| dc.description.abstract | Acute respiratory infections (ARIs) are the leading cause of morbidity and mortality in children and are often viral in origin. Viral and bacterial ARIs have overlapping clinical presentations, causing uncertainty in diagnosis and management and leading to inappropriate prescribing of antibiotics in children with viral ARIs, which can cause antibiotic resistant infections, medication-related adverse events, and unnecessary healthcare costs. Viral ARIs can be identified through respiratory viral testing; however, issues remain with accessibility and implementation of viral testing for children with ARIs. To understand the clinical value of testing for management of children with ARIs, we evaluated factors influencing provider-ordered viral testing and the association of viral testing with antibiotic prescribing across healthcare settings. We found children seeking medical care from the emergency department (ED) with a clinical pattern mimicking influenza/rhinovirus-like illness had 44% increased odds (OR:1.44; 95% CI:1.24, 1.68) of receiving viral testing compared to children with dissimilar clinical patterns. In contrast, children seen in the ED or hospitalized with a rhinovirus-like clinical characteristics had 71% (OR:0.29; 95% CI:0.24, 0.34) and 39% (OR:0.61; 95% CI:0.49, 0.76) decreased odds of receiving viral testing, respectively. In the ED, a child who received a viral test had 25% reduced odds (aOR:0.75; 95% CI:0.54, 0.98) of being prescribed a narrow-spectrum antibiotic compared to a child who did not receive viral testing. Hospitalized children had 57% increased odds (aOR:1.57; 95%:1.01, 2.44) of being administered a broad-spectrum antibiotic compared to a child who did not receive viral testing. A systematic review identified 18 studies that developed ARI prediction models in children. Due to differences seen across healthcare settings, we developed viral ARI diagnostic prediction models in the ED (c-indexED:0.74) and inpatient (c-indexinpatient:0.70) settings. Although viral testing varied by healthcare setting, it did not impact clinical management and is currently used according to recommended protocols. However, the diagnostic prediction model developed as part of this dissertation may serve as a cost-effective strategy to predict viral ARI in children (via mathematical formula or nomogram). Efforts to improve the widespread availability and implementation of diagnostic prediction models are still needed. | |
