HEALTH ECONOMICS Health Econ. 7: 595–603 (1998) ECONOMIC EVALUATION MAGNETIC RESONANCE IMAGING FOR THE INVESTIGATION OF KNEE INJURIES: AN INVESTIGATION OF PREFERENCES STIRLING BRYAN a, *, MARTIN BUXTON a , ROBERT SHELDON b AND ALISON GRANT b a Health Economics Research Group, Brunel Uniersity, Uxbridge, UK b Accent Marketing and Research, London, UK SUMMARY The conventional approach to the diagnosis and treatment of severe knee injuries is arthroscopy, a minimally invasive surgical procedure. Since arthroscopy is an invasive technique that carries risks, magnetic resonance imaging (MRI) is increasingly being used for diagnosis. MRI is potentially associated with diagnostic and therapeutic ‘impacts’, in that arthroscopy can be avoided. This paper reports a discrete choice conjoint analysis exercise that assessed the value placed on such ‘impacts’ by potential patients and investigated the degree to which respondents were willing to trade between process and outcome. The marginal rates of substitution between attributes were estimated. The results suggest that the diagnostic and therapeutic ‘impacts’ of MRI were valued by many respondents. The study has highlighted a number of important issues for the design and analysis of future health-related conjoint studies, including the use of treatment cost as an attribute, dealing with data from lexicographic respondents, and distinguishing between points of indifference and missing data. © 1998 John Wiley & Sons, Ltd. KEY WORDS — knee injuries; MRI; conjoint analysis INTRODUCTION This paper reports a discrete choice conjoint anal- ysis exercise. The use of conjoint analysis is well- established in a number of fields of economics, most notably transport and environmental eco- nomics [1 – 4]. The application of conjoint analysis to the field of health economics is relatively new, although there is growing interest in the use of this technique in the health area and there have been a number of studies undertaken recently [5–11]. It is appropriate to view choice-based conjoint analysis models as falling within the overall scope of random utility theory. The aim in developing such a model might be either to evalu- ate certain attributes in terms of others, thus allowing marginal rates of substitution to be esti- mated, or to forecast demand. The use of conjoint analysis in health economics is likely to be con- cerned principally with the former objective. One of the objectives of the study reported here was to address study design issues in health economics. The clinical focus of this paper is the diagnosis and treatment of severe knee injuries. The conven- tional approach to the diagnosis of such injuries that has now become accepted practice is arthroscopy [12,13]. This is a minimally invasive surgical procedure commonly undertaken as a day-case under general anaesthetic. One of its main advantages as a diagnostic approach is that if the surgeon discovers a problem requiring surgi- * Correspondence to: Health Services Management Centre, University of Birmingham, 40 Edgbaston Park Road, Birmingham B15 2RT, UK. Tel.: +44 121 4147706; fax: +44 121 4147051; e-mail: s.bryan@bham.ac.uk CCC 1057–9230/98/070595 – 09$17.50 © 1998 John Wiley & Sons, Ltd. Receied 22 August 1997 Accepted 29 May 1998