A new study found that COVID-19 symptoms come in six different “clusters,” which may help predict which patients are more at risk and in need of respiratory support.
For the study, researchers used a machine learning algorithm to analyze data from more than 1,600 patients in the U.S. and the U.K. who had tested positive for COVID-19 and had regularly logged their symptoms on the Zoe health app in March and April. Researchers then tested the algorithm by using it on an additional 1,047 patients in the U.S., the U.K. and Sweden, who entered their symptoms on the app throughout May.
The study identified six clusters of symptoms:
Cluster 1 (“flulike” with no fever): Headache, loss of smell, muscle pains, cough, sore throat, chest pain, no fever
Cluster 2 (“flulike” with fever): Headache, loss of smell, cough, sore throat, hoarseness, fever, loss of appetite
Cluster 3 (gastrointestinal): Headache, loss of smell, loss of appetite, diarrhea, sore throat, chest pain, no cough
Cluster 4 (severe level one, fatigue): Headache, loss of smell, cough, fever, hoarseness, chest pain, fatigue
Cluster 5 (severe level two, confusion): Headache, loss of smell, loss of appetite, cough, fever, hoarseness, sore throat, chest pain, fatigue, confusion, muscle pain
Cluster 6 (severe level three, abdominal and respiratory): Headache, loss of smell, loss of appetite, cough, fever, hoarseness, sore throat, chest pain, fatigue, confusion, muscle pain, shortness of breath, diarrhea, abdominal pain
“Although continuous cough, fever and loss of smell (anosmia) are usually highlighted as the three key symptoms of COVID-19, data gathered from app users shows that people can experience a wide range of different symptoms, including headaches, muscle pains, fatigue, diarrhea, confusion, loss of appetite, shortness of breath and more,” notes a statement published on the Zoe app. “The progression and outcomes also vary significantly between people, ranging from mild flulike symptoms or a simple rash to severe or fatal disease.”
The researchers also looked at which clusters of patients were more likely to require ventilators or supplemental oxygen and found that patients in clusters 6, 5, and 4 (19.8 percent, 9.9 percent, and 8.6 percent, respectively) needed the most respiratory support. But “only 1.5 percent of people with cluster 1, 4.4 percent of people with cluster 2, and 3.3 percent of people with cluster 3 COVID-19 required breathing support,” according to a statement by King’s College of London, whose researchers, along with scientists at Massachusetts General Hospital, provided input during the development of the app for the study.
The study authors suggested that these results “could be used to monitor at-risk patients and predict medical resource requirements days before they are required,” adding, for example, that “patients who fall into cluster 5 or 6 at day 5 of the illness have a significant risk of hospitalization and respiratory support and may benefit from home pulse oximetry with daily phone calls from their general practice to ensure that hospital attendance occurs at the appropriate point in the course of their illness.”
Study co-author Claire Steves, a clinical senior lecturer at King’s College London, tells Yahoo Life: “This study helps us to see the different ways COVID can present. It alerts us to symptoms which are particularly worrying in the first few days if someone has a positive test — like confusion, abdominal symptoms and severe fatigue. More importantly, if we did start to monitor the symptoms in the first five days, we could get help to those that are on a more severe disease path earlier, and perhaps prevent them getting really sick.”
Linda Anegawa, an internist with virtual primary care platform PlushCare, tells Yahoo Life that the study results are “interesting because, up until now, we have mostly focused on patient characteristics, which [put] people at risk for severe disease. This is the first instance I’ve seen where actual symptoms are correlated with disease severity.”
She adds: “While further research is needed, identifying the six clusters now can start to help doctors risk-stratify patients they are seeing and allocate resources toward those patients who are felt more likely to be severely affected. This is critical since resources are scarce and doctors are spread thin, especially in surge areas.”
Anegawa also says that being able to identify which cluster corresponds with your symptoms “may help patients seek care earlier if they find their symptoms put them into one of the high-risk” categories.
Dr. Matthew Exline, a pulmonary and critical care specialist at the Ohio State University Wexner Medical Center, tells Yahoo Life that the research is “very helpful,” including for patients with mild coronavirus symptoms. “A lot of patients, when they’re told they have COVID, there’s obviously a lot of anxiety. Appropriately, we focus on what’s the worst-case scenario — the dangers if you end up on a ventilator, things like that.” But knowing you’re in a low-risk cluster could help ease that anxiety. Doctors being able to say, “‘Hey, a lot of people with your type of symptoms end up not needing much [medical] help’ is reassuring to patients and is very helpful,” he says.
Both Exline and Anegawa point out that the study, however, hasn’t been peer-reviewed. Anegawa also expressed some concern about trying to group patients in clusters: “I have seen many patients who do seem to fluctuate between multiple symptom clusters or who have characteristics of multiple clusters — for example, a patient with diarrhea only. How would we classify these types of individuals? It’s a tricky question.”
But overall, Exline says that using an app so COVID-19 patients can log their daily symptoms is “a great idea,” adding: “I think it’s an awesome use of the data. One of the things that’s really a challenge with COVID research is [that scientists have] so many questions and there’s just a finite number of people to try to answer those questions. There’s difficulty in human beings making hundreds of phone calls [to collect data on patients]. By asking willing people to enter their symptoms as it’s happening, you can get a lot of data.”
Anegawa agrees, adding: “In this era where virtual medicine is rapidly growing, health apps are a great way to help patients gather data for their telemedicine visits. In the future, I anticipate apps helping to screen patients with symptoms, again to save time and resources.”