Olfactory Training Impacts Olfactory Dysfunction Induced by COVID-19: A Pilot Study

Three years after the outbreak of the coronavirus disease 2019 (COVID-19), SARS-CoV-2 has infected more than 280 million people worldwide, and the clinical manifestations of those affected vary [1, 2]. Olfactory dysfunction (OD) which includes anosmia, hyposmia, parosmia, and phantosmia is considered a specific indicator of early infection of SARS-CoV-2 and surpasses fever, cough, and shortness of breath [3, 4]. It should be noted that this is more true for earlier variants and less with the latest Omicron [5] and Delta [6] variants. In the acute phase, it affects between 50% and 75% of people diagnosed with COVID-19 [7‒9]. COVID-19-induced OD is predominant among youth, women, and people with milder COVID-19 [10‒12].

While the effect of acute COVID-19 on olfactory function is well established, a relative lack of data persists regarding the recovery rate of olfactory function in COVID-19 patients [3, 13]. Among 704 individuals who contracted PCR-confirmed COVID-19 in spring 2020, 137 (19.5%) still exhibited OD 3–7 months after the infection [14]. Similarly, at 6 months, 4.7% of patients with OD have not objectively recovered olfaction [10]. Although most patients are expected to achieve a full recovery, a proportion of them exhibit persistent post-viral OD after 1 year [15, 16].

Many individuals with OD report a reduced quality of life and display higher rates of anxiety and depression [17‒19]. Further, OD can also trigger dysfunctional eating behaviors such as increased salt and sugar intake or anorexia leading to nutritional deficiencies and significant weight disturbances [20, 21].

In recent years, researchers have investigated possible treatments for post-viral OD, including olfactory training (OT). This training consists of repeated daily exposure to a range of odorants with the aim of potentially promoting the regeneration of olfactory neurons [22, 23]. This regeneration, called neurogenesis, is thought to be mediated by an increase of neurotrophic factors in the olfactory mucosa caused by the constant stimulation of olfactory receptors [24]. This stimulation then promotes neuronal plasticity [25]and even leads to an increase in olfactory bulb volume, among other changes, in patients who underwent OT [26, 27]. These changes are consistent with studies that have demonstrated improved olfaction in patients with post-infectious OD after OT [22, 28, 29]. Usually, OT involves repeat sniffing of a set of four common odorants for approximately 20 seconds each at least twice a day for minimally 3 months (or longer if possible). This inexpensive and safe method may therefore be considered for patients with persistent COVID-19-related OD. Studies already published on post-COVID OT have found it to be an effective treatment in improving olfactory function when associated with oral corticosteroid [30]. Moreover, a single course of 28 days of OT has shown a subjective improvement in olfactory function [31] whereas 6–9 consecutive months was necessary to resolve parosmia [32]. Awaiting further knowledge on this topic, the British Rhinological Society recommended OT for all anosmic patients [33] based on the efficacy of this treatment in post-viral OD [34].

Here, we report an evaluation of efficacy of OT in patients with persisting OD due to COVID-19. We hypothesized that OT could improve olfactory function compared to placebo in people with OD post-COVID-19.

This study took the form of a double-blind randomized pilot study. The protocol, its amendments, and other documents were approved by the Medical Research Ethics Committee of the CIUSSS MCQ (MP-2021-486). Data were collected from 2021 May 6 to 2021 October 27.

All participants were either self-referred or referred by healthcare professionals via a dedicated email address. A member of the research group identified potential participants who met the inclusion and exclusion criteria and presented the research project to them over the phone. The inclusion criteria were (1) over 18 years old, (2) resident of the province of Quebec, (3) a positive COVID-19 PCR test, and (4) OD since COVID-19 for 2 months or more (OD was established by means of the University of Pennsylvania Smell Identification Test [UPSIT]). Exclusion criteria were (1) a diagnosis of chronic rhinosinusitis with or without polyps, (2) OD pre-COVID-19, (3) previous nasal sinus surgery, (4) diagnosis of a neurologic disorder such as Alzheimer’s or Parkinson’s disease, and (5) normosmia or probable malingering as identified with the UPSIT-40 testing. A recent meta-analysis on the effectiveness of OT reports an effect size of 1.10 for the effect of OT on olfactory function [25]. A statistical power analysis (GPower 3.1.9.2) reveals that, to detect such an effect size with a probability of error (alpha) of 0.05 and a power (1- beta) of 0.9, samples of n = 19 per group are needed. To make up for a potential loss of follow-up participants, we aimed for recruiting 25 participants per group for a total of 50 participants. Before enrollment, participants signed an informed consent, which was in line with the declaration of Helsinki.

Following recruitment, all participants who had not been seen by an otorhinolaryngologist following their newly onset OD were referred to their local otorhinolaryngology department. No compensation or incentive was offered for participation. We collected demographic information from all participants including age, sex, occupational and environmental exposure, lifestyle habits, date of the diagnosis of COVID-19, and OD onset.

Participants were randomized in a 1:1 ratio and no restrictions of groups were made. Randomization was done by a member of the research group who did not take part in the data collection process. After randomization, participants were mailed a kit containing two smell tests (UPSIT-40) along with an OT kit (intervention group) or a placebo kit. Videoconference was used to provide information as well as completing testing and the first OT session. They were advised not to share their training kits and not to let partners or family members smell the vials. Importantly, participants were told to sniff the substance contained in every bottle and at no point where they indicated that the substances had a smell or not. In other words, the placebo group could not conclude having received a placebo because their substance was odorless since they did not know the rationale of the intervention being based on the smelling of odors. Further, patients received a phone call from one of the experimenters after 6 weeks into the training period (1) to assess the patient’s olfactory function and (2) to verify and maintain compliance with the training procedure.

OT was performed over a period of 12 weeks [35]. Patients exposed themselves twice daily to four odors. Specifically, we used floral (phenylethanol: rose odor), fruity (orange aroma: orange odor), aromatic (eugenol: cloves odor), and resinous (eucalyptol: eucalyptus odor). In the original publication, lemon was used as a fruity odor. However, due to supply chain issues, we could not use lemon as a stimulus. In accordance with the original publication, we used orange, another citrus odor with similar properties [35]. All odorants were purchased from Sigma-Aldrich, USA. Training patients received four amber opaque glass vials (30 mL, Fisherbrand Inc., USA); each contained one odorant (5 mL, soaked in cotton pads to prevent spilling) and identified with a number. Each session included a rotating exposure of each odor for 10 s, with 10 s rest intervals between each scent; we instructed participants to do this for a total of 5 min.

Participants in the placebo group were asked to sniff four glass vials that were identical in appearance to the ones distributed to the intervention group. However, instead of odorants, we used odorless propylene glycol (Sigma-Aldrich, USA). The procedure was in all respects identical to that of the OT. Both participants and experimenters were blinded to the participants’ condition.

Primary outcome was the amelioration of objective olfactory function. We evaluated olfactory function by using the UPSIT-40, a standardized and well-validated odor identification test [36]. This test consists of 40 microencapsulated odors that are released by rubbing with the tip of a pencil. The identification of each smell is made among four choices. Scores are compared to gender and age norms. A score of 35 and 34 or above represents normosmia for female and male, respectively. Mild microsmia ranges from 31 to 34 and 30–33. Moderate microsmia ranges from 26 to 30 and 26–29. Severe microsmia ranges from 19 to 25 for both sexes. Total anosmia ranges from 6 to 18. A score of 5 or less indicates probable malingering. Importantly, the test can be posted and applied remotely. Participants received two tests, one to complete at the initial meeting and the other at the final meeting; both tests were carried under videoconference supervision by a member of the research group who was blinded to participant’s allocation.

Secondary outcomes were increases in subjective chemosensory function, decreases in prevalence of parosmia, and increases in quality of life. We asked participants to (1) self-evaluate smell and taste sensitivity at questionnaire completion (0–10 points). Next, we asked participants if (2) they suffered from parosmia (yes/no). Finally, we assessed (3) the influence of chemosensory dysfunction on quality of life using a short version of the Questionnaire of Olfactory Disorders (0–28 points) [37]. We also used the Nasal Obstruction Symptom Evaluation (NOSE), a visual analogue scale (scale from “not a problem” to “severe problem”), to rate participants’ nasal obstruction [38]. These were carried under videoconference supervision by a member of the research group who was blinded to participant’s allocation.

We used SPSS 25 (IBM Corp, Armonk, NY, USA) for data analysis. Differences in demographic characteristics between the 2 groups of patients (mean, standard deviation) were analyzed using independent samples t tests; differences in proportions between groups were assessed using the χ² test. Effects of group (intervention, placebo), time (before training, after training), and modality (subjective smell, subjective taste) were assessed using a repeated measures ANOVA. We corrected post hoc t tests with the Bonferroni-Holm procedure to control for multiple comparisons. We set the alpha value at 0.05.

A total of 50 participants were included in the study. The intervention group consisted of 16 women and 9 men (mean age 44.9 [SD = 7.4] years); the placebo group consisted of 17 women and 8 men (mean age 44.5 [SD = 10.1] years). Time in between first symptoms of OD and initiation of OT was 273 days (SD = 108) for the intervention group and 241 days (SD = 110) for the placebo group (Table 1). We did not observe any clinically meaningful group differences in any of these variables.

At follow-up, a total of 41 participants returned, namely 19 from the intervention group and 22 from the placebo group. For an additional participant from the intervention group, we could not obtain the UPSIT score at follow-up (shown in Fig. 1).

The average UPSIT score before training was 24.3 (7.01) in the intervention group and 24.6 (5.58) in the placebo group. Following training, these numbers were 25.8 (7.95) and 25.6 (6.13), respectively. We did not observe any significant effect of group or time, nor any interaction on the UPSIT scores (shown in Fig. 2). The number of days between onset of OD and difference in UPSIT scores were significantly and positively correlated (r(40) = 0.38; p = 0.016).

We observed a significant effect of time: (F(1,39) = 17.5; p < 0.001) on the auto-evaluation of smell and taste function (average scores before training: (mean [SD]) 4.3 (0.29) points; after training: 5.2 (0.36) points). Furthermore, we observed a significant effect of modality (F(1,39) = 10.57; p = 0.002; average scores for smell: 4.2 [0.28] points; taste: 5.3 [0.40] points). We did not observe an effect of group, but the interaction of group*time showed a trend (F(1,39) = 2.99; p = 0.091). We therefore decided to carry out post hoc tests by comparing scores for both groups. These post hoc tests showed that participants from the intervention group exhibited a significant increase in their self-evaluation of smell (before: 3.8 [1.9] points; after: 5.4 [1.8] points; p = 0.002) (shown in Fig. 3a) and taste (before: 5.3 [2.5] points; after: 6.5 [2.5] points; p = 0.01) (shown in Fig. 3b); no such effect was found in the placebo group (smell: before 3.5 [1.8] points; after: 4.2 [2.4] points; taste: before: 4.4 [2.6] points; after: 4.9 [3.0] points, ns).

Before training, 16/20 participants in the training group and 19/22 in the placebo group indicated to suffer from parosmia (χ²(1, 42) = 0.33, p = 0.58). After training, 14/19 participants from the trained group indicated parosmia, while this number was 21/22 in the placebo group (χ²(1, 42) = 3.87, p = 0.049; shown in Fig. 4).

We observed an effect of time (F(1,39) = 13.3; p = 0.001) on quality of life impairment but no effect of group or interaction. In fact, QOD scores before training (12.4 [SEM = 0.74] points) were significantly larger than after the training (10.5 [0.78] points), indicating that quality of life improved in both groups significantly (shown in Fig. 5).

Here, we describe a pilot study on the effectiveness of OT on chemosensory function. Our major results are as follows: (1) OT increases subjective olfactory (and gustatory) function in both intervention and placebo groups; the effect is significant in the intervention group but not in the placebo group. Nevertheless, the subjective improvement was not paralleled by any effects on scores in olfactory testing. (2) The frequency of parosmia was significantly lower after OT. (3) Quality of life was improved after the training in both intervention and placebo groups. (4) Longer intervals between OD onset and OT are associated with better improvement of UPSIT scores.

Several studies showed that OT has a better outcome than spontaneous recovery in post-viral OD [22, 28, 29, 39‒41], before the COVID-19 pandemic. In this line of thought, the British Rhinological Society drafted a consensus guideline to treat anosmia during the COVID-19 pandemic and recommend OT for all anosmic patients [33]. However, this recommendation is based on efficacy of OT in post-viral OD rather than post-COVID due to the lack of knowledge on its efficacy for olfactory recovery in the literature, highlighted by Jafar et al. [34] in their systematic review. Our results – even if limited to subjective olfactory function – are in line with another report that showed improvement of olfactory function after a 10-week OT period [30]. However, this improvement was noted only in a group of people who had a course of oral corticosteroid along with OT and they did not have a control group to assess the efficacy of OT alone. Moreover, another study also found a subjective improvement in patients undergoing OT for at least 28 days [31]. These results complement well ours since we found similar results in terms of subjective olfactory function; however, our report adds the complement of objective testing to self-assessment.

We have not yet completely understood how OT improves olfactory function; potential underlying mechanisms include the regeneration of olfactory receptor neurons in the olfactory epithelium (neural plasticity of the olfactory system [25]). In fact, histologic studies on anosmic mice show that olfactory stimulation leads to increased expression of olfactory receptors as well as neurotrophic factor resulting in olfactory regeneration. More specifically, when mice undergo OT, they express more genes associated with neurogenesis. In other words, stimulation of receptors in the olfactory epithelium, when performing OT, promotes expression of genes and neurotrophic factors implicated in neuronal plasticity and then regeneration of affected neurons can occur [42, 43]. In fact, olfactory receptors are involved in neuronal connectivity regulation and contribute to the refinement and plasticity of the peripheral olfactory system [24]. In line with this, neurogenesis occurs in the adult olfactory epithelium [44]. Therefore, repetitive exposure to an odorant increases peripheral (i.e., from the olfactory mucosa) and central (i.e., from the scalp) electrophysiological responses while decreasing the threshold for that substance in humans [23, 45]. On a macroscopic level, OT leads to an increase of the olfactory bulb volume [26, 27], possibly the result of more afferent neurons relaying olfactory input to the brain. In addition, OT is associated with the reorganization of functional cerebral networks [46, 47] and an increase in gray matter volume in olfactory-related regions of the brain, specifically the hippocampus and the thalamus [26], and can even provoke connectivity changes in the visual cortex [48]. Still, the exact mechanism is not known; the continuous influx of patients of COVID-19-induced OD may provide a reservoir for future studies to understand the effects of OT on different levels of processing in the human olfactory system.

Parosmia prevalence increases with the duration of COVID-19-related OD [49‒51]. In fact, parosmia may be one of the most debilitating long-term consequences of COVID-19 on olfactory function as it has a strong impact on quality of life [52]. The pathogenesis mechanism of parosmia is not completely understood, but it may be linked to cognitive processing of incomplete afferent sensory information [53]. According to this hypothesis, COVID-19 disrupts peripheral neurons in the olfactory bulbs, altering the olfactory information relayed to the brain. In this line of thought, parosmia can be seen as an early sign of recovery from post-COVID anosmia, with partially re-established input to central nervous structures [53]. Our data suggest that OT reduces parosmia frequency. One can therefore speculate that OT increases the number of afferent neurons in the periphery and hence provides complete information to the brain thereby regularizing the underlying problem causing parosmia. Another study has shown that OT was efficient in resolving parosmia [32], reinforcing our results. Moreover, their study had a course of 6–9 months of OT while ours had 3 months. This may suggest that parosmia may benefit from a shorter course of OT. Future studies must investigate this closer and should also evaluate the severity of parosmia.

We further observed an improvement in quality of life following OT. This may very well be a placebo effect, as patients may have felt empowerment in face of their olfactory deficits, hoping it would help to improve their condition. Nevertheless, it is important to point out that patients with post-COVID-19 OD are often desperately looking for medical support, thus making OD a more common motive for consultation in the otolaryngology office. As aforementioned, OD can lead to anxiety, depression, and dysfunctional eating behaviors outlining the importance of validating the effectiveness of OT in post-COVID-19 OD [22]. In our opinion, this result shows that patients’ complaints about OD need to be taken seriously.

Therapy duration is still a matter of debate and there is no consensus on how long OT needs to be sustained to attain therapeutic effects since benefits have been established in non-COVID-19 post-viral OD from 12 to 56 weeks [54]. Timing of initiation of OT is also critical. Kattar et al. [54] stipulated that best results were seen with a shorter interval between symptom onset and initiation of OT (<12 months). In contrast to this, our data suggest that longer intervals may be associated with better outcomes. One may hypothesize that in shorter intervals between COVID-19 diagnosis and OT, ongoing inflammatory processes limit the OT’s potential. In fact, patients infected with COVID-19 exhibit inflammation of the olfactory cleft visible in MRI which may play a role in OD pathogenesis [55]. As corticosteroids reduce inflammation, patients who underwent a trial of oral corticosteroid along OT had a larger increase in olfactory function in a pilot prospective study [30], but careful considerations have to be taken while selecting patients because of potential risk of oral steroids [41]. Future studies need to clarify the role of ongoing inflammation in post-COVID-19 OD. Nevertheless, one has to keep in mind the earlier studies have shown that the probability of a successful OT decreases with a longer interval between onset of OD and onset of OT. Future studies should investigate the potential overlap between these two potentially independent mechanisms to evaluate the possibility of a “sweet spot” with the highest possibility of a positive outcome.

This study has some limitations. First, we did not assess patients’ compliance. The time required to complete daily OT might have discouraged individuals to fully undergo the training; future studies should find appropriate means to evaluate patient compliance (e.g., daily log apps). Second, we assessed only the presence or absence of parosmia; future studies should use tools such as the parosmia questionnaire or the Sniffin’ sticks parosmia test to evaluate the severity of parosmia [14, 56]. Third, placebo control trials with OT are difficult because we can suppose participants could ask a third party to sniff the vial and tell them whether it has a scent or not. This shortcoming was addressed by telling both groups the studied effect was sniffing the content of the vial and not exposure to smells. Thus, they were not informed the rationale of OT is based on sniffing four scents. However, our study is, to the best of our knowledge, the first prospective study with randomization involving a control group regarding the efficacy of OT for patients with post-COVID OD and we believe it reports important new information that may help guide future studies. As this is a pilot study, more data, with the suggested modifications, are required to assess the efficacy and the external validity of OT for treatment of post-COVID-19 OD.

This double-blind pilot study shows that OT leads to improved subjective olfactory function and a reduced rate of parosmia. We did not observe any effect of OT on scores in a validated test of olfactory function; furthermore, OT improved quality of life in both the intervention and the placebo group. Especially when considering that OT has virtually no side effects, it should therefore be recommended as a first-intention treatment option to all patients presenting with post-viral OD.

The protocol, its amendments, and other documents were approved by the Medical Research Ethics Committee of the CIUSSS MCQ (MP-2021-486). Participants gave their informed and written consent prior to enrollment in the study.

Authors have no conflicts of interest to declare.

Fonds de Recherche du Québec – Santé (FRQS). No roles of funding in analyzing the data nor preparing the manuscript.

Simon Bérubé, Claudia Demers, Nicholas Bussière, Frank Cloutier, Josiane Bolduc-Bégin, and Johannes Frasnelli designed the study. Data collection was conducted by Simon Bérubé, Claudia Demers, Valérie Pek, and Angela Chen. Johannes Frasnelli analyzed the data. Simon Bérubé and Claudia Demers drafted the manuscript, which was critically revised by all authors, and conducted participants’ recruitment. All authors gave final approbation for the article to be published.

All data analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.