Criteria for valid preclinical trials using animal stroke models.

Stroke welcomes Letters to the Editor and willpublish them, if suitable, as space permits. They should not exceed 1,000 words (excluding references) and may be subject to editing or abridgment. Please submit Letters to the Editor in duplicate, typed double-spaced. Include an appropriate title and a fax number for the corresponding author. The American Heart Association requires that all authors sign a Copyright TransferAgreement before publication. For a copy of this agreement, contact the Stroke Editorial Office. The editorial by Hsul presents rigorous criteria for valid animal studies of stroke treatments. These criteria could also be advocated for all animal experiments. Consequently, a careful examination of the cost-benefit relationship of his recommendations is in order. Dr Hsu is concerned about observer bias and about bias by the person producing an intervention (eg, carotid ligation). He believes unconscious bias could influence outcome if the person measuring end points knows the treatment or if the performer of the intervention knows whether the animal will be treated. In human studies there are two reasons for "blind" doctors. Not only may their bias affect their measurements, but through a placebo effect, it may actually influence the performance of the patient. The placebo effect is not a factor in animal studies. Consequently, we have only the risk of observer bias to worry about. In most instances this can be avoided by objective end points supplemented at least initially by ignorance on the part of the observer concerning "expected" or hypothesized outcome. In histological studies complete observer blindness can be maintained without hiring additional staff, simply by coding slides. For physiological studies there would often be the expense of hiring staff to make observations with no knowledge of treatment or hypothesis concerning drug effect. The unconscious systematic bias-driven procedural errors (eg, giving "short-weighted" injections ; ligating vessels partially rather than completely) seem rather far-fetched to me. In the case of stroke models, an objective measurement of flow immediately after ligation would certainly reveal biased preparation of animals and is a mandatory part of the experimental design. Dr Hsu has a rather cavalier approach to current funding problems. He says that since large sums are already spent for human trials, the National Institutes of Health ought not balk at expanded spending for the animal experiments to enable us to hire the additional staff required for blinded studies. This is simply unrealistic. Investigators should eliminate as many opportunities …