Scientific Methodology in Clinical Medicine

Abstract

In the methods used to evaluate prevention of disease, epidemiologic teams have generally not had the scope or size to obtain and investigate appropriate, large, healthy populations. In the prevention of clinical progression for an established disease, the nusances of natural history in many neoplastic and chronic illnesses are still not known well enough for clinicians to divide patients into subgroups with comparable prognoses. The agents of therapeutic trials are therefore often poorly assigned because the patients are allocated by methods that are more haphazard than random. In the alteration of illnesses, the major defect of contemporary clinical methodology is an unsatisfactory assessment of post-therapeutic changes in symptoms and signs. These changes often are either neglected in favor of laboratory data, or are evaluated by informal, unstandardized techniques of "clinical judgment". For treatment to be assessed scientifically, symptoms and signs must receive a formal status as indices of therapeutic response, and changes in the indices must be interpreted with specific criteria. Medical records, often ignored or disdained in therapeutic science, can be markedly helpful in clarifying aspects of natural history that cannot be feasibly determined by any other mode of investigation. The records can be used effectively in prognostic research by applying appropriate chronologic analyses to avoid the hazards of retrospection, and by recognizing the importance of qualitative clinical observations.