Abstract
BACKGROUND: Cardiovascular disease (CVD) is a common comorbidity associated with chronic obstructive pulmonary disease (COPD), but few studies have been conducted to identify CVD risk in COPD patients. This study was to develop a predictive model of CVD in COPD patients based on the National Health and Nutrition Examination Survey (NHANES) database. METHODS: A total of 3,226 COPD patients were retrieved from NHANES 2007-2012, dividing into the training (n = 2351) and testing (n = 895) sets. The prediction models were conducted using the multivariable logistic regression and random forest analyses, respectively. Receiver operating characteristic (ROC) curves, area under the curves (AUC) and internal validation were used to assess the predictive performance of models. RESULTS: The logistic regression model for predicting the risk of CVD was developed regarding age, gender, body mass index (BMI), high-density lipoprotein (HDL), glycosylated hemoglobin (HbA1c), family history of heart disease, and stayed overnight in the hospital due to illness last year, which the AUC of the internal validation was 0.741. According to the random forest analysis, the important variables-associated with CVD risk were screened including smoking (NNAL and cotinine), HbA1c, HDL, age, gender, diastolic blood pressure, poverty income ratio, BMI, systolic blood pressure, and sedentary activity per day. The AUC of the internal validation was 0.984, indicating the random forest model for predicting the CVD risk in COPD cases was superior to the logistic regression model. CONCLUSION: The random forest model performed better predictive effectiveness for the cardiovascular risk among COPD patients, which may be useful for clinicians to guide the clinical practice.