Widespread smell testing for COVID-19 has limited application – Authors' reply

Since the beginning of the COVID-19 pandemic, there has been much debate worldwide about which of the COVID-19 symptoms should be used for contact tracing to contain the viral spread. In April, 2020, we showed that loss of smell was the strongest single predictor of COVID-19,¹ and anosmia was finally added to Public Health England's COVID-19 case definition on May 18, 2020.²

In our previous Correspondence,³ we also suggested that low-cost so-called smell the difference screening tests could be implemented in some settings to capture a larger number of positives than temperature sensors do. We highlighted the greater potential importance of anosmia, as fever was present only in 42·7% of individuals testing positive, versus anosmia being present in 64·6% (34·7% of those not suffering from fever).²

However, we agree with Claire Hopkins and Barry Smith that recommendations for screening should be more cautious as a high percentage of the population, especially older people, have a pre-existing diminished sense of smell⁴ and should not be discriminated against. There is also the problem of long duration symptoms, increasingly known as long COVID.

Using the COVID Symptoms Study app, which now has more than 4 million users,⁵ we confirm that anosmia is still the single most predictive symptom of a positive swab test across different age groups, with odds ratios ranging from 13·67 (95% CI 11·65–16·02) for the older group to 20·86 (18·62–23·4) for the younger one (appendix).

We have also been able to collect data on symptoms duration. Of the 4182 adult app users who (1) were healthy at the time of sign-up, (2) tested positive for severe acute respiratory syndrome coronavirus 2 after registration on the app, with onset of symptoms occurring between 14 days before and 7 days after PCR test, and (3) logged regularly (ie, no gap in reporting of more than 7 days), we found that 54·5% reported anosmia in the first week. In this subset, anosmia appeared on day 2 (IQR 0-5) and had a median duration of 6 (3–11) days, but in 10% of cases, anosmia could persist for 3 weeks or more. In contrast, fever lasted a median of 3 days (1–7), although 10% reported it lasting more than 11 days and only 3·6% for more than 21 days.⁵

These data suggest that those people with new-onset anosmia should self-isolate and seek testing. However, as anosmia or dysosmia can often be present long after the first 10 days when transmission is less likely,⁶ when used as a screening tool, it is crucial to consider the onset of symptoms so as not to discriminate against older people or those with long-term symptoms.