Multidimensional nomogram for prediction of cardiovascular disease risk in cancer survivors.

Ten factors were significantly associated with CVD risk among cancer survivors on univariate analysis.

• The study included 1766 cancer survivors from the 2011-2018 NHANES database.
• Significant variables identified on univariate analysis included age, marital status, family income poverty index ratio, total cholesterol, hypertension, diabetes, smoking, depression degree, sedentary time, and sleep time.
• All associations were statistically significant at P < 0.05.
• The cross-sectional design of NHANES was used, precluding causal inference.

The XGBoost model identified six factors independently associated with CVD in cancer survivors.

• The XGBoost model was used to assess variable importance among the candidate predictors.
• The six factors identified were age, marital status, total cholesterol, hypertension, diabetes, and depression degree.
• This variable selection step reduced the predictor set from ten univariate candidates to six for inclusion in the final model.

Multivariate logistic regression identified increased age, divorced or cohabitation status, hypertension, and moderately severe or severe depression as significantly associated with increased CVD likelihood.

• Elevated total cholesterol was associated with a reduced likelihood of CVD risk in the multivariate analysis.
• The nomogram prediction model was constructed based on these multivariate logistic regression results.
• Psychosocial variables, specifically depression degree and marital status, were retained as independent predictors alongside traditional biomedical indicators.

The nomogram model demonstrated moderate discriminative performance with an AUC of 0.734 for assessing CVD risk in cancer survivors.

• Model performance was evaluated using receiver operating characteristic (ROC) curve analysis.
• The area under the ROC curve (AUC) was 0.734.
• The calibration curve showed high consistency between the predicted and actual risks.
• Decision curve analysis confirmed that the model had a net clinical benefit over a wide range of threshold probabilities.

The study supports integrating psychosocial variables with traditional biomedical indicators for CVD risk stratification in cancer survivors.

• Depression degree and marital status were included as psychosocial predictors in the final nomogram alongside biomedical variables.
• The authors concluded that this integration is useful for risk stratification in this population.
• The cross-sectional design limits causal inference from the findings.
• The study population comprised cancer survivors drawn from the NHANES 2011-2018 database (n = 1766).