| Objective: The research on blood lipid and metabolic syndrome(MetS)mostly focuses on a single indicator,few studies evaluate how lipid indicators can be used together to estimate the risk of MetS.This study was based on eight-year longitudinal data from healthy physical examination populations,used principal component analysis(PCA)and joint model to summarize blood lipid information,reveal its inherent physiological connections,and analyze the association between dynamic changes in principal components and the occurrence of MetS.Method: Collect the medical examination data from a medical examination center of a top three hospital in Dalian from January 2010 to December 2017,and establish a longitudinal medical examination database.IBM SPSS 23.0 software was used for data collation and basic description,SPSS was used to draw PCA and ROC curves of blood lipid indicators,and R 3.2.5 software was uesd to establish joint model,and analyze the dynamic changes of PC and its correlation with MetS.Results: 1.A total of 5,486 study subjects were included in this follow-up study,including 1875 men(34.18%)and 3611 women(65.82%).The baseline age was 19 to 64 years.At the end of the follow-up,there were 772 study subjects occurred MetS(the incidence of MetS was 14.07%,3.85 / 100 person-years),included 395 men(incidence was 21.07%,6.11 / 100 person-years),and 377 females(incidence was10.44%,2.79 / 100 person-years).2.In this study,principal component analysis was performed on the annual mean value of blood lipid indicator of individuals who did not develop MetS during the followup process.Two principal components were extracted.The contribution rate of PC-1 was 51.33%,and the contribution rate of PC-2 was 34.15%.The cumulative contribution rate of the two main components is 85.48%.The factor load of PC-1 on TC and LDL-C is large.It is composed of TC and LDL-C at a ratio of approximately 1: 1.It is considered that it reflects "the retention level of blood lipid";the absolute factor load of PC-2 on TG and HDL-C is larger,consisting of 0.6 times TG and 0.9 times HDL-C,where HDL-C is positively related to PC-2,and TG is negatively related to it,which is considered to reflect "the level of blood lipid clearance".3.Based on the occurrence of MetS as a positive diagnosis,the ROC curves of different lipid indicator and main components of blood lipid were plotted for compare the diagnostic efficacy.As a result,the diagnosis of PC-1 was better than TC but worse than LDL-C,and PC-2 was better than TG and HDL-C.4.In the PC-1 joint model results,the LME sub-model part: at baseline,the average level of PC-1 in 34-year-old female population was 6.7644 mmol / L,while the baseline PC-1 level of male in the same age was 0.2175 mmol / L higher than that of female;the coefficient of age shows that PC-1 increases by 0.0518 mmol / L for each year of the individual's baseline age;the negative interaction between age and gender indicates that the growth rate of PC-1 level in male increased with baseline age is lower than that of female;the coefficient of time indicates that PC-1 increases by 0.0144 mmol / L for each additional year of follow-up time;there is a negative interaction between time and gender,and the annual average change rate of PC-1 in men with follow-up time is 0.0287 mmol/ L lower than that of women,PC-1 increases faster for women.The results of the combined model showed that the correlation coefficient ? was statistically significant(p <0.05),indicating that for each additional unit of PC-1 longitudally,the risk of MetS increased by 1.3002 times(95% CI: 1.1271,1.4998).5.In the results of the joint model of PC-2,the LME sub-model part: the average level of PC-2 in 34-yaer-old female population was 0.8766 mmol / L,while average level of PC-2 in men who is the same age was 3.9977 mmol / L lower than that of women;PC-2 decreases by 0.0045 mmol / L for each each year of the individual's baseline age;due to the negative interaction between age and gender,the reduction value of men with baseline age is lower than that of women;PC-2 decreases by 0.0125 mmol / L for each additional year of follow-up time;there is a negative interaction between time and gender,indicates that the amount of PC-2 change in men with unit follow-up time is lower than that of women 0.0108 mmol / L,that is,the PC-2 of female decreases faster.In the submodel of the Cox model,age was statistically significant(P <0.05),indicating that age and has an impact on the incidence of MetS.For each one-year increase in the individual's baseline age,the risk of MetS is 1.0119 times(95%CI: 1.0024,1.0215).The results of the joint model show that the correlation coefficient ? is statistically significant(p <0.05),which indicates that for each unit of PC-2 decreased longitudinally,the risk of MetS increased by 2.9586 times(95% CI: 2.3691,3.6955).Combined with the results of the joint model analysis of PC-1 and PC-2,it can be seen that the longitudally decrease of PC-2 has a greater impact on the occurrence of MetS than the longitudally increase of PC-1.Conclusion: PCA is performed on the annual mean value of the blood lipid indicator,and 2 main components are extracted,which are considered to reflect the retention level of blood lipid and clearance level of blood lipid respectively;the longitudal increase of PC-1 and the longitudal decrease of PC-2 are risk factors of MetS,especially the clearance level of blood lipids,has a greater impact on the occurrence of MetS.It is suggested to pay more attention to the dynamic changes of blood lipid indicator,improve the ability of blood lipid clearance through exercise and other methods,and intervene early to prevent the occurrence of MetS. |
PDF Full Text Request