- Email: [email protected]
The Double Contrast Barium Enema: A Retrospective Single Centre Audit of the Detection of Colorectal Carcinomas
Clinical Radiology (2002) 57: 29±32 doi:10.1053/crad.2001.0724, available online at http://www.idealibrary.com on
The Double Contrast Barium Enema: A Retrospective Single Centre Audit of the Detection of Colorectal Carcinomas DA NI E L J . A . CO N NO L LY, Z O EÈ C . T R A I L L , H E L E N S . R E I D , SU S A N J . CO P L E Y, DANIEL J. NOLAN Department of Radiology, John Radclie Hospital, Oxford, U.K. Received: 30 November 2000 Accepted: 23 January 2001 AIM: To determine retrospectively the sensitivity and speci®city of the double contrast barium enema (DCBE) as performed in one institution for the detection of colorectal carcinoma. SUBJECTS AND METHODS: Eight hundred and eighty barium enema reports were reviewed of consecutive adult patients who underwent DCBE and also had hospital case notes with a minimum follow up of two years, a later diagnostic colonoscopy, or operative and histological ®ndings. RESULTS: Seventy-four true positive cases of colorectal carcinoma diagnosed at DCBE were con®rmed at surgery and histological examination. There were four false positive diagnoses of carcinoma at DCBE. Eight false negative cases at DCBE were demonstrated within a two-year followup period. The sensitivity of the DCBE for detecting colorectal carcinoma was therefore 90.2% and the speci®city was 99.5%. CONCLUSION: DCBE is a sensitive and highly speci®c investigation for the detection of colorectal carcinoma. Connolly D. J. A., et al. (2002) Clinical Radiology 57, 29±32. # 2002 The Royal College of Radiologists Key words: barium enema, colorectal cancer, sensitivity, speci®city, double reporting.
Colorectal carcinoma is responsible for over 19 000 deaths per annum in the U.K. [1]. The double contrast barium enema (DCBE) has been a standard method of examining the large intestine for many years. The emergence of ¯exible sigmoidoscopy and colonoscopy with their high reported sensitivities and speci®cities for the detection of colorectal carcinoma, has led to their adoption in many centres as the method of choice for investigating the large intestine. The previously reported sensitivity and speci®city of the DCBE has generally been found to be below that of colonoscopy although some of the study populations have been very small and large variance between practitioners has been demonstrated with the colonoscopy technique [2,3]. Audit from England and Wales suggests that completion rates for colonoscopy in the NHS may be as low as 50% [4]. Two studies from the U.S.A. have claimed completion rates of 97% for colonoscopy but in selected healthy and relatively young (mean ages 62.9 and 59.8 years respectively) populations [5,6]. A retrospective single centre study of 307 unselected patients undergoing DCBE found a sensitivity of 96% for colorectal carcinoma detection, but Author for correspondence and guarantor of study: Dr D. J. A. Connolly, Department of Neuroradiology, Newcastle General Hospital, Westgate Road, Newcastle-upon-Tyne NE4 6BE, U.K. Tel: 44 (0)191 273 8811; Fax: 44 (0)191 273 1613. A grant to fund this research was obtained from Oxfordshire Health Clinical Audit Services. 0009-9260/02/010029+04 $35.00/0
by the criteria used in our study there were two falsenegative examinations and the sensitivity would be 92% [7]. In a separate study, combination of the DCBE technique with ¯exible sigmoidoscopy in 462 patients appeared to raise the sensitivity for colorectal carcinoma to 100%, though no follow-up period was speci®ed [8]. The aim of this study was to determine the sensitivity and speci®city of the DCBE when performed at a single centre with a single examination technique. If optimum conditions are achieved, how accurate is the DCBE compared to colonoscopy, an examination technique which is operator dependent and has a higher complication rate [4].
METHODOLOGY
In a retrospective study, we reviewed the medical case notes of 1201 patients who had had a DCBE performed over a three-year period in a single institution as documented on the Radiology Department data base (John Radclie Hospital, Oxford, U.K.). During this time period, a total of approximately 3600 DCBE studies were performed, but many of the patients were referred directly by GPs and did not have hospital case notes. Comparison of the radiology report was made with the subsequent clinical outcome over a two-year follow-up period and where possible with ®ndings at laparotomy and histology in # 2002 The Royal College of Radiologists
30
CLINICAL RADIOLOGY
order to assess the sensitivity and speci®city of this procedure for the detection of colorectal carcinoma. The DCBE studies were performed by either a radiology registrar under the supervision of a single gastrointestinal radiology consultant, or by the consultant gastrointestinal radiologist, and all used a standard technique which is described below. Bowel preparation involved a combination of a low residue and mainly ¯uid diet, two sachets of sodium picosulphate (Picolax, Nordic, Langley, U.K.) and an increased ¯uid intake on the day before the examination. Smooth muscle relaxants, either hyoscine butylbromide 20 mg (Buscopan, Boehringer Ingelheim, Bracknell, U.K.) or glucagon 1 mg (Lilly, Basingstoke, U.K.), were routinely given intravenously. Barium suspension was prepared according to the manufacturer's instructions with addition of `bubble-breaker' (Lafayette Healthcare, South Normanton, U.K.). Using an overcouch tube, an AP supine view of the entire colon, a prone angled view of the rectosigmoid colon, and right and left decubitus views were obtained. The image intensi®er was used to obtain a lateral view of the rectum, a supine right anterior oblique view of the sigmoid, a prone left posterior oblique view of the sigmoid, a supine or supine oblique view of the caecum, and upright oblique views of the splenic and hepatic ¯exures. Further spot views were obtained as required. We used a follow-up time period of two years after the DCBE as the limit for later detection of colorectal carcinoma to be termed false-negative studies. Previous studies of colorectal carcinoma detection by barium enema have shown a variable follow-up period between one and three years [3,9±11]. RESULTS
Symptomatic patients were referred from both outpatient clinics and GPs. One thousand two hundred and one sets of medical case notes of adult patients who had undergone a DCBE were reviewed. Three hundred and twenty-one patients were excluded from the study because the followup period was less than 2 years. This left a study population of 880 patients, mean age 64.8 years. Seventy-four cases of con®rmed colorectal carcinoma were diagnosed at DCBE. Four cases were found to have been false-positive for colorectal carcinoma at DCBE. In two of the false positive patients, a diagnosis of severe sigmoid diverticular disease was made on histology after sigmoidectomies. In the other two false positive patients, normal endoscopy and two years' follow-up demonstrated no evidence of colorectal carcinoma in the caecum or sigmoid colon respectively. There were 794 true-negative cases at two years' follow-up. Eight patients with colorectal carcinoma had a falsenegative DCBE study. In six of these cases, the subsequent cancer was in the recto-sigmoid colon. Six of the eight falsenegative studies were single rather than double reported, ®ve of these by a registrar. In one of the false-negative studies a ¯exible sigmoidoscopy was advised due to a tortuous sigmoid colon. The ¯exible sigmoidoscopy which
followed gave the diagnosis of colorectal carcinoma with little delay in patient management. Poor bowel preparation in two cases and sigmoid colon spasm in a further case were noted in the radiology report of the false-negative studies. The sensitivity and speci®city of the DCBE for the detection of colorectal carcinoma were therefore calculated at 90.2% and 99.5% respectively. The accuracy of the DCBE for colorectal carcinoma was 94.9%. DISCUSSION
Glick et al. reviewed studies of the sensitivity and speci®city of the DCBE and found most sensitivities above 85% and speci®cities of about 98% [12]. Glick et al. also reviewed the bias in many of these studies dependent upon the length of follow-up and selection of cases. For example, a longer follow-up period would increase the false-negative rate in a study of this type. This study has a two-year follow-up period, the same time period in which a patient with large bowel symptoms could be expected to be diagnosed with a colorectal carcinoma after a false-negative DCBE [10,13]. The selection of patients is positively biased in our study by the requirement of the patient to have hospital case records, thus eliminating from our study the large population of patients referred by a GP, who proceeded to have a negative study and then no further follow-up. The geography of the local region means that few if any patients would have been referred to another hospital for later management of a possible colorectal carcinoma, thus reducing the possibility of other falsenegative reports. Many studies of the investigation of suspected colorectal carcinoma have positive selection bias as only patients who were known to have colorectal carcinomas were included and speci®city could not be calculated. We believe that the design of this study has at least partially negated positive selection bias by reviewing all consecutive sets of case notes that were available, thereby allowing assessment of both sensitivity and speci®city [3,11,14±20]. Thomas et al. completed a multicentre audit of positively selected patients with proven colorectal carcinomas who had a barium enema in the 12 months preceding the histological diagnosis [18]. They discovered an 85% sensitivity for the detection of colorectal carcinomas. Strùm et al. reviewed 571 patients with colorectal carcinoma [11]. They demonstrated a sensitivity of 89.8% for the DCBE in the detection of colorectal carcinoma compared to 80.0% for colonoscopy with a three-year follow-up period. This surprisingly low sensitivity for colonoscopy is presumed to relate to increased diculty in performing this procedure in an older symptomatic population. These and other papers describing the DCBE have shown that most missed tumours were located in the caecum/ ascending colon or rectosigmoid colon [9,11,18,21,22]. Winawer et al. reported that the con®dence level for reporting DCBEs is lowest in these two regions due to redundancy and diverticulosis in the sigmoid colon and the presence of faecal residue and poor mucosal coating in the ascending colon and caecum [23]. Flexible sigmoidoscopy
31
THE DOUBLE CONTRAST BARIUM ENEMA
may have reduced the number of missed tumours if routinely employed in DCBE examinations where the sigmoid colon is not entirely normal, e.g. where there is moderate to severe diverticular disease. Law et al. reviewed 488 patients who underwent both DCBE ( performed by a radiographer and then double reported with a radiologist) and either colonoscopy or ¯exible sigmoidoscopy [24]. They assumed that the two techniques between them detected all carcinomas, found 34 cases of colorectal carcinoma within the study population and demonstrated a sensitivity for detection of colorectal carcinoma by DCBE of 97%. However, there is positive selection bias in this study as the 488 patients were selected from a total population having barium enemas of 5227 over 5 years, who had also to have undergone lower GI endoscopy. No follow-up period was speci®ed. If the optimal conditions and standards for the DCBE are achieved, then how signi®cantly dierent are the sensitivity and speci®city of this test when compared to colonoscopy? Unfortunately, most directly comparative studies use colonoscopy as the gold standard rather than clinical follow up or randomization and thus few studies have calculated speci®city [23,25]. The completion rates for colonoscopy vary hugely depending upon the clinical setting of the study and the expertise of the operator. Rex et al. noted that there was a large variation in sensitivity of colonoscopy in colorectal cancer detection depending on whether a gastroenterologist or non-gastroenterologist performed the study (97% versus 87% respectively) [3]. Crawley et al. reviewed radiographers' performance of the barium enema and found that diagnostic accuracy was not compromised, but that radiation dose increased [26]. The sensitivity and speci®city of this study thus compare favourably with the published data for colonoscopy in the detection of colorectal carcinoma, 80±97% and 91% respectively [3,11,25]. Flexible sigmoidoscopy alone will miss colorectal carcinomas in the signi®cant proportion of patients with proximal neoplasms [5,6]. Flexible sigmoidoscopy, with barium enema if negative, has been shown to be as sensitive as colonoscopy alone [8]. The cost of a barium enema and a ¯exible sigmoidoscopy is about the same as a colonoscopy [12]. However, typical complication rates of diagnostic colonoscopy are 1/1 000 for perforation, 3/1 000 for major haemorrhage and 1±3/10 000 for death. The risk of bowel perforation of the DCBE is approximately 1/25 000 [4,27]. DCBE may, however, lead to the recommendation of colonoscopy and polyp removal, thus raising the complication rate of this diagnostic algorithm. CT `virtual colonoscopy' may prove the most accurate method of investigating the large bowel in the near future but large study data is not yet available [28]. There were 8 false-negative examinations in this study, of which 6 were single reported (5 of these 6 by a registrar). When we reviewed 403 consecutive cases entered into the study 176 (43.6%) were found to be single reported, with 85 of these (21.1%) being single reported by a registrar rather than a consultant. This data supports the argument for
double reporting of all DCBE studies with at least one reporter being a consultant gastro-radiologist [11,29]. The double-contrast barium enema demonstrates a very high speci®city (4 99%) and a greater than 90% sensitivity for the detection of colorectal carcinoma in a symptomatic population with a two-year follow-up period. Double reporting of all DCBE studies may raise sensitivity. If DCBE was routinely combined with ¯exible sigmoidoscopy sensitivity may be further increased.
Acknowledgements. We wish to thank Miss Jacqueline Wright and her colleagues for their extensive secretarial, administrative and practical assistance with the completion of this project.
REFERENCES 1 OPCS Mortality statistics. Cause. 1993. HMSO: London, 1995. Series DH2. 2 Durdey P, Weston PMT, Williams NS. Colonoscopy or barium enema as initial investigation of colonic disease. Lancet 1987;2: 549±551. 3 Rex DK, Rahmani EY, Haseman JH, Lemmel GT, Kaster S, Buckley JS. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology 1997;112:17±23. 4 Executive NHS. Guidance on Commissioning Cancer Services: Improving Outcomes in Colorectal Cancer ± the Manual. London: Department of Health, 1997. 5 Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. N Engl J Med 2000;343:162±168. 6 Imperiale TF, Wagner DR, Lin CY, Larkin GN, Rogge JD, Ransoho DF. Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal ®ndings. N Engl J Med 2000;343:169±174. 7 Bloom®eld JA. Reliability of barium enema in detecting colonic neoplasia. Med J Aust 1981;1:631±633. 8 Brewster NT, Grieve DC, Saunders JH. Double-contrast barium enema and ¯exible sigmoidoscopy for routine colonic investigation. Br J Surg 1994;81:445±447. 9 Foster DR. Audit of barium enema examinations in the diagnosis of colorectal carcinoma. Australas Radiol 1998;42:197±198. 10 Dixon PM, Roulston ME, Nolan DJ. The small bowel enema: a ten year review. Clin Radiol 1993;47:46±48. 11 Strom E, Larsen JL. Colon cancer at barium enema examination and colonoscopy: a study from the county of Hordaland, Norway. Radiology 1998;211(1):211±214. 12 Glick S, Wagner JL, Johnson CD. Cost eectiveness of double contrast barium enema in screening for colorectal cancer. AJR Am J Roentgenol 1998;170:629±636. 13 Ott DJ, Gelfand DW. Colorectal tumors: pathology and detection. AJR Am J Roentgenol 1978;131:691±695. 14 Rex DK, Lehmann GA, Lappas JC, Miller RE. Sensitivity of double-contrast barium study for left-colon-polyps. Radiology 1986;158:69±72. 15 Thoeni RF, Petras A. Double-contrast barium-enema examination and endoscopy in the detection of polypoid lesions in the cecum and ascending colon. Radiology 1982;144:257±260. 16 Fork F-T. Radiographic ®ndings in overlooked colon carcinomas. A retrospective analysis. Acta Radiologica 1988;29:331±336. 17 Johnson CD, Carlson HC, Taylor WF, Weiland LP. Barium enemas of carcinoma of the colon: sensitivity of double- and singlecontrast studies. AJR Am J Roentgenol 1983;140:1143±1149. 18 Thomas RD, Fairhurst JJ, Frost RA. Wessex Regional Radiology Audit: barium enema in colorectal carcinoma. Clin Radiol 1995;50: 647±650.
32
CLINICAL RADIOLOGY
19 Steine S, Stordahl A, Lunde O, Loken K, Laerum E. Doublecontrast barium enema versus colonoscopy in the diagnosis of neoplastic disorders: aspects of decision-making in general practice. Fam Pract 1993;10:288±291. 20 Williams CB, Hunt RH, Loose H, Riddell RH, Sakai Y, Swarbrick ET. Colonoscopy in the management of colon polyps. Br J Surg 1974;61:673±682. 21 Anderson N, Cook HB, Coates R. Colonoscopically detected colorectal cancer missed on barium enema. Gastrointest Radiol 1991;16(2):123±127. 22 Bolin S, Franzen L, Nilsson E, Sjodahl R. Carcinoma of the colon and rectum. Tumors missed by radiologic examination in 61 patients. Cancer 1988;61(10):1999±2008. 23 Winawer SJ, Stewart ET, Zauber AG, et al. A comparison of colonoscopy and double-contrast barium enema for surveillance after polypectomy. N Engl J Med 2000;342:1766±1772.
24 Law RL, Longsta AJ, Slack N. A retrospective 5-year study on the accuracy of the barium enema examination performed by radiographers. Clin Radiol 1999;54:80±84. 25 Winawer SJ, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997;112:594±642. 26 Crawley MT, Shine B, Booth A. Radiation dose and diagnosticity of barium enema examinations by radiographers and radiologists: a comparative study. Br J Radiol 1998;71:399±405. 27 MacCarty RL. Colorectal cancer: the case for barium enema. Mayo Clin Proc 1992;67:253±257. 28 Ferrucci JT. Colonoscopy and barium enema: radiologist's response. Gastroenterology 1997;112:294±297. 29 Brady AP, Stevenson GW, Stevenson I. Colorectal cancer overlooked at barium enema examination and colonoscopy: a continuing perceptual problem. Radiology 1994;192(2): 373±378 [see comments].
The Double Contrast Barium Enema: A Retrospective Single Centre Audit of the Detection of Colorectal Carcinomas DA NI E L J . A . CO N NO L LY, Z O EÈ C . T R A I L L , H E L E N S . R E I D , SU S A N J . CO P L E Y, DANIEL J. NOLAN Department of Radiology, John Radclie Hospital, Oxford, U.K. Received: 30 November 2000 Accepted: 23 January 2001 AIM: To determine retrospectively the sensitivity and speci®city of the double contrast barium enema (DCBE) as performed in one institution for the detection of colorectal carcinoma. SUBJECTS AND METHODS: Eight hundred and eighty barium enema reports were reviewed of consecutive adult patients who underwent DCBE and also had hospital case notes with a minimum follow up of two years, a later diagnostic colonoscopy, or operative and histological ®ndings. RESULTS: Seventy-four true positive cases of colorectal carcinoma diagnosed at DCBE were con®rmed at surgery and histological examination. There were four false positive diagnoses of carcinoma at DCBE. Eight false negative cases at DCBE were demonstrated within a two-year followup period. The sensitivity of the DCBE for detecting colorectal carcinoma was therefore 90.2% and the speci®city was 99.5%. CONCLUSION: DCBE is a sensitive and highly speci®c investigation for the detection of colorectal carcinoma. Connolly D. J. A., et al. (2002) Clinical Radiology 57, 29±32. # 2002 The Royal College of Radiologists Key words: barium enema, colorectal cancer, sensitivity, speci®city, double reporting.
Colorectal carcinoma is responsible for over 19 000 deaths per annum in the U.K. [1]. The double contrast barium enema (DCBE) has been a standard method of examining the large intestine for many years. The emergence of ¯exible sigmoidoscopy and colonoscopy with their high reported sensitivities and speci®cities for the detection of colorectal carcinoma, has led to their adoption in many centres as the method of choice for investigating the large intestine. The previously reported sensitivity and speci®city of the DCBE has generally been found to be below that of colonoscopy although some of the study populations have been very small and large variance between practitioners has been demonstrated with the colonoscopy technique [2,3]. Audit from England and Wales suggests that completion rates for colonoscopy in the NHS may be as low as 50% [4]. Two studies from the U.S.A. have claimed completion rates of 97% for colonoscopy but in selected healthy and relatively young (mean ages 62.9 and 59.8 years respectively) populations [5,6]. A retrospective single centre study of 307 unselected patients undergoing DCBE found a sensitivity of 96% for colorectal carcinoma detection, but Author for correspondence and guarantor of study: Dr D. J. A. Connolly, Department of Neuroradiology, Newcastle General Hospital, Westgate Road, Newcastle-upon-Tyne NE4 6BE, U.K. Tel: 44 (0)191 273 8811; Fax: 44 (0)191 273 1613. A grant to fund this research was obtained from Oxfordshire Health Clinical Audit Services. 0009-9260/02/010029+04 $35.00/0
by the criteria used in our study there were two falsenegative examinations and the sensitivity would be 92% [7]. In a separate study, combination of the DCBE technique with ¯exible sigmoidoscopy in 462 patients appeared to raise the sensitivity for colorectal carcinoma to 100%, though no follow-up period was speci®ed [8]. The aim of this study was to determine the sensitivity and speci®city of the DCBE when performed at a single centre with a single examination technique. If optimum conditions are achieved, how accurate is the DCBE compared to colonoscopy, an examination technique which is operator dependent and has a higher complication rate [4].
METHODOLOGY
In a retrospective study, we reviewed the medical case notes of 1201 patients who had had a DCBE performed over a three-year period in a single institution as documented on the Radiology Department data base (John Radclie Hospital, Oxford, U.K.). During this time period, a total of approximately 3600 DCBE studies were performed, but many of the patients were referred directly by GPs and did not have hospital case notes. Comparison of the radiology report was made with the subsequent clinical outcome over a two-year follow-up period and where possible with ®ndings at laparotomy and histology in # 2002 The Royal College of Radiologists
30
CLINICAL RADIOLOGY
order to assess the sensitivity and speci®city of this procedure for the detection of colorectal carcinoma. The DCBE studies were performed by either a radiology registrar under the supervision of a single gastrointestinal radiology consultant, or by the consultant gastrointestinal radiologist, and all used a standard technique which is described below. Bowel preparation involved a combination of a low residue and mainly ¯uid diet, two sachets of sodium picosulphate (Picolax, Nordic, Langley, U.K.) and an increased ¯uid intake on the day before the examination. Smooth muscle relaxants, either hyoscine butylbromide 20 mg (Buscopan, Boehringer Ingelheim, Bracknell, U.K.) or glucagon 1 mg (Lilly, Basingstoke, U.K.), were routinely given intravenously. Barium suspension was prepared according to the manufacturer's instructions with addition of `bubble-breaker' (Lafayette Healthcare, South Normanton, U.K.). Using an overcouch tube, an AP supine view of the entire colon, a prone angled view of the rectosigmoid colon, and right and left decubitus views were obtained. The image intensi®er was used to obtain a lateral view of the rectum, a supine right anterior oblique view of the sigmoid, a prone left posterior oblique view of the sigmoid, a supine or supine oblique view of the caecum, and upright oblique views of the splenic and hepatic ¯exures. Further spot views were obtained as required. We used a follow-up time period of two years after the DCBE as the limit for later detection of colorectal carcinoma to be termed false-negative studies. Previous studies of colorectal carcinoma detection by barium enema have shown a variable follow-up period between one and three years [3,9±11]. RESULTS
Symptomatic patients were referred from both outpatient clinics and GPs. One thousand two hundred and one sets of medical case notes of adult patients who had undergone a DCBE were reviewed. Three hundred and twenty-one patients were excluded from the study because the followup period was less than 2 years. This left a study population of 880 patients, mean age 64.8 years. Seventy-four cases of con®rmed colorectal carcinoma were diagnosed at DCBE. Four cases were found to have been false-positive for colorectal carcinoma at DCBE. In two of the false positive patients, a diagnosis of severe sigmoid diverticular disease was made on histology after sigmoidectomies. In the other two false positive patients, normal endoscopy and two years' follow-up demonstrated no evidence of colorectal carcinoma in the caecum or sigmoid colon respectively. There were 794 true-negative cases at two years' follow-up. Eight patients with colorectal carcinoma had a falsenegative DCBE study. In six of these cases, the subsequent cancer was in the recto-sigmoid colon. Six of the eight falsenegative studies were single rather than double reported, ®ve of these by a registrar. In one of the false-negative studies a ¯exible sigmoidoscopy was advised due to a tortuous sigmoid colon. The ¯exible sigmoidoscopy which
followed gave the diagnosis of colorectal carcinoma with little delay in patient management. Poor bowel preparation in two cases and sigmoid colon spasm in a further case were noted in the radiology report of the false-negative studies. The sensitivity and speci®city of the DCBE for the detection of colorectal carcinoma were therefore calculated at 90.2% and 99.5% respectively. The accuracy of the DCBE for colorectal carcinoma was 94.9%. DISCUSSION
Glick et al. reviewed studies of the sensitivity and speci®city of the DCBE and found most sensitivities above 85% and speci®cities of about 98% [12]. Glick et al. also reviewed the bias in many of these studies dependent upon the length of follow-up and selection of cases. For example, a longer follow-up period would increase the false-negative rate in a study of this type. This study has a two-year follow-up period, the same time period in which a patient with large bowel symptoms could be expected to be diagnosed with a colorectal carcinoma after a false-negative DCBE [10,13]. The selection of patients is positively biased in our study by the requirement of the patient to have hospital case records, thus eliminating from our study the large population of patients referred by a GP, who proceeded to have a negative study and then no further follow-up. The geography of the local region means that few if any patients would have been referred to another hospital for later management of a possible colorectal carcinoma, thus reducing the possibility of other falsenegative reports. Many studies of the investigation of suspected colorectal carcinoma have positive selection bias as only patients who were known to have colorectal carcinomas were included and speci®city could not be calculated. We believe that the design of this study has at least partially negated positive selection bias by reviewing all consecutive sets of case notes that were available, thereby allowing assessment of both sensitivity and speci®city [3,11,14±20]. Thomas et al. completed a multicentre audit of positively selected patients with proven colorectal carcinomas who had a barium enema in the 12 months preceding the histological diagnosis [18]. They discovered an 85% sensitivity for the detection of colorectal carcinomas. Strùm et al. reviewed 571 patients with colorectal carcinoma [11]. They demonstrated a sensitivity of 89.8% for the DCBE in the detection of colorectal carcinoma compared to 80.0% for colonoscopy with a three-year follow-up period. This surprisingly low sensitivity for colonoscopy is presumed to relate to increased diculty in performing this procedure in an older symptomatic population. These and other papers describing the DCBE have shown that most missed tumours were located in the caecum/ ascending colon or rectosigmoid colon [9,11,18,21,22]. Winawer et al. reported that the con®dence level for reporting DCBEs is lowest in these two regions due to redundancy and diverticulosis in the sigmoid colon and the presence of faecal residue and poor mucosal coating in the ascending colon and caecum [23]. Flexible sigmoidoscopy
31
THE DOUBLE CONTRAST BARIUM ENEMA
may have reduced the number of missed tumours if routinely employed in DCBE examinations where the sigmoid colon is not entirely normal, e.g. where there is moderate to severe diverticular disease. Law et al. reviewed 488 patients who underwent both DCBE ( performed by a radiographer and then double reported with a radiologist) and either colonoscopy or ¯exible sigmoidoscopy [24]. They assumed that the two techniques between them detected all carcinomas, found 34 cases of colorectal carcinoma within the study population and demonstrated a sensitivity for detection of colorectal carcinoma by DCBE of 97%. However, there is positive selection bias in this study as the 488 patients were selected from a total population having barium enemas of 5227 over 5 years, who had also to have undergone lower GI endoscopy. No follow-up period was speci®ed. If the optimal conditions and standards for the DCBE are achieved, then how signi®cantly dierent are the sensitivity and speci®city of this test when compared to colonoscopy? Unfortunately, most directly comparative studies use colonoscopy as the gold standard rather than clinical follow up or randomization and thus few studies have calculated speci®city [23,25]. The completion rates for colonoscopy vary hugely depending upon the clinical setting of the study and the expertise of the operator. Rex et al. noted that there was a large variation in sensitivity of colonoscopy in colorectal cancer detection depending on whether a gastroenterologist or non-gastroenterologist performed the study (97% versus 87% respectively) [3]. Crawley et al. reviewed radiographers' performance of the barium enema and found that diagnostic accuracy was not compromised, but that radiation dose increased [26]. The sensitivity and speci®city of this study thus compare favourably with the published data for colonoscopy in the detection of colorectal carcinoma, 80±97% and 91% respectively [3,11,25]. Flexible sigmoidoscopy alone will miss colorectal carcinomas in the signi®cant proportion of patients with proximal neoplasms [5,6]. Flexible sigmoidoscopy, with barium enema if negative, has been shown to be as sensitive as colonoscopy alone [8]. The cost of a barium enema and a ¯exible sigmoidoscopy is about the same as a colonoscopy [12]. However, typical complication rates of diagnostic colonoscopy are 1/1 000 for perforation, 3/1 000 for major haemorrhage and 1±3/10 000 for death. The risk of bowel perforation of the DCBE is approximately 1/25 000 [4,27]. DCBE may, however, lead to the recommendation of colonoscopy and polyp removal, thus raising the complication rate of this diagnostic algorithm. CT `virtual colonoscopy' may prove the most accurate method of investigating the large bowel in the near future but large study data is not yet available [28]. There were 8 false-negative examinations in this study, of which 6 were single reported (5 of these 6 by a registrar). When we reviewed 403 consecutive cases entered into the study 176 (43.6%) were found to be single reported, with 85 of these (21.1%) being single reported by a registrar rather than a consultant. This data supports the argument for
double reporting of all DCBE studies with at least one reporter being a consultant gastro-radiologist [11,29]. The double-contrast barium enema demonstrates a very high speci®city (4 99%) and a greater than 90% sensitivity for the detection of colorectal carcinoma in a symptomatic population with a two-year follow-up period. Double reporting of all DCBE studies may raise sensitivity. If DCBE was routinely combined with ¯exible sigmoidoscopy sensitivity may be further increased.
Acknowledgements. We wish to thank Miss Jacqueline Wright and her colleagues for their extensive secretarial, administrative and practical assistance with the completion of this project.
REFERENCES 1 OPCS Mortality statistics. Cause. 1993. HMSO: London, 1995. Series DH2. 2 Durdey P, Weston PMT, Williams NS. Colonoscopy or barium enema as initial investigation of colonic disease. Lancet 1987;2: 549±551. 3 Rex DK, Rahmani EY, Haseman JH, Lemmel GT, Kaster S, Buckley JS. Relative sensitivity of colonoscopy and barium enema for detection of colorectal cancer in clinical practice. Gastroenterology 1997;112:17±23. 4 Executive NHS. Guidance on Commissioning Cancer Services: Improving Outcomes in Colorectal Cancer ± the Manual. London: Department of Health, 1997. 5 Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. N Engl J Med 2000;343:162±168. 6 Imperiale TF, Wagner DR, Lin CY, Larkin GN, Rogge JD, Ransoho DF. Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal ®ndings. N Engl J Med 2000;343:169±174. 7 Bloom®eld JA. Reliability of barium enema in detecting colonic neoplasia. Med J Aust 1981;1:631±633. 8 Brewster NT, Grieve DC, Saunders JH. Double-contrast barium enema and ¯exible sigmoidoscopy for routine colonic investigation. Br J Surg 1994;81:445±447. 9 Foster DR. Audit of barium enema examinations in the diagnosis of colorectal carcinoma. Australas Radiol 1998;42:197±198. 10 Dixon PM, Roulston ME, Nolan DJ. The small bowel enema: a ten year review. Clin Radiol 1993;47:46±48. 11 Strom E, Larsen JL. Colon cancer at barium enema examination and colonoscopy: a study from the county of Hordaland, Norway. Radiology 1998;211(1):211±214. 12 Glick S, Wagner JL, Johnson CD. Cost eectiveness of double contrast barium enema in screening for colorectal cancer. AJR Am J Roentgenol 1998;170:629±636. 13 Ott DJ, Gelfand DW. Colorectal tumors: pathology and detection. AJR Am J Roentgenol 1978;131:691±695. 14 Rex DK, Lehmann GA, Lappas JC, Miller RE. Sensitivity of double-contrast barium study for left-colon-polyps. Radiology 1986;158:69±72. 15 Thoeni RF, Petras A. Double-contrast barium-enema examination and endoscopy in the detection of polypoid lesions in the cecum and ascending colon. Radiology 1982;144:257±260. 16 Fork F-T. Radiographic ®ndings in overlooked colon carcinomas. A retrospective analysis. Acta Radiologica 1988;29:331±336. 17 Johnson CD, Carlson HC, Taylor WF, Weiland LP. Barium enemas of carcinoma of the colon: sensitivity of double- and singlecontrast studies. AJR Am J Roentgenol 1983;140:1143±1149. 18 Thomas RD, Fairhurst JJ, Frost RA. Wessex Regional Radiology Audit: barium enema in colorectal carcinoma. Clin Radiol 1995;50: 647±650.
32
CLINICAL RADIOLOGY
19 Steine S, Stordahl A, Lunde O, Loken K, Laerum E. Doublecontrast barium enema versus colonoscopy in the diagnosis of neoplastic disorders: aspects of decision-making in general practice. Fam Pract 1993;10:288±291. 20 Williams CB, Hunt RH, Loose H, Riddell RH, Sakai Y, Swarbrick ET. Colonoscopy in the management of colon polyps. Br J Surg 1974;61:673±682. 21 Anderson N, Cook HB, Coates R. Colonoscopically detected colorectal cancer missed on barium enema. Gastrointest Radiol 1991;16(2):123±127. 22 Bolin S, Franzen L, Nilsson E, Sjodahl R. Carcinoma of the colon and rectum. Tumors missed by radiologic examination in 61 patients. Cancer 1988;61(10):1999±2008. 23 Winawer SJ, Stewart ET, Zauber AG, et al. A comparison of colonoscopy and double-contrast barium enema for surveillance after polypectomy. N Engl J Med 2000;342:1766±1772.
24 Law RL, Longsta AJ, Slack N. A retrospective 5-year study on the accuracy of the barium enema examination performed by radiographers. Clin Radiol 1999;54:80±84. 25 Winawer SJ, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997;112:594±642. 26 Crawley MT, Shine B, Booth A. Radiation dose and diagnosticity of barium enema examinations by radiographers and radiologists: a comparative study. Br J Radiol 1998;71:399±405. 27 MacCarty RL. Colorectal cancer: the case for barium enema. Mayo Clin Proc 1992;67:253±257. 28 Ferrucci JT. Colonoscopy and barium enema: radiologist's response. Gastroenterology 1997;112:294±297. 29 Brady AP, Stevenson GW, Stevenson I. Colorectal cancer overlooked at barium enema examination and colonoscopy: a continuing perceptual problem. Radiology 1994;192(2): 373±378 [see comments].