Simulation Study On Longitudinal Dietary Cadmium Exposure Data Based On Mixed Model

Objective:Chronic dietary exposure assessment need to collect the crowd dietary consumption data for a long time in theory.However,the implemeantation of this survey is difficult How to use statistical method and computer simulation technology based on existing short-term dietary survey data to obtain longitudinal dietary data worth exploring,which is also the hotspot research in this field.This study proposes a new longitudinal data simulation method based on the mixed model,that the short-term data "stretch" for the long-term results,to provide new idea for the calculation of usual intake in chronic dietary exposure assessmentMethod:This study integrates consumption data and contaminant concentration data for the dietary exposure assessment.The consumption data are derived from the"China Health Nutrition Survey" in Jiangsu Province,selecting the 2004,2006,2009,2011 non-continuous four years of three consecutive days of adult consumption survey data.The contaminant concentration data are derived from the data of cadmium in the Jiangsu Province Contaminant Monitoring Network from 2004 to 2009.Construct the bridge database,and calculate the daily intake of cadmium.According to the short-term exposure amount results,the mixed model can be used to calculate the fixed effect and random effects,and calculate the autocorrelation coefficient.According to these model parameters,set the number of simulated people and simulated days,using Monte Carlo simulation random sampling,to obtain long-term individual daily intake,and verify the data structure of the simulation data.Analyze the variability of the exposure data,and the uncertainty of the mean and P95.Calculate the usual intake distribution,and compare it with the OIM,BBN,to analyze the feasibility of this new method.Result:This study has a total of 535 adults in the four years for 3 consecutive days survey,including 247 men,288 women,178 urban adults,357 rural adults,the results are as follows:1.The mean and distribution of daily intake:Adults in Jiangsu province and its subgroups of the daily intake(unit:?g kg-1 bw day-1)of the cadmium are as follows:Total population(535 people)mean value of 0.36,P2.5?P99.9 is(0.04?7.51);Men mean value of 0.36,P2.5?P99.9 is(0.04?7.59);Women mean value of 0.36,P2.5?P99.9 is(0.04?7.51);City adults mean value of 0.34,P2.5?P99.9 is(0.04?5.54);Rural adults mean value of 0.37,P2.5?P99.9 is(0.04?10.60).2.Model parameters related:The gender,urban and rural,years,and the level of interaction is analyzed in mixed model.Finally,the years,urban and rural,and the level of interaction between them is of significance.Variance between individual ?b2=0.054,variance within the individual ?s2= 0.732,one-lag autocorrelation coefficients p=-0.059.3.Simulation method validation:This study set 3 simulated groups,the number of simulated people and simulated days is(300 people,365 days),(500 people,365 days),(1000 people,365 days).Compare the logarithmic conversion of the original data and the simulation data from several aspects,such as mean,standard deviation,variance component ratio and autocorrelation coefficient.Above several indicators,the simulation data are very close to the original data.4.Uncertainty analysis:Repeat 200 Bootstrap samples,the total population(all adults),urban adults,rural adults(mean and P95)and their confidence intervals are calculated for the three groups of simulated data.95%confidence interval of statistics included Monte Carlo simulation results.5.Usual intake distribution:The new method based on the mixed model calculates the mean of the usual intake is 0.34?g kg-1 bw day-1,the standard deviation is 0.09?g kg-1 bw day-1,P2.5?P99.9 is(0.20-0.72)?g kg-1 bw day-1,BBN model calculates the mean of the usual intake is 0.34?g kg-1 bw day-1,the standard deviation is 0.09?g kg-1 bw day-1,P2.5?P99.9 is(0.20?0.73)?g kg-1 bw day-1.P99.9 of the usual intake of population is less than ADI.Conclusion:1.The difference of the daily intake of Cadmium is mainly reflected in the high-end percentile,rural adults is higher than urban adults.2.The simulation method based on the mixed model can use short-term data to obtain long-term longitudinal data,which is consistent with the original data in data structure.3.Using this method to simulate the long-term longitudinal exposure data to calculate the usual intake,the result is similar to BBN model.4.Long exposure assessment based on the samples don't see foods containing cadmium cause harm to the health of adults in Jiangsu province.