and other steroids in samples from saliva (Aardal and Holm 1995; Kiess et al. 1995) and plucked hair (Davenport et al. 2006). Even less handling is involved when samples are taken from voided urine, feces, expired air, and shed hair (Poon and Chu 1999), if these methods are validated for the species under study. Other noninvasive techniques for data collection include sound recordings (Holy and Guo 2005), cameras (Hobbs et al. 1997), or noninvasive, sensor-laden apparel simply worn by the animal (Jarrell et al. 2005).
Validated endpoints that occur earlier in the course of the protocol and involve no detectable indication of disease, injury, or abnormal behavior can prevent or minimize distress in experimentation and testing. The use of humane endpoints (i.e., “end a study earlier to avoid or terminate unrelieved pain and/or distress”; Stokes 2000) or surrogate endpoints (i.e., those that can reliably substitute for more distressing or painful phenomena) is especially applicable in scientific disciplines that focus primarily on molecular and cellular phenomena associated with disease (Morton 2000; Hendriksen and Morton 1999). In these cases, biochemical changes may be detectable at early stages in the disease process, prior to the manifestation of clinical signs consistent with distress. For example, elevated white blood cell counts are detectable in leukemia models before illness becomes obvious and serum biochemical values often change in early stages of toxicity before animals appear ill (Poon and Chu 1999). Thus, taking measurements or collecting samples from animals before the appearance of any clinical signs (including all clinical manifestations, not only those related to distress) is desirable, especially when the signs themselves are not the study’s focus. In such cases, the predictive value of validated endpoints may permit early euthanasia of these animals and postmortem collection of data or samples (for additional information see Appendix). Alternatively, a clinically normal animal could be anesthetized before a distressful procedure and euthanized before regaining consciousness.
Familiarity with certain procedures or experimental protocols often allows for predicting the course of adverse clinical signs and distress. In many instances death results from indirect effects such as dehydration and is not related to the response variable under study. In mice, for example, progressive hypothermia due to low food intake will cause an animal’s death over several days. However, distress can be minimized through the use of validated humane endpoints, such as euthanizing the animals at the first recording of low body temperature (Morton 1998; Soothill et al. 1992). The choice and use of endpoints should be part of the experimental protocol whenever possible.