• Fengming Ding, Lei Han, Dongning Yin, Yan Zhou, Yong Ji, Pengyu Zhang, Wensheng Wu, Jijing Chen, Zufang Wang, Xinxin Fan, Guoqing Zhang, and Min Zhang.
• Department of Respiratory and Critical Care Medicine, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.
• Bmc Med. 2022 Oct 18; 20 (1): 360.

BackgroundAcute febrile respiratory illness (AFRI) patients are susceptible to pneumonia and suffer from significant morbidity and mortality throughout the world. In primary care settings, the situation is worse. Limited by computerized tomography resources and physician experiences, AFRI patients in primary care settings may not be diagnosed appropriately, which would affect following treatment. In this study, we aimed to develop and validate a simple prediction model to help physicians quickly identify AFRI patients of pneumonia risk in primary care settings.MethodsA total of 1977 AFRI patients were enrolled at two fever clinics in Shanghai, China, and among them, 727 patients who underwent CT scans were included in the analysis. Acute alveolar or interstitial infiltrates found on CT images were diagnosed with pneumonia. Characteristics and blood parameters were compared between pneumonia and non-pneumonia patients. Then a multivariable model for pneumonia prediction was developed through logistic regression analysis. Its value for pneumonia prediction was prospectively assessed in an external multi-center population, which included 1299 AFRI patients in primary settings from 5 different provinces throughout China.ResultsIn the model development population, pneumonia patients (n = 227) had a longer duration of fever; higher frequencies of purulent sputum, dyspnea, and thoracic pain; and higher levels of respiration rates and C-reactive protein (CRP) than non-pneumonia patients (n = 500). Logistic regression analysis worked out a model composed of items on dyspnea, respiration rates > 20/min, and CRP > 20 mg/l (DRC) for pneumonia prediction with an area under curve (AUC) of 0.8506. In the external validation population, the predictive accuracy of the DRC model was the highest when choosing at least one positive item (1 score) as a cut-off point with a sensitivity of 87.0% and specificity of 80.5%. DRC scores increased with pneumonia severity and lung lobe involvement and showed good performance for both bacterial and viral pneumonia. For viral pneumonia, dyspnea plus respiration rates > 20/min had good predictive capacity regardless of CRP concentration.ConclusionsDRC model is a simple tool that predicts pneumonia among AFRI patients, which would help physicians utilize medical resources rationally in primary care settings.© 2022. The Author(s).

14 keywords...

hide…