EXPERT-SYSTEMS - ASSISTING FORMULARY DECISION-MAKING IN THE AMBULATORY SETTING

In this article, the author reviews the application of a computer-assi sted decision support system to their formulary decision-making proces s. Basic information is presented describing expert systems, which are a type of computer-assisted decision support system, and their advant ages and disadvantages. A specific example of an expert system, 'RXPER T', is described. 'RXPERT' is a prototype expert system that models th e decision-making process for an ambulatory (nonhospital) formulary. T his formulary is the underpinning of the prescription drug benefit pro gramme for the nearly 1 million residents of Saskatchewan, Canada. In the current formulary decision process, each drug product is evaluated by 2 separate committees, with the third and final decision resting w ith the Ministry of Health. The first committee, the Drug Quality Asse ssment Committee (DQAC), comprises members with expertise in medicine, pharmacology, clinical pharmacy, pharmaceutics, statistics, and regul atory processes. The DQAC evaluates information from the drug manufact urer and other independent sources, and makes an initial assessment wi th respect to clinical aspects of alternative therapies and generic in terchangeability. The committee then makes its recommendation to the S askatchewan Formulary Committee (SFC). The SFC reviews the recommendat ion of the DQAC and considers the administrative and economic implicat ions of accepting the product for the patient, the programme, and heal thcare professionals' practice. The SFC either reaffirms the recommend ation of the DQAC or modifies it based on further review, and forwards its recommendation to the Ministry of Health. Finally, the Ministry o f Health reviews the evaluation and determines the drug's formulary st atus. The 'RXPERT' prototype is based on specific variables, rules, an d criteria used by the members of the DQAC and SFC to evaluate drug pr oducts for formulary inclusion. In modelling the decision-making proce ss, knowledge acquisition techniques consisted of retrospective analys is of DQAC and SFC meeting minutes, observation of the 2 committees' d ecision making, and structured personal interviews with individual mem bers. Preliminary assessment demonstrated that the 'RXPERT' recommenda tions agreed with those of the DQAC in about 90% of cases. As the 'RXP ERT' prototype expert system has demonstrated, computer-assisted decis ion support systems can be readily applied to pharmacy practice. Howev er, before these systems are accepted, several issues need to be addre ssed. These issues include healthcare professionals' acceptance of the technology, the credibility and validity of the expert system recomme ndation, the legal liability asociated with system derived decisions a nd, most importantly, the economic and personnel resources required fo r the development and implementation of these systems.