Researchers from the University of Georgia’s College of Public Health have found challenges in the diagnosis of influenza through telehealth, particularly concerning the accuracy of clinical decision rules (CDRs) in a virtual environment. In traditional clinical settings, CDRs are diagnostic tools that combine symptom presentation and lab tests to assess the likelihood of a patient having a disease like the flu. These tools are generally designed for in-person assessments, where clinicians can perform physical examinations such as listening to a patient’s breathing or taking their temperature. In a telehealth setting, clinicians must rely heavily on patients’ self-reports of symptoms, which raises questions about the accuracy and reliability of these diagnostic methods.
The study was conducted on a cohort of 250 college students who visited a university health center between December 2016 and February 2017, aimed to evaluate the effectiveness of existing CDRs for flu in a telemedicine setting using only patient-reported symptoms. The researchers compared the predictions of five previously developed CDRs for flu against each patient’s polymerase chain reaction (PCR) diagnosis. This comparison was important in assessing the agreement between symptom reports by clinicians and patients. The findings showed a frequent mismatch between patient and clinician assessments of symptoms, suggesting that current CDRs might not be as effective in a telehealth context where physical examination is not possible.
The implications of these findings are large for telemedicine, a mode of healthcare delivery that has seen a substantial increase in use, especially during the COVID-19 pandemic. Telehealth has been extremely useful in keeping sick individuals out of community spaces, thereby reducing the spread of illnesses. This change necessitates a reevaluation of how clinicians make diagnostic decisions in the absence of physical patient interaction. Zane Billings, an epidemiology and biostatistics doctoral student at UGA Public Health, detailed the need to improve telemedicine’s capability in identifying high-risk patients.
To explore the possibility of creating a new CDR based solely on patient-reported symptoms, the research team employed machine learning techniques. They developed prediction models using patient-reported data and clinician-reported symptoms to determine the likelihood of flu. The model using clinician-reported symptoms was more accurate than the one relying on patient reports. This finding points to the challenges in creating an effective CDR for telehealth that depends solely on patient-reported information. The researchers noted that the study’s sample, comprising mostly healthy college students at low risk for flu, might have contributed to the difficulty in deriving accurate diagnoses, as their symptoms were less pronounced compared to other demographics like young children and the elderly. The research team expressed a desire to replicate the study with a more diverse sample, including individuals at higher risk for severe flu cases. Further research could be more beneficial in creating accurate flu CDRs for telehealth applications. Accurately identifying patients who require more intensive in-person care and those who can safely receive care at home is important for effective healthcare management, especially during flu season, as the distinction helps prevent the spread of respiratory diseases in community settings and ensures that resources are allocated efficiently.