Cancer National Specialist Advisory Group. Welsh Breast Cancer Clinical Audit for Patients Diagnosed 2008

Transcription

1 Cancer National Specialist Advisory Group Welsh Breast Cancer Clinical Audit for Patients Diagnosed 2008 Published 2012

2 Cancer NSA Welsh Breast Cancer Clinical Audit: patients diagnosed in 2008 Foreword Last year we carried out a pilot audit of all breast cancers recorded on Canisc as diagnosed in 2007 to ascertain what information could be extracted, and how that would help to identify possible variations in treatment across the various breast units in Wales. The data presented in this audit are for patients diagnosed in We have taken the opportunity to make some comparisons with the CSCG Breast Cancer Audit of patients diagnosed in Wales in With advance advice having been given that this audit would be used to make genuine comparisons of practice across breast units in Wales, it was hoped that units would have put systems in place to ensure accuracy and ownership of the data being collected. Following on from the successful NHS Breast Screening Programme (NHSBSP) UK audits of screen detected breast cancers over a number of years in which Wales has shown consistent leadership in the standards that could be achieved, it was anticipated that this comprehensive audit of screen detected and symptomatic breast cancers would provide benchmarks for breast cancer management in general. Enormous energy has been involved in developing, maintaining and interrogating this database and the efforts of all those involved at all Wales and local levels is highly recognised and commended. As the results show, however, individual units and the local health boards who have come into existence since the time of this audit period will need to ensure that appropriate resources are made available to ensure accurate data entry into Canisc. Otherwise unit performance may appear in detrimental light. It is important to recognise that the management of breast cancer evolves and new techniques and treatments often need to be introduced progressively so that some degree of variation in practice is both inevitable and healthy. However this audit demonstrates variations that need to be locally investigated. A multidisciplinary audit committee reporting to the Cancer NSAG Breast Subgroup is being formed to expand the next audit cycle into new areas, which is an exciting development. This will present a more comprehensive picture of the patient pathway. In addition to tracking developments over time, it will be responsive to changing clinical practice, providing a valuable resource to clinicians and organisations in achieving and maintaining best practice and outcomes. Ian Monypenny Audit Lead for Breast Cancer Sub-group of the Cancer NSAG, Consultant Breast Surgeon, University Hospital of Wales, Clinical Audit Lead for Breast Cancer for Wales 2 P a g e

3 Cancer NSA Welsh Breast Cancer Clinical Audit: patients diagnosed in 2008 Acknowledgments This audit was designed and produced on behalf of the Cancer National Specialist Advisory Group (NSAG) by: Ian Monypenny Rebecca Thomas Suzanne Jenkins Consultant Breast Surgeon, University Hospital of Wales, Audit Lead for Breast Cancer Sub-group of the Cancer NSAG Senior Statistician, Welsh Cancer Intelligence and Surveillance Unit Lead Breast Cancer Information Specialist for Wales Additional data analysis and data quality checks carried out by; Julie Howe Statistician, Welsh Cancer Intelligence and Surveillance Unit Editorial input and approval for publication Breast Cancer Sub-group of the Cancer NSAG Cancer NSAG Core Team. 3 P a g e

4 Cancer NSA Welsh Breast Cancer Clinical Audit: patients diagnosed in 2008 Table of Contents Executive Summary... 1 Recommendations... 2 Introduction... 3 Aims and Objectives... 3 Audit Organisation... 3 Audit Process... 4 WCISU Case Matching... 4 Overall Summary of data extracted from Canisc... 5 Male Breast Cases... 9 Invasive Female Breast Cancers Non-Invasive Cases Breast Cancer Clinical Outcome Measures Outcome Measure 1: Number and percentage of breast cancers for which complete information is recorded Outcome Measure 2: Number of symptomatic and screen detected cancers treated in a breast unit per annum Outcome Measure 3: Number and percentage of symptomatic cases with a non-operative histological/cytological diagnosis (KCI 1) Outcome Measure 4 & 5: Number and percentage of invasive cases having primary medical or surgical treatment Outcome Measure 6: Mastectomy rate by whole size of tumour Outcome Measure 7: Number and percentage of surgically treated invasive cancers with known histological nodal status Outcome Measure 8: Number and percentage of histologically node negative invasive breast cancers for which less than 7 nodes were removed (KCI 2) Outcome Measure 9: Number and percentage of invasive cancers treated by conservation surgery and receiving adjuvant radiotherapy (KCI 5) Outcome Measure 10: Number and percentage of node positive patients with invasive breast cancer, aged 60 or under receiving adjuvant chemotherapy (KCI 4) Outcome Measure 11: Number and percentage of ER positive patients with invasive breast cancer receiving adjuvant hormone therapy (KCI 3) New Outcome Measure: Number and percentage of patients having breast conservation surgery with margins of excision free of tumour Survival Summary and Next Steps Appendix 1- Case matching Appendix 2- Breast Cancer MDT Lists P a g e

5 of patients diagnosed in 2008 Executive Summary This audit describes the patterns of presentation, diagnosis and treatment of patients diagnosed with breast cancer in Wales in 2008 using data extracted from Canisc, designated as the central repository of breast cancer data across Wales. It follows on from the pilot audit of patients diagnosed in Data on 2617 patients (symptomatic 62% and screen-detected 38%) were identified for analysis from 14 hospitals where the first definitive treatment was carried out. The mean number of cases was 187 (range ). Comparison was made with breast cancer registrations held by the Welsh Cancer Intelligence and Surveillance Unit (WCISU) and showed that almost 100% of Canisc cases were known to WCISU although there were variations in key items of information such as date of diagnosis and morphology code. WCISU had a further 440 cases that were not registered on Canisc within the defined range of date of diagnosis and the reasons for this have been evaluated further. Although overall data quality in the 2007 audit had been reasonable it was hoped that marked improvements would be seen in the current audit to give more certainty about any apparent differences in patient management. Unfortunately this has not been the case despite excellent work being carried out with regular feedback reports to breast units about data quality. Although half of the breast units have made modest progress, in many cases it is clear this is not being acted upon at unit level. Even simple data about preoperative diagnosis for symptomatic patients appears to be missing in a considerable proportion of patients. The capture of oncological treatment data is noticeably poor in north and southwest Wales. Where patient management by MDTs (multidisciplinary teams) is split across sites for surgical and subsequent oncological treatment, a pathway of responsibility for accurate and timely data entry to Canisc needs to be defined and maintained. Key Clinical Indicators (KCI) have been agreed by the Cancer National Specialist Advisory Group (NSAG) and are refinements of those produced by the UK Breast Cancer Clinical Outcome Measures project (BCCOM). These show variation between breast units of 1 : 1% to 89% (mean =70%) of patients having a preoperative diagnosis for invasive breast cancer (Figure 18). 45% to 81% (mean =61%) of patients aged 70+ have surgery as primary treatment for invasive breast cancer; the remainder may or may not have had primary drug therapy (Figure 21). 22% to 89% (mean = 41%) of all patients having mastectomy as primary treatment for invasive breast cancer (12%-89% by individual surgeon, Figure 22) 18% to 100% (mean = 82%) of patients who had breast conserving surgery for invasive breast cancer with known clear excision margins (Figure 37). 0% to 94% (mean = 22%) of patients having more extensive (>7 nodes removed) axillary surgery for histologically node negative invasive breast cancer (Figure 31). Although some of this variation relates to the training phase of the introduction of sentinel node biopsy, it needs watching in the future. 67% to 100% (mean = 90%) of patients receiving radiotherapy after breast conserving surgery for invasive breast cancer (Figure 33). 25% to 100% (mean = 77%) of node positive patients aged </=60 with invasive breast cancer receiving adjuvant chemotherapy (Figure 35). 0% to 87% (mean = 38%) of ER+ve patients with invasive breast cancer receiving adjuvant hormone therapy (Figure 36). 1 Not all outcome indicators and Key Clinical Indicators are included in this list 1 P a g e

6 of patients diagnosed in 2008 One year relative survival of the 2007 and 2008 cohorts treated surgically are 99.78% and 99.55% respectively. Of those not treated by primary surgery, in most cases being those with late stage at presentation or significant co-morbidity, the estimates are 83.76% and 84.78% respectively. The data suggest that older patients with breast cancer may be being undertreated, or their treatment under-recorded and an in-depth audit of the underlying reasons should be carried out. While missing data will account for some of these variations and small numbers may be relevant in some cases as well, it is too easy to hide behind this mantle. This audit contains a great deal of valuable information about possible variation in practice and a feedback mechanism to complete the audit loop is necessary. This has existed for many years in breast screening through the quality assurance programme and a similar system is needed for symptomatic breast cancers as well. It is hoped that this will now come through the National Clinical Audit Advisory Committee in the Welsh Government. Recommendations It is recommended that LHBs: Review their performance against each outcome measure and set internal targets to achieve in the future, including: o o Addressing variation from best practice Assessing, and if necessary addressing, variation from the average identified within this audit report Audit treatment rates for older patients. Implement a system of exception reporting for patients who do not receive adjuvant radiotherapy following local excision. Review the appropriateness of advice given to patients on adjuvant therapies in the absence of data to calculate the NPI. Nominate clinical audit/information leads within each MDT to support the MDT lead. Ensure that complete and accurate clinical data is entered onto Canisc by each Breast MDT including; o o o Needle core biopsy recording Full data on all outcome measures Treatments given It is recommended that the Breast subgroup of the Cancer NSAG repeat this audit for patients diagnosed in 2009 and 2010 to give larger patient numbers for subset analysis and to identify time trends in changes in patient management. This will advise LHBs on their progress and it will allow proposal of new All Wales outcome indicators. 2 P a g e

7 of patients diagnosed in 2008 Introduction Aims and Objectives Canisc is an all Wales electronic patient record used for the clinical management of cancer patients. It allows any number of organisations involved in a patient s care to record clinical information into a common patient case note which can then be accessed by any authorised health care professional. Canisc can provide a full picture of each patient s care whenever and wherever required. As Canisc is the nominated data repository for cancer data on an All Wales basis, this audit has been carried out to evaluate the data on new breast cancers that can be extracted from the system. A pilot audit was carried out on 2007 data, the aim of this being to assess the quantity and quality of data collected. As a result of lessons learnt, improvements were made to try to obtain better data validation before extracting the data for The monitoring of treatment annually has long been part of the Quality Assurance programme of Breast Test Wales and the NHSBSP/ABS 2 UK audits. It has helped identify areas where practice needs to be improved to achieve target standards. These audits however only look at screendetected breast cancers and data collection was funded from the out set of screening. Although efforts have been made through similar UK audits for symptomatic patients, obtaining data on over 50% of patients has historically been difficult to achieve. There have also been wide variations in the quantity and quality of data collected in this way, making comparisons between units difficult. It is hoped that a more consistent approach to data collection across Wales using a single system and regular data validation will allow better audit, thus helping to identify areas for improvement. This audit is primarily concerned with clinical management outcomes as opposed to waiting times and issues surrounding these. A set of clinical outcome measures for symptomatic breast cancers were published by the BCCOM group and have been adopted with some modifications as the key clinical indicators for breast cancer by the breast cancer sub-group of the Cancer NSAG. They are not totally comprehensive but can attempt to benchmark individual units. These outcome measures have been adapted in this audit to allow similar comparisons with screen-detected cancers. Audit Organisation The audit has been carried out by a small project group who have: Utilised the existing data analysis tools within Canisc, designed for this purpose, to extract the appropriate data for analysis. Organised the download and distribution of data to individual Health Boards for validation, including providing summaries of missing or incomplete data, plus liaising with clinical teams and cancer networks to improve data quantity and quality.. Carried out a detailed analysis of the final dataset. Developed this report to inform clinicians and Health Boards. The report and its contents are primarily the property of the Breast sub-group of the Cancer NSAG and will be published on the Cancer NSAG web site. 2 National Health Service Breast Screening Programme/Association of Breast Surgery 3 P a g e

8 of patients diagnosed in 2008 Audit Process All patients with a new diagnosis of breast cancer (screening and symptomatic) with a date of diagnosis between 01/01/2008 and 31/12/2008 were included in the audit cohort Initial WBCCA (Welsh Breast Cancer Clinical Audit) extract taken from Canisc by Lead Breast Information Specialist. Audit information pack and generated patient data summary sent to each Breast MDT Coordinator at each LHB/Trust for validation. Validation period: Local Health Board (LHB)/Trust carried out verification with clinical team adding missing information to patient records where appropriate. WBCCA data extracted from Canisc by the Data Quality Manager and submitted to audit team. 4. Detailed data analysis performed by audit team. 5. Statistical analysis carried out by WCISU statisticians in consultation with the audit team. 6. Iterative draft report preparation 7. Finalised WBCCA Report signed off by Breast subgroup of the Cancer NSAG for publication WCISU Case Matching As a data quality exercise, the breast cancer cases from Canisc were matched and compared to cases held by WCISU. This was carried out in two stages; the first stage was to match WCISU cases to the Canisc breast cancer cases; the second stage was to identify all 2008 breast cancer cases from the WCISU database and match the Canisc cases to them. The methodology for this and results are detailed in appendix 1. 4 P a g e

9 Overall Summary of data extracted from Canisc Figure 1 : Flow chart of breast cancer cases diagnosed in 2008 * No breast surgery (or none recorded) 5 P a g e

10 Table 1 : Overview of screen detected and symptomatic cases for each hospital Note: Throughout the report surgical treatment refers to breast surgery (conservation or mastectomy). It should be noted that some of the hospital units are part of a wider MDT as detailed in Appendix 2- Breast Cancer MDT Lists. In the 1997 audit 1990 patients with breast cancer were identified, 21% having been detected by the Breast Test Wales screening programme, compared with 38% of 2617 cases in 2008 (Table 1). There has of course been an extension of the routine screening age from 64 to 69 during the interim and the added effect of the post war baby boomers reaching the screening age may account for this increase. Overall the proportion with invasive disease has remained very similar as has that of those having surgical treatment. 56 patients were recorded as having primary medical treatment either as neoadjuvant chemotherapy or endocrine therapy before or instead of surgery. However there are a further 591 cases where no surgical treatment is recorded and they are not flagged as having primary medical treatment. These are considered in Table 2. 6 P a g e

11 Table 2 : Treatment received for the 591 unknown cases by hospital HOSPITALS Treatments recorded chemo chemo & horm chemo & imm horm horm & imm imm radio radio & chemo radio & horm radio & chemo & horm radio & horm & imm no treatment Total BRONGLAIS GENERAL HOSPITAL GLAN CLWYD CANCER SERVICES LLANDUDNO GENERAL HOSPITAL NEATH PORT TALBOT HOSPITAL NEVILL HALL HOSPITAL PRINCE CHARLES HOSPITAL PRINCE PHILIP HOSPITAL PRINCESS OF WALES HOSPITAL ROYAL GLAMORGAN HOSPITAL ROYAL GWENT HOSPITAL SINGLETON HOSPITAL UNIVERSITY HOSPITAL OF WALES WITHYBUSH GENERAL HOSPITAL YSBYTY MAELOR WREXHAM ALL HOSPITALS Many of these patients have some recording of treatment including endocrine (hormonal) therapy, chemotherapy or trastuzumab (immunotherapy). Some were also recorded as receiving radiotherapy soon after diagnosis. Although this may have been for advanced disease at presentation it is also possible that data on surgical procedures is missing. In some cases there is information about invasive size or nodal status which can only be acquired if the patients had surgery. The 184 cases with no treatment recorded have been notified to the Health Boards and Trusts. 7 P a g e

12 Table 3 : Characteristics of the cases Screen detected cases Symptomatic cases All cases No. % No. % No. % Total No N/A Sex: Female Male Tumour side: Left Right Bilateral Not recorded Invasive status: Invasive Non-invasive Unknown Age group: < Median age for patients diagnosed through symptomatic presentation was 68 compared with 62 for the screen-detected patients. The age distribution of screen-detected cancers reflects the invited population (Table 3). It is concerning that even the most basic data for cancer registration such as side or invasive status is not being correctly recorded in some patients, albeit a small number. Table 4: Histological type of breast cases Status Type Screen detected cases Symptomatic cases All cases Invasive Invasive tumour:not assessable Invasive ductal carcinoma Invasive lobular carcinoma Invasive medullary carcinoma Invasive tumour: mixed Invasive mucinous/colloid carcinoma Invasive tubular carcinoma Non-specific malignant tumour of breast Other specified malignant tumour of breast Total Non-invasive Ductal carcinoma in situ Lobular in situ neoplasia Paget's disease of nipple Total Unknown Total % of invasive breast cancers were classified as ductal of no special type (Table 4), and 13% were lobular. Non-invasive cancer accounts for 21.4% of screen detected cases compared with 5.4% of those presenting to the symptomatic service. This is as expected since most cases of in situ cancer are only identified on mammography. However there is a slight increase in both groups compared with 2007 cases. 8 P a g e

13 Male Breast Cases There were 12 cases of male breast cancer diagnosed within the audit period, 0.5% of the total cases diagnosed; all invasive: 9 had a mastectomy of whom 44% were node positive, and 3 had endocrine therapy only. This is a similar number (11) and proportion (0.6%) to the cases of male breast cancer identified in Because of small numbers, male breast cancers are not considered further in this report apart from in Outcome Measure 1, looking at completeness of data items. To obtain any meaningful outcome measures would require whole UK data, probably over several years. 9 P a g e

14 Proportion of cases Welsh Breast Cancer Clinical Audit Invasive Female Breast Cancers Figure 2 : Variation in age and first treatment modality for invasive breast cancers 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% < Unknown Primary medical surgery Age group As expected, most invasive cancers have surgical treatment as the first recorded treatment - the present normal standard of care. Across all groups, 65% were recorded as having primary surgical treatment, which is in contrast to the 1997 data (a casenote audit) where 78% of patients had primary surgery recorded. Primary medical treatment can be used in three different scenarios: For reducing tumour bulk in locally advanced tumours to enable subsequent surgical treatment (chemotherapy is the most common modality). For attempting to shrink tumour to make breast conservation a more viable option for the patient - used increasingly either with chemotherapy, endocrine therapy or trastuzumab in HER2 positive tumours. It is also a way to fast track potential new therapeutic agents within clinical trials. For treating patients where the risks of surgery are considered high due to patient comorbidity (endocrine treatment is most common). There are significant numbers of patients (23%) where the first treatment is not identified on Canisc as either surgical or medical as defined above (Figure 2). Again this contrasts to the casenote audit from 1997, where only 2% of patients had no primary treatment recorded. This is most evident in older age group, with 22% of year old having no primary treatment recorded rising to 63% of those aged 80 and over. It is believed that almost all patients will have some form of breast cancer specific therapy and, as discussed in Outcome Measure 4, missing primary or neoadjuvant endocrine therapy is the most likely explanation. If this is the case and those known to have had primary medical treatment are combined with those where first treatment is unknown, it can be seen that non-surgical first treatment rises with age, reaching 64% of those aged 80 or over at diagnosis. This is of concern, whether the explanation is lack of treatment and/or lack of adequate recording of treatment, and is revisited in more depth on pages 25 and P a g e

15 Proportion of cases Welsh Breast Cancer Clinical Audit Figure 3 : Variation in surgical treatment with age for invasive breast cancers 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% < < Screen detected Symptomatic Conservation Mastectomy Age group Overall mastectomy rates rise with age (Figure 3). In part may be due to patient choice rather than more widespread disease at presentation and higher rates of comorbidity may influence decisions. The rate for screen-detected patients under 50 must be interpreted with caution due to extremely small numbers and also because some of these patients may have a high family history risk, influencing decisions about treatment. Compared with the 1997 audit when overall breast conservation was 40% in surgically treated cases, in 2008 it was achieved in 59.2% of patients. Table 5 : Characteristics of invasive breast cancers surgically treated Surgically treated invasive cancers Screen detected (710) Symptomatic (990) All (1700) No. % % of total known No. % % of total known No. % % of total known Nodal status: Negative Positive Unknown N/A N/A N/A Grade: G1 - Well differentiated G2 - Moderately differentiated G3 - Poorly differentiated Unknown/blank/not assessable N/A N/A N/A Invasive size: <15mm <20mm <50mm mm Blank/unknown N/A N/A N/A Surgery: Conservation Mastectomy The pathology results for patients having surgical treatment are in line with other series, such as the UK All Breast Cancer Report October As expected the node positivity rate is lower for screendetected cancers compared with symptomatic (17.5% compared to 40.2%), (Table 5). In total, node positivity for patients in Wales (32.9%) is slightly lower than for the UK as a whole (38%). These rates are slightly higher when expressed as a percentage of those in whom the node status is known as shown in Figure 4 where 0, 12, 46, 36 and 19 patients were excluded in each age group respectively. The lower rate for screen-detected cancers had been recognised since the start of the screening programme and it is assumed that it will lead to better survival in these patients in the long term. However this alone cannot be used as a direct argument for extending the screening age further since the older patients are much more likely to die of other non breast cancer related causes. The overall known node positivity rate of 32.9% is similar to that in the 1997 audit (35%). 11 P a g e

16 Proportion of known cases Proportion of known cases Welsh Breast Cancer Clinical Audit Figure 4 : Variation in lymph node status by age for invasive breast cancers surgically treated 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% < Node negative Node positive Age group Figure 5 : Variation in grade for invasive breast cancers surgically treated 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Screen detected Symptomatic G3 - Poorly differentiated G2 - Moderately differentiated G1 - Well differentiated Grade is an important prognostic factor in breast cancer, with well-differentiated cancers having better predicted outcomes. There is an obvious trend towards lower grade tumours being identified through screening (Figure 5). 9 and 14 patients with unknown grade are excluded in each group respectively. 12 P a g e

17 Proportion of known cases Proportion of cases Welsh Breast Cancer Clinical Audit Figure 6 : Variation in lymph node status by grade and size of invasive breast cancers surgically treated (where size and grade are known) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <15mm 15- <20mm 20- <50mm 50+mm <15mm 15- <20mm 20- <50mm 50+mm <15mm 15- <20mm 20- <50mm 50+mm G1 - Well differentiated G2 - Moderately differentiated G3 - Poorly differentiated Node negative Node positive Size of invasive tumour Figure 6 summarises the trend towards higher node positivity as both size and grade increase. These factors are used in the Nottingham Prognostic Index 3 (Figure 7). Figure 7 : Variation in Nottingham Prognostic Index (NPI) grouping for invasive breast cancers surgically treated by screening status 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Screen detected Symptomatic EPG GPG MPG MPG PPG screen-detected and 125 symptomatic NPI values could not be calculated due to missing data. All of the data used to calculate NPI are needed to properly assess adjuvant therapy recommendations the fact that they are missing raises significant doubt about whether all patients are receiving appropriate advice. 3 EPG excellent prognosis group (NPI score < 2.4) GPG good prognosis group (NPI score > 2.4 and <3.4) MPG1 moderate prognosis group (NPI score > 3.4 and <4.4) MPG2 moderate prognosis group (NPI score > 4.4 and <5.4) PPG poor prognosis group (NPI score > 5.4) 13 P a g e

18 Proportion of cases Proportion of known cases Welsh Breast Cancer Clinical Audit Figure 8 : Variation in Nottingham Prognostic Indicator (NPI) score with age for all invasive breast cancers surgically treated, where NPI is known 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% < EPG GPG MPG MPG PPG Age group Although patients under 50 in age generally have a worse NPI at presentation compared with those 50 or over (Figure 8), this trend is reversed again in the 80 and over group breast cancer does not become a better disease in older patients. Some of this apparent effect is due to screening in the middle groups. Figure 9 : Variation in mastectomy rates with increasing whole size of invasive tumour for surgically treated patients 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <15mm 15- <20mm 20- <50mm Screen detected 50+mm blank/un known <15mm 15- <20mm 20- <50mm Symptomatic 50+mm blank/un known Conservation Mastectomy Whole tumour size Note: Invasive size used if whole size missing or not measurable. Not surprisingly the mastectomy rate is higher for larger tumours whether screen-detected or symptomatic (Figure 9). The invasive size alone is not the best measure to determine when more radical surgery is needed because some patients may have a moderate sized invasive tumour surrounded by a large area of insitu cancer. 14 P a g e

19 Proportion of cases Proportion of cases Welsh Breast Cancer Clinical Audit Non-Invasive Cases Non-invasive or in situ cancer accounts for less than 12% of all cases reported in this audit and the following graphs indicate some of the differences in the presentation and treatment of this disease. Figure 10 : Variation in method of presentation by age for all non-invasive breast cancers 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% < Symptomatic Screen detected Age group As expected the majority of non-invasive breast cancers are found through screening in the screening age group (50-69), (Figure 10). A small number, however, are identified in symptomatic patients under 50 where it may be coincidental because the patients are having mammograms as part of their evaluation. Figure 11 : Variation in mastectomy rate for non-invasive DCIS tumours with increasing whole size 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <15mm 15- <20mm 20- <50mm Screen detected 50+mm blank/un known <15mm 15- <20mm 20- <50mm Symptomatic 50+mm blank/un known Conservation Mastectomy Whole tumour size As for invasive cancers there is an obvious increase in the mastectomy rate for larger areas of DCIS (Figure 11). There are a considerable number of cases where the whole size is not recorded properly and this needs improving. 15 P a g e

20 Proportion of cases Welsh Breast Cancer Clinical Audit Figure 12 : Variation in radiotherapy rates for DCIS cases following conservative surgery by grade 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% G1 - Low grade G2 - Intermediate grade G3 - High grade (blank) All DCIS cases No radiotherapy Radiotherapy Nuclear grade Current protocols recommend giving radiotherapy after local excision of DCIS when high grade or >20mm of intermediate grade to reduce the risk of local recurrence (see Figure 12). As was reported in the 2007 cohort audit, it is highly likely that there is significant under-recording of the use of radiotherapy in this situation and this needs attention to ensure that patients are receiving appropriate therapy. 16 P a g e

21 Breast Cancer Clinical Outcome Measures Potential surrogate measures have been developed and refined nationally by the Cancer NSAG Breast Subgroup in collaboration with the Cancer Information Framework team. It is hoped these may be indicative of long term outcomes for patients in terms of recurrence and survival. The analysis of the clinical outcome measures refer to invasive cancer only, unless otherwise stated. Outcome Measure 1: Number and percentage of breast cancers for which complete information is recorded. Comparison has been made of the levels of data completeness for individual Breast Units between 2007 cases as reported last year and the new 2008 data for both male and female breast cancers. Eight areas were analysed for completeness and have also been scored on relative importance as follows. Demographics - NHS number, surname, forename, date of birth, postcode and gender examined. If any of these items were missing then the cases was defined as incomplete. score +1 for every 20% complete (0-5) Tumour details - diagnosis date, histology and laterality examined. If any were missing then the case was defined as incomplete. score +1 for every 20% complete (0-5) Treatment final type of surgery, radiotherapy flag, chemotherapy flag, hormone therapy flag and immunotherapy flag examined. If all fields were empty then the case was defined as incomplete. score +1 for every 20% complete (0-5) Staging Invasive size, grade and node status were examined for surgically treated invasive cases only. If any were missing then the case was defined as incomplete. (An invasive size of 0 or 0 nodes examined or a stage of G0 (unknown) was treated as incomplete. Stage GX (not appropriate) or unmeasurable for invasive size were treated as complete). score +1 for every 20% complete (0-5) but 4 for 80-95% or 5 for above Oestrogen Receptor (ER) status ER status examined. If missing or not recorded then the case was defined as incomplete. score +1 for every 20% complete (0-5) but 4 for 80-95% or 5 for above HER-2 status HER-2 status examined for patients aged < 70. If missing, the case was defined as incomplete. If not done recorded for an invasive tumour, then also treated as incomplete. score +1 for every 20% complete (0-5) but 4 for 80-95% or 5 for above Data for Outcome measure 9 Case defined as incomplete if invasive cancer treated with conservative surgery has no record of radiotherapy. score +1 for every 20% complete (0-5) but 4 for 80-95% or 5 for above Data for Outcome measure 10 Case defined as incomplete if a node positive patient who has received surgery and is less than 61 years old has no record of chemotherapy. score +1 for every 20% complete (0-5) but 4 for 80-90% or 5 for above 17 P a g e

22 Data for Outcome measure 11 Case defined as incomplete if an ER positive patient has no record of hormone therapy. score +1 for every 20% complete (0-5) but 4 for 80-90% or 5 for above 90%. The radar plots below (Figure 13) are categorised at all Wales, Cancer Network and Hospital level. The red areas identify the data completeness in 2008 compared with the blue in 2007 Figure 13 Radar plots of data completeness for patients diagnosed in 2007 and 2008 Outcome measure 11 Outcome measure 10 All Wales Demographics Tumour details Treatment Outcome measure 9 Pathology data HER-2 status ER status Regions 4 Outcome measure 11 Outcome measure 10 North Wales Demographics Tumour details Treatment Outcome measure 11 Outcome measure 10 South West Wales Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status Outcome measure 11 Outcome measure 10 South East Wales Demographics Tumour details Treatment Outcome measure 9 Pathology data HER-2 status ER status Regions synonymous with the three cancer networks extant at the time these patients were diagnosed. 18 P a g e

23 North Wales Hospitals Outcome measure 10 Glan Clwyd General Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 10 Llandudno General Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status Outcome measure 10 Ysbyty Maelor Wrexham Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data HER-2 status ER status South West Wales Hospitals Outcome measure 10 Bronglais General Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 10 Withybush General Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status Outcome measure 11 Outcome measure 10 Prince Philip Hospital Demographics Tumour details Treatment Outcome measure 11 Outcome measure 10 Singleton Hospital Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status P a g e

24 Outcome measure 10 Neath Port Talbot Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 10 Princess of Wales Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status South East Wales Hospitals Outcome measure 11 Outcome measure 10 Prince Charles Hospital Demographics Tumour details Treatment Outcome measure 10 Royal Glamorgan Hospital Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status Outcome measure 11 Outcome measure 10 Royal Gwent Hospital Demographics Tumour details Treatment Outcome measure 11 Outcome measure 10 Nevill Hall Hospital Demographics Tumour details Treatment Outcome measure 9 Pathology data Outcome measure 9 Pathology data HER-2 status ER status HER-2 status ER status Outcome measure 10 University Hospital of Wales Outcome measure 11 Demographics Tumour details Treatment Outcome measure 9 Pathology data HER-2 status ER status In order to give a better view of data completeness, a scoring system has been developed and could be refined in the future. It is disappointing that overall there has been no significant improvement in data quality. Glan Clwyd Hospital particularly stands out as having the worst data quality and steps need to be taken to improve this. However there are also others where data quality has become worse: Singleton, Llandudno, Prince Charles, Royal Glamorgan and Princess of Wales (Figure 14; those highlighted in red have deteriorating data quality between the two audits). 20 P a g e

25 Figure 14 Completeness score by hospital Collection of data relating to chemotherapy, endocrine and radiotherapy remains unsatisfactory across the networks. The quality of pathology data is critical for determining adjuvant therapy for all patients having surgical treatment and tumour grade, size, and node status are all needed to produce an NPI score. Increasing reliance is placed on ER and Her2 as well and these results need to be available at the MDT meeting. Figure 15 : Variation in NPI recording for invasive breast cancers surgically treated by hospital (black line indicates mean) Figure 15 shows the proportion of cases where NPI cannot be calculated by hospital and highlights the hospitals with a particularly poor record. Table 6 identifies which components of NPI are not available. 21 P a g e

26 Table 6 : Variation of unknown tumour grade, size, nodal status and NPI score for surgically treated invasive breast cancers diagnosed by hospital In case there might be unforeseen bias in missing data due to age or stage presentation differences, Table 7 shows the median NPI (where available) and median age for each hospital. Table 7 : Median age and median NPI score of surgically treated invasive breast cancers by hospital 22 P a g e

27 Outcome Measure 2: Number of symptomatic and screen detected cancers treated in a breast unit per annum Figure 16 Invasive breast cancers by hospital (black line indicates mean) It is important to know the ratio of screen-detected to symptomatic cancers for an MDT because the long term outcomes may be much better for screen-detected cancers and this would be reflected in survival figures for an MDT (Figure 16Figure 17). In 2008 there were no surgeons at Prince Charles and Neville Hall working at BTW and hence patients with screen-detected cancers from their natural catchment areas were treated at other hospitals. This is a direct consequence of the way in which screening has been organised to concentrate the expertise needed to manage impalpable breast cancers. Figure 17 : Non-invasive breast cancers by hospital (black line indicates mean) Figure 17 confirms that most cases of DCIS are diagnosed through screening. (Prince Charles had no cases.) 23 P a g e

28 Outcome Measure 3: Number and percentage of symptomatic cases with a non-operative histological/cytological diagnosis (KCI 1) 5 Figure 18 : Proportion of symptomatic cases with a pre-operative diagnosis for hospitals (black line indicates mean) The current Cancer NSAG key clinical indicator that applies to this data is as follows: Pre-operative diagnosis rate (%) in symptomatic breast cancer cases. (minimum: 90%, target 95%) Additionally the Association of Breast Surgery standard is to achieve a pre-operative diagnosis rate of 95% or greater for symptomatic cancers. The mean rate across Wales is 70% using the recorded data (Figure 18), with only one hospital reaching the minimum key clinical indicator level, which on face value is very poor, and an apparent decrease over the decade since the first audit, where only 3% had no preoperative histology or cytology. However data capture for needle biopsy is thought to be poor across hospitals in Wales and should be improved with a high priority. This needs to be urgently addressed by LHBs. Data for screen-detected cancers will be available from Canisc for patients diagnosed in 2009, although the rates are well known within Breast Test Wales as part of its QA process. 5 Cancer NSAG: breast key clinical indicator 1 24 P a g e

29 Outcome Measure 4 & 5: Number and percentage of invasive cases having primary medical or surgical treatment It is important to be able to distinguish those patients who do not follow the usual pathway of surgery before drug treatment since many of the subsequent quality measures are specifically used for patients following this normal pathway. Table 8 : Type of treatment recorded for invasive breast cancers by hospital HOSPITALS No. invasive cases Primary surgical treatment Primary medical treatment Unknown No. % No. % No. % BRONGLAIS GENERAL HOSPITAL GLAN CLWYD CANCER SERVICES LLANDUDNO GENERAL HOSPITAL NEATH PORT TALBOT HOSPITAL NEVILL HALL HOSPITAL PRINCE CHARLES HOSPITAL PRINCE PHILIP HOSPITAL PRINCESS OF WALES HOSPITAL ROYAL GLAMORGAN HOSPITAL ROYAL GWENT HOSPITAL SINGLETON HOSPITAL UNIVERSITY HOSPITAL OF WALES WITHYBUSH GENERAL HOSPITAL YSBYTY MAELOR WREXHAM ALL HOSPITALS A significant number of patients are not recorded as having surgical or primary medical treatment (Table 8). Although a very small number of patients will have presented with terminal metastatic disease and therefore have no breast specific intervention, this is a rare event. Many may have had these methods of treatment but the data had not been recorded correctly. Five patients had some form of surgical data entry but it is thought that this was after primary chemotherapy or the dates of surgery were outside pre-defined limits. Table 9 : Hospitals with possible missing surgery Tumour grade may be available from a pre-operative core biopsy but not tumour size (i.e. as measured histologically after excision). Therefore the presence of size suggests that surgery has been carried out (Table 9). The presence of radiotherapy records may also suggest this since it is usually only given after surgery. Chemotherapy, hormone therapy and immunotherapy may be given 25 P a g e

30 as primary medical treatment or after surgery and do not therefore help in identifying which was the first treatment. Patients who are assumed to have had surgery for the reasons discussed above but where it was not recorded are excluded from Figures 19 to 21. Figure 19 : Variation by hospital of invasive breast cancers where no surgery is recorded and have no surgical information - all ages Even allowing for missing surgery records when it appears to have been carried out from the pathology information available, and the 30 patients known to have primary medical treatment, nearly 20% of patients still have no reliable primary treatment pathway recorded as shown in Figure 19. Figure 20 and 21 break this down by age group: there is a higher rate in the over 70s suggesting strongly that these patients are likely to have been treated with primary hormonal therapy only. There is no inherent reason to think that data from surgery should be missing in this age group compared with the age group. Figure 20 : Variation by hospital of invasive breast cancers where no surgery is recorded and have no surgical information age group(black line indicates mean) Figure 21 : Variation by hospital of invasive breast cancers where no surgery is recorded and where there is no surgical information age group (black line indicates mean) If this is true, there appears to be considerable variation across breast units in the use of primary endocrine treatment in the management of patients above the age of 70. This has been observed in several national audits such as the ABS UK Symptomatic Breast Audit showing unit variation 0%-75%). This may be variation in comorbidity and patient choice but could also be due to variation in policies and practice, and warrants further investigation into the reasons by LHBs. 26 P a g e

31 Mastectomy Rate Welsh Breast Cancer Clinical Audit Outcome Measure 6: Mastectomy rate by whole size of tumour The ideal mastectomy rate is unknown and will also depend on patient/tumour characteristics, particularly when patient choice is offered in an open manner. Modern oncoplastic techniques may enable breast conservation for larger tumours while still maintaining good cosmetic outcomes. Nevertheless it may be possible to identify units where the rates are so far from the norm that further investigation of reasons should be considered. In the future it will be possible to track trends over time in case small numbers in an individual year are skewing results (see Figure 22 and 23). Figure 22 : Variation by hospital in mastectomy rates for all surgically treated invasive breast cancers (all sizes) (black line indicates mean) Funnel plots allow many points to be plotted simultaneously, with information about whether each point is significantly above or below the expected, or average, value. In the funnel plots within this report, the y axis shows the mastectomy rate and the x axis is the number of cases in total. Each point represents a hospital. The dotted lines show 95% (2 standard deviations) and 99.7% (3 standard deviations) confidence intervals and the straight line is the average rate. Any hospital that lies outside of the confidence intervals is significantly different from the average. Caution is advised when dealing with small numbers. Figure 23 : A funnel plot to show the variation by hospital in mastectomy rates for all surgically treated invasive breast cancer (all sizes) Data Average 2SD limits 3SD limits Total number of surgically treated invasive female breast cancer cases 27 P a g e

32 Mastectomy Rate Welsh Breast Cancer Clinical Audit The Welsh mean (see Figure 22 and Figure 23) is very similar to the UK mean of 43%. One unit with a large caseload lies below the 3SD line, indicating a breast conservation rate that differs to the norm for all sizes of breast cancer as well as for tumours <20mm whole size as shown in Figure 24. This may indicate a difference in practice. The availability of more radical oncoplastic techniques for achieving good cosmesis may contribute to this but this unit should ensure that the cosmetic results, margin clearance and local recurrence rates have been fully evaluated and shown to match the best published results. Figure 24 : A funnel plot to show the variation by hospital in mastectomy rates for all surgically treated patients with invasive breast cancer of <20mm invasive size (for hospitals with >5 cases) Data Average 2SD limits 3SD limits Total number of surgically treated invasive female breast cancer cases <20mm invasive size With smaller breast cancers, it would be anticipated that more patients would have breast conservation and this plot demonstrates the mean to be 26% for tumours <20mm compared with 41% for all invasive breast cancers. 28 P a g e

33 Mastectomy Rate Welsh Breast Cancer Clinical Audit Figure 25 : A funnel plot to show the variation by surgeon in mastectomy rates for all surgically treated invasive breast cancer (all sizes, for surgeons with >10 cases) Data Average 2SD limits 3SD limits Total number of surgically treated invasive female breast cancer cases Figure 25 demonstrates the wide variation in mastectomy rates by surgeon. One surgeon appears to have a mastectomy rate statistically different to the norm in Wales. Rates range from c.89% to c.12%, which compares with a range of 79 to 39% in 1997 (for surgeons >10 cases). As previously stated, the ideal rate is unknown, and deviation from the norm should not be used to infer sub-optimal practice, however, it is recommended that both clinical and patient satisfaction outcomes are reviewed. Figure 26 : Variation by hospital in mastectomy rates for surgically treated screen detected, invasive breast cancers (all sizes) (black line indicates mean) Figure 26 and 27 demonstrate higher mastectomy rates overall for symptomatic breast cancers compared with screen detected, likely mainly due to smaller sizes of the latter. This may however be due to the greater expertise available in units with wide exposure to screening. Bronglais and Neville Hall have very small numbers of selected screen-detected cancers being treated. 29 P a g e

34 Mastectomy Rate Welsh Breast Cancer Clinical Audit Figure 27 : Variation by hospital in mastectomy rates for surgically treated, symptomatic invasive breast cancers (all sizes) (black line indicates mean) Figure 28 : Variation by hospital in mastectomy rates for all surgically treated non-invasive DCIS breast cancers (black line indicates mean) Where the type of surgery was recorded there is wide variation in the mastectomy rate for DCIS as shown in Figure 28. Figure 29 however does not identify any surgeon more than 3 standard deviations from the mean. Figure 29 : A funnel plot to show the variation by surgeon in mastectomy rates for surgically treated DCIS cases (for surgeons with >10 cases) Data Average 2SD limits 3SD limits Total number of surgically treated DCIS female breast cancer cases 30 P a g e

35 Outcome Measure 7: Number and percentage of surgically treated invasive cancers with known histological nodal status For the correct management of patients with invasive breast cancer, it is essential that the histological axillary node status is known. This enables correct management of the axilla to minimise axillary recurrence and also informs the selection of adjuvant systemic therapy and radiotherapy. Figure 30 Nodal status recorded by hospital for all surgically treated invasive breast cancers (black line indicates mean) Although the availability of node status is generally good (mean 93%), (Figure 30), there are concerns about the recorded level in Glan Clwyd and to a lesser extent Withybush and Bronglais. LHBs will need to establish whether this is due to missing information or lack of appropriate surgery and implement changes accordingly. 31 P a g e

36 Outcome Measure 8: Number and percentage of histologically node negative invasive breast cancers for which less than 7 nodes were removed (KCI 2) 6 The purpose of this measure is to identify where patients may have received excessive dissection of the axilla thus leading to higher complication rates, particularly lymphoedema. This has been identified in the National Institute for Clinical Excellent (NICE) Early Breast Cancer Guidelines (2009) where minimal dissection of the axilla should be offered unless preoperative assessment of the axillary nodes has proved histological or cytological involvement. Therefore the measure specifically looks at the number of nodes harvested in patients where all the removed nodes are negative. The figure of greater than 7 was selected to allow for both sentinel node biopsy (SNB) and axillary sampling. However occasionally a complete axillary clearance can be carried out and less than 8 nodes identified. The Cancer NSAG key clinical indicator states that, at minimum 90% of histologically node negative patients should have 1-7 nodes removed with a target of 95% Figure 31 : Variation by hospital in node negative invasive breast cancers treated surgically (black line indicates mean). Occasionally patient choice or comorbidity may mean that an axillary clearance rather than sentinel node biopsy is carried out even without positive preoperative confirmation of positive nodes but this should be a rare event. In the future it is hoped that this will occur for less than 10% of patients. The mean for 2008 of 22% (Figure 31) has improved compared with 33% for 2007 as SNB has become mainstream but there are still some notable exceptions as indicated in Figure 31. Figure 32 excludes cases where an axillary lymph node clearance is known to have been carried out and improves the overall figures considerably, although the numbers per unit are very small for some units. In many cases the patients having less than 7 nodes removed would have been during the training phase of SNB. This requires careful monitoring in the future when all units have completed SNB training. Figure 32 : Variation by hospital in node negative invasive breast cancers surgically treated (excluding those cases where an axillary lymph node clearance is known to have been carried out at the same time)(black line indicates mean). 6 Cancer NSAG key clinical indicator 2 32 P a g e

37 Outcome Measure 9: Number and percentage of invasive cancers treated by conservation surgery and receiving adjuvant radiotherapy (KCI 5) 7 It is standard practice to give breast radiotherapy to all patients after breast conservation surgery since in trials the local recurrence rate is much higher if this is not done. Figure 33 : Radiotherapy given after conservation surgery for invasive breast cancers, by hospital (black line indicates minimum level required) Key clinical indicator: % of patients who had breast conservation surgery for invasive breast cancer having breast radiotherapy. (minimum: 90%, target 100%), (Figure 33 ). The percentage of patients with no record of radiotherapy after breast conserving surgery has improved from a mean 17% in 2007 to 10% in This is likely to be due to better recording but breast units need to review the cases where no radiotherapy is recorded to establish the reasons and to make sure all patients are being offered appropriate treatment. In Figure 34, comparing age above and below 65 for patients apparently not receiving radiotherapy shows that older age is not generally the reason, with the possible exception of Glan Clywd. Figure 34 : Variation by hospital in age of invasive breast cancers receiving no radiotherapy after conservative surgery (black line indicates average split between age groups) 7 Cancer NSAG Breast key clinical indicator 5 33 P a g e

38 Outcome Measure 10: Number and percentage of node positive patients with invasive breast cancer, aged 60 or under receiving adjuvant chemotherapy (KCI 4) 8 It is good practice to give adjuvant chemotherapy to the majority of node positive patients aged 60 or under but there will be some for whom the added benefit is quite small, some who are not medically fit enough, and others who do not wish to have it. Thus the actual ideal percentage to aim for is uncertain but probably a minimum of 80% is realistic. Key clinical indicator: % of node positive women (aged <=60 years) who received adjuvant chemotherapy (minimum: 80%, target 90%) Figure 35 : Variation by hospital in chemotherapy treatment of node-positive patients with invasive breast cancer 60 years of age or less following surgical treatment to the breast(black line indicates minimum level required) A mean of 61% patients in this category were recorded as having received chemotherapy in 2007 and this has increased to 77% in 2008 which may well be due to better data recording. Over 60% of hospitals now appear to at least exceed an 80% rate which is encouraging (Figure 35). However for Wrexham, Neath Port Talbot, Withybush and Princess of Wales Hospitals, it is not possible to tell whether patients are getting optimum management. These hospital teams need to review their cases. It has been suggested that for future audit rounds we should examine whether this indicator should be extended to all patients under Cancer NSAG Breast key clinical indicator 4 34 P a g e

39 Outcome Measure 11: Number and percentage of ER positive patients with invasive breast cancer receiving adjuvant hormone therapy (KCI 3) 9 Most patients with ER+ve invasive breast cancer should be given adjuvant hormonal therapy after surgery. The benefits in small well differentiated node negative cancers are very small and in 2007 some MDTs in Wales did not routinely prescribe it for these cases. The new All Wales endocrine 10 guidelines released in February 2008 recommended offering endocrine therapy to all ER+ve patients with invasive breast cancer and we would expect at least 90% of such patients to receive endocrine therapy. Key clinical indicator: % ER positive women who received adjuvant hormonal therapy (excluding those in the Excellent Nottingham prognostic index group) (minimum: 90%, target 95%) Figure 36 : Variation by hospital for ER positive patients with invasive breast cancer that received hormone therapy following surgical treatment to the breast (black line indicates mean) It is therefore disappointing that the mean of 36% of ER+ve patients recorded as receiving adjuvant endocrine in 2007 has only increased to 38% in 2008 (Figure 36). There is huge variation in the recorded use of adjuvant endocrine therapy and those hospitals identified in Figure 36 as having low or non-existent recording need to improve dramatically particularly Royal Glamorgan, Wrexham and Princess of Wales. The problems of collecting adjuvant endocrine therapy have been highlighted at a UK level (NCIN All Breast Cancer Report 2009). 9 Cancer NSAG Breast key clinical indicator 3 10 Available on Cancer NSAG Internet: And Intranet: 35 P a g e

40 New Outcome Measure: Number and percentage of patients having breast conservation surgery with margins of excision free of tumour It is widely accepted that one factor for increased local recurrence after breast conserving surgery is incomplete clearance of the primary tumour. Measuring this requires close cooperation between surgeon and pathologist to ensure the surgical margins are properly orientated in the specimen. When tumour cells are identified at, or very close to, the radial excision margins by careful histological evaluation, it is normal practice to carry out further surgery to ensure the tumour has been adequately excised. There may be occasions when MDT evaluation and discussion concludes that further surgery is not in the patient s best interests or the patient, with informed advice, decides against this. However this is likely to be a rare event. The NICE Early Breast Cancer Guidelines (2009) have made the specific recommendation of a minimum 2mm clearance of DCIS, although the margin for invasive cancer was not one of the topics evaluated in these guidelines. Figure 37 : Variation in clearance of margins by hospital for invasive breast cancers following breast conservation surgery Figure 37 for invasive cancers and Figure 38 for DCIS shows a considerable number of patients with no margin information recorded (11% and 13% respectively). A smaller number of patients had involved or uncertain excision margins. All hospitals should review these cases to ensure that patients are getting optimum management. Figure 38 : Variation in clearance of margins by hospital for DCIS following breast conservation surgery 36 P a g e

41 Survival For the first time information about relative survival is shown for the 2007 and 2008 cohorts. One and two year survival estimates are a very short timescale for breast cancer and reflect patients presenting at a late stage of disease, particularly in those not treated surgically (Table 10 : 1 and 2 year relative survival (RS) of the 2007 female invasive breast cases Table 10Table 11). However they represent a benchmark for the future. It will be the surgically treated patients who will require monitoring, possibly in association with WCISU Triennial Report. Table 10 : 1 and 2 year relative survival (RS) of the 2007 female invasive breast cases Table 11 : 1 year relative survival (RS) of the 2008 female invasive breast cases However, using the chi square distribution, there is a significant difference in the distribution of the ages of the surgically treated group and the non-surgically treated group. This would be expected because of existing comorbidity at presentation, particularly in older patients and perhaps also because late presentation is more common in the older patients. The following table shows the proportions: Table 12 : Age distribution The International Cancer Benchmarking Project (ICBP) has published data on invasive breast cancers in a number of countries including Wales 11. The methodology does not directly compare to that used here, however the publication reports that for patients diagnosed in Wales between 2005 and 2007, 1 year relative survival for breast cancer, although similar to other UK regions, was worse than all other countries participating. 11 Lancet Vol 377 January 8, 2011, pg P a g e

42 Summary and Next Steps In summary, this audit builds on the excellent work of the previous year and presents analysis of a large body of data to support clinicians and LHBs in optimising care for breast cancer patients. Positive developments are evident in a number of areas; however, variation exists in the clinical indicators of good practice. Audit is an iterative process, at its best evolving and maturing with each round, as increasing data quality and clinical engagement interact in a positive feedback loop to drive up quality. Whilst often apparent in the early cycles, the current lack of complete data is frustrating; limiting useful analysis and therefore feedback to clinical teams. It is to be hoped that initiatives and developments since 2008 will have improved data completeness: the wider roll-out of Canisc, the implementation of CHIRP, and the drive to improve capture of staging data, participation in international research (International Cancer Benchmarking Project) requiring full clinical data for breast cancer and this audit itself. More accurate data will allow expansion of the audit beyond the primarily surgical scope that has been the case to date, facilitating analysis of other areas of clinical interest, and therefore wider clinical ownership of the audit. Suggestions of areas to cover in the next audit include: Outcomes and stage of presentation. Characteristics, management and outcomes of patients with triple negative cancers in Wales. Axillary node staging. Local recurrence rates for breast conservation. Comparison of pathways and outcomes in the screening age range between screening and symptomatic patients. Metastatic disease This will require collaboration between organisations and data sources, including pathology and radiology. It is proposed the next audit round presents data for patients diagnosed in both 2009 and 2010 to bring the audit more up to date. It is recommended that each MDT nominate an audit lead, to support the MDT lead in ensuring that the data submitted accurately reflects practice and to liaise with the Breast subgroup of the Cancer NSAG on areas for inclusion in the next analysis. 38 P a g e

43 Appendix 1- Case matching Figure 1 : Matching WCISU cases to Canisc breast cancer cases 2166 (82.8%) morphology match (The differences were largely due to the specificity of coding rather than having a completely different type of breast cancer. This was looked at in further detail below.) 2449 (93.6%) diagnosis match (within 45 days) (From checking a sample of these cases, it showed both databases could be updated with each others earlier diagnosis dates) 2491 (95.2%) laterality match (The majority of the 25 cases had the correct side recorded on WCISU) 2516 (96.1%) recorded in 2008 (The majority of the 94 cases had the same laterality and morphology code. They are to be looked at in further detail) 2610 (99.7%) recorded breast cancer (The 2 patients were recorded with unknown primary cancers on WCISU and will be updated to breast cancer) 2612 (99.8%) NHSNo match (3 patients were English so wouldn t be recorded on WCISU and 2 patients need to be registered with WCISU) 2617 Canisc patients As shown in can be seen from the Figure 1, 2617 patients were extracted from Canisc for the diagnosis year of these patients were found on WCISU s database. Of the 5 patients not found, 3 were English patients and the other 2 were clinically diagnosed and have now been registered with WCISU. Of the 2612 patients that matched on NHS number, 2610 patients had a breast cancer recorded on WCISU s database. The remaining 2 had an unknown primary recorded. Further investigation into these found that the primary cancer was breast and so the 2 records have been amended on WCISU s database cases were a breast cancer diagnosed in 2008 on WCISU s database. A sample of the remaining 94 patients was looked at in more detail and reasons for the differences in diagnosis dates included pathology reports for WCISU showing an earlier diagnosis date and diagnosis dates on Canisc with no proof or evidence of where they came from. However the majority were the same cancer, having the same morphology and laterality patients further matched on laterality. The 25 patients that did not match were also looked at in further detail and for the majority of cases WCISU had the correct side recorded. The remaining 2491 cases were then matched further on diagnosis date within 45 days of each other matched with the reasons not matching being those mentioned above. Both databases could be updated. The last match looked at morphology codes. Of the 2449 patients, 2166 matched. A sample of the 283 difference was looked at in further detail and the reasons included a more specific morphology being coded on either of the databases and slightly different codes being used. 39 P a g e

44 Figure 2 : Matching Canisc breast cancer cases to the WCISU breast cancer cases from 2008 diagnosis year WCISU had 2956 breast cancer patients diagnosed in This figure included out of area (OOA) and inactive cases i.e. English patients treated at a Welsh hospital of these were recorded in the Canisc extract. WCISU had a further 440 patients which required further investigation to determine why they were missing from the Canisc extract. 45 of these patients were found in the 2007 Canisc extract. Reasons behind different diagnosis dates were discussed earlier. Of the remaining 395 patients, 42 were OOA or inactives (hence English patients), 122 were diagnosed at an English hospital and 24 did not have a hospital recorded and the majority of these 24 were death certificate only (DCO) registrations. This explains why they were not in the 2008 Canisc extract. 207 of the 395 patients were diagnosed at Welsh hospitals with 156 being in one of the hospitals in this report. Of these 207 patients, 131 had pathology, 15 were clinical, 1 came from an English registry and 60 came from PEDW. These 207 cases were further investigated to see if they were recorded on Canisc for another diagnosis year other than 2007 and 2008 or to see if there were reasons why they weren t extracted. 40 P a g e