Prediction models for the risk of cardiovascular diseases in Chinese patients with type 2 diabetes mellitus: a systematic review

Highlights

• Cardiovascular diseases prediction model in Type 2 Diabetes Mellitus patients.

• Summary of common and novel factors in prediction of CVDs.

• Renal function parameters in CVDs prediction.

Abstract

Objectives

Diabetes mellitus (DM) is a serious public health issue worldwide, and DM patients have higher risk of cardiovascular diseases (CVDs), which is the leading cause of DM-related deaths. China has the largest DM population, yet a robust model to predict CVDs in Chinese DM patients is still lacking. This systematic review is carried out to summarize existing models and identify potentially important predictors for CVDs in Chinese DM patients.

Study design

Systematic review.

Methods

Medline and Embase were searched for data from April 1st, 2011 to May 31st, 2018. A study was eligible if it developed CVD (defined as total CVD or any major cardiovascular component) risk prediction models or explored potential predictors of CVD specifically for Chinese people with type 2 DM. Standardized forms were utilized to extract information, appraise applicability, risk of bias, and availabilities.

Results

Five models and 29 studies focusing on potential predictors were identified. Models for a primary care setting, or to predict total CVD, are rare. A number of common predictors (e.g. age, sex, diabetes duration, smoking status, glycated hemoglobin (HbA1c), blood pressure, lipid profile, and treatment modalities) were observed in existing models, in which urine albumin:creatinine ratio (ACR) and estimated glomerular filtration rate (eGFR) are highly recommended for the Chinese population. Variability of blood pressure (BP) and HbA1c should be included in prediction model development as novel factors. Meanwhile, interactions between age, sex, and risk factors should also be considered.

Conclusions

A 10-year prediction model for CVD risk in Chinese type 2 DM patients is lacking and urgently needed. There is insufficient evidence to support the inclusion of other novel predictors in CVDs risk prediction functions for routine clinical use.

Introduction

Diabetes Mellitus (DM) is a well-recognized public health issue.¹ It has been estimated to affect 463 million people worldwide by 2019 and to cost 760 billion US dollars (USD) in global health expenditures annually.¹ China is a major epicenter in the diabetes epidemic,² with an estimated 116.4 million adults (25.1% of total cases globally) living with the condition by 2019.¹ The prevalence of DM in China has increased rapidly, rising from less than 1% in 1980 to 10.9% in 2013.³ In 2019, diabetes-related health expenditure in China was 109 billion USD, ranking only second to the United States.¹ It has been found that DM patients have a two- to fourfold increase in the risk for cardiovascular events when compared with the general population.⁴ Approximately 70% of DM-related deaths have been attributed to cardiovascular diseases (CVDs)⁵ with CVDs significantly increasing the medical costs of patients with DM.⁵ The American Heart Association (AHA) guidelines have therefore recommended that clinical care providers conduct regular assessments, as well as management of, cardiovascular risk factors for DM patients.⁶ Management should also be tailored to the CVD risk level of the patient as suggested by the American Diabetes Association (ADA).⁷

A number of CVD risk prediction models for patients with diabetes have been developed to assist clinicians to estimate the risk so that management can be tailored to the needs of the patient. However, the majority of these models were developed using data from predominantly Caucasian participants and do not perform well when applied to the Chinese population with diabetes due to ethnic differences in the prevalence of CVD events.⁸ For example, the risk of coronary heart disease (CHD) and heart failure (HF) is 50% lower among the Chinese population when compared with Caucasians,^9,10 while the risk of stroke is higher.¹¹ Such differences may be explained by variations in lifestyle behaviors and genetic factors as well as environmental influences.¹² Previous evidence has shown that the common risk prediction models, including the Framingham [10-year],^13,14 the United Kingdom Prospective Diabetes Study (UKPDS) [5-year],^15,16 and the Joint Asia Diabetes Evaluation (JADE) [5-year]^17,18 models, tend to overestimate the risk and show poor calibration power in predicting the observed events in the Chinese population with diabetes.^8,19 A robust model to accurately predict CVD in Chinese DM patients is still lacking but urgently needed so as to enable accurate risk stratification and management to prevent CVD complications in the world's largest DM population. Thus, we carried out a systematic review to summarize the existing models and to identify potentially important predictors for CVD in Chinese patients with diabetes. We follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist²⁰ in the reporting of this systematic review, which is displayed in Appendix 1.