Smells like a variant: How the association between COVID-19 and olfactory dysfunction changed between 2019 and 2022

Abstract

Dear Editor,

It is well established that olfactory dysfunction, including partial smell loss (hyposmia), total smell loss (anosmia), and distorted smell (parosmia), is associated with COVID-19. Case studies reporting this association appeared as early as April 2020. It was estimated that olfactory disturbances affected more than 60% of patients with symptomatic SARS-CoV-2 infection, with olfactory symptoms often developing after other infectious symptoms [1, 2]. Loss of taste or smell, if persistent, can affect eating habits and overall wellness and has even been linked to serious mental health disorders like anxiety and depression [3]. Predictors of these symptoms in the setting of COVID-19 include female sex, younger age, and fever [4, 5]. More recent data suggest that the positive predictive value of olfactory dysfunction for COVID-19 diminished with the emergence of vaccines and later variants [6]. Most notably, two related studies indicate that chemosensory loss due to COVID has declined dramatically, suggesting prevalence rates of 3%–4% during omicron waves [7, 8]. We sought to assess variations in the prevalence of olfactory disturbance diagnoses in a large healthcare system before and during the COVID-19 pandemic and compare these to population rates of SARS-CoV-2 infection.

We conducted a retrospective cohort study among patients 18 years or older between 1/1/2019 and 10/31/2022 with active membership in Kaiser Permanente Northern California (KPNC) and with at least one olfactory disturbance diagnosis with or without taste disturbances (ICD-10 codes R43.0, R43.1, R43.8, and R43.9) for any encounter type (inpatient, outpatient, in-system, and claims). Our cohort includes patients with diagnoses at any time during our study period, not limited to those temporally associated with a documented SARS-CoV-2 infection. We assigned 32 months to five distinct periods of variant dominance and examined temporal trends in olfactory disturbance diagnoses alongside the population incidence of SARS-CoV-2 infection. We calculated the monthly rate of olfactory disturbance diagnoses per COVID-19 diagnoses (per 100,000 health plan members) for each variant and tested for differences with a Kruskal–Wallis test.

Our retrospective review identified 66,067 olfactory disturbance diagnoses among a cohort of 23,570 patients, with 72.1% of patients receiving more than one related diagnosis during the study period. The most common encounter types were outpatient clinic visits and scheduled telephone visits. Patient median age was 46.1 (IQR 32.1–61.4) years, and 61.0% were female, with a median of 2 encounters for olfactory disturbance diagnoses per patient (IQR 1–4). Figure 1 depicts temporal trends of index olfactory disturbance diagnoses alongside the population incidence of COVID-19 (per 100,000 patients). The median monthly rate of olfactory disturbance diagnoses varied with statistical significance across periods of variant dominance: initial variant, 2.32; epsilon, 1.38; alpha, 2.58; delta, 1.31; and omicron, 0.37 (p = 0.0006).

Our results reveal a varying association between olfactory disturbance diagnoses and COVID-19-dominant variants. Prior to the pandemic, care-seeking for olfactory dysfunction was rare. Subsequently, sharp increases in the rates of olfactory disturbance diagnoses coincided with an increasing incidence of SARS-CoV-2 infection. Figure 1 demonstrates this temporal association. The declining rates of olfactory disturbance diagnoses relative to COVID-19 between the delta and omicron periods are consistent with prior studies reporting differing symptomology of these variants [9] and further validate observations of declining prevalence of COVID-related chemosensory loss during later stages of the pandemic [7, 8]. Given the timing of the emergence of different SARS-CoV-2 variants, our data reinforce that pre-omicron variants of the disease were more likely to lead to olfactory dysfunction. These changes may reflect the rapidly evolving genetic profile of the SARS-CoV-2 virus and the presence of several novel mutations in the spike protein of the omicron variant, affecting the mechanism, frequency, and location of viral entry [10, 11]. Despite higher observed transmissibility, omicron may be less effective at recognizing and fusing with certain surface receptor proteins and produce a lesser viral load, resulting in less inflammation and direct damage in the olfactory epithelium as seen in patients with COVID-related olfactory deficits [11, 12]. The immunity conferred by prior infection and vaccinations may also have mitigated the risk of SARS-CoV-2-related olfactory dysfunction.

This study is limited to patients included in the observed data set and study period. The established cohort was identified using specific ICD-10 codes (from a single-healthcare system and geography) and may not include all patients affected by olfactory disturbances. Additionally, smell loss is a self-reported symptom; the setting does not employ a standardized, objective collection of sensory deficits.

Further, changing temporal trends in care-seeking and reporting behavior throughout the pandemic likely contributed to the observed declines in both diagnoses. The association between olfactory dysfunction and COVID-19 was widely reported and understood by the public; therefore, the expectation of associated sensory symptoms with COVID-19 likely reduced patients’ concern with and desire to seek care for such symptoms. The widespread availability of rapid at-home testing for SARS-CoV-2 during later stages of the pandemic may also have reduced confirmed cases in the health record.

In conclusion, olfactory disturbance diagnoses were differentially associated with population COVID-19 variants and waves, becoming less consistently associated with the infection over time. These data reinforce that each variant presents with a distinct symptom profile, which may or may not include sensory symptoms; therefore, olfactory dysfunction may no longer be a reliable indication of SARS-CoV-2 infection. The observed trends may be explained by varying disease symptomology across variants, mitigation of the disease by immunizations and treatments, increases in natural immunity from prior infection, and changes to COVID-19 care-seeking behavior over time. Further study will be required to track this association in the future as new variants emerge.

All the authors declare no conflicts of interest.

The Permanente Medical Group Delivery Science Research Program

The Kaiser Permanente Northern California Institutional Review Board approved this study with a waiver of informed consent.