Biennial fecal occult-blood screening reduced colorectal cancer mortality
ACP J Club. 1997 May-Jun;126:62. doi:10.7326/ACPJC-1997-126-3-062
Kronborg O, Fenger C, Olsen J, Jørgensen OD, Søndergaard O. Randomised study of screening for colorectal cancer with faecal-occult-blood test. Lancet. 1996 Nov 30;348:1467-71.
To determine whether screening with biennial fecal occult-blood tests (FOBTs) reduces mortality from colorectal cancer.
Randomized controlled trial with 10-year follow-up.
A community in Denmark.
In 1985, residents of Funen, Denmark, aged 45 to 75 years were assessed. Exclusion criteria were known colorectal cancer, precursors of colorectal cancer (adenomas), or distant spread of all types of malignant disorders.
30 967 adults were assigned to biennial screening (Hemoccult-II tests). FOBTs were done with dietary restrictions (no red meat, fresh fruit, iron preparations, vitamin C, or aspirin or other nonsteroidal antirheumatic drugs for 3 d before testing) and collection of 2 samples from 3 consecutive stools. 20 672 adults had their first screening, and 14 203 attended all 5 screening rounds. 30 966 adults were assigned to the control group and received no information about the study.
Main outcome measures
Death from colorectal cancer and incidence of colorectal cancer and adenomas were ascertained from regional and national databases. The incidence and mortality ratios were calculated using the data from the control group as the denominator.
Analysis was by intention to treat. 481 patients in the screening group had colorectal cancer compared with 483 patients in the control group; the incidence ratio was 1.00 (95% CI 0.87 to 1.13). Fewer patients in the screening group died from colorectal cancer and complications of treatment (205 vs 249 deaths; number needed to treat [NNT] 704, CI 360 to 12 854; mortality ratio 0.82, CI 0.68 to 0.99) and from colorectal cancer alone (182 vs 230 deaths; NNT 645, CI 352 to 3689; mortality ratio 0.79, CI 0.65 to 0.96). More patients in the screening group had Dukes stage A colorectal cancer (22% vs 11%, P < 0.01), and more patients in the control group had more advanced colorectal cancer (Dukes stage C, distal spread, or no classification) (52% vs 44%). More adenomas were identified in patients in the screening group (413 vs 174; P < 0.01). The groups did not differ for death from all causes (6228 in the screening group vs 6303 in the control group, mortality ratio 0.99, CI 0.95 to 1.02).
Biennial screening with the fecal occult-blood testing reduced mortality from colorectal cancer. It did not reduce the incidence of colorectal cancer or total mortality.
Sources of funding: Danish Cancer Society; Helsefonde; County of Funen; Danish Research Council; University of Odense; Albani Foundation.
For article reprint: Professor O. Kronborg, Department of Surgery, Odense University Hospital, DK 5000 Odense C, Denmark; and Danish Epidemiology Science Centre, Aarhus University, Denmark. FAX 45-6613-2854.
With the publication of these 2 studies along with the publication of the Minnesota trial in 1993 (1), 3 of 3 randomized controlled trials from 3 countries have shown that screening with FOBTs reduces mortality from colorectal cancer. The test in all 3 studies was a guaiac-based FOBT used on 2 samples of 3 consecutive stools while the patient was receiving a special diet. The results of screening were considered positive if any sample was positive. Follow-up of positive screening tests was done with diagnostic colonoscopy and excision of polyps or early tumors. These trials do not indicate whether an alternative diagnostic test, such as double-contrast barium enema with or without flexible sigmoidoscopy, would have led to a similar reduction in mortality from colorectal cancer.
The effect for the Danish (Kronborg and colleagues) and British (Hardcastle and colleagues) studies was only about half that reported in the Minnesota trial: The Danish and British studies found an 18% and 16% reduction in mortality from colorectal cancer, respectively, whereas the Minnesota trial found a 33% reduction in the group that was screened yearly. Screening every other year was not found to be effective beyond chance in the Minnesota trial, but the observed effect in that trial was consistent with that in the Danish and British trials (95% CI for the cumulative mortality ratio was 0.68 to 1.31).
The 2 new studies are similar to each other in design but differ from the Minnesota trial in 4 ways, all of which might account for the lower effect size observed. First, the interval between screening examinations was different. In the Danish and British trials, screening was done every other year; in the Minnesota trial, both yearly and every- other-year screenings were studied. Second, stool samples were not rehydrated in the European studies, and so both sensitivity and false-positive rates were lower than those in the Minnesota study. Third, the Danish and British trials studied persons sampled from the general population, whereas the Minnesota trial comprised volunteers. Less-selected persons might be expected to be less compliant with testing. Finally, follow-up was somewhat shorter in the 2 recent trials, and effect size tends to increase with duration of time in the screening program. It is difficult to know which of the differences in research questions—screening interval, rehydration, population sampling, or length of follow-up—accounted for the difference in effect size, because all were either present or absent together in these 3 trials.
Although mortality from colorectal cancer was reduced, the incidence of colorectal cancer was not. FOBT use is most effective in detecting cancer and is not a very sensitive test for detecting adenomatous polyps, the precursor of most cases of colorectal cancer. Adenomatous polyps were detected more frequently in the screened groups; some were large enough to bleed and others were found incidentally during the work-up of false-positive FOBT results. But most polyps would not have been expected to progress to cancer during the several years of follow-up in these studies. Perhaps incidence will be found to be reduced after longer follow-up.
It can now be stated that FOBT screening prevents deaths from colorectal cancer beyond a reasonable doubt. Although screening every other year with nonrehydrated specimens is effective, many clinicians and patients will probably prefer the greater effectiveness of annual screening with rehydrated specimens, despite the higher cost and the trouble associated with the evaluation of false-positive screening test results. They may prefer to add sigmoidoscopy to FOBT or to choose other screening tests (double-contrast barium enema or colonoscopy), which are probably even more effective but are currently less well supported by evidence. Because the screening test options are so different from each other—not only in the strength of the evidence but also in potential effectiveness, convenience, cost, and risk—clinicians and individual patients should decide together which approach is right for them (2).
Robert H. Fletcher, MD, MSc
Harvard Medical SchoolBoston, Massachusetts, USA