Carryover bias in clinical investigations.

If the effect of a treatment continues after the treatment is withdrawn then the response to a second treatment may well be due in part to the previous treatment. This, so called, carryover effect may bias any type of study in which subjects are tested more than once. Crossover studies can be routinely checked for this bias. In other study designs, however, common sense and alertness for unusual patterns in the data are the only defenses against it. The amount of carryover bias can be somewhat minimized by the following measures. Dose-response studies, dose-titration studies, and open-evaluation studies should require a sufficient washout period between the administrations of drugs. The use of duplicate standard deviations for the estimation of intraindividual reproducibility of a test should always be combined with a statistical test for differences between the duplicate data. Subjective variables, which are frequently influenced by psychologic carryover effects, should be validated together with objective variables whenever possible. In spite of these measures, many cases of carryover effect remain unpreventable. We shall simply have to accept them.