| Background:High sodium intake is an important risk factor for cardiovascular disease,and approximately 1.65 million cardiovascular deaths per year can be attributed to high sodium diets,accounting for 9.5%of total cardiovascular deaths.It is an important content of clinical research to actively evaluate patients’salt intake,so as to carry out targeted prevention and treatment measures such as salt restriction.The24-hour urine sodium can accurately assess salt intake,but its clinical application is limited by its long urine retention time,tedious operation and low patient compliance.In recent years,the exploration of spot urine has been gradually deepened,and spot urine is expected to replace 24-hour urine.Known spot urine sodium methods include INTERSALT method,Kawasaki method,Tanaka method,SUN＿C method,etc.However,the accuracy of spot urine sodium method is affected by race,gender,age,salt intake and other factors,and the established spot urine methods for the population in the southern Liaoning region of China are not perfect,therefore,it is of clinical significance to establish a new spot urine model suitable for the assessment of salt intake in the population in the southern Liaoning region of China.Objective:To establish a spot urine model suitable for the assessment of salt intake in the southern Liaoning region of China,and to compare and analyze it with other models at the population level and individual level.Method:A total of 2271 patients,aged 18 to 80 years old,admitted to the Department of Hypertension of the First Affiliated Hospital of Dalian Medical University from January 2019 to October 2022 were included.All of them received urine samples for 24 hours and fasting urine samples for a second time in the morning.Demography and clinical data were collected,including spot urine sodium concentration,spot urine creatinine concentration,spot urine potassium concentration,24-hour sodium quantification,blood creatinine concentration,etc.All patients were divided into a training set(2044 cases)and a validation set(227 cases).In the training set,independent variables were screened by correlation analysis and least absolute shrinkage and selection operator(LASSO)regression analysis,and a model was established by stepwise linear regression analysis and curve fitting.In the validation set,the estimation accuracy of the new model and the other 10 models at the population level was evaluated by correlation coefficient(r),intraclass correlation coefficient(ICC),mean deviation and 95%confidence interval(CI),and Bland-Altman plot.The estimation performance of the model at the individual level was evaluated by relative difference distribution,absolute difference distribution,and classification error rate.Patients were divided into elderly(≥65 years old)and young and middle-aged(18-65 years old)groups,and the accuracy of the model was assessed in different subgroups.Results:The 50%of 2271 patients were male,aged 58(46-67)years,and their24-hour sodium excretion was 2645(1794-3634)mg/d.There were no significant differences in gender,age,height,weight,liver function,kidney function and blood glucose between the training set and the validation set(P>0.05).A new model was established with gender,age,body weight,spot urine sodium concentration,spot urine creatinine concentration,spot urine potassium concentration and blood glucose concentration as independent variables(R~2=0.294,P<0.001).In the validation set,the estimated value of the new model was significantly correlated with the measured value,with the highest correlation coefficient(r=0.523,P<0.05),while the correlation coefficients of the other 10 models ranged from0.386-0.520(P<0.05).The mean deviation between the estimated value and the measured value of the new model was the smallest(-15.0 mg,95%CI:-2618 to 2588mg),and the highest(ICC=0.419)for ICC.Bland-Altman plots show that most subjects in the new model are fairly evenly distributed within the 95%CI range.The correlation coefficient between the estimated value and the measured value of the new model was highest in the young and middle-aged group(r=0.494,P<0.05)and the old group(r=0.548,P<0.05).At the population level,the estimation accuracy of the new model is better than other models.The absolute difference distribution plot showed that the proportion of absolute difference estimated by the new model was the highest within 393 mg(1g salt),786mg(2g salt)and 1179 mg(3g salt),which were 23.8%,49.8%and 71.8%respectively.Other models were 7.0%to 22.0%,15.4%to 44.9%and 28.2%to 67.8%,respectively.The relative difference distribution diagram showed that the relative difference of the new model is highest within 10%,20%,30%and 40%,which were 17.2%,33.9%,48.9%and 61.7%,respectively.Other models were respectively 6.6%to 16.7%,16.3%to 33.5%,24.7%to 45.4%,28.2%to 58.1%.The overall classification error rate of the new model was 46.7%,while that of the other models was 47.6%to 70.5%.It can be seen that the estimation accuracy of the new model was better than other models at the individual level.Conclusion:A new model was developed to estimate 24-hour sodium excretion in the southern Liaoning region of China by sex,age,body weight,spot urine sodium concentration,spot urine creatinine concentration,spot urine potassium concentration,and blood glucose concentration.Compared with the previous models,the new model shows better estimation ability at both the population level and the individual level,and has clinical application value. |
PDF Full Text Request