The Effect Of Interaction Between Social Ecological Risk Factors And Polygenic Risk Score On Clustering Of Health Risk Behaviors In Adolescents

ObjectivesTo provide an effective evaluation tool for the study of social ecological risk factors of Chinese adolescents,the Youth Social Ecological Risk Factor Assessment Questionnaire was compiled and the psychostatistical evaluation was carried out.To explore the relationship between social ecological risk factors and adolescent clustering of health risk behavior（HRB）,and to clarify the impact of interaction between social ecological risk factors and genes on adolescent clustering of HRB,so as to provide theoretical basis for improving adolescent psychological behavior problems.MethodsIn the study 1,based on literature review,group discussion and expert interview,and based on the conceptual framework of social ecological model and ecosystem theory,the initial Adolescent Social Ecological Risk Factors Assessment Questionnaire（hereinafter referred to as questionnaire）with 55 items in 7 dimensions was constructed.Sample 1:From October to December 2019,1 386 middle and high school students from 2 middle schools in Hefei city,Anhui Province were enrolled in a pre-experimental survey,which was adjusted according to the questions answered by the students.Sample2:From October 2020 to December 2020,a multi-stage convenient cluster sampling method was adopted to select 5 500 middle and high school students from 13 middle schools in Zhengzhou city,Henan Province for questionnaire survey.After discarding invalid questionnaires,5 188 questionnaires were screened by independent sample t test,correlation analysis,factor analysis and other methods to form a questionnaire for assessing adolescent social ecological risk factors（SERF）consisting of 50 items.In exploratory factor analysis,the results of Kaiser-Meyer-Olkin（KMO）and Bartlett sphericity test were used to determine.The reliability and validity of the questionnaire were evaluated by homogeneity reliability,split-half reliability,structural validity and discriminative validity.In the discriminative validity,sample 2 was used to compare the scores of social ecological risk factors from different groups such as gender,family economic status,parental education level and psychological sub-health.Sample 3:Confirmatory factor analysis was conducted in 2 610 samples from Shenyang,Liaoning province in December 2020.Normed fit index（NFI）,relative fit index（RFI）,comparative fit index（CFI）,goodness of fit index（GFI）,adjusted goodness of fit index（AGFI）and root mean square error of approximation（RMSEA）were used to evaluate the fit degree of the model.In study 2,from March 2021 to May 2021,a multi-stage cluster sampling method was adopted,and 3 cities including Beijing,Yangjiang city of Guangdong Province and Zhengzhou City of Henan Province were selected as the survey cities.A total of 39schools were selected from 6 ordinary middle schools and 7 ordinary high schools in each city.A total of 17 800 questionnaires were sent out in 3 cities,947 invalid questionnaires were excluded,and 16 853 valid questionnaires were collected.The questionnaire on adolescent SERF assessment and the questionnaire on HRB were used to evaluate the level of adolescent HRB and the status of social ecological risk factors.One-way ANOVA and independent sample t test were used to analyze the differences of SERF scores among different populations.Latent category analysis（LCA）was used to explore the clustering mode of HRB.The clustering of HRB was determined according to the model fitting indexes,including Akaike’s information criteria（AIC）,Bayesian information criteria（BIC）and sample size adjusted BIC（ssa BIC）,Entropy and Lo-Menell-Rubin Test（LMRT）.Spearman correlation analysis,Chi square analysis,Logistic regression and Bayesian statistics were used to analyze the association between SERF and the clustering of HRB.Study 3,using the mapping knowledge domain map to identify risk factors for HRB,using VOSviewer software for risk factors for adolescent HRB related literature text recognition and clustering analysis,the results data export and combed the related risk factors,risk factors for adolescent HRB for identification.A questionnaire survey was conducted in 4 schools（2 middle schools and 2 high schools）in Xuzhou,Jiangsu Province.Oral and pharyngeal swab samples were used to extract DNA for low depth whole genome sequencing（3.48×～10×）.A total of 264 middle and high school students were collected.The LCA was used to explore the clustering patterns of HRB,and Spearman correlation analysis,Chi-square analysis and Logistic regression model were used to analyze the associations between social ecological risk factors and polygenic risk score（PRS）and the clustering of HRB.The Process model explores associations between SERF and psychopathological symptoms,as well as biological clock PRS and the clustering of HRB.ResultsIn sample 1 of study 1,1 500 questionnaires were issued,1 450 were recalled,invalid questionnaires were excluded,and the final valid questionnaires were 1 386.Questions were adjusted according to students’degree of completion and understanding.A total of2 631 male subjects（50.7%）and 2 557 female subjects（49.3%）were included in the second sample.The average age of the subjects was（15.33±1.82）years.Through project analysis,5 items were deleted and 50 items from 7 dimensions,including individual,family,school,community,policy,culture and chronosystem,were formed into the Youth Social Ecological Risk Factor Assessment Questionnaire（hereinafter referred to as the Questionnaire）.The cumulative contribution rate of variance of the Questionnaire was 52.35%.In the independent sample t test,the difference between the high and low groups in each item was statistically significant.The Pearson correlation coefficients between each item and the total score ranged from 0.298 to 0.724,and between each item and its dimension ranged from 0.478 to 0.865,with statistically significant differences,indicating that each item had good discrimination.In the questionnaire reliability analysis,the Cronbachαcoefficient of the total questionnaire was 0.897,and the Cronbachαcoefficient of each dimension was above 0.5,ranging from 0.525 to 0.848.The partial coefficient of the total questionnaire was 0.716,and the partial coefficient of each dimension ranged from 0.801 to 0.907.Confirmatory factor analysis of sample 3 showed thatχ~2/df value was 19.319（P<0.01）,NFI was 0.838,RFI was 0.826,CFI was 0.841,TLI was.852,AGFI was 0.851,RMSEA was 0.058,indicating good goodness of fit.In sample 2,the discriminative validity of different gender,family economic status,parental education level and mental subhealth groups was good,and the differences were statistically significant.In study 2,16 853 subjects were analyzed by questionnaire of study 1.The results showed that there were 8 390 boys（50.2%）and 8 463 girls（49.8%）,and the average age of adolescents was（15.21±1.74）years old.The total score of social ecological risk factors of the subjects was（106.06±22.93）.The total scores of social ecological risk factors and individual and cultural dimensions of female students were higher than those of male students,and the differences were statistically significant.The scores of family,community,policy and chronosystem of male students were higher than those of female students,and the differences were statistically significant except for family and culture.The total score of social ecological risk factors and the scores of all dimensions of adolescents with lower educational level of parents,poor family economy,poor academic performance and fewer partners were higher than those of other groups.High school students’individual,family,school,community,policy,culture and total scores were higher than junior high school students,while junior high school students’chronosystem was slightly higher than senior high school students,the differences were statistically significant.The younger the age,the higher the score of social ecological risk factors,the difference was statistically significant（P<0.01）.According to the model fitting index,LCA showed that the three categories were the optimal ones,so the clusters of HRB were named as low,medium and high levels.The Chi-square test suggests that,high-risk individuals（χ~2=1258.24）,families（χ~2=1452.72）,schools（χ~2=909.37）,communities（χ~2=282.41）,policies（χ~2=197.75）,culture（χ~2=1020.33）and total scores（χ~2=1640.77）were associated with higher clustering of HRB（P values all<0.01）.Logistic results showed that SERF were associated with adolescent HRB clustering and co-occurrence,among which,high-risk individual factors（OR=5.34,95%CI=4.81,5.92）,family factors（OR=7.02,95%CI=6.27,7.85）,school factor（OR=4.32,95%CI=3.89,4.79）,community factor（OR=2.23,95%CI=2.01,2.46）,policy factor（OR=1.94,95%CI=1.76,2.14）,cultural factors（OR=4.84,95%CI=4.36,5.38）and chronosystem（OR=2.65,95%CI=2.39,2.93）were positively correlated with high level of clustering of health risk behavior.OR value was 8.80（95%CI=7.81,9.90）between high score of total SERF and high level of co-occurrence HRB（P values all<0.01）.In study 3,according to the clustering results of HRB risk factor research,four prominent areas are displayed by clustering information fields in network viewable and density viewable.Including"student","behavioral risk factor surveil","drug use"and"cardiovascular disease".The aggregation results also found information fields related to genes（"genes"）.Overlay visualization project the same as the network visualization,the difference is that the color of the corresponding project,the project focus year by far and near is according to the color of the corresponding project scope,from violet to green to red,by the diagram can be found that the direction of the researchers from sexual behavior and infectious diseases to suicidal self-injury behavior,sedentary behavior and psychological problems such as noncommunicable diseases.The total score of high risk socio-ecological risk factors and high level adolescents HRB clustering（OR=4.78,95%CI=1.94,11.76）,co-occurrence（OR=2.58,95%CI=1.40,4.74）and psychopathological symptoms（depression OR=5.41,95%CI=2.77,10.55;anxiety OR=6.07,95%CI=2.90,12.70;mental sub-health OR=4.88,95%CI=2.17,10.98）was positively correlated,and the difference was statistically significant.After controlling covariates,the results remained consistent.There were three-way interactions between SERF,age and psychopathological symptoms,and the differences were statistically significant.Genome-wide association study（GWAS）of the clustering of HRB revealed a total of 372 significant SNPs loci.Through GWAS and literature review,biological clock PRS was constructed to identify genetic susceptibility to the clustering of HRB.The interaction of high risk social ecological risk factors and high genetic risk biological clock PRS was positively correlated with high level of the clustering of HRB in adolescent（OR=8.33,95%CI=1.56,44.64）and co-occurrence HRB（OR=3.01,95%CI=1.03,8.76）,and the difference was statistically significant.The interaction between the sub-dimensions of social ecological risk factors and the high genetic risk biological clock PRS is essentially correlated with the outcome of the clustering of HRB and co-occurrence.There were three way interactions between social ecological risk factors,PRS and psychopathological symptoms,and the differences were statistically significant.ConclusionsThe Social ecological Risk Factor Assessment Questionnaire for Adolescents meets the requirements of psychostatistics with good reliability and validity,and can be used as a tool to assess the level of adolescents’social psychological environment.High risk social ecological risk factors were positively associated with high levels of clustering of HRB.PRS plays an important role in the relationship between SERF and adolescent HRB,and there is also a relationship between gene and environmental interaction and adolescent clustering of HRB.