Background

In the wake of China’s massive economic development, attention has turned to the enormous treatment gap that exists for mental health problems. Phillips et al.’s foundational study [1], representing 12% of China’s population, assessed the prevalence of any mental disorder at 17.5%, with estimates indicating that more than 91% of 173 million Chinese citizens with a mental disorder fail to reach professional help [2]. In 2011, China passed its first National Mental Health Law, articulating ambitious reform for the mental health service delivery system [10, 25]. These ethnographically-grounded vignettes were developed to present symptoms and behaviors of hypothetical persons with two major mental illnesses, schizophrenia and major depression according to Diagnostic & Statistical Manual (DSM-IV) criteria (physical health vignette of asthma not used here) [26]. All vignettes were evaluated for accuracy by a psychiatrist with expertise in cross-national studies of psychiatric disorders, for acceptability by the research team at Renmin University, and for cultural understanding by Chinese native speakers (see Pescosolido et al. for more detail) [27] [26]. Each vignette was constructed to vary by gender, education, and the majority/minority race/ethnicity. Respondents received one randomly assigned vignette. Because all respondents in these analyses were of Han ethnicity, only responses towards Han vignette individuals are used here. Vignettes are provided in Appendix 1.

Measures

Mental illness recognition

[NAME] is used here as a generic term to refer to the vignette person; in the survey, typical male and female Han names were used. Participants were asked, “Would you say that [NAME’s] situation is caused by depression, asthma, schizophrenia, stress, or something else?” Exact/specific mental illness recognition was defined as correctly identifying the vignette character’s specific mental illness diagnosis (i.e., depression or schizophrenia). This variable was recoded from the original response categories into a binary variable (correct/ incorrect). We also assessed “general” mental illness recognition, which was defined as identifying some type of mental illness diagnosis, even if the specific diagnosis was incorrect (e.g., identifying the vignette as depression when it depicted schizophrenia). This variable was also dichotomized.

Treatment recommendations

The endorsement or recommendation for three types of formal help-seeking outcomes, “a general medical doctor,” “a psychiatrist,” and “mental health professional,” were assessed. Participants were asked, “For each statement, please tell me using a scale from 1 to 10, where 1 is not at all important and 10 is very important, how important you think each would be for NAME to use in order to get help with his/her situation.” Mental health professionals include psychologists, therapists, social workers, or counselors; examples were provided if respondents asked for clarification. In a fourth response, respondents were asked, “How likely is it that [NAME’s] situation will improve on its own?” The original response set, a 4-point Likert likelihood scale, was recoded into a binary outcome (likely/not likely).

Biological/genetic causal belief

Respondents were asked the likelihood that “[NAME’s] situation” was caused by a “brain disease or disorder” or a “genetic or inherited problem.” Endorsing either one of these attributions was coded as a neurobiological binary variable (likely/not likely).

Perceived illness severity

Respondents were asked, “How serious would you consider [NAME’s] situation to be?”. Participants responded using a 4-point Likert scale of “very serious, moderately serious, not very serious, to not at all serious”, which was reverse-coded so that a higher score indicates higher severity.

Sociodemographic control variables

Age was coded in years. Gender was dichotomized (male = 1) as was rural/urban residence (urban = 1). Income was measured as a continuous variable. Marital status and educational attainment were originally included as dummy variable sets in the preliminary analysis (categories listed in Table 1). However, because of their high inter-correlation with rural/urban status, they were dropped from final analyses.

Analysis

Simple descriptive data, univariate tests and multivariate analyses are used to provide the profile of recognition and response options and to examine their correlates. Specifically, differences in the percentage of respondents who accurately recognize the general and specific disorder described in the vignette is assessed by vignette illness type using a chi-square test. Results are further stratified by urban versus rural residence. Bivariate analysis between social demographics and outcome variables using ANOVA and Pearson Chi-square Test were also conducted to identify key sociodemographic factors to be included in the model. Given that education, income, and marital status were highly related to rural/urban residence in both the literature and in our analysis, the final model included only rural/urban residence. Multiple series of adjusted regression models estimated the effects of each predictor towards the three types of help-seeking. Ordinary least square (OLS) and logistic regression modeling were used depending on whether the outcome was dichotomous or continuous. The four steps look at independent effects of key social demographic factors (Model 1), with accurate mental illness recognition added (Model 2), considering biological/genetic causal beliefs (Model 3), and finally adding perceived illness severity (Model 4). Lastly, because age suggested a nonlinear relationship with mental health recognition in prior studies [11, 14] we explored interaction effects between: a) age groups (defined categorically as ≤35 years old, 36–64 years old, and ≥ 65 years old, with 36–64 years acting as the reference category because this group might be expected to have the greatest recognition of mental illness [11, 14]); and b) gender, to determine whether the effects of recognition varied (i.e., were modified) by these two factors (Model 5 and 6). These models were run both for specific and general mental illness recognition. Statistical significance was set at p < 0.05. All analysis was conducted using Statistical Package for Social Sciences (SPSS) Version 22.

Results

Levels of accurate mental illness recognition

Table 2, Panels A-B reports the percentage of Chinese Han respondents who could recognize the vignettes as problems of mental illness (Panel A) and, more specifically, as the DSM disorder described (Panel B). Three findings stand out. First, the rates of recognition are low but substantial. Just under one-third of respondents (32.6%) were able to generally recognize the vignette as a mental illness. Second, only 18.6% were able to accurately recognize the specific mental illness depicted. When stratified by urban versus rural residence, a significantly greater percentage of urban respondents (23.3%) were able to accurately recognize the exact diagnosis in the vignette compared to rural residents (16.4%). Third, across mental illness types, accurate recognition of depression (25.5%) was significantly greater than accurate recognition of schizophrenia (11.2%). Among urban respondents, this difference was even greater; 38.0% of respondents correctly identified depression while only 8.7% correctly identified schizophrenia (p < 0.001). In contrast, among rural respondents, correct recognition of depression (20.0%) and schizophrenia (12.4%) did not significantly differ.

Table 2 Correct recognition and treatment recommendations of vignette overall and by urban and rural residence
Full size table

Levels of treatment recommendations

Table 2, Panels C-F reports the public’s assessment of appropriate response to the vignette situation. Just over 36% indicated that the situation will not improve on its own. In terms of the importance of medical professionals, the population average for mental health professionals was highest (Mean = 7.57, SD = 2.68), followed by medical doctors (Mean = 6.43, SD =2.76), and lastly, by psychiatrists (Mean = 6.09, SD = 3.26).

Correlates of recognition and recommendations

Table 3 presents the multivariate results. All models were run with both exact and general mental illness recognition, but only the former are presented. Preliminary results indicated almost no difference in significance and no meaningful differences in magnitude between the recognition variable types nor significant change in the effects of other correlates, though differences will be noted.

Table 3 Adjusted logistic regression analysis of treatment recommendations
Full size table

Panel A considers the correlates of individuals who believe that the vignette condition will not improve on its own. In Model 1, those living in urban areas, compared to those living in rural areas, report that the condition described in the vignette would not get better on its own. Adding the main effects of exact recognition does not change the sociodemographic effects, and exact recognition was significantly associated with believing that the vignette condition would not get better on its own. The next two steps, endorsing neurobiological causes for the vignette situation and perceived severity, respectively, increased the likelihood of seeing the vignette condition as not getting better on its own, with only perceived severity emerging as a significant correlate in Model 4. The effects of exact recognition were not modified by gender nor age categories (with the middle age category of 36–64 years acting as the reference category; results not shown for interactions for gender or age). In general, the results did not differ when exact recognition was substituted with general recognition.

Panel B examines the correlates of respondents’ assessment of the importance of different providers. In this case, younger individuals (≤35 years old) are less likely to see medical doctors and psychiatrists as important. However, this age effect is reversed for older individuals who are significantly less likely to recommend mental health professionals. Exact recognition is significantly associated with lower recommendations for medical doctors; and, it increases the public’s endorsement of specialty providers (both psychiatrists and mental health professionals; results similar in magnitude and significance for general recognition). Endorsement of biological or genetic attributions is significantly associated with greater importance of medical doctors and psychiatrists but with less importance of mental health professionals. Across the board, severity increases the recommendations of psychiatrists and mental health professionals.

The effects of exact recognition on respondents’ assessment of the importance of different providers were not modified by gender. When examining interactions between exact recognition and age, the effects of exact recognition were modified by age group for recommendations for medical doctor and for mental health professionals. The relationship between exact recognition and lower recommendation for medical doctor is stronger for older individuals (≥65 years old) than middle adults (36–64 years old; see Fig. 2a). The relationship between exact recognition and higher recommendation for mental health professional is weaker for younger people (≤35 years old) than middle age individuals (36–64 years old; see Fig. 2c). Interaction results remained consistent for age group and recommendation for medical doctor when substituting general recognition for exact recognition (see Fig. 2b) but not for recommendation for mental health professional (results not shown).

Fig. 2
figure 2

a Exact Recognition and Importance of Medical Doctor (Older versus Middle Age). b General Recognition and Importance of Medical Doctor (Older versus Middle Age). c Exact Recognition and Importance of Mental Health Professional (Younger versus Middle Age)

Full size image

Discussion

Mental health recognition in China

Overall, the recognition of mental illness can be characterized as relatively low in our sample. This may be expected, as is the lower recognition of schizophrenia relative to depression, a finding that remained consistent after stratifying by geographic residence. This level of recognition is much lower than in Western nations (Switzerland, Germany, and Australia) [28,29,30], where depression awareness in nationally-representative studies is between 37.5–73.7% and schizophrenia is 22.4–73.6%. However, this is consistent with other recognition findings within China (schizophrenia = 8.5–10.0%; depression = 16.1–36.3%). The results of this study suggest continued efforts are still needed, particular to address the gap in mental health recognition between urban and rural residents. The dedicated efforts by China’s 2002–2010 Mental Health Work Plan to increase public recognition of common mental disorders, specifically depression and dementia, might partially account for the greater observed accurate recognition of depression in our study.

Prior explanations of relatively poor mental health recognition identify several key contributory factors, including lack of public mental health education and awareness [11], culturally-specific idioms of mental illness that provide a socially-accepted interpretation of symptoms (i.e., “excessive thinking”) [31]; and severe stigma of mental illness [32]. The negative stigma towards mental illness among Chinese populations may have contributed to a reluctance to associate symptoms with mental illness and to label the individual as having a mental disorder [33]. Instead, there is a greater tendency to attribute socially or culturally appropriate labels (e.g., work problems or stress) to mental illness symptoms [3, 13, 34]. As the first study, to our knowledge, to utilize a nationally-based sample to capture baseline mental health literacy and its influence on different treatment recommendations for China, our findings complement the emergence of China’s 2002–2010 Mental Health Work Plan [35]. Adopting a purely diagnostically-based approach, the plan aims to improve public awareness of the characteristics of mental illnesses in order to build both help-seeking and advocacy in Chinese society. Other evaluations of this plan have not been confined to Western vignettes of psychiatric disorders. Rather, some have evaluated mental health literacy through respondent’s understanding of multiple facets of mental health (i.e., assessing both “mental disorder” and “mental health” literacy via the 20-item Mental Health Knowledge Questionnaire [MHKQ]). These broader measures of mental health literacy do indeed result in relatively higher levels of mental health literacy, as defined in a multi-component manner among both rural (58%, [14]) and urban (60–72%; [36,37,38,39]) Chinese community respondents. We believe that both approaches—a diagnostically-focused one that assesses recognition and response to psychiatric syndromes, and a broader strategy incorporating awareness of features and prevention of mental disorders—hold utility in improving recognition of specific mental illnesses as well as improving broader knowledge and promotion of mental health and mental disorders.

Core components of treatment recommendations

The SGC-MHS also allowed for an analysis of how different aspects of mental illness recognition and beliefs are associated with endorsement of formal treatment options. The ability to recognize a scenario as a mental health problem, whether generally or specifically, consistently emerged as a significant predictive factor even while controlling for other demographic and key attitudinal factors. Respondents who were able to accurately recognize the vignette tended to see a greater importance in seeking help from psychiatrists and mental health professionals. This extends research where respondents who did not see mental illness as a personal failure, but rather as a disorder, were more like to endorse treatment [8, 19, 40].

The prominent somatization of mental illness symptoms, made popular during the Cultural Revolution [34, 41], may have influenced why some respondents attributed the cause of the vignette to be biological/genetic in nature and why physical/genetic causal belief was a significant predictor in all of the treatment recommendations [8, 20]. As in many mental health services research studies, greater perceived severity was associated with treatment recommendations for psychiatrists and/or mental health professionals [42]. But the differential response to mental health professionals among some variables suggests that special effort should be targeted to their acceptance, especially as more of these providers are trained to help address China’s mental health gap. As currently stated in the 2015–2020 National Mental Health Work Plan [43], improving the lay public’s attitude towards, and understanding of when to utilize, specialized mental health providers, is not specifically addressed.

We did not find any gender differences in the relationship between recognition and treatment recommendation. Gender effects on treatment recommendation have not been previously examined in China. One study [11] did demonstrate that females were more likely to recognize alcohol abuse. However, several other studies in China [11, 15, 36] found that gender was not associated with accurate recognition of schizophrenia or depression. These findings contrast with studies conducted in Western contexts where females generally tend to show better mental health recognition [44, 45]. One potential explanation may be a floor effect in China; since the recognition of schizophrenia and depression is very low overall, gender-related effects upon recognition may be less detectable.

Our study did find age-specific differences. Prior studies in Western contexts and China have suggested that older age was associated with lower mental health literacy [11]. However, our findings suggest that relationships with age and recognition of mental illness are more nuanced in regards to specific forms of mental health treatment in China. There is a greater tendency to see the importance of seeking out a mental health professional when a middle-aged person (i.e. ages 36–64 years old) can recognize any form of mental illness. This may reflect greater life exposure to mental illnesses and health care, when compared with a younger person (≤35 years old), who may lack such experiences [46]. Counterintuitively, our findings also showed that older individuals (≥65 years old) showed a greater tendency, when recognizing either the specific or general form of mental illness, to not recommend a medical doctor for treatment, compared with a middle-aged person (i.e. ages 36–64 years old). It is possible that these older groups may not subscribe to general medical treatment when recognizing mental illness, opting instead for informal network support (e.g., from family), or other alternative (e.g., Chinese herbal medicine) forms of treatment [13]. Regardless, our results demonstrate that the relationship of mental illness recognition, age, and treatment recommendations is not linear in China. Further, older populations, in particular, might benefit from education campaigns encouraging treatment-seeking from mental health professionals when mental illness is recognized.

Limitations

The SGC-MHS is cross-sectional thus limiting causal inference. We also acknowledge the possibility that Chinese participants responded to vignettes according to their culturally-informed model of psychiatric distress (e.g., stress). This fits the cultural context, but does not match a traditional Western psychiatric diagnosis. However, since mental health service use in the Chinese medical system is delivered based upon an international psychiatric diagnostic system (ICD 10) [47], we do not view this as impacting the validity of our findings. Further, respondent burden limited the ability of the SGC-MHS to more deeply investigate cultural beliefs or social influence processes linked to networks [42]. Finally, this is an analysis of recommendations for others. Recommendations might be substantially different from what individuals would do when actually confronted themselves with the situation described since lack of an attitude-behavior linkage has been well-documented. We conceptualize this study as a description of the contours of the “cultural toolbox” of beliefs, attitudes, and cultural scripts among the Han Chinese.

These issues should be the focus of future study in China and elsewhere because they are problem generic to much mental health utilization research. In particular, given that the National Mental Health Law (2012) and subsequent reform took place just after the 2011 date of the SGC-MHS, there needs to be a follow-up to document possible shifts in contemporary recognition of mental illness and public support for treatment options. In essence, our study provides a valuable baseline by which to characterize mental health recognition among China’s population and provides a valuable data point to facilitate scale-up efforts of China’s mental health system.

Conclusion

These mental health literacy findings provide important new data to guide scale-up of China’s expansion from hospital-based delivery systems to a community-based mental health care system, as well as its shift from involuntary admissions of primarily psychotic disorders to voluntary treatment of a range of mental health problems. We confirmed an expected low overall level of accurate mental illness recognition nationally, with a particular gap between urban and rural respondents. These effects also vary for different age groups. In sum, this study captures baseline mental health literacy at a unique moment in China’s mental health reform history. Although the country has continued to make reforms since the time our study was conducted, these findings can still help shape China’s future development of mental health systems. In addition, the work may also benefit other lower middle income countries (LMIC) planning to undergo similar reforms to develop community mental health care delivery systems [38, 48]. Through a continued understanding of the processes that shape effective mental health treatment utilization, we intend our findings to promote the basic human right to mental health care and to alleviate the global burden of mental illness.