UKACR QUALITY AND PERFORMANCE
                                  INDICATORS 2009: FINAL

CONTENTS                                                                         PAGE

INTRODUCTION                                                                     1

COMMENTARIES FROM THE REGISTRIES                                                 2

APPENDIX 1:Technical Notes     2009                                              38

APPENDIX 2: Conversion of TNM of Dukes Staging                                   46


TABLE 1:    Registrations and timeliness reported by registries as at 30/06/09   47

TABLES 2:   Ascertainment and quality of the data held

            TABLE 2 A         % Change in registrations (males)                  48

            TABLE 2 B         % Change in registrations (females)                48

            TABLE 2 C         Childhood cancer incidence rates                   48

            TABLE 2 D         % Death Certificate Only (DCO)                     49

            TABLE 2 E         % "True" DCO                                       49

            TABLE 2 F         % Zero survival                                    50

            TABLE 2 G         % Microscopically verified                         50

            TABLE 2 H         % Non-specificity of morphology code
                              for cases which are microscopically verified       51

            TABLE 2 I         Mortality : Incidence ratios                       51

TABLES 3:   Completeness of the dataset

            TABLE 3 A         Demographic and diagnostic details                 51

            TABLE 3 B         Treatment and screening information                52

            TABLE 3 C         Specific staging information and grade             53

UKACR report2009_final.doc
CONTENTS                                                                PAGE

TABLE 4:    Trends in timeliness (as reported by registries)            54

TABLE 5:    Trends in timeliness (as reported by ONS)                   55

TABLE 6:    Trends in % Death Certificates Only                         56

TABLE 7:    Trends in % microscopically verified                        57

TABLE 8:    Trends in Mortality : Incidence ratios                      58

TABLES 9:   Trends in Completeness of the Dataset

            TABLE 9 A         Demographic and diagnostic details        59

            TABLE 9 B         Treatment and prognostic tumour details   60

UKACR report2009_final.doc

The following commentaries were supplied by each cancer registry in the United Kingdom
and concern the tables of results for the UKACR Annual Performance Indicators 2009. The
majority of registries do not attain the target of 100% for unique health identifier. This target
is not achievable since some patients are military personnel or in prison and thus will have
no unique health identifier. Thus, if a registry does not attain this target no comment has
been made for this reason.

Additionally, a large number of registries will report high proportions of death certificate only
registrations (and zero survival rates) for lung cancer and ill-defined sites due to the poor
survival seen for lung cancer and hence being registered after death and unspecific death
causes that are described on death certificates, hence these being ill-defined. Thus, again if
a registry has not attained a target of 2%, a comment was not necessary for this reason.

Eastern Cancer Registry and Information Centre (ECRIC)

The ECRIC now covers the population for the entire East of England of 5.66 million. Over
the last year we have standardised the registration process across the Region; all sites now
send electronic pathology and PAS data which is matched and quality assured in our new
cancer registry database. From next year our staff will be able to access the database
securely over NHS net, improving our collection at remote registry sites. We are working
hard to improve our clinical data sources and now have weekly electronic feeds of MDT data
from many of our Trusts who use the Somerset system. By early next year all our clinical
data providers should be able to supply regular outcome data, which we will quality assure
and feedback to their MDTS.

Table 1
There has been a slight increase in the total number of registrations for this year. This most
likely reflects improved ascertainment by the registry, rather than a genuine increase in
malignancy in our population. The increased ascertainment comes from richer and higher
quality electronic sources of data, combined with better engagement between the Registry
and the whole cancer community across our Region.

Table 2a/2b
Four malignancies show a significant increase on previous years.

The number of malignant melanomas reported in men has also increased based on the 3-
year average (13.5%). While this might be a genuine trend, the number of cases is very
small and there is considerable variation year-on-year – indeed the absolute figure for this
year is down by 25 on the 2006 figure. There is no significant increase in the number of
melanomas reported in women.

In situ breast carcinoma (12.7% increase), we feel is a genuine rise in the rate of this
disease and reflects the high level of reporting of ductal carcinoma in situ in those women
presenting to both the screening and symptomatic breast services.

In situ cervical carcinoma also shows an increase (12.6%) and this rise we believe is due to
improved ascertainment by the Registry.       Several factors contribute to this.    In some
laboratories we have clarified the SNOMED codes that are used to select pathology reports
that are sent to the Registry – this has raised the numbers of borderline/in situ records we
receive. We have use the KC53/61 data from the QARC for total numbers of in situ cervical
malignancy and with this we have an indication of the target we should expect. In some
cases, the QARC can directly query the laboratory data; in the future we should be able to
pull cytology/histology data for the entire regional cytology service.

The number of cases of colorectal cancer in women also shows a significant increase this
year (7%) and this continues a trend seen for the last four years. The reasons for this are
not entirely clear – they are at odds with the overall national trend which shows little change
in the incidence in women. The figures also predate the introduction of the bowel cancer
screening programme, which might well be expected to create an apparent increase in the
number of cases due to lead-time bias effects.

Table 2e
We are pleased to see that our DCO rates have improved significantly from those in 2006.
Previously, our practice was aggressively to pursue through a search of the notes in
secondary care as soon each death card notice that was potentially a DCO was received at
the Registry. Although this was often rewarding, it was very labour intensive. Over the last
year we have adopted the practice used by the West Midland Cancer Intelligence Unit and
now write to the GP with a proforma asking for details on any patient who is a potential DCO.
The number of requests to individual GPs is small and most complete the data willing and at
no charge. Where necessary, we will also request GP notes, although these tend to be of
increasingly limited value as many of the recent GP records are only available electronically.

Table 2h
Following examination of the 7 NOS morphology cases that were histologically verified for
cervical cancer, 4 cases have subsequently been updated to a specific morphology and so
this figure is now 1.5% and within the specified target.

Table 3a
Data completeness for core data items remains good. The main deficiency continues to be
ethnicity coding. Although over 96% of our records have a valid ethnicity code, because of
changes a few years ago in the national coding vocabulary for ethnicity, over half our
records have old codes, and these are classified in the PIs as invalid. It may be possible to
update some of these historic codes from the new CR-HES linked dataset, or in others map
them to the new codes; this could produce some improvement in the figures. Otherwise it is
likely to take some years to purge these old codes from the data.
Table 3b
We continue to have an issue with the full screening histories. Data transfer between the two
organisations is now working well, and data has been sent by the QARC to the individual
units for checking. 7 of the 11 breast Units have checked all their historic data from 1999-
2005 and the others are due to complete this within the next few months. The backlog of
work was considerable. The QARC has also employed additional staff and they will work
with the Breast Units to collate all interval and screening history data. This issue will be
resolved in time for the PI indicators report in 2010.

Table 3c
Staging data for the four malignancies measured is in generally well collected by the registry,
but is still not available in all cases. We anticipate that the emphasis placed on staging and
outcome data in the Cancer Reform Strategy and the new MDT data feeds that we receive in
the registry will significantly improve the completeness of these data items. Even with these
changes the effects are likely to take a few years to appear in the PI reports.

North West Cancer Intelligence Service (NWCIS)

General comment
Developments over the past year in improving data quality and processing have included:
(i)     The use of PRAXIS software at Liverpool and Manchester Offices
(ii)    Use of CWT, breast and cervical QARC and MDT data
(iii)   Development of a National Registry Best Practice document to be used to improve
the PI indicators for NWCIS

Table 1: Timeliness of registrations
Our reported completeness of registrations is 96%; outside the target range of 98% to 102%.
We have identified 706 cases that were on our database on 30.6.2009 which were not
included in our returns. These have since been sent and accepted by ONS.             The 706
patients were made up of three groups: two of whom were not sent because of teething
problems with PRAXIS which have been rectified. The Manchester and Liverpool offices still
have separate databases and the third group were residents in one area who were managed
in another. The specific problem has been rectified but this shows the importance of moving
to one database as soon as possible.

We believe that the revised figure of 98.1% is still an underestimate of the true number of
cases in the Region. For 2005 and 2006 we used PAS as a source for identifying cases, we
did not use it for this purpose in 2007 due to initial difficulties with PRAXIS, we will do so
again for 2008 cases.

Tables 2A and 2B: %Change in registrations
Lung male
Though underascertainment in general and a real drop almost certainly do contribute to the
decrease, there is evidence of underreporting of lung cancer cases with a good prognosis.
This will be investigated by comparing the numbers of cases reported by hospitals for 2005,
2006 and 2007, checking against their PAS extracts and writing to those hospitals with a
significant decrease in numbers in 2007. For 2008 we will also write out to hospitals for
details on cases mentioned on PAS but for which we receive no further details (such cases
are not registered without further details).

We will investigate the drop of 10% in a similar manner to lung.

Haematology: males and females
121 of the 706 cases that were not returned for the PIs were haematological; their inclusion
would have led to a considerable improvement in our results.         However, we do have a
problem in that many haematology laboratories in the NW do not allocate a diagnostic code
to specimens.      This makes it very difficult to identify those patients with a malignancy
diagnosed on examination of a peripheral blood specimen, leading to underreporting of
cancers such as leukaemia, especially those with a good prognosis. We undertook work on
2006 cases which ceased due to lack of funding; this resulted in the numbers for 2007 not
being as good as those for 2006.        We are taking a number of steps to improve data flows
(i)    Providing timely feedback to labs on their performance and if not getting the required
response, writing to chief executives
(ii)   Seeking funding for an individual who will set up systems for obtaining data from
haematology labs

Breast invasive
We believe that the reported decrease of 4% can be explained by the underascertainment
and by a small true decrease in rates.

Cervix in situ
This large decrease will be investigated by the actions outlined under lung.

All xnmsc
See explanation under Table 1.

Table 2D and 2E: % Death certificate only (DCO)
Most of our DCO rates are above the 2% level; they are also higher than last year’s rates.
The explanation for the rise in the % of DCOs is that the introduction of PRAXIS led to a
problem with inputting data provided by GPs on patients identified by death certificates. This
is being addressed.

Rectifying this problem probably would decrease our DCO rates to about the level of 2006 –
just below 4%. To decrease these further we need better data from primary care on patients
who are identified via death certificates. We are developing better links with PCTs through
data quality managers and setting up systems whereby GPs receive requests for information
while they still have ready access to the deceased patient’s notes.

Table 2F: % Zero Survival
Our zero survival rates are higher than last year for the same reason as for DCOs and will
be addressed using the same measures. The percentages for lung and haematology are
particularly high due to the specific problems for these sites of underascertainment of
patients with a good prognosis.

Table 2G: % Microscopically verified (MV)
Though all of our values for MV except for haematology have improved since last year, our
rates still do not meet the target. This is due primarily to incomplete reporting by several
pathology labs in our Region. We have been working on this for some time with limited
success. The steps outlined under haematology will also be used to improve data flows
from histopathology laboratories.

Table 2H: Non-specific morphology for MV cases - Bladder
We have reviewed the cases and the coding is correct.

Table 2I: Mortality : Incidence Ratios (MI)
Our MI ratios are high for lung and haematological malignancies due to underreporting of
patients with a good prognosis.

Table 3A: Completeness of dataset
The only value that does not meet the target is unique identifier at 99.8% which is above the
national average.

Table 3B: Treatment and screening information
Most of our values are similar to the national average and the results for % identified by
screening are a great improvement on last year. The one poor value is that only 62.5 % of
patients aged 0-24 are recorded as having had any treatment. This will be investigated as a
matter of urgency.

Table 3C: Stage
Most of our staging values have improved since last year and we now reach the national
target for half of the measures. This is a credit to the hard work of our staff. We are still
below the national target for the other four due to incomplete reporting of surgical specimens
by several path labs. We are running a large project to obtain staging details from MDTs
which we believe is better than attempting to determine stage from pathology reports where
it is not explicitly stated.

Northern and Yorkshire Cancer Registry and Information Service (NYCRIS)

General comment
Despite a considerable amount of hard work by our registration and QA staff, NYCRIS did
not meet the completeness target for 2007 registrations. As at 30/6/09, registrations were
96.4% complete. The shortfall can largely be explained by difficulties in gaining access to
case notes at a small number of Trusts. For 2008 registrations, we are hoping to reduce the
requirement to access to case notes by combining Cancer Waiting Times data with that from
additional notifications or links to Trust systems.

Late completion impacts, in turn, on several other indicators such as stability of incidence,
M:I ratio and completeness of the dataset.

We aimed to reach the 98% target by mid-July and were 98.8% complete as at the 17th July

Table 1 - Registrations and Timeliness
The figure reported has improved from that reported in the previous two years.

Table 2A/2B - % Change in registrations
The increased incidence of Melanoma on males is consistent with national trends.

The shortfall for in-situ cervix cases has improved from last year’s position as a result of
audit work undertaken with those pathology laboratories showing shortfalls in provision;
some of this work is still ongoing.

The reported reduction in incidence of invasive xnmsc female cancers may be due to a
cumulative effect of marginal differences in incidence resulting from NYCRIS not yet having
completed all its 2007 registrations.

Table 2C – Childhood incidence rates
At the time of reporting we had been unable to obtain data from one of our regional
childhood registers. We have subsequently received this data, which has potentially
identified a further 10 cases. However, only 1 case is in the male 5-9 age group.

Table 2D/2E – Death Certificate Only - “Current” and “Previous”
We are consistent with the rates reported by other registries, with the majority of DCO cases
being in the 75 plus age group. The rates for 2007 registrations are expected to improve
when trace back has been completed.

Table 2F - % Zero Survival
We are consistent with the rates reported by other registries for ill-defined sites

Table 3A/3B – Completeness of dataset
The proportion of cases with an NHS number is expected to rise once the registration
process has been completed for 2007 cases although there are always a number of records
that we would not expect to find on PDS i.e. armed forces personnel, prisoners etc.

Diagnosis date was not 100% as the registration process for 2007 was not complete as at

NYCRIS only began to collect NPI score for 2008 diagnosed cases onwards. For 2007
cases we collected breast staging according to the TNM classification and this is available
for 89% of cases. Although we do not collect the number of positive nodes, we know that
83% of cases had positive nodes after node sampling.

Oxford Cancer Intelligence Unit

General comment
In October 2009 the OCIU will be going live with a new database, CBASE, shared with
South West Cancer Intelligence Service (SWCIS). The past year has seen much additional
data cleaning in preparation for our merger with the SWCIS database, and some coding
changes have been implemented to the current data to facilitate this merger. During this
period of transition we have striven to maintain high levels of data quality and timeliness.

Quality Assurance (QA) programme
OCIU has a comprehensive QA programme aimed at maintaining and improving data
quality. The programme includes:
•      Monthly validation and duplicate checks.
•      Monthly 2% quality checks of all new registrations/amendments on the database.
•      Cancer Waiting Times (CWT) data quality checks.
•      Patient Administration System (PAS) data quality checks.
•      Death Certificate Only (DCO) follow up.
•      ONS pre-download checks.
•      Quarterly data completeness checks.
Other ad hoc checks e.g. against clinical databases are undertaken as indicated.

Table 2A and 2B (% change in registrations)
The number of cancers registered in 2006 (for both sexes combined and excluding non-
melanoma skin) has fallen by 1.0% compared to the three previous years. There have been
no significant changes in men, but for women there has been a significant increase in cervix
in situ and a significant decrease in haematological malignancies.

A problem collecting cervix in situ cases from one of our local acute trusts in 2006 has been
successfully addressed, and we are now receiving data as expected. This years increase of
9.6% has been investigated and reflects the natural fluctuation in the number of new cases
registered each year rather than any specific problem with the data.

We have had continuing problems receiving some haematological notifications (eg. blood
films, bone marrow reports) from some of our trusts. We are working with data providers to
improve ascertainment. Further analysis of haematological cases show that the number of
cases for females is down, for most diagnostic groups and for most trusts. New data feeds
from MDTs (including haematology MDTs) are being established and should help to address

Case ascertainment is regularly reviewed by hospital trust and tumour site to identify any
potential problem areas.

Table 2D and 2E (% DCO)
All DCO rates for 2007 are within target except lung at 2.3%. The 30 lung DCO cases are
being reviewed. In 2006, there were three cervix invasive DCO cases. These have been
followed up and no additional information has been found.

Table 2H (Non-specificity of morphology codes for microscopically verified cases)
Fourteen microscopically verified bladder cases have been reported as having a non-specific
morphology code. These were re-examined, and eleven cases have had their morphology
code changed to a more specific code.

Table 2I (Mortality:Incidence)
Our M:I ratios for cervix invasive and prostate cancers have been consistently low for a
number of years. We do not believe that we are missing cases. This may reflect better
survival in the registry catchment population.

Table 3A (Completeness of dataset - demographic and diagnostic details)
All registrations without an NHS number are sent to the NHS Strategic Tracing Service as
part of a batch trace but a small number cannot be traced and therefore cannot be assigned
an NHS number. For 2007 registrations, only 48 registrations (0.4%) did not have an NHS
number. This may be due to anomalies with name spellings or dates of birth and postcodes
etc. It also has to be kept in mind that some individuals (e.g. prisoners) do not have an NHS
number so achieving 100% coverage will always be challenging. These 48 cases have
since been traced individually and 25 NHS numbers have been added.

The level of ethnicity coding is lower this year because the 2007 HES data was received too
late to add ethnicity to this years cases. This is now being addressed.

Table 3B (Completeness of the dataset - treatment and screening information)
As a registry using largely electronic sources and with no routine access to case notes, there
have been few reliable sources of information on hormonal treatments.            We are now
collecting information on hormone treatments for prostate cancer via the cancer waiting
times data and this has led to an increase in the proportion of prostate cases recorded as
receiving hormonal treatments.

OCIU’s level of recording of screening history and screening category for both breast and
cervical cancer cases is higher than the national average. This reflects the fact that we have
well established working exchanges of data with both the the South Central Breast and
Cervical Screening Quality Assurance Reference Centres (QARCs) which cover most of our
catchment area. Data flows between OCIU and the East Midlands Breast Screening QARC
are also well established. We are continuing to work on the links with the East Midlands
Cervical Screening QARC and have been sending them data regularly for the last two years.
We have not yet received any data back from them but have recently agreed a schedule for
data exchanges.    Service Level Agreements are now in place with all four QARCs.

Table 3C (Completeness of the dataset - specific staging information including grade)
The level of recording of stage (NPI score) is lower than the target because of problems with
the recording of tumour size. Current practice is that if a cancer was incompletely excised,
or the patient had chemotherapy or radiotherapy prior to surgery, then the size is recorded
as unknown. For 2008 cases onwards, the UKACR guidance on recording breast tumour
size will be fully implemented.

We are continuing to look at improving our level of recording of colorectal stage as part of
our ongoing QA programme. New data feeds from MDT meetings are being established and
should help to address this.

Most melanoma pathology reports include a record of Breslow thickness. Some reports also
record Clark’s level and a small proportion only record the Clark’s level. OCIU’s practice is
to record Breslow thickness if present and to record the Clark’s level only if the Breslow’s is
missing. On our current computer system, we are unable to record both if both are present.
We will start recording both if present, when we move to a new cancer registry computer
system in autumn 2009.

South West Cancer Information Service (SWCIS)

General comment

The Merging of Oxford and South West cancer registration databases
The Oxford and South West cancer registries are currently working on a NCIN funded
project to merge the two cancer registration databases into a single shared application
accessible via a secure VPN link within NHSNet.
This is the first time that two registries will be able to share a common database, work on the
same cases, utilise common national data sources and share data processing methods and
technical expertise.
The project will also facilitate the aims of both registries to provide a cancer intelligence
function covering the combined regions of South West and South Central England.

Data Visualisation Projects
The South West cancer registry is also working with the NCIN on some data visualisation
projects, namely:
-    to transfer the annual UKACR Quality and PI Reports into data visualisation objects to
show comparisons (and trends) of registry performance against sets of indicators, and
-   to prototype a tool to specifically show comparative data across Trusts and Networks in
relation to data completeness (for different data sources) and PI performance across Trusts

NHS Standard Contract for Acute Services
The SWCIS have been working closely with trusts in meeting the requirements of the NHS
Contract for Acute Services. Progress to date:-
Services Level Agreements set up with all trusts
PAS electronic extracts agreed and submitted from 2008 onwards
Pathology electronic extracts agreed and submitted from all trusts
MDT data collected and submitted in an electronic format
-      Somerset Cancer Register              13 trusts      data submitted 2006 >
       •       Further 3 trusts purchased
-      Dendrite and MDI                      2 trusts       data submitted 2006>
-      Info flex                             6 trusts       datasets agreed – extract
                                                            routines being developed
-      In house systems                      2 trusts       datasets agreed – extract
                                                            routines being developed
Radiotherapy electronic extracts agreed and submitted from 9 of the 11 centres within the
region. Discussions are in progress to ensure full coverage is achieved.
Chemotherapy is currently received through various routes, namely CWT, PAS, MDT’s and
other individual databases. Work is currently underway to establish the gaps and work
towards achieving full coverage.

Table 1 Timeliness
The overall 2007 ONS ready figure shows a decrease on 2006.
The Royal United Hospital in Bath implemented a new pathology system early last year but
due to ongoing software issues it was not possible to provide ourselves with an extract of
pathology reports for diagnosis year 2007.
All cases registered from HES have been followed up through the online Avon Virtual
Pathology System. Once the issue is resolved an extract will be submitted and the data

Table 2B – Stability of Incidence - % increase in registrations Female Haematology
There was an increase in ascertainment from 2004 – 2006 which is having an impact on the
overall figure. Checks are also being carried out to ensure that changes to our HES business
rules have not impacted on this figure in some way though overall the figure is consistent
with the regional 10-year trend in haematological cancer incidence.

Table 2C Childhood cancer Incidence rates – females age 10 – 14
The actual numbers are very small hence the potential variation is quite large. The
percentages suggest they are within the range of natural variation for the period.

Table 2D - % Death Certificate only
All have been followed up through hospital systems and GP notes. We feel that this is a true
reflection within our region.

Table 2E - % Death Certificate only for previous diagnosis year
See comment for Table 2D above.

Table 2F - % Zero Survival
This has been decreasing year on year with improved follow up through various hospital
systems including radiology and increased data sources.

Table 2I – Mortality:Incidence Ratios
The affluence of our area and the use of HES are likely to reduce the M:I ratio overall.
Cervix Invasive is in-line with the UK average. Colorectal is 0.01% down on the value
reported last year.

Table 3A
Unique Health Identifier: All registrations without an NHS number are sent to the NHS
Demographic Batch Service (DBS). Of those patients registered 42 cases remain without an
NHS number.

Table 3B
Breast Cancer – Hormone
The figure is lower for 2007 as we are currently In the process of writing a procedure to
automate the uploading of specialist data and when this is complete we expect to see an
increase on the 2006 figures.
The manual input of data was completed on 20th August which resulted in the figure being
Negotiations are currently underway to improve the completeness of radiotherapy
information. We recognise this as an issue and hope to see some major improvements over
the coming year.
Cervical Screening
2006 Cervical Screening Data was not processed prior to the 30th June cut off, this data is
now complete and the % screen detected is now 15% which is in-line with the UK average.
Breast cancer - % with full screening category for ages 50-67
The implementation of the Interval Cancers Database (developed by SWCIS) will allow
potential interval cancers to be easily identified and sent to the BSQARC’s in order that they
can confirm the individual screening status with the relevant screening office and then
feedback to SWCIS to update the screening category. This has recently been implemented
by the South West BSQARC and South Central BSQARC will also take the system on end
2009/beginning 2010. This should start to have an impact on this category from AY08
onward – prior to this the identification of known intervals has been incomplete.

Thames Cancer Regsitry

The sections below discuss only those measures which have been highlighted as in some
way divergent from observed national figures.

Table 2a/2b - % change in registrations – males & females
The percentage change for all tumour sites for both male and female falls within the
expected upper and lower threshold of + or – 2%. Variation for individual sites that fall
outside of the expected ranges can be explained by the greatly increased access to, and
processing of, pathology data. This is particularly marked for melanoma of skin and in-situ
cervix registrations. In most cases (e.g. prostate) this increase is also in line with the
direction of national trends in incidence.

The percentage increase in female breast in-situ cancers is consistent with both the England
and UK trend and averages and the expected impact of broader inclusion in the breast
screening program.

We have no specific information which would absolutely explain the 5.8% reduction in male
lung cancer but would note that this follows the direction of the national trend and may also
have been impacted by the success of efforts to trace back DCO cases.

Table 2d/2e - % death certificate only - males & females
This year TCR continues the year on year improvement of DCO rate across all tumour sites,
reporting 2% overall for cancer in all sites for 2007, reduced from 3% for 2006 (See table 6).
We continue to do all we can to improve our primary case ascertainment and follow up of
DCI cancer cases.

Table 2g - % microscopically verified - males & females
The percentage of microscopically verified female breast in-situ registrations falls short of the
nominal expected target of 100% by 1.6%. This is primarily due to information received from
screening centres on in-situ diagnoses without a clear indication of MV status. We continue
to work with our screening QARCs to improve and clarify our understanding of the data
received from the screening service.

Table 2h - % Non-specificity of morphology codes for cases which are
microscopically verified
It has been indicated that there may be some difference between average UK figures for
non-specificity of histology coding for bladder, prostate, and colorectal cancers. We have no
specific information to indicate a likely cause for this or if this apparent difference is
significant. We will continue to examine and quality assure our processes and coding
practice to ensure that such variation does not arise out of systematic error.

Table 3a - Completeness of the dataset - demographics and diagnostic details
Completeness of data on anniversary date was recorded as 98.2%. This is due to a strict
interpretation of the reporting guidelines. Thames uses a partial date where it is not possible
to ascertain the exact date of diagnosis (usually a month and year of diagnosis). This allows
more precise data received after initial registration to take precedence in automated
systems. A completed date of diagnosis is created by algorithm in data for analysis.

Table 3b - Completeness of the dataset - screening information
This year we have exchanged information on screen detected cancers with both of our
screening QARCs for 2006. Data from one QARC arrived very close to closedown and is not
yet reflected in figures. Data on full screening history for 2006 has not been received.

No information is received from the cervical screening service on screening category
although TCR does supply information to the screening QARC.

Table 3c - Completeness of the dataset - specific staging information including grade
TCR continues to improve the quality and completeness of staging information captured
from clinical notes and electronic data. The recording of clinical stage for the 2007
registrations increased to 49% which is an improvement of 5 percentage points from 2006.
The Registry is working with data suppliers and its own staff to increase the ascertainment of
stage, and to reduce the variation in the proportions of staging by TCR and by clinicians.

Trent Cancer Registry

General Comment
The priority over the last twelve months has been to continue the improvement in levels of

In the last year we have made considerable progress identifying gaps in our ascertainment
of in situ cervical cancers. This has increased the number of in situ cases by well over 10%
for each diagnosis year since 2004. This has masked the percentage increase that would
have been seen if ascertainment had only improved for the 2007 diagnosis year (of 20%).

We are now following up all Cancer Waiting Times records for patients in our catchment
area where we have not received any other data from a trust. We do not end up registering
all of these ‘missing’ CWT cases as some are not registerable or are not new cancers. This
follow up however has led to another increase in ascertainment.

Very recently we started to receive feeds from the remaining pathology system not supplying
data and now have 100% pathology coverage of the population. This is not yet reflected in
the figures for 2007.

Tables 1-2B (Registrations and timeliness)
Generally all increases of more than 2% are a result of better ascertainment of cases due to
increases in electronic pathology data feeds, following up CWT cases where we have no
registration and improved working practices.

We are currently investigating the apparently high increase in the number of female lung
cancers and plan to investigate the change in numbers of registrations for haematology and
male bladder cases in the near future.

Table 2C (Childhood Cancer Incidence Rates)
We have reviewed all female registrations in the 5 to 9 age group. This has resulted in the
cancellation of a number of tumours. The rate now stands at 12.9, which falls within the
acceptable range.

Table 2D-2E (% Death Certificate Only (males and females))
Our overall %DCO trend has been improving consistently since 2003 and all the relevant
cancer sites, except lung and haematology, now have DCO rates of less than 2%. Our
overall DCO rate is now 2.0%. This is largely due to the thorough follow-up procedure in
place. Increases in pathology data have also contributed and the use of Cancer Waiting
Times has helped us to identify the most relevant trusts to contact when other follow-up
procedures have been unsuccessful.

Table 2G (% Microscopically Verified (males & females))
The number of prostate registrations that do not have a microscopic basis of diagnosis is
high and amounts to around 100 more cases than expected.             We intend to review all
prostate registrations where the basis of diagnosis is not microscopic.

Table 2H (% Non-specificity of morphology codes for cases which are microscopically
We plan to review the 7 cases of in situ cervical cancer that do not have a specific
morphology code.

Table 2I (Mortality : Incidence ratios)
The M:I ratio is low for lung and cervical cancers, suggesting that we have more registrations
than expected for these sites.

We have made considerable progress improving the ascertainment of in situ cervical
cancers, which has had a knock-on effect of increasing ascertainment of invasive tumours,
and is the most likely explanation for our low M:I ratio. If we exclude micro-invasive tumours
from the calculation, the M:I for cervix is 0.32, which is the same as the UK average.

The lung M:I ratio suggests that we have an excess of 200 registrations in this year, which is
approximately 6% of the total.     We will decide a way forward with this following our
investigation of female lung cancers.

Table 3A (Completeness of the dataset – demographics and diagnostic details)
We batch trace all registrations with the National Strategic Tracing Service which allows us
to obtain missing NHS numbers. On 30th June 2009, 31 of our 2007 registrations did not
have an NHS number.

Table 3B (Completeness of dataset – treatment information)
For many years our primary sources of treatment information have been PAS/HISS
notifications which are submitted by clinical coders soon after the patient’s discharge. The
notifications include treatment indicators (Yes, No or Not known) for performed and intended
treatments and these details may or may not be complete or accurate, particularly in relation
to intended treatments.

For 2007 registration data we have started to use information from HES to update our
surgery and chemotherapy indicators.

Surgery indicators were updated for cancers of the colon, rectum, lung, breast, cervix,
bladder and prostate. This increased the number of cases with surgery for 2007 diagnoses
from 26.0% before the update to 35.7% afterwards.

Chemotherapy has been updated for all tumour sites and we believe that the resulting
percentage of 22.7% is more accurate than our figure of 8.5% for the previous calendar

Notifications from PAS are still the only route by which we obtain information on radiotherapy
and hormone treatments. The lack of completeness of this information can be seen clearly
when comparing our figures with those of other registries and is particularly obvious for
cancer sites where these treatments are common.

We intend to do more work using HES to update treatment information and also aim to
improve the data using information received directly from MDTs, CWT and RES.

We have recently agreed SLAs with both our Cervical Screening QA Reference Centres.
These agreements will result in us obtaining screening information for cervical cases
diagnosed from January 2002 onwards by the end of April 2010.

Table 3C (Completeness of dataset – specific staging information including grade)
We do not currently record NPI score (Breast), number of positive nodes (Breast), Clark
level (Melanoma) or Breslow thickness (Melanoma) as separate data items. Invasive size of
tumour for breast cancer is now collected, but was not collected for cases diagnosed in

We collect stage using TNM, although the individual components (T, N and M) are not
recorded, meaning that the figures for all sites other than colorectal could not be presented
in this table. The Registry continues to improve the completeness of all staging information
on the database and the figures for breast (73.4), colorectal (64.1), cervix (70.6) and
melanoma (73.0) are generally higher than those submitted in the previous report.

West Midlands Cancer Intelligence Unit (WMCIU)

General Comment
The WMCIU is pleased that the actions taken to reduce DCO rates has shown early
successes this year, with rates falling from 11% to 7%. The WMCIU will continue to improve
and monitor this scheme, and hope to see ongoing reductions towards the target.

The WMCIU is aware of issues with data submission from a number of Trusts in the region,
but is working closely with Trusts to resolve issues and arrange access to new data sources.
The WMCIU has improved the monitoring of data submitted to the unit to enable problems to
be identified earlier.

Table 2a/2b
The WMCIU showed no significant variation in total incidence of all cancers combined
(excluding non-melanoma skin cancer). However, there were significant variations recorded
in males for prostate cancer (-5.3%). Significant variations were recorded in females for in
situ breast cancer (+17.6%), in situ cervical cancer (-12.7%), and bladder cancer (-14.8%).

Prostate cancer
The number of prostate cancer cases in males fell in 12 of the 17 West Midlands PCTs, and
rose in 5. Numbers have now returned to 2003 levels, after 4 years during which they rose,
fell, rose, and fell once more. The fall in the number of cases is due mostly to a decreased
number of cases in men in their early 60s and early 80s, but there is no clear trend with age.
It is not known if the provision of PSA testing has changed in the region, but this may be
relevant to the changes in the younger age group.

In situ breast cancer
There has been a gradual increase over time in the number of in situ breast cancers
registered by the WMCIU.       The relatively large increases in 2005, 2006 and 2007 are
primarily in the age range 65 – 75, and are due to the NHS Breast Screening Programme
(NHSBSP) age expansion. As a result of the age expansion, the NHSBSP has detected
greater numbers of in situ breast cancers in women aged 65 – 70. It has also increased
‘breast awareness’ in 70 – 74 year olds resulting in increased self referral for screening with
an accompanying rise in the detection of in situ breast cancers. The WMCIU is surprised
that the rise in the number of in situ breast cancers is not apparent in all English registries.

In situ cervical cancer
The WMCIU has had a major problem with data submission for 2007 from one of the main
cancer centres in the region. The Trust installed a new pathology system which resulted in a
failure to report all relevant cancers to the WMCIU. An ongoing solution to this problem is
being sought, and as an interim measure pathology data for all invasive tumours were
obtained manually by WMCIU staff visiting the Trust and photocopying the original pathology
reports. However, this manual exercise did not cover in situ cancers as it was hoped that an
electronic solution to the problem would be developed. Unfortunately, this is still outstanding
and as a result the number of in situ cervical cancers received from this Trust is very low. If
this Trust had reported its 2005 numbers of in situ cancers (over 250) instead of the 20 sent
in 2007 there would not be a significant difference in the overall number of in situ cervical
cancer cases recorded.

As well as this specific problem, there are still three other Trusts where in situ cervical
cancer numbers have reduced noticeably since 2005. It is likely that this reduction is also
linked to the problem of coding and extracting in situ cancers on pathology laboratory
computer systems. The WMCIU is intending to use the good links between the Cervical
Screening QA Reference Centre and the laboratories to resolve these problems in 2008.

Bladder cancer
The number of bladder cancers fell in 11 PCTs and increased in 6. There was a fall in the
number of cases across all age groups, except for the 60-64 group where incidence was

Table 2d        DCO rates/ zero survivors
The WMCIU failed to meet the 2% Death Certificate Only (DCO) standards for 2007
registrations for all cancer sites (except breast (invasive), cervix (invasive) and melanoma
skin cancer).    These high DCO rates also explain the high number of zero survivors

The DCO rates are greatly improved from those reported in the 2006 UKACR Performance
Indicators. The WMCIU implemented a revised death certificated follow up process in 2007,
which involves requesting information from the patient’s GP, requesting information from
hospices, matching death certificates to the Cancer Waiting Times database and HES
records, and requesting data from the Trust named on the death certificate where
appropriate. Using the new follow up methods, overall DCO rates have fallen from 11% to
7%, and rates are reduced for all sites.       Analyses show that the use of GPs as an
information source has allowed the WMCIU to obtain data for patients who did not receive
treatment, but who had diagnostic imaging procedures for which there is currently no clear
data source within Acute Trusts. Of patients successfully followed up using this method,
around 25% died within 30 days of diagnosis, and 50% died within 60 days.

The highest DCO rates are found in groups where late diagnosis or no treatment are most
expected, such as tumours of ill defined sites and patients aged over 75. The exception is
haematological malignancies.     Unlike solid haematological malignancies, leukaemia and
other reports on blood cell malignancies are not generally sent to the WMCIU by
histopathology laboratories, which are the WMCIU’s primary source of data. The data on
these cases are therefore incomplete.      The WMCIU will be working with haematology
network site specific groups in 2009 to address this issue.

Table 2i Mortality : Incidence Ratios
Invasive cervical cancer
All UK registries have a M:I ratio lower than the 0.38 target. This suggests that there may
genuine changes in incidence, mortality and survival rather than a data quality issue.

Malignant melanoma
The UK average of 0.20 is higher than the 0.18 target. The WMCIU value of 0.22 is within
0.02 of the UK average.

Table 3a: Completeness of the dataset – demographics and diagnostic details
The WMCIU achieved 99.95% completeness for the postcode and therefore failed to meet
the 100% Cancer Peer Review Measure Standard. The UK average was 99.9%. There
were 14 cases without a postcode in the West Midlands in 2007.

Unique health identifier
The WMCIU achieved 99.0% completeness for the unique health identifier (NHS Number)
and therefore failed to meet the 100% Cancer Peer Review Measure Standard. The UK
average was 99.0%. There were 262 cases without an NHS number in the West Midlands in
2007. No English cancer registry met the 100% target. The WMCIU will continue to work to
improve completeness of NHS number, but notes that there are year on year improvements
in completeness of this field.

Anniversary (diagnosis) date
The WMCIU achieved 99.9% completeness for anniversary (diagnosis) date and therefore
failed to meet the 100% Cancer Peer Review Measure Standard. The UK average was
99.0%. There were 28 cases which had an incomplete diagnosis day and month in the West
Midlands in 2007. The majority of these have a clinical diagnosis basis and were registered
from death certificates. Obtaining definitive diagnostic dates is more difficult in these cases
than when the tumour is microscopically verified.

Table 3b: Completeness of the dataset – treatment and screening information
Treatment data - chemotherapy
Only 11.4% of cases registered on the WMCIU’s database received chemotherapy
compared with the UK average of 21.1%. Chemotherapy data in the West Midlands are not
routinely recorded onto computer systems that can produce data in an electronic formal
suitable for submission to the WMCIU, and there is not a routine flow of paper based
chemotherapy records from all Trusts.

The WMCIU is therefore aware that it does not achieve complete capture of all
chemotherapy data and is continuing to explore various methods to improve chemotherapy
data completeness. Electronic downloads of chemotherapy data are being supplied by one
outside region cancer centre that treats a relatively large number of West Midlands
residents. The WMCIU has also developed a process which extracts chemotherapy data
from HES data and automatically updates the cancer registration database. This has been
successfully trialled for 2006 data, but has not been rolled forward yet into 2007. Direct
access to pharmacy databases such as Bookwise is being explored with Trusts in the Pan
Birmingham Cancer Network.         The Greater Midlands Cancer Network and the Pan
Birmingham Cancer Network have now implemented the Somerset Cancer Register in all
their MDT meetings, and the WMCIU intends to collect details of chemotherapy treatment
from this. A pilot of paper based collection of chemotherapy data at one NHS hospital is
also underway. Although the WMCIU wants to focus its efforts on efficient, electronic data
collection, if paper proves to be a quick stop-gap, these records could act as a temporary
source of data while long term solutions are found. Site specific chemotherapy information
for breast cancer has been imported from the additional data supplied by clinicians to the
BCCOM Project, and it is hoped that access to the NCASP data will enable chemotherapy
data for other sites to be improved. The WMCIU is also looking forward to accessing the
new extended Cancer Waiting Times database which will include the start dates for all the
chemotherapy treatment provided to primary tumours and recurrences. This diverse range
of strategies should ensure that chemotherapy data collection is more complete in the future.

Table 3C: Completeness of the dataset – specific staging information including grade
Colorectal cancer
Only 71.9% of the colorectal cancers registered in 2007 have a known pathological Dukes
stage. This is 2.1% below the 74% Cancer Peer Review Measure Standard.

Of the 959 tumours with no Dukes stage, Dukes stage was not applicable for 171, with the
majority of these (138) being polyps. If these polyps had been recorded as Dukes A, then
the peer review measure would have been met. As well as the tumours for which Dukes
stage is inappropriate, 133 tumours were diagnosed either clinically or from clinical imaging,
and a further 135 were DCOs. There thus appear to be valid explanations to account for the
WMCIU’s failure to meet the Cancer Peer Review Standard.

Scottish Cancer Registry

Table 1 - Registrations and timeliness
The current backlog of cancer registrations in Scotland contrasts with the situation several
years ago when some of the English regional cancer registries had substantial backlogs.
For example, ONS first published data for the year 1994 in the year 2000. The backlog in
England was largely cleared following an injection of specifically earmarked resources from
the Department of Health (DoH) Public Health Development Fund in early 2000, followed by
uplifts to regional cancer registry budgets through Cancer Action Plan funding. There has
been no equivalent, targeted injection of resources in Scotland. The origin of the current
backlog in Scotland is multifactorial, but in large part can be traced back to a major re-
organisation of cancer registration that was initiated by the Scottish Office around 1995.
Five autonomous regional cancer registries were disbanded and ISD took over responsibility
for all aspects of cancer registration, including the implementation of a more electronic-
based method of data capture, and active validation by a team of outposted peripatetic
cancer registration officers. Unfortunately, this re-organisation did not attract dedicated new
resources, and initially there were inadequate funds to employ the full complement of
outposted staff envisaged in the original re-organisation plan. Thus, the development of a
backlog at this stage was almost inevitable, compounded in due course by a gradually
increasing annual tally of registrations, largely due to ageing of the population. Although
there was some evidence (from fortnightly tallies of registrations) that the backlog was
reducing during 2004-2005 (see Table 4), this has recently been set back by staffing issues
(for example, the sudden tragic death in 2005 of one of the staff based in the west of
Scotland), and importantly by two recent additional requirements falling on outposted cancer
registration staff, namely (a) collection of waiting times data for the approximately 14% of
patients with cancer who are not currently included in cancer audit data collection; and (b)
collection of morbidity data for a sample of cancer patients for a SEHD-funded research
project. A further factor that may have affected the timeliness of the cancer registration data
temporarily was the recent introduction of a new computer system, although in the longer
term, it is anticipated that this will make the process more efficient. Also in the longer term,
successful implementation of an electronic patient record (EPR) may offer the prospect of
achieving cancer registration in a more timely and less labour-intensive manner.

Table 2A and 2B - % change in registrations
Variations in the % change in registrations across the UK may reflect changes in the
efficiency of ascertainment by some registries (skewing the UK average), the play of chance,
or genuine differences in incidence patterns. Although a number of our % changes differ
from the UK average, these relate mainly to lower volume neoplasms, and none of these
give us particular cause for concern.

Table 2C - Childhood cancer incidence rates
Although our childhood cancer incidence rates are lower than expected for males in the 5-9
year age group, this may just reflect the variability associated with rare diseases. We are
reasonably confident that our figures reflect reality (and we have, for many years, cross-
checked our data with the NRCT at Oxford).

Table 2D and 2E - % death certificate only (males & females)
It is perhaps unsurprising that our highest DCO % relates to ill-defined sites, but it still
compares favourably to the UK average. Having already made strenuous efforts to reduce
our DCO % in recent years, it is difficult to see how we could reduce this further. In this case,
the target may be inappropriate and unattainable.

Table 2H - % non-specific morphology codes for microscopically verified cases
Scotland has a slightly higher (2.5%) than expected (2.2%) proportion of non-specific
morphology codes among microscopically verified cases of prostate cancer. We shall
investigate whether there is scope to reduce it (electronic access to the full text of the
pathology report, coupled with nationally agreed standards for completing urological cancer
pathology reports should help).

Table 2I - Mortality : Incidence ratios
Variations in mortality:incidence ratios across the UK may reflect, for example, changes in
the efficiency of ascertainment or elimination of duplicate registrations by some registries
(skewing the UK average), the play of chance, or genuine differences in survival patterns.
The M:I ratio for bladder cancer is particularly dependent on the extent of misclassification
between invasive and non-invasive tumours, both in the cancer registry and in mortality
records. This makes it difficult to compare results between cancer registries.

Table 3A - Completeness of the data set
Ethnicity - Scotland is using the Scottish Census 2001 categories (1A-99) from 1 April 2006,
but the old codes for 1 January 2006 to 31 March 2006. To adhere to the definition, the old
codes found within the 2006 diagnosis year were treated as invalid values in this table.
However, even if any information on ethnicity is accepted, the completeness of recording is
very low. Completeness is dependent on the completeness of recording in medical records
and hospital discharge records (usually very low).        There is already a national effort to
increase the availability and recording of this variable in all health records.
Unique health identifier - Again this is partly dependent on the completeness of recording
of this variable on other health records and there is a national effort to increase the
availability and recording of this variable in all health records.          We have increased
completeness from a very low baseline by a retrospective “seeding” exercise by record
linkage, and by establishing online access to the CHI database in Scotland. In fact our
completeness for this variable is now similar to the UK average.

Welsh Cancer Intelligence and Surveillance Unit (WCISU)

General comment
WCISU continue to validate all malignant neoplasms for the following cancer sites for the
current diagnosis year (2008) to be published in December 2009:
•      Bone Cancer (ICD 10 codes: C40-C41 or morphology 918** to 934**)
•      Brain Cancer (ICD 10 code: C71)
•      Childhood Cancer (ICD 10 codes: all C and D codes, ages 0-14 years)
•      Liver Cancer (ICD 10 code: C22)
•      Mesothelioma (ICD10 code: C45 or morphology 905**)
•      Pancreatic Cancer (ICD 10 code: C25)1
•      Ill-defined, Secondary and Unspecified Cancer Sites (ICD 10 codes: C76-C80)1
       1   All validations to be sent out for those aged under 50 but only those validations with difference
           between anniversary date and death date greater than 14 days to be sent out for those aged 50
           years and over.

Additionally, if a death certificate only registration has been received with a hospital name,
then these cases are also sent out to try and find any previous information relating to the
cancer diagnosis.

WCISU usually send a 10% sample of all malignancies, in-situ neoplasms, benign
neoplasms (bladder, brain, endocrine and central nervous system) and uncertain or
unknown neoplasms excluding non-melanoma skin cancer (ICD 10 code C44) and
carcinoma in-situ of skin (D04) for the diagnosis year 2007 to enable at least 5% of these
registrations to be validated.       However, from this year, for alternating years, a specific
cancer(s) will be chosen of specific relevance. For this year, skin cancers (ICD code C43-
C44) and leukaemia (ICD 10 codes C91-C95) have been chosen.

WCISU took part in the analysis of the Data on Head and Neck Oncology (DAHNO) Audit,
along with Oxford Cancer Intelligence Unit and Trent Cancer Registry for the second year
and it is intended that WCISU will continue to participate in the analysis for the next year.

Table 1
The number of registrations for all malignancies excluding non melanoma skin cancer
continues to increase year on year in Wales due to an increase in the number of sources
used. This is the reason as to why the percentage of initial registrations and ONS ready
registrations in 2007 compared with the period 2004-2006 is over the 2% target at 103.4%.
However, the WCISU are currently investigating if any changes in coding could have
affected the accrual of a large number of registrations in 2007.

Table 2a - % change in registrations (males)
There has been a significant increase in registrations for males in Wales for malignant
melanoma of skin, colorectal cancer, prostate cancer and all malignancies excluding non
melanoma skin cancer and a significant decrease in registrations for bladder cancer. The
WCISU believe that this significant increase for malignant melanoma of the skin is real due
to the general increasing trend for this cancer and is also seen by other registries in the UK.
A similar reasoning can be said for colorectal cancer.          In fact, if 15 less cases were
diagnosed for both cancers, these results would have been non significant. The WCISU are
currently investigating the large increase for prostate cancer which has a knock on effect for
all male malignancies. There has been a large decrease for bladder cancer. This is due to
the WCISU implementing the new coding rule for bladder cancer for those diagnosed in
2007 onwards.     However, the WCISU are investigating possible changes in practice in
coding which may have an effect on the above results.

Table 2b - % change in registrations (females)
There has been a significant decrease in registrations for bladder cancer with the same
reasoning as for males with the new coding rules being implemented. There has been a
significant increase in registrations for breast in-situ, cervix in-situ and colorectal cancer.
However, note that this significant increase is removed if just 1 breast in-situ case was
removed. Similarly, if 12 less cases were diagnosed for colorectal cancer, this result would
have been non-significant. However, since the general trend is increasing, the WCISU feel
that this increase is real. Since all breast in-situ and cervix in-situ registrations in 2007 have
been microscopically verified, we have to assume that this figure for 2007 is correct and that
in the past may not have been complete. The WCISU are investigating possible changes in
practice in coding which may have an effect on the above results. As already stated earlier,
the increased number of sources will also affect the figures for those cancers that are
continually increasing year on year for males and females.

Table 2c – Childhood Cancer Incidence Rates
Childhood cancer incidence rates are higher than that expected for males and females aged
0-4 year and lower than expected for females aged 5-9 years. These are also outside the
“normal” range using data from CI5 Volume VIII. There are four cases in this age band (2

male and 2 females) that have yet to be validated – two of which are English registrations.
One of the cases has since been validated and removed from the WCISU database.

Table 2d - % death certificate only (males and females) “traditional” DCO
The percentage of death certificate only registrations continues to fall in Wales and are
within the target of 2% for the majority of cancers examined. Death certificate only rates are
above the 2% target for lung cancer, colorectal cancer, ill defined sites and all malignancies
(excluding NMSC) for both sexes. Historically, death certificate only rates have been high
for lung cancer. However, comparing this year’s rates with the previous diagnosis year,
nearly all sites have shown a decrease in the DCO rate.             The DCO rates continually
decrease year on year by site in Wales. The DCO rate for ill-defined sites is high at 10.9%
but has improved from last year. This cancer site will inevitably produce high rates.

Table 2f - % zero survival rates
The zero survival rates for ill defined sites are high due to a very high DCO rate for this
cancer site.

Table 2g - % microscopically verified (males and females)
Although this indicator is generally lower than other registries, the trend for this indicator in
Wales is continuing to improve year on year from previous reports. It was found that if the
registrations from English hospitals were excluded, the rates increase but do not still hit the
target required – the rate for haematology is attained if English data is excluded.

Table 2h - % non specific morphology codes that are microscopically verified
Compared to last year, the proportions of those cases that are microscopically verified but
have a non specific morphology has decreased for the majority of cancer sites examined.
The WCISU have since found a problem with some hospitals that have been assigned as
histologically verified when it has not been – this is currently being investigated.

Table 2i – Mortality:Incidence Ratios
M:I ratios are particularly low for prostate cancer and bladder cancer. For prostate cancer,
this appears to be due to the large increase in incidence of prostate cancer in 2007
compared to previous years which will then have a “knock on” effect for all male
malignancies excluding non melanoma skin cancer. The WCISU have recently adopted the
new coding rules for coding bladder cancer. However, this can only be done if the source of
data is pathology.    Those cases that only have a source of registration via the Patient
Episode Database for Wales (PEDW) cannot be coded this way. The WCISU are currently
investigating how this can be rectified.

Table 3a – Completeness of the dataset – demographics and diagnostic details
Known ethnicity is low at 31.9%, but is higher than that submitted last year. WCISU are
currently examining how this measure can be improved. Similarly, WCISU improve year on
year with the percentage known for basis of diagnosis when looking back at past reports and
are now just 1.6% off the target. This target is passed if data from English hospitals were

Table 3b – Completeness of the dataset – treatment
The number of those patients having therapeutic surgery is slightly higher than in past
reports at 56.0% but other measures are in line with that expected and appear in line with
other registries results. The results for the new indicator “Any treatment” appear consistent
with other registries.

Table 3c – Completeness of the dataset – specific staging information including grade
The percentage of those with known Bloom and Richardson grade continues to improve year
on year and has now passed the target of 78% for the first time. Nodes positive, size and
NPI score were collected at the WCISU for breast cancer with a diagnosis date of 2008
onwards. However, a small number of cases diagnosed in 2007 have this information which
is seen in the results for this years report.

Staging information for the cancer sites examined are particularly lower than the targets
since WCISU usually stage via TNM. Dukes (colorectal cancers), FIGO (cervical cancer)
and Clarkes and Breslow (melanoma) are only recorded when documented, hence the low
percentages. WCISU are working towards collecting the specific staging information with
new sources as opposed to just TNM staging. It is hoped that these staging figures will
improve in next year’s report.

Northern Ireland Cancer Registry (NICR)

General Comment
The overall cancer incidence has risen in Northern Ireland in 2007. In addition to an ageing
population, we have noticed an increased detection in certain cancers due to screening

Tables 1 & 2 Ascertainment
Stability of incidence
The rise in incidence is most notable for prostate cancer which has continued to rise so that
we have over twice the levels of this cancer we had in the early 1990’s. We believe this is
entirely due to increased use of PSA testing.
This year has also shown a rise in colorectal cancer which is significantly higher in males.
This matches a rise we have noticed in early colorectal tumours and in benign tumours. Note
that colorectal cancer screening has not yet been introduced to Northern Ireland but is
planned in the near future.
The rise in breast cancers cannot be readily explained except by the ageing population plus
underlying risk factor changes e.g. obesity, or HRT use. Breast screening increase is
unlikely to be the cause, as all ages have shown an increase. It is hoped that both the BASO
and BCCOM projects will enable comparison with other data to confirm the numbers.
The instability in bladder cancer for females is probably explained by the small numbers

Childhood Cancers
The childhood incidence rates for boys in the age group 0-4 have risen sharply. We have
rechecked all 22 cases and they correspond with the data from the CCRG. Since the return
was submitted, we found subsequent information on one boy which showed that he did not
have cancer. The rate has now fallen to 35 which is still very high. This may reflect year-on-
year fluctuation, but we will continue to monitor the situation.

Other Ascertainment Measures
DCO rates have continued to fall to under 1% for both males and females. Most cancer sites
have shown a decrease except for lung which shows a small increase and ill defined sites
which has risen considerably. However the latter is based on only 12 cases and is not
considered a cause for concern.

Microscopic Verification rates have continued to rise with most sites over 90% with
microscopic verification. The exception to this is lung cancer which continues to be around
70%. We are fairly confident that we are receiving the majority of the histology and cytology
for these patients. The non-specificity of morphological types was increased this year as
laboratories did not send us the full written cytology reports. This can to be checked when
we are able to process data from the MDT databases.

Table 2i Mortality : Incidence Ratios
Most of these are within acceptable limits except the cervix is much lower than expected.
However since numbers for both mortality and incidence are small, wide variation in this ratio
is expected.

Table 3a: Completeness of the dataset – demographics and diagnostic details
Demographic and Diagnosis details have improved from last years return. The increasing
use of the Health and Social Care Number has improved greatly which has also improved
our patient matching. Post codes for some areas of Northern Ireland are rarely used as so it
is unlikely that Post code coverage will be 100% in the near future. Type of growth is also
relatively high which is reflected in the non-specificity of morphologies in some sites as noted
in comments on Table 2.         Ethnicity remains an issue but at present there seems no
significant progress on the hospitals collecting this information.

Table 3b: Completeness of the dataset – treatment and screening information
This has remained fairly stable from previous years. However we did not receive any
hormone treatment information for 2007. It is hoped that the MDT databases may help in
collecting this. It is difficult to interpret the treatment data for specific cancer sites for trends,
but it does seem that the older age groups receive less treatment. We have not been able to
explain the low levels of treatment for elderly prostate patients. This may reflect our coding
of treatments and we are currently investigating the reasons for this by comparisons with
audit data collected on 2006 patients.

We have screening categories for cervical cancer for 2006, but it does not formally indicate
“screen detected” cases. It is hoped that we will be able to get this for next and subsequent
years as a new system has been brought in. Breast screening categories have yet to be
assigned by our QARC for 2006 data.

Table 3c: Completeness of the dataset – specific staging information including grade
Improvements have been made in the collection of Breast and Colorectal staging information
since last year. Cervical cancer and melanomas remain fairly high with over 90% staged.

                      UKACR Quality and Performance Indicators 2009



1.    Years
      Data should be provided for 2007 for tables 1 to 3 (except Scotland who are submitting 2006
      data for tables 1, 2 and 3).

2.    Tumours
      Registrations should include all malignant tumours, excluding non-melanoma skin cancer (All
      ICD 10: C00 - C97       Excluding C44

      Breast and cervix in situ (ICD10: D05, D06) have been added separately to Table 2 only, but
      should not be counted in the “All sites” total.

3.    Population
      The population covered should be the residents of the registry’s catchment area only as
      defined by the Cancer Postcode Directory. (The list of Lower Super Output Areas (LSOAs)
      can be found on the UKACR website in the member’s section under
      projects/shared resources/postcodes/Cancer Postcode Directory April 2009) using the April
      2009 file.       The population figures should be taken from projects/shared
      resources/Populations/Version 5.4/2007 LSOA estimates.


4.    The ratio of registered (malignant tumours all xnmsc) to expected cases for the year 2007
      should be measured at 30/06/2009, where the expected number of cases is estimated using
      the average number of registrations for the years 2004-2006.

      The number of registrations in each year 1998-2006 should also be provided.

      Some registries may wish to distinguish between “Initial registrations” and registrations that
      are “ONS ready”. The term “ONS ready” means that the registrations are of a standard
      usable for analysis and in publications. They are of a standard to send to ONS - although
      they do not actually have to have been sent to ONS. This does not preclude registrations
      with missing fields and ones where more data may be received (e.g. on treatment) at a later


      The following ICD10 codes should be used: “Lung” (trachea, bronchus & lung: C33-C34),
      invasive breast (C50 - females only), in situ breast (D05 - females only), invasive cervix
      (C53), in situ cervix (D06), melanoma of the skin (C43), colorectal (C18-C20), prostate (C61),
      bladder (C67), haematology (C81–C96), ill-defined sites (C76-C80), all sites (C00-C97,
      excluding C44) for the diagnosis year 2007 (except Scotland) unless otherwise stated.

      For the all sites calculations, these should be broken down by sex and age at diagnosis. The
      denominator should include all malignant tumours xnmsc for the particular cancer site(s) of

5.       Stability of incidence
         This measure is defined here as the % increase in the number of registrations this year
         compared with the average of the previous 3 years, as currently on the database as at

               Number of registrations in current year  − 1 *100
                                                           
              Number of registrations in previous 3 years 3  
                                                              

         Alternative approach to determining if stability of incidence figures are significant (as opposed
         to 2% and 5% cut off points) – the method to use will be discussed at the Exec Committee on
         June 26 :

         We replace the system of fixed thresholds with one based on 95% confidence intervals.
         These increase in size as the number of cases used to calculate them shrinks. We take
         account of the extra variation caused by the estimate of the expected number of cases based
         on three previous years of data .

         Define n1, n2, n3, n4 as the number of cancers of interest in four consecutive years, with year 4
         being the year of the current PI, and years 1 to 3 being the previous three years.

         Define a mean number of cases in years 1 to 3, nm, as                          nm = (n1 + n2 +n3)/3

         The standard error of nm is σm with                                            σm = sqrt(nm/3)

         and the standard error of n4 is σ4 with                                        σ4 = sqrt(n4)
                                                                                                                  2     2
         Use a 2-sided z-test, which gives a z-statistic z such that                    z = (nm – n4) / sqrt(σm + σ4 )

         We find that the absolute difference between nm and n4, i.e., abs(nm - n4), is significant at the
         95% level if
                                 abs(nm - n4) > 1.96 × sqrt(nm/3 + n4)

         If this is true (i.e., there is a significant change) then most recent year’s data is significantly
         high if n4>nm and significantly low if n4<nm.

         The total number of registrations for each type of cancer for the current year of analysis is
         required along with the three previous years to determine whether there has been a
         significant increase (or decrease) in the stability of incidence (Excel will automatically
         determine this) after the four data items have been inserted for each sex. The figure for
         stability of incidence will also be automatically calculated following the insertion of each of the
         figures for the four years of diagnosis.

6.       Childhood incidence rates
         Age-specific incidence rates in the age groups 0-4 years, 5-9 years and 10-14 years should
         be used.

         All malignant tumours xnmsc should be included.

         The rates should be calculated using mid year 2007 LSOA derived populations as the
         denominator (also 2007 for Scotland).

  This method does not take account of trends in cancers (e.g. the steep fall in male lung cancer incidence, or the rise in
melanoma incidence). While it would be technically possible to do this it would add complication to the methodology but more
importantly would disguise real and notable changes in cancer incidence.

7. 1   “DCO” rates
       These should exclude GP only (GPO) and Post Mortem only (PMO) cases. If this is not
       possible then please specify what is included in the calculation. (or see Library of
       Recommendations, Po/03/03.)

       These traditional “DCO” rates will also be calculated for the previous year of diagnosis to
       determine whether those registries that had not completed follow up at the time of the
       previous report were able to find other sources of information for the outstanding “Death
       Initiated” cases.

7.2    Zero survival rates
       This should include any cases where the date of diagnosis = date of death.

8.     Microscopic verification
       Registrations should be considered microscopically verified where diagnosis is based on a
       malignant histology report for either the primary or of metastases of the primary tumour,
       cytology report, bone marrow report or blood report. (N.B. The registry may not have seen, or
       possess the report). Using the national codes from the web link below, these relate to codes

9.1    Specificity of morphology: the proportion with a non-specific morphology code for
       those cases which have been microscopically verified
       The denominator should include all tumours which have been microscopically verified (point
       8) for the relevant cancer site(s) of interest.

       The number of cases whose morphology is M8000, M8001 or M8010 should form the
       numerator for this measure for the relevant cancer site of interest, except for haematology.

       For haematology, the non-specific morphologies counted in the numerator should be either
       M9800 or M9590.

       For the “All xnmsc” total, the denominator should include only all xnmsc cases that have been
       microscopically verified (i.e. not including the in situ cases). The non-specific morphologies
       counted in the numerator should be M8000, M8001, M8010, M9800 and M9590.

9.2    (OPTIONAL EXTRA: please note that several registries have requested that the additional
       internal check for looking at the proportion of cases with a specific morphology code which
       have not been microscopically verified, also be included in the template to allow for internal
       QA checks. This has therefore been left in table 2 in the template, and if you wish to complete
       the fields, please do so using the following definitions:

       The denominator should include all tumours which have NOT been microscopically verified
       (point 8) for the relevant cancer site(s) of interest.

       The number of cases whose morphology is NOT one of M8000, M8001 or M8010 should form
       the numerator for this measure for the relevant cancer site of interest, except for

       For haematology, the specific morphologies counted in the numerator should be those whose
       code is not one of M9800 or M9590.

      For the ‘All xnmsc’ total, the denominator should include only all xnmsc cases that have not
      been microscopically verified. The specific morphologies counted in the numerator should be
      those cases whose codes are not one of M8000, M8001, M8010, M9800 and M9590.

10.   Mortality to incidence ratios

      M:I = deaths in specified period, with diagnosis C as underlying
                      incident cases in period with diagnosis C

      For England and Wales, mortality data supplied for CIS by Thames via ONS for the registered
      deaths should be used for the numerator, not deaths as recorded by the registry.

      Equivalent data of the numbers of deaths should be provided by Northern Ireland and
      Scotland. Mortality data for 2007 should be used by all registries to remain consistent.


11.   Database check - Table 3a - demographics and diagnostic details:

      All database records for         all   malignant   tumours   xnmsc   should   be   checked for
      completeness and validity:

      % Blank                 The record contains no entry for that data item.

      % Invalid               The record contains a code for that data item, but it is not a valid

      % Valid Unknown         The record contains a valid code for that data item, but the code
                              means that data item unknown.

      % Valid Known           The record contains a valid code for that data item, and the data item
                              takes a known value.

      For each row of the database the four columns should sum to 100%.

      Name, Address, Postcode, Date of Birth, Anniversary (diagnosis) date, Behaviour of Growth:
      There should be no Valid Unknown cases for these data items – please identify what they are
      if you have some.

      Sex: All options are possible.

      Unique Health Identifier (NHS Number for England and Wales; Community Health Index
      (CHI) for Scotland and Unique Health and Social Care number for Northern Ireland): There
      should be no Valid Unknown cases for these data items – please identify what they are if you
      have some.

      Ethnicity: For English registries, only the standard code set as defined by the relevant DSCN
      issued in 2000 (212000.pdf) and supplemented in 2001 (022001.pdf), both found at in line with the NCDS
      should be counted as a Valid Known code; except for code ‘Z’ or ‘99’ = not stated which
      should be counted as Valid Unknown. Thus, only cases with either one of the 16 other single
      letter codes, or one of the 89 other double letter codes should be counted as being a Valid
      Known code. Similar definitions for the non-English registries should be used based on the
      2001 Census in each of these countries.

      Any other ethnicity systems being used which do not conform to these codes, should be
      described as being Invalid.

      Site of primary: Valid Unknown is defined as C76-C80.

      Date of death: These should be calculated using the total number recorded as dead as the
      denominator. There should be no Valid Unknown cases for this data item – please identify
      what they are if you have some.

      Type of growth: Valid Unknown is defined as M8000, M8001, M9800, M9590 or M8010 with
      no microscopic verification (see point 8).

      Basis of diagnosis: All options are possible. Valid unknown is defined as “9” (using national
      codes) from the basis of diagnosis codes from point 8.

12.   Database check - Table 3b – treatment and screening information:

      All database records for all malignant tumours xnmsc should be checked for completeness
      and validity, UNLESS a different denominator is stated for the particular data field:

      % Blank                 The record contains no entry for that data item.

      % Invalid               The record contains a code for that data item, but it is not a valid

      % Valid Other           The record contains a valid code for that data item, but the code
                              means that data item is either NO or Unknown.

      % Valid “Yes”           The record contains a valid code for that data item, and the data item
                              takes a known value of YES.

      For each row of the database the four columns should sum to 100%.

      For all four treatment variables, the treatment information should be as defined by ONS and
      relate to the first 6 months from diagnosis only (include short negative diagnosis to treatment
      days from -1 days up to -90 days only and exclude any over -90 days). If it is known that the
      patient did not have any treatment then this should be included in Valid Other rather than
      Valid “Yes”.

      % Surgery: Valid Known should be the % who received surgery as a proportion of all xnmsc
      registrations. If it is known that the patient did not have any surgery, then these cases should
      be included in Valid Other rather than Valid “Yes”.

      % Radiotherapy: Valid Known should be the % who received radiotherapy of any intent as a
      proportion of all xnmsc registrations. If it is known that the patient did not have any
      radiotherapy, then this should be included in Valid Other rather than Valid “Yes”.

      % Chemotherapy: Valid Known should be the % who received chemotherapy of any intent as
      a proportion of all xnmsc registrations. If it is known that the patient did not have any
      chemotherapy, then this should be included in Valid Other rather than Valid “Yes”.

      % Hormone treatment: The denominator for these cases should be either only female
      breast cancers or only prostate cancers, rather than all xnmsc      registrations (and

      these should be reported separately). Valid “Yes” should be the % who received hormone
      therapy of any intent as a proportion of either all female breast cancer or prostate
      registrations, respectively. If it is known that the patient did not have any hormone therapy,
      then this should be included in Valid Other rather than Valid “Yes”.

      % Any treatment: The denominator should be the total number of site specific cancers for the
      specified cancer site. For all malignancies excluding non melanoma skin cancer, ages 0-24,
      25-64 and 65 years and over should be analysed. For breast cancer, prostate cancer and
      colorectal cancer, only ages 0-64 and 65 years and over should be analysed. Valid “Yes”
      should be the % who received any surgery, chemotherapy, radiotherapy, hormonal treatment,
      palliative treatment or “Watch and Wait”. If a registry is able to report that the patient received
      no treatment then this should be reported in Valid Other. For Scotland, their “Other Therapy”
      category on SOCRATES does not correspond with other treatment with regards to the
      definition required here, hence they should exclude from their analysis.

      Only data provided directly by the QARCs should be reported on and for 2006 cases NOT
      2007 (also 2006 data for Scotland).

      Screening status for invasive female breast tumours:
      For the age-group, 50 to 67 inclusive at diagnosis (upper age limit + 3 years to allow for
      possible diagnosis of interval cancer – we have chosen to stick with 67 as the upper limit
      because not all screening programmes will have rolled out the age extension yet), the two
      following measures of breast cancer are required:

      % Screen-detected of the valid age group: Only breast cancer cases flagged as being screen
      detected should be counted as Valid “Yes”.

      % of cancers with a full screening category within the valid age group:. The list below
      identifies the Valid “Yes” and Valid Other categories (derived from the NCDS). Null values
      are counted as Blank and any other values are Invalid.

      Screen Detected - valid “Yes”
      Interval Cancer - valid “Yes”
      Other (eg lapsed attended) - valid “Yes”
      Not known - valid “Other”
      Uncategorised - Invalid

      Screening status for invasive cervical tumours:
      For the screening age-group, 25 to 67 inclusive at diagnosis (upper age limit + 3 years to
      allow for possible diagnosis of interval cancer), the following measure should be reported:

      % Screen-detected of the valid age group: Only cervical cancer cases flagged as being
      screen detected should be counted as Valid “Yes”.

14.   Database check - Table 3c - specific staging information including grade:
      All database records for the particular cancer sites of interest should be checked for
      completeness and validity against these specified staging systems for the particular cancer
      sites of interest:

      % Blank                  The record contains no entry for that data item.

      % Invalid                The record contains a code for that data item, but it is
                               not a valid code.

% Valid Unknown        The record contains a valid code for that data item, but the code
                       means that data item unknown. Where a tumour is unstageable this
                       should count as Valid Unknown and not staged.

% Other Known          The record contains a valid code for that data item which belongs to
                       a different staging system to that specified for the particular cancer
                       site of interest. Use this if you are not sure which staging system is
                       being used; ie you do not know that the specific staging system for
                       the cancer site is definitely being reported.

% Valid Known          The record contains a valid code for that data item, and the data item
                       takes a known value for the specified staging system for the specified
                       cancer site of interest. You must only report this (rather than ‘Other
                       Known’) if you definitely know that the stage has been reported using
                       this staging system excluding cervical cancer and colorectal cancer.

For each row of the database the five columns should sum to 100% (as you will either have
included the case as having a valid known (if using specified staging   system) or Other
Known (if you are not sure whether the staging system is the specified one; or you know that
it is definitely not the specified staging system).

The following specific measures should be reported for female breast cancer, colorectal
cancer, cervical cancer and melanoma of the skin:

Female Breast cancer - % with known Bloom and Richardson grade: All invasive female
breast cancer cases should be examined. Of these only those with known Bloom and
Richardson grade (i.e. G1-G3) should be counted as Valid Known.; G4 should be included as
being as Invalid Known (so put in the invalid column); GX should be included as being Valid

For any cases where it is not known whether or not the grade is according to the modified
Bloom & Richardson histological grading system should be reported as being Other Known
rather than Valid Known.

Female Breast cancer - % with known number of positive nodes: All invasive female breast
cancer cases should be examined. Of these only those cases who are known to have had
nodes removed and for whom it is known how many of these were positive (ie a known
number of positive nodes) should be counted as Valid Known. Those cases where the
number of positive nodes is zero should be included as Valid Known as long as the women
definitely had nodes removed. Cases which are known to be node positive, but the exact
number of nodes is unknown should be counted Valid Unknown.

Any cases using TNM or other staging systems, or where the staging system is unknown,
should be recorded as Other Known rather than Valid Known.

Female Breast cancer - % with known invasive size: All invasive female breast cancer cases
should be examined. Of these only those with the known exact invasive size should be
counted as Valid Known.

Any cases using TNM or other staging systems, or where the staging system is unknown,
should be recorded as Other Known rather than Valid Known.

Female Breast cancer - % with known NPI score: All invasive female breast cancer cases
should be examined. Of these only those with a known NPI score should be counted as Valid
Known. However, if the three separate components are known then these cases can be

included as having a valid known NPI score because it is acceptable to apply the calculation
used to derive the NPI score from the three separate components. This calculation is:

NPI score = Histological Grade + Nodes + 0.2 x Invasive Size of Tumour (in cm)

        Grade takes values 1, 2 or 3 (using Bloom and Richardson grade as defined
        Nodes = 1 if zero nodes are positive
              = 2 if 1–3 nodes are positive
              = 3 if >=4 nodes are positive

Any cases using TNM or other staging systems, or where the staging system is unknown,
should be recorded as Other Known rather than Valid Known.

Colorectal cancer - % with known pathological Dukes stage: - All invasive colorectal cancer
cases should be examined. Cases with a known pathological Dukes stage, or one derivable
from a TNM stage according to the conversion table in Appendix 1, should be counted as
Valid Known; except cases known to have metastases at diagnosis can be taken as being
Dukes Stage D and hence have a Valid Known code.

Any cases using other staging systems, or where the staging system is unknown, should be
recorded as Other Known rather than Valid Known.

Cervical cancer - % with known Complete FIGO stage: - All invasive cervical cancer cases
should be examined. Of these only those with a known complete FIGO stage, or known pT
stage, should be counted as Valid Known.

Any cases using other staging systems, or where the staging system is unknown, should be
recorded as Other Known rather than Valid Known.

Melanoma skin cancer - % with known Breslow thickness: - All invasive melanoma skin
cancer cases should be examined. Of these only those with a known Breslow thickness
should be counted as Valid Known.

Any cases using TNM or other staging systems, or where the staging system is unknown,
should be recorded as Other Known rather than Valid Known.

Melanoma skin cancer - % with known Clarke level: - All invasive melanoma skin cancer
cases should be examined. Of these only those with a known Clarke level should be counted
as Valid Known.

Any cases using TNM or other staging systems, or where the staging system is unknown,
should be recorded as Other Known rather than Valid Known.


     This report will need to cover the reasons for any measures not met, either against peer
     review targets, or against target/expected values for non-peer review measures as detailed in
     the separate measurestemplate2009.xls file.

     Collated results will be sent to each registry using the new UKACR measures template. The
     UKACR Task and Finish Group will determine whether a comment should be made by a
     registry regarding the figures for a specific measure. If any measure for a particular registry is
     below the Peer Review, expected or target value, or is “not in line” with other registries then
     the registry should assess whether this is a genuine shortfall. All genuine shortfalls must be
     commented upon. Note that a commentary should be made for the stability of incidence if all
     malignancies excluding non melanoma skin cancer are NOT within 2% of the average of the
     three previous years and if all other site specific cancers are NOT within 5% of the average of
     the three previous years.

     You should also include the information you supplied in tables 3d and 3e in the commentary.

     Details of progress made since last year, in particular with registration processes and data
     quality can also be included in the report.


     TNM profile can be converted to Dukes Staging by using the table below. The “T”, “N” and “M”
     components refer to any pathological (e.g. pT1pN0pM0) or clinical (e.g. cT2cN0cM0) certainty

       TNM stage Profile            Summary Stage                  Dukes Stage
       T1N0M0                       I                              A
       T3N0M0                       II                             B
       Any T N1M0                   III                            C*
       Any T N2M0
       Any T Any N M1               IV                             D

     * Dukes C can be further divided into two categories - C1 where the apical node is NOT
     involved and C2 where the apical node is involved. However this is not normally part of the
     TNM staging and cannot be derived from the TNM profile.

     “X” in the TNM profile

     In the Case of “MX”, it can be assumed in the absence of any other data that the clinician has
     decided that the patient has no distant metastases i.e. “cM0”

       TNM stage Profile            Summary Stage                  Dukes Stage
       T1N0MX                       I                              A
       T3N0MX                       II                             B
       Any T N1MX                   III                            C
       Any T N2MX
       TXN0M0                       II                             B
       TXN1M0                       III                            C


To top