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

527

PATEL et al.

in reporting the presence or absence of the signal, ie, the individual’s response criterion. The advantage over thresh old testing is that signal detection analysis can indepen dently differentiate an individual’s response criterion from his or her sensitivity, per se. Thus, a more emotional indi vidual may believe that they perceive a stimulus but the response actually reflects greater liberalness in reporting its presence. Such tests are exquisitely sensitive to very subtle deficits in smell function, but typically take more time than threshold tests given the large number of trials needed for stable measures and the need to titrate the stim ulus concentrations for each individual. Moreover, norma tive data for olfactory signal detection tests are lacking. Some shorter signal detection tests have been employed clinically. Reliability of Olfactory Test Measures In general, the more items or trials in an olfactory test, the higher its reliability, ie, measurement consistency over time. 1116 Reliability is a prerequisite for validity. However, reliability coefficients, which are the main measure of such consistency among individuals of a group, depend on the variation in test scores and can be misleading when distri butions of scores are restricted, eg, by being grouped into too few categories. Although test-retest reliability coeffi cients are reported for numerous tests, differences among such coefficients are rarely assessed for statistical signif icance. In a study in which this was done, the reliabil ity coefficients of tests that ranged from 0.90 to 0.76 did not differ significantly from one another. 1116 These coef ficients did differ from those ranging from 0.71 to 0.67, which, in turn, differed significantly from those ranging from 0.53 to 0.43. Hence, when subtle differences in reli ability coefficients are reported among tests, one cannot assume that the differences are statistically meaningful. That being said, reliability coefficients are among the few metrics to which tests can be compared, and, despite con founding factors, need to be considered in context when choosing a test for administration. Reliability coefficients are a guide, but not the sole determinate of the value of an olfactory test, and comparisons among tests can be enig matic. As can be seen in Table VIII.6, of 73 tests that were surveyed, a significant number failed to provide this very basic psychometric measure. Relationships Among Nominally Different Types of Olfactory Tests In general, tests of odor identification, detection, discrim ination, and memory are correlated with one another (Tables VIII.10-VIII.12), with the sizes of the correlation

being theoretically bound by the less reliable test and the range of test scores used in the computation. Because of such relationships, many authors default to the most reli able of the tests as the only needed indicator of smell func tion. While a case can be made that nominally different tests may be differentially sensitivity to a number of dis orders, for most practical purposes, more than one type of test is not needed. Despite their being correlated, comparison of results from nominally distinct tests must be interpreted con servatively, since different psychophysical tests rely on several odorants at variable concentrations, have differ ent cognitive demands, 1117,1118 and vary in terms of their reliabilities. 1116,1119 In one study employing SS felt-tip pen markers to present stimuli, demographic and cognitive fac tors accounted for 15% of the variance in odor identification values, 23% of the variance of discrimination values, and 9% of the variation in threshold values. 1118 It is important to recognize that operational terms used to describe olfactory tests (eg, detection, identification, dis crimination, memory) are not pure representatives of inde pendent physiologic or psychologic chemosensory pro cesses signified by their names. 1105 The correlations among such tests are a testament to this fact (Table VIII.10). For example, if an odor is to be identified or remembered, it must first be detected. The ability to remember odor qualities is a prerequisite for discriminating among them, assuming they are of equivalent intensity. Discrimination requires discerning odor qualities although identification is not required. As noted earlier, even threshold tests rely on some level of cognitive processing. Unilateral or Bilateral Testing? In general, bilateral tests reflect the better functioning side of the nose and for this reason are not sensitive to unilateral deficits. Testing each side of the nose is useful for detect ing deficits confined to one side of the nose, although, in most cases, deficits are bilateral and unilateral testing can be confounded by the nasal cycle, which impacts airflow to the OC in some individuals. A common way to test each side of the nose separately is to occlude the nontested side with a piece of tape. Microfoam tape (3M Corporation) is commonly used since it is odorless, easy to apply and remove, and leaves no residue. Normative unilateral data are available for some tests. General Recommendations The choice of an olfactory test depends on the purpose that is intended. In general, forced-choice tests of odor identification are preferred to other types of tests based on