Dukes B adjuvant chemotherapy risk prognostication tool

Decision support system designed to provide individualised risk prognostication in postoperative chemotherapy treatment of Dukes B stage patients with bowel cancer.

 

Background

Decisions about cancer treatment have a significant impact on people’s lives and place a considerable burden on patients and clinicians. On occasions the balance of harms and benefits may be marginal. Decision aids can inform the decision process and help people making decisions according to their values, goals and capabilities as reported in a recent Cochrane review (O’Connor et al, 2009). Decision aids appear particularly valuable for decisions where there is no single best choice.

Bowel cancer is one of the most common cancers in the UK with 35,000 people diagnosed every year. More than a third of all bowel cancer patients are offered postoperative chemotherapy to reduce the risk of recurrence and prolong survival. It is not possible, however, to predict the effects of such treatment on an individual: patients risk exposure to chemotherapy with the risk of side effects but negligible benefit. This is particularly true for patients without lymph node involvement (Dukes B), where the risk of recurrence is comparably lower.  Whilst currently clinicopathological risk factors such as T-stage and vascular invasion are frequently used to determine high risk groups, there is no consistent way of identifying patients with Duke’s B who would benefit from chemotherapy. In the consultation about adjuvant treatment, clinicians need to convey complex information about prognosis, benefits and side effects of chemotherapy to facilitate the decision process based on the best available evidence and consistent with patients’ values, goals and capabilities. To help with this we have developed a computerised decision support tool for this group of patients. The tool provides information on prognosis, pros and cons of chemotherapy treatment options and includes a value clarification tool. 

 

Knowledge Base

The system’s knowledge base uses the most recent guidelines and textbooks and expert interpretation of the current knowledge regarding the value of adjuvant chemotherapy in Dukes B stage patients. 

 

Knowledge Modelling

A project data model (or dataset), developed by the  National Bowel Cancer Audit Programme (NBOCAP) has been used. SNOMED has also been employed when the NBOCAP dataset was found insufficient. A database schema has been prepared based on the above model.

The PROforma language has been used to model the workflow as a series of tasks. The model suggests steps to help the consultant and the patient make an informed decision. The process consists of a case review, an optional data entry/correction phase, value clarification, risk calculation and decision support.

 

The risk and decision logic has been modelled using two approaches that will be combined for comprehensive risk estimation and decision making:

1. 1. A quantitative approach that is based on the risk model published by Adjuvant online. Our quantitative model extends Adjuvant online by presenting the evidence-based support in the form of evidence-derived arguments;
2. A qualitative approach that encodes the knowledge derived from publications (guidelines, textbooks, trial results, meta analyses etc) into a set of for and against rules. This approach has been proven to be very efficient in communicating the decision rationale to patients in the past and as stated above, it supplements the quantitative risk calculators.

The decision process has been modelled in four steps that we believe will help the patient make a more informed decision:

1. Calculation and rationale for the risk of recurrence;
2. Estimation of the potential benefit from adjuvant chemotherapy;
3. Choice of regime taking patient preferences into account;
4. Use of the value clarification tool (filled in by the patient).

The data, workflow and logic models have been implemented using the PROforma language (see screen below). The data model has been additionally encoded in a database schema and the risk calculation model as algorithms in the source code.

 

Data, workflow and decision modelling in PROforma using the TALLIS composer

 

Architecture

An important aspect of the architecture of the system’s software framework is its generic design. The Dukes B colorectal service, designed to deliver decision support in complex problems such as the choice of adjuvant treatment, is a specialisation of the platform.

This approach allows us to benefit from and contribute to the work of other projects that use the same platform, such as MATE which has developed a system to provide decision support to breast MDT meetings. The Dukes B framework has seen some of its technology used in the breast MDT tool and the experience gained in the development of the latter system has provided invaluable knowledge in our modelling of medical decisions.

 

System in Use: example screens

Web-based user interface of the system showing the patient
database browser (left) and a case review (right)

 

 

Decision step 1. The risk of recurrence is shown based on both quantitative and qualitative decision
models i.e. the Adjuvant Online calculator combined with evidence-derived rules 

 

The system's chemotherapy regime recommendation (which may be "do nothing")
 based on a combination risk graphs, evidence based rules and patient preferences

In use

The system is designed to be web-based but can run on a remote site, on the hospital’s Intranet or as a standalone laptop application. It follows a component-based approach and can use its own database or data from an EHR system.

 

Study

A study is planned to evaluate the potential of the system for patients having a consultation on adjuvant treatment after resection of Dukes B bowel cancer. The study will aim to quantify the patient benefit of using a combination of both decision support based on clinicopathological information and gene expression profiling in the decision-making process.

The system has been designed to provide the patient with all available treatment options, and shows how each may modify relapse or mortality risks, and also how each could affect future quality of life. The study will also therefore seek to evaluate the system's ability to help the patient make an informed treatment decision with the assistance of the expert clinician. 

Reference
 

O'Connor AM., Bennett CL, Stacey D et al. Decision aids for people facing health treatment or screening decisions. Cochrane Database of Systematic Reviews 2009, Issue 3. 

 

Demonstrator

• Demonstrator (NOTE: this is NOT for clinical use) 

 

Support

Royal Free Hampstead NHS Trust Charity. [DATES]

 

 

Acknowledgements

Astrid Mayer, leader of the colorectal multidisciplinary team at the Royal Free NHS Trust, London

Ioannis Chronakis, senior software engineer, Department of Academic Oncology, UCL and Royal Free NHS Trust, London.