Prevalence and Recovery of Olfactory Dysfunction in 1,363 patients with coronavirus disease 2019: A Multicenter Longitudinal Study.

Olfactory dysfunction (OD) is a key symptom of coronavirus disease 2019 (COVID-19). Currently, a few data are available about the recovery of OD after the infection resolution. In this study, we investigated both prevalence and recovery rate of OD with subjective and objective clinical tools in 2,581 patients. First, our data showed that the prevalence of OD was signicantly higher in mild form (85.9%) compared with moderate-to-critical forms (4.5-9.7%; p=0.001). Second, focusing on patients with OD who completed the 2-month follow-up period (N=1,363), we observed that 328 patients (24.1%) did not subjectively recover olfaction 60 days after the onset of the dysfunction. The mean duration of self-reported OD was 21.6±17.9 days. Third, the objective olfactory evaluations performed on a subset of patients (N=233) reported hyposmia or anosmia in 54.7% and 36.6% of mild and moderate-to-critical forms, respectively (p=0.001). At the end of follow-up, 15.3% of anosmic/hyposmic patients did not objectively recover olfaction. The higher baseline severity of objective olfactory evaluations was strongly predictive of persistent OD (p<0.001). OD disappeared in 75% to 85% of patients regarding self-reported or objective olfactory evaluations.


Introduction
As of June 24th 2020, there have been 9 million con rmed cases of coronavirus disease 2019 (COVID- 19) worldwide, with 469,587 con rmed deaths. 1 The clinical picture of the infection may vary regarding the disease severity and usually includes general, otolaryngological and neurological symptoms. 2,3 The olfactory dysfunction (OD) is one of the most prevalent symptoms. 2 The prevalence of OD may vary regarding the clinical setting, with rates of total loss of smell as high as 70% in patients with mild COVID-19 form. 2,4,5 The prevalence of OD in moderate-to-critical COVID-19 forms was poorly investigated. Moreover, there is, to date, a paucity of studies prospectively studying the recovery rates of smell sense in COVID-19 patients.
The aim of this study is to investigate both prevalence and recovery of OD in COVID-19 patients through subjective and objective clinical tools.

Subjects & Setting
From March 22, to June 3, 2020, 2,581 ambulatory and hospitalized patients with laboratory-con rmed diagnosis of COVID-19 (nasal swabs-RT-PCR) were consequently identi ed from 18 European hospitals. Among them, 1,916 patients reported subjective OD (74.9%), de ned as partial of total loss of smell. The de nition of mild, moderate, severe and critical patients was based on the COVID-19 Disease Severity Scoring of World Healthy Organization. 6 Mild patients were de ned as patients without evidence of viral pneumonia or hypoxia and were commonly home-managed and followed. Moderate COVID-19 patients had clinical signs of pneumonia (fever, cough, dyspnea, fast breathing) but no sign of severe pneumonia (including SpO2 ≥ 90% on room air). Severe COVID-19 patients were de ned as individuals with clinical signs of pneumonia plus one of the following: respiratory rate > 30 breaths/min; severe respiratory distress; or SpO2 < 90% on room air. According to the center and the availability of local healthcare resource, moderate and severe patients were home-managed (moderate) or hospitalized in non-intensive care units (ICU) versus ICU. Patients with critical disease had acute respiratory distress syndrome (ARDS), sepsis or septic shock and were hospitalized in ICU. Patients with OD were followed to assess the recovery olfactory rates and the duration of OD.

Epidemiological and Clinical Outcomes
Epidemiological (gender, age, ethnicity) and clinical data (comorbidities and symptoms) were collected with a standardized online questionnaire at the end of the disease (de ned as the general symptom resolution) or at the hospital discharge. Both patients (home-managed) and physicians (hospitalized patients) may ful ll the questionnaire. For patient who completed the study, olfactory and gustatory questions were based on the smell and taste component of the National Health and Nutrition Examination Survey (Appendix 1). 7,8 A subset of patients (EpiCURA Hospital, Belgium & Foch Hospital, Paris, France) bene ted from objective olfactory tests within the 2 (mild, moderate) or 3 (severe to critical) weeks of the onset of the olfactory disease. Olfactory objective evaluations consisted of Sni n'Sticks tests (Medisense, Groningen, Netherlands), which is a standardized and validated psychophysical olfactory evaluation using 16 smell pens. The patient had to choose the adequate term describing the smell between 4 given options. 7 The total score ranges from 0 (no olfaction) to 16 (perfect olfaction). Regarding results, three categories were de ned: normosmia (score between [12][13][14][15][16], hyposmia (score between 9-11) and anosmia (score < 9). 9 Patients with hyposmia or anosmia were invited to attend for repeated evaluation with Sni n'Sticks until scores returned to normal levels. Patients who bene ted from psychophysical olfactory evaluation also ful lled the French version of the sinonasal outcome tool-22 . 10 More details about the data collection, inclusion and exclusion criteria are available in the ow chart (Fig. 1).

Biology and Imaging Findings
The following admission outcomes were collected from the hospitalized patients: O2 saturation, chest computed tomography ndings and biology, including blood formula, liver, renal, heart functions, in ammatory molecules and ionogram. The 1-month serology (IgG) has been realized in patients who bene ted from objective olfactory evaluations.

Statistical Analyses
Statistical analyses were performed using the Statistical Package for the Social Sciences for Windows (SPSS-v22,0; IBM Corp, Armonk, NY, USA). The outcome comparison between severity patient groups and the evolution of olfactory evaluations through the follow-up period were made through Kruskal-Wallis, Mann-Whitney U test, Wilcoxon Rank test and Chi-square. The relationship between epidemiological, clinical and olfactory outcomes was analyzed through multivariate analysis. According to the WHO classi cation, the statistics were realized considering 3 groups: mild, moderate, and severe-to-critical patients.

Prevalence of Olfactory Dysfunction
Among the 2,581 patients, 1,916 reported self-reported OD (74.2%). The prevalence of self-reported OD was 85.9%, 4.5% and 6.9% in mild, moderate and severe-to-critical patients, respectively (p = 0.001). The clinical presentation signi cantly varied between mild and moderate-to-critical patient groups. Patients with moderate-to-critical COVID-19 were older than patients with mild COVID-19 (p = 0.001). Moderate-tocritical forms had higher prevalence of the following comorbidities: hypertension, diabetes, gastric disorders, renal, respiratory, heart, liver and neurological disorders (p < 0.05). Among the usual symptoms, OD was more prevalent in mild form compared with moderate-to-severe forms (p = 0.001). Severe and critical COVID-19 patients had more frequently cough (severe and critical), dyspnea and fever than the others (p < 0.05) ( Table 1).

Evolution of Subjective Olfactory Dysfunction
Among the patients with OD, 1,363 (71.1%) completed the follow-up subjective evaluations (Fig. 1). The high majority of these patients had mild COVID-19 form (98.0%). Their clinical and olfactory features are described in Table 3. The most prevalent comorbities were hypertension (8.4%), asthma (6.5%) and gastroesophageal disorders (5.0%). According to our clinical tools (National Health and Nutrition Examination Survey and SNOT-22), the most prevalent symptoms of patients were asthenia, headache and rhinorrhea. OD consisted of self-reported total loss of smell in 81.6% of patients, while 18.4% of patients reported partial loss of smell (Table 3). The mean duration of COVID-19 symptoms (excluding OD) was 13.8 ± 6.1 days. The OD developed after the other symptoms in 44.7% of cases and disappeared within the month following the onset of OD in 54.3% of patients. Dysgeusia, de ned as the impairment of salty, sweet, bitter and sour, was reported by 55.9% of patients, whereas 83.9% of patients reported aroma perception dysfunction (Table 3). A total of 328 patients (24.5%) did not subjectively recover olfaction 60 days after the onset of the dysfunction. The mean duration of self-reported OD was 21.6 ± 17.9 days.

Prevalence and Evolution of Objective Olfactory Dysfunction
Among the 2,581 patients, 233 patients completed objective olfactory evaluations (Table 4). There were 52 patients with moderate-to-critical COVID-19 and 181 patients with mild form of the disease (77.7%). Self-reported anosmia, cacosmia, phantosmia and aroma dysfunction were signi cantly more prevalent in mild than moderate-to-critical COVID-19 forms (p < 0.02). The mean values of the Sni n'Sticks tests were signi cantly lower in mild patient group compared with moderate-to-critical patient group (p = 0.001; Table 5). Moreover, the prevalence of objective olfactory dysfunction was signi cantly higher in mild forms compared with moderate-to-critical forms (36.6 versus 54.7; p = 0.001).  The baseline mean value of Sni n-Sticks tests of anosmic and hyposmic patients (N = 118; 7.6 ± 3.0) signi cantly improved after 30 (10.6 ± 3.7) and 60 days (11.3 ± 3.5) of follow-up (p = 0.001). At the end of follow-up, 18 anosmic/hyposmic patients did not objectively fully recover olfaction (15.3%).

Clinical and Objective Olfactory Associations
Among the cohort of 2,581 patients, there were no signi cant association between clinical data (biology, CT-scan ndings) and the development of OD.
Among the cohort of 233 patients, individuals with fever exhibited a signi cantly higher level of IgG In this study, we observed that both self-reported and objective ODs were more prevalent in mild patients compared with individuals presenting moderate-to-critical COVID-19. Vaira et al. recently observed that anosmia and hyposmia accounted for 70% of COVID-19 mild-to-moderate patients. 5 However, they only observed a trend of signi cant differences between severe and mild forms regarding the objective olfactory disorder. In the study of Moein et al. 60 COVID-19 hospitalized and home-managed patients bene ted from objective olfactory evaluations. 14 Using a different anosmia de nition (microsmia), the authors reported a prevalence of objective OD in 98% of patients. Although a low number of hospitalized patients (N = 6), the study ndings support a higher prevalence of OD in mild patients (45%) compared with severe patients (10%). 14 The comparison with these two studies is however limited because authors did not classify the patients according to the WHO classi cation and they used different olfactory tests.
The main hypothesis underlying the higher prevalence of anosmia in mild COVID-19 would consist of differences in the immune response to the infection in mild and moderate-to-critical patients. In this hypothesis, patients with mild COVID-19 could have a better local immunological response through a higher production of IgA, which could limit the virus spread into the organism. The limited virus spread in the host body could therefore be associated with a mild clinical form of the disease. Due to the local in ammatory reaction and the well-demonstrated olfactory cell expression of Angiontensin Coverting Enzyme-2 (ACE2) and TMPRSS2, 15,16 the patients with mild disease could have stronger impairment of olfactory cells. In addition, we observed that severe and critical patients had a signi cantly higher level of IgG than mild patients, which may corroborate some ndings of the literature. 17 However, this hypothesis requires additional studies involving immunological sera, saliva, and nasal secretion analyses.
According to our analysis, young patients could have a higher rate of anosmia compared with elderly individuals. Similar ndings have been reported in the study of Speth et al. who investigated self-reported OD in 103 COVID-19 patients. 18 Although a signi cant p-value, we need to remain cautious in the interpretation of this results for two reasons. On the one hand, the association is signi cant but exhibited a low correlation coe cient (r s =0.246).
The high prevalence of OD in COVID-19 patients supports the need for primary care, ear, nose, and throat (ENT) and neurology physicians to be able to counsel patients regarding the likelihood of recovery, and to identify those at risk of persistent OD, such that therapeutic strategies can be targeted appropriately.
Considering both subjective and objective data, we may suggest that the 60-day recovery rate ranges from 75-85%. Interestingly, we may identify several pro les of OD severity because over a third of patients reported smell recovery within the 14 days following the development of OD, while one third did not recover within the 45 days. Typically, OD occurring as part of the common cold is related to nasal congestion, rhinorrhea, olfactory cleft edema and lasts 2-3 weeks. The high prevalence of nasal symptoms could partly explain the occurrence of short OD in some patients who rapidly recovered olfaction once the nasal symptoms disappeared. However, for patients with mid-to-long term or persistent OD, the pathophysiological mechanisms underlying the development of OD could be more complex.
According to recent ndings, 15,19 the OD could be additionally associated with injury of the olfactory neuroepithelial and a virus spread into the olfactory bulb where sustentorial cells and; in some patients, stem neurons express ACE2 and TMPRSS2. Because the expression of ACE2 and TMPRSS2 varies between individuals, 20 the long duration of OD in some patients could be due to higher protein expression and more extensive injuries of the olfactory cells. The neurogenesis of the olfactory cells is possible but may take several months. 21 The neural hypothesis of OD related to COVID-19 infection is supported by the lack of signi cant association between olfactory evaluation results and nasal complaints. Moreover, post-viral anosmia was observed in some infections related to viruses of the Coronaviridae family. 22 In sum, the mechanisms underlying the OD development could associate olfactory cleft congestion in short-term anosmic patients, injury of the olfactory neuroepithelium and virus spread into the olfactory bulb in mid-to-long terms anosmic patients. Future studies are needed to con rm these hypotheses.
The present study has several strengths and limitations. The main strength is the high number of included patients, which allows to con rm the higher prevalence of OD in mild over moderate-to-critical patients. The data collected in this large cohort allowed us to evaluate the 2-month subjective and objective recovery rate of smell sense. The main limitations are the lack of clinical olfactory examination or imaging at the onset of the disease to assess the olfactory cleft and the olfactory bulb. These observations could provide useful information to better understand the pathophysiological mechanisms underlying the development of anosmia. However, performing nasal beroptic examination during the pandemic was prohibited. Moreover, the taste evaluations reported low rates of taste dysfunction in hospitalized patients. Many severe-to-critical patients had nasogastric feeding tube, which, additionally to the delay to assess taste function, may bias the assessment of taste dysfunction. Another limitation is related to the delay (2 to 3 weeks) between the OD onset and the realization of the olfactory evaluations.
This delay was particularly long in hospitalized patients who had to be able to undergo olfactory evaluations. Although this possibility is not supported by patient-reported symptoms, the delay between the onset of symptoms and the objective olfactory testing may underestimate the incidence of olfactory dysfunction.

Conclusion
OD is a prevalent disorder in COVID-19 patients with a higher prevalence in patients with mild forms of the disease. At the 2 months follow-up, 75-85% of patients recovered olfaction according to subjective and objective olfactory evaluations. Future studies are needed to determine the long-term recovery rate of COVID-19 patients.
Declarations Figure 1 Flow Chart. 1The diagnosis tests of ambulatory and hospitalized patients were centralized in Hospital Laboratories, allowing the identi cation of patients with a positive diagnosis. 2Physicians directly completed patient information on an online questionnaire for hospitalized patients while ambulatory patients ful lled the evaluation at home (home-managed patients). 3The rst objective of the study was to investigate the prevalence of olfactory dysfunction according to the severity of the disease (N=2,581).
4The second objective of the study was to investigate the recovery of olfaction, which was made on 1,363 patients who completed the evaluations. To be included, patients had to be followed over the 60 days post-COVID-19 to assess (potential) occurrence and evolution of olfactory dysfunction. 5The third objective of the study was to objectively assess the olfactory dysfunction in a subset of patients of the 2,581 included patients. Abbreviations: COVID-19= coronavirus disease 2019; RT-PCR= reverse transcription polymerase chain reaction.