Odorant confusion matrix: the influence of patient history on patterns of odorant identification and misidentification in hyposmia Daniel B. Kurtz a, *, Theresa L. White a , Paul R. Sheehe a , David E. Hornung b , Paul F. Kent a a Departments of Neuroscience, Physiology, and Neurology, SUNY Upstate Medical University, 750 East Adams St, Syracuse, NY 13210, USA b Biology Department, St. Lawrence University, Canton, NY 13617, USA Received 24 March 2000; received in revised form 5 October 2000; accepted 4 December 2000 Abstract The odorant confusion matrix (OCM) is an odorant identification test in which the number of correct odorant identifications quantifies the level of olfactory function. As with other confusion matrices, the OCM reflects distortions of sensory perception as errors in identification. Previous work with the OCM suggests that, within an individual, hyposmia is associated with a stable shift in odorant perception. The current study examined whether consistent shifts in odorant perception are also characteristic of the various pathologies that lead to an olfactory loss. In a retrospective study, OCM response patterns for 135 hyposmic patients were fit into a five-dimensional space in which the distances between subjects reflected the dissimilarities between their OCM response patterns. Multivariate regression was performed relating position in the five-dimensional space to each of 11 factors representing 33 demographic and medical history variables. One factor, named congestion (gathering the variables of past polyposis, current polyposis, and current nasal obstruction due to swelling), was significantly indicative of patterns of responses on the OCM, independent of the level of hyposmia. These data suggest that conductive olfactory loss may be associated with alterations in odorant perception, which are reflected in consistent odorant confusions. Such alterations in perception may eventually serve as a basis for a clinical test to provide differential diagnoses as to the sources of olfactory losses. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Olfaction; Hyposmia; Confusion matrix; Multivariate regression; Multidimensional scaling; Information theory; Odorant identification; Polyposis; Conductive olfactory loss; Aging To date, the role of clinical olfactory testing has been to provide a quantitative estimation of the level of loss in a patient's sense of smell. Such evaluations have been used to document a reported deficit and, when it occurs, follow the time course of recovery. The most commonly utilized clinical tests have been measurements of odorant threshold and odorant identification. In general, threshold tests estab- lish the minimum amount of an odorant that a subject can perceive. In the clinical evaluation of dysosmia, the stimulus for the threshold test is traditionally butanol [4] or phenyl± ethyl alcohol [28], but thresholds may be examined for any odorant. For example, a recent variant of the threshold test uses propanol [9]. Although threshold tests can be useful in clinical olfactory evaluations, they have three intrinsic difficulties. The first of these difficulties is that an inability to detect the presence of a particular odorant does not necessarily indicate anosmia to all other odorants [1]. The second difficulty is that normal performance on a threshold test does not imply that a patient is free from olfactory complaints, as the ability to identify odorants may still be impaired [32]. And, finally, because of difficulties in con- trolling stimulus concentration, olfactory thresholds seem to be more variable than threshold measurements for other sensory systems [3]. Consequently, threshold tests are gen- erally used in conjunction with tests of odorant identifica- tion, such as the University of Pennsylvania Smell Identification Test (UPSIT). The UPSIT has proven to be a valuable clinical tool because of its ease of administration, extensive normative data [12], and high reliability [11]. In addition, both the Connecticut Chemosensory Clinical Research Center Smell Test [5] and the Medical College of Virginia Olfactory Screening Test [8] combine odorant * Corresponding author. Department of Neuroscience and Physiology, State University of New York Upstate Medical University (formerly the Health Science Center at Syracuse), 750 East Adams Street, Syracuse, NY 13210, USA. Tel.: +1-315-464-5396; fax: +1-315-464-7712. E-mail address: kurtzd@mail.upstate.edu (D.B. Kurtz). Physiology & Behavior 72 (2001) 595 ± 602 0031-9384/01/$ ± see front matter D 2001 Elsevier Science Inc. All rights reserved. PII:S0031-9384(01)00410-3