Acute Lymphoblastic Leukaemia (ALL) is the commonest childhood cancer. Treatment for lasts up to 3 years and is usually carried out within the context of clinical trials. In the UK, these trials are being co-ordinated by Cancer Research UK's Children's Cancer Group (CCG).
A large and diverse team of staff participates in each child's treatment, with care typically shared between a central specialist unit, a smaller hospital nearer the patient's home, and community liaison teams. Treatment involves multiple drugs and complex dosage and scheduling protocols. Given the large number of staff involved, ensuring that these protocols are followed correctly presents considerable challenges.
The potential for computerised decision-support systems to improve physician compliance within clinical trials, as well as supporting safety and efficiency, is well-recognised. However, there has been little progress in realising this potential - largely due to the difficulties of translating promising experimental systems into routine practice in complex clinical environments. In this project, we have been able to exploit an infrastructure developed to support electronic data gathering to additionally support the deployment of a decision-support system across a large number of institutions.
The Children's Cancer Group and the ACL collaborated on the development of a prototype decision-support system to assist clinicians with prescribing decisions during the critical maintenance phase of treatment. The DSS was integrated with a remote-entry database (developed by the Cancer Research UK's IS department) designed to record detailed clinical information about each patient and provide basic prescribing support. This system was piloted in the summer of 2001, and deployed in routine practice in late 2002.
The decision-support tools developed are targeted at recognised clinical needs - specifically dosage adjustments and scheduling of interventions during collaboratively delivered Continuing Therapy. A number of clinicians have expressed a desire to see the system extended to provide more detailed support for scheduling and co-ordination of interventions during Continuing Therapy. At present, the decision-support component informs users what tasks would normally be due in a given week, but this information is not patient-specific. We are exploring ways of extending the PROforma language and inference engine to enable users to query, for example, whether tasks are overdue or have been cancelled, or whether a task's intended rescheduling is within the bounds defined by the protocol.
The key software components of the LISA system are: