The preceding article proposed an assured treatment design that would address certain difficulties in recruiting persons who are at greater risk into randomized clinical trials.
The purpose of this article is to illustrate the statistical validity of the design in a pratical setting.
Three actual randomized clinical trials were considered as case studies ; in each, the data that would have been obtained under assured allocation were identified.
Then, with only these data, together with a reasonable choice of model describing the reponse of subjects under standard treatment as a function of initial severity, the treatment effect was estimated for the subjects at greater risk.
The estimates were compared with conventional estimates for the sicker patients randomized in the original trials.
In each case, the estimates produced in the assured treatment trial were close to those observed in the randomized trial.
Risk-based allocation trials deserve serious consideration when randomized clinical trials are difficult or impossible to execute.
The proposed designs to offer persons at greater risk assurance that they would receive the new treatment, while researches would retain the ability to draw valid statistical conclusions about treatment efficacy.
Mots-clés Pascal : Nouveauté, Hypolipémiant, Zidovudine, Cardiopathie coronaire, Recrutement, SIDA, Virose, Infection, Essai clinique, Distribution, Risque élevé, Homme, Méthodologie, Traitement, Chimiothérapie, Pyrimidine nucléoside, Appareil circulatoire pathologie, Immunopathologie, Immunodéficit, Modèle statistique
Mots-clés Pascal anglais : Novelty, Antilipemic agent, Coronary heart disease, Recruitment, AIDS, Viral disease, Infection, Clinical trial, Distribution, High risk, Human, Methodology, Treatment, Chemotherapy, Pyrimidine nucleoside, Cardiovascular disease, Immunopathology, Immune deficiency, Statistical model
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 96-0273301
Code Inist : 002B30A08. Création : 199608.