Evaluation of Atherosclerotic Cardiovascular Risk Prediction Models in China: Results From the CHERRY Study

JACC Asia. 2022 Jan 4;2(1):33-43. doi: 10.1016/j.jacasi.2021.10.007. eCollection 2022 Feb.

Authors

Xiaofei Liu^{1

2}, Peng Shen³, Dudan Zhang⁴, Yexiang Sun³, Yi Chen¹, Jingyuan Liang¹, Jinguo Wu⁵, Jingyi Zhang⁵, Ping Lu⁵, Hongbo Lin³, Xun Tang¹, Pei Gao^{1

2

6}

Affiliations

¹ Department of Epidemiology and Biostatistics, Peking University Health Science Center, Beijing, China.
² Center for Real-world Evidence Evaluation, Peking University Clinical Research Institute, Beijing, China.
³ Yinzhou District Center for Disease Control and Prevention, Ningbo, China.
⁴ MRC/BHF Cardiovascular Epidemiology Unit, Department of Public Health and Primary Care, University of Cambridge, Strangeways Research Laboratory, Wort's Causeway, Cambridge, United Kingdom.
⁵ Wonders Information Co., Ltd, Shanghai, China.
⁶ Key Laboratory of Molecular Cardiovascular Sciences (Peking University), Ministry of Education, Beijing, China.

Abstract

Background: Updated American or Chinese guidelines recommended calculating atherosclerotic cardiovascular disease (ASCVD) risk using the Pooled Cohort Equations (PCE) or Prediction for Atherosclerotic Cardiovascular Disease Risk in China (China-PAR) models; however, evidence on performance of both models in Asian populations is limited.

Objectives: The authors aimed to evaluate the accuracy of the PCE or China-PAR models in a Chinese contemporary cohort.

Methods: Data were extracted from the CHERRY (CHinese Electronic health Records Research in Yinzhou) study. Participants aged 40 to 79 years without prior ASCVD at baseline from 2010 to 2016 were included. ASCVD was defined as nonfatal or fatal stroke, nonfatal myocardial infarction, and cardiovascular death. Models were assessed for discrimination and calibration.

Results: Among 226,406 participants, 5362 (2.37%) adults developed a first ASCVD event during a median of 4.60 years of follow-up. Both models had good discrimination: C-statistics in men were 0.763 (95% confidence interval [CI]: 0.754-0.773) for PCE and 0.758 (95% CI: 0.749-0.767) for China-PAR; C-statistics in women were 0.820 (95% CI: 0.812-0.829) for PCE and 0.811 (95% CI: 0.802-0.819) for China-PAR. The China-PAR model underpredicted risk by 20% in men and by 40% in women, especially in the highest-risk groups. However, PCE overestimated by 63% in men and inversely underestimated the risk by 34% in women with poor calibration (both P < 0.001). After recalibration, observed and predicted risks by recalibrated PCE were better aligned.

Conclusions: In this large-scale population-based study, both PCE and China-PAR had good discrimination in 5-year ASCVD risk prediction. China-PAR outperformed PCE in calibration, whereas recalibration equalized the performance of PCE and China-PAR. Further specific models are needed to improve accuracy in the highest-risk groups.

Keywords: ACC/AHA, American College of Cardiology/American Heart Association; ASCVD, atherosclerotic cardiovascular disease; Chinese; EHR, electronic health record; EMR, electronic medical record; PCE, pooled cohort equations; atherosclerotic cardiovascular disease; electronic health records; primary prevention; risk assessment.

© 2022 The Authors.