Algoritmo para identificación de consultas por infección respiratoria aguda baja en pediatría en registros clínicos electrónicos

BACKGROUND: Due to ambiguities in terminology, acute lower respiratory infections (ALRI) in childhood are frequently not properly recorded, especially during outpatient visits. A tool that accurately identifies them, would assess the impact on respiratory health of massive harms, and design policies to prevent or mitigate their effects. We aimed to design an algorithm that allows identifying children with ALRI based on data from the electronic clinical record (ECR) of the Government of the City of Buenos Aires (GCBA). METHODS: From the ECR-GCBA database, we randomly selected 1000 outpatient visits of patients aged under 2 years. Terms showing that the visit was due to LARI were searched using an algorithm based on hard rules. Another dataset including 800 visits was used to adjust the algorithm and, finally, its performance was tested in a third dataset of 800 queries corresponding to the entire year 2018. RESULTS: In the validation set, our tool identified LARI with sensitivity 88.24%, specificity 97.5%, PPV 86.07% and NPV 97.93%. CONCLUSION: Our search algorithm allows us to identify with acceptable precision the outpatient visits related to LARI in children under 2 years of age from electronic clinical records.