Common mental disorders among medical students: systematic review and meta-analysis of Brazilian studies

ABSTRACT BACKGROUND: Common mental disorders (CMDs) have been correlated with consequences in different domains of life. OBJECTIVE: To summarize the prevalence rates of CMDs and factors associated with them among students at Brazilian medical schools. DESIGN AND SETTING: Systematic review and meta-analysis of studies developed in Brazilian medical schools. METHODS: In October 2021, searches were carried out in seven electronic databases, in Google Scholar and in reference lists. Observational studies reporting prevalence rates of CMDs among students at Brazilian medical schools were sought. Variables associated with CMDs arising from multivariate regression models were included in the synthesis. A meta-analysis was developed using a random-effects model and the risk of bias was assessed using an instrument developed from previous references. RESULTS: Fourteen original studies were included. The pooled prevalence rate of CMDs among undergraduate students at Brazilian medical schools was 43.3% (95% confidence interval = 38.9% to 47.6%; I2 = 87%; n = 3,927). Among the nine studies in which multivariate analyses were conducted, five showed risk associations between CMDs and medical school-related dissatisfactions, among which the desire to abandon the medical course can be highlighted (n = 3). In three studies, CMDs were associated with sleep indicators. CONCLUSION: Considering that the prevalence of CMDs among medical students is higher than in the general population, we recommend that Brazilian medical schools should give greater attention to this topic, and should enable expansion of care offerings relating to mental health. SYSTEMATIC REVIEW REGISTRATION: Prospective Register of Systematic Reviews (PROSPERO) database (CRD42020142184).


INTRODUCTION
Medical courses are, as a rule, characterized by an integral routine of theoretical and practical activities, early insertion in a praxis that has little margin for errors and permanent contact with illness and death. In the Brazilian context, it is recognized that these demands are associated with different health risk behaviors, such deprivation of sleep 1 and leisure, 2 which, in turn, are determinants of mental health. 3,4 Since the mental health indicators observed among Brazilian medical students are generally poorer than those observed in the general population, 5 the mental health of medical undergraduate students is an emerging agenda in Brazilian research, enhanced by recognition of the most drastic outcomes associated with it. 3 For example, common mental disorders (CMDs) are one of the mental health indicators that have been studied in these populations. 5 CMDs present as a mixture of somatic, anxiety-related and depressive symptoms, such as insomnia, irritability, forgetfulness and difficulty in concentrating, among others. 6 However, these symptoms do not meet sufficient formal criteria to be diagnosed as depression or anxiety, according to the classifications of the DSM-V (Diagnostic and Statistical Manual of Mental Disorders, 5 th edition) and the International Classification of Diseases, 11 th revision (ICD-11). 7 Nonetheless, the literature available suggests that CMD symptoms among Brazilian medical students are already observed in the early stages of the medical course. 8 In this sense, it can be understood that early recognition of CMD prevalence rates, as well as the factors associated with these disorders, can provide support for efforts to address them at different levels. These can range from the strategies available for mental healthcare in medical schools to a broader view of the curricula of Brazilian medical schools.
Considering the distinct biopsychosocial impairments associated with CMDs, identifying the prevalence rates of CMDs and factors associated with them can support development of preventive strategies involving these populations.

OBJECTIVE
Thus, the aim of this study was to identify and statistically summarize CMD prevalence rates and factors associated with these disorders among students at Brazilian medical schools.

METHODS
We developed a systematic review of the literature, with metaanalysis. Its protocol was previously registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42020142184) and its design and report were developed from the items of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. 9 Our inclusion criteria were formulated from the possible items of the "PICOS" structure, considering the following: (I) Population: students at Brazilian medical schools, without restrictions regarding the stage of the course and the profile of the university (i.e. public or private); (II) Outcome: "common mental disorders", not considering studies that addressed other terminologies and other specific psychiatric disorders/diseases, such as anxiety, stress and depression; and (III) Study design: observational studies (e.g. cross-sectional or cohort studies), without restriction as to their representativeness (i.e. whether local, regional or national), reported in English, Portuguese or Spanish. It is worth mentioning that, in accordance with the study designs of interest, the PICOS items "Intervention" and "Control" were not applicable.
On the other hand, dissertations, theses, abstracts and preprints were not considered for the synthesis. Nor was research involving students on healthcare courses in which no stratified analysis of CMD prevalence rates among medical students was presented.
As our secondary outcome, we sought to identify the variables which tool was used, a report on its previous validation and information that enables replication of the measurement, in the case of questionnaires developed specifically for the study); (IV) Losses and dropouts (i.e. information on losses and dropouts, along with the percentage of students whose data were analyzed, compared with the initially proposed number); and (V) Analysis protocol (e.g. analysis plan and adequacy of the method used to identify the prevalence of CMDs in the sample). This instrument may be requested from the corresponding author.
A random-effects meta-analysis was conducted, based on the original prevalence rate data from each individual study and the respective 95% confidence interval (CI). Given that variability data is not reported with the 95% CI in many studies, these data were manually calculated from the sample size and the prevalence of identified CMDs and subsequently checked in the statistical software.
To conduct the analysis, the Review Manager software was used (version 5.4; Cochrane Collaboration, 2020). Thus, the summarized effect was constructed from the random model, considering the differences between the samples (e.g. phase/year of the course and type of institution). The I 2 statistic was used to assess heterogeneity between studies: this was classified as "moderate heterogeneity" when the summarized effects using I 2 were between 50% and 74%, or as "high heterogeneity" when I 2 ≥ 75%, as suggested in the study by Higgins et al. 11

RESULTS
Overall, the searches retrieved 325 potential studies (Figure 1).
After identification and removal of duplicates (n = 71), 254 studies remained for assessment using their titles and abstracts.

Studies included in descriptive synthesis and meta-analysis (n = 14)
Identification Screening Eligibility Included Figure 1. Flowchart of systematic review and meta-analysis.
Multivariate regression models to identify factors associated with CMDs were developed in nine studies (64.3%). Despite the differences relating to the types of regression and variables used for adjustment, the results suggest that CMDs were frequently associated with dissatisfaction relating to the medicine course. 12,15,19,22,24 The factors of "desire to quit the course" 19,22,24 and "feeling uncomfortable with the course" 12,15 can be highlighted. Furthermore, risk associations between CMDs and sleep indicators 8,12,22 were also shown ( Table 2).

DISCUSSION
Based on the data from 14 original studies, the pooled prevalence rate of CMDs was 43.3% among undergraduate students at Brazilian medical schools. Compared with a previous study, 5 which reported a pooled prevalence of 31%, our study had two specificities: inclusion of studies that specifically used the 'common mental disorders' terminology and inclusion of studies that assessed CMDs through questionnaires other than the SRQ-20.

Specifically regarding CMD assessment, it is important to highlight that the GHQ-12 is classified by the Brazilian Federal
Council of Psychology as an "unfavorable psychological test" and, therefore, is not indicated for psychologists' professional activities. 25 However, these studies using the GHQ-12 were kept in our synthesis, 13,21 for two reasons: our recognition of a Brazilian validation study on the GHQ-12 in which the SRQ-20 was used as    a comparison; 26 and our recognition that the GHQ-12 has been used in other population-based studies. 27,28 Even though no nationwide survey exists, our findings corroborate the understanding that the prevalence of CMDs observed among Brazilian undergraduate medical students is higher than that of the general adult population. 7,29 We believe that this result is important and can contribute to efforts relating to the debate about the structure of medical courses offered in Brazil, through highlighting the need for provision of specialized mental healthcare for undergraduate students. Identifying these indicators in the early phases of the medicine course can be of interest, in order to avoid chronicity among them, along with the more deleterious outcomes mentioned above. 3 Another suggestion is to investigate whether there are phases, course cycles (i.e. basic, clinical or internship) or even moments in each semester when the risk of developing CMDs is higher. It is worth mentioning that in the longitudinal studies by Ferreira et al. 8 and Costa et al., 16 it was suggested that the prevalence of CMDs increases over the semester 8 and over the years, 16 respectively.
As our secondary result, we also showed that course-related dissatisfaction 12,15,19,22,24 and sleep indicators 19,22,24 were associated with CMDs. Even though three studies pointed out that CMDs were associated with the desire to drop out of medical school, it has been recognized in the literature available that medicine courses have lower dropout rates than nursing, pharmacy and dentistry courses. 31 This finding can largely be explained by different perspectives, such as remuneration, job expectations and social recognition. 32 Nonetheless, apart from the social role of medical doctors and the social constructs that permeate the profession, these associations relating to dissatisfaction and frustration with the course suggest that there is a need for periodic assessment of course workloads.
Adoption of innovative strategies that go beyond purely technicist and poorly humanized approaches can also be highlighted. 33 Regarding sleep indicators, it has been recognized that sleep disorders are associated with other mental health indicators, such as anxiety and depression. 34 The routines required by different curricular components and the pressure for better performance in tasks can lead to constant sleep deprivation. In addition, there is also high and recurrent consumption of psychostimulant substances (e.g. energy drinks and caffeine) among medical students, to prolong their state of alertness. 35 Thus, beyond guidance about the harm of sleep deprivation, better care in the internal organization of courses is suggested, so that overlapping of activities, tests and/or important tasks on specific dates or in specific weeks can be avoided.
In addition, since most of the studies included here had cross-sectional designs, we would recommend that longitudinal studies should be conducted. These would not only investigate whether there are phases/cycles of higher risk of CMDs during the medical course, but also provide understanding of the possible causal relationships between the variables. Therefore, we would emphasize that caution is required in interpreting the findings from this study, since most of the studies included were cross-sectional. This formed a limitation on deeper discussion of temporality and causality, i.e. whether CMDs are the cause or the consequence of sleep disorders.

CONCLUSION
Our study showed that the pooled prevalence rate of CMDs was 43.3% among the undergraduate medical students. It also showed that risk associations existed between CMDs and course-related dissatisfaction and sleep indicators. Considering that the prevalence of CMDs among medical students is higher than in the general population, we recommend that Brazilian medical schools should give greater attention to this topic and should enable expansion of care provision relating to mental health.