Smell, Taste and Covid-19: Testing is Essential

Abstract During the Covid-19 pandemic is became clear that smell and taste (chemosensory) disturbance is very common in the early stages of disease. This article addresses: 1) why Covid-19 specifically targets the modalities of smell and possibly taste and what is the mechanism 2) what is the frequency of smell and taste loss and, 3) what is the overall prognosis. It is suggested that mouth breathers may be at particular risk of Covid-19. Symptom-based questionnaires are likely to under-estimate the prevalence of chemosensory impairment by as much as 50%. The prevalence of smell loss is so high that a person who has normal olfaction on formal testing is unlikely to be infected significantly with Cov-2. Furthermore, someone without symptoms who has an abnormal smell test could still be infected and liable to spread the disease. Brief, low cost, olfactory tests are available that would permit a high throughput in field stations and airports. A normal result might obviate the need for a nasopharyngeal swab for the Cov-2 virus

bulb. Indeed MRI-based studies have shown oedema of the olfactory bulb 9 (Figure 3) as well as more central changes, namely in the gyrus rectus 10 and by CT/PET, in the orbitofrontal cortex 11 . There is provisional evidence that ACE2 receptors are present in the tongue (Figure 4) particularly taste buds and to lesser degree in the lingual epithelium 12,13 . Cov-2 can be isolated from saliva 14 thus there is a plausible mechanism for such infection to involved taste bud receptors. Less is known about ACE2 expression in the major taste nerves, namely the chorda tympani and glossopharyngeal nerves. 2. What is the frequency of chemosensory loss?
There have been numerous estimates worldwide, but with a few exceptions detailed below, most have been based on questionnaire surveys without objective measurement and several have not contained a control group. Samples have been varied: some are based predominantly on outpatients others reliant on in-patients with testing at varying stages of illness. The largest investigation 15,16 employed a smartphone-based app to retrieve symptomatic data on over 2 million people in UK and USA and found that in those reporting chemosensory impairment, 65% had a subsequent positive PCR for Covid-19. When this was combined with fever, cough, fatigue and loss of appetite the correlation with PCR for Cov-2 was very high. A large meta-analysis totalling 38,198 subjects 17 documentated an overall prevalence of smell impairment in Caucasians of 49% and 16.7% in Asians. Taste symptoms occurred in 51% Caucasians and 18% Asians. Other studies show wide estimates of prevalence -up to 70% 18 with an approximately equal rate for smell or taste.
Sometimes isolated impairment of smell or taste is documented as a presenting symptom. A study from San Diego based on ambulatory individuals with influenza-like symptoms, noticed that subjective report of smell impairment was associated with a 10-fold lower risk of hospital admission for Covid-19 19 . This finding is discussed further below Surveys that have relied on patient reports are susceptible to multiple confounders including recall bias 20 and a tendency to over-representation of female respondents 21 . Even more importantly, less than 40% of individuals are actually aware of a proven olfactory defect 22 . For subjective awareness, the defect needs to be bilateral and of at least moderate severity. Furthermore, smell loss and taste loss are very frequently confused. Most people who complain of impaired taste have reduced olfaction 23 whereas it is unusual for someone with primary taste loss to complain of smell impairment. The mechanism of this phenomenon has not been satisfactorily explained.
Patient reports of olfactory impairment are therefore intrinsically unreliable and will tend to underestimate the true picture due to lack of awareness and confusion with taste. Furthermore, if taste is really affected in Covid-19 any such deficiency would inflate estimates of smell impairment where based on subjective reports. According to PubMed, at the time of writing there have been 14 articles worldwide where various objective olfactory measurements have been made (Table 1). Case numbers range by centre from 14-345 individuals. In nearly all instances a confirmatory polymerase chain reaction (PCR) test for Cov-2 has been undertaken. In only 9 cases was there a control group and where present the PCR test is not stated in 7 of these. Matching by age and gender was performed in just 4 instances. Subjective awareness of olfactory loss was indicated in 12 studies with a prevalence ranging from 28%-86% (mean 54%). Some authors have used non-standard olfactory measurement e.g. modified ethyl-alcohol threshold test 24 or an in-house identification test of 10 odours 25 neither containing details of control data. In one article, patients quarantined at home were instructed on how to make up their own smell and taste ingredients 26 , despite the existence of readily available standardised commercial test kits for home use. The 4-odour Pocket Smell Identification Test used by one group 27 or the 3-odour Quick Smell Identification Test (Q-SIT) employed by others 28,29 are more appropriate for rapid screening in the clinic, rather than large research projects. For example, a score of 3/3 correct answers on the Q-SIT is likely to indicate normal olfaction 30 but as emphasised by the authors, a value of 2/3 could represent either hyposmia or a normal result because of wide variance. Nonetheless, a score of 3/3 would help exclude anosmia where a low-cost, high-volume survey is required. The data from Iran 31,32 have been criticised unfairly because many of the 40 odours used in the identification kit were allegedly unfamiliar to Iranians 33 . However, the test used was in fact specifically modified to account for cultural differences 34 .
It is important to be aware of the time of olfactory assessment in calculating the prevalence of Cov-2 related smell impairment, whether based on questionnaires or psychophysical tests. Clearly the closer to acute symptom onset, the more chance of an abnormal result. In four instances this information is not supplied. Where the time of testing is supplied, this ranges from 4-37 days.
Taking into account the above reservations, there are just four more robust studies that have used standardised smell tests, have a control group, time of examining 14 days or less and adequate patient numbers, arbitrarily set at 45 or more 28,31,32,35 . With this reservation, it may be inferred a) that subjective awareness of smell impairment is highly variable i.e. 28%-49% b) olfactory impairment on objective testing is present in 84%-98%. c) in general, hospitalised patients who are assessed within 14 days of symptom onset have more abnormal smell tests (71%-98%). The picture for outpatients is less clear Compared to subjective patient reports, smell measurement will therefore uncover a further 40% -50% of proven Cov-2 infected people, indicating that the olfactory defect is near universal. In practical terms this means that an abnormal smell test may be present in someone with no symptoms and yet be capable of spreading the virus. Conversely, a normal standardised smell test such as the Sniffin' Sticks or UPSIT should help exclude the presence of Covid-19 and would be valuable for mass population screening.
A less clear picture is available for the sense of taste. Only 6 studies report taste measurement (Table 1) and details of a control group are not given in four of these. Just three centres 27,34,36 .
implemented a standardised measurement (taste strips) and documented a normal result in a total of 40 patients from two centres 27,36 with an abnormal value from one unit 34 (5/72; 7%). The other three 25,26, 37 used in-house tests and observed abnormalities ranging from 27%-49%. No reliable conclusion can be drawn from these limited observations.  The olfactory neuroepithelium has considerable capacity for regeneration, provided the stem cell layer is not damaged 41 . This process is unlikely to account for the rapid subjective recovery that in some instances appears complete in as little as 2 weeks ( Figure 5). Such swift improvement is more in keeping with resolution of inflammation/oedema surrounding the nasal neuro-epithelium as shown on by MRI (Figure 3) . There are insufficient reports relating to the prevalence and recovery of taste impairment. Potential risk for mouth breathers. There are multiple causes of mouth breathing. It relates usually to nasal obstruction from a displaced septum, congestion, polyps and a variety of developmental abnormalities of the nasal cavity including Down's syndrome. In some it is just a bad habit. Most snorers breathe through the mouth and there is evidence that people with obstructive sleep apnoea are mouth breathers 42 . In the San Diego study of olfaction and Covid-19 19 it was speculated that milder cases of COVID-19 may present with severe anosmia and higher self-reporting, compared to the undetected or slight hyposmia associated with moderate to severe COVID-19 cases. If correct, this dichotomy may relate in part, to an individual's pattern of inspiration. Thus, habitual nosebreathers would direct airborne virus into the nasal passages where there are multiple immunebased defence functions that serve as a primary mucosal immune barrier. e.g. the nasopharynxassociated lymphoid tissue 43 (NALT) known collectively as Waldeyer's ring. A mouth-breather would therefore bypass the nasal component i.e. the adenoid and tubal tonsils and have to rely on the laryngeal and lingual tonsils. In theory, those who have had tonsillectomy or adenoidectomy might be more susceptible to subsequent viral infection although the consensus view is against this contention 44,45 . A further defence mechanism favouring nose breathers relates to increased synthesis of sino-nasal nitric oxide (NO) which is an integral and highly conserved part of the host immune response. It acts as a first-line of defence against micro-organisms and upregulates ciliary motility. At low concentration, NO acts as a signalling molecule that promotes growth and activity of immune cells. At high concentrations it binds DNA, proteins and lipids, thereby inhibiting or killing target pathogens 46 . In support of this in the clinical setting 47  Olfactory testing elevates the detection rate of a defect by about 50% i.e. from around 30-40% according to symptoms, to more than 90% where based on measurement. The importance of smell testing as opposed to smell questioning cannot be emphasized more strongly The prevalence of smell symptoms and signs is so high that a person who has normal olfaction on testing by procedures listed below, is unlikely to be infected with Cov-2 or if so, their viral load would be low and unlikely to result in transmission to others. Where resources are limited it is suggested that a rapid screening test of olfaction could be used in field stations or airports as a