App Data May Sniff Out Early Symptoms of COVID-19


People who tested positive for COVID-19 infection were more than six times more likely to report loss of smell and taste as early symptoms compared with people who tested negative, according to data from a smartphone-based app.

Among more than 18,000 participants using a smartphone-based app to report COVID-19 symptoms in the U.K. and the U.S. and who had undergone a test for SARS-CoV-2, 65% of those with a positive test result reported loss of smell and taste versus 21% of those with a negative test result (OR 6.74, 95% CI 6.31-7.21), reported Christina Menni, PhD, of King’s College London, and colleagues.

Moreover, odds were even higher when using only the U.S. subset of data (OR 10.01, 95% CI 8.23-12.16), the authors wrote in Nature Medicine.

Other unusual, potentially predictive symptoms of COVID-19 infection included fatigue and persistent cough with loss of appetite, they noted.

Widespread population testing for COVID-19 is not yet available in most countries, making it “important to identify the combination of symptoms most predictive of COVID-19,” the authors said.

They noted anosmia in particular may be present in absence of other symptoms, according to case reports and mainstream media articles, meaning, “this symptom could be used as a screening tool to help identify people with potential mild cases who could be recommended to self-isolate,” they wrote.

Researchers examined data from a symptom tracker smart phone app used by over 2.6 million individuals in the U.K. and the U.S. Individuals ranged in age from 16 to 90, with a BMI of 14-45, and over 168,000 individuals were from the U.S.

Interestingly, after adjusting for age, sex and BMI, the authors found out of 10 potential COVID-19 symptoms, only loss of smell and taste, fatigue, and skipped meals were associated with a positive test for COVID-19 in the U.S. cohort.

Using the U.K. cohort, researchers identified that a combination of loss of smell and taste, fatigue, persistent cough, and loss of appetite resulted in the best symptom prediction model for COVID-19. This model was validated in both the U.K. and the U.S. cohort with an area under the curve of the receiver operating characteristic curve (ROC-AUC) of 0.76 (95% CI 0.74-0.78) for each cohort, which may be considered “fair.”

However, when applying the predictive model to the over 800,000 individuals in the U.K. and U.S. who reported symptoms, but had yet to be tested for COVID-19, the authors found that, according to the model, about 17% of those reporting symptoms were likely to be infected, representing about 5% as a proportion of overall responders to the app, they noted.

Despite the self-reported nature of their data, the authors concluded that loss of smell and taste should be added to the World Health Organization’s COVID-19 symptom list and should be part of routine screening for the infection.

Last Updated May 11, 2020

  • author['full_name']

    Molly Walker is an associate editor, who covers infectious diseases for MedPage Today. She has a passion for evidence, data and public health. Follow


The study was supported by Zoe Global and the U.K. Research and Innovation London Medical Imaging & Artificial Intelligence Centre for Value-Based Healthcare.

Menni disclosed no relevant relationships with industry. Co-authors disclosed support or employment with Zoe Global, the company that helped develop the smartphone app.

Leave a Reply

Your email address will not be published. Required fields are marked *