APPENDIX E
OPTIMAL RADIOTHERAPY UTILISATION MODELLING - METHODOLOGY
Radiotherapy Utilisation Modelling is an attempt to try to establish the proportion of cancer patients within a population who should receive radiotherapy based on best available evidence. For a number of years it has been widely accepted across the commonwealth that about 50% of cancer patients should receive radiotherapy. However, in reality radiotherapy utilisation does and will vary across the countries as each has a different case-mix of cancer types, stage of cancer at presentation and co-morbid diseases, all of which independent of resource availability, impact on treatment delivery.
In an attempt to accurately quantify the requirement for radiotherapy, Professor William MacKillop and his team at the Queen's Cancer Research Institute, Kingston, Ontario devised a technique of establishing radiotherapy indications; defined as 'a set of clinical parameters that suggest that radiotherapy is necessary, expedient or advisable'. They conducted a systematic review of the literature, including published guidelines, such as, Scottish Intercollegiate Guidelines ( SIGN), Cancer Care Ontario Clinical Practice Guidelines, National Cancer Institute PDQ guidelines etc and from this established evidence based clinical scenarios where radiotherapy is indicated. Then, using population data the proportion of cases in a cancer population matching these scenarios was applied to calculate optimal radiotherapy utilisation.
In 2001, the Australian Department of Health and Ageing funded the Collaboration for Cancer Outcomes Research and Evaluation ( CCORE) of the Liverpool Health Service to carry out a project titled Radiotherapy in cancer care: estimating the optimal utilisation from a review of evidence-based guidelines. This group, chaired by Professor Michael Barton, conducted a systematic review of the radiotherapy indications for all cancer sites. Any indications for radiotherapy identified in clinical practice guidelines or other literature was included in the analysis. These were then entered into a software programme (TreeAge TM) and for each cancer radiotherapy utilisation trees were constructed. Then using, wherever possible, Australian population data optimal utilisation rates for all cancer sites were calculate. A full report has been published and is available at http://www.ncci.org.au/pdf/radiotherapyreport.pdf
The Radiotherapy Activity Planning Group for Scotland gratefully acknowledge the permission of Professor Barton and Dr Delaney from CCORE, to use their radiotherapy utilisation trees to calculate the optimal radiotherapy utilisation for Scotland.
In order to achieve radiotherapy indications that would more accurately reflect Scotland and provide a sounder basis for forward planning for the Scottish Radiotherapy Service 2011-2015, the evidence base applied to the modelling was altered if there was a widely different approach to cancer management. In brief this is because interpretation of data and clinical practice varies around the world and highest priority evidence applied by the CCRORE group was at times at variance with clinical practice in Scotland. One example is the use of post-operative radiotherapy for localised stomach and pancreatic cancer. While this may be standard practice in Australia, it is rarely practiced in Scotland mainly as a result of different surgical practice and interpretation of the evidence.
Once completed, the Scottish Radiotherapy Utilisation trees were circulated to site-specialist Clinical Oncologists to obtain their agreement that they accurately reflected optimal clinical practice based on the current clinical community and national consensus views on currently available evidence.
Once the indications for radiotherapy were established, data from the Scottish Cancer Registry and National and Regional Audits were used to calculate the proportions of patients within each category. Scottish data was not available for all cancer types and in these cases the Northern and Yorkshire Cancer Registry and Information Service database ( http://www.nycris.org.uk) was searched to obtain comparable data for a UK population. Failing this, the data used in the original Australian report were used.
The data which are required for this process are
- data on distribution of the various pathological subtypes
- data on the stage distribution of the different cancer sites
- data on surgical clearance
- patient fitness and co-morbid diseases
- presence or absence of focal symptoms
Data was obtained from the following Scottish Cancer Audits.
- Breast: South East Scotland Cancer Network ( SCAN) breast cancer audit 2003. SCAN includes patients from Borders, Dumfries and Galloway, Fife and Lothian and covers a population of 1.4 million, 27% of the Scottish population. It is a mixed urban and rural population. All patients with breast cancer are referred to a specialist breast cancer surgeon and so the audit case ascertainment is high. In accordance with SCAN clinical guidelines all patients see a Clinical Oncologist pre-operatively to assess their suitability for breast conservation, therefore the rate of mastectomy for 'social' reasons e.g. travelling, is low. Data on tumour stage (including number of lymph nodes involved) and type of surgery were used.
- Lung: SCAN lung cancer audit 2002-2003. The case ascertainment is not as complete as the Breast cancer audit with about 80% (800 out of 1000 cases) of the population being identified. However, as all surgery, radiotherapy and chemotherapy is delivered by clinician's participating in this audit it is unlikely that any patient receiving treatment for their lung cancer would have been missed. The 'missing cases' have been distributed according to the same proportions as the known cases. Data on tumour stage and performance status were used. The respiratory physician making the diagnosis enters the latter data onto the registration form.
- Urological cancers : Scottish Urological Cancer Audit 1999 - 2002. This was a large national Audit (in preparation for publication), which looked at a case distribution and treatment of Urological Cancers across Scotland, including bladder, kidney, penis, prostate, renal pelvis & ureter and testis.
- Head and Neck: Scottish Audit of Head and Neck Cancers 1999-2002 . This was a large national Audit (in preparation for publication), which looked at the case distribution and treatment of head and neck cancer across Scotland. The case distribution was used. In addition to these data, other data on the suitability for local resection was obtained from 2003 SCAN audit data.
- Gynaecological Cancers: Edinburgh Cancer Centre database 2001. In this year all patients with pathologically confirmed gynaecological invasive cancer from the SCAN region were discussed at the multi-disciplinary meeting and their details entered onto the Cancer Centre database (subsequently only those seen at the Cancer Centre have been entered). As the 'trigger' for discussion was the pathological diagnosis of cancer, no cases should have been missed. Data on stage was used.
- CNS: 1999 CRAG audit. This was commenced as a Scottish National audit, but unfortunately Greater Glasgow had to stop collecting data but the data still covers over half the Scottish population. This audit looked at the management of patients presenting with a CT scan, which suggested a malignant brain lesion. The data on biopsy rate and case distribution were used.
- Upper Gastro-intestinal: Scottish Audit of Upper Gastrointestinal Audit and Oesophageal Cancers ( SAGO) 1997-2000. This was a large national Audit including Stomach, Pancreas, Oesophagus, Gall bladder.
- Melanoma data is collected nationally and was supplied via personal correspondence
- Colorectal : SCAN 2002 Colorectal Cancer Audit. This identifies all patients referred to colorectal surgeons and oncologists. This primarily collects data according to Duke's stage and the data on TNM have only recently been collected but are available for 74% of rectal cancer cases. The 'missing cases' have been distributed according to the same proportions as the known cases.
In the Scottish modelling we primarily looked at radiotherapy use in the initial phase following diagnosis. Form the audit of treatments performed in 2003 it can be seen that radical treatments consume in excess of 80% of the fractions. However, in order to enable comparison with International figures, using the relapse figures from the Australian data, the rate of deferred radiotherapy has also been calculated.
Horizon Scanning Group of the Radiotherapy Activity Planning Group
November 2004