Sample and data collection
We conducted a large population-based, cross-sectional study among 18 colleges/universities in Wuhan, Central China, from October 2017 to February 2018. A multi-stage cluster random sampling method was applied in this survey. Firstly, according to subject settings, we categorized the 18 colleges/universities into seven groups: five comprehensive universities, seven universities of science and technology, two universities of finance and economics, and one university of teacher-training, agronomy, nationalities as well as sports, respectively. Secondly, we randomly selected, in the proprotion of students sizes, several classes from each grade (from undergraduate to doctoral degree) in every college/university. Then, all students in selected classes were encouraged to participate in this survey with the voluntary principle and ensured no less than 500 questionnaires were received from each college/university. All participated students were asked to fulfill an online questionnaire on their computers or cellphones. The questionnaire was used to collect students’ information including socio-demographic characteristics, individual lifestyles or behaviors, perceived social support, physical and mental health status. Ultimately, a total of 11750 college students participated in the survey and 11093 questionnaires were collected on a computer terminal, with a response rate of 94.41%. After excluding those completed in less than five minutes, 10676 qualified questionnaires were included in final statistical analyses, yielding a 96.24% qualification rate.
This study was approved by the ethics committee of Tongji Medical College institutional review board, Huazhong University of Science and Technology, Wuhan, China. All participants signed informed consent before filling out the questionnaire.
Multidimensional Scale of Perceived Social Support (MSPSS)  consists of 12 items with response options scoring from 1 (very strongly disagree) to 7 (very strongly agree). It estimates SSQ from three sources: family (item 3, 4, 8, and 11), friends (item 6, 7, 9, and 12) and significant others (item 1, 2, 5, and 10) . Scores of all items are added up and then divided by 12. The mean scores ranging from 1 to 2.99, 3 to 5 and 5.01 to 7 are classified as low, medium, and high perceived support levels, respectively . MSPSS has a sound factorial validity (with Cronbach’s alpha coefficients of 0.953), and internal consistencies for the full scale and subscales are both satisfactory . The Chinese version has been suggested as a reliable tool for assessing SSQ .
Mental health status
The 12-items General Health Questionnaire (GHQ-12)  is one of the most widely used versions measuring individual mental health problems. It includes 12 items corresponding to three dimensions: anxiety/depression (item 1, 2, 7, and 10), social dysfunction (item 3, 4, 5, 6, 8, and 9) and deficiency of confidence (item 11 and 12) . There are four answers ranging from “better/healthier than normal” to “much worse/more than usual”. The GHQ scoring method (the four options were scored by 0-0-1-1, respectively) has been adopted in our study . Higher score corresponds to worse mental health status. A total score of 4 or more was classified as having notable mental problem . GHQ-12 had satisfactory reliability (with Cronbach’s alpha coefficients of 0.886) and extensive sensitivity, which had been previously validated.
Socio-demographic characteristics and lifestyles
The questionnaire includes the following demographic variables: age, gender, ethnicity, religious belief, place of residence, single-parent/-child family or not, paternal/maternal education level, and family economic status. Family economic status was assessed by asking the question of “what do you think of your family economic condition?” with optional responses of “very affluent”, “more affluent”, “the general”, “less affluent”, or “non-affluent”. Based on the responses, family economic status was categorized as good, general, and poor.
Lifestyle variables refer to physical exercise, regular work-rest or not, sleep duration, smoking and alcohol drinking status. Physical exercise was judged from the question of “do you have chronic aerobic exercise (e.g. setting-up exercise, jogging, walking) for 30 minutes and longer three times a week?”, and the responses were classified into “never/seldom”, “somtimes”, and “usually/always”. Regular work-rest was estimated by the question of “do you have a regular daily routine?”, and the options were also classified into three categories: “never/seldom”, “somtimes”, and “usually/always”. Sleep duration was divided into “<7 hours”, “7-8.9 hours”, and “≥9 hours” based participants’ answers to “In recent three monthes, you sleep for XX hours, XX minutes every day on average.” Smoking and alcohol drinking were dichotomized as “yes” and “no” according to participants’ responses.
Data analyses were performed using the SPSS software (Version 22 for Windows, SPSS Inc, Chicago, IL, U.S.A.). Descriptive analyses included means (standard deviations [SDs]) for continuous variables and frequencies and percentages for categorical data. We analyzed respondents' demographic characteristics, and compared the differences of SSQ and mental health statuses among various demographics by χ2 tests. Potential influencing factors of psychological problem were identified via multivariate logistic regression analyses. Furthermore, we described the correlation between MSPSS and GHQ-12 by matrix analysis and estimated relationships between SSQ and mental problems under different adjustments using trend analysis. Potential confounders included age, gender, ethnicity, religious belief, residence area, single-parent/-child family, parental education level, family economic status, work-rest routine, sleep duration, alcohol drinking, and smoking. Significance level was accepted as P<0.05 (two-tailed) for all tests.