*3.4. Physiological Measures*

The major physiological measures investigated were those reflecting fear or arousal, via the hypothalamic–pituitary–adrenal (HPA) axis response [2,7–11,13,24,31,35,36,38] and the associated release of stored glucose [1,22,23,25,29,32,34,37]. Stress almost invariably activates the HPA axis, which, via a sequence of steps, leads to glucocorticoid production—the principle rodent glucocorticoid being corticosterone. Corticosterone serves to regulate glucose, allowing for its release from reserves in the short term, and carrying out other metabolic actions with the goal of establishing homeostasis [48]. Typically, blood serum or plasma is used to measure corticosterone [49] or glucose concentrations [50], with an increase implying an acute stress response. Measurement of faecal corticosterone metabolites (FCM) has recently been proposed due to the advantage of being non-invasive. FCM provide a retrospective measure of the HPA axis response with lag time from peak in blood to faecal excretion being 9 h in mice [51], thus reflecting sampling method and subsequent recovery [2].

### *3.5. Clinical and Pathological Parameters*

General health parameters such as fur condition, inactivity or dehydration status are commonly implemented in severity score sheets for rodent experiments. A change in these parameters can provide an indication of reduced welfare or disease but may not on their own be a sensitive indicator of a stress response. Studies evaluating welfare typically combine these measures with other physiological or behavioural parameters, as occurred in the included studies. Procedure-specific adverse events were important to include as outcome measures, since these may have considerable impact on individual welfare. Furthermore, an increased incidence of such events would prevent recommendation of a technique for practice. Events considered in the studies included haemorrhage from the ear and nares, ocular lesions, circling and convulsions.

Mortality rate is a commonly used indicator providing a retrospective measure of welfare, since it may be influenced by disease, trauma or environmental problems [52].

Other measures utilised are arguably less indicative of animal a ffective state, but may provide insight by inference—for example, the quantification of tissue damage through post-mortem or histology [1,4,8,24,27–29,32,33,36,38] may imply pain or loss of function.
