Most computer-based decision support systems (DSSs) in medicine have been developed in hospital settings, intended for use in hospitals. In this study, a DSS for general practitioners (GPs) in primary care is designed, taking into consideration that primary care is the first level in a health care organization. Female genitourinary (GU) infections is chosen as prototype area for the study of decisionmaking.
There are three primary data sources used for this study: 1. Decision protocols were obtained from 11 GPs after 139 GU consultations. A decision certainty estimate and an estimate by the GP of the patient's desire to go through GU work-up was included. 2. A nine-physician panel evaluated 63 of these protocols. Individually, the panel completed a similar decision protncol without access to the GP's decisions. 3. Questionnaires were responded to by 186 primary care physicians regarding information needs and attitudes towards computer support. The critical incident technique is used to identify information dilemmas.
Discriminant analysis is used to identify dati items used by the GPs to differentiate between decision alternatives. The kappa coefficient is used as measure of inter-physician decision variability in the panel.
From a theoretical review, a model is establisbed,of which knowledge types the GP uses and the forms in which this knowledge is used in daily practice:Not all types of knowledge relevant to the GP are available in forms amenable to computer manipulation. Doctor-patient communication skills are, for instance, tacit and acquired through professional experience.
The main empirical results of this study are that: I. The GPs rely heavily on laboratory data in their decisions. However, they fail to use negative evidence. Orthogonal patient desire is a major source of uncertainty. 2. The urethritis diagnosis is used inconsistently. 3. There are considerable differences between individual physicians in their use of medical concepts. In one case out of four, no consensus diagnosis is available at all. 4. Dilemmas in general inte'rnal medicine are the most prevalent medicaldilemmas for the GP, and support for drug prescription and access to full-text databases are the computer applications most desired.
A design of the DSS is described, which consists of five integrated components: a hypertext module, a critiquing program for support of drug prescriptions, diagnosis support of reconsider type, an interface to computer-based library and communication resources, and a central database. The design is implemented in experimental form. Organizational changes to facilitate decision-making and a theoretical model of the GP's information use arc discussed.
Vimmerby: VTT Grafiska , 1989. , 74 p.
Papers, included in the Ph.D. thesis, are not registered and included in the posts from 1999 and backwards.