Research design and sample
This study was a cross-sectional survey, conducted from April to July, 2018. The participants were recruited from Hunan Province, China by a multi-stage cluster sampling. First, the prefecture-level cities in Hunan Province are stratified according to their geographical location (eastern Hunan, southern Hunan, western Hunan and northern Hunan), and one prefecture-level city (Zhuzhou City, Yongzhou City, Shaoyang City and Changsha City) is randomly selected in each area. Then 4 senior high schools (2 in urban areas and 2 in rural areas) are randomly selected in each prefecture-level city as sample schools, and a total of 16 high schools are selected for investigation. Finally, according to the proportion of students in the school, five classes were randomly selected from the selected schools, and all the students in the selected classes were selected as the subjects of the survey. Due to the heavy academic tasks of the students in the third grade of senior high school, the acceptance of this survey is low, so only the students in the first and second grades of senior high school are selected as the objects of this study. Before the field investigation, we requested permissions from the principals of each school. Once the permissions were granted, investigators conducted the research in each class with the help of the headteachers. A total of 4050 questionnaires were distributed and 3934 valid questionnaires were collected, with an effective response rate of 97.1%.
All participants signed informed consent forms, and the purpose of the study as well as the questionnaire sections were explained to them by investigators. The students were assured of the anonymity and confidentiality of the information provided in the self-reported questionnaires, and the respondents were free to discontinue their participation at any time of the study. The study received the approval from the ethical committee of Xiangya School of Public Health, Central South University. (XYGW-2017-056).
Data collection and measurements
General characteristic
Potential confounding factors were selected based on the literature with regard to the association between bullying victimization, emotional problems and sexual orientation[19-20], which included gender, grade, the type of school (public high schools, or private high schools), school location(rural or urban), family type (single-parent family, Remarriage / foster family, or Complete family), parental education (lower than a bachelor's degree ,or a bachelor's degree and above) and whether the respondent comes from one-child family (yes or no). Besides, parental abuse were considered as confounding variables in this study according to prior work [21]. Parental abuse were measured by asking participants “how often your parents hit or abused you”. The frequency coded on a 5-point scale in the last half year, ranging from 1 to 5 (1= never happened, 2 = only once or twice, 3 = two or three times a month, 4 = about once a week, 5 = several times a week). Those who chose options 2, 3, 4 and 5 were defined as being abused by their parents, and those who chose option 1 were defined as not abused by their parents.
Bullying victimization
Traditional bullying victimization was surveyed via the suffering subscale of Chinese version of Olweus Bully/Victim Questionnaire (OBVQ). The reliability of the subscale is good, and the Cronbach's α coefficient is 0.80[22]. Second, bulling victimization was assessed by asking that have you ever been bullied by peers in the last 6 months using the following 6 items: (1) Having been hit, kicked, pushed or shoved; (2) Having belongings been taken or damaged; (3) Having been called nasty name; (4) Having been made fun of; (5) Having been kept out of things on purpose, excluded from the group or completely ignored; (6) They told lies or spread rumors about you and/or tried to make others dislike you. The frequency was coded on a 5-point scale ranging from 1 to 5 (1= never happened, 2 = only once or twice, 3 = two or three times a month, 4 = about once a week, 5 = several times a week). Students were considered to be involved in traditionally bullying victimization if the frequency of bullying behavior mentioned above happened more than two or three times a month [23].
Cyber school bullying was surveyed via the question, "In the past six months, have any students ridiculed, abused, threatened you or maliciously spread your privacy on the Internet through electronic products such as text messages, e-mails and chat platforms?. " Students who answered “ yes ” were classified as victims of cyber bullying.
For the analysis, the participants were placed into categories based on their responses to traditional and cyber bullying victimization: 1) no form of bullying, 2) traditional bullying only, 3) cyber bullying only 4) a combination of traditional and cyber bullying.
Depression symptoms
The Chinese version of the 9-item Patient Health Questionnaire (PHQ-9) was used to screen depression of participants [24]. The PHQ-9 has 9 questions with a score ranging from 0 to 3 for each item, and total score ranging from 0 to 27. A higher score indicates that the participant has more depressive symptoms. A total score of 10 or more is considered to meet the criterion of screening for depressive symptoms (1 = Yes, 0 = No)[25, 26]. The prior study showed good reliability of the Chinese version of PHQ-9 in children and adolescents population[27]. In the present study, the Cronbach’s alpha of PHQ-9 was 0.88.
Anxiety symptoms
The Chinese version of the 7-item Generalized Anxiety Disorder Scale (GAD-7) was used to assess anxiety disorders of children[28]. The GAD-7 has 7 questions with a score ranging from 0 to 3 for each item. Therefore, total score of GAD-7 ranges from 0 to 21[29]. A total score ≥ 10 of the GAD-7 is considered to meet the criterion of screening for anxiety disorder (1 = Yes, 0 = No)[29, 30]. The Chinese version of GAD-7 had good reliability among Chinese population according to early findings[31]. In this study, the alpha Cronbach for GAD-7 was 0.93.
Sexual orientation
First, participants were provided with a definition of sexual orientation[32]. Then, sexual orientation was measured using a single item on a 4-point scale. Participants were asked to circle a number which best describes their sexual orientation, the responses included: 0 = Heterosexual, 1 = Lesbian/gay, 2 = Bisexual, and 3 =Questioning[33]. Sexual minority included adolescents who self-identified as lesbian/gay, bisexual, or questioning about their sexual orientation (LGBQ).
Statistical analysis
EpiData 3.1 and SPSS 19.0 software were used for data entry and statistical analysis. Prevalence of participants involved in different types of bullying victimization , depression and anxiety symptoms were summarized by descriptive statistics [n (%)]. The χ2 test and multinomial logistic regressions were conducted to examine the associations between sexual orientation and traditional, cyber and combined bullying. The χ2 test and multivariate binary logistic regression were conducted to examine the associations between sexual orientation and depression and anxiety symptoms. The crude odds ratio (COR) , adjusted odds ratio (AOR)and 95% confidence intervals (95% CI) were estimated. The gender and grade of the respondent, the type of school, whether the school was rural or urban, parental education , whether the respondent comes from one-child family, whether the respondent was beaten by a parent or guardian were included as covariates in the logistic regression. P values of less than 0.05 were considered statistically significant.