The Findings and Impact of Nonrehydrated Guaiac Examination

ORIGINAL INVESTIGATION The Findings and Impact of Nonrehydrated Guaiac Examination of the Rectum (FINGER) Study A Comparison of 2 Methods of Screening for Colorectal Cancer in Asymptomatic Average-Risk Patients

Edmund J. Bini, MD; Roshini C. Rajapaksa, MD; Elizabeth H. Weinshel, MD

Background: Testing stool for occult blood at the time ity fee for endoscopy, and the pathology fee for the of digital rectal examination (DRE) has been discour- biopsy specimens. aged because it is thought to increase the number of false- positive test results. Results: During the 5-year study period, 672 patients were evaluated and a colonic source of occult bleeding was identified in 145 patients (21.6%). The predictive value of a posi- Objective: To compare the diagnostic yield of colonos- tive fecal occult blood test result (22.0% vs 21.3%, P = .85) copy and the cost per cancer detected in asymptomatic andthecostpercancerdetected($7604.80vs$7814.54)were patients with a positive fecal occult blood test result ob- no different in the DRE and SPS groups, with carcinomas tained by DRE with that obtained from spontaneously being detected in 11.7% and 11.3% of patients, respectively. passed stool (SPS) samples. Conclusions: Testing stool for occult blood at the time Methods: We reviewed the medical records of con- of DRE does not increase the number of false-positive test secutive asymptomatic patients at average risk for colo- results or the cost per cancer detected in asymptomatic rectal cancer who were referred for colonoscopy to patients at average risk for colorectal cancer. In this pa- evaluate a positive fecal occult blood test result tient population, all individuals should be evaluated by full obtained by DRE (n = 282) or SPS samples (n = 390). colonoscopy regardless of the method of stool collection. The cost of colonoscopy was estimated by adding the physician fee under Medicaid reimbursement, the facil- Arch Intern Med. 1999;159:2022-2026

CREENING FOR fecal occult In clinical practice, asymptomatic in- blood has been shown to re- dividuals with a positive FOBT result ob- duce mortality from colorec- tained by DRE are often referred for a colo- tal cancer,1-3 the second lead- noscopic evaluation.10 To date, there are ing cause of cancer death in almost no data on the diagnostic yield of theS United States.4 Because 70% to 80% of colonoscopy in asymptomatic patients with colorectal cancers are diagnosed in “av- a positive FOBT result obtained by DRE. The erage-risk” patients,5 the American Can- aims of this study were to compare the di- cer Society recommends annual fecal oc- agnostic yield of colonoscopy in patients cult blood testing (FOBT) of stool samples with a positive FOBT result obtained by DRE in all patients aged 50 years or older us- with that obtained by SPS sampling and to ing 3 specimens obtained from spontane- determine the cost of identifying a source ously passed stools (SPS).6 To decrease the of occult gastrointestinal bleeding by colo- number of false-positive FOBT results, pa- noscopy in these 2 groups of patients. tients are placed on dietary restrictions, and nonsteroidal anti-inflammatory drug RESULTS (NSAID) and aspirin use is discontinued 1 week before testing. Despite these rec- During the 5-year study period, 672 ommendations, clinicians often test stool asymptomatic average-risk patients with samples for occult blood at the time of digi- From the Division of 7 Gastroenterology, New York tal rectal examination (DRE). Although University Medical Center, widely practiced, this screening method has been discouraged because it is thought This article is also available on our Bellevue Hospital, and New Web site: www.ama-assn.org/internal. York Veterans Affairs Medical to increase the number of false-positive Center, New York, NY. FOBT results.8,9

ARCH INTERN MED/ VOL 159, SEP 27, 1999 2022

Downloaded From: https://jamanetwork.com/ on 09/27/2021 PATIENTS AND METHODS ENDOSCOPY All colonoscopic examinations were performed by gastro- PATIENTS enterology fellows with an experienced faculty member in attendance. Informed consent was obtained from each pa- Consecutive asymptomatic patients older than 50 years at tient before every procedure. All endoscopic abnormali- average risk for colorectal cancer with a positive FOBT ties were noted, and multiple biopsy specimens were ob- result obtained by DRE or SPS sampling who were tained from each one. All polyps were removed whenever referred to the gastroenterology service at Bellevue Hospi- possible. No major complications directly related to en- tal Center, New York, NY, between January 1992 and doscopy were noted. Lesions that were considered a source January 1997 were identified. Patients were excluded of occult gastrointestinal blood loss included: adenoma- from this study if they had any abdominal signs or symp- tous polyps greater than or equal to 1 cm in diameter, car- toms. Average-risk individuals were defined as patients cinoma, active colitis or inflammatory bowel disease, co- who did not have a history of colon polyps or colorectal lonic ulcers greater than or equal to 1 cm in diameter, and cancer, a family history of colorectal cancer, or inflamma- multiple vascular ectasias. Hemorrhoids and diverticula were tory bowel disease. not considered a source of occult gastrointestinal blood loss.

DATA COLLECTION COST OF ENDOSCOPY

Data on each patient, which were collected by means of a The cost of endoscopy was estimated by adding the phy- review of endoscopy records, patient charts, and pathol- sician fee under Medicaid reimbursement, the facility fee ogy reports, included age, sex, race, location at the time for endoscopy, and the pathology fee for the biopsy speci- of endoscopy (inpatient vs outpatient), presence of mens. Medicaid reimbursement values were used in the cost comorbid illness, history of peptic ulcer disease, history analysis because this payment method is most applicable of upper or lower gastrointestinal hemorrhage, presence to the Bellevue Hospital Center patient population. The to- of anemia, and use of alcohol, aspirin, NSAIDs, antico- tal estimated cost, including the physician fee, facility fee, agulants, or tobacco. Comorbid illness was defined as and pathology fee, was $739.64 for colonoscopy and $999.64 the presence of any of the following: cardiovascular dis- for colonoscopy with biopsy and/or polypectomy. ease (congestive heart failure, recurrent angina, or endo- carditis), pulmonary disease (pneumonia, pulmonary STATISTICAL ANALYSIS embolus, or chronic obstructive pulmonary disease), liver disease (acute hepatitis, chronic hepatitis, or cir- The diagnostic yield of colonoscopy in patients with a posi- rhosis), renal disease (creatinine level Ͼ177 µmol/L [Ͼ2 tive FOBT result obtained by DRE was compared with that mg/dL]), neurologic disease (meningitis or central ner- in patients who were found to have occult blood in SPS vous system disease with loss of independence), meta- samples. The cost of identifying a source of occult bleeding, static cancer, or the presence of a systemic bacterial or cost per neoplasm, cost per adenoma, cost per adenoma greater fungal infection. Anemia was defined as a hemoglobin than or equal to 1 cm in diameter, and cost per cancer de- level of less than 140 g/L in men or less than 120 g/L in tected by these 2 methods of screening were calculated. Con- women. Fecal occult blood testing was performed by tinuous variables were compared using a t test or a nonpara- testing 2 samples from each of 3 SPS or 1 specimen metric test, as appropriate. Categorical variables were obtained during DRE using commercially available test compared using the Fisher exact test. A 2-tailed P value of kits (Hemoccult II; SmithKline Diagnostics Inc, San Jose, less than .05 was considered statistically significant. All data Calif) without rehydration. The presence of occult blood are expressed as mean ± SD. Statistical analysis was per- was identified when a positive test result was noted on at formed using a commercially available software package (SPSS least 1 slide. version 7.5 for Windows; SPSS Inc, Chicago, Ill).

a positive FOBT result were evaluated by colonoscopy. in diameter, and adenocarcinoma were similar in The patient characteristics are shown in Table 1. Pa- both groups of patients. Also, 7 patients had vascular tients referred for a positive FOBT result obtained by SPS ectasias, 3 had ulcerative colitis, and 1 had Crohn dis- sampling were more likely to be outpatients than those ease. Five of the 7 patients with vascular ectasias had who were tested during DRE. The remaining demo- multiple lesions located in the cecum and ascending graphic and clinical characteristics were similar in both colon; the remaining 2 patients had isolated lesions in groups of patients. the sigmoid colon. Of the 3 patients with ulcerative coli- A colonic source of occult gastrointestinal bleed- tis, 1 had pancolitis, 1 had proctosigmoiditis, and 1 had ing was identified in 145 patients (21.6%). The predic- disease limited to the rectum. The patient with Crohn tive value of a positive FOBT result was no different disease had involvement of the ileum and proximal between those who were tested by DRE and those who colon. Lesions that were thought not to be a cause of had occult bleeding detected by SPS sampling (22.0% occult bleeding included hemorrhoids (125 patients vs 21.3%, P = .85). The lesions detected by colonos- [18.6%]), diverticulosis (49 patients [7.3%]), and both copy are shown in Table 2. The number of patients hemorrhoids and diverticulosis (37 patients [5.5%]). with neoplastic lesions (adenoma or adenocarcinoma), The findings of colonoscopic examination were normal adenomas, adenomas greater than or equal to 1 cm in 249 patients (37.1%).

ARCH INTERN MED/ VOL 159, SEP 27, 1999 2023

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Table 1. Characteristics of the 672 Patients Referred Table 2. Colonoscopic Findings in 672 Asymptomatic for Colonoscopy* Patients With a Positive Fecal Occult Blood Test Result*

DRE SPS No. (%) (n = 282) (n = 390) P DRE SPS Age, mean ± SD, y 63.2 ± 0.6 62.3 ± 0.5 .21 (n = 282) (n = 390) P Male 150 (53.2) 221 (56.7) .39 Race Source of occult 62 (22.0) 83 (21.3) .85 White 80 (28.4) 109 (27.9) bleeding identified Hispanic 84 (29.8) 100 (25.6) Neoplastic lesions 125 (44.3) 175 (44.9) .94 Black 50 (17.7) 82 (21.0) .73 All adenomatous polyps 111 (39.4) 151 (38.7) .87 Asian 45 (16.0) 65 (16.7) Adenoma Ն1 cm 29 (10.3) 47 (12.1) .54 Other 23 (8.2) 34 (8.7) Adenocarcinoma 33 (11.7) 44 (11.3) .90 Outpatient 237 (84.0) 357 (91.5) .003 Vascular ectasias 5 (1.8) 2 (0.5) .14 Comorbid illness present 64 (22.7) 70 (17.9) .14 Ulcerative colitis 2 (0.7) 1 (0.3) .56 Prior peptic ulcer disease 15 (5.3) 23 (5.9) .87 Crohn disease 1 (0.3) 0 (0.0) Ͼ.99 Prior upper GI bleeding 8 (2.8) 16 (4.1) .41 Normal results on 107 (37.9) 142 (36.4) .69 colonoscopy Prior lower GI bleeding 7 (2.5) 7 (1.8) .59 Current alcohol use 35 (12.4) 45 (11.5) .81 NSAID or aspirin use 66 (23.4) 93 (23.8) .93 *DRE indicates digital rectal examination; SPS, spontaneously passed stools. Anticoagulant use 20 (7.1) 24 (6.2) .64 Current smoker 73 (25.9) 90 (23.1) .41 Anemia 84 (29.8) 105 (26.9) .44 Table 3. Cost per Lesion Detected by Colonoscopy*

*GI indicates gastrointestinal; DRE, digital rectal examination; SPS, spontaneously passed stools; and NSAID, nonsteroidal anti-inflammatory DRE SPS drug. All values other than age are presented as number (percentage). (n = 282), $ (n = 390), $ Cost per source of occult bleeding 4047.72 4142.65 Cost per neoplasm 2007.67 1964.80 The average cost of performing colonoscopy was Cost per adenoma 2260.89 2277.08 $885.12 per patient, and there was no difference in cost Cost per adenoma Ն1 cm 8653.74 7315.74 between the DRE and SPS groups ($889.92 vs $881.64). Cost per adenocarcinoma 7604.80 7814.54 The cost of identifying a colonic source of occult gastrointestinal bleeding is shown in Table 3. The cost of detect- *DRE indicates digital rectal examination; SPS, spontaneously passed ing an adenoma equal to or greater than 1 cm in diameter stools. was slightly higher in the DRE group, while the cost per adenocarcinoma found was slightly higher in the SPS supplements such as ascorbic acid, the fact that only a group. However, these differences were not substantial. single specimen is obtained, and interpretation of the test results by a physician rather than by an experienced COMMENT laboratory technologist. In a retrospective study of 202 patients who under- Numerous factors compromise the effectiveness of went a DRE and stool guaiac test on admission to the hos- screening for colorectal cancer with FOBT. In struc- pital, Gomez and Diehl7 found that the results of clini- tured rigorous trials, the rate of return of screening cally indicated tests were positive more often than those guaiac cards has ranged from 40% to 68%.11,12 Further- of routinely performed tests (35% vs 11%). Based on these more, this estimate decreases to approximately 30% in findings, they concluded that FOBT performed during community screening programs, with patient educa- physical examination should be reserved for patients with tion, aversion to stool sampling, and the rigor of the clinical indications such as symptoms of acute or chronic recruitment effort all affecting patient compliance.11,13,14 gastrointestinal bleeding, anemia, weight loss, or change Even if patients are willing to collect multiple stool in bowel habits. samples, adherence to dietary restrictions and absti- To our knowledge, there are no studies demonstrat- nence from NSAIDs and aspirin are often less than ing that the sensitivity and specificity of FOBT of stool ideal. During the DRE, which is an established part of samples obtained by DRE are less than those of SPS sam- the physical examination, many physicians obtain stool pling. In an earlier study, Eisner and Lewis15 reviewed samples to perform FOBT,7 as this may be the only the records of 270 patients who underwent colonos- opportunity to screen for colorectal cancer. However, copy for a positive FOBT result obtained at the time of the use of FOBT of stool samples obtained by DRE to DRE (144 patients) or after testing of 3 SPS samples (126 screen for colorectal cancer is controversial. patients). In this mixed population of symptomatic and In an editorial on this subject, Longstreth9 asymptomatic patients, the authors found a similar describes the testing of stool samples obtained by DRE frequency of colonic abnormalities with both stool col- for fecal occult blood as a “knee-jerk” procedure that lection methods and no statistically significant differ- has little sensitivity or specificity in screening for colo- ences in the rates of detection of adenomatous polyps or rectal cancer. He attributes the limitations of this screen- colon cancers. ing method to hemorrhoidal trauma by DRE, the lack of In another study, Brint et al16 retrospectively evalu- control over the use of medications such as NSAIDs or ated 185 asymptomatic patients who underwent colo-

ARCH INTERN MED/ VOL 159, SEP 27, 1999 2024

Downloaded From: https://jamanetwork.com/ on 09/27/2021 noscopy for a positive FOBT result obtained by DRE and screening method is thought to increase the number of found neoplastic lesions in 28% of them. Based on these false-positive test results, thereby increasing the cost per findings, they concluded that screening for colorectal can- cancer detected as well as the complications associated cer by DRE does not increase the rate of false-positive with unnecessary endoscopic procedures. However, in results, and that positive test results should prompt a struc- our study, we found the cost per adenoma larger than tural evaluation of the colon. Interestingly, a recent study 1cm in diameter ($8654 vs $7316) and the cost per can- found that the likelihood of a gastrointestinal lesion be- cer detected ($7605 vs $7815) to be unaffected by the ing detected was significantly higher in patients who had method of stool collection. The cost per adenoma larger fecal occult blood detected during DRE than among those than 1 cm in diameter is similar to the cost per polyp who had a positive test result obtained by the tradi- ($8689) that has been reported in other studies.13 It is tional method of testing SPS samples.17 interesting to note, however, that the cost per cancer de- In the current study of 672 asymptomatic patients tected in the present study was less than half of that pre- at average risk for colorectal cancer, we found the pre- viously reported.13 Furthermore, colonoscopy did not dictive value of a positive FOBT result obtained by DRE result in any major complications in either group. to be no different from that obtained by SPS sampling If our findings are confirmed in prospective stud- (22.0% vs 21.3%). Furthermore, there was no differ- ies, screening by DRE would have a positive impact on ence in the number of adenomas, adenomas larger than colorectal cancer screening. Testing stool samples ob- 1 cm in diameter, or carcinomas between these 2 tained at the time of DRE may reduce the cost of screen- groups of patients. To our knowledge, this is the first ing because only 1 guaiac test kit would be needed in- study to compare these 2 methods of screening in a stead of 3, and the amount of time that nurses and cohort of asymptomatic patients at average risk for physicians spend with patients teaching them about di- colorectal cancer. etary restrictions would be reduced substantially. More Although other investigators have suggested that test- importantly, the number of patients who would be ing stool for fecal occult blood at the time of DRE may screened would increase from the current rate of ap- increase the number of false-positive test results,9 we found proximately 30%11,14 to nearly 100% of patients who are no difference between these 2 methods of screening for seen by a physician (assuming that all physicians offer colorectal cancer. One possible hypothesis to explain this testing to their patients). This can potentially reduce the lack of difference is that DRE does not cause a false- mortality from colorectal cancer considerably. positive test result and that there is truly no difference In conclusion, we found that FOBT of stool samples between these 2 methods. Alternatively, patients who were obtained by DRE does not increase the number of false- found to have fecal occult blood present when SPS samples positive test results or the cost per neoplasm detected in were tested may have been noncompliant with dietary, asymptomatic patients at average risk for colorectal can- NSAID, and aspirin restrictions. Noncompliance may have cer. Although we do not recommend FOBT at the time increased the number of false-positive test results in the of DRE as a routine method of screening for colorectal SPS group, resulting in a positive predictive value that cancer, our data would suggest that in this patient popu- was similar to that of the DRE group. However, this lat- lation, a positive FOBT result should be evaluated by full ter hypothesis would be impossible to validate in a ret- colonoscopy. Prospective randomized studies compar- rospective study. ing FOBT and SPS sampling are warranted to validate these In the present study, 527 patients (78.4%) did not findings before testing stool samples at the time of DRE have a colonic source of occult gastrointestinal bleeding can be routinely recommended. identified. The low predictive value of a positive FOBT result in our patient population (21.6%) is a well- Accepted for publication January 15, 1999. described limitation of screening for colorectal cancer with Presented in part at the annual meeting of the Ameri- guaiac-based tests.14 Since the positive predictive value can Society for Gastrointestinal Endoscopy during Digestive is highly dependent on the prevalence of disease in the Disease Week, May 20, 1998, New Orleans, La, and published population studied, it is not surprising to find a high in abstract form (Gastrointest Endosc. 1998;47:AB95). number of false-positive test results in a cohort of asymp- Reprints: Edmund J. Bini, MD, Division of Gastroen- tomatic patients. In addition to the test result being truly terology (111D), New York Veterans Affairs Medical Cen- false-positive, colonic lesions may have been missed dur- ter, 423 E 23rd St, New York, NY 10010. ing our endoscopic examination. Alternatively, the positive FOBT result may have detected occult bleeding from REFERENCES a site in the gastrointestinal tract proximal to the colon. Indeed, we18 and others17,19-23 have found a considerable number of patients with a positive FOBT result and nor- 1. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med. 1993;328:1365-1371. mal colonoscopic findings who had lesions detected dur- 2. Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard O. Randomised study ing upper gastrointestinal endoscopy. of screening for colorectal cancer with faecal-occult-blood test. Lancet. 1996; The cost of screening for colorectal cancer with 348:1467-1471. FOBT, including the cost of colonic evaluation of posi- 3. Hardcastle JD, Chamberlain JO, Robinson MH, et al. Randomised controlled trial of tive test results, has been estimated to be approximately faecal-occult-blood screening for colorectal cancer. Lancet. 1996;348:1472-1477. 4. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, 1998. CA Cancer J 1.2 billion dollars per year for the 60 million Americans Clin. 1998;48:6-29. 24 older than 50 years. Testing of stool samples obtained 5. Winawer SJ, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guide- at the time of DRE has been discouraged because this lines and rationale. Gastroenterology. 1997;112:594-642.

ARCH INTERN MED/ VOL 159, SEP 27, 1999 2025

Downloaded From: https://jamanetwork.com/ on 09/27/2021 6. Byers T, Levin B, Rothenberger D, Dodd GD, Smith RA. American Cancer Soci- 16. Brint SL, DiPalma JA, Herrera JL. Is a Hemoccult-positive rectal examination clini- ety guidelines for screening and surveillance for early detection of colorectal pol- cally significant? South Med J. 1993;86:601-603. yps and cancer: update 1997: American Cancer Society Detection and Treat- 17. Rockey DC, Koch J, Cello JP, Sanders LL, McQuaid KR. Relative frequency of ment Advisory Group on Colorectal Cancer. CA Cancer J Clin. 1997;47:154-160. upper gastrointestinal and colonic lesions in patients with positive fecal occult- 7. Gomez JA, Diehl AK. Admission stool guaiac test: use and impact on patient man- blood tests. N Engl J Med. 1998;339:153-159. agement. Am J Med. 1992;92:603-606. 18. Bini EJ, Valdes MT, Weinshel EH. Predictive value of a positive fecal occult blood 8. Lieberman DA. Colon cancer screening: the dilemma of positive screening tests. test for upper gastrointestinal lesions in asymptomatic patients with a negative Arch Intern Med. 1990;150:740-744. colonoscopy [abstract]. Gastrointest Endosc. 1998;47:AB80. 9. Longstreth GF. Checking for “the occult” with a finger: a procedure of little value. 19. Thomas WM, Hardcastle JD. Role of upper gastrointestinal investigations in a J Clin Gastroenterol. 1988;10:133-134. screening study for colorectal neoplasia. Gut. 1990;31:1294-1297. 10. Rex DK. Colonoscopy: a review of its yield for cancers and adenomas by indi- 20. Zuckerman GR, Benitez J. A prospective study of bidirectional endoscopy (colo- cation. Am J Gastroenterol. 1995;90:353-365. noscopy and upper endoscopy) in the evaluation of patients with occult gastro- 11. Ahlquist DA. Occult blood screening: obstacles to effectiveness. Cancer. 1992; intestinal bleeding. Am J Gastroenterol. 1992;87:62-66. 70:1259-1265. 21. Hsia PC, Al-Kawas FH. Yield of upper endoscopy in the evaluation of asymptom- 12. Hardcastle JD, Thomas WM, Chamberlain J, et al. Randomised, controlled trial atic patients with Hemoccult-positive stool after a negative colonoscopy. Am J of faecal occult blood screening for colorectal cancer: results for first 107,349 Gastroenterol. 1992;87:1571-1574. subjects. Lancet. 1989;1:1160-1164. 22. Geller AJ, Kolts BE, Achem SR, Wears R. The high frequency of upper gastro- 13. Trehu EG, Cooper JN. Cost of screening for colorectal cancer: results of a com- intestinal pathology in patients with fecal occult blood and colon polyps. Am J munity mass screening program and review of the literature. South Med J. 1992; Gastroenterol. 1993;88:1184-1187. 85:248-254. 23. Chen YK, Gladden DR, Kestenbaum DJ, Collen MJ. Is there a role for upper 14. Simon JB. Occult blood screening for colorectal cancer: a critical review. Gas- gastrointestinal endoscopy in the evaluation of patients with occult blood- troenterology. 1985;88:820-837. positive stool and negative colonoscopy? Am J Gastroenterol. 1993;88:2026- 15. Eisner MS, Lewis JH. Diagnostic yield of a positive fecal occult blood test found 2029. on digital rectal examination: does the finger count? Arch Intern Med. 1991;151: 24. Ransohoff DF, Lang CA. Screening for colorectal cancer. N Engl J Med. 1991; 2180-2184. 325:37-41.

ARCH INTERN MED/ VOL 159, SEP 27, 1999 2026

Downloaded From: https://jamanetwork.com/ on 09/27/2021