Postgrad Med J (1994) 70, 469 - 474
Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
Review Article
Screening for colorectal cancer
D.H. Bennett and J.D. Hardcastle
Department ofSurgery, E Floor, West Block, University Hospital, Queen's Medical Centre, Nottingham
NG72UH, UK
Introduction
Colorectal carcinoma is the third commonest cause
of death from malignant disease in England and
Wales - approximately 25,000 deaths in 1992. The
incidence of the disease appears to be increasing
while there has been a slow decline in mortality
since 1985.' The result has been a relatively con-
stant annual mortality from colorectal cancer
throughout the Western world over the last 30
years. The aim of colorectal cancer screening is to
reduce this mortality by both preventing its
development and by detecting its presence at a
stage when curative resection is possible.
Screening for colorectal cancer is justified on the
premise that prognosis is mainly related to the
extent of tumour spread at the time of diagnosis.2'3
and diagnosis when the tumour is at a less
advanced stage may offer the best hope of reducing
the mortality. For a screening programme to be
effective, the disease in question must be common,
the epidemiology and natural history of the condi-
tion known and a screening test be available which
detects the condition while presymptomatic.
Similarly, the screening test should be simple,
acceptable to the patient and clinician, sensitive,
specific, safe and, preferably, cheap. Finally, the
investigation and treatment of positive test results
should be effective, readily available and cost-
effective.
Around 1970 several procedures were intro-
duced which changed the management ofcolorectal
cancer - colonoscopy, colonoscopic polypectomy
and modern techniques of cancer surgery4 - enab-
ling screening for colorectal cancer to be under-
taken using simple and inexpensive faecal occult
blood (FOB) testing.5
Epidemiology and natural history
Colorectal cancer is a major problem in most
developed countries. During the period 1980-1984
Correspondence: D.H. Bennett, F.R.C.S.
Received: 17 January 1994
A) The Fellowship of Postgraduate Medicine, 1994
recorded mortality from all member states show
that the average annual number of colorectal
deaths in European Community residents was over
85,000.6 Similarly, in the USA, there are approx-
imately 60,000 deaths and 152,000 new cases
reported annually.7 In the UK it is the second
commonest cause of death from malignant disease
coming after lung cancer in men and women.8
As mentioned above, survival rates have im-
proved slightly over recent decades, 5-year relative
survival of all registered cases diagnosed in 1971
being 30%,9 and of all cases registered in
copyright.
1979- 1981 being 35%.'1 It has been calculated that
a 50 year old person has a 5% risk of having
colorectal cancer by the age of 80 and a 2.5% risk of
dying from it."
The natural history of the disease is continuing
to be elucidated as the genetic changes underlying
the development of malignancy are discovered. It is
believed that the majority of colorectal cancers
arise from adenomas, evidence supporting this
argument including firstly, the demonstration that
polyps sometimes contain sites of adenocar-
cinoma12 and, secondly, the incidence of both
adenomas and carcinomas increase in parallel with
age. Cancer of the large bowel is a disease of older
people (except in the rare familial form), the risk
doubling with every decade over 40 years.'3 The
incidence of adenomas also increases with age.
Adenomas are common only in countries with a
high incidence of colorectal cancer.'4
However, only a minority of adenomas progress
to cancer formation. Necropsy studies of asymp-
tomatic patients show that the prevalence of
adenomas in Britain is 34% in 50-60 years olds,
rising to 40-60% in the over-75s. However, the
prevalence of cancers in age-matched asymp-
tomatic patients is only 1.6% and 3%.'5
Is it possible to predict which adenomas will
progress? In a retrospective study, Stryker et al.
suggested a cumulative risk of polyp progression at
5 years, 10 years and 20 years of 2.5%, 8% and
24%, respectively.'6 There appear to be three
indices which relate to possible progression -
470 D.H. BENNETT & J.D. HARDCASTLE

histological characteristics, size and the degree of
dysplasia. The villous adenoma is the most likely to
transform, along with polyps greater than 2 cm in
diameter and those exhibiting severe dysplasia.
Eide calculated the annual risk of an adenoma
converting to a carcinoma to be 0.25% for all
adenomas, 3% for adenomas greater than 1 cm,
17% for villous adenomas and 37% for villous
adenomas with severe dysplasia.'7
The recent advances in genetic research provide
perhaps the strongest evidence supporting the
adenoma - carcinoma progression. The discovery
of the gene for familial adenomatous polyposis,
known as APC, has provided direct evidence for a
genetic cause for the development of polyps.
Similarly, somatic mutations in the APC gene have
been demonstrated in colorectal cancer tissue in
patients with no known heritable syndrome.'8"9
Further support comes from identification ofki-ras
oncogene mutations which occur in premalignant
adenomas as well as carcinomas and a series of
genetic mutations have been suggested as necessary
before malignant invasion occurs.20
In summary, colorectal cancer is a common
condition with a significant burden of illness and is
an important cause of mortality. Its natural history
suggests that detection at an early stage or in its
precancerous stage should improve outcome,
Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
effectiveness of screening by FOBTs. The com-
monest type of FOBT employed is the Haemoccult
test, a slide preparation of guaiac gum which relies
on the peroxidase activity of haematin to catalyse
the phenolic oxidation of the substrate. Based on
unrehydrated tests performed 2-yearly, the sen-
sitivity of this test in the various trials has ranged
from 48% to 75%,2325 while the specificity can be
as high as 99%.23
I
Controlled trials - results (Table I)
The first trial to report4 mortality data was a
non-randomized study which systematically as-
signed participants to screening or a control group.
This study showed a 43% reduction in mortality in
the screened group after 10 years of follow-up. This
reduction, however, failed to reach statistical
significance due to an inadequate number of deaths
from colorectal cancer.
The three European randomized population
trials23'24'26 have all reported interim results. The
Swedish trial randomly divided 27,700 participants
aged 60-64 years into test and control groups, and
rescreened the test group after a mean interval of 20
months. The compliance in the study was 66% at
the first screen and 58% at the second. Investiga-

meeting the criteria to make screening effective.
Three principle tests have been used to screen for
colorectal cancer - digital rectal examination,
faecal occult blood tests (FOBT) and sigmoid-
oscopy. While digital examination is an easy test, it
is of minimal value for screening since the majority
of tumours are out of reach of the examining
finger2' and soft polypoid tumours may be difficult
to feel.
Since Greegor's proposal for the use of a home
test involving guaiac-impregnated paper slides,22
five large trials have been set up to evaluate the
copyright.
tion of 322 subjects with a positive test result
revealed 16 cancers and 58 individuals with one or
more adenomas at the first screen. In the interval
between screens, 26 carcinomas presented symp-
tomatically in the screened group and 16 in the
control group. The rescreen revealed 19 car-
cinomas and 92 adenomas in the test group com-
pared with two carcinomas and one adenoma in the
control group. There was no statistical difference
between the Dukes' staging of the carcinomas,
46% of the screened group having stage A or B
tumours compared with 40% of the control group.

Table I Randomized controlled trials of Haemoccult testing in screening for colorectal cancer
Nottingham, Goteborg, Fuhnen, Burgundy, New York, Minnesota,
England Sweden Denmark France USA* USA
Cohort size 155,000 52,000 62,000 94,000 22,000 48,000
Positivity rate (%) 2.1 1.9t 1.0 2.1 1.7 2.4t
5.81 9.8t
Compliance 54% 65% 67% 52% 75%§ 90%§
Predictive value (% 50 22 57 31 30 31
neoplasia)
Dukes' A cancers (%)
Screened group 52 50 51 52 43 34
Test group 30 30 27 n/a 35 n/a
Control group 13 12 9 n/a 27 35
*Trial included rigid sigmoidoscopy in both arms of trial; tnon-rehydrated Haemoccult slides; trehydrated
Haemoccult slides; §subjects recruited were volunteers rather than identified from Practice lists and then randomized.

In the light of these inconclusive results, the trial
has been extended to 51,477 participants and
further results are awaited.
The Danish study randomly allocated 61,938
participants aged 50-74 to the screening and
control groups. The rescreening interval was
approximately 2 years. Sixty-seven per cent of
those invited completed the first screen and 93% of
those initially screened completed the second test.
'here were 215 positive test results at the first
screen with 37 carcinomas and 86 adenomas
detected compared with 159 positive tests at the
rescreen with a further 13 carcinomas and 76
adenomas detected. During the follow-up, 40 inter-
val cases and 39 non-responders presented with a
carcinoma compared with a total of 115 car-
cinomas in the control group. The screened group
showed a statistically significant earlier Dukes'
stage than the controls (59% stage A or B in the
screened group compared with 45% in the control
group). The estimated death rate from colorectal
cancer was 1.2 per 1,000 persons in the screening
group versus 1.6 per 1,000 persons among controls,
a trend which does not reach statistical significance
(P= 0.16).
The largest of the randomized studies is taking
place in Nottingham where 155,034 persons aged
50-74 years have been randomly allocated to test
and control groups. Rescreening has been carried
out at 2-yearly intervals. The average initial com-
pliance 54%. At the initial screen 842 of the tests
were positive and further investigation detected 91
cancers and 305 persons with adenomas. At the
first rescreen, the compliance was 78% and
revealed 54 carcinomas, and 139 patients with
adenomas while the second rescreen (compliance
88%) revealed a further 25 carcinomas and 77
people with adenomas. There was a significantly
greater number of screened people with less
advanced tumours than among the controls (56%
Dukes' A or B compared with 46%) and this was
also found to be the case on rescreening (62%
Dukes' A and B). This difference in the proportion
of early Dukes' stage carcinomas between the
screen-detected and control groups accords with
the findings of the Danish trial. The Nottingham
study also reported fewer emergency procedures,
unresectable tumours and fixed tumours in the
screened group.
The Minnesota Colon Cancer Control Study,27
has recently reported the first statistically sig-
nificant reduction in mortality from colorectal
cancer by FOBT screening. In a prospective ran-
domized trial, 46,551 participants were assigned to
screening once a year, every 2 years or to a control
group. Compliance within the study was very good
with 75.2% of annual and 78.4% of biennial
screenings completed. There were 1,002 cases of
colorectal carcinoma detected over the 13-year
SCREENING FOR COLORECTAL CANCER 471
Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
period of follow-up and 320 deaths attributed to
this cause, with a cumulative annual mortality rate
of 5.88 per 1,000 in the annually screened group
and 8.33 per 1,000 in the biennially screened group
compared with the control group rate of 8.83 per
1,000. These results show a 33% decline in mor-
tality in the annually screened group compared
with the control group. There is also a decline in
mortality in the biennially screened group com-
pared to the control but this does not reach
statistical significance.
While this study appears to vindicate supporters
of colorectal screening programmes, there are
several important points to consider. Firstly, the
screening programme was interrupted for 3 years
and then recommenced when, at that stage, the
power of the study was too low to show a
statistically significant result. There was both a
higher incidence and higher mortality from col-
orectal cancer in the biennially screened group
early in the study which Mandel and colleagues
admit could have resulted from a chance imbalance
in the randomization procedure. This was followed
by the 3-year hiatus in screening and therefore it is
impossible to know whether chance influenced the
biennially screened group at this stage.
Secondly, and more importantly from the
screening point of view, this study utilized rehy-
copyright.
drated Haemoccult slides which resulted in a better
sensitivity (92.2%) than non-rehydrated slides but
a much lower specificity (90.4%). The practical
effect of using this technique was a much higher
rate of colonoscopy - 38% of the annually and
28% of the biennially screened participants under-
went at least one colonoscopy. It could be argued
that with over one-third of participants undergoing
at least one colonoscopy in the 13-year period, the
study demonstrates the value of screening by
colonoscopy!
A third comment relates to the positive predic-
tive value for the detection of cancer. In the
Minnesota study, the positive predictive value is
2.2% compared with 17% in the Danish study,
4.6% in the Swedish study (who also used rehy-
drated Haemoccult), 10% in the Nottingham study
and 7.5% in the New York study. Thus it can be
seen that rehydrating Haemoccult slides signi-
ficantly reduces the positive predictive value for the
test with a consequent increase in the cost of
diagnostic investigation.
There are other problems with Haemoccult as a
screening test. Red meat and peroxidase-con-
taining vegetables such as tomatoes and bananas
may give false-positive results therefore dietary
restriction for 3 days prior to the test is sometimes
recommended.
When the Nottingham study introduced dietary
restriction the rate of positive results fell from 3.4%
to 1.3%.
472 D.H. BENNETT & J.D. HARDCASTLE

Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
Newer faecal occult blood tests subjects with 1 .1% Haemoccult positive and 9.7%
HemeSelect positive. Nine cancers were detected by
Since the original Haemoccult test was developed, HemeSelect but only one of these was Haemoccult
alternative FOB tests have been produced such as positive. Similarly, 49 patients with adenomas were
more sensitive peroxidase-based tests, immuno- identified, 48 were HemeSelect positive but only
chemical tests specific for haemoglobin and eight were Haemoccult positive. While this study
heme-porphyrin assays. The more sensitive confirms the greater sensitivity of HemeSelect, its
peroxidase-based tests improve the detection of lower specificity results in a greater number of
neoplasia but have a lower specificity.28'3' The investigations with the concomitant increase in the
heme-porphyrin assay, HemoQuant, detects cost of screening and risk of complications. Further
blood loss throughout the gastrointestinal tract data are required to prove the benefits of changing
and, although provisional results were en- to a newer faecal occult blood test before the use of
couraging,32 subsequent studies have been disap- either HaemoccultSENSA or HemeSelect can be
pointing.3334 The immunological tests have the supported.
theoretical advantage that they should only detect To summarize, screening by FOBTs leads to the
blood lost into the large bowel, since any blood lost detection of less advanced Dukes' stage tumours
into the upper tract will have lost the immuno- with better prognostic features. One randomized
logically reactive site recognized by the antibody by trial has shown a significant reduction in mortality
the time the blood reaches the large bowel by and it remains to be seen whether the European
digestion. trials confirm these findings. Newer FOBTs have a
In the most recent work to date, St John et al. higher sensitivity for the detection of both car-
compared Haemoccult with a newer peroxidase- cinomas and adenomas but their lower specificities
based test, HemoccultSENSA, HemeSelect (an have important implications for the potential cost
immunological test) and HemoQuant both in of public screening programmes.
symptomatic patients and in participants of a
colorectal screening programme. The sensitivities
for these tests in patients with known colorectal Sigmoidoscopy as a screening technique
malignancies were 88.8%, 93.5%, 97.2% and

copyright.
71.0%, respectively. Their results demonstrate Sigmoidoscopy has not been popular as a screening
significantly better positivity rates with the Hemo- technique due to the medical input required and the
ccultSENSA and HemeSelect tests than with inconvenience and discomfort to the patient. A
Haemoccult (their specificities were comparable). uniphase screening programme was initiated by the
Due to the expense of immunological testing, it is Memorial Sloan-Kettering Hospital who reported
suggested that a two-tier testing strategy be their results in 1960.37 Cancers were diagnosed in 58
adopted, with slides being prepared for both patients out of 26,126 individuals who underwent
HemoccultSENSA and HemeSelect, but the sigmoidoscopy, a detection rate at the initial screen
HemeSelect slide only being developed if the of 0.002. The majority of these tumours were at an
HemoccultSENSA test was positive. Evaluation of early stage and 5-year patient survival was 88%.
this proposal showed that it would have detected More recently, the Kaiser Permanente Multi-
the single carcinoma in the screened participants phasic Evaluation Study reported its results for
but would have missed seven out of 13 large screening sigmoidoscopy.38 Members of the med-
adenomas. ical care programme were randomized to a group
Another interesting feature of this study was the who were actively encouraged to schedule annual
failure to demonstrate any difference in the ability health checkups or a control group who were not so
of Haemoccult to detect left- and right-sided encouraged. The study group subjects had both a
lesions. The sensitivity of Haemoccult for detecting lower cumulative incidence (4.3 versus 6.7 cases per
carcinomas proximal to the splenic flexure was 1,000 persons) and a better stage distribution (86
89.3% compared with 88.6% for lesions at or distal versus 54% Stage B or better) than non-
to the splenic flexure. This contrasts with the encouraged controls for colorectal cancers arising
Nottingham data where 81% of tumours in the within reach of the sigmoidoscope. However, there
descending and sigmoid colon were detected but was only a small difference between the two groups
Haemoccult only had a sensitivity of 45% for rectal in their exposure to sigmoidoscopy (30% in the
carcinomas, and 47% for carcinomas of the screened group and 25% in the control group) and
caecum or ascending colon.35 it was concluded that the difference in mortality
Further support for an immunological FOB test was too great to be accounted for purely by
comes from a study performed in Brighton and screening sigmoidoscopy.
Guildford36 where Haemoccult and HemeSelect In a retrospective case-control study,39 previous
were directly compared in an asymptomatic exposure to sigmoidoscopic screening was com-
population. Both tests were completed by 1,489 pared in subjects who had died ofrectal cancer with

a general population sample. Eight point eight per
cent of the case group had undergone sigmoid-
oscopy in the preceding 10 years compared with
24.2% of the control group, suggesting screening
by rigid sigmoidoscopy reduces the risk of develop-
ing colorectal cancer within the next 10 years by
over two-thirds. The design of this study has been
criticized due to selection bias in the case group but
the study contained an internal control, tumours
above the reach of the sigmoidoscope occurring
with equal frequency in the two groups.
The development of flexible sigmoidoscopy has
focused attention on this technique as a possible
solution to the poor sensitivity of Haemoccult and
the limited examination possible with a rigid endo-
scope. Data from the Nottingham and Danish
studies suggest that over 70% of screen-detected
tumours are within reach of the flexible sigmoido-
scope and it has been suggested that combining
flexible sigmoidoscopy with a FOBT would result
in a more realistic sensitivity for the detection of
colorectal carcinoma for a screening programme.
References
1. Levin, B. Colorectal cancer screening. Cancer 1993, 72:
1056-1060.
2. Slaney, G. Results of treatment of carcinoma of colon and
rectum. Mod Trends Surg 1971, 3: 69-89.
3. Stower, M.J. & Hardcastle, J.D. The results of 11 15 patients
with colorectal cancer treated over an 8-year period in a single
hospital. Eur J Surg Oncol 1985, 11: 119-123.
4. Winawer, S.J. & Schottenfeld, B.J. Colorectal cancer screen-
ing. J Natl Cancer Inst 1991, 83: 243-253.
5. Greegor, D.H. Diagnosis of large-bowel cancer in the asymp-
tomatic patient. JAMA 1967, 201, 943-945.
6. Chamberlain, J. Personal communication.
7. Ransohoff, M.D. & Lang, C.A. Screening for colorectal
cancer. N Engl J Med 1991, 325: 37-41.
8. Office of Population Censuses and Surveys. Mortality Statis-
tics, Cancer Series DH2, No. 14. HMSO, London, 1989.
9. Office of Population Censuses and Surveys. Cancer Statistics,
Studies on Medical and Population Subjects. No. 43. HMSO,
London, 1981.
10. Office of Population Censuses and Surveys. Cancer Survival,
1979-1981 Registrations. Series MBI, No. 16. HMSO,
London, 1986.
11. Boring, C.C., Squires, T.S. & Tong, T. Cancer Statistics,
1993. CA Cancer J Clin 1993, 43: 7-26.
12. Morson, B.C. Evolution of cancer of the colon and rectum.
Cancer 1974, 34: 845-849.
13. Cancer Research Campaign. Facts on Cancer. Mortality in the
UK. Cancer Research Campaign, London, 1988.
14. Burkitt, D.P. Epidemiology of cancer of the colon and
rectum. Cancer 1974, 28: 3-13.
15. Pollock, A.M. & Quirke, P. Adenoma screening and colorec-
tal cancer. Br Med J 1991, 303: 3-4.
16. Stryker, S.J., Wolff, B.G., Culp, C.E. et al. Natural history of
untreated colonic polyps. Gastroenterology 1987, 93:
1009-1013.
17. Eide, T.J. Risk of colorectal cancer in adenoma bearing
individuals within a defined population. Int J Cancer 1986,
38: 173-176.
18. Nishisho, I., Nakamura, Y., Miyoshi, Y. et al. Mutations of
chromosome 5q2 1 genes in FAP and colorectal cancer
patients. Science 1991, 253: 665-669.
SCREENING FOR COLORECTAL CANCER 473
Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
In the Nottingham study, 15 of 22 interval cases
presenting within 2 years of screening were in the
rectum and sigmoid colon; if all these cases had
been detected by sigmoidoscopy at the time of
screening, the sensitivity of combined FOBT and
sigmoidoscopic screening would have been 93% - a
gain of 18% over the sensitivity of FOBT alone.
Again, there are some unanswered questions.
How many of the interval cancers arise de novo in
the interval between screening and are not actually
detectable endoscopically at the time of FOBT
screening? While the specificity of sigmoidoscopy
should be 100%, can it be assumed that the
sensitivity will also approximate 100%? What
effect will the addition of sigmoidoscopy have on
compliance rates? In an attempt to answer some of
these questions, a European Multicentre Ran-
domised Control Trial has been initiated to com-
pare colorectal screening by FOBT with screening
by combined FOBT and flexible sigmoidoscopy -
the results are eagerly awaited.
19. Kinzler, K.W., Nilbert, M.C., Vogelstein, B. et al.
Indentification of a gene located at chromosome 5q2 1 that is
copyright.
mutated in colorectal cancers. Science 1991, 251: 1366-1370.
20. Vogelstein, B., Fearon, E.R., Hamilton, S.R. et al. Genetic
alterations during colorectal-tumour development. N Engl J
Med 1988, 319: 525-532.
21. Winawer, S.J., Prorok, P., Macrae, F. & Bralow, S.P.
Surveillance and early diagnosis of colorectal cancer. Cancer
Detect Prev 1985, 8: 373.
22. Simon, J.B. Occult blood screening for colorectal cancer: a
critical review. Gastroenterology 1985, 88: 820-837.
23. Hardcastle, J.D., Thomas, W.M., Chamberlain, J. et al.
Randomised, controlled trial of faecal occult blood screening
for colorectal cancer: results for first 107,349 subjects. Lancet
1989, 1: 1160-1164.
24. Kronborg, O., Fenger, C., Olsen, J., Bech, K. & Sondergaard,
0. Repeated screening for colorectal cancer with faecal occult
blood tests: a prospective randomised study at Funen,
Denmark. Scand J Gastroenterol 1989, 24: 599-606.
25. Allison, J.E., Feldman, R. & Tekawa, I.S. Haemoccult
screening in detecting colorectal neoplasm: sensitivity,
specificity and predictive value. Long term follow-up in a
large group practice setting. Ann Intern Med 1990, 112:
328-333.
26. Kewenter, J., Bjork, S., Haglind, E., Smith, L. et al. Screening
and rescreening for colorectal cancer. Cancer 1988, 62:
645-651.
27. Mandel, J.S., Bond, J.H., Church, T.R., Snover, D.C. et al.
Reducing mortality from colorectal cancer by screening for
faecal occult blood. N Engl J Med 1993, 328: 1365-1371.
28. Adlercreutz, H., Liewendahl, K. & Virkola, P. Evaluation of
Fecatest, a new guaiac test for occult blood in faeces. Clin
Chem 1978, 24: 756-761.
29. Frommer, D. & Logue, T. Comparison of five guaiac resin
paper tests for demonstrating the presence of blood in faeces.
Aust NZ J Med 1981, 11: 494-496.
30. Feneyrou, B., Bories, P., Pomier-Layrargues, G., Michel, H.
& Gravagne, G. Discrepancy in results from three guaiacum
resin tests. Br Med J 1982, 284: 235-236.
474 D.H. BENNETT & J.D. HARDCASTLE
31. Pye, G., Jackson, J., Thomas, W.M. & Hardcastle, J.D.
Comparison of Coloscreen Self-test and Haemoccult faecal
occult blood tests in the detection of colorectal cancer in
symptomatic patients. Br J Surg 1990, 77: 630-631.
32. Ahlquist, D.A., McGill, D.B., Schwartz, S., Coles, M.E.,
Hubert, T.W. & Thomas, D.W. Fecal blood levels in health
and disease: a study using HemoQuant. N Engl J Med 1985,
312: 1422-1428.
33. St John, D.J.B., Young, G.P., McHutchinson, J.G., Deacon,
M.C. & Alexeyeff, M.A. Comparison of the specificity and
sensitivity of Haemoccult and HemoQuant in screening for
colorectal neoplasia. Ann Intern Med 1992, 117: 376-382.
34. Ahlquist, D.A., McGill, D.B., Fleming, J.L. et al. Patterns of
occult bleeding in asymptomatic colorectal cancer. Cancer
1989, 63: 1826-1830.
35. Thomas, W.M., Pye, G., Hardcastle, J.D. & Walker, A.R.
Screening for colorectal carcinoma: an analysis of the
sensitivity of Haemoccult. Br J Surg 1992, 79 833-835.
Postgrad Med J: first published as 10.1136/pgmj.70.825.469 on 1 July 1994. Downloaded from http://pmj.bmj.com/ on February 10, 2020 by guest. Protected by
36. Robinson, M.H.E., Marks, C.G., Farrands, P.A., Thomas,
W.M. & Hardcastle, J.D. Population screening for colorectal
cancer: comparison between a guaiac and immunological
faecal occult blood test. Br J Surg 1994, 81: 448-451.
37. Hertz, R.E.L., Deddish, M.R. & Day, E. Value of periodic
examinations in detecting cancer of the colon and rectum.
Postgrad Med 1960, 27: 290-294.
38. Selby, J.V., Freidman, G.D. & Colen, M.F. Sigmoidoscopy
and mortality from colorectal cancer: the Kaiser Permanente
Multiphasic Evaluation Study. J Clin Epidemiol 1988, 41:
427-434.
39. Selby, J.V., Freidman, G.D., Quesenberry, C.P. & Weiss,
N.S. A case-control study of screening sigmoidoscopy and
mortality from colorectal cancer. N Engl J Med 1992, 326:
653-657.
copyright.