Numerical Analysis of Clinical Experience

Abstract

In this chapter, we consider some history of numerical analysis applied to clinical observations on individuals rather than to the massed data of the official statistics. The history of such numerical analysis is short; perhaps 1790 could be chosen as a beginning, with the works of Alexander Gordon (§9.1) and Philippe Pinel (§16.2) following soon after. Some reasons for this late development can be given:

Specificity of disease. The difficulties of classifying diseases to form homogeneous groups for comparisons could only be solved after the bacteriological revolution initiated by Pasteur and Koch in the last quarter of the nineteenth century; before that time, diseases were not thought of as entities (see §8.6 and Chapter 16).

Individuality of patients. There were doubts that patients could be classified in a meaningful way, the academic view being that the individuality (variance) of the patient was of a greater order than that of the disease, for example, Double’s view as set forth in the debate in the French Academy of Science (see §§17.2 and 17.10).

Numbers of patients observable. For comparisons between treatments of any disease, large numbers of patients are usually necessary, with the size of the necessary experience being dependent on the relative effectiveness of the treatments. The assembly of such large numbers would have been possible in the great hospitals but we have seen that conditions in them were usually chaotic.

Lack of effective medicines. In the early nineteenth century, there were few effective remedies; often the supposed remedy was actually harmful.

Opposition from the academic physicians. There was a general attitude that “mathematics” did not have anything to give to medicine.