Clinical Usefulness of Self-Rated Olfactory Performance—A Data Science-Based Assessment of 6000 Patients

• Published in: Chemical senses Vol. 44; no. 6; pp. 357 - 364
• Main Authors: Lötsch, Jörn, Hummel, Thomas
• Format: Journal Article
• Language: English
• Published: England 17.07.2019
• Subjects: olfactory testing; data science; odor identification; anosmia; diagnostic test performance
• DOI: 10.1093/chemse/bjz029
• ISSN: 0379-864X; 1464-3553

In clinical practice, with its time constraints, a frequent conclusion is that asking about the ability to smell may suffice to detect olfactory problems. To address this question systematically, 6049 subjects were asked about how well they can perceive odors, with 5 possible responses. Participants presented at a University Department of Otorhinolaryngology, where olfactory testing was part of the routine investigation performed in patients receiving surgery at the clinic (for various reasons). According to an odor identification test, 1227 subjects had functional anosmia and 3113 were labeled with normosmia. Measures of laboratory test performance were used to assess the success of self-estimates to capture the olfactory diagnosis. Ratings of the olfactory function as absent or impaired provided the diagnosis of anosmia at a balanced accuracy of 79%, whereas ratings of good or excellent indicated normosmia at a balanced accuracy of 64.6%. The number of incorrect judgments of anosmia increased with age, whereas false negative self-estimates of normosmia became rarer with increasing age. The subject’s sex was irrelevant in this context. Thus, when asking the question “How well can you smell odors?” and querying standardized responses, fairly accurate information can be obtained about whether or not the subject can smell. However, this has to be completed with the almost 30% (355 subjects) of anosmic patients who judged their ability to smell as at least “average.” Thus, olfactory testing using reliable and validated tests appears indispensable.