D.Litt. et Phil. / The practice of medicine changes constantly and rapidly. Consequently, it is difficult for clinicians to learn about innovations, given the vast quantity of information available. Evidence-based medicine (EBM) is the process by which practitioners turn clinical problems into questions, and then systematically integrate personal clinical expertise with the best available external evidence as the basis for clinical decisions. To practice EBM, the practitioner is required to search the literature for relevant material, and then to synthesise knowledge and apply findings to each patient. Clinicians require fast and specific access to multiple data sources, but the availability of electronic full text documents has substantially exacerbated the lack of time to read the clinical literature owing to the demands of clinical practice, and is further compounded by the fact that the Web contains much health-related misinformation. Clinicians therefore require a means of searching the literature that will enhance the retrieval of accurate and evaluated clinical data from ranked resources, whereby the most relevant information is retrieved first from the most likely source. Strong correlations exist between four primary steps in EBM, and the formula commonly used in search strategy design in the field of information seeking. The similarities inherent in these steps suggest that an evidence-based approach to information seeking might enable endusers in the health professions to enhance their searching skills and to translate the clinical question into an appropriate information-seeking strategy. A main problem and two sub-problems were investigated, namely whether: · a Web-based EBM information-seeking model could be designed to enhance the information-seeking skills of healthcare practitioners · it was possible to design an information-seeking model more closely aligned with the clinical decision-making model familiar to healthcare practitioners · it was possible to design such a model in a manner that could further enhance the translation of the clinical question into an appropriate information-seeking strategy. Various models in medicine and the domain of information seeking were investigated. It was found that the model of the clinical decision-making process accorded with all six phases of the information-seeking process (ISP), whereas other information-seeking models only addressed the ISP from the formulation of the problem onwards, thus ignoring prior stages of initiation, selection and exploration in the ISP. A Web-based EBM information-seeking model (Model C) was devised and tested for compatibility against a general Web-based information-seeking model, and was found to be valid. Model C was further empirically assessed against a Web site design methodology, and was again found to be compatible. A unique approach to EBM information seeking is thus offered by Model C, which incorporates all aspects of the clinical-decision-making metaphor, as well as the “PICO” EBM filters (Patient/Problem, Intervention, Comparative Intervention and Outcome), into a facet analysis template for the design of a clinical search strategy. Prior to selection of the EBM information resource, Model C further allows for the ranking of each resource and for the design of individual browsing and/or analytical search strategies, as appropriate, so as to enhance EBM information seeking amongst healthcare practitioners.

The relationships between medical libraries and the main users of medical information in the teaching hospitals and University departments in Oxford is examined. Systems ideas are used to define the sort of model or picture of the users that a library manager needs in order to provide the appropriate services, and to provide a formal means of incorporating users and their information needs into a management control system. Data were gathered by several methods and combined into a rich picture of the users and their information-seeking behaviour. The systems methodology developed by Checkland at the University of Lancaster was used to test this rich picture and link it with monitoring for library effective- ness. Application of the Checkland methodology was a crucial step which shifted the emphasis of the project from quantitative to conceptual modelling. The methods of data collection and the results are described as the User Survey. The following techniques were used: questionnaires, semi-structured interviews, direct observation, feedback forms (a critical incident technique), a reference tracing experiment and analysis of existing library records. The data gathered by those methods presented a consistent picture in which the nature of the users' work, ie research or clinical practice, was the dominant influence on information-seeking patterns. Application of the Checkland methodology and the conceptual models derived from it are described as the Systems Study. This revealed that the formal processes for monitoring and control expected by the conceptual models did not appear in identifiable form in the real world. Further examination showed that a detailed description of the library function was necessary and that this statement could be used to generate performance criteria. In addition, the rich picture from the User Survey was found to be a fair representation of reality. Conclusions for systems thinking, user studies, library managers and medical librarianship are presented.