Less frequent coughs linked to increased risk for intubation, death in COVID-19 patients
Click Here to Manage Email Alerts
Patients hospitalized with COVID-19 who coughed 3.4 times per hour or less based on digital cough monitoring faced a greater risk for intubation or death, according to study results published in Journal of Biomedical Informatics.
“For the everyday physician, these findings provide an easily observed prognostic indicator (frequency of coughing) for hospitalized patients: if a patient is not coughing that much after they are admitted, they need to be more carefully monitored for clinical deterioration,” J. Glenn Morris, MD, MPH&TM, director of the University of Florida’s Emerging Pathogens Institute, told Healio.
The study was conducted simultaneously at University of Florida Health Shands Hospital by Morris and at Centre Hospitalier de l’Université de Montréal under the guidance of Simon Grandjean Lapierre, MD, MSc, FRCPC, assistant clinical professor in the microbiology, infectious diseases and immunology department at Université de Montréal. In total, Morris and Grandjean Lapierre evaluated 123 patients hospitalized with COVID-19 from two hospitals between Dec. 17, 2020, and June 15, 2021, to determine if time patterns of coughs monitored digitally could predict clinical prognosis and detect patients who face a higher risk for intubation or death.
Of the total cohort, 98 patients (mean age, 61 years; 52% men; 61% white) came from the University of Florida Health Shands Hospital, and 25 patients (mean age, 55 years; 60% men; 76% white) came from the Centre Hospitalier de l’Université de Montréal.
The Hyfe Research smartphone app, which features an artificial intelligence (AI) model that can differentiate coughing sounds from non-coughing sounds, was used to continuously record patients’ coughs until discharge, intubation or death.
Researchers tracked the hourly rate of coughs, as well as the time since hospitalization and study enrollment, in both cohorts. Using logistic models fitted on hourly coughs, researchers predicted outcomes among patients.
Of the total cohort, 16 patients required intubation and 13 patients died during the study period.
Researchers found that the digital cough monitoring system recorded 549.52 days and an average of 395 coughs per patient (282 per Florida patient; 822 per Montreal patient) over the study period.
Over the study period, researchers found that 3.4 coughs per hour was the maximum hourly cough rate of patients who went on to need mechanical ventilation or died (transitional cough rate). In the first 6 hours of monitoring, the transitional cough rate was 9.5 coughs per hour, but this decreased to 3.42 coughs per hour in the first 24 hours.
Researchers further found that frequent coughing (more than 3.4 coughs per hour) was related to less instances of intubation and death across the first 6 hours, 24 hours and total study period, which was unexpected, Morris told Healio. Notably, the Hyfe model had 94.1% sensitivity and 97.4% specificity, according to researchers.
“Starting with the descriptions of the first COVID cases, cough has always been recognized as a key element of COVID’s clinical presentation,” Grandjean Lapierre said. “Intuitively, one might think that patients who were coughing more frequently had a more serious infection. Instead, at least for hospitalized patients, our data show that patients who were coughing less frequently in the period shortly after admission were more likely to have a poor outcome, with greater oxygen needs and a higher likelihood of being admitted to the ICU.”
In terms of predicting unfavorable outcomes, the area under the curve (AUC) for cough frequency was 0.761 over the study period, and this measurement was similar to the AUC of 0.792 in the first 24 hours and of 0.715 in the first 6 hours. Researchers noted that since early measurements of cough frequency that predicted poor outcomes came close to the overall frequency measurement, monitoring coughs early on in hospitalized COVID-19 patients could be useful when making clinical decisions.
Further, sex, race, smoking status, diabetes, hypertension, cardiac disease, chronic obstructive pulmonary disease or asthma were not identified as predictors of unfavorable outcomes/potential biases. The only factor that was associated with unfavorable outcomes was age.
“At this point, this is a simple clinical observation,” Morris told Healio. “There is a need for further studies to get a feel for the underlying pathophysiologic mechanisms — put another way, why do people who cough less tend to have a poorer outcome? Understanding these underlying mechanisms, may, in turn, help in development of strategies to optimize clinical management of COVID-19 patients in hospital settings.”
For more information:
J. Glenn Morris, MD, MPH&TM, can be reached at jgmorris@epi.ufl.edu.
Simon Grandjean Lapierre, MD, can be reached at simon.grandjean.lapierre@umontreal.ca.