O2202: Antibiotic management of pediatric acute respiratory tract infections when telemedicine is integrated into primary care

Background: Previous work raised concerns about the quality of antibiotic management for pediatric acute respiratory tract infections (ARTIs) in direct-to-consumer telemedicine visits with virtual-only vendors. However, emerging data from single health systems suggest antibiotic prescribing may be more judicious when telemedicine is integrated within primary care practices. Our objective was to examine telemedicine and in-person antibiotic management for children presenting to their primary care practice with ARTI symptoms.

Methods: We performed retrospective, cross-sectional analyses of 2023 electronic health record data from 843 U.S. pediatric and family medicine primary care sites. We identified telemedicine and in-person visits by children < 18 years for bacterial and viral ARTIs and constructed episodes of care, defined as an index visit (visit with no ARTI visit in prior 21 days) along with any follow-up visits (visits in subsequent 14 days). We created a propensity score model to estimate the probability of telemedicine use at each index visit, and we used inverse probability weights to balance confounding covariates between telemedicine and in-person visits. After applying propensity score weights, we compared antibiotic receipt, diagnoses, guideline concordant antibiotic management, and follow-up care for ARTI episodes beginning with telemedicine vs in-person index visits. We calculated 95% confidence intervals for outcomes using robust standard errors accounting for weighting and clustering within practices.

Results: We analyzed 540,607 ARTI episodes (Table 1). After application of propensity score weights, absolute standardized bias was < 0.15 for all variables. Compared to in-person index visits, telemedicine index visits were less likely to receive an antibiotic prescription (35% telemedicine [95% confidence interval (CI) 27-43%] vs 47% in-person [95% CI 45-48%] Table 2) and more likely to receive a viral diagnosis (66% telemedicine [95% CI 58-73%] vs 56% in-person [95% CI 54-57%]). Telemedicine index visits had similar rates of guideline concordant antibiotic management (86% [95% CI 80-91%]) as in-person index visits (86% [95% CI 85-87%]). In the 14 days after the index visit, telemedicine index vs in-person visits had higher rates of follow-up visits (14% telemedicine [95% 9-20%] vs. 9% in-person [95% CI 8-9%]), but no significant difference in subsequent antibiotic prescriptions (3% telemedicine [95% 1-5%] vs. 4% in-person [95% 3-4%]).

Conclusion: Despite concerns about antibiotic prescribing for children via telemedicine with direct-to-consumer virtual-only vendors, we found evidence of high-quality antibiotic management for children during telemedicine integrated within primary care practices. Our results suggest that future telemedicine approaches should account for varying quality of telemedicine care in different telemedicine models and support ongoing integration of telemedicine within pediatric primary care.