Establishment of an Individual-Specific Nomogram for Predicting the Risk of Left Ventricular Hypertrophy in Chinese Postmenopausal Hypertensive Women.

Background and

Objectives

The physiological phenomenon peculiar to women, namely menopause, makes the occurrence of left ventricular hypertrophy (LVH) in postmenopausal hypertensive women more characteristic. Less is known about the risk of developing LVH in Chinese postmenopausal hypertensive women. Thus, the present study was intended to design a nomogram for predicting the risk of developing LVH in Chinese postmenopausal hypertensive women. Materials and

Methods

Postmenopausal hypertensive women aged between 49 and 68 years were divided into either the training set (n = 550) or the validation set (n = 284) in a 2:1 ratio. Patients in the validation set were followed up for one year. A stepwise multivariable logistic regression model was used to assess the predictors of LVH in postmenopausal women with hypertension. The best-fit nomogram was executed using R software. The calibration and decision curve were employed to verify the predictive accuracy of the nomogram. The results were evaluated in the validation set.

Results

Menopause age (OR = 0.929, 95% CI 0.866-0.998, p = 0.044), BMI (OR = 1.067, 95% CI 1.019-1.116, p = 0.005), morning systolic blood pressure (SBP: OR = 1.050, 95% CI 1.032-1.069, p = 0.000), morning diastolic BP (DBP OR = 1.055, 95% CI 1.028-1.083, p = 0.003), angiotensin II receptor blocker (ARB) utilization rate (OR = 0.219, 95% CI 0.131-0.365, p = 0.000), LDL-C (OR = 1.460, 95% CI 1.090-1.954, p = 0.011) and cardio-ankle vascular index (CAVI) (OR = 1.415, 95% CI 1.139-1.757, p = 0.028) were associated with LVH in postmenopausal hypertension patients. The nomogram model was then developed using these variables. The internal validation trial showed that the nomogram model described herein had good performance in discriminating a C-index of 0.881 (95% CI: 0.837-0.924) and high quality of calibration plots. External validation of LVH-predictive nomogram results showed that the area under the ROC curve was 0.903 (95%CI 0.900-0.907).

Conclusions

Our results indicate that the risk prediction nomogram model based on menopausal age, BMI, morning SBP, morning DBP, ARB utilization rate, LDL-C and CAVI has good accuracy and may provide useful references for the medical staff in the intuitive and individualized risk assessment in clinical practice.