Why has prevalence of mental disorders not decreased as treatment has increased?

Serious concerns have been voiced that prevalence of common mental disorders has not decreased in recent years (Jorm et al., 2017) and might even be increasing (Mulder et al., 2017), despite expanded provision of effective treatments.

Prescriptions of antidepressants have doubled in the high-income countries between 2000 and 2015 (Organisation for Economic Co-operation and Development [OECD], 2017). The proportion of people with diagnosable mental disorders currently receiving treatment has increased from around 12% to 20% in the 1990s in the United States, from 37% to 46% in the 2000s in Australia and from 28% to 33% between 1993 and 2007 in the United Kingdom. However, the total Years Lived with Disability (YLD) due to depressive disorders have been estimated to increase by 22% in high-income countries and by 49% across the world between 1990 and 2017 (James et al., 2018).

If these trend data are accurate and if the treatments that have increased really are effective, how can we reconcile the fact that increased treatment has not resulted in decreased prevalence? Although not providing an answer, it is relevant to note that the situation is not dissimilar for some non-communicable diseases (NCDs) other than mental disorders. For example, prescriptions of antihypertensive drugs doubled and those of cholesterol-lowering drugs quadrupled in high-income countries between 2000 and 2015 (OECD, 2017). Yet deaths due to cardiovascular diseases (CVDs) increased by 49% and YLDs associated with CVDs increased by 88% between 1900 and 2017 across the world, with the respective figures for high-income countries being −6% (i.e. a decrease in CVD deaths) and +22% (but an even higher increase in CVD-related YLD; James et al., 2018).

Do results such as these mean that treatments are failing? Several other possibilities exist

First, the apparent explosion of YLDs due to the above disorders largely disappears when the figures are adjusted for age and recalculated per population. Globally, the age-standardized YLDs due to depressive disorders have decreased by 6% and those due to CVDs by 1% between 1990 and 2017 (James et al., 2018). A systematic review of 44 studies that used the same epidemiological survey methodology in the same geographical region repeatedly has found only a modest increase in the prevalence of common mental disorders, while the prevalence of CVD has stayed virtually constant through 2000s in England and Scotland. In other words, the apparent increase in the YLDs due to these disorders was driven mainly by population growth and aging.

Second, estimation of prevalence in psychiatric epidemiological surveys is only approximate. The measurements have not focused on incidence and may not have distinguished between actively symptomatic major depression and chronic and controlled depression. Furthermore, the calculations of YLDs reported above were based on extrapolations from complex regression models applied post hoc to survey data collected for other purposes. Slight changes in the assumptions might lead to meaningful changes in these estimates. Typically, the response rates in these surveys ranged between 50% and 75%, which means that changes in the relative participation of people with depression and/or changes in the readiness of these people to admit to being depressed could have substantively affected estimates of trends in prevalence.

Third, it is important to recognize that treatment of NCDs does not reduce lifetime prevalence and might not reduce 12-month prevalence. Instead, treatment would be expected to have its greatest effect in reducing point prevalence either by shortening length of episodes in the case of episodic disorders like depression or by reducing their recurrences. Antidepressants have been found to shorten major depressive disorder (MDD) episodes by 2–14 days in comparison with placebo in randomized controlled trials. Antidepressants can halve the probability of relapse up to 3 years if these medications are taken continuously after successful acute phase treatment. While the efficacy reported in randomized controlled trials (RCTs) is likely an underestimate in comparison to the natural course, actual treatments provided in the real world are often suboptimal. For example, more often than not patients in real-world settings terminate their antidepressant treatments prematurely. It would not be surprising if the limited efficacy of the current treatments coupled with lack of optimization would not result in a meaningful decrease in the point prevalence of depression, especially if there was a small rise in prevalence in the absence of treatment. Similar calculations might apply to CVD prevention treatments, whose typical relative risk (RR) is 0.7 and whose typical event rate in the high-risk population is around 10% over a period of 5 years (and therefore, the expected reduction in incident CVD is less than 1% per year and even less for lower risk populations): even extended provision of antihypertensive or cholesterol-lowering drugs would probably not by itself significantly affect the annual prevalence of around 15%.

Assuming that the prevalence of common mental disorders is hard to reduce, would it mean that their treatments are pointless? There has been implicit support for this position on the part of some task forces based on concerns about the generally low quality of depression treatment in the primary case, the sector where the vast majority of the increase in depression treatment has occurred over the past two decades. However, this position has slowly changed over time to the point that some (Siu et al., 2016) although not all guidelines now recommend broad-based screening and primary care treatment for depression based on accumulating evidence about the effectiveness of integrated care in primary care settings where mental health care providers are co-located. Consensus has been slower in accepting the value of maintenance treatment to reduce risk of remission, although this is likely to be of great importance in reducing overall illness burden. Actuarial evidence of reduced prevalence is likely to emerge only after mental health services are scaled up to include effective use of information and communication technologies for outreach and we realize the synergistic effects of expanded screening, integration of evidence-based acute treatment into primary care and maintenance treatment.