Full text of DOI:10.1007/s003840100316

Int J Colorectal Dis (2001) 16:313–317
DOI 10.1007/s003840100316
O R I G I N A L A RT I C L E
M. Lamah
J. Norris
Effect of faecal occult blood testing
on colorectal cancer mortality in the
surveillance of subjects at moderate risk
of colorectal neoplasia: a case-control study
S. M. Caffarey
M. Broughton
C. G. Marks
Abstract Colonoscopy is the estab-
lished method of surveillance of sub- lection system of the screening pro-
jects at high risk of developing colo- gramme. Cases were less likely than
ers was obtained from the data col-
Accepted: 12 April 2001
Published online: 22 June 2001
© Springer-Verlag 2001
rectal neoplasia. Its role in the sur-
controls to have ever been screened,
veillance of a population at moderate with an odds ratio of 0.64 (95% con-
risk is less clear, however, as the
procedure is expensive, time con-
suming and occasionally hazardous.
fidence interval 0.34–1.15) for expo-
sure to at least one FOB testing.
There was no significant difference
The aim of this study was to estimate between the sub-groups according to
by case-control methods the effect of gender, age at diagnosis or location
faecal occult blood (FOB) screening of the cancer. The inverse associa-
on colorectal cancer (CRC) mortality tion between screening for faecal oc-
in a population at moderate risk of
developing CRC. Screening by FOB suggests that screening in a popula-
testing prior to diagnosis in patients tion at moderate risk of CRC can re-
over the age of 45 years who died of duce mortality from CRC in this
cult blood and fatal colorectal cancer
M. Lamah ( ) · J. Norris · S.M. Caffarey
✉
M. Broughton · C.G. Marks
CRC diagnosed in 1989–1998 was
compared with screening in controls
matched with the case for age and
sex. Information about episodes of
FOB testing and potential confound-
group.
Royal Surrey County Hospital, Surrey, UK
Present address:
Keywords Colorectal cancer ·
Screening · Faecal occult bloods
M. Lamah, 10 Langley Road,
London SW19 3NZ, UK
e-mail: marclamah@hotmail.com
duction in CRC mortality in the screening group of the
order of 15–20%. By itself however, such evidence of ef-
fectiveness in reducing mortality is insufficient to justify
the implementation of a national mass population screen-
ing programme: logistical, ethical, psychological and
economic issues are only some of the considerations that
must be addressed in order that the chance that the pa-
tient will benefit from the screening procedure substan-
tially outweighs the risk of harm from it. Indeed several
authors have argued against the use of FOBT in screen-
ing the general population [13, 14].
Targeted screening may be a more efficient and cost-
effective method of detecting early neoplasms than pop-
ulation screening. Colonoscopy is the established meth-
od of surveillance of subjects at high risk of developing
colorectal neoplasia [15]. These are subjects with a well-
Introduction
Colorectal cancer (CRC) is a leading cause of illness and
death in the Western world, with an estimated 169,400
new cases in the 12 countries of the European Communi-
ty in 1990 [1], and resulting in approximately 20,000
deaths annually in the United Kingdom. In the absence
of effective measures of primary prevention of CRC, and
with the knowledge that surgical treatment at the early
stages of malignancy is effective in improving survival
rates [2], the past two decades have witnessed increased
interest in CRC screening using faecal occult blood test-
ing (FOBT). Four randomised controlled trials of general
population screening [3, 4, 5, 6], together with data from
several case-control studies from Europe, the United
States and Japan [7, 8, 9, 10, 11, 12] have shown a re-

314
Table 1 Characteristics of cases (deaths from colorectal cancer)
defined genetic predisposition, or a strong family history
of CRC [16, 17]. There remains a broader ‘middle’
group of subjects, not so well-defined as the high-risk
group, with characteristics placing them at slightly high-
er risk of developing CRC than the general population. A
general practice based programme was initiated in
Guildford (U.K.) to offer such individuals screening of
faecal occult blood. This paper reports the results of a
case-control study of the efficacy of FOBT screening
among such a population.
and controls
Characteristics
Patients
(n=146)
Controls
(n=292)
n
%
n
%
Age at diagnosis (years)
45–55
55–64
65–74
75–85
22
23
61
40
15.1
15.8
41.8
27.4
43
46
122
81
14.7
15.8
41.8
27.7
Gender
Male
Female
Materials and methods
89
57
61.0
39.0
178
114
61.0
39.0
Study population
Anatomical location of cancer
Colon
Rectum
Undefined
The design of the study has been reported previously [18]. Since
1987 individuals at higher than average risk of developing large
bowel cancer have been identified from general practices in the
Guildford area and offered opportunistic screening in the form of
FOBT. Subjects with a positive result were invited to undergo a
colonoscopy or double-contrast barium enema where complete en-
doscopic exploration of the colon was not possible.
56
74
16
38.4
50.6
11.0
–
–
–
–
–
–
been living in the same area since the screening programme was
started; (d) had not had previous histories of CRC before the
screening programme was started in 1987. Cases were identified
through the data collection system of the screening programme.
The histological diagnosis of adenocarcinoma, its anatomical loca-
tion, and the cause of the patient’s death were verified by review-
ing the computerised data base of the cases diagnosed as CRC.
Death was attributed to CRC if advanced, or metastatic CRC was
present at the time of death, or if the person died in the post-opera-
tive phase.
For each case, two controls were randomly selected from the
files of the general practitioners in such a way as to reflect the
level of screening activity in the population from which the cases
arose and, specifically, were matched with cases for age and sex.
They were required to have been alive at the time when the
matched case patient died; the date of diagnosis of the case was
applied to the case-control matched set as the reference date, so as
to ensure comparable intervals for the screening history of cases
and controls; each control in the matched set thus had the same
opportunity as the corresponding case of undergoing a screening
For the purpose of this study we defined cases as individuals at
risk those who were over 45 years of age with one or more of the
following symptoms: rectal bleeding, change in bowel habit, ab-
dominal pain, tenesmus, anal or rectal discomfort and anaemia, or
a family history of one or more first-degree relatives with cancer
of the bowel (at age 50 years or younger), breast, endometrium,
liver, ovary or stomach. Patients with other potential sources of
occult rectal bleeding such as haemorrhoids or ulcerative colitis
were not excluded as it was felt that these diagnoses might conceal
bleeding from a carcinoma.
The FOBT was performed using the haemoccult test, accord-
ing to the manufacturers’ instructions. A kit for testing FOB was
given to patients, who were then asked to telephone the surgery
for results. Completed kits were returned to the surgery and col-
lected weekly, developed at the hospital, and the results posted
back to the surgeries.
Screening history
The screening histories of the cases and controls were retrieved test.
retrospectively from hospital or general practitioners’ files; details
With this choice of case and control, one is estimating the rela-
of each participant’s instances of faecal occult blood testing with tive risk of dying from the disease between groups with different
respect to date, indication and outcome were recorded.
screening histories. Efficacy of FOBT screening in reducing mor-
By definition, after the initial diagnosis of disease, an individu- tality from CRC is suggested if a screening history is more com-
al is no longer eligible for screening. The relevant screening histo- mon among controls than cases. A previous history of adenoma-
ry of the cases therefore consisted of all screening tests performed tous polyps or nonfatal cancer was not a grounds for exclusion.
from the time screening started until the time of diagnosis, but no
Table 1 shows the main characteristics of cases and controls.
tests performed between diagnosis and death. The relevant screen- The mean age at diagnosis was 68.5 years for cases and 68.6 years
ing history of a matched control was therefore also limited to the for controls. A total of 146 fatal cases of CRC (89 men, 57 wom-
time interval during which screening tests on the case would have en) met the case definition; 292 controls were eligible according
been included, i.e. until the time of diagnosis. For fatal cases of to the criteria set above. The anatomical distribution of the fatal
CRC detected by the screening FOBT, the FOBT that led to diag- cancers included 94 (64.4%) originating in the rectum or sigmoid
nosis was included with other screening FOBTs in all analyses.
colon, 14 (9.6%) in the left colon, 10 (6.8%) in the transverse co-
lon, and 28 (19.2%) in the right colon. Of these cancers, seven
(4.8%) were of Dukes’ A histological staging, 38 (26.0%) were
Dukes’ B, and 101 (69.2%) Dukes’ C or D.
Identification of cases and controls
Case subjects were defined as those who: (a) were diagnosed with
adenocarcinoma of the colon or rectum between 1989 and 1998,
who subsequently died before December 1998 as a result of this
Statistical analysis
cancer; (b) were diagnosed as having CRC between the ages of 45 Conditional logistic regression analysis was used to estimate odds
and 85 years after the screening programme was started; (c) had ratios and the corresponding 95% confidence intervals (95% CI)

315
for death from CRC. For any screening test that leads to early de-
tection, it is less likely that a case patient would have subsequent
tests, because the earlier tests would have had some chance of
detecting the cancer, thus making further screening impossible. An
adjustment for previous screening examinations controls for this
bias. Separate odds ratios were also calculated for gender, age at
diagnosis, and anatomical location of the cancer (colon and
rectum).
and unscreened persons, especially the issue of compara-
bility of individuals regarding their underlying risk of
the occurrence of the cancer in question. Also, the retro-
spective ascertainment of screening history is often inac-
curate, sometimes differentially so between cases and
controls. Notwithstanding this, several aspects seem to
be consistent with true efficacy. Firstly, the sensitivity of
FOBT in patients is probably no better than 50% for can-
cer [21, 22]; the efficacy of a screening test cannot be
higher than its sensitivity, and our data are consistent
with this. Also, the analysis in this study was conducted
without regard to the time that screening occurred, and
this might have obscured a possible benefit of the tests
performed within the relatively short period during
which screening for faecal occult blood is capable of
achieving detection.
The generally higher reduction in mortality in case-
control studies noted above may be due to the fact that in
a case-control study, attenders are compared with non-
attenders, while in randomised clinical trials the compar-
ison is between an entire study group and an entire con-
trol group, so that compliance to testing and contamina-
tion of the control group can affect the final result.
Although randomised controlled trials and case-con-
trol studies of population screening have all shown a sig-
nificant reduction in CRC mortality, the indications for
implementation of a screening programme using FOBT
remain far from clear. In North America, for example,
considerable controversy remains as to what represents
the most cost-effective approach to screening for CRC.
The American Cancer Society recommends annual
FOBT and flexible sigmoidoscopy every 3 to 5 years be-
ginning at the age of 50 years in asymptomatic, average-
Results
Of the 146 cases, 17 (11.6%) had been screened and 129
had not, whereas of the 292 controls, 50 (17.1%) had
been screened and 242 had not. Thus a significantly
smaller proportion of cases than of controls underwent
screening by FOBT during the study period. This was
statistically significant (P<0.01). The odds ratio of hav-
ing been exposed to one or more FOBT during the whole
study period relative to unscreened persons was 0.64
(95% CI 0.34–1.15). Thus, relative to unscreened per-
sons, the risk for fatal CRC was reduced by 36% among
those who had an FOBT.
The OR for developing fatal cancer did not differ sig-
nificantly between the sub-groups according to gender
(0.64 and 0.63 for men and women, respectively), age at
diagnosis (0.68 and 0.60 for individuals aged under
60 years and those over 60, respectively) or anatomical
location (0.62 and 0.69 for cancers of the rectum and
colon, respectively).
Discussion
This study suggests a 36% reduction in CRC mortality risk individuals [23], whereas the American Gastroen-
for individuals screened by FOBT. The magnitude of the terological Association Panel recommends flexible sig-
reduction in this study is considerably greater than that moidoscopy every 5 years and/or yearly FOBT [24].
reported in randomised controlled trials of general popu-
The implementation of a national screening pro-
lation screening, where the combined evidence from a gramme of the general population using FOBT would
meta-analysis review suggests a point estimate reduction add a considerable burden to health service resources in
of 16% [19], although the reduction could be as great as terms of follow-up examinations, such as colonoscopy.
23% or as little as 7%. Our figures also show a slightly The additional workload can be estimated by extrapolat-
greater reduction in mortality than in case-control studies ing the data from the Nottingham study to the United
of general population screening [8, 9, 10, 11, 12], where Kingdom as a whole [25]. Using a biennial screening
reductions in the order of 20–30% have been reported. protocol, and assuming a compliance rate of 55%, it has
There are no published results of any other case-control been estimated that an additional 77,000 colonoscopies
studies of targeted screening using FOBT, but our results would be required each year. Notwithstanding the practi-
are very close to the projected 33% reduction in CRC cal and logistical implications of such a burden, a further
mortality rate that has been calculated for asymptomatic problem is the finite risk of colonic perforation and re-
individuals with a family history of the disease by em- sultant mortality [20]. Calculations of the risks of colo-
ploying annual FOB screening [20].
noscopy in screening patients who have undergone pre-
Our figures must be interpreted with caution, how- vious polypectomy have shown that if the residual risk to
ever. The confidence intervals were wide, and chance, a 50-year-old man is 1 in 40 and the effectiveness of co-
confounder factors and other unmeasured factors may lonoscopy at 3-year intervals is 100% in preventing CRC
explain at least in part the observed association. The death, 283 colonoscopies would be required to prevent
price when randomisation is abandoned is the probable one cancer death, incurring 0.6 colonic perforations and
introduction of factors that may differ between screened 0.04 perforation-related deaths [26]. Hence screening

316
would be 25 times more likely to prevent a cancer death ulation screening and should have a more favourable
than to cause a death. This level of benefit may be ac- risk-benefit ratio. How this is best achieved remains con-
ceptable, but if the residual risk of the screened popula- troversial. The gold standard for screening is endoscopic
tion is only 1 in 80, and colonoscopy is 50% effective in visualisation of the colon. Thus, a reduction of 85% mor-
preventing cancer death (a more likely estimate than tality using colonoscopic screening, compared to 33%
100%), the benefit becomes even less obvious, as screen- using FOBT has been calculated. This is due to the far
ing would be only 6 times more likely to prevent a death higher sensitivity of colonoscopy, and the beneficial early
than to cause one: although 1131 colonoscopies would detection and removal of benign and early malignant co-
be required to prevent one cancer death, there would be lorectal neoplasms. In a previous study by the same
2.3 colonic perforations, 0.17 procedure-related deaths group of screening by FOBT for colorectal neoplasia in a
and a cost of £190,000. Similarly, using a complex math- targeted high-risk population, a sensitivity and specific-
ematical model and certain fundamental assumptions, it ity of 63% and 96%, respectively, and a positive predic-
has been calculated that an annual FOB test would pre- tive value for all neoplasia of 29% were found [18]. This
vent 71 per 10,000 men screened dying from the cancer is still considerably lower than equivalent colonoscopic
that they would have developed had they not been figures, and it remains contentious whether a group of
screened. This, however, translates into a mean increase patients who have been identified as high risk, and who
life expectancy of only 30.6 days for the screened popu- know that fact, should be offered a test whose sensitivity
lation as a whole.
is demonstrably lower than that of colonoscopy, where a
Deeper scrutiny of the results of mass screening pro- reduction in CRC mortality of 85% has been calculated
grammes raises further interesting points. Firstly, reduc- [20]. Immunological tests for FOBT which are highly
tions in CRC mortality achieved by the immense UK and sensitive but lack the specificity required for population
Danish trials were modest, in the order of 15–18%. screening may become valuable if used more widely. A
Translated in absolute terms, these figures indicate that case-control study using an immunochemical haemag-
for 1000 persons invited for FOBT screening once every glutination test in Japan has reported a reduction in mor-
2 years during 10 years, one death due to CRC would be tality from CRC of 60% [10] although the specificity of
avoided [27]. Secondly, in both the UK and Danish tri- this test is not well established at a population level, and
als, interval cancers were more numerous than screen- the test is expensive and more difficult to analyse.
detected CRCs, probably because among screened sub-
When screening an intermediate risk group as in this
jects, more attention was paid to early symptoms, study, the drawbacks associated with colonoscopy be-
prompting earlier diagnosis of malignant lesions. The come more apparent. This procedure is expensive, time
modest reduction in CRC mortality may have therefore consuming and carries a finite risk of colonic perforation
been in part due to better medical attention for those sub- and resultant mortality [20, 26, 28, 29]. In addition, other
jects who were randomised in the screening group. Fi- harmful screening effects include disruption to lifestyle,
nally, it cannot be assumed that the modest gains in CRC the stress and discomfort of testing and further investiga-
mortality can be replicated in other European countries tions, and the anxiety caused by false-positive tests [30].
given the important discrepancies in health care systems
throughout Europe.
Further studies which take these considerations as
well as economic factors into account are needed to
As a screening measure it would therefore seem that compare the overall efficacy of colonoscopic screening
targeted screening may be a more efficient and cost- of a high-risk population with that of a FOBT screening
effective method of detecting early neoplasms than pop- programme.
References
1. Esteve J, Kricker A, Ferlay J, Parkin
DH (1993) Facts and Figures on Can-
cer in the European Community. Lyon,
IARC
2. Jatzko G, Lisborg P, Wette V (1992)
Improving survival rates for patients
with colorectal cancer. Br J Surg 79:
588–591
3. Hardcastle JD, Chamberlain JO,
Robinson MHE, et al (1996) Random-
ised controlled trial of faecal-occult-
blood screening for colorectal cancer.
Lancet 348:1472–1477
4. Kronborg O, Fenger C, Olsen J,
Jorgensen OD, Sondergaard O (1996)
Randomised study of screening for co-
lorectal cancer by faecal occult blood
test. Lancet 348:1467–1471
6. Kewenter J, Brevinge H, Engaras B,
Haglind E, Ahren C (1994) Results of
screening, rescreening, and follow-up
in a prospective randomized study for
detection of colorectal cancer by fecal
occult blood testing. Results for 68:308
subjects. Scand J Gastroenterol 29:
468–473
5. Mandel JS, Bond JH, Church TR, et al
(1993) Reducing mortality from colo-
rectal cancer by screening for fecal
occult blood. N Engl J Med 328:
1365–1371

317
7. Wahrendorf J, Robra B-P, Wiebelt H,
Oberhausen R, Weiland M, Dhom G
(1993) Effectiveness of colorectal can-
cer screening: results from a popula-
tion-based case-control evaluation in
Saarland, Germany. Eur J Cancer Prev
2:221–227
8. Selby JV, Friedman GD, Quesenberry
CP, Weiss NS (1993) Effect of fecal
occult blood testing on mortality from
colorectal cancer. A case-control study.
Ann Intern Med 118:1–6
9. Lazovich D, Weiss NS, Stevens NG,
White E, McKnight B, Wagner EH
(1995) A case-control study to evaluate
efficacy of screening for faecal occult
blood. J Med Screen 2:84–89
10. Saito H, Soma Y, Koeda J, et al (1995)
Reduction in risk of mortality from co-
lorectal cancer by fecal occult blood
screening with immunochemical
hemagglutination test. A case-control
study. Int J Cancer 61:465–469
11. Zappa M, Castiglione G, Grazzini G,
Falini P, Giorgi D, Paci E, Ciatto S
(1997) Effect of faecal occult blood
testing on colorectal mortality: results
of a population-based case-control
study in the district of Florence, Italy.
Int J Cancer 73:208–210
12. Faivre J, Tazi MA, Mrini T El, Lejeune
C, Benhamiche AM, Dassonville F
(1999) Faecal occult blood screening
and reduction of colorectal cancer mor-
tality: a case-control study. Br J Cancer
79:680–683
14. Whynes DK, Neilson AR, Walker AR,
Hardcastle JD (1998) Faecal occult
blood screening for colorectal cancer:
is it cost-effective? Health Econ 7:
21–29
15. King’s Fund Forum (1990) Cancer of
the colon and rectum. Br J Surg 77:
1063–1065
16. Lynch HT, Lynch PM, Albano WA,
Lynch JF (1981) The cancer family
syndrome: a status report. Dis Colon
Rectum 24:311–322
17. Lovett E (1976) Family studies in can-
cer of the colon and rectum Br J Surg
63:13–18
18. Caffarey SM, Broughton CIM, Marks
CG (1993) Faecal occult blood screen-
ing for colorectal neoplasia in a tar-
geted high-risk population. Br J Surg
80:1399–1400
19. Towler B, Irwig L, Glasziou P,
Kewenter J, Weller D, Silagy C (1998)
A systematic review of the effects of
screening for colorectal cancer using
the faecal occult blood test, hemoccult.
BMJ 17:559–565
20. Eddy DM, Nugent FW, Eddy JF, et al
(1987) Screening for colorectal cancer
in a high risk population: results of a
mathematical model. Gastroenterology
92:682–692
21. Ahlquist DA, Wieand HS, Moertel CG,
et al (1993) Accuracy of fecal occult
blood screening for colorectal neopla-
sia. A prospective study using hemoc-
cult and hemoquant tests. JAMA 269:
1262–1267
23. Byers T, Levin B, Rothenberger D,
et al (1997) American Cancer Society
guidelines for screening and surveil-
lance for early detection of colorectal
polyps and cancer: update 1997.
CA Cancer J Clin 47:154–160
24. Lang CA, Ransohoff DF (1994) Fecal
occult blood screening for colorectal
cancer. Is mortality reduced by chance
selection for screening colonoscopy?
JAMA 271:1011–1013
25. Hardcastle JD, Thomas WM,
Chamberlain J, et al (1989) Random-
ised controlled trial of faecal occult
blood screening for colorectal cancer:
the results of the first 107 349 subjects.
Lancet I:1160–1164
26. Ransohoff DF, Lang CA, Kuo HS
(1991) Colonoscopic surveillance after
polypectomy: considerations of cost
effectiveness. Ann Intern Med 114:
177–182
27. Gotzsche P (1997) Screening for colo-
rectal cancer (Letter). Lancet 349:356
28. Macrae FA, Tan KG, Williams CB
(1983) Towards safer colonoscopy: a
report on the complications of 5000 di-
agnostic or therapeutic colonoscopies.
Gut 24:376–383
29. Habr-Gama A, Waye JD (1989) Com-
plications and hazards of gastrointesti-
nal endoscopy. World J Surg
13:193–201
30. Stewart Brown S, Farmer A (1997)
Screening could seriously damage your
health. Decisions to screen must take
account of the social and psychological
costs. BMJ 314:533–534
13. Faivre J, Tazi MA (1998) Should there
be mass screening using faecal occult
blood tests for colorectal cancer? Eur J
Cancer 34:773–776
22. Allison JE, Feldman R, Tekawa IS
(1990) Hemoccult screening in detect-
ing colorectal neoplasm; sensitivity,
specificity, and predictive value. Ann
Intern Med 112:328–333