Moderating Effects Of Eating Behavior In The Association Of Urinary Phthalate Metabolite Concentrations With Emotional Symptoms In Chinese University Students

Objectives This study aimed to describe the urinary phthalate(PAEs)metabolite levels and latent growth trajectories in Chinese university students,assess the cumulative exposure risks and sources of exposure to PAEs,analyze the longitudinal association between PAEs metabolite concentration and emotional symptoms,and explore the moderating effects of eating behavior in the association.Methods This school-based survey was carried out among university students from Anhui Medical University in Anhui Province and Shangrao Normal University in Jiangxi Province of China from December 2018 to May 2019.We used purposive sampling to select the participants.1,239 participants completed questionnaires and provided 1,017 urine samples(urina sanguinis).A follow-up cohort was established on the basis of the baseline survey,and 906 freshmen were included in the longitudinal study.893 participants completed questionnaires(13 participants were lost to follow-up,and lost rate was 1.4%)and provided 555 urine samples in follow-up survey from June 2019 to November 2019.Additionally,90 university students were recruited in Anhui Medical University for a 7-days continuous survey in June 2019.630 questionnaires were completed and 630 urine samples were collected in 7-days survey.For the survey,participants used smart phone We Chat to scan the specified QR code to complete the electronic anonymous questionnaire.The survey included general demographic variables,lifestyle behaviors(eg,eating behavior,personal care product use,physical activity,mobile phone use,etc.),stressful life events,family history of depression,and emotional symptom assessment.Eating behavior was assessed by food frequency questionnaire(FFQ).The Chinese version of 21-item Depression Anxiety Stress Scales(DASS-21)was used to assess university students' emotional symptoms included depressive symptoms,anxiety symptoms and stress symptoms.Moreover,physical examination was performed in participants in hospital.Six PAEs metabolites were analyzed by high-performance liquid chromatography-tandem mass spectrometry and urine specific gravity(SG)analyzed with a handheld refractometer.The researchers strictly carried out quality control during the investigation and laboratory testing.This study was approved by the Ethics Committee of Anhui Biomedical University(batch number: 20170291).All statistical analyses were performed using SPSS 23.0,Mplus 7.4 and R 3.5.1.Analytical methods include descriptive statistical analysis,chi-squared test,spearman correlation analysis,rank sum test,repeated measures analysis of variance,generalized linear model(GLM),latent growth curve model(LGCM),latent profile analysis(LPA),autoregressive cross-lagged model(ARCLM),generalized estimation equation(GEE),Bayesian mediation effect model and latent moderation model.Results The results of the study showed that the detection rates of six PAEs metabolites include monobutyl phthalate(MBP),mono-2-ethylhexyl phthalate(MEHP),monoethyl phthalate(MEP),mono(2-ethyl-5-hydroxyhexyl)phthalate(MEHHP),monomethyl phthalate(MMP)and mono(2-ethyl-5-oxohexyl)phthalate(MEOHP)in urine ranged from 79.6% to 99.7%.The median concentrations of above PAEs metabolites ranged from 2.90 to 155.04 ng/m L.The concentrations of PAEs metabolites corrected for SG were highly correlated with those before correction(r = 0.82-0.96,all P < 0.01).Repeated measures analysis of variance showed that there were statistically significant differences in concentrations of MEP,MMP,MEHHP,MEOHP,low molecular weight phthalate(LMWP)and high molecular weight phthalate(HMWP)of 7-days measurement(P < 0.05).Also,result of LGCM suggested that latent growth trajectories of different PAEs observed in the study of 7-days repeated measurement.For instance,there were individual differences in the initial concentrations of MBP,MEP,MMP,MEHP and LMWP for all participants(Intercept factor(I)= 0.04-0.42,all P < 0.01).Not only were there individual differences in initial concentrations and growth rates of MEHHP,MEOHP and HMWP for all participants(I = 0.12-0.14,slope factor(S)< 0.01,all P < 0.01),but I and S were negatively correlated,the higher the initial level,the lower the growth rate.The median of estimated daily intake(EDI)of di 2-ethyl hexyl phthalate(DEHP),diethyl phthalate(DEP),dibutyl phthalate(DBP)and dimethyl phthalate(DMP)in participants were 3.29,0.82,6.31 and 0.46 ?g/kg·bw/day,respectively.According to the daily tolerable intake(TDI)recommended by the European Food Safety Authority(EFSA),32.2% of participants had a PAEs exposure hazard index(HI)?1.Meanwhile,according to the reference dose(Rf D)recommended by the US Environmental Protection Agency(EPA),3.5% of participants had a PAEs exposure HI ?1.Result of LGCM showed that the latent growth trajectories of different EDI and HI in 7-days repeated measurement.There were individual differences in initial concentrations and growth rates of EDI of DBP,DEP,DMP and DEHP for all participants(I = 0.29-333.95,all P < 0.01;S =-0.09-10.36,all P < 0.05).Except for DBP,I and S were negatively correlated,the higher the initial level of EDI,the lower the growth rate.Besides,there were individual differences in the initial concentrations of Rf D-HI and TDI-HI(I = 0.90-1.43,P < 0.01),and growth rates of Rf D-HI(S = 0.03,P < 0.01).I and S of Rf D-HI and TDI-HI were negatively correlated,the higher the initial level of HI,the lower the growth rate.GLM analysis results of baseline data suggested that fast food consumption,milk consumption and personal care product use were associated with MBP(OR = 1.48,95%CI:1.07-2.04;OR = 1.66,95%CI:1.08-2.57;OR = 1.40,95%CI:1.03-1.90;P <0.05)and LMWP(OR = 1.46,95%CI:1.06-2.02;OR = 1.34,95%CI:1.01-1.76;OR = 1.54,95%CI:1.02-2.32;P < 0.05).Likewise,fast food consumption and personal care product use was associated with MEP(OR = 1.44,95%CI:1.05-1.98;OR = 1.72,95%CI:1.15-2.58;P < 0.05).The positive rates of depressive symptoms,anxiety symptoms and stress symptoms were 17.4%,24.8% and 9.5%,respectively.These observations in LGCM imply that potential growth trajectories of different emotional symptoms score observed in the 7-days survey.There were individual differences in initial score of the emotional symptom(I = 15.87-32.10,all P < 0.01),I and S were negatively correlated,the higher the initial score,the slower the growth rate.In the longitudinal study of association between PAEs exposure and emotional symptoms in university students,GLM analysis results found that MBP was associated with depressive symptoms,anxiety symptoms and stress symptoms in girls(? = 2.24,95%CI:0.38-4.10;? = 2.08,95%CI:0.14-4.01;? = 2.89,95%CI:0.59-5.19;all P < 0.05),respectively;MEHP was associated with depressive symptoms,anxiety symptoms and stress symptoms in girls(? = 0.82,95%CI:0.23-1.41;? = 0.84,95%CI:0.23-1.45;? = 0.85,95%CI:0.13-1.56;all P < 0.05),respectively.However,no association of PAEs metabolites with emotional symptoms was observed in boys.Nevertheless,GEE analysis results suggested that the inverse association analysis of PAEs and emotional symptoms was not statistically significant.In the longitudinal study of association between eating behavior and emotional symptoms in university students,GEE analysis results showed that fast food consumption was associated with three emotional symptoms(depressive symptoms,anxiety symptoms and stress symptoms),respectively(? = 0.11,95%CI: 0.08-0.14;? = 0.10,95%CI: 0.07-0.13;? = 0.09,95%CI: 0.05-0.12;all P < 0.01);packaged food consumption was associated with three emotional symptoms,respectively(? = 0.08,95%CI: 0.04-0.11;? = 0.08,95%CI: 0.05-0.12;? = 0.07,95%CI: 0.04-0.11;all P < 0.01);street food consumption was associated with three emotional symptoms,respectively(? = 0.13,95%CI: 0.09-0.16;? = 0.12,95%CI: 0.09-0.16;? = 0.12,95%CI: 0.08-0.16,all P < 0.01).It is notable that reverse associations were observed in GEE of longitudinal study.Depressive symptoms and anxiety symptoms were associated with fast food consumption,respectively(? = 0.22,95%CI: 0.08-0.35;? = 0.17,95%CI: 0.02-0.32;all P < 0.05).Depressive symptoms and anxiety symptoms were associated with street food consumption,respectively(? = 0.13,95%CI: 0.00-0.25;? = 0.16,95%CI: 0.02-0.30;all P < 0.05).The moderating effects of eating behavior in the longitudinal association between PAEs exposure and emotional symptoms in university students were analyzed in latent moderation model.The findings indicated that packaged food consumption moderated the association between LMWP and depressive,anxiety and stress symptoms(? = 0.23-0.89,all P < 0.05).Fast food consumption moderated the association between LMWP and depressive symptoms(? = 0.32,P < 0.05).Street food consumption moderated the association between LMWP and depressive symptoms(? = 0.31,P < 0.05).Unexpectedly,gender stratification analysis did not observe the moderating effects in the association between PAEs and emotional symptoms.Conclusions 1.The results suggest that there is a greater health risk in Chinese university students due to their generally exposed to DEHP,DEP,DBP and DMP.The higher the initial value of the latent growth trajectories of PAEs metabolites,EDI and HI,the lower the growth rate.Eating behavior(such as takeaway fast food consumption)and the use of personal care products are the important source of PAEs exposure for university students.2.What is important is that a higher positive rate of emotional symptoms was found in Chinese university students.This study demonstrates that there is a one-way,non-linear association between PAEs exposure and emotional symptoms in university students,and no longitudinal associations are observed in boys.No reverse association is observed in the longitudinal study.3.The findings indicate that there is a two-way association between eating behavior and emotional symptoms in university students.It is important to note that eating behavior moderate the association between PAEs exposure and emotional symptoms,high-frequency eating behavior has a positive moderating effect,and low-frequency eating behavior has a negative moderating effect.