Decreasing Incidence of Chemosensory Changes by COVID-19 Variant

Abstract

Anecdotal clinical observation suggests that rates of chemosensory dysfunction associated with COVID-19 infection may be decreasing. To investigate, the National COVID Cohort Collaborative database was queried for all patients with and without smell and taste loss within 2 weeks of COVID-19 diagnosis. Six-week periods of peak variant prevalence were selected by using CoVariants.org for analysis. Of 3,678,214 patients with COVID-19 in the database, 616,318 met inclusion criteria during the time intervals of interest, with 3431 having an associated smell or taste disturbance diagnosis. With the initial/untyped variant set as the baseline, the odds ratios for alpha, delta, and omicron (December 27, 2021–February 7, 2022) were 0.50 (95% CI, 0.45–0.55; P < .0001), 0.44 (95% CI, 0.41–0.48; P < .0001), and 0.17 (95% CI, 0.15–0.18; P < .0001), respectively. These data strongly support the clinical observation that patients infected with more recent variants are at a significantly lower risk of developing associated chemosensory loss.

From the earliest days of the COVID-19 pandemic, chemosensory changes have been recognized as a cardinal symptom of infection. For most patients, recovery of normal or near-normal smell and taste function is expected.¹ For the roughly 20% of patients with persistent chemosensory dysfunction, the consequences to quality of life and safety can be substantial.² Yet as the pandemic persists and new variants of the virus arise, anecdotal clinical observation suggests that rates of associated smell and taste loss may be decreasing over time. The objective of this study is to use a large national database to determine whether rates of chemosensory dysfunction differ by variant type.

Methods

Based on data from CoVariants.org, a peak period for each variant (initial/untyped, alpha, delta, omicron) was defined as the 6 weeks with the highest prevalence of that particular variant in the United States.³ The N3C database⁴ (National COVID Cohort Collaborative) was queried for patients having a positive COVID-19 test result and an ICD R43 diagnosis of smell and taste disturbance occurring within 2 weeks of the test date. Demographic data were recorded. Relative rates of smell and taste loss were calculated by using the earliest group (initial/untyped) as the baseline. Statistical analysis among groups was performed by odds ratio calculations (95% CI) and chi-square testing. This study was approved by the Institutional Review Board at Virginia Commonwealth University (HM20022747).

Results

Of 3,678,214 patients with COVID-19 in the database, 616,318 met inclusion criteria during the time intervals of interest, with 3431 having an associated smell or taste disturbance diagnosis. Demographic data are displayed in Table 1. The total number of patients with COVID-19 and those with COVID-19 and chemosensory changes by variant are presented in Table 2. With the initial/untyped variant (June 22–August 3, 2020) set as the baseline, the odds ratios for alpha (April 19–May 31, 2021), delta (September 20–November 1, 2021), and omicron (December 27–February 7, 2022) are 0.50 (95% CI, 0.45–0.55; P < .0001), 0.44 (95% CI, 0.41–0.48; P < .0001), and 0.17 (95% CI, 0.15–0.18; P < .0001), respectively.

Table 1.

Demographics of Patients With COVID-19 and Smell/Taste Loss by Variant.

	Initial/untyped	Alpha	Delta	Omicron
Age, y (range, SD)	35.60 (18–83, 13.91)	39.18 (18–94, 15.18)	42.56 (18–93, 16.41)	41.21 (18–94, 15.92)
Sex, %
Female	58	58	58	58
Male	42	42	42	42
Race, %
White	42	47	64	51
Black	22	33	18	34
Other/unknown	36	19	18	16

Table 2.

Incidence of Chemosensory Change During 6-Week Variant Peaks.

	Initial/untyped	Alpha	Delta	Omicron
COVID-19	117,839	73,476	157,858	367,145
COVID-19 + S/T	1499	470	892	570
%	1.27	0.64	0.57	0.21
Odds ratio	—	0.50a	0.44a	0.17a
95% CI	—	0.45–0.55	0.41–0.48	0.15–0.18

Abbreviation: S/T, smell or taste disturbance.

^aP < .0001.

Discussion

Disparate symptomatology, spread, disease course, and rates of reinfection among variants have been described.^5,6 However, to our knowledge this is the first study to examine the relative rates of smell and taste change among COVID-19 variants, supporting clinical observations. Such findings may shed insight into the epidemiology and potential pathophysiology of COVID-19–associated chemosensory loss. Although studies comparing recovery rates of smell and taste loss among variants are ongoing, the present data can aid clinicians in patient counseling.

It should be noted that one shortcoming of this method is that rates of smell and taste loss are substantially less than the roughly 50% prevalence reported in a recent meta-analysis.⁷ All databases, including the N3C, depend on coding, which itself relies on clinicians asking about and documenting chemosensory changes. The low rates observed in our data suggest a uniform systematic limitation in how smell and taste losses are coded. It is conceivable, though unlikely, that as chemosensory dysfunction became well recognized and as the pandemic lingers, clinicians may feel less inclined to document such findings, leading to the observed downward trend with successive variant surges. Finally, although 100% of patients’ infections could be attributed to the specific variant during the delta and omicron periods, alpha cases likely represented only 60% of the cohort during the alpha period, so inclusion of other variants (iota, gamma, and “others”) may have altered the observed rate of chemosensory changes. Likewise, no specific variant can be associated with the initial/untyped cohort. As such, the data for these periods are not ideally homogenous to make as strong conclusions for these 2 variants as for the others. Nonetheless, taken together, these powerful data strongly support the clinical observation that patients infected with more recent variants are at a significantly lower risk of developing associated chemosensory loss.