Abstract:
Objective To explore the factors associated with poor prognosis of cohort patients, and to test the ability of bronchiectasis severity index (BSI) and FEV1-Age-Chronic colonization by Pseudomonas aeruginosa-extension-dyspnoea (FACED) scoring systems in predicting poor prognosis of cohort patients.
Methods From January 1, 2017 to December 31, 2019, 160 patients diagnosed with bronchiectasis were recruited in Zhongshan Hospital, Fudan University to analyze this retrospective cohort study. All patients were followed up for 12 months by telephone to analyze the prognosis. Patients were divided into a good prognosis group (n=85) and a poor prognosis group (n=75) based on their prognosis. Multivariate logistic regression analysis model was used to identify risk factors related to prognosis. Baseline records provided data for determining BSI and FACED. ROC curve was drawn to analyze the predictive value of BSI and FACED scoring system for prognostic risk of these patients.
Results In the multivariate logistic regression analysis model, the disease duration≥10 years (OR=3.142, 95%CI 1.325-7.451), FEV1%pred < 50% (OR=5.988, 95%CI 1.715-20.833), BMI < 18.5 kg/m2 (OR=4.762, 95%CI 1.247-18.120), Pseudomonas aeruginosa positive (OR=3.534, 95%CI 1.135-11.007)and hemoptysis (OR=2.551, 95%CI 1.070-6.097) were independent predictors for poor prognosis (all P < 0.05). In ROC analysis, BSI (AUC=0.890)was more accurate in predicting poor prognosis than FACED (AUC=0.753)in clinical application. CT score, Modified British medical research council (mMRC) and BMI were combined to evaluate poor prognosis in patients with bronchiectasis, which had a good accuracy (AUC=0.842).
Conclusions Disease duration≥10 years, FEV1%pred < 50%, BMI < 18.5 kg/m2, Pseudomonas aeruginosa positivity and hemoptysis are independent predictors of poor prognosis. Both BSI and FACED scoring systems have a strong ability to predict poor prognosis in patients with bronchiectasis. The combination of CT score, mMRC and BMI is a simpler prognosis prediction model.