# Establishment of a Predictive Model for Chronic Cough after Pulmonary Resection

## TL;DR

This study creates a model to predict chronic cough after lung surgery, using factors like breathing tests and surgical details to improve patient care.

## Contribution

The novel contribution is the development and validation of a predictive model for chronic cough after pulmonary resection using clinical and physiological factors.

## Key findings

- The model achieved an AUC of 0.954, indicating strong predictive accuracy.
- Key predictors include FEV1/FVC ratio, surgical approach, and lymph node dissection.
- The model showed high calibration and clinical utility via decision curve analysis.

## Abstract

背景与目的 肺部切除术后慢性咳嗽是最常见的并发症之一，严重影响患者术后生活质量，目前国内尚无关于肺部切除术后慢性咳嗽预测模型。因此，本研究旨在探讨肺部切除术后慢性咳嗽相关危险因素，构建预测模型并进行验证。 方法 回顾性分析2021年1月至2023年6月于中国科学技术大学附属第一医院接受肺部切除术的499例患者的临床资料和术后咳嗽情况，按7:3随机分配原则分为训练集（n=348）和验证集（n=151），根据训练集患者术后是否慢性咳嗽分为咳嗽组和非咳嗽组。使用中文版莱斯特咳嗽问卷（The Mandarin-Chinese version of Leicester cough questionnare, LCQ-MC）评估术前、术后咳嗽的严重程度及其对患者生活质量的影响，采用咳嗽视觉模拟量表（visual analog scale, VAS）和自拟的数字评分法（numerical rating scale, NRS）评估术后慢性咳嗽，采用单因素和多因素Logistic回归分析独立危险因素和模型构建，受试者工作特征（receiver operator characteristic, ROC）曲线评估模型区分度，校准曲线评估模型的一致性，绘制决策曲线分析（decision curve analysis, DCA）评估模型的临床应用价值。 结果 多因素Logistic分析筛选出术前用力呼气第1秒呼气量与用力肺活量比（forced expiratory volume in the first second/forced vital capacity, FEV1/FVC）、手术方式、行上纵隔淋巴结清扫、行隆突下淋巴结清扫、术后胸腔闭式引流时间是术后慢性咳嗽的独立危险因素，基于多因素分析结果构建列线图预测模型。ROC曲线下面积为0.954（95%CI: 0.930-0.978），最大约登指数所对应的临界值为0.171，此时敏感度为94.7%，特异度为86.6%。Bootstrap法抽样1000次，校准曲线图预测的肺部切除术后慢性咳嗽与实际发生风险高度一致。DCA显示当预测模型概率的预概率为0.1-0.9之间，患者表现为正的净收益。 结论 肺部切除术后慢性咳嗽严重影响患者生活质量。列线图的可视化展现形式有助于准确预测肺部切除术后慢性咳嗽，为临床决策提供支持。

Results of univariate analysis between cough group and non-cough group (training set)

Results of univariate analysis between cough group and non-cough group (training set)

A: Training set; B: Validation set. ROC: receiver operating characteristic; AUC: area under the curve.

A: Training set; B: Validation set.

A: Training set; B: Validation set. DCA: decision curve analysis.