Abstract: Objective To construct a predictive model of pulmonary hypertension from acute exacerbation of chronic obstructive pulmonary disease and evaluate its predictive value. Methods A total of 276 AECOPD patients admitted to the respiratory department of Chengde Central Hospital, Hebei Province from May 2020 to October 2022 were retrospectively selected, and the patients were randomly divided into a training set (n=196) and a validation set (n=80) according to a ratio of 7: 3. Clinical data and laboratory examination indexes of patients within 24 hours of hospitalization were collected. The effective risk predictors were obtained by LASSO regression, and the multi-factor logistic regression risk prediction model was constructed. The predictive power, differentiation, calibration, and clinical efficacy of the predictive model were evaluated using receiver ROC curve, C-index, calibration curve, and decision analysis curve, respectively. Validation sets were used for external validation. Results Lower BMI, elevated PaCO₂, BNP and combined lower limb venous thrombosis were independent risk factors for AECOPD combined with PH (P<0.05). In the training set and validation set, the area under ROC curve of the model was 0.938 [95% CI(0.907~0.969)] and 0.826 [95% CI(0.738~0.915)], respectively, and the threshold probability fluctuation of net benefit was 0.06~0.96 and 0.10~0.83. Conclusion The progression of PH in patients with AECOPD is related to lower BMI, elevated PaCO₂, BNP and combined lower limb venous thrombosis, and the prediction model established based on this is highly accurate and has great clinical application prospect.

Key words: chronic obstructive pulmonary disease, pulmonary hypertension, nomogram

摘要： 目的 构建慢性阻塞性肺疾病急性加重期（AECOPD）患者进展至肺动脉高压（PH）的预测模型并评估其预测价值。方法 回顾性选取2020年5月1日—2022年10月31日河北省承德市中心医院呼吸内科收治的AECOPD患者276例,将患者按7: 3比例随机分为训练集（n=196）和验证集（n=80）。收集患者住院24 h内的临床资料和实验室检查指标。使用最小绝对收缩选择法（LASSO）回归获得有效风险预测因子,并构建多因素回归风险预测模型。分别使用受试者工作特征曲线（ROC）、C指数、校准曲线及决策分析曲线评估预测模型的预测能力、区分度、校准度以及临床效能,使用验证集进行外部验证。结果 低体质量指数（BMI）、PaCO₂升高、B型脑钠肽（BNP）、合并下肢静脉血栓是AECOPD合并PH的独立危险因素（P<0.05）。在训练集和验证集中,模型的ROC曲线下面积分别为0.938[95% CI(0.907~0.969)]、0.826[95% CI(0.738~0.915)],净获益的阈概率波动分别在0.06~0.96、0.10~0.83。结论 AECOPD患者进展至PH与低BMI、PaCO₂升高、BNP、合并下肢静脉血栓相关,据此建立预测模型准确性较高,临床应用前景大。

关键词: 慢性阻塞性肺疾病, 肺动脉高压, 列线图