HipMod: Development of a Multi-Agent Audit-Based Computer Simulation of Hip Fracture Care

This paper describes the development of a computer simulation of the care of hip fracture, a common and serious injury with a complex journey of care. The project made use of a national, evidence-based guideline on hip fracture care, together with data from a national hip fracture audit and from service and research sources. To overcome the recognized limitations of such data in the modelling of care, clinicians from different specialties and disciplines working on hip fracture care participated by means of multi-agent-based modelling techniques. The model followed the journey of care (admission, surgery, rehabilitation, discharge or death), and, by incorporating the clinical reasoning of experienced practitioners, was developed to reflect the realities of day-today clinical decision-making. It was evaluated in terms of the credibility of its outputs with clinical participants, and by formal statistical comparisons of its outputs with real outcomes from comparable groups of patients in the national audit. The model was used to explore the impact of guideline compliance on care and resource use. Further modelling might serve to explore the implications of different service configurations for hip fracture care; to support service planning for demographic change; and to assist clinical training. Introduction Planning health care services requires the acknowledgement of different interests: the 'consumer' or patient, clinicians, managers, politicians, accountants, public health specialists and more. Decision-making strategies should therefore be broadly based, allow consideration of multiple options and testing of outcomes and also encourage the involvement of stakeholders in terms that they understand. Health care service planning must often be done with incomplete or inadequate information , and in a context of severe clinical, economic and political pressures. Health care provision is also expected to be increasingly evidence-based, and evidence-based guidelines are emerging for many clinical conditions. But questions remain about how service delivery can be aligned to guideline recommendations, how compliance with them can be measured and how this will affect outcomes. Computer simulation might be more extensively used both to inform planning and answer such questions. Health care systems typically have large amounts of data describing their service activities – such as patient numbers, surgical procedures, length of stay and hospital mortality – while other important domains such as clinical casemix, details of process and clinical outcomes are often poorly described. The limitations of computer simulation of services based on routine service data are substantial, but have been addressed over recent years in ways …