Multi-state Models For The Identification Of Relevant Factors Of Cardiovascular Disease In Patients With Prediabetes

ObjectiveThis study aims to explore the natural history of the development of cardiovascular disease in the patients with prediabetes,evaluate the influencing factors and influencing levels at different transitions across a follow-up period,and provide scientific evidence for the preventive strategies of cardiovascular disease in patients with prediabetes.MethodsWe performed a cohort study of database records for residents who entered the health check-up program for the elderly people aged 60 years and above,in Xinzheng City,Henan province.The patients with prediabetes who participated in the health check-up program from January 2011 to May 2012 were enrolled in the study based on the inclusion and exclusion criteria.The information on the physical examination and chronic disease were collected during the follow-up period until December 2017.All statistical analyses were performed using R3.5.0 and Stata 14.2.The multi-state framework included two states(prediabetes and cardiovascular disease),three states(prediabetes,diabetes,and cardiovascular disease),and four states(prediabetes,normal blood glucose,diabetes,and cardiovascular disease).When existing two states,the Cox proportional hazards regression model was used to explore the risk factors associated with the onset of cardiovascular disease in patients with prediabetes.The non-parametric,semi-parametric,and parametric multi-state models were used in the multi-state framework with three states to explore the natural history of disease and relevant factors in different pathways.The parametric multi-state model was used in the four-state multi-state framework to explore the relevant factors in different pathways.Results1.A total of 1246 participants with prediabetes were included in the study.The median age(interquartile range)was 69.00(64.98,74.24)years.Individuals were observed for a mean of 5.44 years(median of 6.00 years).A total of 163 cases of incident cardiovascular disease occurred,and the incidence of cardiovascular disease was 240/10,000 person years.2.Multivariate Cox proportional hazard regression analysis showed that age,hypertension,overweight or obesity,smoking,increased resting heart rate,increased low-density lipoprotein cholesterol,increased triglyceride glucose index were significantly associated with the incident cardiovascular diseases(P < 0.05).3.In the non-parametric multi-state model with three states(prediabetes,diabetes,cardiovascular disease),patients with prediabetes who developed diabetes had a higher state occupation probability of cardiovascular disease than those who did not develop diabetes.In the parametric multi-state model with three states,hypertension,smoking,and triglyceride glucose index had statistically significant effects on the pathway from prediabetes to diabetes;age,sex,overweight or obesity,smoking,resting heart rate,low-density lipoprotein cholesterol had statistically significant effects on the pathway from prediabetes to cardiovascular diseases;drinking,smoking,and high-density lipoprotein cholesterol had statistically significant effects on the transition from diabetes to cardiovascular diseases.4.In the parametric multi-state model with four states(prediabetes,normal,diabetes,and cardiovascular diseases),the increases in total cholesterol and triglyceride glucose index were significantly associated with the pathway from prediabetes to normal blood glucose;high blood pressure and smoking had statistically significant effects on the transition from prediabetes to diabetes;smoking and low-density lipoprotein cholesterol had statistically significant effects on the transition from prediabetes to cardiovascular diseases;overweight or obesity had statistically significant effects on the transition from normal blood glucose to diabetes;men,high blood pressure,overweight or obesity,smoking,resting heart rate had statistically significant effects on the pathway from normal blood glucose to cardiovascular diseases;drinking,smoking,and resting heart rate were also significantly associated with the pathway from diabetes to cardiovascular diseases.ConclusionsMany factors such as hypertension,smoking,overweight or obesity,and triglyceride glucose index are associated with various endpoints describing the course of cardiovascular diseases in patients with prediabetes.In addition,the factors can play different roles in the different stages of the natural history of cardiovascular diseases.The multi-state model with states ≥ 3 can allow for the transition-specific prognostic factors and explore the changes in disease progression.