[Non-traditional sources for health surveillance: using emergency data for early-detection]

Preparation and response to public health emergencies involve efforts in developing systems for early detection, alert and response. In this paper, we present the strategy implemented in Rio de Janeiro municipality, where data from urgency and emergency visits were acquired and modeled, in order to detect trend shifts and generate alerts. From the ICD-10 field in electronic records, time series representing events of interest were created. A generalized additive model (GAM) was fitted for smoothing, slope determination in each point, and alert generation. When comparing trends for COVID-19 between SIVEP/e-SUS systems and emergency attendances, a gain in speed was identified. On diarrhea monitoring, the generated alert led to coordinated communication and actions in the territory. We draw attention to the potential in the use of this type of data on identifying events of interest in a timely manner, promoting a more modern approach to health surveillance.