Back

Risk factors for ICU admission due to severe acute respiratory infection in Brazil: A multicenter case-control study using national data

de Paula, R.; Cintra, P.; Andrade, E.; Pala, L. O.; Murata, R. M.; Pereira, L.

2025-11-22 epidemiology
10.1101/2025.11.20.25340681 medRxiv
Show abstract

ObjectiveTo identify risk factors associated with Intensive Care Unit admission due to Severe Acute Respiratory Infection in Brazil between 2021 and 2023, across different geographical levels during the COVID-19 pandemic. MethodsA retrospective case-control study was conducted using data from the OpenDataSUS system. Logistic regression models, within the framework of Generalized Additive Models for Location, Scale, and Shape, incorporated demographic, clinical, and vaccination variables. Model performance was assessed through cross-validation. ResultsMale sex, age between 60 and 70 years, dyspnea, respiratory distress, oxygen saturation <95%, diabetes, and obesity significantly increased the risk of Intensive Care Unit admission admission. Cough and sore throat were protective factors in the national, regional, and state-level analyses. COVID-19 and Influenza vaccination were consistently associated with reduced risk across all geographical levels, particularly among elderly patients. At the municipal level (Lavras, Minas Gerais), obesity had the greatest impact, while fever and odynophagia emerged as additional predictors, suggesting the influence of local factors such as socioeconomic conditions and hospital infrastructure. The models demonstrated moderate performance (AUC: 0.60-0.81). ConclusionsSevere respiratory symptoms, comorbidities, and male sex are key determinants of Severe Acute Respiratory Infection -related Intensive Care Unit admission. COVID-19 and Influenza vaccination showed a protective effect, especially in older adults, which may explain the observed reduction in risk after age 60. Regional differences underscore the importance of considering sociodemographic and structural factors in health planning and resource allocation.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.