xRead - Olfactory Disorders (September 2023)

20426984, 2022, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/alr.22929, Wiley Online Library on [04/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

526

INTERNATIONAL CONSENSUS ON OLFACTION

tures. However, compared with identification tests, they require more administration time, are typically of lower reliability, and are limited in terms of the spectrum of odor ants that can be evaluated. Despite the fact that variations in intertrial intervals do not meaningfully impact thresh old values, the procedures used to present the odorants, such as volumes of sniff bottles, do have such impact. 1106 Although, in general, persons with high thresholds (ie, low sensitivity) to one odorant tend to have high thresholds to other odorants, and vice versa, this is not the case with all odorants. This is particularly evident for odorants for which some people are relatively insensitive (ie, so-called specific anosmias). Unfortunately, the concepts of detec tion and recognition are commonly confounded in thresh old test procedures (eg, having a patient smell a higher con centration of a threshold series so the odor can be identi fied and then claiming detection thresholds are being mea sured), thereby increasing variability. 1107 Failure to provide specific instructions can lead to such confounding. Thresh old tests can be frustrating for patients given that many tri als are weak or below threshold, leading even those with a normal sense of smell to believe they performed poorly on the test. It is commonly stated that threshold tests are solely a measure of peripheral, ie, epithelial, OF. However, this is clearly not the case. Even detection threshold tests require cognitive processes such as working and short term memory (eg, discerning a stimulus from blanks in a temporal sequence 1108 ) and are impacted by top-down centrally mediated decision processes. 1108 Indeed, thresh old tests, like tests of odor identification and discrimina tion/memory, have been shown to correlate with neuropsy chological measures of verbal and visuospatial memory. 852 Importantly, threshold measures are sensitive to lesions in higher order brain structures such as those observed inAD, 549 multiple sclerosis, 1109 and epilepsy. 424 Moreover, given the greater variability and lower reliability of most threshold tests compared with identification tests, observa tions of weaker cognitive associations with threshold tests than with identification tests do not necessarily imply a meaningful differential cognitive load. Methods to obtain threshold measures vary, and, despite assumptions often made by regulatory agencies, there is no single threshold value for a given odorant. Hence, like other psychophysical measures, threshold values depend on the procedures employed in estimating them and mul tiple subject factors including age and sex. In the method of constant stimuli, a range of odorant concentrations are randomly presented and an ogive-like function (cumula tive frequency graph) is fitted to the stimulus-response function (concentrations on the abscissa and performance, eg, percent trials that are correct, on the ordinate). When a blank comparison is provided at each concentration in

a forced-choice task, the concentration where 75% per formance occurs is commonly calculated as the thresh old, since by chance alone 50% of the trials would be per formed correctly. Although this method can also provide information about an odorant’s psychophysical dynamic range, ie, the sharpness of the buildup in performance among a given concentration gradient, only rarely is the method of constant stimuli used clinically. This is because of the need for a large number of trials to obtain a reliable measure. Nonetheless, this is the gold standard method to which other threshold tests are commonly compared and there are a few clinical applications of this technique. In the initially ascending methods of limits procedure, stim uli are started at below-threshold concentration levels and then increased in concentration until they are detectable. Repeated trials are required. This approach has been cod ified as the ASTM International E679 procedure. 1110 Ver sions of this procedure have employed methods to blast boluses of odorants into the nose to minimize impact of sniffing or breathing, the so-called blast-injection tech nique. In initially ascending series staircase procedures, stimuli are increased in concentration from below thresh old levels systematically until they are detected, then decreased and increased according to the correctness of the individual’s responses within the perithreshold region. An average of the reversals, ie, points of upward or downward transitions, provides the threshold estimate. Although double staircase procedures, 1111 ie, procedures in which two staircases are performed simultaneously (one initially descending from higher concentrations and the other initially ascending from lower concentrations) are commonly used in other sensory systems and are gener ally preferable, 1112,1113 they are rarely employed in olfaction because of time considerations and concerns about adap tation. In general, staircase procedures are preferred over other methods, resulting in relatively stable and reliable thresholds with a minimum number of trials. 1114 Signal Detection Tests Signal detection tests require individuals to differentiate between low levels of an odorant, usually a single concen tration established for each patient separately, and blank stimuli, although subtle quality differences between stim uli can also be measured. Instead of conceptualizing sen sitivity as a border between no sensation and sensation, as occurs in threshold measurement, signal detection the orists view the detection task as discriminating between noise and signal plus noise. Signal is viewed largely as a constant, whereas noise reflects physiological and psycho logical variations of the individual, including the liberal ness or conservativeness of the individual at any one time