Medical student depression, anxiety and distress outside North America: a systematic review

North American medical students are more depressed and anxious than their peers. In the UK, the regulator now has responsibility for medical students, which may potentially increase scrutiny of their health. This may either help or hinder medical students in accessing appropriate care. The prevalences of anxiety, depression and psychological distress in medical students outside North America are not clear. A better understanding of the prevalence of, risk factors for and results of psychological distress will guide the configuration of support services, increasingly available for doctors, for medical students too.


INTRODUCTION
Doctors report high levels of stress, common mental health disorders and alcohol misuse, 1,2 sometimes with catastrophic consequences. 3 A number of recent reports have focused attention on the area, with calls for better services for doctors with health problems, emphasising that patients need healthy doctors. 4 The Practitioner Health Programme is one model of such care and its early results are very positive. 5 Time as a medical student, although enjoyable and rewarding, is associated with significant pressures such as those imposed by long hours of study, a high workload 6 and considerable financial pressures. 7 Suicides are not unknown 8 and the death of Daksha Emson 9,10 in 2000 has been a major driver for change within the National Health Service (NHS). 11 A recent study of doctors on long-term sick leave suggested that interventions should be delivered in medical school to make doctors aware of their own health needs and of how to access appropriate support. 12 In the UK the General Medical Council (GMC) now has a regulatory responsibility for medical students and medical education. 13 It is not yet clear what impact this will have on either the health or the behaviour of medical students. It seems likely that ill health which might impact on professional performance will be better identified and managed. The substantially increased scrutiny of students who are going through significant life changes may be less helpful.
A recent systematic literature review described high levels of anxiety, depression and psychological distress amongst medical students in North America. 14 This distress was shown to be higher than in the general population, including in age-matched peers. Perhaps even more concerning is the finding that results from this review also revealed an association between psychological distress and a decline in academic performance, and impairments of professionalism and empathy towards patients. 14 However, the extent to which these results can be extrapolated to other educational settings is not clear. In the USA, medicine is taught to graduate students who are, accordingly, older and perhaps more committed. They may, however, have higher levels of debt and different social stressors, such as parental status. 15,16 The aim of this study was to systematically review the literature on depression, anxiety and psychological distress in medical students outside North America. In order to obtain a sufficiently detailed picture of the issues, whilst maintaining a degree of homogeneity with regard to the educational systems studied, we included studies from Europe and the Anglophone world outside Europe. We aimed to identify not only prevalence data, but also risk factors that might be associated with such common mental disorders.

METHODS
Using OvidSP, the following databases were selected: (i) Ovid MEDLINE (R) from 1948 to October 2013; (ii) PsycINFO from 1806 to October 2013, and (iii) EMBASE from 1980 to October 2013.
Databases were searched for primary research data pertaining to medical student distress using a combination of free-text and medical subject heading (MeSH) search terms (Appendix S1, online). No date limits were applied.
Searches were evaluated against strict inclusion criteria: (i) the study population must comprise undergraduate medical students; (ii) the study must be located in a European or English-speaking country outside North America; (iii) the study exposure must include depression (and its subtypes), anxiety, adjustment disorder, or emotional distress, and (iv) the study must use a validated assessment tool. There are many assessment tools. Those validated to identify cases of depression or anxiety were included as such. Tools not validated to assess 'caseness' but nonetheless including an assessment of sadness, anxiety and other negative mood states were included under the heading 'psychological distress'. This categorisation, which provides an inclusive concept of distress, has been used elsewhere. 17,18 Studies that examined student mental health difficulties arising solely from psychosis, schizophrenia or a significant life event were excluded.
An additional hand search of the reference lists of the final selection of papers was conducted.
A set of search terms was agreed between the authors. After the initial search, each subsequent stage was carried out independently by each author. Areas of disagreement were resolved by discussion. Data were extracted into identical spreadsheets by each author. These spreadsheets were then merged and discrepant findings were again resolved by discussion with reference back to the text of the paper.
There is no single set of quality measures that can be used for all systematic reviews. Our focus in assessing quality was to highlight where bias may have been an issue. Bias can be introduced to prevalence studies in the way that participants are selected, by low study numbers and by low response rates. We extracted data from each paper on each of these measures. To further assess the quality of the research process, we checked whether it was made clear that ethical approval had been sought. As a final measure of overall quality, we noted whether the study appeared in a peer-reviewed publication on the basis that in general peer review supports better-quality studies. We made no attempt to combine our measures of quality. Figure 1 shows a full flow diagram illustrating the systematic review process.

RESULTS
A total of 29 papers met the inclusion criteria (Appendix S2, online) and were included for review. These papers included 9784 medical students. Sample sizes ranged from 60 19 to 1743. 20 The median sample size was 273 medical students. Almost all studies were cross-sectional.

Assessment tools used
The tools most frequently used to identify depression, anxiety and psychological distress were the Beck Depression Inventory (BDI) 21 and the 12-item General Health Questionnaire (GHQ-12). 22 Other less frequently used tools were the Hospital Anxiety and Depression Scale (HADS) 23 and the Beck Anxiety Inventory (BAI). 24 Many other tools, some developed by the researchers themselves, were also used to assess the prevalence of psychological distress ( Table 1).

Prevalence of depression
We identified 14 studies that measured the prevalence of depression (Table 2). These studies used seven validated depression assessment tools. Even Figure 1 Selection of studies for this systematic review studies using the same tool used different cut-offs. For example, with reference to the BDI, Alvi et al. 25 classed a score of > 14 out of 63 as representing a case, whereas Mancevska et al. 26 used a score of > 17 out of 63. Five studies did not state their cutoff score for depression. [27][28][29][30][31] The majority identified a high prevalence of depression, although prevalences ranged from 6.0% 30 to 66.5%. 31 One study 32 identified a lower rate of depression amongst 255 medical students than amongst a control group of health science, nursing and architecture students, although the actual prevalence was still high at 17.0% (p < 0.045).

Prevalence of anxiety
We found 11 studies that measured the prevalence of a number of aspects of anxiety (Table 3), using seven validated anxiety assessment tools. Three studies did not state the scores used as cut-offs for anxiety. 28,33,34 Prevalences ranged from 7.7% to 65.5%.
With the notable exception of that by Samaranayake and Fernando, 32 few studies examined specific anxiety disorders such as generalised anxiety disorder in any formal way and thus these studies described results that were not always directly comparable. Some studies examined the prevalence of social anxiety 34 or trait anxiety 26 as opposed to state anxiety or an anxiety disorder. Several studies did not clearly differentiate between anxiety symptoms and a clinical diagnosis of an anxiety disorder. 26,32 Prevalence of psychological distress We identified 16 studies that measured the prevalence of psychological distress using eight validated assessment tools (Table 4). All except one 35 study identified high levels of psychological distress. Recorded prevalences ranged between 12.2% 19 and 96.7%. 41 The weighted mean prevalence was 29.6%. There is likely to be significant publication bias in the reporting of symptoms of psychological distress.

Associated factors
A number of papers included data on risk factors associated with reports of psychological distress, depression and anxiety.

Change during time as a medical student
In the UK and Europe, by contrast with the USA, most medical students are in their first degree course; typically they will start in their late teens. This is a time of significant change, growth and development. A number of the papers included in our study found that levels of common mental disorders and psychological distress differed depending on how far students had progressed through their studies.
We found evidence that depression, 36 anxiety 33,36 and psychological distress may be more common as the student progresses through the course. Intriguingly, we found that more studies demonstrated that these difficulties became less prevalent towards qualification. This was the case for depression, 25,26,30 anxiety 25,26,37 and psychological distress. 19,29,[38][39][40] Gender In all three areas, we identified studies that reported higher levels of difficulty in female stu-dents. 25,29,37 However, a number of other studies reported either no difference between the sexes 28 or higher levels in male students. 28,41 Social and economic factors Karaoglu and Seker 28 identified low family income as a risk factor for medical student depression and anxiety.

Quality of studies
Prevalences of depression in the full sample ranged from 6.0% to 66.5%. Studies reporting a response rate of > 80% reported prevalences of 12.9-32.0%. Among only those studies that clearly reported the provision of ethical approval, 32,42,43 prevalences ranged from 6.0% to 32.0%. Only one study 43 randomly sampled a population of medical students, reporting a mean prevalence of depression of 14.0%.
Prevalences of anxiety across the studies ranged from 7.7% to 65.5%. Amongst studies with response rates of > 80%, prevalences ranged from 8.0% to 50.3%. Amongst only those studies reporting ethical approval, 32,34,43 prevalences ranged from 8.0% to 50.5%. Only one study 43 randomly sampled a population of medical students, reporting a mean prevalence of anxiety of 43.0%.
Prevalences of psychological distress across the studies ranged from 12.2% to 96.7%. In those studies with response rates of > 80%, prevalences ranged from 13.7% to 96.7%. Amongst those studies that reported the provision of ethical approval, prevalences ranged from 13.7% to 68.6%. DISCUSSION We carried out a systematic review of the literature describing prevalences of depression, anxiety and psychological distress in medical students in the UK, Europe and the wider Anglophone world excluding North America. We found 14 studies describing depression, 11 describing some form of anxiety and a further 16 describing what we have called 'psychological distress'. We found prevalences of depression to range from 6.0% to 66.0%, those of anxiety from 7.7% to 65.5% and those of some form of psychological distress to range from 12.2% to 87.3% of students.
In the prevalence studies included in this review, we also examined which factors if any might be associated with common mental disorders. We found a very mixed picture, which to some degree represented the range of study quality we observed. More studies showed that depression and anxiety reduced in prevalence during the course of a medical degree, although there were few truly longitudinal studies. Whereas in the general population common mental disorders are more often seen in women, our studies described a more mixed picture in which only psychological distress was clearly distributed in this way. A small number of papers described an association between depression and anxiety and academic stressors such as examinations, although no causal inferences could be drawn. The few studies that examined wider social and economic risk factors showed a positive association with depression and psychological distress.
The results of our review suggest that depression and anxiety are more prevalent amongst medical students than among peers of a similar age. However, this finding should be interpreted with care. Firstly, the better-quality studies tended to report lower prevalences and the extent to which the higher-than-background results are related to methodological shortcomings is not clear. Secondly, all of the studies we included were specifically described as studies of students' mental health. Goodwin et al. 44 have shown that such a strategy can lead to an overestimation of the prevalence of common mental disorders. However, our findings are in keeping with Dyrbye et al. 14 's review of students in North American medical schools, which found a higher prevalence of depression, anxiety and distress than in the background population.
Given the low number of studies, and the methodological limitations of the studies we did find, results regarding factors associated with common mental disorders are hard to interpret. There is a suggestion that common mental disorders may decrease in prevalence over time, by contrast with Dyrbye et al.'s 14 finding. This may relate to chance or methodological weaknesses, especially the lack of longitudinal studies outside North America. It may also relate to the differences in the student body as medicine in North America is a graduate degree course. Hence, by the end of the course, American students are older, likely to be more indebted, and more likely to be exposed to additional psychosocial stressors such as parenthood. 15,16 Our study has a number of strengths. We examined a large number of databases and included studies from a range of countries and educational systems outside North America. We included only those studies in which a validated tool was used to assess the psychological health of medical students.
Regrettably, the studies we included are far too heterogeneous to merit any form of meta-analysis.
Most of the studies we found had methodological limitations. With a few notable exceptions, sample sizes were small, many studies gave little or no information on methods of recruitment, and, to our great surprise, a number of studies did not give a clear statement on whether or not they had been given ethical approval. It should be noted that prevalences of common mental disorders were lower in the better-quality studies. All except one study showed a prevalence of depression or psychological distress higher than that in the background population where applicable, and there is likely to have been significant publication bias. Only three of the studies included in our final analysis had any longitudinal component, which greatly limits any conclusions on the impacts of time, progression and increasing maturity on the prevalence of psychological disorders in this population.
To maximise the data available for our study, we included research from the UK, Europe and Anglophone countries across the world. The students included held a number of things in common, but came from and studied in a range of different cultural settings. It is well recognised that prevalences of and risk factors for depression vary among different countries. A recent review by Kessler and Bromet 45 highlights evidence showing that depression is more common in high-income countries, and that the female preponderance is found across different countries. An adverse impact on education is another cross-national outcome. We were not able to incorporate data on individual countries or cultures in our review and we cannot be certain that assessment tools developed in the UK or the USA are capturing the same issue when they are used in Africa. Nevertheless, we feel that our inclusive approach has merit inasmuch as no particular pattern emerged among high-and middle/low-income countries, yet high prevalences of depression were found in medical schools across the world.
We think on balance that there is evidence to suggest that medical students suffer higher levels of depression, anxiety and psychological distress than their non-medical student peers. We are unable to say conclusively whether this is more of a problem at the beginning or the end of medical training and methodologically sound longitudinal studies are required to clarify this. We suggest that the associations, albeit limited, with the socio-economic status of the medical student merit a greater level of interest. Medicine needs to draw upon all sections of society and must not be the preserve of the most affluent students. If the occurrence of common mental disorders represents one reason why those from less advantaged backgrounds find it harder to become doctors, then we must establish structures to identify those at risk and provide them with appropriate support.
One key area missing from our study refers to what happens to those who do have psychological difficulties as medical students, in terms of both their academic results and the provision of care. In an era in which, in the UK, medical students now come under the auspices of the GMC, issues of stigma and the fear of jeopardising a promising career may lead a medical student to not seek appropriate care, lest he or she is negatively labelled. Overwork, 46 social isolation 47 and alcohol misuse 48,49 are commonly used coping strategies in qualified doctors. These are not especially helpful in the short term and, more worryingly, may become more firmly established later. Medical students need guidance on the development of more appropriate ways of managing distress.
The time has now come for a high-quality, multicentre longitudinal study of medical students. Such a study should be able to tease apart individual risk factors for poor mental health that derive from the stress imposed by the study of medicine. This will require the collection of detailed personal information about students, not just the handing out of a tool to assess psychological distress. Therefore, trust in the investigators and assurances regarding confidentiality will be key. Researchers will need to consider carefully how to maximise not only sample size, but also response rate. Validated assessment tools for depression and anxiety will be required. We would suggest that assessments are carried out in line with at least three time-points over the course of a medical student's career in order to assess the impact of the course on the student's mental health.
Ideally, consent to follow the cohort into their professional careers would be sought at the start so that the follow-up phase could then examine the longerterm impact of poor mental health as a student on future mental health, professional attainment and involvement with the regulator.
Contributors: VH had the initial idea for the study. VH and MH worked together on the design of the study, independently analysed the data, and together interpreted the results. VH drafted the article and MH revised it critically for intellectual content. Both authors approved the final manuscript for submission.