Associations between socioprovincial factors and self-reported mental disorders among students from grade 4 to 8 in rural China

Background The focus on child mental health in developing countries was increasing. However, little was known in China. This study aimed to explore the associations between socioprovincial factors and self-reported mental disorders in rural China. Methods Data were from a publicly available survey with 54,498 students from Grade 4 to 8 in rural China. Chi-square test was used for descriptive analysis. Self-reported mental disorders included overall mental disorder, study anxiety, personal anxiety, loneliness, guilt, sensitivity, symptomatic psychosis, phobia, and impulsivity. Multiple logistic regressions and errors-in-variables regression models were employed to explore the associations between socioprovincial factors and mental disorders. Poisson regressions and errors-in-variables regression models were adopted to reveal the associations between socioprovincial factors and number of self-reported mental disorders. Results Descriptive statistics showed that mental health was poor in rural adolescents in China. Logistic regression showed that the odds of overall mental disorder and study anxiety were 189% (AOR = 2.89, 95%CI: 2.76, 3.02) and 92% (OR = 1.92, 95%CI: 1.84, 2.00) in Gansu more than those in Anhui, while the odds of personal anxiety, guilt, sensitivity, symptomatic psychosis, and phobia were 92% (AOR = 0.08, 95%CI: 0.08, 0.09), 71% (AOR = 0.29, 95%CI: 0.27, 0.30), 88% (AOR = 0.12, 95%CI: 0.11, 0.13), 69% (AOR = 0.31, 95%CI: 0.29, 0.32), and 78% (AOR = 0.22, 95%CI: 0.21, 0.23) in Gansu less than those in Anhui. Moreover, Gansu (Poisson regression: IRR =1.45, 95%CI: 1.42–1.47; errors-in-variables regression: Coefficient = 0.26, 95%CI: 0.16, 0.36), Ningxia (Poisson regression: IRR =1.63, 95%CI: 1.60–1.67; errors-in-variables regression: Coefficient = 0.43, 95%CI: 0.32, 0.53), Qinghai (Poisson regression: IRR =1.65, 95%CI: 1.60–1.69; errors-in-variables regression: Coefficient = 0.44, 95%CI: 0.34, 0.55), and Shaanxi (Poisson regression: IRR =1.28, 95%CI: 1.25–1.30; errors-in-variables regression: Coefficient = 0.11, 95%CI: 0.00, 0.21) were significantly associated with the number of self-reported mental disorders. Conclusion Class and provincial disparities in self-reported mental disorders were reported among the students from Grade 4 to 8 in rural China. Mental health care supported by governments and schools could be an effective way to reduce the disparities in mental disorders among the adolescents.


Background
There was a growing body of evidence suggesting that adolescents were susceptible to a variety of mental disorders among adolescents. Currently, mental disorders including anxiety disorders [1] and depressive symptoms [2] in adolescents had been attracted by academic circle. For example, an epidemiological study found a high incidence of comorbidity in children and adolescent psychiatry [3]. Another study indicated that depressive and anxiety symptoms often co-occurred [4]. Empirically, early adolescent anxiety disorders were related to lower self-esteem from early adolescence through young adulthood, with social phobia having the greatest impact [5].
Regarding socio-geographical factors, prior research reported that socioregional factors [6], social gradient in antisocial behavior [7], teacher perspectives [8], and income change [9] had associations with mental disorders among adolescents. Furthermore, a study reported race moderated the link between physical discipline and externalizing behavior problems for European American adolescents during Grades 6 and 8 [10]. Another large epidemiological study in Israel found an association between religion/ethnicity and internalizing and externalizing disorders in Muslim adolescents [11].
Multiple studies documented the associations between household wealth and mental health changes among adolescents worldwide. For instance, a study concluded that national wealth had associations with poor mental health at the aggregated level [12]. Regarding the association between wealth inequality and the risk of mental disability in the Chinese population, a study with nationally represented, population-based data from the second China National Sample Survey on Disability 2006 suggested that wealth was a significant predictor in the distribution of mental disability [13]. Furthermore, a study with data from 2060 young adults aged 18-27 in 2005-2011 from the Panel Study of Income Dynamics suggested family wealth affected mental health [14]. Likewise, a study with data (2001-2012) from the Household, Income and Labour Dynamics in Australia survey reported that low household wealth prior to disability acquisition in adulthood resulted in a greater deterioration in mental health than those with high wealth [15]. Additionally, a study with the case of three distinctive communities in Haiti found household agricultural wealth was significantly and strongly associated with both reductions in depression symptoms and anxiety symptoms [16].
However, there were a limited number of studies to report the associations between socioprovincial factors and self-reported mental disorders in rural China. Prior research on the associations of socioprovincial factors with mental health were seldom documented because the data employed did not include provincial variable. This study tried to fill into the gaps.

Data source
This study employed a publicly available survey data [17]. The dataset was aggregated from 10 different school-level surveys that conducted in rural areas of five provinces of from 2008 to 2015, which covered 54,498 students from Grade 4 to 8 across 65 counties in rural China. The dataset included information on student characteristics (gender, grade, and surveyed provinces) and family background (household asset) as well as mental health situations.

Main variables
Here, dependent variables referred to overall mental disorder, mental subscales (study anxiety, personal anxiety, loneliness, guilt, sensitivity, symptomatic psychosis, phobia, and impulsivity), and number of self-reported mental disorders. Additionally, overall mental disorder and mental subscales were reflected by the questions: "Percentage of kids who have 7+ points in any category" and "Whether or not kids have 7+ points in study anxiety, personal anxiety, lonely, guilty, sensitive, symptomatic, phobic, or impulsive subcategory", respectively. Their response options were yes (=1) and no (=0). Number of self-reported mental disorders was calculated by arithmetic sum of response options of study anxiety, personal anxiety, loneliness, guilt, sensitivity, symptomatic psychosis, phobia, and impulsivity.

Statistical analysis
The associations between socioprovincial factors and self-reported mental disorders were mainly calculated by multiple logistic regressions. Due to response options with rare "yes" relative to "no" in loneliness and impulsivity, the associations between socioprovincial factors and mental disorders were calculated by complementary log-log regression (stata program: cloglog). The associations between socioprovincial factors and number of self-reported mental disorders were explored by Poisson regression with incidence-rate ratios (IRR).
Due to survey data, measurement errors could be considered in the errors-in-variables regression models when the subjective response variables were possibly measured with errors. Here, the assumed reliability of standardized index for household wealth was defined as 0.85. Subsequently, errors-in-variables regressions on associations between socioprovincial factors and mental disorders were conducted.

Associations between socioprovincial factors and mental disorders
Regarding gender, in Tables 2 and 3, significant adjusted odds ratios (AOR) and significant coefficients of male indicated male students were less susceptible to mental disorders than female students. Considering school grade, in Tables 2 and 3, significant adjusted odds ratios and significant coefficients indicated students in Grade 5 were less susceptible to mental disorders than students in Grade 4. In addition, students in Grades 7 and 8 were more susceptible to mental disorders than students in Grade 4.

Associations between socioprovincial factors and number of self-reported mental disorders
In Table 4

Discussion
Notably, high prevalence of self-reported mental disorders was reported among the sampled students. The associations between socioprovincial factors and selfreported mental disorders and the number of selfreported mental disorders were confirmed. Besides gender disparities in mental disorders, there existed class and provincial disparities in mental disorders among the sample. Regarding gender, this study was consistent with a series of early studies. For example, a Jordan adolescents study reported significant associations of age and gender with high prevalence of mental disorders [19]. A study with data from China National Sample Surveys on Disability in 1987 and 2006 indicated that gender of children was consistently associated with psychiatric disability among children [20]. The finding in this study was also in line with a study in South Korea which reported adolescent depression experience was highly associated with gender mix in the school [21]. In this study, there might be gender disparities between mental disorders.
Regarding schooling, this study was in line with a series of early studies. For example, a study demonstrated important mental health issues with a high incidence of comorbidities in poverty-stricken areas and left-behind children and adolescents in 40 primary and middle schools [22]. Globally, a substantial study also documented high mental disorders prevalence among adolescent students in Brazilian [23], left behind adolescents in Anhui province [24], adolescent age groups in Saudi Arabia [25], Australian child and adolescents [26], Hungarian children [27], and Lithuanian youth [28].   Additionally, an available research on the prevalence of child and adolescent psychiatric disorders in India was reviewed, synthesized, and evaluated [29]. The novel finding in this study was there existed class disparities in mental disorders among the students. Meanwhile, the students in higher class were more likely to be susceptible to mental disorders than those in lower class in junior school. The students in elementary school were less likely to be susceptible to mental disorders than those in junior school. IRR of standardized index for household wealth was equal to 1, which indicated that household wealth was not the main determinant of the number of mental disorders. This finding could be verified by a current research which reported family functioning had associations with adolescent health and emotional well-being [30]. Furthermore, the research outcome was not in congruent with the published associations between wealth inequality and positive emotion [31,32].
To the best of the knowledge of the author, this was the first study to report provincial disparities of mental disorders in rural China. Some early studies documented mental health of school students in Hong Kong, Shanghai, and Beijing [33], Shanghai [34][35][36], and Sichuan [22]. Likewise, mental health in rural China also was reported in northwestern China [37] and western China [38]. Compared with these studies, the novel finding in this study highlighted provincial disparities of mental disorders in rural China.

Policy implications
This study highlighted the importance of socioprovincial factors in mental health. Adolescent mental disorders were confirmed to represent a risk marker for a number of later adverse outcomes [39]. Socioeconomic disadvantage and psychological deficits were found to contribute to criminal offending independently and with roughly equal magnitude [40]. Thus, adolescent psychopathology and support need be provided to minimize adverse outcomes. Regarding the provincial differences among ORs and IRRs, to my opinion, governments and schools should concern the differences of mental disorders between province-level units in China. Obviously, governments and schools were responsible for provincial imbalance of mental health service in rural China.
Understanding the associations between socioprovincial factors and self-reported mental disorders could help identify high-risk adolescents and take steps to minimize their disparities in mental disorders. Some similar studies recommended school mental health resources [41], developing personalized approaches to clinical assessment [42], adaptive interventions [43], and increasing health workforce [44] should be adopted to prevent and reduce the prevalence of mental disorders Note: *p < 0.1, **p < 0.05, ***p < 0.01. SIHW Standardized index for household wealth among the students. On the basis of the results in this study, optional prevention measures against mental orders need to smooth the difference of funds, human resource, and facilities among schools and provinces. Mental health care policies aimed at a redistribution of health resources at the provincial level also could decrease health service utilization inequalities in the adolescent mental well-beings.

Limitations
Some limitations in this study should be highlighted. First, this study employed a second-handed survey data which did not provided specific items of MHT. Thus, reliability (split-half/Cronbach's alpha, test-retest reliability, alternate form reliability, inter-rater reliability) and validity (content validity, construct validity, convergent or concurrent validity, discriminant validity, criterion (or predictive) validity) could not be calculated. Second, only five provinces were analyzed in this study. The research outcomes could be limitedly generalized to the whole country. Finally, several key definitions were ambiguous. Taking household wealth as an example, whether the wealth was parental or students' was not defined. Meanwhile, a study with data from the UK Millennium Cohort Study indicated children's mental health was not influenced by parental housing wealth but family permanent income and socioeconomic characteristics [45]. In addition, some key variables including socioeconomic inequalities [46], socioeconomic status [47], length of residence [48], unhealthy behaviours [49], and physiological engagement in social contexts [50], which were confirmed to be associated with adolescent mental disorders, were not included. Thus, mediating and moderating analyses could not be conducted. Accordingly, further research needs more variables to gain a better understanding of the disparities in student mental health.

Conclusion
In conclusion, this study reflected the associations between socioprovincial factors and self-reported mental disorders among students from Grade 4 to 8 in rural China. Especially, this study reported disparities in gender, class, and province regarding mental disorders. Regarding the situation reflected by this study, public policy intervention should be highlighted and redesigned to solve the mental disorders of the adolescents in rural areas of China.