Development and validation of a hyperlipidemia risk prediction model for middle-aged and older adult Chinese using 2015 CHARLS data.

OBJECTIVE

To develop and validate a predictive model for hyperlipidemia risk among middle-aged and older adult individuals in China, this study aims to offer an effective screening tool for identifying those at risk.

METHODS

In this study, we included 6,629 middle-aged and older adult individuals, aged 45 and above, who met the inclusion criteria from the 2015 China Health and Retirement Longitudinal Study (CHARLS) as our research subjects. Utilizing the LASSO regression and multivariate Logistic regression method, we analyzed the independent risk factors associated with hyperlipidemia among these subjects. Subsequently, we established a risk prediction model for hyperlipidemia in the middle-aged and older adult population using statistical software Stata 17.0.

RESULTS

The prevalence rate of hyperlipidemia among the 6,629 middle-aged and older adult participants was 26.32% (1,745 out of 6,629). The LASSO regression and multivariate Logistic regression analysis all revealed that Body Mass Index (BMI), fasting blood glucose, serum uric acid, C-reactive protein, and white blood cell count were independent risk factors for hyperlipidemia in this demographic (with Odds Ratios (OR) greater than 1 and -values less than 0.05). From these findings, a nomogram prediction model was constructed to estimate the risk of hyperlipidemia for middle-aged and older adult individuals. The Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) for the nomograms was 0.717 (95% Confidence Interval (CI): 0.703-0.731), indicating good discrimination. The Decision Curve Analysis (DCA) demonstrated that when the probability of hyperlipidemia in the middle-aged and older adult population falls between 0.11 and 0.61, the application of the nomogram yields the highest net benefit, suggesting that the nomogram model possesses good clinical applicability. The Spiegelhalter's z-statistic test confirmed that the predicted probabilities from the nomogram model are in good agreement with the observed frequencies of hyperlipidemia ( = 0.560). The Brier score for the nomogram model was 17.1%, which is below the threshold of 25%, indicating good calibration. To internally validate the nomogram model, we performed bootstrap resampling 500 times. The C-statistic for the nomogram model from this internal validation was 0.716, and the Brier score was 11.4%, suggesting that the model not only has good predictive efficiency but also good stability.

CONCLUSION

The nomogram model, which incorporates the identified risk factors for hyperlipidemia in middle-aged and older adult individuals, has demonstrated good predictive efficiency and clinical applicability. It can serve as a valuable tool to assist healthcare professionals in screening for high-risk groups and implementing targeted preventive interventions. By doing so, it has the potential to significantly reduce the incidence of hyperlipidemia among this demographic.