ABSTRACT
Background Anosmia is a recognized symptom of COVID-19, but the relationship of SARS-CoV-2 exposure with olfactory dysfunction remains enigmatic. This report adds unique data from healthy students tested as the virus emerged locally.
Methods Standard psychometric measures assessed odor detection, identification and hedonics in healthy university students. Data from asymptomatic students (N=22), tested as SARS-CoV-2 unknowingly emerged locally, were compared to students tested just prior to local virus transmission (N=25), and our normative sample (N=272) tested over the previous 4 years.
Results The exposed cohort demonstrated significantly reduced odor detection sensitivity compared to the students in the prior group (P=.01; d=0.77; CI 0.17, 1.36), with a distribution skewed towards less detection sensitivity (P=.03). Categorically, the exposed group was significantly more likely to have hyposmia (OR=7.7; CI, 3.1, 19.4), particularly the subset assessed in the final week before campus closure (OR=13.6; CI, 3.4, 35.7). The exposed group also rated odors as less unpleasant (P<.001, CLES=0.77, CI, 0.51, 1.56) which also showed a skewed distribution (P=.005). The groups had similar odor identification performance.
Conclusion Psychometric measures of odor detection sensitivity and hedonics may detect early SARS-CoV-2 exposure in asymptomatic and pre-symptomatic persons with normal odor identification. Viral detection by nasal associated lymphoid tissue is known to trigger systemic immune effects, but its activation may also reduce smell sensitivity and shift perception of the environment towards unpleasant, increasing the social isolation that may mitigate viral infection or transmission. Regular testing of odor detection and hedonics may have value for identifying regional viral exposure.
INTRODUCTION
Coronaviruses (CoVs) are a group of enveloped single stranded RNA viruses associated with the common cold in adult humans and upper respiratory viruses in other mammals and birds.1 The viruses were considered to produce mild symptoms until the highly lethal SARS-Cov- 1 presented in 2003, followed by Middle East respiratory syndrome (MERS-CoV) in 2012. The novel SARS-CoV-2 virus emerged in Wuhan in 2019 and rapidly became a worldwide pandemic,2 challenging health services and economic stability worldwide.
Self-reports of diminished smell (hyposmia/anosmia) and taste (ageusia/dysguesia) led to the inclusion of chemosensory impairment as official COVID-19 symptoms, fueling efforts to identify the pathobiology, particularly for the olfactory symptoms.3 While the proportion of cases with olfactory complaints varies by ascertainment and assessment approach and with sample age and sex,4 there is a consensus that olfactory symptoms can present early in the course of the infection, sometimes as the only symptom,5 and some evidence suggests that early olfactory impairment may be associated with better outcomes.6
SARS-Cov-2 may be spread through respired droplets, even by asymptomatic persons;7 thus, indicators of latent infection are of crucial interest. Entry of SARS-CoV-2 into cells involves spike (S) protein binding to angiotensin-converting enzyme 2 (ACE2) along with host proteases, principally transmembrane serine protease 2 (TMPRSS2). Olfactory receptor neurons do not express these receptors,8 so the earliest olfactory changes likely reflect the responses non- neuronal olfactory epithelial cell entry, including cells that initiate ultra-rapid immune responses.9 The nasal epithelium contains innate immune-associated tissue, the nasal- associated lymphoid tissue. This phylogenetically ancient system is the first line of immune defense against inspired or ingested viruses and other pathogens, organizing systemic inflammatory processes10 that may be protective against more severe pulmonary involvement. As diminished odor sensitivity in rhinosinusitis is correlated with cytokine responses,11,12 diminished sense of smell may not entail overt pathology to olfactory receptor neurons.
Discerning the mechanism for the earliest loss of olfactory sensitivity may be a key to early detection of SARS-CoV-2 exposure and infection, although prospective studies on the earliest olfactory impairment is limited. Most studies have relied on self-reported symptoms, which are significantly less sensitive to dysfunction than standard olfactory measures. To date, no studies have comprehensively assessed odor detection, identification, and hedonic judgments in asymptomatic and pre-symptomatic persons exposed to COVID-19. Odor detection and identification involve different olfactory pathways, but to our knowledge, they have not yet been distinguished in the COVID-19 literature. Additionally, no studies have compared olfactory function in exposed persons to a large normative sample of age- appropriate and comorbidity-matched participants, nor had the availability of comparison measures from a cohort tested prior to the emergence of the pandemic.
We have applied our unique data, drawn from four years of systematic olfactory function assessments in healthy undergraduate students, to address these significant gaps in the literature. Data from an asymptomatic cohort of students exposed to SARS-CoV-2 virus as it became endemic were compared to data from a cohort tested in the weeks prior to the arrival of the virus and to ecologically valid comparison groups.
METHOD
Participants
Olfaction Tested Cohorts
Undergraduate students (N=367) participated in ongoing studies of odor detection sensitivity and identification at a private Midwestern university from 2016–2020 to earn research experience for class credit. Participants were asked to not wear perfume, cologne, or strongly scented lotions, smoke or vape at least four hours prior to their appointment or consume food or beverages 15 minutes prior to testing. The studies were approved by the university Institutional Review Board (IRB) and all participants provided voluntary informed consent.
We retrospectively separated participants into cohorts (see Figure 1a) based on campus exposures to SARS-CoV-2, exclusive of 64 students who reported or exhibited signs of illness (N=50) or smoked cigarettes or vaped (N=14) within four hours of testing. Students (Cohort IV), tested February 14–26 (N=9), when COVID-19 exposure is uncertain, were excluded from all analyses except where noted. Cohort I (N=25) was comprised of those tested before local transmission of the virus, from January to early February 2020. Cohort II (N=22) included students tested from late February to early March 2020, as the virus unfolded in the area. A subset of this group (Cohort IIa) was tested immediately before social distancing was enforced on campus. To assess the potential impact of seasonal factors, comparisons were made to the students tested late February and early March 2016–2019 (Cohort III; N=52). Normative performance was determined from an ecologically valid comparison group of all students (N=272) tested in this laboratory prior to February 7, 2020.
Sample sizes and assessment time periods for a) olfactlon tested cohorts, and b) health survey participants.
Health Survey
Separate IRB approval was obtained to conduct an online health survey following reports that anosmia/hyposmia was potentially linked to COVID-19. The survey was administered online via Qualtrics, and after participants provided informed consent. They were asked to provide information on their health and any chemosensory symptoms from January 1, 2020 through date of participation (Figure 1b). A subset of students from the aforementioned cohorts completed this survey (N=9), including four students from Cohort II, provided consent to link their survey and olfaction data.
Measures
Olfaction
All measures were administered by systematically trained undergraduate and graduate research assistants. The Sniffin’ Sticks n-Butanol Threshold Test (STT)13 assessed odor detection sensitivity, using a single-staircase forced-choice design. Scores range from 0 (low) to 16 (high). Participants were classified as anosmic, hyposmic, or normosmic based on published data.13
Odor identification was tested with the Sniffin’ Sticks Identification Test, 13 which uses 16 odor sticks to deliver common odorants (e.g. garlic, rose, orange) with a forced multiple- choice response. Scores range from 0 (low) to 16 (high). Participants also rated each the pleasantness and unpleasantness of each odor on two five-point Likert scales ranging from “0” (not pleasant/unpleasant) to “4” (extremely pleasant/unpleasant). Responses for each scale range from “0” (not pleasant/unpleasant) to 64 (extremely pleasant/unpleasant). More detailed methodology can be found in the supplementary appendix.
Health Survey
Participants were surveyed between April 14th and May 10th, 2020 as to whether they had experienced fever, sore throat, lymphadenopathy, cough, dyspnea, nasal congestion, myalgias, fatigue, diarrhea, and/or emesis (excluding alcohol related incidents) since January 1, 2020.
A modified version of the Self-Report Inventory to Measure Olfactory Dysfunction and Quality of Life14 assessed changes in the sense of smell during a one-month period and the impact of any change on daily functions, modified to also assess changes in taste, including sweet, salty, bitter, and sour. Additional information about the health survey measures can be found in the supplementary materials.
Data analysis
Data analyses were conducted with SPSS (version 23.0). Participants with missing data were excluded from analyses, as detailed in the legends of Tables 1 and 2. Independent sample t-tests and chi-squares were used to test for cohort differences in demographic characteristics, and independent sample t-tests and Mann-Whitney U tests were employed to compare olfaction measures between Cohorts I, II, IIa, and III.
To ensure that potential group differences were not the result of anomalous Cohort I data, a two-sample Kolmogorov-Smirnov (K-S) test was used to compare the frequency distributions of Cohorts II and IIa to the normative sample distribution. Distribution fits are displayed in in Figure 2. One-sample t-tests were used to compare the exposed cohorts to published norms. Effect sizes and 95% confidence intervals [CI] were computed using the Psycometrica online calculator.15 Odds ratios were employed to examine risk of hyposmia over time in the weeks leading up to the closure of campus.
Cumulative distributions of odor detection sensitivity for the normative sample (blue) and Cohort II (orange), fit to a gamma distribution (dotted line), and normal distribution(solid line). Dlstrlbu11on fits were applied as follows: Norma11ve distribution gausslan flt mean=B.26113 (8.01615, 8.50612,] slgma=1.95481 [1.79628, 2.14426]; Cohort II Gamma distribution flt mean=4.91205, a=S.99561 [3.37288, 10.6578], b=1.22059 [0.669847, 2.22415] Inset: histograms for odor senstiivity threshold for the normative sample (blue; normal flt) and for Cohort II (orange; gamma flt).
RESULTS
Odor Detection
Means and standard deviations for the olfaction measures are presented in Table 2. Significantly reduced odor detection sensitivity was demonstrated by Cohort II, the exposed cohort, compared to Cohort I, which was assessed prior to SARS-CoV-2’s emergence in the area (t=2.60; P=.01; d=0.77; CI, 0.17, 1.36). Cohort IIa, the subsample tested in the final week before social isolation, was also significantly impaired (U=58.50; P=.007; CLES=0.77; CI, 0.31, 1.81). Comparisons of odor sensitivity thresholds to those of students during the same time period in the previous four years (Cohort III) confirmed the absence of a seasonal explanation, with Cohort IIa showing less odor detection sensitivity (U=162.50; P=.03; CLES=0.75; CI, 0.30, 1.64), and a similar trend for the entire Cohort II (U=411.50; P=.06; CLES=0.63; CI=-.06, 0.95).
A comparison to the normative sample distribution showed significant rightward skewing, indicating reduced odor detection sensitivity for Cohorts II (D=0.38; P=.005; Figure 2) and IIa (D=0.52, P=.007). Similarly, Cohorts II (t=-2.92., P=.008) and IIa (t=-3.39 P=.007) had diminished odor detection in comparison to published normative data.13
Smell Diagnosis
Categorical diagnoses were examined to define the proportion of students with hyposmia in comparison to our normative sample. From mid-February to March, which included Cohort II and Cohort-IV, the latter assessed between February 14–26, hyposmia was 4- fold more likely (OR=4.13; CI, 1.8, 9.50). This increased by 7.7-fold for Cohort II (OR=7.74; CI, 3.1, 19.40) and by 13.6-fold for Cohort IIa (OR=13.61; CI, 3.40, 35.66; Figure 3).
Odor Identification and Hedonics
There were no significant differences in odor identification (all P’s >.13) or odor pleasantness ratings (all P’s >.92). Cohort II rated odors as significantly less unpleasant than Cohort III (U=257.0, P<.001, CLES=0.77, CI, 0.51, 1.56). The distribution of unpleasantness ratings for Cohort II was also positively skewed compared to the normative sample (D=0.429, P=.001). Odor detection threshold and unpleasantness ratings were not correlated across the entire sample or within any cohort (all P’s >.16).
Health Survey
Approximately 15% (N=16) of participants who completed the health survey reported being tested for influenza by clinicians between December 31, 2019 and March 9, 2020. Four had a positive test for influenza (25%), none of whom reported chemosensory changes. Six participants (38%) with negative influenza tests did report chemosensory changes at the time of their flu-like symptoms, consistent with SARS-CoV-2. The last positive influenza test was resulted on February 22, 2020. The only participant formally diagnosed with COVID-19 by a physician reported having anosmia and ageusia two weeks prior other recognized COVID-19 symptoms. Six additional participants suspected they were infected with SARS-CoV-2 but did not obtain testing and four students reported sudden onset anosmia in the absence of other COVID-19 symptoms.
Four Cohort II participants completed the survey, two of whom showed the greatest impairments in odor detection sensitivity (scores of 4 and 4.25). Neither of them reported any subjective change in olfaction, but one experienced nasal congestion and a sore throat a week after olfactory testing, followed by a cough a week later. Another participant (odor detection sensitivity=7) reported the onset of persistent muscle aches almost a month after olfaction testing.
Public Records
Publicly available health records confirm that SARS-CoV-2 was endemic to the area as early as January 13, 2020 and became more widespread by the end of February.16
DISCUSSION
This is the first report of significantly reduced odor detection sensitivity and altered hedonic ratings of odors in asymptomatic students assessed prior to knowledge that the SARS- CoV-2 virus was spreading in the local area. The study employed a comprehensive, psychometrically validated olfaction testing battery administered prior to and during the unique window in which the virus emerged. These findings suggest that olfactory impairments are the earliest manifestation of SARS-CoV-2 exposure and infection and may be useful for early identification of affected persons to limit contagion.
These findings were detected in asymptomatic healthy students. The follow-up health questionnaire confirmed that some participants who subsequently developed COVID-19 symptoms had shown the most severe sensitivity decrements yet had no subjective awareness of olfactory impairment. This finding highlights the superiority of psychometric olfaction tests over self-reports, particularly in asymptomatic persons. Also noteworthy is the intact odor identification of the exposed cohort, showing spared peripheral olfactory processing despite diminished detection sensitivity.
Our results are consistent with another study that found reduced odor detection sensitivity with intact odor discrimination in confirmed COVID-19 patients.17 However, our findings differ from studies reporting odor identification impairments associated with COVID- 19, although we only assessed apparently healthy persons and the other studies did not assess odor detection sensitivity.6,18 Odor identification requires accurate transmission of olfactory stimuli to higher processing regions, indicating intact olfactory receptor neuron function and the fidelity of transmission through the olfactory bulb. Later viral infection and inflammatory pathology may damage the peripheral olfactory pathways, but our findings suggest that the earliest blunting of taste and smell may have different underpinnings.
While olfactory receptor neurons do not express the ACE2 receptor and TMPRSS2 that are requisite for viral entry into cells, these are expressed in the epithelial support cells.19 In particular, ACE2 protein expression is found in the basal layer of the squamous epithelium of the nasal mucosa and nasopharynx.20 Sensory cells in the nasal mucosa detect viral pathogens and very rapidly organize countering immune responses21 through the activation of nasopharynx-associated lymphoid tissue (NALT). This ancient, innate immune system is the first line of immune defense against airborne pathogens and can activate B and T cell receptor expression.
Humans can also detect sickness in other individuals through an olfactory route, even when they are not ill themselves, activating a specific multisensory threat circuitry that involves the primary and secondary olfactory areas, including entorhinal/piriform cortex, orbitofrontal cortex, inferior frontopolar gyrus, middle frontal gyrus, and thalamus.22 Activation of this pathway would involve the amygdala, which is consistent with altered perceptions of unpleasantness in the hedonic odor assessments. A resultant reduction of smell and taste sensitivity may be a protective response to pathogen exposure rather than pathology. Along with a shift in the perception of the environment as more unpleasant, behavioral alterations could act to limit social interactions and feeding, preventing exposure to a more pathogenic load of viral particles or transmitting the virus to conspecifics.23 Social withdrawal is a known component of sickness behavior directed through the central nervous system, but protection against environmental threats through decreasing odor sensitivity is unexplored.
Our findings are consistent with the involvement of some pathways previously associated with immune activation and sickness behavior. It is notable that females and younger persons, who have enhanced chemodetection and immune responses compared to their male and older counterparts, also have better COVID-19 outcomes.24,25 Better detection of airborne pathogens may preempt more severe illness though personal immune activation and social withdrawal to prevent viral spread through the social group. Although correlation does not imply causation, olfaction or gustatory dysfunction in COVID-19 cases is associated with a milder course of illness.26 Potentially, viral detection in the environment may preclude infection through behavioral and inflammatory mechanisms.
However, without restoration of homeostasis this system could drive inflammation, social dysfunction, and contribute to psychiatric conditions. Our laboratory has shown reduced odor detection sensitivity in association with social withdrawal in males and females with schizophrenia, bipolar disorder, and in a non-clinical sample of persons with olfactory complaints.27-30 A coordinated immune and behavioral response is not unexpected as infectious diseases account for the greatest amount of human mortality over evolution,31 highlighted again by the COVID-19 pandemic. Viral threats are estimated to underlie a third of mammalian genetic adaptations.32
The uniqueness of this report rests in the natural experiment wherein SARS-CoV-2 arrived on a campus already having ongoing studies of olfactory function. This research will never be replicated; yet it is not without limitations. The sample sizes of the 2020 cohorts are small, although we still found moderate to large effects and statistical significance. Although our subjects were healthy and without nasal symptoms, we did not explicitly assess COVID-19 symptoms at the time of olfaction testing as the studies were not designed to test hypotheses related to infectious diseases. The absence of SARS-CoV-2 positivity confirmation also limits our ability to definitively draw the conclusion about exposure to SARS-CoV-2 at the time Cohort II participated. The retrospective nature of our health survey and lack of local and national access to COVID-19 testing for young adults also limits our health survey results. The State of Ohio, where at almost half of the students reside, only allowed access to COVID-19 testing for non- high-risk persons <60 and non-healthcare workers on June 11, 2020.16
This report adds a scientifically rigorous perspective to the early clinician and patient reports associating anosmia/hyposmia and ageusia/dysgeusia with COVID-1933-37 that propelled the American Academy of Otolaryngology to advocate for the CDC and WHO to add these symptoms to COVID-19 screening and urge precautionary isolation for individuals experiencing chemosensory impairment. Crucially, we found that odor detection can be significantly impaired without nasal obstruction or subjective awareness, as participants did not report impaired olfaction at the time of their assessment. Regular systematic testing of odor detection could aid in identifying COVID-19 infection in asymptomatic persons and also contribute to our understanding of how olfactory changes may be associated with infection severity.
Data Availability
We are willing share the data presented in this manuscript.
Role of the Funding Source
This work was supported by: University of Dayton Research Council Grants-in-Aid IGRU16 and IGRV16 (JWM); NIMH R01MH110418 (DM)
Conflict of Interest
The authors report no financial conflicts of interest.
Supplementary Appendix
Standardized Testing Demonstrates Altered Odor Detection Sensitivity and Hedonics in Asymptomatic College Students as SARS-CoV-2 Emerged Locally
Detailed Methods
Olfaction Testing Procedures
The Sniffin’ Sticks Threshold Test utilizes pen-like odor dispensing devices, 32 of which are blanks (no odor) and 16 of which contain varying concentrations of n-butanol (alcohol). The test employs a single- staircase forced-choice design in which the test administrator presents odor triplets directly below the blindfolded participant’s nostril (two blanks, one with n-butanol) one at a time in random order for five seconds each. After each triplet presentation, the participant was asked to identify the pen that contained the odor.
The Sniffin’ Sticks Identification Test is comprised of 16 pen-like devices that deliver common odorants (e.g. orange, garlic, fish, etc.). The test administrator presented each odor pen directly under the participant’s nostril, one at a time for 3-4 seconds. After each presentation, the test administrator read four response choices and the participant choose the one that best described the odor. Participants then rated each the pleasantness and unpleasantness of each odor on two five-point Likert scales ranging from “0” (not pleasant/unpleasant) to “4” (extremely pleasant/unpleasant). Responses across all 16 odors were summed to yield total pleasantness and unpleasantness scores.
Health Survey
Participants were surveyed between April 14th and May 10th, 2020 as to whether they had experienced fever, sore throat, lymphadenopathy, cough, dyspnea, nasal congestion, myalgias, fatigue, diarrhea, and/or emesis (excluding alcohol related incidents) since January 1, 2020. For each symptom endorsed, follow-up queries assessed severity, timing, and frequency. Additional questions asked chronic health conditions and whether they suspected or formally diagnosed with infectious diseases (influenza, COVID-19, strep pharyngitis, mononucleosis, and/or common cold) during this period, followed-up with questions regarding test results (if any), timing, concurrent symptoms, and
A modified version of the Self-Report Inventory to Measure Olfactory Dysfunction and Quality of Life15 assessed changes in the sense of smell during a one-month period and the impact of any change on daily functions, modified to also assess changes in taste, including sweet, salty, bitter, and sour. Participants were asked whether they experienced any changes in their sense of smell or taste during January, February, March, or April 2020; if yes, they were asked to complete the measure for that month.
Acknowledgements
Thank you to the following students for their assistance with data collection, entry, and verification: Brooke Lipnos, Madison Degnan, Rhiannon Gibbs, Wyatt Kaiser, Jenna Kilian, Michael A. Lee, Russell Mach, Maia McLin, Julie Mitchell, Lauren Olson, Piper Sereno, Madeline Scherer, Lisa Stone, Lucia Zook. We also thank Michelle Phares for her assistance in executing online data collection, and Drs. Lee Dixon and Jackson Goodnight for their support in developing the health survey.
References
