Operative mortality in colorectal cancer: prospective national study
BMJ 2003; 327 doi: https://doi.org/10.1136/bmj.327.7425.1196 (Published 20 November 2003) Cite this as: BMJ 2003;327:1196All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
The following statement, made by the consultant surgeon who is
"Chair National Bowel Cancer Project", alarms me. "Whereas", the statement
reads, "higher- risk patients may have a better outcome if the cancer is
by-passed or stented" (1). I can just see the registrars using this
statement to justify bypassing resectable cancers and gastroenterologists
using it to justify stenting. Poor patients.
I cannot recall ever having seen a primary colorectal cancer that was
not resectable and in most cases suitable for anastomotic reconstruction.
On occasions, however, tumour deposits may be left behind and
postoperative radiotherapy and adjuvant therapy may be advisable. I have
resected more than one case that had been deemed irrectable by other
surgeons after exploration and a trial dissection or bypass. I have,
however, seen recurrences which I considered to be irresectable or
unsuited for surgery because of the advanced stage of the disease.
My experience in large bowel cancer might not be as large as the
authors of this rapid response but I was once told by someone who had seen
the figures that I had for a time the largest surgical practice in a major
university tertiary referral center in the US where I had restricted my
surgical practice to gastrointestinal diseases.
I cannot recall co-morbidities ever having been grounds for not
operating on a patient with a resectable colo-rectal cancer. It is,
however, possible that chronic lung diseases might be far more prevalent
in the old coal mining communities in Wales and other industrial parts of
the UK than in the communities in which I have worked. [The community in
which I worked in the US was largely supported by the motor industry]. I
have, however, repeatedly stressed the fact that the risk of surgery is
determined by the presence or absence of incomplete resuscitation before,
during or after surgery. In those who are permitted to have an operation
in this state the risk of surgery in those with co-morbidities is indeed
higher (2,3).
The appallingly bad results reported in this paper (4) cannot be used
to justify bypassing and stenting. They should be used to justify
improving the standard of care for this common surgical disease. The
outcomes are even worse in the subgroup treated electively but admitted
through the A&E.
Unless they are able to provide objective evidence of incomplete
resusctation, defined as an abnormally low gastric intramucosal pH, before
surgery to support their conservative recommendations the influential
authors of this rapid response should retract their statement. Failure to
do so could deprive many more patients of definitive treatment without
risk of death or a permanent colostomy.
1. Jeffrey D Stamatakis, Paris P Tekkis, Jan D Poloniecki and Michael
R Thompson The ACPGBI colorectal cancer model – implications and usage
in surgical practice
(Rapid response on 23 December 2003)
2. Fiddian-Green RG. Gut mucosal ischemia during cardiac surgery.
Semin Thorac Cardiovasc Surg. 1990 Oct;2(4):389-99.
3. Poeze M, Takala J, Greve JW, Ramsay G. Pre-operative tonometry is
predictive for mortality and morbidity in high-risk surgical patients.
Intensive Care Med. 2000 Sep;26(9):1272-81
4. Paris P Tekkis, Jan D Poloniecki, Michael R Thompson, and Jeffrey D
Stamatakis
Operative mortality in colorectal cancer: prospective national study
BMJ 2003; 327: 1196-1201
Competing interests:
None declared
Competing interests: No competing interests
We would like to thank Mr Glen (e-BMJ 4th December 2003) for his
valuable comments on the potential contribution of social deprivation in
the ACPGBI colorectal cancer risk model1. We are interested to read the
results from the Monklands Hospital where a combination of cancer stage
and ASA appeared to act as markers for deprivation. This is an interesting
topic which requires further investigation and we aim to evaluate the role
of social deprivation on short-term outcomes from the current year’s data
of more than 10,000 cases.
Mr Cheetham (e-BMJ 23rd November 2003) and Mr Khoury (e-BMJ 14th
December 2003) have rightly questioned the use of the ACPGBI risk model in
the process of informed consent, as they believe that only three risk
factors can be identified preoperatively. We must draw attention to the
fact that, in addition to the three factors, patient’s age, ASA grade and
mode of surgery, the presence of metastatic disease (i.e. Dukes’ D) should
be routinely determined as part of pre-treatment staging, using ultrasound
or CT scanning for both elective and emergency cases. (Please refer to the
ACPGBI guidelines on the management of colorectal cancer[1]). Dukes’ D is
the discriminating stage in the model as this has the main effect on 30-
day postoperative survival. Dukes’ stages A – C, where cancer is confined
to the bowel and lymph glands, has little effect on early postoperative
survival and therefore any effect in the predictive model is minimal.
Having four out of the five factors available, the surgeon and patient can
then utilise the fifth factor (cancer excision) to balance the risk of
major surgery against palliative treatment. On those occasions when the
decision to resect cannot be made before the patient is anaesthetised, the
possibility of non-resection and related risk can be included in consent.
In general, patients with minimal co-morbidity have a survival advantage
if the cancer is removed, whereas higher- risk patients may have a better
outcome if the cancer is by-passed or stented, as suggested by Mr Nash (e-
BMJ 14th November 2003).
Mr Khoury draws attention to the level of completeness of ASA
recording in the study. We have previously carried out a sensitivity
analysis of patients in this study with and without missing ASA values and
found no effect on the precision of the model estimates. This is an
advantage of population studies utilizing large sample sizes. We
anticipate that data completion will improve, as more cases are entered
into the model and clinicians find practical benefit in the routine use of
such data. The use of ASA, both as an independent risk factor and as an
interaction with cancer excision, proved to be critical to the model.
The three levels used in the published hierarchical model were the
patient, the unit and the region. We specifically studied the unit at
level 2, rather than the individual surgeon, to take account of current
concepts of multidisciplinary team care and allow for unit variables such
as critical care bed provision, nursing numbers, anaesthetic services etc.
We agree with Mr Khoury that, for the individual patient, their surgeon’s
personal mortality rate is important, but suggest that this should be
known by the surgeon and their MDT, as an issue of clinical governance
within that Unit, rather than the subject of National Audit. However we
also recognise the need for patients and healthcare providers to have
specific information on risk and post-operative outcomes.
May we draw attention to an online site, www.riskprediction.org.uk,
which includes a number of models for risk-adjusting surgical mortality,
including that described in the recent BMJ[2] publication. The models are
all automated, so that having entered the fields; the clinician can obtain
immediate information on the level of risk. This can be used in the
clinical situation for direct patient care and for audit purposes.
1. Guidelines for the management of Colorectal Cancer (2001). The
Association of Coloproctology of Great Britain and Ireland.
2. Tekkis PP, Poloniecki JD, Thompson MR, Stamatakis JD Operative
mortality in colorectal cancer: prospective national study. BMJ. 2003;
327(7425):1196-201.
Competing interests:
None declared
Competing interests: No competing interests
Editor
The authors conclude (1) that post-operative death can be predicted with
a numerical table derived from the statistical model of the ACPGBI and
consent to surgery can be truly informed. The model was derived from
dataset of 73 selected centres (7,374 patients) over a 12month period,
representing 25% of all new rectal cancer diagnoses in the UK. Only
4,491 patients (60%) had complete data for the 5 risk factors. In 30% of
7,374 patients, ASA grade was missing, a significant flaw since it is used
twice in the multivariate analysis, both independently and its interaction
with cancer resection. Using imputation techniques for missing values of
such scale occurring twice is suspect in excluding bias. Although
operative mortality for 2,216 patients with missing ASA data is similar to
the remaining 5,158 patients, the adjusted odds ratio for each separate
group may alter significantly if data were complete.
It is questionable that the model assists in consenting patients
since half
the variables in predicting patient mortality, are only known after
surgery. Further, all variables are patient related but the surgeon is
probably the most important single factor influencing anastomotic
integrity (5-30%) and mortality (2).
Patients are not just concerned with 30 day mortality, but with safe
discharge from hospital. What matters in patients’ perception, is
“hospital and surgeon effect” which prevail. Cormorbidity and cancer
extent are important but so is the judgement of the surgeon in careful
patient selection avoiding inappropriate surgery and ensuring good life
quality, few complications, low recurrence and long-term survival (3).
Individual regression lines for each surgeon and centre in hierarchical 3
level regression models, are desirable to include patient (level 1),
surgeon (level 2), and hospital (level 3) risk factors. The model is
destined to evolve further as surgeons’ patient numbers increase
sufficiently to be meaningful.
The ACPGBI model is preferable to the P-POSSUM equation (4) as data
is
easily verified and provided ASA data is complete, it is invaluable for
comparative audit. Distinction is necessary between its application for
comparative audit as opposed to informed consent. The former
excludes the surgeon (the variable under investigation), whilst the latter
should include the surgeon as a significant independent variable in
mortality for individual patients. The recently reported (5) four times
higher mortality for a cohort of patients undergoing major surgery in the
UK compared with those in the USA for a given level of risk, would
suggest that the “hospital and surgeon effect” are paramount.
References
(1) Tekkis PP, Poloniecki JD, Thompson MR, Stamatakis JD. Operative
mortality in colorectal cancer:prospective national study. BMJ 2003;327:
1196-9
(2) Fielding LP, Stewart-Brown S, Blesovsky L, Kearney G. Anastomotic
integrity after operations for large-bowel cancer: a multicentre study.
BMJ 1980;281:411-4
(3) Fielding L P, Stewart-Brown S, Dudley HAF. Surgeon-related variables
and the clinical trial. Lancet 1978;2:778-81
(4) Khoury G A. Both components of POSSUM ratio require critical
analysis. BMJ 2003;326:1397
(5) Bennett-Guerrero E, Hyam JA, Shaefi S, Prytherch DR, Sutton GL,
Weaver PC, Mythen MG, Grocott MP, Parides MK. Comparison of P-
POSSUM risk adjusted mortality rates after surgery between patients in
the USA and the UK. Brit J Surg 2003;90:1593-8
Competing interests:
None declared
Competing interests: No competing interests
The authors are to be congratulated on producing a simple model that
will allow comparisons of operative mortality.
It is noted however, that the authors do not include deprivation
catagory in their scoring system. We provide a colorectal service in a
district general hospital setting where 84% of the patients have a
deprivation catagory of 4 or more (Morris R, Carstairs V. Which
deprivation? A comparison of selected deprivation indices. J Public Health
Med 1991;13:318–26), and we were concerned that the model suggested by
Tekkis et al (Paris P Tekkis, Jan D Poloniecki, Michael R Thompson, and
Jeffrey D Stamatakis
Operative mortality in colorectal cancer: prospective national study BMJ
2003; 327: 1196-1201)would not translate to our patient population. The
predicted mortality did not specifically take into account socio-economic
status and no mention of this was made in the work done to produce this
model.
However, when we applied the predicted mortality score to our
patients, whose data was collected prospectively over the last 5 years, we
demonstrated that the model appears to hold fast in a deprived population
(Table below). Cancer stage and ASA seem to provide sufficient
information on deprivation.
Table n= 170 Predicted 30 day mortality all cases= 6.0% Actual 30 day mortality all cases= 6.5%
Competing interests:
None declared
Competing interests: Tablen= 170Predicted 30 day mortality all cases= 6.0%Actual 30 day mortality all cases= 6.5%
One should congratulate Mr. Tekkis et al [1] for their operative risk
predictor for colorectal cancer (CRC), although its use in allowing
informed consent is limited by its retrospective nature. A truly
predictive scoring system may allow a balanced decision whether surgery is
indeed beneficial to some patients, as one of the greatest surgical skills
is deciding who not to operate on. The data not included in the study from
the 499 patients with CRC who avoided surgery may have made a suitable
control group to investigate the comparative risk of operating versus not
operating.
In addition, it was interesting to note that in those patients least
fit (ASA IV & V) there was little survival difference whether the CRC
was resected or not. This point perhaps highlights the role of stenting in
those unfit patients presenting with obstructing CRC. Although not
divulged in the study, if indeed there were patients stented (either prior
to or in place of surgery) a subset analysis may further our knowledge of
the benefit of stenting in such a poor risk group.
1.Tekkis PP, Poloniecki JD, Thompson MR, Stamatakis JD. Operative
mortality in colorectal cancer: prospective national study, BMJ
2003;327:1196-1201.
Competing interests:
None declared
Competing interests: No competing interests
Tekkis et al. (1) report the development of a model to predict
operative mortality following surgery for colorectal cancer based on a
dataset 7374 patients, the data having been collected post-operatively
from a number of centres throughout the United Kingdom.
The authors suggest that this model can be used in the “preoperative
counselling of patients and their carers”. However of the five variables
used in the model, only three (age, ASA score and urgency of surgery) can
be reliably assessed prior to surgery. Of the remaining two variables, one
will only be known with certainty at the conclusion of the operation
(resection or no resection) and the Dukes’ staging may not be available
for some days after surgery. Although preoperative imaging studies may
help in assessing the likelihood of resectability and the presence of
distant metastases, this data was not used in the development of the
model. These flaws suggest that, while the model may useful in comparative
audit between centres or individual surgeons, it is unlikely to be helpful
for counselling patients prior to surgery.
1. Tekkis PP, Poloniecki JD, Thompson MR, Stamatakis JD. Operative
mortality in colorectal cancer: prospective national study, BMJ
2003;327:1196-1201.
Competing interests:
None declared
Competing interests: No competing interests
Critical care provision and operative mortality in colorecectal cancer
Tekkis et al. describe a model for prediction of operative mortality
following surgery for colorectal cancer (1). We were interested to note
that access to intensive care (ITU) facilities was not one of the analysed
risk factors. The Association of Coloproctology of Great Britain and
Ireland (ACPGBI) recommends that access to ITU facilities is necessary in
hospitals undertaking colorectal cancer resections (2), although there is
no evidence base to underlie this recommendation. Admission to an
intensive care unit is associated with substantial morbidity and mortality
(3). In the light of service reorganisation throughout the NHS there will
be many hospitals that may not have access to ‘on-site’ ITU facilities
(4). Thus elucidation of whether admission to an ITU facility is an
independent risk factor predicting increased mortality following
colorectal cancer surgery may have a significant impact on the working
practices of many NHS trusts (and individual surgeons).
In our own NHS trust (consisting of three hospitals on separate
sites), service reconfiguration has resulted in the loss of ITU facilities
from one hospital. A high dependency unit has been maintained in this
hospital. Transfer of patients (either pre- or post-operatively) to a
different hospital within the trust is thus necessary for any patient
requiring ITU support. Elective surgery for patients with colorectal
cancer has continued despite this and has allowed investigation into the
effect of availability of ITU facilities on operative (thirty-day)
mortality following colorectal cancer resection. Interestingly, there was
no difference between the operative mortality in the six month period
immediately prior to loss of ITU facilities (3/29 patients [10.3%]; p-
POSSUM predicted mortality 4.2%) compared to that in the same time period
following the loss of ITU facilities (1/26 patients [3.8%]; p-POSSUM
predicted mortality 3.9%); P=0.36, Mann-Whitney U test). There was no
difference between the two cohorts with respect to age (P=0.77), p-POSSUM
physiological score (P=0.22), ASA grade (P=0.53) or Dukes’ stage (P=0.69;
all Mann-Whitney U test). Lack of ‘on-site’ ITU facilities required the
transfer of a single patient post-operatively, and none pre-operatively.
These data suggest that elective colorectal cancer surgery can be
performed in units without ‘on-site’ ITU facilities with an acceptable
operative mortality rate.
1. Tekkis PP, Poloniecki JD, Thompson MR, Stamatakis JD. Operative
mortality in colorectal cancer: prospective national study. BMJ
2003;327:1196-1201.
2. The Association of Coloproctology of Great Britain and Ireland.
Guidelines for the management of colorectal cancer. London: The
Association of Coloproctology of Great Britain and Ireland, 2001.
3. Vincent JL. Nosocomial infections in adult intensive-care units.
Lancet 2003;361(9374):2068-77.
4. Expert group on critical care services, NHS Executive.
Comprehensive critical care. A review of adult critical care services.
London: Department of Health, 2000.
Competing interests:
None declared
Competing interests: No competing interests