard to Dr. Duane's final point, the interpretation of all the Hemoccult tests was performed by two laboratory technicians who were unaware of the patients' diagnoses.
In response to Drs. Heneghan and McCarthy: stool collections from all patients were obtained during the initial stage of the diagnostic investigation, and these collections always preceded intestinal biopsies. Thus, no patients were at risk for false positive Hemoccult tests due to bleeding from mucosal biopsies.
1 ReferencesKenneth D. Fine, M.D.
Baylor University Medical Center, Dallas, TX 752461
Allison JE ,Tekawa IS ,Ransom LJ ,Adrain AL . A comparison of fecal occult-blood tests for colorectal-cancer screening. N Engl J Med 1996;334:155-159
Full Text | Web of Science | Medline
