Technology and Health Care, 1 (1993) 89-92 0928-7329/93/$06.00 © 1993 Elsevier Science Publishers B.V. All rights reserved

89

Clinicians' needs and requirements. 2. Conceptual comments Joaquim Henriques Hospital Militar Principal, Lisbon, Portugal

Introduction Dr. Zoltie, in the previous presentation, has described clinicians' requirements from a practical viewpoint; and I am in broad agreement with what he says. Certainly there is need for rapid, easy to use, systems which relate to clinical practice and where the main system emphasis is on the clinician and his work, and not upon the technology involved. In this article however, I want to discuss a further problem; because we are talking about decision support, and it is not much use to discuss decision support without discussing what sort of support we are going to provide and what sort of decisions we are doing it for!

The clinical decision process For a start, let us criticise severely the term often used to describe decision support, namely computer aided diagnosis. With hindsight, it is difficult to think of a more inappropriate term! Quite apart from the sensitivity of doctors towards computer systems taking over their traditional role, every practising clinician knows that there is far more to patient management than mere diagnosis.

Of course diagnosis is important, but is merely a step towards appropriate management. Also important are prognosis and risk evaluation [1]; that is to say in many medical situations the diagnosis is obvious (and no diagnostic help is needed from computers or anything else), but appropriate management depends on assessment of risk and likely prognosis after different forms of therapy (Fig. 1). A good example of this situation is acute chest pain. Here the diagnosis is often obvious, for example, a typical myocardial infarct. The real challenge in management is often (a) the prediction of prognosis, and (b) allocation of the most appropriate form of therapy thereby. There are many such examples in clinical medicine: another good example (which we shall later discuss) concerns risk evaluation in elderly patients before surgery.

Risk evaluation systems It is not surprising therefore that, in relation to acute chest pain, and many other areas of medicine, a whole variety of "risk evaluation systems" have been proposed. Some of these are computerised systems, some are merely scoring systems which are worked out on a piece of paper

90

J. Henriques / Technology and Health Care 1 (J 993) 89 - 92 Usual (incorrect) concept of Medical Management

IPatient Arrives

----'tI~ Diagnosis -----I\{TilITa~YJ

1-\

r-\

More complex (realistic) concept of Management

Patient presen t5 with problem

Overall Assessment

Fig. 1. Two concepts of the medical decision making process. The first is simple but not accurate. The second (more complex), more nearly reflects what doctors actually do when making decisions.

by clinicians in real time, some simply lists of "risk factors". All have the aim of providing individual patient prognosis; but not all have been successful. A typical example of the difficulties encountered by risk evaluation systems is found in Crohn's disease. The prognosis of Crohn's disease in an individual patient remains a mystery even to experienced clinicians. In 1976 the American National Co-operative Crohn's Disease study produced an index (the Crohn's Disease Activity Index) aimed to provide a mathematical framework for decision support. This admirable idea was soon followed by publication of at least six other systems, all mutually incompatible, and all subsequently shown to be unrelated to patient prognosis! Another area where risk evaluation systems have had more success has been in the area of surgical and intensive care, particularly in elderly patients. This is an area where considerable attention has been paid in Portugal to risk evaluation. In Portugal, the proportion of patients over 90 undergoing surgery has increased by a factor of 5 in the last 15 years, and the estimated number of individuals over 60 years old by the year 2000 will be 2000000. There is thus considerable impetus for risk evaluation systems in these patients [2]. In

practice, several systems predicting prognosis after surgery and in the intensive care situation are well known (such as the Apache system [2] and the systems of Vac anti [4] and Siegel [3] and their colleagues). In this clinical area we see why "computer aided diagnosis" is such a restrictive term. After surgery in the elderly, the diagnosis is almost always perfectly clear. What the surgeon wants to know is the relative risk faced by the patient. Here the situation is more hopeful. In our own on-going studies [1] (involving patients with a mean age of 65 coming to surgical operation), we have been able to discriminate (on the basis of calculating probabilities) between a group of patients in a predicted high risk group (of whom 35% died) and a similar low risk group of patients (of whom 8% died). The benefits (should these results be maintained and reproduced) both for patient care and for patient counselling are obvious. The help information technology can give If we accept this broader definition of decision making, then a role can clearly be seen for information technology and computers - perhaps not that which the enthusiasts would wish, but useful

1. Henriques / Technology and Health Care 1 (J 993) 89 - 92

91

MEANS by which COMPUTER may assist CLINICAL DECISION Area of Support

Example

Simple Calculation

Calculating "Risk Scores"

Evaluation of Therapy

Evaluation

O.M.D.M.

Data Standardisation Peer recommended Modus Operandi

Decision Support

Diagnostic Suggestions Prognostic Predictious

Predictive Systems Technology Alternate Therapies

Fig. 2. Listing of some modalities by which computers may aid medical decisions. Note that the list is profoundly different from simple "computer diagnosis" proposed in the 1960s.

nevertheless. The various aspects of this role are set out in Fig. 2. First, computers can do simple calculations. I fully agree with Dr. Zoltie when he says that computers can save duplication. In this case, since many of the predictive systems are quite complicated, computers can do the simple mathematical calculations and save doctors the bother of doing so. Moreover computers can retain previous results and display them either as a chart or graphically so that it is easy for doctors to see at a glance whether the patient's index is going up or going down. All of this computers do very well. Computers can also be used to conduct simulation studies and thereby evaluate not only the various aspects of predictive systems (for example each symptom and sign in the Crohn's Disease Activity Index or the Apache physiological score) they can also evaluate the systems themselves; and can (by extension of this idea) be used to evaluate on a more objective basis each innovation in management or surgical procedure. Clearly there are other areas where information technology and decision support systems can provide useful help. One such stems from the very nature of the current technology systems themselves. Their use on a wide scale involves uniformity of data and harmonisation, which is in itself a valuable aspect as we have seen in the Concerted Action project.

I have left until last the help which directly concerns patients. Clearly, as in our own studies mentioned earlier, an accurate risk evaluation system can help both with patient care and also with patient counselling. These are valuable attributes of risk evaluation; this applies obviously to immediate patient management; but it also applies to such matters as stratifying patients for controlled research studies.

Summary In conclusion, as regards the conceptual aspects of decision support, we may make a number of observations. First, there probably is going to be an increased role for information technology in terms of medical decision making; but such help should not be limited to the narrow concept of "computer aided diagnosis". Management of individual patients depends not only on diagnosis but on an accurate assessment of patient risk; and many doctors who feel they need no help with diagnosis are extremely happy to make use of risk evaluation systems. Those systems which are currently available are becoming increasingly accurate; and we may expect to see considerable expansion of this aspect of decision support in the future.

92

References 1 Henriques, J. (1992) Surgical risk evaluation in the elderly. Presented at the 1st Congresso Medicina Medico-Militair, Rio de Janeiro, Brasil. Unpublished. 2 Knaus, W., Knaus, A., Williams, Z.E., Wagner, P.L. (1981) APACHE a phsiologically based classification system. Crit. Care Med. 591-597.

1. Henriques / Technology and Health Care 1 (1993) 89-92

3 Siegel, J.H., Arrahan, R.I., Samir, D., Goodaez, S. (1990) Early physiologic predictors of injury severity and death in blunt multiple trauma. Arch. Surg. 125. 498-508. 4 Vacanti, C.J. (1970) A stistical analysis of the relationship of physical status to postoperative mortality in 63,388 patients. Anesth. Analg. 49. 564.

Clinicians' needs and requirements. 2. Conceptual comments.

Clinicians' needs and requirements. 2. Conceptual comments. - PDF Download Free
506KB Sizes 3 Downloads 5 Views