R. J. Stusser, MD

Division of Clinical Biostatistics and Epidemiology,

Clinical Research Centre, Havana, Cuba

The formation of scientific hypotheses and the design of statistical hypotheses are essential to modern clinical and surgical research: in fact, the design of statistical hypotheses depends upon the formation of scientific hypotheses that are based on facts coherent to clinical theory and logic. The aim of this work is to found why and how should be much more focused scientific hypothesis formation before the statistical hypothesis design in clinical research, in order to make the advanced statistics methods scientifically more effective. Over the past 55 years, the application of efficient experimental, quantitative, epidemiological and computing algorithmic methods has furthered the progress of confirmatory applied clinical trials and surveys. At the same time, however, the development of observational, qualitative, clinical, and pre-computing heuristic methods for exploratory, descriptive, and explanatory clinical investigations have remained virtually stationary. Several factors could explain this. Because clinical judgement is used primarily to discover more operational scientific hypotheses about a particular state or process of an individual patient during clinical care, it has lost appreciation for its value in the discovery of more conceptual scientific hypotheses about general states or processes in one or few more patients, hypotheses that could be useful in a broader clinical sense. Although part-time clinical investigators now have better scientific training, most of them still lack adequate conceptual and general scientific background for training in scientific method, logic, as well as in statistics and computation, to know how much they can expect of the latter so developed technologies. If these clinical investigators fail to do their own fundamental clinical research, they will lose the scientific skills necessary to perform the rigorous clinical trials and surveys suggested by the general scientific hypotheses developed by the full-time experimental and epidemiological investigators. To help formulate general scientific hypotheses, computer programs have been proposed to use searching algorithms of scientific literature databases. Although this approach is useful, some suggestions are important: 1) heuristic rules about clinical judgement have to be studied and made more explicit, 2) training programs in clinical research process must offer a more thorough conceptual background for scientific method and logic, including statistics and computation, and 3) there must be full-time clinical investigators who are dedicated to the formulation of new scientific hypotheses that aim at improving basic knowledge in the clinical sciences. Each of these will help the clinical judgement method regain an appreciation for being able to discover new general scientific hypotheses. To improve the efficiency of clinical research projects and programmes, the clinical biostatistician, the experimentalist, and the epidemiologist must help the clinical and surgical investigators study and develop scientific heuristic rules to formulate general scientific hypotheses within the framework of clinical judgement. By drawing on progress in the basic research practice of clinical sciences, the now implicit procedures used to discover new general scientific hypotheses will become explicit, and their future formalisation and computerisation will become possible.

Integration Science Program for Medicine, Nursing, Technology & Health
Back to the Index Page