Relationship between odor intensity estimates and COVID-19 prevalence prediction in a Swedish population

Abstract

In response to the COVID-19 pandemic, countries have implemented various strategies to reduce and slow the spread of the disease in the general population. For countries that have implemented restrictions on its population in a step-wise manner, monitoring of COVID-19 prevalence is of importance to guide decision on when to impose new, or when to abolish old, restrictions. We are here determining whether measures of odor intensity in a large sample can serve as one such measure. Online measures of how intense common household odors are perceived and symptoms of COVID-19 were collected from 2440 Swedes. Average odor intensity ratings were then compared to predicted COVID-19 population prevalence over time in the Swedish population and were found to closely track each other (r=-0.83). Moreover, we found that there was a large difference in rated intensity between individuals with and without COVID-19 symptoms and number of symptoms was related to odor intensity ratings. Finally, we found that individuals progressing from reporting no symptoms to subsequently reporting COVID-19 symptoms demonstrated a large drop in olfactory performance. These data suggest that measures of odor intensity, if obtained in a large and representative sample, can be used as an indicator of COVID-19 disease in the general population. Importantly, this simple measure could easily be implemented in countries without widespread access to COVID-19 testing or implemented as a fast early response before wide-spread testing can be facilitated.

Keywords: COVID-19; Coronavirus; anosmia; olfactory dysfunction; population prevalence.

Figure 1.

Odor intensity perception relate to COVID-19 prevalence. (A) Mean intensity ratings of the 5 odor categories (blue line and axis) in relation to population prediction (black line and axis) of COVID-19 prevalence in the Stockholm region. (B) Mean intensity ratings of unimodal odors (odor categories 1 and 2; blue line and axis) in relation to population prediction of COVID-19 prevalence in the Stockholm region. (C) Mean intensity ratings of bimodal odors (odor categories 3–5; blue line and axis) in relation to population prediction of COVID-19 prevalence in the Stockholm region. (D) Mean intensity ratings of odors (categories 1–5), separated into individuals without (green squares, blue axis) and with (purple squares, blue axis) reported COVID-19 symptoms, in relation to population prediction (black line and axis) of COVID-19 prevalence in the Stockholm region. Error bars in all panels indicate standard error of the mean (SEM).

Figure 2.

Odor intensity perception relates to COVID-19 symptoms. Individual mean rated intensity of odors in relation to number of reported COVID-19 symptoms, excluding loss of smell/taste. Dots represent individuals and red dotted line indicates the regression line. Blue color indicates the number of overlapping individuals.

Figure 3.

Olfactory dysfunction in relationship to COVID-19 symptoms. (A) Percentage of olfactory dysfunction within subsample that indicated either COVID-19 symptoms (sessions, n = 2469) or had undergone COVID-19 testing (Covid-19 + = positive [n = 16], Covid-19 − = negative [n = 25]). (B) Shift in intensity ratings between sessions for individuals that progressed from indicated “No Symptoms” to indicating “Symptoms”. Dots indicate individual values (n = 107) and lines connects the values for the same individual. Error bars indicate standard error of the mean (SEM).