Biocompatability studies usually involve a control group of animals (untreated and/or formulation treated) and usually two or more treated groups with different levels of exposure to the device of interest with exposure approximately that used in man (as closely as possible). Occasionally there also may be recovery groups to determine if any observed effects are reversible (and if so, to what extent) In most instances the high dose level is expected to elicit some toxic effects in the animals, often expressed as decreased food consumption and/or below-normal bodyweight gain, and has been selected after consideration of earlier data, perhaps from dose range-finding studies, or at least to dose as high as possible by the intended route. The other two dose levels are anticipated not to cause toxic effects. Generally, but not always (e.g., non-steroidal anti-inflammatory drugs in rodents), the “low” dose level is a several-fold multiple of the expected human therapeutic or exposure level (preferably five-fold for non-rodent and ten-fold for rodents). However, without knowing the true relationship of these dose levels to each other with respect to the absorption, distribution and elimination of the new chemical entity as reflected by its pharmacokinetics, it is difficult to see how meaningful extrapolations concerning safety margins can be made from the toxicity data obtained. Also, without pharmacokinetic data from the positive control group, its inclusion is of limited value and the results obtained could lead to erroneous conclusions.