Patient-Reported Symptom Data May Hinder Telehealth-Based Flu Diagnosis

Researchers from the University of Georgia’s College of Public Health (UGA Public Health) demonstrated that clinical decision rules (CDRs) for influenza may be less accurate in a telemedicine setting, in which diagnosis relies on accurate symptom reporting by patients.

The research team noted that telehealth visits can help keep sick people out of community spaces and reduce the spread of illnesses like the flu, but virtual visits require that clinicians modify how they make diagnostic decisions.

“We know that telemedicine is working identifying high-risk patients, but we know that we can do better also,” said Zane Billings, an epidemiology and biostatistics doctoral student at UGA Public Health, in the press release.

In the clinical setting, care teams rely on established CDRs—diagnostic tools that typically leverage symptom presentation and lab tests to determine the likelihood that a patient has a disease—to guide their assessment of a patient with a suspected case of the flu.

Many of the CDRs for flu assume that a clinician is assessing the patient in person and can do things like listen to the patient’s breathing or take their temperature. However, during a virtual visit, clinicians must rely on patients’ self-reports of their symptoms.

The researchers sought to determine whether CDRs hold up whenever the clinician-patient interaction is limited by the parameters of a telemedicine visit.

The study, published recently in the Journal of the American Board of Family Medicine, evaluated the accuracy of existing CDRs for flu in a telehealth setting using only patient-reported symptoms.

The research team performed a secondary data analysis of a prospective, nonrandomized cohort of 250 college students who visited a university health center between December 2016 and February 2017. Each patient reported various symptoms associated with the flu.

Those who reported an upper respiratory complaint were required to report their symptoms, and their clinician was also required to report the same list of symptoms. From there, the researchers assessed the performance of five previously developed CDRs for flu on both symptom reports.

These predictions were then compared with each patient’s polymerase chain reaction (PCR) diagnosis. The research team then analyzed the agreement between symptom reports.

In doing so, the researchers could see how often clinicians and patients agreed on the likelihood of flu. Disagreement could undermine the validity of flu CDRs, they noted.

“What we found was that for most of these symptoms, patients and clinicians did not agree very often,” said Billings. “We want to know if a patient goes to a telemedicine appointment, can the patient self-report their symptoms and we still get an accurate diagnosis with that CDR?”

“If patients and clinicians are agreeing with each other on what symptoms the patient has almost all of the time, that means you can pretty much drop in those patient-reported symptoms instead. Whereas if patients and clinicians systematically disagree, we then wouldn’t expect those diagnostic rules to work anymore,” he continued.

After analyzing patient-clinician agreement, the research team built machine learning-based prediction models using both sets of data to establish whether a new CDR for flu could be developed based only on patient-reported symptoms.

One of the models used information from patient reports to predict the likelihood of flu, while the other leveraged clinician-reported symptoms to do the same.

The model using clinician-reported system data yielded more accurate diagnoses than the model using patient-reported data. The CDRs assessed in the study also often provided different predictions for the same patient due to disagreement in symptom reporting.

The researchers indicated that deriving an accurate diagnosis from these data is challenging in part because of the study’s sample. They noted that college students are typically healthy overall and at low risk for flu, making their symptoms more difficult to assess.

“If they do get the flu, it usually won’t be very severe. They might not be reporting as many symptoms as we might see for other people. Whereas if you have little kids and elderly people who get the flu, they’re often reporting more symptoms. It would be easier to diagnose them with the flu based on the information that we can get,” Billings stated.

The study authors would like their work to be replicated in a cohort with more individuals at higher risk for severe cases of flu, as doing so would potentially be more useful for creating accurate flu CDRs for telehealth. Those insights would also help determine which patients need more intensive, in-person care and which should stay home.

“Because if those people stay at home, people who aren’t in danger of complications stay at home, they’re not spreading the flu or whatever respiratory disease to other people,” said Billings.