Population based randomised study of uptake and yield of screening by flexible sigmoidoscopy compared with screening by faecal occult blood testingBMJ 1998; 317 doi: http://dx.doi.org/10.1136/bmj.317.7152.182 (Published 18 July 1998) Cite this as: BMJ 1998;317:182
- Julia E C W Verne, consultant in public health medicinea,
- Roger Aubrey, general practitionerb,
- Sharon B Love, medical statisticianc,
- Ian C Talbot, consultant histopathologistd,
- John M A Northover, honorary directora
- aICRF Colorectal Cancer Unit, St Mark's Hospital, Northwick Park, Middlesex HA1 3UJ
- bBridge Cottage Surgery, Welwyn, Hertfordshire AL6 9EF
- cICRF Medical Statistics Group, Centre for Statistics in Medicine, Institute of Health Sciences, Oxford OX3 7LF
- dDepartment of Histopathology, St Mark's Hospital
- Correspondence to: Dr J E C W Verne, North Thames Regional Office, London W2 3QR
Objectives: To compare the feasibility of mass screening by flexible sigmoidoscopy with screening by faecal occult blood testing (Haemoccult) and both tests combined.
Design: Patients were randomised to screening by flexible sigmoidoscopy, faecal blood testing, or both tests. The flexible sigmoidoscopy examinations were performed by a general practitioner.
Setting: General practice.
Subjects: 3744 patients aged 50-75 years.
Main outcome measures: Uptake, positive results, detection of neoplasia, complications, and recall for diagnostic colonoscopy.
Results: Uptake was significantly higher in the flexible sigmoidoscopy group (46.6%) than in the faecal blood test group (31.6%; P<0.001) or than in the group having both tests (30.1%; P<0.001). Telephone reminders increased uptake of sigmoidoscopy to 61.8%. In total, 1116 sigmoidoscopy examinations were performed without major complication. Polyps were found in 19.3% (95% confidence interval 17.0% to 21.6%) but only 6.8% (5.3% to 8.3%) had adenomas and 2.4% (1.5% to 3.3%) “high risk” adenomas. Cancer was detected in four subjects. The faecal blood test yielded positive results in 0.8% (0.2% to 1.4%) but missed at least one cancer and 30 cases of adenoma which were found by sigmoidoscopy in the combined group. Use of histological criteria—shown elsewhere to correlate with future risk of colorectal cancer —to select “positive” patients could reduce recall for diagnostic colonoscopy from about 20% to less than 5%.
Conclusions: Some of the predicted obstacles to screening with flexible sigmoidoscopy are surmountable. Clear evidence relating to efficacy will be obtained only from a randomised controlled trial.
Colorectal cancer is the second highest cause of death from cancer in England
Detection of premalignant adenomas by screening with flexible sigmoidoscopy offers the chance of reducing incidence as well as mortality
High uptake of flexible sigmoidoscopy screening is achievable provided accurate call up lists are used
Flexible sigmoidoscopy detects more adenomas and cancer than screening with a faecal occult blood test
Haphazard introduction of flexible sigmoidoscopy screening should be discouraged until there is more substantive evidence of its effectiveness
Colorectal cancer is the second highest cause of death from cancer in England and Wales.1.Recent evidence suggests that removal of neoplastic lesions at sigmoidoscopy can reduce the incidence of and mortality from distal colorectal cancer.26This has prompted calls for mass screening by flexible sigmoidoscopy. These data, from cohort and case-control studies, however, may be subject to biases,78 giving an overoptimistic impression of efficacy. Given the negative aspects of screening programmes,9policy makers have an obligation to ensure that the benefits (primarily lives saved) outweigh the costs before programmes are introduced.7This could be most objectively shown through a randomised controlled trial.7 This feasibility study was conducted to determine key features which could influence the design of such a trial.
Subjects and methods
The study was approved by East Hertfordshire ethics committee.
The study was conducted in one general practice. The catchment area had a higher proportion of patients from social classes I and II (56%) than in England and Wales as a whole (23.3%) and a lower proportion from ethnic minority groups.10 A list of practice patients from the family health services authority suggested that 3933 (29%) of the practice population were in the study age range (50-75 years). The general practitioner (RA) identified and removed from the list 189 (4.8%) patients who had died or moved or were ineligible for the study because of a previous diagnosis of colorectal neoplasia, investigation of the colon and rectum within the previous 2 years, and physical or mental disease contraindicating screening.
The remaining 3744 (50% men) nominally asymptomatic subjects were randomised. Throughout the study inaccuracies on the list and postal returns were recorded.
Households were randomised by using the random number generator in Minitab and invited, by post, to undergo flexible sigmoidoscopy, faecal occult blood testing (Haemoccult), or faecal occult blood testing plus flexible sigmoidoscopy. Reminders were not routinely sent.
It was found that inclusion of all eligible subjects would give more than 90% power to estimate the true prevalence rate of adenoma within 2%11 and to detect a 10% difference in compliance between groups (where one was 50%) at the 5% significance level.12
Telephone survey of non-responders to flexible sigmoidoscopy
As little was known about reasons for non-uptake of flexible sigmoidoscopy compared with faecal occult blood testing,13 a telephone survey of a random sample of 184 non-responders in the flexible sigmoidoscopy group was conducted to ascertain eligibility and, when appropriate, to make a second offer of flexible sigmoidoscopy screening.
An appointment (date and time), a sachet of laxative (sodium picosulphate-magnesium citrate (Picolax; Nordic)), and an explanatory booklet were sent 2 weeks in advance. Appointments could be changed or cancelled by telephone. Subjects were asked to give written consent to the examination.
All examinations were performed by the general practitioner (RA). To minimise recall for colonoscopy a pragmatic approach was adopted so that when diminutive polyps (<5 mm) that seemed hyperplastic were found only in the rectum these were removed at screening for histological examination. Subjects were recalled for colonoscopy only if histological examination showed an adenoma; those with other lesions were recalled for colonoscopy or surgery as appropriate.
Faecal occult blood test
The 3 day, six sample, diet restricted faecal occult blood test (Haemoccult, Rohm Pharma) was sent with a prepaid reply envelope and instruction booklet. The test was developed without rehydration. Patients were recalled if one or more windows yielded a positive result.
Faecal occult blood testing and flexible sigmoidoscopy
Subjects were asked to complete the faecal occult blood test before attending for flexible sigmoidoscopy. The faecal test was developed blind to the results of the flexible sigmoidoscopy examination and vice versa. Subjects were recalled for colonoscopy if either the faecal test yielded positive results or the findings at flexible sigmoidoscopy fulfilled the criteria described above.
Colonoscopy and histology
Colonoscopy was performed by the general practitioner. Subjects with adenomas were classified into high and low risk groups on the basis of lesions found at screening (high risk if at least one adenoma was cm, villous, or tubulovillous or showed features of severe dysplasia) on histological examination.5
Crude uptake rates were calculated as the number of responders per group and the number of invited per group.
The information collected on ineligibility (from postal returns and the telephone survey of non-responders) was used to estimate the uptake rate for flexible sigmoidoscopy which could be achieved given an accurate register of eligible subjects as the number responding to the postal or telephone invitations (number invited minus number found to be ineligible).
The implications for diagnostic colonoscopy workloads of using various histological criteria for referral were examined according to polyps seen (no histology); at least one adenoma; and only high risk adenoma(s).
The χ2 test for contingency tables was used to compare proportions. All reported P values are two tailed.
Uptake of screening tests
Crude uptake rates are shown in table 1. In the flexible sigmoidoscopy group the crude uptake rate (46.6%) was significantly higher than in the faecal occult blood test group (31.6%; P<0.001). Similarly, it was significantly higher than in the combined test group whether subjects did both tests (30.1%; P<0.001) or only one of the two tests (39.5%; P<0.001). Of the subjects in the combined test group doing only one test, 80% chose flexible sigmoidoscopy (94 24). The uptake of flexible sigmoidoscopy in the combined test group was 37.6%. Although this was significantly lower than the rate in the flexible sigmoidoscopy only group (46.6%; P<0.001) it was significantly higher than in the group that underwent faecal occult blood testing only (31.6%; P<0.01). Conversely, the crude rate for the faecal test in the combined test group was 32.0%, which was not significantly different from the comparable rate in the faecal test group.
The telephone survey of a random sample of non-responders in the flexible sigmoidoscopy group revealed that up to 16% of invitations could have been sent inappropriately. If we take account of both the number of ineligible subjects and the increased uptake after the telephone survey this gives an estimated uptake rate of 61.8% (95% confidence interval 57.3% to 66.3%) for flexible sigmoidoscopy in those eligible to be screened.
Endoscopic findings at flexible sigmoidoscopy
In total 1116 patients (51% men) underwent flexible sigmoidoscopy screening without major complication. Polyps were found in 138 (24.2%) men and 81 (14.9%) women (P<0.001), and two men had overt malignancy. Three subjects were referred for surgical resection (a woman with a 3 cm adenoma and two men with cancer). One hundred and ten subjects with polyps were recommended to return for colonoscopy. One man refused so his polyps were removed during the screening examination. A further 90 (41% of those with polyps) subjects with diminutive rectal polyps had these removed at screening. Seventeen (7.7%) subjects with diminutive (<5 mm) polyps had no intervention (two were taking warfarin and one had acute rectal prolapse after the laxative; contraindications for the others were not elucidated).
Findings at colonoscopy
Altogether 123 subjects(78 men) underwent colonoscopy; adenomas were removed from 14 at flexible sigmoidoscopy, and 109 were recalled directly without biopsy at flexible sigmoidoscopy. Eleven (10% of those who underwent colonoscopy) were found to have adenomas proximal to the sigmoid colon, 10 of whom had only a single adenoma and one who had three.
Clinical significance of neoplasias detectable by flexible sigmoidoscopy
The data on polyps and cancer of the sigmoid colon and rectum collected at flexible sigmoidoscopy or colonoscopy, or both, were combined. A diagnosis was assigned according to the most prognostically significant lesion in the 197 (91%) subjects with cancer or polyps for whom histological data were available (table 2). Four subjects had carcinoma (three Dukes' stage A, one Dukes' stage B) of the sigmoid colon or rectum, and 76 (35% of those with distal polyps or cancer) had at least one adenoma. Although polyps were detected in one in five subjects, the prevalence rates of neoplasia detected at screening were 0.4% for cancer and 6.8% for adenomas. Of the 76 patients with adenoma, a third were classified as “high risk” (2.4% (1.5% to 3.3%) of all subjects).
Comparison of rates of positive diagnosis
In total 854 patients underwent faecal occult blood testing alone or combined with flexible sigmoidoscopy. Seven (0.8%; 0.2% to 1.4%) had positive results and all underwent colonoscopy. One had a Dukes' stage C rectal carcinoma, one had a 2 cm adenoma, and a third had a 2 mm adenoma. A fourth subject had two diminutive adenomas, one detected at screening and the other at colonoscopy. The three remaining patients did not have neoplasia.
In the combined test group 401 subjects underwent both tests, and a comparison of the performance of the two tests was made. Only one subject in this group in whom a polyp was seen at flexible sigmoidoscopy had a positive result of the faecal blood test. In 81 subjects with negative test results polyps were found at flexible sigmoidoscopy; of these 30 had one or more adenomas (all less than 1 cm diameter) and one had a Dukes' stage A cancer (<2 cm diameter).
The aim of this study was to evaluate whether these obstacles are surmountable.
The estimated achievable uptake rate of flexible sigmoidoscopy (on the basis of an accurate list of eligible patients and a telephone reminder to non-compliers) of just over 60% compares favourably with the 29% in subjects offered flexible sigmoidoscopy after negative results of the faecal blood screening test in the United Kingdom.16.It is, however, lower than the 81% achieved in a population based Norwegian study in which reminders and press releases were used to boost uptake.17 In an Irish study 68% of volunteers preselected for their eligibility and willingness actually attended for flexible sigmoidoscopy.18
Certain features of the practice (enthusiasm of the primary care team for screening, social class profile of patients) could be expected to encourage higher uptake rates of screening. Both higher and lower uptake rates than those observed in this study would probably be reported, however, if screening were to be offered under different circumstances. It will be important to ascertain how widely rates differ and their most important determinants.
Comments made by subjects in the combined test group revealed possible reasons for the differential uptake rates for the two tests. These included the perceived immediacy of the flexible sigmoidoscopy screening and its results, less distaste for idea of sigmoidoscopy, the additional perceived benefit of consulting the general practitioner while undergoing sigmoidoscopy, and concern that failure to attend for screening might be noted by the doctor. There is support for the latter two factors from other studies of screening. 1319
Need for colonoscopy
A fundamental prerequisite for the introduction of screening is that there should be sufficient facilities for diagnosis and treatment of any lesions detected.20 Colonoscopy services at present cannot meet diagnostic and follow up needs in many districts.21
With data from our study it can be seen that the proportion recalled for diagnostic colonoscopy could be reduced from 20% to about 7% (5.3% to 8.3%) if polyps are biopsied at screening and only patients with adenoma are recalled. The prevalence rates for cancer and adenomas detected by screening in this study were similar to those found in other studies in asymptomatic subjects screened by 60 cm flexible sigmoidoscopy.2223 Recall of only those classified as at “high risk” of future colorectal cancer would result in a further halving of the numbers (2.4%; 1.5% to 3.3 %). This is similar to the proportion of subjects who would be recalled for colonoscopy as a result of a positive faecal blood test,24 but in contrast, “histological positivity” has more biological relevance and hence efficiency than “haematin positivity.”
The rate of positive results of the faecal occult blood test of 0.8% (0.2% to 1.4%) in this study was at the lower end of the range reported in the literature24and is probably due to close adherence to the dietary restrictions. This study confirms the greater sensitivity of flexible sigmoidoscopy compared with faecal occult blood testing for distal neoplasia but refutes the hypothesis that neoplasia detection by flexible sigmoidoscopy can be considerably enhanced by the addition of faecal occult blood testing.25 The observation that the test result was negative in many subjects who were found at flexible sigmoidoscopy to have adenomas or cancer was not surprising; all of the adenomas were less than 1 cm in diameter and the malignant polyp was less than 2 cm. It has been shown that faecal occult blood testing detects only a quarter of polyps greater than 1 cm.26
We found that given an accurate list of eligible subjects and a telephone reminder an uptake rate of over 60% is achievable even without the use of mass media campaigns. We have also shown that if the result of flexible sigmoidoscopy screening is defined as “positive” on the basis of the histological characteristics of polyps removed during the procedure rather than simply their detection, this will result in selection of subjects whose current and future risk of large and villous adenomas or cancer is considerably increased 52729 and will also reduce the recall rate for colonoscopy from about 20% to under 5%.
In this study the offer of screening by both faecal occult blood testing and flexible sigmoidoscopy had a detrimental effect on uptake and did not increase detection of neoplasia so we conclude that the synchronous offer of both tests is not worth while. Priority should now be given to completing a randomised trial and discouraging the haphazard introduction of flexible sigmoidoscopy screening without more substantive evidence of its effectiveness.
We thank the nurses and volunteers at the screening clinic, in particular Sister G Jellis, Mrs K Pearce, Mrs R Ashwell, and Mrs R McMillan; staff of the practice, in particular Mrs S Woods; and staff of the ICRF Colorectal Unit, in particular Mr K Miller and Mrs S Epstein for their assistance with and support for the study and Miss H Crowne for designing the database. We also thank the late Lt Col Sir Martin Gilliat.
Contributors: JECWV had the original idea for the study, formulated the hypothesis, designed the protocol, planned and coordinated the study, and wrote the paper. RA enabled the study to be undertaken in the general practice in which he is a principal; undertook all the flexible sigmoidoscopies and colonoscopies; and contributed from the stage of hypothesis formulation to all aspects of the study design. He contributed suggestions to the paper. SBL provided statistical advice from the stage of formulation of the hypothesis (including initial power calculations) to the study design and performed the randomisation and statistical analyses. She contributed suggestions to the paper. ICT examined all the samples histologically and provided diagnoses; he also contributed suggestions to the paper. JMAN directed the overall programme at the unit on screening for colorectal neoplasia and will act as guarantor. In this study, he provided guidance and support throughout, particularly in the development of the original idea, hypothesis formulation, protocol design, and writing of the paper.
Funding: Smith-Kline-Beecham donated the flexible sigmoidoscopes.
Conflict of interest: None.