Discussion

This is the first study to demonstrate a consistent underrepresentation of mental health compared with three major areas of physical health in the publication of international, high-quality research in general medical journals over a decade. Psychiatry exhibited the greatest burden of disease, and consistently the fewest articles published of the four medical disciplines. This disparity may be most pronounced for original research articles, which perhaps highlights the problem further. Whereas the other three disciplines were overrepresented based on the burden estimate, cardiology was reliably the closest to parity. Within mental health, results showed fairly stable trends across the 10-year period, with dementia and psychotic disorders over-represented and those in the ‘other disorders’ category under-reported. There are indications that mood disorders (which represent the most burdensome category within mental health) have become less well represented in recent years. Publication rates of mental health-focused research varied between journals, and this may reflect our choice of publication sources: although this is the general medical category, some of these are journals specifically communicating internal medicine (of which mental health is not the focus). Despite this, the internal medicine journals we investigated did not demonstrate lower proportions of psychiatry articles than the other, more general journals.

Previous reports have demonstrated that low- and middle-income countries produce substantially less psychiatric research than Europe and America,^{Reference Patel and Sumathipala5} thus the research published may also demonstrate ethnocentricity. Less funding for psychiatry than for other disciplines has been reported,^{Reference Byrne6} and this is probably a contributory factor to the reduced research output that we have identified. Low numbers of psychiatric researchers may be partly a consequence of funding scarcity, or of social stigma (which appears to contribute to an overall unpopular perspective of psychiatry within the medical profession).^{Reference Lyons7,Reference Korszun, Dharmaindra, Koravangattu and Bhui8} Another putative contributor to low parity of publication is a reduced willingness of high-impact journals to publish psychiatric articles, perhaps derived from stigma-related factors. Psychiatric research might be further complicated by its nature: an absence of clear biological markers of psychiatric disorders has led to the use of a criteria-based diagnostic system, which gives rise to challenges including the comparability of symptom ratings and diagnoses. This often leads to increased heterogeneity in research findings and reduced statistical significance of results, and ultimately can succumb to publication bias.^{Reference Gilbody and Song9} The inability to make consistently accurate diagnoses in psychiatry could also perpetuate stigma across medical and academic professions.^{Reference Stuart, Sartorius and Liinamaa10} Stigma within mental health itself may impede researcher willingness to submit papers to high-impact general medical journals (if mental health is perceived to be too niche an interest), and there may be a higher frequency of publication opportunities in psychiatry-specific journals than is the case in other areas of medicine. Methodological challenges encountered with much psychiatric research can result in underpowered studies which are less likely to be accepted for publication in high-impact research journals. It is possible that (whatever the causes for low parity of publication), the underrepresentation of psychiatric research may self-perpetuate owing to ongoing maintaining factors.

Within psychiatry, the category of ‘other disorders’ was the most underrepresented. This group includes personality disorders, eating disorders and developmental disorders, as well as a large number of conditions that are generally less well understood and/or prevalent, which we were not able to analyse. It must be noted that articles pertaining to more than one psychiatry category were also categorised in the ‘other disorders’ group, which only serves to magnify our finding, and it is essential that the specific diagnoses within this group are addressed in future investigations. Compared with other areas of psychiatry, anxiety disorders do not appear to require increased prioritisation, as has been suggested by Christensen et al.^{Reference Christensen, Batterham, Griffiths, Gosling and Hehir11} Their article also documented the recently increased funding in Australia in areas such as psychosis and dementia, both of which we found to be over-represented. Psychotic disorders have notoriously been neglected in research in the past, and dementia has well-documented projections of burden increases in the future, which may explain the above phenomena. Future methods for burden calculations of disease may yield greater parity within psychiatry if accounting for projected change in illness burden (although the measurement of this presents a challenge). Alternatively, the marked frequency of dementia articles compared with other areas within psychiatry may denote a different kind of burden, as dementia has a high contribution to mortality in terms of years of life lost, rather than DALYs.^{Reference Prince, Patel, Saxena, Maj, Maselko and Phillips3} This could reflect the priority given to early mortality rather than suffering with a debilitating condition, which Prince et al speculates is typically prioritised in medicine. We note that our use of DALYs to approximate burden might have been altered if mortality had been prioritised. However, the Global Burden of Disease Study in 2010 implied that the values reported by Prince et al in 2007 might be an underestimation of the true burden of mental illness.^{Reference Vos, Flaxman, Naghavi, Lozano, Michaud and Ezzati12}