Between January 2018 and January 2019, a total of 1700 young adults, aged 18–35 years, were enrolled in the online national survey using a convenience sampling method. The popular Chinese online survey platform Wenjuanxing (http://www.wjx.cn) was used. The survey recruitment information was posted to six administrative regions in China, including the northeast, east, north, south-central, southwest, and northwest areas. First, the link for the online survey was sent to the research partners or friends who reside in six administrative regions by email. Then, they posted this link in different forums (e.g., QQ, WeChat). To excluded participants who have a diagnosis of mental illness, such as MDD, bipolar disorder, psychotic disorders, and others, the participants would be asked whether or not receiving a diagnosis of mental illness from a psychiatrist. The sample size was determined by the subject to item ratio of 5–10:1 , and the total number of survey items was 166.
The modified Chinese version of the Leiden Index of Depression Sensitivity (LEIDS-RR-CV) is a 26-item self-report measure of CR to sad mood . Participants are asked to imagine the last time they felt a mild state of dysphoria, and then to indicate the degree to which a list of statements describes their typical cognitions and behaviors in response to a sad mood. The LEIDS-RR-CV contains 26 items from 5 subscales, including hopelessness/suicidality, acceptance coping, aggression, control/perfectionism, and avoidant coping. All of the items are rated using a 5-point Likert-scale (0=not at all to 4=very strongly). Items are all positively worded for CR, and the total score is obtained by summing the scores of all items. Considering the differences in the numbers of items among the five subscales, the average score for each subscale was calculated. A higher total score indicates stronger CR. In this study, the Cronbach’s α coefficient was 0.95 for the overall scale. Huang et al. identified a cut-off score of 60 for LEIDS-RR-CV to screen for healthy individuals at risk for depression in China .
2.2.2. Social support
The Multidimensional Scale of Perceived Social Support (MSPSS) is a 12-item self-report scale used to measure perceived social support from family, friends, and significant others [23,24]. The scale employs a 7-point rating scale ranging from 1 (very strongly disagree) to 7 (very strongly agree). The total scores of the scale range from 12–84, with higher scores indicating greater levels of social support. The social support is classified into low, middle, and high support levels according to the cut-off score ranges of the MSPSS of 12–36, 37–60, and 61–84, respectively . In this study, the Cronbach’s α coefficient was 0.97 for the overall scale.
The Neuroticism Subscale of the Chinese Big Five Personality Inventory (NEO-CBF-PI) is the most comprehensive self-report questionnaire measuring the five dimensions of personality, including neuroticism. The CBF-PI consists of 40 items and has been extensively validated . The 8-item neuroticism subscale of the CBF-PI is rated on a 6-point Likert scale (1–6), with the total score ranging from 8 to 48. Higher scores are indicative of a higher level of neuroticism. Based on previous studies , levels of neuroticism are classified into high and low according to a cut-off score of 36 for the CBF-PI. In this study, the Cronbach’s α coefficient was 0.87 for the NEO-CBF-PI.
The Chinese version of the 14-item Resilience Scale (RS-14) developed by Wagnild and Young is one of the most reliable tools in measuring resilience in various age groups and different conditions [26,27]. It is composed of 14 items representing the “Personal Competence Factor” and “Acceptance of Self and Life Factor.” Each item is graded from 1 (strongly disagree) to 7 (strongly agree). Graded items are summed to provide a total score, with lower scores indicating less resilience. According to the cut-off value of 74, resilience levels are classified into high and low . The RS-14 has satisfactory internal consistency with Cronbach’s α ranging from 0.87 to 0.91, and stability with test-retest reliability ranged from 0.65 to 0.84 . In this study, the Cronbach’s α of the RS-14 was 0.96.
The Self-Compassion Scale (SCS) is the most commonly used scale to measure self-compassion at times of perceived difficulty . It is composed of 26 items and 6 subscales, including self-kindness, self-judgment, common humanity, isolation, mindfulness, and over-identification. Each item rated on a 5-point Likert-type scale for frequency (1=almost never; 5=almost always). The total score is calculated with the average of the individual subscales, and all negatively scored items are transformed. Levels of self-compassion are classified high and low according to 75% of the total SCS scores (130*0.75) as based on a previous study  (i.e., a score of 98). In this study, the Cronbach’s α of the SCS was 0.77.
2.2.6. Life events
The 48-item Life Events Scale (LES) is used to evaluate negative and positive life events that have occurred during the previous year or longer, including family, work or study, and social events . Each of the 48 life event items was anchored to 4 questions: (i) when it happened, measured by “never,” “in the past 1 month,” “in the past 1 year or longer”; (ii) whether it was positive or negative for the target person; (iii) the impact on the target person’s mental health, measured by a 5-point scale ranging from “no impact” to “very severe impact”; (iv) the duration of the event, measured by a 4-point scale ranging from “3 months,” “6 months,” “≤1 year” to “longer.” The intensity score of each life event is calculated by the time when it happened (i) multiplied by the duration (iii) and impact (iv). The total intensities of positive and negative life events are summed by the intensity score of each positive or negative life event. Based on the 75% value of this score , the total intensities of positive and negative life events are further classified into high and low levels. In this study, the Cronbach’s α of the LES was 0.94.
2.2.7. Socio-demographic and clinical characteristics
Participants were asked about their socio-demographics including residential area (northeast/ eastern/ north/ south central/ southwest/ northwest area), residential location (urban/suburban/rural), age, sex (male/female), marital status (married/ unmarried/ others), educational level (less than high school degree/high school degree /bachelor's degree or higher), religious belief (no/ yes), monthly household income (Yuan, RMB)(<1000/1000–2999/3000–4999/5000+), employment status (students/ full-time employment/ unemployment/ farmer/ other), and living mode (living by oneself/ living with spouse/ living with family/ others).
We also collected clinical characteristics including smoking status(yes/no), BMI, family history of mental illness (no/yes/unclear), whether they had previously experienced depression (yes/no), the frequency of sad mood in the past month (none/ occasionally/ sometimes/ often/ always), and their sleep quality (very good/ good/ average/ bad/ very bad). In this study, according to the WHO-recommended BMI cut-off values for adults , BMI is classified as underweight, BMI <18.5 kg/m2; normal weight, BMI 18.5 to 25.0 kg/m2; and overweight, BMI ≥25.0 kg/m2.
All procedures were approved by the ethical committee of Fujian Medical University (No. FMU2017024), and informed consent was obtained from all participants. The study adhered to the STROBE (Strengthening the Reporting of Observational studies in Epidemiology) statement . All measures were completed via the Wenjuanxing platform. The participants read and signed the written informed consent form on the platform before they completed the questionnaire in approximately 20–30 min. The questionnaire could not be submitted if it was less than half complete or contained repeated answers. Participants who completed the survey were remunerated with a RMB 10 gift card.
2.4. Statistical analysis
Data analyses were conducted using SPSS 24.0 (IBM, Armonk, NY, USA). Approximately 5% of missing data were replaced using mean value substitution, and p<0.05 was considered statistically significant. The data met the assumptions of normality as one-sample Kolmogorov-Smirnov tests were not statistically significant. Continuous variables are expressed as means and standard deviations (SDs) and were dichotomized to improve comparison . Categorical variables are expressed as proportions or percentages.
Young adults with a LEIDS-RR-CV total score <60 were considered the normal group (NG), while those with a score ≥60 were considered the risk for depression group (RDG).
We performed three analysis steps to identify influencing factors of CR. First, chi-square tests were used to compare the differences in socio-demographic and clinical variables, self-compassion, resilience, social support, neuroticism, and life events between the two groups. Second, the collinearity of the independent variables was examined by the variance inflation factor (VIF) before conducting binary logistic regression. The VIFs of the 11 variables ranged from 0.45–2.63 (ideal is <4.0), suggesting no violations of the regression assumptions .
Third, binary logistic regression with a forward conditional method was conducted to determine the influencing factors associated with CR. There is evidence in the logistic regression literature that backward selection is often less successful than forward selection because the full model fit in the first step is the model most likely to result in a complete or quasi-complete separation of response values . The dependent variable was whether the young adults were at risk for depression. The variables that were statistically significant in chi-square tests were input as independent variables.