The logical description of a disease class as a Boolean function with special reference to the irritable bowel syndrome.

Although some disease classes can be defined by a single defining characteristic, in terms of traditional logic one that is both necessary and sufficient, other disease classes, such as the so-called "irritable bowel syndrome' (IBS), have only a set of symptoms, signs etc. which are said to characterize it in some undefined way. This characterization might be made definite by using simple Boolean algebra to give a logical description of the disease class. This method can only be used if the individual doctor can be shown to be consistent in his assignment of a set of data to the disease class and also if his assignment agrees independently with that of another doctor. To test such consistency and concordance, 100 case records were collected of supposed IBS and not-IBS; 20 of these were replicated to test consistency, and the total of 120 records was examined by five consultants independently, who assigned them either to an IBS or a not-IBS class. From the 1500 pair-wise comparisons, analysis revealed that the disagreement in assignment between doctors was only slightly greater than the disagreement within doctors, suggesting that the group of doctors acted as if there were an implicit description of IBS with which each of them broadly agreed. This implicit description, if made explicit, could form a basis for diagnosis by logical implication. This method of logical definition has general application in medicine and a set of defined cases could be used to establish entry criteria for multicentre trials of a previously ill-defined class.

Mathematics and medicine. Test reduction: III-Practical applications.

While there is no generally applicable method of test reduction, ways in which the problem can be tackled can usually be suggested. These include the examination of error rates of tests and the use of current statistical methods. When the cost-effectiveness of some more expensive test has to be measured, a model of the decision problem needs to be developed, with an attempted estimation of the values or utilities of the states of health that result from treatment.

Test reduction: I--Introduction and review of published work.

The word "test" is used in a more general sense to describe the process of eliciting evidence of any kind from a patient, and this paper explores the thesis that much evidence is unnecessary and that therefore much test reduction is possible. The value of test reduction can be measured by its efficiency--that is, its effect on the misclassification of disease--or, preferably, by some measure of its cost-effectiveness.

Evidential value of the hospital record in clinical decision making.

A simulated retrospective exercise in the diagnosis and management of 53 readmissions to a gastrointestinal unit was undertaken by two consultants. Diagnosis of the illness at readmission was made on evidence sought from a referee, who also supplied, on request, items of relevant evidence from the past medical record. Patient management was agreed from these sources. For each item of evidence the evidential weight, the irrecoverability, and the expected benefit accruing to the patient of its availability was calculated. It was concluded that the evidence worth recording in the event of subsequent hospital admission could be largely specified for each diagnosis and each operation. It would be brief and could be numerically coded.

What is the monetary value of a human life?

The resources available to the health service are limited and so the amount the NHS can spend on saving human life is also limited. Rational allocation of resources requires a decision theory model, which in turn demands some monetary valuation of human life. Each of three approaches discussed--basing value on productive capacity, the NHS's implied values, or individuals' values--rests on an underlying set of non-monetary values. Choice of the underlying value will determine the method to be used in placing a monetary value on life. As the Health Service implicitly places certain values on life already, a means of making this valuation more rational and explicity can only improve the quality and quantity of health care.

Computer interrogation of patients.

A system of routine interrogation of patients using a computer has been developed. It includes a visual display unit with a specially designed response keyboard, and the program has been designed to adapt to the individual patient. The system was evaluated objectively, using the criteria of accuracy in eliciting symptoms, acceptability to the patient, and cost. While doctors will always take the ultimate management decisions, it seems that machines can be programmed to undertake the routine interrogation of patients, elicit evidence accurately and acceptably, and calculate the probabilities of disease as effectively as doctors.

Development of a formal structure for clinical management decisions: a mathematical analysis.

Decision theory and the calculating power of the computer now enables us to contemplate the development of formal methods for making decisions about clinical management. In the simplest model, it is first necessary to define all treatment decisions as an exhaustive and mutually exclusive set and similarly to define the set of consequences or outcomes of treatment. The probability of each outcome conditional on treatment has to be estimated and this consequent state of health has to be quantified as a utility. Possible methods of estimating utilities of states of health are discussed and the construction of a unidimensional utility function based on a sequence of wagers. The states of health consequent on severe brain damage can only be described multidimensionally and the model has to be extended to include this case. While such a model would allow simple treatment decisions to be formalized, it could not decide whether the cost of treatment was worth while nor whether it would pay to carry out further investigative tests and thus buy more evidence. If these additional variables are to be included in the model, it is necessary to introduce the motion of an equivalence between monetary values and utilities. This implies attaching a monetary value to any given state of health.