Abstract
Background
Olfactory dysfunction (OD) can occur from a variety of etiologies. However, there are few reports examining whether varying etiologies have unique profiles of psychophysical testing that may provide insight into the pathophysiology of OD.
Methods
Adults with chronic rhinosinusitis with and without nasal polyps (CRSwNP/CRSsNP) and healthy control adults with no sinus complaints underwent olfactory assessment with Sniffin’ Sticks. Profiles of identification and discrimination were compared between CRS and non-CRS subjects across the spectrum of OD.
Results
Normosmics with or without CRS identified apple, pineapple, and turpentine less frequently than expected (range 52%-68% correct). Hyposmics with CRS correctly identified orange more frequently than control hyposmics (83%-93% vs 68% for controls) with similar findings for rose. Hyposmics of all cohorts were unable to identify apple (26%). Discrimination profiles were similar across the spectrum of OD and between diagnostic groups.
Conclusions
Identification and discrimination rates of specific odorants may provide unique information regarding the etiology of OD, however psychophysical testing is a complex interplay of olfactory and trigeminal function, the strength of target odorant, distractor choices, and familiarity with odorants.
Keywords: olfaction, olfactory disorder
Introduction
Otolaryngologists are familiar with looking at patterns of sensory function and testing, most commonly with audiograms. Loss of function at certain stimulus frequencies predicts etiology, anatomic localization of pathology, and subsequent treatment of hearing loss. For example, 2 patients may have normal pure tone thresholds at 50% of tested frequencies, but one with Meniere’s disease usually presents with low-frequency sensorineural loss, while another with presbycusis presents with high-frequency sensorineural loss. So while the overall percentage of normal responses is similar, patterns of normal and abnormal function for specific stimuli provide insight into the pathogenesis and subsequent treatment of sensory loss.
Similar to hearing loss, olfactory dysfunction (OD) can occur from a wide variety of etiologies, however, the relationship between patterns of olfactory testing and etiologies for OD have not been studied. 1 Patients with impaired airflow to specific regions of the olfactory cleft, such as chronic rhinosinusitis (CRS) with or without nasal polyps (CRSwNP/CRSsNP), may have differential impairments for specific odorants, depending upon where in the olfactory neuroepithelium these odorants are detected and how the local inflammation triggers olfactory neuronal cell death. In contrast, patients without CRS may have normal airflow, but experience an olfactory loss due to widespread neural degeneration, fibrosis, or central processing deficits. One could hypothesize that those patients with underlying global neural degeneration may have similar impairments across all odorants. In addition to differential impact upon olfactory neuron function, detection of many odorants occurs as a complex interplay between olfactory and trigeminal stimulation. 2 Thus, it is plausible that etiologies that differentially impact trigeminal function may also have unique olfactory testing profiles. The goal of this study is to investigate unique individual odorant profile responses between CRS and control patients.
Materials and Methods
Study Design and Sample Population
Study enrollment was completed as part of an observational research study of human subjects funded by the National Institute on Deafness and Other Communication Disorders (Bethesda, MD; Federal grants R01-DC005805 and 5R01DC019078). Study participants were prospectively recruited from a patient population presenting to academic, rhinology practices located in the United States at Oregon Health and Science University (OHSU, Portland, OR), the Medical University of South Carolina (Charleston, SC), the University of Utah (Salt Lake City, UT), the University of Colorado (Aurora, CO), and the University of Virginia (UVA, Charlottesville, VA). The Institutional Review Board affiliated with each enrollment location approved study protocols, annual review, and clinical data safety. Patients provided written, informed consent after initial enrollment meetings to ensure voluntary participation involving minimal risk and no deviations from the standard of care (SOC) for the treatment of CRS.
Diagnosis and Treatment Modality
Adult case study participants received a confirmed diagnosis of CRS, with nasal polyposis (CRSwNP) or without (CRSsNP) confirmed via sinonasal endoscopy, from a fellowship-trained rhinologist following criteria established by current clinical practice guidelines of the American Academy of Otolaryngology-Head and Neck Surgery. 3 Study participants reported experiencing at least 2 of the cardinal symptoms associated with CRS including, but not limited to: nasal congestion, mucopurulent drainage, facial pain and/or pressure, and/or impairment in olfactory function.
Control Population
Control participants were simultaneously enrolled from a community-based sample of healthy subjects within MUSC and UVA without a history of CRS or previous endoscopic sinus surgery (ESS). Control participants were prospectively interviewed and enrolled on a voluntary basis using standard, adult consent procedures. Adult, study volunteers were recruited locally using advertisements, word-of-mouth, and self-referral.
Exclusion Criteria
Study participants were not considered for study inclusion if they disclosed a history of ESS in the 6 months preceding enrollment or comorbid conditions associated with a higher prevalence of OD including previous physician diagnoses of granulomatosis with polyangiitis, sarcoidosis, dementia/Alzheimer’s disease, Parkinson’s disease, immunodeficiency, and/or major head trauma/traumatic brain injury. All participants were also required to demonstrate strong fluency in English as a first or second language.
Olfactory Dysfunction Measurement
For the primary outcome of interest to this investigation, comprehensive bilateral olfactory function was assessed using Sniffin’ Stick pens (Burghart Messtechnik, Wedel, Germany)4,5 which evaluate 3 separate domains of olfactory function including (1) odorant threshold (T score range: 1–16), (2) odorant discrimination (D score range: 0–16), and (3) odorant identification (I score range: 0–16). Odorant threshold (n-butanol target). Odorant threshold (T) was evaluated in a 7-step staircase procedure using pen triplets in which odorant thresholds are detected on a continuum of dilution steps until the odorant can be accurately distinguished from 2 blanks offered in random sequence. Odorant discrimination (D) was conducted using a presented sequence of pen triplets in which 2 pens have the same odorant. Study participants are directed to identify 16 odorant targets within the presented sequence. Odorant identification (I) was evaluated using 16 different pens containing common odorants presented individually. Respondents are directed to identify the correct odorant from 4 multiple choice options. Correctly identified threshold (T), discrimination (D), and identification (I) scores, as well as a composite TDI total score, are summarized from all response scores (score range: 1–48) with higher scores reflecting superior olfactory function. Conventional diagnostic interpretations of TDI normative value scores are normosmic (>31), hyposmic (16–30), and anosmic (<15). Study participants also consented to medical history interviews during enrollment.
Database Management and Statistical Methods
Study data for case and control subjects was secured using unique study identification number assignment and removal of all protected health information prior to transfer to a centralized database at OHSU in compliance with the Health Insurance Portability and Accountability Act of 1996. Descriptive and statistical comparisons were completed using SPSS software (version 28.0; IBM Corporation, Armonk, NY). Independent, between-group, comparisons were completed using two-sided t-testing for scaled measures or chi-square (χ2) testing for nominal measures, employing either 2 × 2 or 2 × 3 contingency tables. Type-I error rates (P-values) are provided for each independent group comparison. Due to the consideration of multiple independent outcomes evaluated simultaneously, which can result in higher false positive rates, family-wise error rates were controlled using the Benjamini–Hochberg stepwise adjustments (q-values) and a conservative false discovery rate (FDR) of 5.0% to control for the potential for elevated false positive associations. 6
Results
Final Study Cohort
A total of 452 study participants met final inclusion criteria and enrolled between November 2016 and February 2020, consisting of 288 (63.7%) case subjects with CRS and 164 (36.3%) controls. Within cases with CRS, 161/288 (55.9%) were identified as having either unilateral or bilateral nasal polyposis via sinonasal endoscopy. All study participants completed Sniffin’ Sticks olfactory testing. Study participant characteristics and comorbidity of the final cohort are described and compared between all cases and controls in Table 1. Not surprisingly, there were notable differences in demographics and numerous co-morbidities often associated with CRS to include nasal polyposis, allergic rhinitis and asthma.
Table 1.
Comparison of Demographic, Comorbidity Factors, and Olfactory Function Scores Between CRS Cases and Healthy Control Subjects.
| Characteristic | CRS cases (N = 288) | Healthy controls (N = 164) | Test statistic | Unadjusted P-value |
|---|---|---|---|---|
| Age (years) Mean ± SD |
49.4 [±16.6] Range: 18–92 |
51.5 [±17.3] Range: 21–93 |
t = −1.22 | .223 |
| Males N (%) |
141 (49.0%) | 61 (37.2%) | χ2 = 5.85 | .016 |
| Females | 147 (51.0%) | 103 (62.8%) | ||
| White/Caucasian | 248 (86.1%) | 117 (71.3%) | χ2 = 23.09 | <.001 |
| African American | 33 (11.5%) | 38 (23.2%) | ||
| Asian | 4 (1.4%) | 4 (2.4%) | ||
| Hispanic/Latino ethnicity | 16 (5.6%) | 6 (3.7%) | χ2 = 0.81 | .367 |
| Co-morbidities: | ||||
| Nasal polyposis | 161 (55.9%) | 0 (0.0%) | χ2 = 142.40 | <.001 |
| Asthma | 134 (46.5%) | 14 (8.5%) | χ2 = 68.49 | <.001 |
| Previous sinus surgery/ESS | 138 (47.9%) | 0 (0.0%) | χ2 = 113.12 | <.001 |
| Allergic rhinitis | 142 (49.3%) | 49 (29.9%) | χ2 = 16.16 | <.001 |
| Allergy testing (mRAST/skin prick confirmed) | 153 (53.1%) | 38 (23.2%) | χ2 = 38.43 | <.001 |
| Allergic fungal sinusitis | 16 (5.6%) | 0 (0.0%) | χ2 = 9.45 | .002 |
| Acetylsalicylic acid sensitivity/AERD | 45 (15.6%) | 0 (0.0%) | χ2 = 28.46 | <.001 |
| Depression (self-reported history) | 84 (29.2%) | 27 (16.5%) | χ2 = 9.10 | .003 |
| Anxiety (self-reported history) | 69 (24.0%) | 20 (12.2%) | χ2 = 9.14 | .002 |
| Diabetes mellitus (Type I/II) | 30 (10.4%) | 15 (9.1%) | χ2 = 0.19 | .664 |
| OSA (self-reported history) | 61 (21.2%) | 15 (9.1%) | χ2 = 10.82 | .001 |
| Smoking/tobacco use: former | 75 (26.0%) | 44 (26.8%) | χ2 = 11.96 | .003 |
| Smoking/tobacco use: current | 9 (3.1%) | 18 (11.0%) | ||
| Alcohol use: former | 29 (10.1%) | 14 (8.5%) | χ2 = 7.39 | .025 |
| Alcohol use: current | 136 (47.2%) | 99 (60.4%) | ||
| GERD | 92 (31.9%) | 18 (11.0%) | χ2 = 24.95 | <.001 |
| Autoimmune disorders, NOS | 31 (10.8%) | 5 (3.0%) | χ2 = 8.49 | .004 |
| Olfactory scoring | ||||
| Threshold domain score | 3.8 [±2.9] Range: 1.0–11.75 |
6.1 [±2.7] Range: 1.0–11.50 |
t = −8.45 | <.001 |
| Discrimination domain score | 9.0 [±3.5] Range: 0.0–16.0 |
10.9 [±2.7] Range: 3.0–16.0 |
t = −6.50 | <.001 |
| Identification domain score | 9.3 [±4.2] Range: 0.0–16.0 |
11.8 [±2.8] Range: 3.0–16.0 |
t = −7.51 | <.001 |
| Total TDI score | 22.1 [±9.5] Range: 1.0–40.0 |
28.8 [±7.0] Range: 8.0–41.25 |
t = −7.85 | <.001 |
CRS = chronic rhinosinusitis; SD = standard deviation; ESS = endoscopic sinus surgery; CAMT = continued appropriate medical therapy; mRAST = modified radioallergosorbent testing; AERD = aspirin exacerbated respiratory disease; OSA = obstructive sleep apnea; GERD = gastroesophageal reflux disease; NOS = not otherwise specified; TDI = summary score of the Sniffin’ Stick olfactory test incorporating threshold, discrimination, and identification scores.
Olfactory Identification in Normosmic Cohorts
In order to determine if CRS-related OD impacts various odorants differentially, we first assessed the prevalence of normosmic identification for all 16 odorants within that domain. Prior normative data 5 reports that normosmics usually identify at least 12/16 (75%) of odorants correctly. Table 2 demonstrates that apple, turpentine, and to some extent, pineapple, are correctly identified at rates lower than expected in normosmics regardless of diagnostic category. Thus, among normosmics who miss 3 to 4 odorants within the Identification domain, there is a high likelihood that these 3 odorants account for many of the missed odorants.
Table 2.
Normosmic Identification Rates.
| Correct answer N (%) | |||||||
|---|---|---|---|---|---|---|---|
| Identification target: | All subjects (n = 145) | CRSsNP (n = 49) | CRSwNP (n = 18) | Controls (n = 78) | χ2 | Unadjusted P-value | q-Value |
| Orange | 140 (96.6%) | 48 (98.0%) | 16 (88.9%) | 76 (97.4%) | 3.65 | .161 | 0.604 |
| Leather | 123 (84.8%) | 46 (93.9%) | 16 (88.9%) | 61 (78.2%) | 6.01 | .050 | 0.750 |
| Cinnamon | 124 (85.5%) | 40 (81.6%) | 16 (88.9%) | 68 (87.2%) | 0.94 | .626 | >0.999 |
| Peppermint | 143 (98.6%) | 48 (98.0%) | 18 (100.0%) | 77 (98.7%) | 0.42 | .813 | >0.999 |
| Banana | 132 (91.0%) | 45 (91.8%) | 17 (94.4%) | 70 (89.7%) | 0.45 | .797 | >0.999 |
| Lemon | 109 (75.2%) | 37 (75.5%) | 12 (66.7%) | 60 (76.9%) | 0.83 | .661 | >0.999 |
| Liquorice | 137 (94.5%) | 48 (98.0%) | 16 (88.9%) | 73 (93.6%) | 2.34 | .311 | 0.933 |
| Turpentine | 80 (55.2%) | 24 (49.0%) | 9 (50.0%) | 47 (60.3%) | 1.77 | .413 | >0.999 |
| Garlic | 139 (95.9%) | 49 (100.0%) | 18 (100.0%) | 72 (92.3%) | 5.38 | .068 | 0.340 |
| Coffee | 139 (95.9%) | 47 (95.9%) | 17 (94.4%) | 75 (96.2%) | 0.11 | .947 | >0.999 |
| Apple | 76 (52.4%) | 29 (59.2%) | 8 (44.4%) | 39 (50.0%) | 1.54 | .460 | 0.986 |
| Clove | 129 (89.0%) | 43 (87.8%) | 17 (94.4%) | 69 (88.5%) | 0.64 | .725 | >0.999 |
| Pineapple | 99 (68.3%) | 34 (69.4%) | 13 (72.2%) | 52 (66.7%) | 0.25 | .882 | >0.999 |
| Rose | 128 (88.3%) | 43 (87.8%) | 16 (88.9%) | 69 (88.5%) | 0.02 | .989 | 0.989 |
| Anise | 123 (84.8%) | 42 (85.7%) | 12 (66.7%) | 69 (88.5%) | 5.44 | .066 | 0.495 |
| Fish | 145 (100.0%) | 49 (100.0%) | 18 (100.0%) | 78 (100.0%) | — | — | — |
CRSsNP = chronic rhinosinusitis without nasal polyposis; CRSwNP = chronic rhinosinusitis without nasal polyposis; χ2 = chi-square test statistic; N = sample size. Bold indicates P ≤ 0.05.
Table 2 demonstrates that while leather was correctly identified by patients with CRSsNP and CRSwNP more frequently than control normosmics (P = .050), this became non-significant when corrected for multiple comparisons (q = 0.750).
Olfactory Identification in Anosmic and Hyposmic Cohorts
Anosmics correctly identified odorants at rates ranging from 8.7% (turpentine) to 44% (orange) (Table 3). Otherwise, anosmics missed odorants broadly and across all stimuli. The only significant difference among diagnostic groups was fish odorant which was identified correctly by 40% of anosmic controls and 0% of anosmic CRSsNP patients (data not shown).
Table 3.
Comparisons of Correct Identification Rates Across Spectrum of OD.
| Correct answer N (%) | ||||||
|---|---|---|---|---|---|---|
| Identification target: | All anosmic (n = 104) | All hyposmic (n = 203) | All normosmic (n = 145) | χ2 | Unadjusted P-value | q-value |
| Orange | 46 (44.2%) | 164 (80.8%) | 140 (96.6%) | 97.25 | <.001 | <0.001 |
| Leather | 35 (33.7%) | 149 (73.4%) | 123 (84.8%) | 77.87 | <.001 | <0.001 |
| Cinnamon | 14 (13.5%) | 110 (54.2%) | 124 (85.5%) | 127.05 | <.001 | <0.001 |
| Peppermint | 33 (31.7%) | 191 (94.1%) | 143 (98.6%) | 217.59 | <.001 | <0.001 |
| Banana | 18 (17.3%) | 156 (76.8%) | 132 (91.0%) | 164.64 | <.001 | <0.001 |
| Lemon | 16 (15.4%) | 110 (54.2%) | 109 (75.2%) | 87.44 | <.001 | <0.001 |
| Liquorice | 31 (29.8%) | 156 (76.8%) | 137 (94.5%) | 129.64 | <.001 | <0.001 |
| Turpentine | 9 (8.7%) | 91 (44.8%) | 80 (55.2%) | 58.54 | <.001 | <0.001 |
| Garlic | 34 (32.7%) | 178 (87.7%) | 139 (95.9%) | 160.64 | <.001 | <0.001 |
| Coffee | 34 (32.7%) | 159 (78.3%) | 139 (95.9%) | 128.42 | <.001 | <0.001 |
| Apple | 24 (23.1%) | 52 (25.6%) | 76 (52.4%) | 33.95 | <.001 | <0.001 |
| Clove | 39 (37.5%) | 151 (74.4%) | 129 (89.0%) | 79.82 | <.001 | <0.001 |
| Pineapple | 22 (21.2%) | 116 (57.1%) | 99 (68.3%) | 57.20 | <.001 | <0.001 |
| Rose | 40 (38.5%) | 136 (67.0%) | 128 (88.3%) | 68.25 | <.001 | <0.001 |
| Anise | 23 (22.1%) | 111 (54.7%) | 123 (84.8%) | 97.81 | <.001 | <0.001 |
| Fish | 17 (16.3%) | 173 (85.2%) | 145 (100.0%) | 244.61 | <.001 | <0.001 |
χ2 = chi-square test statistic; N = sample size.
Odorants useful for differentiating hyposmics from normosmics or anosmics should have correct rates that are generally in the 25% to 75% range. As seen below, hyposmic subjects correctly identified odorants at rates ranging from 25% (apple) to 94% (peppermint), limiting the potential utility of these odorants to either end of the spectrum of OD. Figure 1 demonstrates odorants such as cinnamon and lemon, that generally have broad distribution of correct answers across the spectrum of olfactory loss and appear useful for differentiating into various clinical categories of OD. Apple on the other hand, is only identified correctly by 52% of normosmics. This decreased to 25% for hyposmics, which is nearly identical to anosmic rates, limiting its ability to differentiate between anosmia and hyposmia. In addition, there is significant potential for normosmics to incorrectly identify apple, potentially leading to misclassification of normosmic patients. Peppermint lies on the other end of the spectrum, as 94% of hyposmics can still correctly identify this odorant, limiting its utility until patients have severe OD and become anosmic.
Figure 1.
Correct odorant identification rates across sample of odorants and OD spectrum.
Olfactory Identification Rates by Diagnostic Category
We then analyzed if certain odorants vary by diagnostic group as patients lose olfaction. Among hyposmics, orange varies by diagnostic group (Table 4). CRS hyposmic subjects get this odorant correct nearly 20% more often than control hyposmics (P = .001). The rose odorant has similar characteristics. Other odorants are missed in similar patterns among hyposmics regardless of diagnostic group.
Table 4.
Comparisons of Hyposmic Identification Rates.
| Correct answer N (%) | |||||||
|---|---|---|---|---|---|---|---|
| Identification target: | All subjects (n = 203) | CRSsNP (n = 67) | CRSwNP (n = 60) | Controls (n = 76) | χ2 | Unadjusted P-value | q-value |
| Orange | 164 (80.8%) | 62 (92.5%) | 50 (83.3%) | 52 (68.4%) | 13.70 | .001 | 0.016 |
| Leather | 149 (73.4%) | 49 (73.1%) | 41 (68.3%) | 59 (77.6%) | 1.49 | .475 | 0.844 |
| Cinnamon | 110 (54.2%) | 33 (49.3%) | 34 (56.7%) | 43 (56.6%) | 0.98 | .612 | 0.816 |
| Peppermint | 191 (94.1%) | 63 (94.0%) | 57 (95.0%) | 71 (93.4%) | 0.15 | .927 | 0.927 |
| Banana | 156 (76.8%) | 53 (79.1%) | 46 (76.7%) | 57 (75.0%) | 0.34 | .844 | 0.965 |
| Lemon | 110 (54.2%) | 34 (50.7%) | 31 (51.7%) | 45 (59.2%) | 1.25 | .536 | 0.780 |
| Liquorice | 156 (76.8%) | 53 (79.1%) | 45 (75.0%) | 58 (76.3%) | 0.32 | .853 | 0.910 |
| Turpentine | 91 (44.8%) | 25 (37.3%) | 24 (40.0%) | 42 (55.3%) | 5.44 | .066 | 0.352 |
| Garlic | 178 (87.7%) | 56 (83.6%) | 55 (91.7%) | 67 (88.2%) | 1.94 | .379 | 0.758 |
| Coffee | 159 (78.3%) | 53 (79.1%) | 44 (73.3%) | 62 (81.6%) | 1.38 | .502 | 0.803 |
| Apple | 52 (25.6%) | 23 (34.3%) | 14 (23.3%) | 15 (19.7%) | 4.21 | .122 | 0.488 |
| Clove | 151 (74.4%) | 49 (73.1%) | 47 (78.3%) | 55 (72.4%) | 0.71 | .702 | 0.864 |
| Pineapple | 116 (57.1%) | 32 (47.8%) | 39 (65.0%) | 45 (59.2%) | 4.05 | .132 | 0.422 |
| Rose | 136 (67.0%) | 49 (73.1%) | 44 (73.3%) | 43 (56.6%) | 5.96 | .051 | 0.408 |
| Anise | 111 (54.7%) | 32 (47.8%) | 31 (51.7%) | 48 (63.2%) | 3.72 | .156 | 0.416 |
| Fish | 173 (85.2%) | 60 (89.6%) | 48 (80.0%) | 65 (85.5%) | 2.30 | .316 | 0.722 |
CRSsNP = chronic rhinosinusitis without nasal polyposis; CRSwNP = chronic rhinosinusitis without nasal polyposis; χ2 = chi-square test statistic; N = sample size. Bold indicates P < .05.
Olfactory Discrimination in Normosmic Cohorts
Similar to identification, in order to determine if CRS-OD impacts various odorants differentially, we began with normosmic discrimination rates (Table 5). Prior normative data 5 reports that normosmics usually discriminate at least 12/16 or 75% of odorants correctly. Normosmic discrimination rates ranged from 51.7% ([-]-carvone) to 96.6% (n-butanol). Eucalyptol differed between groups, however this lost significance when corrected for multiple comparisons.
Table 5.
Comparisons of Normosmic Discrimination Rates.
| Correct answer N (%) | ||||||||
|---|---|---|---|---|---|---|---|---|
| Discrimination target: | Organoleptic properties: | All subjects (n = 145) | CRSsNP (n = 49) | CRSwNP (n = 18) | Controls (n = 78) | χ2 | Unadjusted P-value | q-value |
| Octylacetate | Waxy floral, fruity | 131 (90.3%) | 45 (91.8%) | 15 (83.3%) | 71 (91.0%) | 1.18 | .554 | 0.886 |
| n-Butanol | Alcohol-like, sweet | 111 (76.6%) | 37 (75.5%) | 15 (83.3%) | 59 (75.6%) | 0.53 | .768 | 0.819 |
| Isoamylacetate | Banana-like, sweet, fruity | 133 (91.7%) | 42 (85.7%) | 17 (94.4%) | 74 (94.9%) | 3.53 | .172 | 0.688 |
| Anethol | Anise, intense, sweet | 115 (79.3%) | 44 (89.8%) | 13 (72.2%) | 58 (74.4%) | 5.00 | .082 | 0.437 |
| Gerniol | Floral, rose, sweet | 119 (82.1%) | 43 (87.8%) | 13 (72.2%) | 63 (80.8%) | 2.35 | .308 | 0.821 |
| 2-Phenyl ethanol | Honey, sweet, floral | 129 (89.0%) | 45 (91.8%) | 16 (88.9%) | 68 (87.2%) | 0.67 | .717 | 0.956 |
| (+)-Limonene | Orange, citrus, fruity | 115 (79.3%) | 39 (79.6%) | 13 (72.2%) | 63 (80.8%) | 0.66 | .721 | 0.887 |
| (-)-Carvone | Mint, spearmint, sweet | 75 (51.7%) | 24 (49.0%) | 10 (55.6%) | 41 (52.6%) | 0.28 | .871 | 0.871 |
| (-)-Limonene | Turpentine, harsh | 128 (88.3%) | 43 (87.8%) | 17 (94.4%) | 68 (87.2%) | 0.77 | .682 | 0.992 |
| 2-Phenyl ethanol | Honey, sweet, floral | 117 (80.7%) | 42 (85.7%) | 14 (77.8%) | 61 (78.2%) | 1.20 | .549 | 0.976 |
| (+)-Carvone | Spice, herbal | 102 (70.3%) | 37 (75.5%) | 10 (55.6%) | 55 (70.5%) | 2.52 | .284 | 0.909 |
| n-Butanol | Alcohol-like, sweet | 140 (96.6%) | 48 (98.0%) | 17 (94.4%) | 75 (96.2%) | 0.57 | .752 | 0.859 |
| Citronellal | Lemon, citrus, fruity | 91 (62.8%) | 33 (67.3%) | 13 (72.2%) | 45 (57.7%) | 1.99 | .370 | 0.846 |
| Pyridine | Sour, fish-like, ammoniacal | 118 (81.4%) | 43 (87.8%) | 14 (77.8%) | 61 (78.2%) | 1.99 | .370 | 0.740 |
| Eugenol | Clove, sweet, spicy | 111 (76.6%) | 39 (79.6%) | 10 (55.6%) | 62 (79.5%) | 5.05 | .080 | 0.640 |
| Eucalyptol | Eucalyptus, herbal, medicinal | 113 (77.9%) | 33 (67.3%) | 13 (72.2%) | 67 (85.9%) | 6.41 | .041 | 0.656 |
CRSsNP = chronic rhinosinusitis without nasal polyposis; CRSwNP = chronic rhinosinusitis without nasal polyposis; X2 = chi-square test statistic; N = sample size. Bold indicates P < .05.
Olfactory Discrimination in Anosmic and Hyposmic Cohorts
Anosmic subjects correctly discriminated odorants between 23% and 39% of the time with no apparent outliers (Table 6). Similar to identification responses, odorants useful in hyposmics should have correct rates that differ from normosmics and anosmics and are generally in the 25% to 75% range. With few exceptions, most odorants performed within this range (Table 6).
Table 6.
Discrimination Rates Across Spectrum of OD.
| Correct answer N (%) | ||||||
|---|---|---|---|---|---|---|
| Discrimination target: | All anosmic (n = 104) | All hyposmic (n = 203) | All normosmic (n = 145) | χ2 | Unadjusted P-value | q-value |
| Octylacetate | 30 (28.8%) | 139 (68.5%) | 131 (90.3%) | 103.35 | <.001 | <0.001 |
| n-Butanol | 34 (32.7%) | 107 (52.7%) | 111 (76.6%) | 48.61 | <.001 | <0.001 |
| Isoamylacetate | 39 (37.5%) | 155 (76.4%) | 133 (91.7%) | 91.96 | <.001 | <0.001 |
| Anethol | 36 (34.6%) | 131 (64.5%) | 115 (79.3%) | 52.28 | <.001 | <0.001 |
| Gerniol | 34 (32.7%) | 121 (59.6%) | 119 (82.1%) | 62.01 | <.001 | <0.001 |
| 2-Phenyl ethanol | 41 (39.4%) | 162 (79.8%) | 129 (89.0%) | 83.85 | <.001 | <0.001 |
| (+)-Limonene | 30 (28.8%) | 138 (68.0%) | 115 (79.3%) | 70.42 | <.001 | <0.001 |
| (-)-Carvone | 28 (26.9%) | 83 (40.9%) | 75 (51.7%) | 15.39 | <.001 | <0.001 |
| (-)-Limonene | 36 (34.6%) | 133 (65.5%) | 128 (88.3%) | 77.40 | <.001 | <0.001 |
| 2-Phenyl ethanol | 27 (26.0%) | 121 (59.6%) | 117 (80.7%) | 74.93 | <.001 | <0.001 |
| (+)-Carvone | 24 (23.1%) | 109 (53.7%) | 102 (70.3%) | 54.64 | <.001 | <0.001 |
| n-Butanol | 32 (30.8%) | 154 (75.9%) | 140 (96.6%) | 132.91 | <.001 | <0.001 |
| Citronellal | 35 (33.7%) | 97 (47.8%) | 91 (62.8%) | 20.88 | <.001 | <0.001 |
| Pyridine | 33 (31.7%) | 137 (67.5%) | 118 (81.4%) | 66.84 | <.001 | <0.001 |
| Eugenol | 38 (36.5%) | 110 (54.2%) | 111 (76.6%) | 41.09 | <.001 | <0.001 |
| Eucalyptol | 38 (36.5%) | 109 (53.7%) | 113 (77.9%) | 44.68 | <.001 | <0.001 |
X2 = chi-square test statistic; N = sample size.
Table 6 lists odorants in the order of testing. It is interesting to note that specific odorants, such as 2-phenyl ethanol, were discriminated correctly by 79.8% of hyposmics when it was the sixth odorant tested, but later only 59.6% of hyposmics correctly discriminated the same odorant when it was the 10th odorant tested. n-Butanol has similar discrepancies.
Olfactory Discrimination Rates by Diagnostic Category
We found that discrimination of specific odorants is lost equally across all odorants between diagnostic groups (Table 7).
Table 7.
Comparisons of Correct Hyposmic Discrimination Rates.
| Correct answer N (%) | |||||||
|---|---|---|---|---|---|---|---|
| Discrimination target: | All subjects (n = 203) | CRSsNP (n = 67) | CRSwNP (n = 60) | Controls (n = 76) | χ2 | Unadjusted P-value | q-value |
| Octylacetate | 139 (68.5%) | 47 (70.1%) | 41 (68.3%) | 51 (67.1%) | 0.15 | .926 | 0.926 |
| n-Butanol | 107 (52.7%) | 37 (55.2%) | 30 (50.0%) | 40 (52.6%) | 0.35 | .841 | >0.999 |
| Isoamylacetate | 155 (76.4%) | 50 (74.6%) | 46 (76.7%) | 59 (77.6%) | 0.18 | .913 | 0.974 |
| Anethol | 131 (64.5%) | 43 (64.2%) | 40 (66.7%) | 48 (63.2%) | 0.19 | .911 | >0.999 |
| Gerniol | 121 (59.6%) | 42 (62.7%) | 36 (60.0%) | 43 (56.6%) | 0.56 | .757 | >0.999 |
| 2-Phenyl ethanol | 162 (79.8%) | 52 (77.6%) | 50 (83.3%) | 60 (78.9%) | 0.70 | .705 | >0.999 |
| (+)-Limonene | 138 (68.0%) | 41 (61.2%) | 43 (71.7%) | 54 (71.1%) | 2.12 | .346 | >0.999 |
| (-)-Carvone | 83 (40.9%) | 30 (44.8%) | 21 (35.0%) | 32 (42.1%) | 1.33 | .515 | >0.999 |
| (-)-Limonene | 133 (65.5%) | 45 (67.2%) | 35 (58.3%) | 53 (69.7%) | 2.05 | .359 | 0.957 |
| 2-Phenyl ethanol | 121 (59.6%) | 41 (61.2%) | 38 (63.3%) | 42 (55.3%) | 1.01 | .603 | >0.999 |
| (+)-Carvone | 109 (53.7%) | 32 (47.8%) | 28 (46.7%) | 49 (64.5%) | 5.69 | .058 | 0.928 |
| n-Butanol | 154 (75.9%) | 54 (80.6%) | 42 (70.0%) | 58 (76.3%) | 1.96 | .376 | 0.859 |
| Citronellal | 97 (47.8%) | 37 (55.2%) | 26 (43.3%) | 34 (44.7%) | 2.25 | .325 | >0.999 |
| Pyridine | 137 (67.5%) | 48 (71.6%) | 40 (66.7%) | 49 (64.5%) | 0.86 | .651 | >0.999 |
| Eugenol | 110 (54.2%) | 43 (64.2%) | 29 (48.3%) | 38 (50.0%) | 4.06 | .131 | >0.999 |
| Eucalyptol | 109 (53.7%) | 40 (59.7%) | 27 (45.0%) | 42 (55.3%) | 2.87 | .238 | >0.999 |
CRSsNP = chronic rhinosinusitis without nasal polyposis; CRSwNP = chronic rhinosinusitis without nasal polyposis; χ2 = chi-square test statistic; N = sample size.
Discussion
This study demonstrated that odorants used in Sniffin’ Sticks identification and discrimination domain tests provide variable utility in assessing olfactory function. This olfactory test has undergone extensive validation in thousands of patients worldwide, thus is a very commonly used tool for clinicians and olfaction researchers. Similar investigations could be done using Smell Identification Test-40 (SIT40), however, Sniffin’ Sticks offers the advantage of also having discrimination assessment. Additionally, this study used Sniffin’ Sticks to classify individuals into olfactory groups, rather than an independent assessment of olfactory status. Thus, by definition it was not surprising that normosmics correctly identified more odorants than anosmics, however, several interesting and unique patterns emerged.
Items that are identified more frequently or less frequently than anticipated, may provide insight into unique sensory loss specific to a given odorant. Ideally normosmics would get close to 100% in order to avoid false negatives (eg missing an odorant but having otherwise normal olfaction). False negatives could be due to lack of familiarity, inadequate strength of stimulus, choice of distractor odorants, or lack of trigeminal stimulus. Prior normative data report that normosmics typically identify and discriminate odorants correctly at least 75% of the time. 5 Sniffin’ Sticks was also validated thoroughly, although in a German population. Thus, some aspects of familiarity with target odorant may differ when used in other populations. In our US population, it is interesting to note that apple, turpentine, and to some extent pineapple, were incorrectly identified in as many as half of normosmics. This occurred regardless of test site, thus limiting the utility of these odorants in any type of screening process to differentiate normosmic from hyposmic subjects. When examining distractors for these specific odorants, incorrect responses for the apple identification include orange, peach and melon. It is possible that these odors are all interpreted by the respondent as “fruity” and may account for the lower rate of correct responses among normosmics. Similarly, pineapple distractors include peach, plum and pear, all “fruity” choices. Turpentine distractors, on the other hand, include mustard, gum and menthol, which may all be interpreted by subjects as pungent. It would be interesting to see if similar results are obtained using SIT40 testing, the most common olfactory test used in the United States.
There were also some differences in normosmic responses based upon diagnostic category. Leather, garlic and anise all had identification differences that approached statistical significance, while eucalyptol had discrimination differences. Presumably, subjects in all diagnostic groups would have similar familiarity, strength of odorant and quality of distractors. Thus, it is unclear if some of these differences between groups are due to differential trigeminal function in each diagnostic group or other factors unique to certain diagnostic groups.
Anosmics, on the other hand, should randomly identify only 25% of all odorants correctly with a 4-item forced choice format, while they should discriminate approximately 33% of odorants in a 3-item forced choice format. In our populations, there were few outliers in identification or discrimination among diagnostic groups, thus all odorants appear to be impacted similarly once patients have severe olfactory loss.
Perhaps the most important group to differentiate and study would be hyposmics. These patients have begun to experience OD, but they are not so severe or widespread that they are classified as anosmic. Hopefully these patients have not yet suffered permanent injury and would respond to medical or surgical interventions. It is also hoped that determining differential odorant responses may ultimately yield some insight into olfactory testing patterns that become clinically useful for better understanding OD. Based upon prior validation studies, hyposmics would likely identify and discriminate 25–75% of odorants correctly. Tables 3 and 4 demonstrate the expected wide variability in identification rates in hyposmic patients. Interestingly, orange and rose identification are missed at a higher rate among non-CRS hyposmics. Identification of these odorants is retained in CRS patients until their OD becomes so severe that they are classified as anosmic. Discrimination performance appears to have fewer unexpected outliers with most odorants in the 25–75% range (Table 6). The sole odorant that performed differently between diagnostic groups was (+)-carvone, a spice/herbal odorant. This odorant trended toward poorer performance among CRS patients compared to non-CRS controls.
This study has some notable weaknesses. While the sample size is moderately large, there are still differences between CRS cases and control cohorts with factors known to impact olfaction including sex, race, allergic rhinitis, and smoking status among others. Thus we are unable to specify that differences in olfactory testing between groups are solely due to the presence or absence of CRS due to these potential confounding factors. CRS patients were also recruited from tertiary practices, while non-CRS controls were recruited from the general population. Applicability to other populations may be limited. Finally, a number of comparisons approached statistically significant type-I errors, thus we may have been underpowered in some cases.
Conclusions
Performance on olfactory testing is extremely complex. Odorant identification or discrimination requires adequate airflow, proper mucus characteristics, intact neuroepithelium and normal central processing. In addition, factors such as strength and familiarity with target odorants, as well as distractor choices all impact test results. This study provides initial data suggesting that certain odors, such as cinnamon or lemon, may be most useful in classifying patients across the spectrum of olfactory loss. Others, such as peppermint or apple, may only be useful when examining specific populations, such as peppermint to differentiate hyposmics from anosmics. Additional studies in patients across the spectrum of olfactory loss with varying etiologies of OD are needed to further understand the utility of olfactory testing.
Footnotes
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This work was supported by the National Institute on Deafness and Other Communication Disorders (NIDCD), one of the National Institutes of Health, Bethesda MD (R01DC019078; PI: RJS, Co-Investigator: ZMS, R01DC005805; PI: ZMS and TLS, Co-Investigator: RJS, JM, JAA, VRR, and K23 DC014747; PI: VRR).
ORCID iDs: Rodney J. Schlosser https://orcid.org/0000-0001-6480-0275
Ryan Rimmer https://orcid.org/0000-0002-9010-8183
Jeremiah A. Alt https://orcid.org/0000-0003-0560-5028
Thomas S. Edwards https://orcid.org/0000-0002-6290-9506
Timothy L. Smith https://orcid.org/0000-0002-6424-7083
References
- 1.Hummel T, Whitcroft KL, Andrews P, Position paper on olfactory dysfunction. Rhinol Suppl. 2017;54(26):1–30. [DOI] [PubMed] [Google Scholar]
- 2.Ganjaei KG, Soler ZM, Storck KA, Rowan NR, Othieno FA, Schlosser RJ. Variability in retronasal odor identification among patients with chronic rhinosinusitis. Am J Rhinol Allergy. 2018;32(5):424–431. [DOI] [PubMed] [Google Scholar]
- 3.Rosenfeld RM, Piccirillo JF, Chandrasekhar SS, Clinical practice guideline (update): adult sinusitis. Otolaryngol Head Neck Surg. 2015;152(2 Suppl):S1–S39. [DOI] [PubMed] [Google Scholar]
- 4.Kobal G, Hummel T, Sekinger B, Barz S, Roscher S, Wolf S. “Sniffin’ sticks”: screening of olfactory performance. Rhinology. 1996;34(4):222–226. [PubMed] [Google Scholar]
- 5.Hummel T, Kobal G, Gudziol H, Mackay-Sim A. Normative data for the “Sniffin’ sticks” including tests of odor identification, odor discrimination, and olfactory thresholds: an upgrade based on a group of more than 3,000 subjects. Eur Arch Otorhinolaryngol. 2007;264(3):237–243. [DOI] [PubMed] [Google Scholar]
- 6.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Royal Stat Soc Ser B. 1995;57(1):449–518. [Google Scholar]