Determinants of persistence and recovery of chronic coronavirus disease 2019 chemosensory dysfunction

Abstract

Background: In 2% to 4% of patients, coronavirus disease 2019 (COVID-19) chemosensory dysfunction (CSD) persists beyond 6 months, accounting for up to 4 million people in the United States. The predictors of persistence and recovery require further exploration.

Objective: We sought to define the predictors of recovery and assess the quality of CSD in registry subjects with self-reported persistent smell and taste dysfunction after COVID-19.

Methods: COVID-19 CSD participants (n = 408) from the 4 major waves of the pandemic completed questionnaires at 4 time points between 2021 and 2023, assessing demographics, sinonasal symptoms, and self-assessed recovery. Objective measurements of smell (UPSIT) and taste (BWETT) were performed on a subcohort (n = 108).

Results: In this chronic CSD cohort, the average symptom duration was 24 ± 5 months, with 70% of those who contracted COVID-19 in 2020 report ongoing dysfunction. Phantosmia and dysgeusia were most prevalent in the early waves of COVID-19, while most participants reported disrupted ability to distinguish scents and flavors as well as undulating chemosensory function. Subjects reported low incidence of subjective sinonasal symptoms but high prevalence of sleep and mood disturbance. Cigarette smoke phantosmia was predictive of persistence of CSD. Conversely, self-reported environmental allergies and hypertension were predictive of recovery, and dust mite allergies specifically were negative predictors of cigarette smoke phantosmia. Finally, no treatment resolved CSD, but nasal steroids were reported to be effective by recovered CSD subjects. Objective measures of both smell and taste were significantly reduced in patients with chronic CSD compared to controls.

Conclusions: Chronic COVID-19 CSD is a syndrome resistant to standard anti-inflammatory therapy. Preexisting environmental allergies and hypertension predict recovery, while cigarette smoke phantosmia predicts persistence.

Keywords: COVID-19; Chemosensory dysfunction; smell disruption; taste disruption.

Fig. 1.. More severe quality of distorted sensory experience in wave 1 and 2 compared to wave 3 and wave 4 of COVID19.

(A, B) Proportion of respondents (x-axis) who agreed or disagreed (color legend) with questionnaire items (y-axis). COVID19 waves grouped based on infection date: wave 1, n = 205; wave 2, n = 108; wave 3, n = 56; wave 4, n = 39. Numbers are the percentage reporting ‘agree.’ (C) Ability to distinguish the five basic tastes. Numbers are percent of participants who reported ‘normal taste.’ (A-C) Cumulative percent is <100% as those listing ‘no response’ are not shown in the bar plots but are accounted for in Table E3 and E4.

Fig. 2.. COVID19 CSD is a persistent syndrome resistant to standard anti-inflammatory treatment.

(A) Participants’ COVID19 infection dates (chronic CSD n = 286, recovered n = 120), categorized by the major waves of COVID19. (B) Persistence of COVID19 CSD in months (x-axis) since COVID19 infection. Y-axis represents frequency of CSD among participants of the indicated groups. (C) Median number of months of CSD (y-axis) by wave of COVID19 infection (x-axis) in the recovered group. Waves were compared using Kruskal-Wallis test with Dunn’s multiple comparisons. Error bars represent IQR. (D) Percent of participants (y-axis) who tried treatments for olfactory function (x-axis). (E) Percent of participants (y-axis) that reported subjective improvement of CSD after trying the denoted treatments (x-axis); P-values: Fisher’s exact test. Significant P-values denoted in the figure.

Fig. 3.. Sleep and mood disturbance and mucosal dryness but not rhinologic and extranasal domains of SNOT-22 are increased in chronic COVID19 CSD.

(A) SNOT-22 items (y-axis) grouped by domain. Bars represent counts of participants (n = 406 in total) who rated each item to varying degrees of severity (color legend). (B) SNOT-22 cumulative score and (C-E) domain scores for rhinologic, extranasal, and ear or facial (C), sleep (D) and affective and mood disturbance (E). Significant P-values from Mann-Whitney U-test denoted in the figures. Recovered (n = 120, black), chronic CSD (n = 286, blue) participants, each dot is a separate participant. (F) Percent of participants in the indicated groups who reported mucosal (mouth and/or nasal) dryness. Fisher’s exact test. Significant P-values denoted in the figure.

Fig. 4. Pre-existing environmental allergies and hypertension are associated with recovery, while cigarette smoke phantosmia is associated with COVID19 CSD persistence.

(A-B) Participants (y-axis) with the following COVID19 related symptom or status (x-axis) for (A) Percent of participants (y axis) and vaccination status by wave of COVID19 infection (x axis). (B) Percent of participants (y axis) with long-COVID19 symptoms. P-values: Fisher’s exact test, recovered (n =120) and chronic CSD (n = 286). (C, E, F) Forest plots indicating multiple logistic regression analysis of (C) pre-existing conditions (y-axis) in relation to recovery from COVID19 CSD; (E) phantom smells (y-axis) in relation to recovery status from COVID19 CSD; and (F) environmental allergies (y-axis) in relation to cigarette smoke phantosmia. Each point denotes the adjusted odds ratio, with error bars indicating 95% confidence intervals, Wald test, P < 0.05 values are shown in red. For (C): conditions reported by less than five participants were excluded from the model (stroke, lung disease, chronic kidney disease, COPD/emphysema, dementia/Alzheimer’s). For (E) “Other” included: sweet/artificial sweetness, wood fire smoke, metallic/hot metal, and chemicals. (D) Percent of participants (y-axis) with phantosmia, Fisher’s exact test.

Fig. 5.. Objective measurements of olfactory and gustatory function validate self-report of COVID-19 CSD.

(A, B) Cumulative UPSIT smell (A) and B-WETT taste (B) test scores controls (n = 23), chronic CSD (n = 75), and recovered CSD (n = 10). (C, D) Cumulative UPSIT (C) and B-WETT (D) test scores for chronic CSD participants based on the COVID19 diagnosis wave. Center lines in A-D denote median and error bars IQR. (E-I) B-WETT taste subscores for sweet (E), bitter (F), sour (G), salty (H), and umami/savory (I) in controls, chronic CSD and recovered (as in A, B). Black center line represents the mean and error bars 95% confidence interval. For all panels P-values for significant comparisons are indicated in the graphs based on Kruskal-Wallis test with Dunn’s multiple comparisons correction.