Retrospective Cohort Study On Relationship Between Blood Pressure Trajectories And Chronic Kidney Disease In Health Management Population

Background:Chronic kidney disease(CKD),a general term for heterogeneous disorders,can irreversibly affect structure and function of the kidney.The average prevalence of CKD in the world is about 13.4%,and that in China was about 10.8%which means the number of patients are as high as 120 million.The burden of disease caused by CKD has become a serious global public health problem,thus its prevention and treatment are urgent.At the early stage of CKD,there are typically no symptoms,while the diagnosis is often made when reaching the middle or late stages.The complications of CKD,including cardiovascular disease,hypertension,diabetes and other serious diseases,increase the risk of death.However,the long-term accumulation of various risk factors continuously damages tissues and organs,which results in severe outcomes in patients.Therefore,understanding the long-term evolution of risk factors will help us to understand the etiology of CKD,and contribute to early intervention and treatment of CKD.Blood pressure is an important modifiable risk factor of CKD.Many studies have demonstrated the importance of controlling blood pressure in preventing the development of CKD.It has been acknowledged that blood pressure varies over time.However,many studies only focused on the relationship between single measurement of blood pressure and the adverse health outcomes,ignoring the dynamic change of blood pressure over time.Understanding the long-term dynamic pattern of blood pressure can predict the risk of malignant outcome more accurately.Objective:To explore the trajectories of blood pressure change over time and its'association with CKD in health management population by utilizing Group-Based Trajectory Modeling method(GBTM),which will help us better understand the etiology of blood pressure on the development of CKD,and therefore allow for an improvement of early prevention and treatment guidance.Methods:Data were obtained from Shandong Multi-center Longitudinal Cohort of Health Management.Individuals aged 18-80 years old,with at least three blood pressure records,no previous history of CKD,and no missing important variables at baseline were selected.The trajectories of blood pressure include systolic blood pressure(SBP),diastolic blood pressure(DBP),pulse pressure difference(PP)and mean arterial pressure(MAP)were explored respectively using GBTM.The differences of baseline indexes,the incidence density of CKD and the survival outcome were investigated in different blood pressure trajectories.The Cox proportional risk regression models were used to explore the relationships between different blood pressure trajectories and CKD outcome.The goodness of fit and the area under ROC curve were used to evaluate the Cox regression models fitted by blood pressure traj ectories compared with traditional indexes.Results:1.Four distinct traj ectory groups of SBP were identified:low stable,middle stable,high stable and extreme-high fluctuation.Increasing trend was observed for baseline variables(age,BMI,SBP,DBP,PP,MAP,percentage of male,diabetes,hypertension and dyslipidemia)over four trajectory categories.Same results were detected for long-term follow-up of mean SBP,mean DBP,mean PP,mean MAP,SBP standard deviation,DBP standard deviation,PP standard deviation and MAP standard deviation.Compared to the low stable trajectory group,high stable group and extreme-high fluctuation group had a higher incidence density of CKD.The multivariate Cox regression model showed that,compared to the low stable trajectory group,the high stable group and extreme-high fluctuation group had greater odds of experiencing CKD.2.Four distinct trajectory groups of DBP were identified:low stable,middle stable,high stable and extreme-high fluctuation.Increasing trend was observed for baseline variables(age,BMI,SBP,DBP,PP,MAP,percentage of male,diabetes,hypertension and dyslipidemia)over four trajectory categories.Same results were detected for long-term follow-up of mean SBP,mean DBP,mean PP,mean MAP,SBP standard deviation,DBP standard deviation,PP standard deviation and MAP standard deviation.Compared to the low stable trajectory group,the high stable group and extreme-high fluctuation group had a higher incidence density of CKD.The multivariate Cox regression model showed that,compared to the low stable trajectory group,extreme-high fluctuation group had greater odds of experiencing CKD.3.Considering SBP and DBP trajectories together in the multivariate Cox regression model,high stable and extreme-high fluctuation trajectory groups had higher risk of CKD than low stable group for SBP,while no significant results were observed for the trajectory of DBP,Individuals were graded based on the trajectory risk groups they belonged to SBP and DBP,respectively.Risk score was calculated by adding up two risk grades for SBP and DBP.After adjusting for potential confounders in Cox model,risk scores showed significant association with the risk of CKD(P<0.05),and dose-response relationship was also observed.Conclusions:1.GBTM could be utilized to identify high risk subpopulation and dynamic change of blood pressure over time in health management population.The traj ectory could reflect the dynamic change of blood pressure over time,which was more representative than single measurement of blood pressure.2.Traj ectories of blood pressure were related to increase the risk of CKD after adjusting for potential confounding factors in Cox model.Compared with the low stable trajectory group,the risk of CKD was higher in high stable and the extreme-high fluctuation groups for SBP.Whereas,only extreme-high fluctuation group showed significant results for DBP.3.The trajectory could reflect the dynamic change of blood pressure over time,which was more representative than single measurement of blood pressure.The blood pressure trajectory performed better results regarding of goodness of fit and predictability in Cox regression model compared to baseline blood pressure and long-term standard deviation of blood pressure,while there was no obvious advantage over the average blood pressure.This was probably due to a short follow-up time of the study cohort and no dramatic change of blood pressure for the participants,which may lead to average blood pressure largely represented blood pressure trajectory.Therefore,the goodness of fit and predictability of the model should be considered,and further researches are needed to investigate these phenomena.