A Sanity Test for talking about Mental Health
Researching mental health is extremely difficult. You can’t see inside people’s heads or at least nowhere near well enough. The cutting edge of neuroscience is still completely blunt when it comes to understanding mental health . You can’t use a brain scan to diagnose a single mental health condition, not even a degenerative neurological disease like dementia.
All psychological research suffers from methodological limitations and flaws which are immediately obvious and very serious. That’s why the physical sciences have always regarded the social sciences as their poor relation. Those difficulties get more severe the further you move beyond the behavioural and try to infer about something as private and subjective as states of mind. And the problems are magnified when you attempt to impose experimental conditions and variables. That’s why the conclusions from experimental research in mental health are invariably expressed to be provisional and tentative. It’s also why very often they’re also inconsistent with each other or directly contradictory.
Given these hurdles, the research that produces the most robust data is the simplest: direct observation of outcomes, without any experimental interference. There’s less to go wrong if you’re just watching what’s happening in the real world and counting. Unfortunately, in the blur of confusion and controversy and consensus around mental health, it’s very easy to lose sight of those more solid data points.
One of the most valuable data sources in the UK is in the process of being refreshed. The Adult Psychiatric Morbidity Survey (APMS) is a survey of households in England conducted every seven years since 1993, most recently in 2014[1]. It provides cross-sectional analysis (who’s suffering from what), on a number of different bases, of a large population sample, it allows us to see trends over time and it uses clinical assessment – consistent use of the same validated diagnostic tools which map directly onto recognised psychiatric classifications – whereas most surveys, and much of the experimental research, rely on direct self-reporting and improvised questionnaires, which aren’t used consistently and haven’t been validated in the same way or at all.
So, as we wait for the results of the new APMS, this seems like a good time to revisit what previous versions have told us and what else we’ve learned from similar sources – the tyres that have been kicked that bit harder. It is a tragedy that so much evidence around mental health is so incoherent but it does provide an unusual opportunity, which is to try to set down in one place quite a lot of what there is that we can more responsibly claim to ‘know’ about this subject. And that offers a useful resource - a small core of more credible, harder data, against which to test what you hear about mental health. Here are some headlines.
The great majority of people will meet the diagnostic criteria for a mental disorder at some point in their lives. 86% of participants in the Dunedin birth cohort study[2] had met the criteria for a mental disorder by age 45.[3] Dunedin is a longitudinal study more than 50 years old in which participants undergo a series of tests and examinations at regular stages through life. It’s a narrow (predominantly white) population sample of around a thousand from New Zealand but Dunedin is unique in supplying the levels of measurable, objective data it does for the length of time it has. Like the APMS, it uses clinical assessment of mental health.
Around 1 in 6 adults suffer from a common mental disorder at any point in time.16% of the 2014 APMS met the criteria for a common mental disorder (different forms of anxiety and depression) during the week before the survey. Many of these will also meet the criteria for other psychiatric diagnosis. There is no sensible basis for the 1-in-4 have a mental illness statistic widely quoted in the UK.
Treatment for mental health doesn’t usually work. The NHS publishes recovery rates for patients suffering from depression or anxiety who access talking therapies (many of whom will also be on medication and/or accessing other interventions). Around half those completing a course of therapy in 2021 were considered to have recovered[4]. That’s the figure you will normally see reported and it’s not a particularly encouraging statistic. But as around half those who start treatment drop out, in terms of recovery from illness, NHS treatment actually doesn’t work for around three quarters of the people who get it.[5]
The NHS reports measurable improvement (where symptoms have improved but recovery hasn’t been achieved) for around two-thirds of patients.[6] But, again, these are patients who have completed the treatment; the half who dropped out are excluded from the calculation.
And where figures do show recovery or improvement, you can’t be certain it was the treatment that was responsible. Mental health fluctuates and the statistical principle of regression to the mean dictates that people will present for treatment when they are experiencing their most acute states of mind and in future tests might be expected to have moved closer to their more normal range of experience. If someone recovers you can’t be sure it was because the treatment worked; if they don’t recover, you know the treatment didn’t work. And if they drop out, you can be pretty sure the treatment didn’t work.
Most people who meet the criteria for a common mental disorder don’t need treatment. More than 60% of 2014 APMS participants who met the criteria for a common mental disorder weren’t receiving any treatment.[7] That’s often taken as evidence of a large unmet need for treatment but since less than 2% of APMS participants reported having asked for treatment and not received it, the evidence is that the great majority of the people who meet the criteria for a common mental disorder and don’t get treatment, don’t want it. And that’s 60% of the people who meet the criteria for a diagnosis.
Getting on for 20% of the UK adult population are receiving mental health treatment. 13% of 2014 APMS participants were receiving some form of mental health treatment. For nearly 12% this included medication. But here we can be confident the APMS figures are an underestimate because there is a more direct reading of raw data available. Government figures stated that 17% of the adult population were prescribed antidepressants in 2017-2018.[8]
The mis-match is likely to be because the APMS only covers private households which excludes a number of groups (the homeless, people in institutional care and the prison system and students) who show significantly higher rates of mental health issues.
Many of them don’t meet the criteria for a mental disorder. 8% of 2014 APMS participants who didn’t meet the criteria for a common mental disorder were receiving mental health treatment. Much of that is likely to be ‘maintenance’ dosage of antidepressants and other medication prescribed to prevent relapse following earlier episodes of mental health problems.
Around 1 in 8 children suffer from a mental disorder at any point in time. 13% of 5 to 19 years old met the criteria for a mental disorder when assessed in the Mental Health of Children and Young People Survey 2017 (the MHCYP is the nearest equivalent to the APMS for children).[9]
Teenage and adolescence are particularly vulnerable times for mental health. 59% of Dunedin participants had met criteria for the initial onset of a disorder by 18.[10]
Overall, mental health (measured by clinical assessment) isn’t getting worse. The APMS shows a modest rise (14% to 17%) in the incidence of common mental disorder in the adult population between 1993 and 2014 (with the biggest rise between 1993 and 2000). It’s a similar picture with the MHCYP, which showed mental disorders amongst 5-15 years olds increasing from 10% in 1999 to 11% in 2017.
Two follow up surveys to the MHCYP to assess the effect of the pandemic have reported a significant deterioration in children’s mental health – a rise in ‘probable mental disorders’ from 11% in the 2017 MHCYP to 17% in 2020 and 2021. But since these studies relied on a brief behavioural screening questionnaire, rather than the clinical assessment used in the MHCYP (and involved a much smaller sample), their results can’t be directly compared with those from the MHCYP. And since the reported spike in rates in 2020 coincided with the change in reporting basis and rates haven’t changed since, you can’t exclude the change in reporting basis being the factor behind the increase, rather than any change in the underlying patterns of mental health.
There is one significant exception to the lack of increase in the prevalence of mental disorder in the APMS: rates of common mental disorder amongst young women aged 16-24 have increased from 19% in 1993 to 26% in 2014.
But treatment levels and reported mental health issues have increased dramatically. Antidepressant prescriptions in England doubled between 2008 and 2018.[11] A 2015 survey from the NUS reported that 78% of students said they had experienced mental health issues in the previous year.[12] ONS figures showed a doubling in self-reported symptoms of depression over the pandemic, from 10% of the adult population in March 2020 to 21% in March 2021.[13]
These and many similar statistics have led to widespread reporting in the media of a mental health epidemic. But, when you apply clinical assessment to large randomized samples of the population in the APMS and MHCYP, the results don’t show the same, or any, increases in mental health issues over a period of around 20 years.
This mis-match between self-report and treatment levels, on one hand, and clinical measurement of mental illness and psychological distress, on the other, has been explored and confirmed in research. A 2018 review of different surveys of the mental health of children and young people over 20 years found consistent increases in self-reported problems (a six-times increase in the case of England) but no increase in the results from clinical questionnaires.[14] Similar mis-matches have been investigated in the US.[15]
This gap between reported experience and clinical assessment is one of the factors behind concern about the ‘medicalisation’ of normal experience.
******************
Put some of these findings together and some reasonably clear pictures start to emerge:
You will almost certainly meet the criteria for a psychiatric diagnosis at some time but it’s more likely than not that you won’t need any treatment
If you have 12 adults in a room you might expect two to meet the criteria for a mental disorder
And you might expect two to be on antidepressants
It’s more likely than not that at least one of the people with a mental health condition wouldn’t be receiving treatment
And there is a very good chance that at least one of the people on antidepressants wouldn’t meet the criteria for diagnosis of a mental health condition
If both people with a mental health condition were receiving treatment, it is unlikely that either would recover by the end of the treatment.
You might conclude from this that our definitions of ‘ill’ aren’t working well: more than half the people we say are ill don’t appear to need treatment and a similar proportion of the adult population are receiving treatment when they don’t fit our definition of ill.
And you might conclude that we’re not very good at treating mental health, since when we say people are ill and treat them, only 1-in-4 get better and only around 1-in-3 derive any measurable benefit.
Those would be very important conclusions to draw which would powerfully affect your perspective on mental health and mental healthcare. But this isn’t about what to think about mental health so much as what to think with.
There can’t be many subjects we devote so much resource and attention to and talk so badly about. Sources like the APMS provide a rare opportunity to find some clarity in those discussions. The information they provide is basic and limited and there’s a lot we need to know that they can’t tell us. But that information is solid and it gives us something against which we can test everything else people want to tell us about mental health. It will take a couple of years to grind the figures for the 2022 APMS before its published. We should look forward to it; we’ve never needed this kind of information more.
[1] https://digital.nhs.uk/data-and-information/publications/statistical/adult-psychiatric-morbidity-survey/adult-psychiatric-morbidity-survey-survey-of-mental-health-and-wellbeing-england-2014
[2] https://dunedinstudy.otago.ac.nz/
[3] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7175086/
[4] https://digital.nhs.uk/news/2021/new-statistics-released-on-talking-therapies-in-england
[5] https://digital.nhs.uk/data-and-information/publications/statistical/psychological-therapies-report-on-the-use-of-iapt-services/january-2022-final-including-a-report-on-the-iapt-employment-advisers-pilot/activity
[6] https://digital.nhs.uk/data-and-information/publications/statistical/psychological-therapies-report-on-the-use-of-iapt-services/december-2021-final-including-a-report-on-the-iapt-employment-advisers-pilot-and-quarter-3-2021-22-data/outcomes
[7] https://digital.nhs.uk/data-and-information/publications/statistical/adult-psychiatric-morbidity-survey/adult-psychiatric-morbidity-survey-survey-of-mental-health-and-wellbeing-england-2014
[8] https://www.gov.uk/government/publications/prescribed-medicines-review-report/prescribed-medicines-review-summary
[9] https://digital.nhs.uk/data-and-information/publications/statistical/mental-health-of-children-and-young-people-in-england/2017/2017
[10] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7175086/
[11] https://www.theguardian.com/society/2019/mar/29/antidepressant-prescriptions-in-england-double-in-a-decade
[12] https://www.theguardian.com/education/2015/dec/14/majority-of-students-experience-mental-health-issues-says-nus-survey
[13] https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/articles/coronavirusanddepressioninadultsgreatbritain/julytoaugust2021
[14] https://www.cambridge.org/core/services/aop-cambridge-core/content/view/AB71DE760C0027EDC5F5CF0AF507FD1B/S0033291718001757a.pdf/mental-health-and-well-being-trends-among-children-and-young-people-in-the-uk-19952014-analysis-of-repeated-cross-sectional-national-health-surveys.pdf
[15] https://journals.sagepub.com/doi/full/10.1177/0022146520984136?s=03