Identifying low-risk patients with acute respiratory tract infections for telehealth triage: A retrospective analysis

Background Acute respiratory tract infections (ARTIs) account for a large share of inappropriate prescribing in the outpatient setting. Telemedicine may offer one strategy to improve prescribing. We aimed to develop a risk prediction model to identify patients at low-risk for complications who can be safely seen via a telehealth visit for an ARTI. Methods We performed a retrospective analysis of adult primary care visits for patients with an ARTI seen in the outpatient setting at three Boston area hospitals from 2017-2021. We used aLasso regression for the final model to predict admission to an observation or inpatient unit within 14 days post ARTI visit in the derivation cohort and validated the model in two external cohorts. We evaluated prediction performance using the area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) along with their 95% confidence intervals. Results Approximately 1% of patients across all three cohorts had either an admission to an observation unit or an inpatient admission. The final prediction model included heart disease, diabetes, cancer, chronic obstructive pulmonary disease, insulin use, prior antibiotic prescription use, prior visit in the past 30 days and age. The AUC across the three cohorts ranged 0.72-0.81. The PPV ranged 0.02-0.04. The majority of patients were classified as low risk 74%-80% and only a small percentage (1%-3%) across all sites were classified as high risk for hospitalization. Conclusions Serious outcomes in patients with ARTIs are rare and most patient with an ARTI can be safely seen via telehealth.