Rectal bleeding and colorectal cancer in general practice: diagnostic studyBMJ 2000; 321 doi: https://doi.org/10.1136/bmj.321.7267.998 (Published 21 October 2000) Cite this as: BMJ 2000;321:998
All rapid responses
Hans Wauters and colleagues report on the diagnostic value of rectal
bleeding in terms of subsequent development of colorectal cancer. We
feel that this study requires clarification for several reasons:
1. They do not report on the prior probability of colorectal cancer
in age specific categories in their population. The diagnostic value of a
symptom such as rectal bleeding and its impact on post-test probability
and subsequent referral threshold is most useful when the prior
probability of disease is known.
2. They fail to mention that less than half of rectal bleeding occurs
in isolation. More often it is associated with other bowel symptoms
which have higher diagnostic value than rectal bleeding alone.
3. The reported positive likelihood ratio of 68.3 "rules in" a
diagnosis of colorectal cancer, irrespective of the pre-test probability
of colorectal cancer. Similarly, the reported specificity of 99.5% has
virtually the same impact of ruling in the target disorder of colorectal
cancer. These findings suggest that any patient attending their general
practitioner with rectal bleeding requires referral and further diagnostic
evaluation. Our own clinical experience and other community based studies
of rectal bleeding suggests that such a high specificity and likelihood
ratio are unlikely and may well be misleading. 
4. We feel that the most likely explanation for their results relates
to general practitioners under-reporting rectal bleeding in the
prospective arm of their study. Wauters et al "chose rectal bleeding as
the reason for visiting a general practitioner" as their inclusion
criterion for their study. They should clarify whether this means that
patients in whom rectal bleeding was not the primary reason for consulting
their general practitioner were excluded. A prospective study in the
Netherlands showed that in patients presenting with rectal bleeding, only
51% stated this as the primary reason for consulting their general
practitioner; 49% consulted for a different reason but rectal blood loss
was subsequently mentioned during the consultation. In another
prospective study it was found that 3% (95% CI 1.4% to 5.8%) of patients
with rectal bleeding subsequently develop rectal cancer, and even in this
study there was an over-representation of patients with "clinically
relevant rectal bleeding". The number of patients with rectal bleeding
who subsequently developed colorectal cancer in Wauters et al's study was
reported as 27 in their table (7%, 95% CI 4.6% to 10%).
In summary it seems likely that there was a systematic bias in
excluding less severe forms of rectal bleeding which may have not been the
primary reason for consulting with a general practitioner. This has
resulted in inflated values for specificity and positive likelihood ratio.
Before their results are incorporated into clinical practice Wauters et al
should clarify their inclusion criteria and provide age specific 2x2
tables so that readers can judge for themselves the diagnostic value of
isolated rectal bleeding in general practice.
Tom Fahey (Senior Lecturer in General Practice)
Alan Montgomery (MRC Training Fellow)
Knut Schroeder (MRC Training Fellow)
We declare no conflict of interest.
1. Wauters, Van Casteren V, Buntinx F. Rectal bleeding and colorectal
cancer in general practice: diagnostic study. British Medical Journal
2. Black ER, Bordley D, Tape TG, Panzer RJ. Diagnostic Strategies
for Common Medical Problems. Philadelphia: American College of Physicians,
3. Douek M, Wickramasinghe M, Clinton JJ. Does isolated rectal
bleeding suggest colorectal cancer? Lancet 2000;354:393
4. Fijten G, Starmans R, Muris J, Schouten H, Blijham G, Knottnerus
JA. Predictive value of signs and symptoms for colorectal cancer in
patients with rectal bleeding in general practice. Family Practice
5. Fijten G, Muris J, Starmans R, Knottnerus JA, Blijham G, Krebber
T. The incidence and outcome of rectal bleeding in general practice.
Family Practice 2000;10:283-287.
Competing interests: No competing interests
The diagnosis of colorectal cancer is an important subject, but the
paper by Wauters et al (1) is flawed on several counts.
To evaluate the value of rectal bleeding in the diagnosis of
colorectal cancer, the authors correctly analysed all patients
retrospectively with a diagnosis of colorectal cancer in 1993-4 to
calculate the sensitivity and prospectively all patients presented with
rectal bleeding in 1993-4 to calculate the positive predictive value.
They then claim to have estimated the specificity and negative predictive
values from these results. However, the retrospective and prospective
parts of the study were on different population of patients and that the
data collected did not give a measure of the relative size of the
population who neither had colorectal cancer nor presented with rectal
bleeding. Hence, it was not possible to make valid estimates of the
specificity and negative predictive values. For the same reasons, valid
estimates of the likelihood ratios cannot be made.
The authors stated that the probability of colorectal cancer
increases greatly in association with fatigue and weight loss. However,
there must be an error in the figure 7.1% (8.3% to 15.8%) for the positive
predictive value associated with fatigue, as the mean is outside the 95%
confidence intervals. The positive predictive value associated with
weight loss of 16.0% (4.5% to 36.1%) is not statistically different from
the overall positive predictive value of 7.0%.
The authors argued that “people should be better informed and
encouraged to seek medical advice if bleeding occurs”. Whilst this advice
might be appropriate, this conclusion was not justified by their results,
which did not show in any way that any of their patients failed to seek
appropriate medical advice.
There are two further typographical errors. First, “positive
prospective values” in the sixth paragraph of the “subjects, methods and
results” section should read “positive predictive values”. Second, the
reference to the table in the seventh paragraph in the “subjects, methods
and results” section after “106 patients had colorectal cancer” is
1. Wauters H, Van Casteren V, Buntinx F. Rectal bleeding and
colorectal cancer in general practice: diagnostic study. BMJ 2000; 321:
Competing interests: No competing interests