Epidemiology Study On The Association Between Atmospheric Particulate Matter Exposure And Metabolic Diseases And Indicator Abnormalities Based On Umbrella Review And Land Use Regression Model

Background and objectives:Atmospheric fine particulate matter(particles with aerodynamic diameters of less than or equal to 2.5μm,PM_2.5)and respirable particulate matter(particles with aerodynamic diameters of less than or equal to 10μm,PM₁₀)pollution is severe globally and is considered a risk factor for several diseases.However,the association between PM_2.5 and PM₁₀ exposure and metabolic diseases and indicator abnormalities,such as diabetes and hypertension,remains unknown.Therefore,this study was designed for the following:1.Umbrella review（UR）was used to grade the strength of evidence for the associations between PM_2.5 and PM₁₀ exposure and metabolic diseases in the systematic reviews and meta-analyses（SRMAs）.Thus provide the first medical evidence research with the highest level of evidence on the subject and guide the following study.2.Land use regression（LUR）models with high accuracy and stability were developed to assess the PM_2.5 and PM₁₀ exposure concentrations of the participants in the Han-Manchu community cohort.The findings will serve as the model foundation and a springboard for further research into the correlation between PM_2.5 and PM₁₀ exposure and metabolic indicator abnormalities of the population.3.Based on the evaluation of the PM_2.5 and PM₁₀exposure of participants in the Han-Manchu community cohort by the LUR model,the association between PM_2.5 and PM₁₀exposure and the four metabolic indicator（blood lipids,blood pressure,blood glucose,and uric acid）abnormalities in this population was investigated.The epidemiological evidence for the association between PM_2.5 and PM₁₀ exposure and the metabolic indicator abnormalities was supplemented.Methods:1.We gathered secondary research evidence（SRMAs）on the association between PM_2.5 and PM₁₀ exposure and metabolic diseases from the Embase,Pub Med,and Web of Science databases and evaluated their quality with the second edition of A Measure Tool to Assess Systematic Reviews（AMSTAR）.UR was used to evaluate the following aspects of SRMAs comprehensively:the summary effect estimates and significance levels,the total number of cases and total sample size of included primary studies（original studies）,the effect estimate of the study with minimum standard error,the 95%prediction interval of the summary effect estimate,heterogeneity among studies,small-study effect,and excess significance bias.According to the aforementioned comprehensive evaluation,UR was used to classify the evidence for the association between PM_2.5 and PM₁₀ exposure and metabolic diseases into four levels,from strong to weak:convincing,highly suggestive,suggestive,and weak.The PROSPERO registration number of UR is CRD42022326088.2.This cross-sectional study（Han-Manchu community cohort）was conducted in Liaoning Province from September 2018 to December 2019,and 10477 participants aged18–79 were recruited.LUR models was constructed to evaluate the exposure concentrations of PM_2.5 and PM₁₀ of the participants.With the effective concentrations of PM_2.5 and PM₁₀from 77 air monitoring stations in Liaoning Province as dependent variables,and geographic information system variables（land use type,population density,road length,number of fire spots,distance from pollution sources,temperature,relative humidity,normalized vegetation index,etc.）as independent variables,LUR models between dependent variables and independent variables were constructed by the forward algorithm and backward algorithm,respectively.The fitting accuracy of the model was characterized by explanatory variance.Furthermore,we tested the performance of LUR models by leave-one-out cross-validation（LOOCV）and then chose the highly accurate and stable LUR models to estimate PM_2.5 and PM₁₀ concentrations at the residential addresses of participants involved in the Han-Manchu community cohort,as a representative of average exposure concentrations of PM_2.5 and PM₁₀.3.The general demographic characteristics of the participants in the Han-Manchu community cohort,such as ethnicity and gender,were obtained through a questionnaire survey.Height,weight,and blood pressure were obtained through physical examination.Biochemical indicators such as blood lipids,blood glucose,and uric acid were obtained through biochemical examination.The Student’s t-test,rank sum test,and Chi-square test were used to compare the difference in baseline characteristics between the abnormal group and the normal group of the four metabolic indicators（blood lipids,blood glucose,blood pressure,and uric acid）.Based on the evaluation of PM_2.5and PM₁₀exposure concentrations by the LUR model,we used logistic regression models to examine the relationship between PM_2.5 and PM₁₀exposure and the four metabolic indicator（blood lipids,blood glucose,blood pressure,and uric acid）abnormalities.Furthermore,gender,age,ethnicity,and other factors of the subjects were used as grouping factors for subgroup analysis,and the interaction was explored.Sensitivity analysis was carried out by two-pollutant logistic regression models.The multiple linear regression model was used to explore the linear relationship between PM_2.5 and PM₁₀ concentrations and four metabolic indicators.Finally,we used a restricted cubic spline model to test the nonlinear relationship between PM_2.5 and PM₁₀ exposure and the four metabolic indicator abnormalities.Results:1.The UR included 41 relationships between atmospheric particulate matter exposure and metabolic diseases.Among the twenty-seven associations between PM_2.5exposure and metabolic diseases,the evidence levels of one,four,and ten significant associations were highly suggestive,suggestive,and weak,respectively,and the remaining twelve associations were nonsignificant.Among the fourteen associations between PM₁₀exposure and metabolic diseases,the evidence levels of one and three significant associations were suggestive and weak,respectively,and the other ten associations were nonsignificant.There were ten associations between PM_2.5 exposure and metabolic diseases with the small-study effect,and eight associations had excess significance bias.There were three associations between PM₁₀ exposure and metabolic diseases with the small-study effect,and no association with excess significance bias was found.The relationships between PM_2.5 and PM₁₀ exposure and metabolic syndrome and obesity/overweight were nonsignificant.None of the evidence of the significant relationship between PM_2.5 and PM₁₀ exposure and diabetes,hypertension,and gestational diabetes mellitus was convincing or highly suggestive.Among them,the evidence of the significant relationship between PM_2.5 and PM₁₀ exposure and diabetes and hypertension was suggestive or weak.Moreover,there was no SRMA on the relationship between PM_2.5and PM₁₀ exposure and abnormal metabolic indicators such as dyslipidemia and high uric acid.2.The modeling research of atmospheric particulate matter exposure in Liaoning Province based on land use regression showed that the interpretation variances of the LUR model of PM₁₀ obtained by backward algorithm,forward algorithm without meteorological factors,and forward algorithm with meteorological factors were 0.63,0.49,and 0.68,respectively.The LUR model interpretation variances of PM_2.5 obtained by the three methods were 0.32,0.30,and 0.38,respectively.The PM_2.5 optimal LUR model included five prediction factors,including land use type,the number of fire points,dew point temperature,etc.The correlation between PM_2.5 concentration and dew point temperature was the highest,followed by land use type.The PM₁₀optimal LUR model included six prediction factors,including land use type,the number of fire spots,atmospheric pressure,wind speed,etc.The correlation between PM₁₀ concentration and wind speed was the highest,followed by atmospheric pressure.PM_2.5 and PM₁₀ concentrations were sensitive to land use type,the number of fire spots,and meteorological factors.The explanatory variances of the optimal LUR models of PM_2.5 and PM₁₀were 0.68 and 0.73,respectively.The adjusted explanatory variances were 0.65 and 0.70,respectively.The coefficients of determination of the LOOCV were 0.62 and 0.66,respectively.The exposure concentration of PM_2.5 and PM₁₀ of 9962 participants in the Han-Manchu community cohort could be assessed by the LUR model,and the average exposure concentrations of PM₁₀ and PM_2.5 were 80.90μg/m³ and 48.56μg/m³,respectively.3.In the Han-Manchu community cohort,the number of people with dyslipidemia,hypertension,hyperglycemia,and high uric acid accounted for 66.13%,36.48%,21.24%,and 11.50%of the total participants,respectively.After adjusting for confounding factors,each 10μg/m³ increase in PM_2.5 was positively associated with the risk of dyslipidemia and hypertension,and the corresponding（odds ratio,OR）and 95%confidence interval（CI）were 1.14（1.06–1.22）and 1.16（1.08–1.25）,respectively.Each 10μg/m³ increase in PM₁₀was positively correlated with the risk of dyslipidemia,hyperglycemia,and high uric acid,and the corresponding OR（95%CI）were 1.08（1.03–1.14）,1.09（1.03–1.16）,and 1.09（1.00–1.17）,respectively.The sensitivity analysis showed that the risk of dyslipidemia,hyperglycemia,and high uric acid might increase with the increase of 10μg/m³ of PM_2.5and PM₁₀exposure concentration.The multiple linear regression model showed that every1μg/m³ increase in PM_2.5 and PM₁₀was positively correlated with the concentration of total cholesterol,low-density lipoprotein cholesterol,and high-density lipoprotein cholesterol.Every 1μg/m³ increase in PM_2.5 was positively correlated with the systolic pressure.Every 1μg/m³ increase in PM₁₀ was positively correlated with the fasting blood glucose and blood uric acid concentration.The restricted cubic spline model showed that there was a nonlinear relationship between the increase in PM_2.5 and PM₁₀exposure concentration and the risk of dyslipidemia,hypertension,hyperglycemia,and high uric acid（P≤0.05）.Conclusions:1.UR demonstrates that the evidence credibility of the association between PM_2.5 and PM₁₀ exposure and metabolic diseases still needs to be improved,and more powerful original research is urgently needed as the most fundamental source of evidence credibility improvement.At the same time,the evidence that PM_2.5 and PM₁₀ are related to dyslipidemia,high uric acid,and other metabolic indicators needs to be supplemented urgently.2.The exposure concentration of PM_2.5 and PM₁₀ of the Han-Manchu community cohort is relatively high,so it is important to study the health effects of PM_2.5 and PM₁₀exposure for disease prevention and control.In addition,the LUR model used to assess the PM_2.5 and PM₁₀ exposure concentrations of this population is highly accurate and stable,which can be used to assess the PM_2.5 and PM₁₀ exposure concentrations of other populations in Liaoning Province.3.The Han-Manchu community cohort shows that the increased PM_2.5 exposure may be related to the increased risk of dyslipidemia and hypertension.PM₁₀ exposure may be related to the increased risk of dyslipidemia,hyperglycemia,and high uric acid.Reducing PM_2.5 and PM₁₀ pollution may be helpful for the early prevention of metabolic indicator abnormalities.