The most effective prevention strategy for atherosclerotic cardiovascular disease (ASCVD) is primary management, with the core measure of risk assessment. The existing prediction models for ASCVD for the elderly are not able to guide TCM primary management well. Therefore, it is necessary to integrate TCM elements into the development of prediction models to guide the primary management of ASCVD with combined traditional Chinese and western medicine.

To construct and validate the ASCVD prediction model for the elderly based on TCM constitution.

A total of 1 418 elderly people who underwent physical examination at Huayuan Street Community Health Service Center, Chentangzhuang Street Community Health Service Center, Xiangyang Road Street Community Health Service Center and Daqiuzhuang Town Central Health Center in 2017 were included as the study subjects. General data of the study subjects were collected and constitution identification was performed. The incidence of ASCVD (clinical outcome) was followed up from 2017 to 2022. The follow-up will end at 2022-11-30. The data of the subjects were randomly divided into a training set (n=1 127) and validation set (n=291) according to 8∶2. In the training set, the conventional ASCVD prediction model for the elderly (model 1) and the conventional ASCVD+constitution prediction model for the elderly (model 2) were constructed by using the forward stepwise method. The nomogram of ASCVD prediction model for the elderly based on TCM constitution was plotted. The calibration curve was plotted and the Hosmer-Lemeshow goodness of fit test was performed to determine the calibration of the model. The receiver operating characteristic curve was plotted and the area under the curve (AUC) was calculated to determine the discrimination of the model. AUC, Net Reclassification Index (NRI), Integrated Discrimination Improvement (IDI), and Decision Curve Analysis (DCA) were used to compare model 2 with model 1 to evaluate the improvement efficacy of model 2.

There was no significant difference in the general data between the training set and validation set (P>0.05). The results of multivariate analysis showed that model 1 included 7 predictors of gender, age, waist circumference, systolic blood pressure, triacylglycerol (TG), BMI, systolic blood pressure×hypertension medication history. model 2 included 8 predictors of gender, age, waist circumference, systolic blood pressure, TG, BMI, systolic blood pressure×hypertension medication history, and constitution type. Hosmer-Lemeshow goodness-of-fit test showed good fit of model 2; Delong test results showed that AUC of model 2 was higher than that of model 1 (Z=2.741, P=0.006), NRI=0.511 (95%CI=0.359-0.663, P<0.001), IDI=0.038 (95%CI=0.024-0.051, P<0.001), suggesting that the addition of constitution predictors could improve the accuracy of model prediction. The clinical utility comparison results showed that the net benefit of model 2 to predict severe ASCVD events in the elderly was better than model 1 at a threshold probability of 5% to 74%.

In this study, a ASCVD prediction model for the elderly was constructed including 8 predictor variables of gender, age, waist circumference, systolic blood pressure, TG, BMI, systolic blood pressure×hypertension medication history, and constitution type. After testing, the differentiation and calibration performed well, which was better than the conventional prediction model, and can be applied to the individualized risk assessment of ASCVD in the elderly and guide the primary management of ASCVD with combined traditional Chinese and western medicine.