RANDOMIZATION BY GROUP IN STUDYING THE EFFECT OF DRUG INFORMATION IN PRIMARY CARE

Drug information is a technology which is rarely evaluated. Practical and ethical considerations limit the use of a classical experimental method (a randomized controlled trial) in studying the effect of drug information in primary care. An alternative approach, randomization by group, is preferable for several reasons: it avoids contamination of the control group; the effect of information can be evaluated in the natural working environment; and the impact of information is increased via diffusion. This article describes the selection of a control group, the Hawthorne effect and 'blindness' in information experiments. Sample size tables and power calculations are presented when randomization by group is used. The study power is influenced by the number of health centres and the variance between them. The number of doctors per health centre plays a less important role, and the number of patients is relatively unimportant. There is also a need to use qualitative methods to prepare information and to understand factors influencing change of behaviour among prescribers.