Influence of hospital and clinician workload on survival from colorectal cancer: cohort studyCommentary: How experienced should a colorectal surgeon be?
BMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7195.1381 (Published 22 May 1999) Cite this as: BMJ 1999;318:1381All 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 article by Kee and colleagues describing the influence of
hospital and clinician workload on survival from colorectal cancer and the
following commentary by Robert Shields (BMJ Volume 318, 22 May 1999, pages
1381-1386) invites further debate on the relationship between volume and
outcome in this important area of both work and policy development. The
Wessex Colorectal Cancer audit has data on 5,177 cases of colorectal
cancer incident within the resident population of Wessex, between 1991
and 1994. These cases are being followed up for 5 years and we have
analysed the 2 year survival in relation to both the number of cases per
surgeon annually and the number of cases seen in the hospital annually.
In order to adjust for important case-mix variables, we selected only
those cases where we had complete information on stage, age, whether the
surgery was elective or emergency, curative or palliative, site of lesion,
sex and number of co-morbid health problems. Thus a total of 3,396 cases
were included. We divided surgeon caseload in two different ways.
Firstly, by ranking surgeons according to the annual number of cases for
which they were responsible, and then sub-dividing the distribution into
fifths, as Kee et al did; and secondly dividing the distribution into
<=10 cases pa, <=20 cases pa, <=30 cases pa, etc. We had only 11
major hospitals, so hospital caseload was divided into three groups.
Kaplan Meier crude 2 year survival was carried out for each of the
different caseload groups. Case-mix was then adjusted for using Cox
Proportional Hazards model. For each of the surgeon caseload groups, a
log rank test confirmed that there was a statistically significant
difference between the groups (p=.0000) with lowest survival in the lowest
caseload group and highest survival in the highest caseload group. After
adjustment for case-mix, this evidence of surgeon caseload heterogeneity
persisted, (p=.0005) and (p=.001) respectively for the two different
methods of sub-division, although it was not strictly linear. Factors
which proved to be significant as well as volume were age, whether
surgery was elective or emergency, curative or palliative, stage, and
number of co-morbid health problems. Hospital caseload had the opposite
effect to that expected with the highest volume hospital having the lowest
survival, even after adjustment for case-mix. The same additional factors
were significant in hospital volume.
We confirm that there is a relationship between surgeon volume and
outcome, though this is not completely linear when adjustment is made for
casemix. We cannot find an explicable relationship between volume of
hospital and outcome, though in reality there is very little difference in
the workload of the hospitals and all are high workload hospitals in
comparison to those referred to by Kee, the lowest volume hospital having
69 cases per annum.
When analysing this data previously we had noted that there were
differences in the actual operations performed by low and high volume
operators as well as the proportion of subsites seen by these groups.
Furthermore, the high volume operators were a mixture of specialist and
non-specialist coloproctologists. Hence we have not concentrated on volume
per se, but on the outcome by specialist coloproctologists. Since analysis
by volume alone does not appear to offer insight into why these
differences should exist, we would suggest that the debate should widen
and we intend publishing data to confirm the advantage of specialisation
in the near future.
Authors
Jenifer A E Smith MRCP FFPHM
Director
South and West Cancer Intelligence Unit
Highcroft
Romsey Road
Winchester S022 5DH
Robert H S Lane MS FRCS
Consultant Surgeon
Royal Hampshire County Hospital
Romsey Road
Winchester S022 5DG
Michael R Thompson MD FRCS
Consultant Surgeon
Queen Alexandra Hospital
Southwick Hill Road
Cosham
Portsmouth
Hampshire PO6 3LY
Philippa M King BSc
Statistician
South and West Cancer Intelligence Unit
Highcroft
Romsey Road
Winchester S022 5DH
1 Kee F, Wilson R H, Harper C,Patterson CC, McCallion K, Houston R
F, et al. Influence of hospital and clinician workload on survival from
colorectal cancer: cohort study. BMJ 1999; 7195: 1381-1385
Competing interests: No competing interests
Editor-
Kee et. al. have identified important issues regarding outcomes in
colorectal surgery. However, we feel the study overlooks several relevant
factors.
Patient details such as adjuvant treatment and concurrent illness,
which may influence prognosis, have not been considered. In addition, we
question whether a 2-year follow-up is sufficient to assess outcome post-
surgery. Although the study identififed years of practice as a reflection
of surgical experience, the number and type of operations undertaken may
have been a more appropriate indicator.
We agree that using hospital workload as a measure of quality of
cancer care may be insufficient. As well as looking at organisation of
hospital services, differences in surgical protocol between individual
hospitals and surgeons could also be considered.
The classification of the study as a cohort may be inaccurate. All
patients were subject to surgery and there was no group without this
exposure. Perhaps the study is more accurately described as a case series.
Lastly, we agree that it may be unwise to make survival data freely
available in the form used in the study. In a different context,
accounting for all other contributing factors, this information may be of
value to a wider audience.
Competing interests: No competing interests
Mesorectal excision plane status is an important determinant of surgical success in rectal carcinoma
Mesorectal excision plane status is an important determinant of
surgical success in rectal carcinoma
We read with interest the cohort study from Kee et al which
investigated the effect of case workload on patient outcome in colorectal
cancer.1 This study of over 3000 patients found that the number of
operations performed by individual surgeons did not alter patient
mortality from colorectal cancer at two years while patients treated in
hospitals dealing with more than 33 such cases annually possessed a
significantly increased odds ratio for mortality over the same follow-up
period. The authors found no obvious explanation for this apparent
paradoxical difference in outcome.
In this large study, comparisons between both surgeons and hospitals
were made using tumour grade (ie differentiation) and Dukes' staging as
pathological determinants. However, Dukes' staging does not take into
account the proximity of tumours to the mesorectal excision margin or the
peritoneal surface of the bowel. Successful treatment, particularly in
carcinoma of the rectum, is dependent upon minimisation of the risk of
local recurrence and this is ideally achieved via complete local removal
of the tumour. The critical importance of complete local tumour excision
has recently been highlighted.2 Assessment of the mesorectal excision
margin has become an essential component of the histopathological
assessment of rectal excision specimens. Indeed, in addition to
measurement of the distance of the tumour from the mesorectal excision
margin, the current MRC-CRO7 trial encourages histopathologists
macroscopically to assess the surgical quality of mesorectal excision in
such cases.3 Tumour involvement of the peritoneal surface of the rectum is
also known to impart a poor prognosis.4
We would be interested to establish whether these additional
variables were taken into account during the comparative analyses in Kee
et al's study. This information is of particular importance since surgeons
may require specific training in mesorectal excision techniques. Surgeons
experienced in this aspect of colorectal surgery may be expected to
achieve higher rates of complete local tumour excision. Patients whose
tumours abut or closely approach the mesorectal excision margin may be
offered local radiotherapy in order to reduce the risk of local recurrence
and the frequency of such adjuvant therapy would also need to be taken
into account during the analysis. Use of the TNM staging system in future
studies may aid inclusion of these factors since the T4 category exists
for primary tumours extending to one or both of these margins of
excision.5
Adrian C Bateman Consultant Histopathologist
Clair EH du Boulay Consultant Histopathologist
G Harry Millward-Sadler Consultant Histopathologist
Department of Histopathology,
Southampton General Hospital, Southampton SO16 6YD
References
1. Kee F, Wilson RH, Harper C, Patterson CC, McCallion K, Houston RF, et
al. Influence of hospital and clinician workload on survival from
colorectal cancer: cohort study. BMJ 1999;318:1381-1385. (22 May).
2. Adam IJ, Mohamdee MO, Martin IG, Scott N, Finan PJ, Johnston D, et
al. Role of circumferential resection margin involvement in the local
recurrence of rectal cancer. Lancet 1994;344:707-711.
3. Pathology-guided treatment in rectal cancer. A randomised trial
comparing pre-operative radiotherapy and selective post-operative
chemoradiotherapy in rectal cancer. MRC-CRO7 Trial. Clinical Protocol
1998: 23.
4. Shepherd NA, Baxter KJ, Love SB. Influence of local peritoneal
involvement on pelvic recurrence and prognosis in rectal cancer. J Clin
Pathol 1995;48:849-855.
5. Hermanek P, Hutter RVP, Sobin LH, Wagner G, Wittekind Ch (eds).
TNM Atlas: illustrated guide to the TNM/pTNM classification of malignant
tumours/UICC, International Union against Cancer, Fourth Edition. Berlin,
Heidelberg, New York. Springer-Verlag 1997:102.
Competing interests: No competing interests
Editor
The number of new colorectal cancer patients (3217, over a five year
period) given by Kee et al (1) would give a crude incidence rate for
colorectal cancer in Northern Ireland of 39.5 per 100,000 persons.
However, as their study included only patients having surgery, correction
needs to be made for this. Although population rates of surgery for
colorectal cancer are not widely published, the range appears to be quite
narrow in the U.K. and Ireland, typically from 80%-85% of all incident
cases.(2 3 4 5) The rate of histological verification of diagnosis in
Northern Ireland, at 87% (6), is typical of UK and Irish registries and so
rates of surgery can be assumed to be consistent with the other published
data.
Using a range of rates for surgical intervention from 80% to 85% the
overall incidence rate for Northern Ireland would be estimated, from the
data of Kee et al, to be between 46.6 and 50.3 per 100,000 persons per
year. This is considerably lower than the population-based incidence
estimates for
Northern Ireland, 1993-1995 (57.2 per 100,000). This discrepancy in
incidence rates raises the possibility that some patients may have been
missed by the colorectal cancer registry. As all hospitals may not have
been equally assiduous in reporting cases to that registry, survival data
from some centres may be biased. Caution is needed in interpreting data on
survival from hospital-based registration, and comparison of survival
between hospitals should be based, where possible, on cases from a
comprehensive population-based registry.
Dr Harry Comber,
Director,
National Cancer Registry (Ireland),
Elm Court,
Boreenmanna Road,
Cork,
Ireland
1 Kee F et al. Influence of hospital and clinician workload on
survival from colorectal cancer: cohort study. BMJ 1999; 318:1381-6.
2 Cancer in South East England 1992. Thames Cancer Registry 1995.
3 Cancer in Ireland 1995. National Cancer Registry, Ireland, 1998.
4 Hospital inpatient care and cancer patients. South and West Cancer
Intelligence Unit, 1996.
5 Commissioning cancer treatment services: the role of the regional
cancer registry. Thames Cancer Registry, 1994.
6 Cancer Incidence in Northern Ireland 1993-95. Northern Ireland Cancer
Registry. Stationery Office, 1999.
Competing interests: No competing interests
When I was young I was a great fan of novels by Capt WE Johns about
Biggles, ace pilot extraordinaire. I have a clear memory of Biggles flying
Sopwith Camels during the first world war talking about overconfident
pilots who had flown just enough hours to get much too cocky but after a
few more hours flying time finally realising they know nothing. More
pilots would make mistakes during this short period and crash than at any
other time
The authors of the paper and commentary have no explanation for the
slightly worse outcome in the hospitals with medium volume workload.
Shields considers it might be " a statistical oddity". I wonder if Biggles
might understand this phenomenon ?
Competing interests: No competing interests
Author's reply: ascertainment of cases operated on is important to volume - outcome debate
Dr Comber draws a distinction between hospital based registration
systems and a population-based registry and raises the possibility that we
have missed cases of colorectal cancer. The leaders of both the Northern
Ireland Cancer Register and the Colorectal cancer Register have been
sharing their data for several years. Those at the Northern Ireland
Colorectal Cancer Register are aware that their ascertainment focuses on
those having surgical intervention. As our paper indicated, we chose to
focus our analysis on patients who had surgical intervention as this seems
to be the most obvious first step to take in analysing the volume-outcome
debate. We know from the hospital Patient Administration Systems that
over the period 1990 to 1994 there were 3,414 inpatient episodes with a
diagnosis of colorectal cancer and an OPCS-IV operation code of HO4-H20
and H30 or H20-H28 or H33-H41. The episode count over the period almost
certainly over-estimates the number of patients having these operations
with a diagnosis of colorectal cancer and so it is likely that our
ascertainment of all those having surgery is at least 92% and probably
materially higher . Dr Comber's letter raises an interesting and
worthwhile subsidiary question and one which we have discussed with the
director of the Northern Ireland Cancer Register i.e. is the outcome of
patients who don't receive surgery any better or any worse if managed in
hospitals with high surgical or low surgical volumes. Our study did not
attempt to answer this question.
The quoted incidence of colorectal cancer drawn from the first report
of Northern Ireland Cancer Register places the Province in an extremely
high position in the league table of colo-rectal cancer incidence in the
UK. However, the mortality to incidence ratio for the Northern Ireland
Cancer Register is among the lowest in the UK and few local clinicians
believe that this has much to do with better survival in Northern Ireland.
While the standardised mortality ratios for colon cancer for men and women
in NI are 20% and 9% in excess of those in the rest of the UK, the
published age-standardised incidence is between 30% and 40% in excess of
that in England and Wales but the excess for rectal cancer (more than 10%
and 20% greater than the rates in England and Wales for men and women) is
even harder to reconcile when the standardised mortality ratios from the
disease are lower than in the rest of the UK. Clearly the incidence trends
over time will merit scrutiny.
ClarkeA, McKee M. The consultant episode: an unhelpful measure.
BMJ 1992; 305: 1307-8
Competing interests: No competing interests