Surgical education can no longer be considered adequate if limited to description of surgical diseases and methods of management. Due to the growth in numbers of surgical diagnostic and therapeutic procedures, the surgeon of the future will find it increasingly necessary to understand the principles by which algorithms are constructed and by which individualized decisions should be made. Several systematic approaches for assisting clinical decision makers have been developed. Decision analysis is particularly appealing because it is flexible and readily adapted to a wide range of clinical situations. It explicitly guides the decision maker in determining the crucial variables in a clinical decision, and permits both objective data and personal preferences to play a part in decision making. Because it provides for personal estimates and preferences, decision analysis is not dehumanizing, even though it is quantitative, explicit, and mathematically rigorous. Topics for a series of seminars or case conferences are suggested. Decision analysis should be part of the intellectual preparation of every clinician.

SummaryReferral decisions of physicians dealing with long-term ambulatory problems are complex phenomena that are not clearly understood. This study was designed to assess the possible rationale behind such decisions in the management of obesity. It examined how well a subjective expected utility (SEU) model accounted for decisions of 45 primary care physicians regarding referral of obese female patients to an endocrinologist. Two patient goals, weight reduction and patient satisfaction, and a two-year time horizon were incorporated in the model.Data were collected using 24 written cases representing 12 patients approximately 100% overweight and 12 about 50% overweight, and a semi-structured interview in which subjective probabilities and importance weights were obtained. Values were calculated by transforming physicians’ ratings of risk of morbidity in the 24 cases into a utility scale.The SEU did not account for the primary care physicians’ referral behavior. Correlations between number of patient cases referred and SEU were analyzed and were not statistically significant, although there was substantial variation across physicians in number of cases referred. Mean subjective probabilities of weight loss and patient satisfaction were essentially identical for referral and non-referral.The formulation of the model, the design of the cases, and the method of value assessment are discussed as potential threats to the validity of the model as an account of referral decisions. Problems of constructing an adequate model are considered.

Decisions regarding estrogen replacement therapy were obtained from 50 physicians for 12 cases representing menopausal women with systematically varying levels of cancer risk, fracture risk, and symptom severity. Their decisions were compared with a decision analytic model for which each physician provided needed quantities—subjective probabilities, utilities of various outcomes, and weightings of the importance of the outcome categories. The majority of observed decisions were not to treat. By contrast, the decision analysis based on physician-provided estimates indicated that the optimal strategy was either to treat or a toss-up. Sensitivity analysis showed that these conclusions would hold over all possible utilities, over all plausible probabilities of cancer, and so long as symptom relief and fracture prevention were also considered as treatment objectives. The increased probability of early detection of cancer by regular follow-up was systematically incorporated into the decision analysis but apparently neglected in unaided clinical judgment, which follows the principle of minimizing the most important risk, regardless of its probability.

Results of two surveys (D and E) that assessed the current structure and status of instruction in medical decision making are reported. Both samples (Survey D, N = 80; Survey E, N = 92) consisted of members of the Society for Medical Decision Making. A consensus was obtained on topics considered important for teaching an introduction to clinical decision analysis to medical professionals. These topics were Bayes' theorem, decision trees, 2 X 2 tables, test sensitivity and specificity, utility, and ROC analysis. There was little agreement on course structure, level, or the preferred method for teaching decision analysis within medical settings. It was concluded that medical educators are in the process of constructing a knowledge base in decision-analytic techniques within academic medicine. It will soon be time to consider the place of more advanced topics within the continuum of medical education.

SummaryWe demonstrate the value of skull x-rays in the assessment of head injured patients. With the use of the simple decision analytic technique of 2 x 2 tables and literature derived figures, it is shown that in patients with a normal neurological examination following head injuries, skull x-rays are a cost-effective method of detecting preventable complications. The positive predictive value (PVP) of skull x-rays in these patients is. 03, in contrast to .001 of the group as a whole. Such cases should then have a CT scan to detect early cases of intracranial haematoma (ICH). The presence of abnormal neurological signs has a high PVP of .2 for complications; in these patients, skull x-rays do not improve the detection of ICH and CT scanning alone should be carried out.

MEDICAL DECISIONS IN PERSPECTIVE: APPUED RESEARCH IN COGNITIVE PSYCHOLOGY ARTHUR S. ELSTEIN* MARGARET M. HOLMES* MICHAEL M. RAVITCH* DAVID R. ROVNERA GERALD B. HOLZMAN,% and MARILYN L. ROTHERT* Medical decisions claim the attention ofthose outside the scientific and medical communities as well as oftheir members. The general pubUc has an abiding concern in the personal and economic consequences ofmedical decisions. Researchers are engaged in analyzing medical decisions and determining the information necessary for action under uncertainty and the impact of high technology on decision making. For these reasons , clinical decision making has become a major area for psychological investigation. The rapid growth of research in clinical decision making has led to interest in improved methods and frameworks for data analysis. The purpose of this paper is to describe and evaluate the primary theories and methods employed in contemporary psychological research on clinical decision making and to compare these approaches with two naturalistic methods of research in this field, chart audit and direct observation. The findings from this body of research wUl not be summarized here since several recent reviews are readUy avaUable [1-4]. Psychological research on medical decision making has generally been experimental or controlled research. Subjects respond to a carefully designed or selected medical problem, either a written or simulated case in an experimental setting. This design results in highly reproducible, reliable information but is subject to question aboutthe relationship ofthese results to the real world. More naturaUstic studies, relying on patient charts or observation in clinical settings, are clearly related to die real world but may produce results that are so much a function of time and Work supported in part by die National Library of Medicine grant LM-03396 and Biomedical Research Support grant SO7RR05656-13 to Michigan State University. "Office of Medical Education Research and Development, Michigan State University, East Lansing, Michigan 48824. tDepartment of Medicine, Michigan State University, East Lansing, Michigan 48824. ^Department of Obstetrics and Gynecology and Reproductive Biology, Michigan State University, East Lansing, Michigan 48824.© 1983 by The University of Chicago. All rights reserved. 0031-5982/83/2603-0343$01.00 486 J Arthur S. Ebtein, Margaret Af. Holmes, et al. * Medical Decisions in Perspective setting that questions of generalizability arise once again. This paper wiU discuss the advantages and limitations of experimental versus naturalistic design as well as the research questions that are characteristic ofthree major psychological theories. Cognitive Theories in Clinical Research The major cognitive theories employed in research on clinical reasoning are information-processing psychology, social judgment theory, and decision theory. The information-processing view of clinical reasoning aims to characterize the reasoning processes by recording and analyzing the steps and thoughts of clinicians as they attempt to solve clinical problems. The goal is to describe the ongoing processes in terms of basic psychological elements and principles. Socialjudgment theory, on the other hand, deals statistically with correlations between original cues and final outcome. The thought processes are regarded as a black box, and linear correlations are used to depict the degree to which each cue enters into the final judgment. Decision analysis approaches the solution of clinical problems from the standpoint of risky choice under uncertainty. The decision problem is carefully structured and bounded. The approach requires one to be explicit and quantitative. It assumes that, given a choice, the physician will act rationally to maximize the best outcome for the most patients. Each approach provides an analytic perspective which suggests questions to be asked and the kind of data best suited to answer these questions . In general, both information processing and social judgment theory rely on descriptive data. These studies aim to identify the organization of factual knowledge and inference rules required for effective clinical judgment [5, 6] or to analyze the determinants of decisions in situations where competent decision makers differ in their recommendations for the diagnostic workup or management of a particular condition [7, 8]. Studies in the information-processing tradition have analyzed expert performance on problems where consensus can be obtained in order to specify the points to be emphasized in instructional materials and the knowledge base or practical skills students should acquire [9, 10]. These studies assume that better understanding of the thought processes...

Problem Solving: Applications of Research to Undergraduate Instruction and EvaluationAbstract The formal reasoning strategy used in medical diagnostic problem solving can be conceptualised as composed of four more elementary processes — cue acquisition, hypothesis generation, cue interpretation and hypothesis evaluation. These processes are closely linked to the clinician's store of medical knowledge. The acquisition, retention and recall of content cannot ensure its effective application, yet training in problem‐solving skills with inadequate attention to factual content will not be effective either. Two educational programmes are described which aim to increase the effective linking of clinical strategy and clinical memory in undergraduate medical students. Clinical problem‐solving sessions use simulated cases to provide experience in blending clinical knowledge and problem‐solving strategies. Problem‐solving examinations offer the opportunity for students to display the reasoning and planning behind their actions. They assess students’ ability to use a problem‐solving model and to document the medical work‐up using the Problem Oriented Medical Record. Both educational programmes use concepts and theory developed in empirical research on medical problem solving. Fundamental research is used to raise questions about educational practice, while the practical concerns of education raise questions for substantive inquiry.