Treatment Access Isn't the Main Issue for Mental Health

A look at the evidence

Photo by Nik Shuliahin

At this point, the global mental health crisis is both well established and well known. According to the World Health Organization, 970 million people worldwide are experiencing some form of mental disorder, making it the most common form of disability. And the numbers are rising. In Canada, major depressive episodes, bipolar disorder, and generalized anxiety disorder have become more common in the past decade.

The playbook for addressing this crisis is also well established. According to Statistics Canada, only half the people who meet the diagnostic criteria for a mood, anxiety, or substance use disorder get professional care, and one-third of people meeting the criteria report having unmet mental healthcare needs. Therefore, if access to treatment increases, people will get better.

This is true, but it misses an important point. For most mental disorders, the evidence consistently shows that treatments have little or no effect for many people.

Let’s focus on depression and anxiety, the two most common mental disorders. A meta-analysis of psychotherapies for depression found that most treatments are equally effective (which is weird). However, the study found that over half of patients experienced little or no benefit, and only a third experienced a reduction in symptoms to the point where they no longer met the criteria for depression. A study of cognitive behavioural therapy for anxiety disorders also found that around half of patients had no response.

Another study, this time an ‘umbrella review’—a review of meta-analyses—reviewed the efficacy of psychotherapies and pharmacotherapies for multiple mental disorders. The studies it included had over 600,000 participants in total. Studies of studies usually report their findings by giving the standardized mean difference (SMD, also known as Cohen’s d), which is a way of measuring effect size when studies use different scales and measures. Typically, 0.2–0.5 is considered a small effect, 0.5–0.8 a medium effect, and above 0.8 a large effect. 

I find it more helpful to use something called the number needed to treat (NNT). NNT estimates the number of patients who need to receive a treatment in order for one patient to get the desired effect. An NNT of 2 means that, for every two patients who receive the treatment, one will benefit. An NNT of 10 means that ten patients have to receive the treatment for one to benefit, or, for every patient who receives the treatment, there’s a ten percent chance the treatment will work. SMD can be converted into NNT, such as in the following table.¹

From “Size of Treatment Effects and Their Importance to Clinical Research and Practice” by Helena Chmura Kraemer and David J. Kupfer.

The umbrella review finds that ketamine is very effective at treating depression in the first few days after treatment. But the SMD for psychotherapies is 0.31 and the SMD for pharmacotherapy ranges from 0.19 to 0.41, depending on the study. This means that between six and nine people need to get treated for one person to benefit.

Some disorders have better treatments than depression and anxiety. Obsessive-compulsive disorder (SMD of 1.03; NNT of ~2) and post-traumatic stress disorder (SMD of 0.54; NNT of ~3) are examples. But others do worse. I wrote about anorexia in a previous article. The umbrella review found an SMD of 0.14, which means that around fifteen people need to be treated for one to benefit.

Summarizing their findings, the authors say the following:

In summary, a systematic re-assessment of recent evidence across multiple meta-analyses on key mental disorders provided an overarching picture of limited additional gain for both psychotherapies and pharmacotherapies over placebo or [treatment as usual]. A ceiling seems to have been reached with response rates ≤50% and most SMDs not exceeding 0.30-0.40. Thus, after more than half a century of research, thousands of RCTs and millions of invested funds, the “trillion-dollar brain drain” associated with mental disorders is presently not sufficiently addressed by the available treatments. 

Since the SMDs are above zero, most interventions have some effect. For some people, therapy or medication can change their lives, including a third of people who no longer experience symptoms of depression. So the point isn’t that mental health interventions don’t work.

Relative to some other common treatments, mental health interventions do pretty well. For instance, 200 million people with heart disease are taking statins, but the NNT is 83, meaning ninety-six percent of them get no benefit.

Still, for mental health, it’s important to realize what the numbers are saying. Treatment access is an issue, but the low effectiveness of treatment is a bigger problem. This has implications for every policy issue involving mental health. So much of the discourse about medical assistance in dying for mental disorders, for example, involves the Better World Argument: since so many people lack access to mental healthcare, we should delay MAID until that access improves.

We should reject this argument for three reasons. First, postponing MAID won’t cause better mental health access on its own, and second, postponement would mean that people with intolerable suffering will be forced to wait for treatment they might never get. 

But suppose that mental health access becomes available to everyone, so someone can get treatment whenever they want without delay. The third reason to allow MAID is that, even then, across all mental disorders, the aggregated data of psychotherapies and pharmacotherapies produces an SMD of 0.35, which means only one in five people would benefit. This is better than nothing, but framing the mental health crisis as a healthcare access problem is missing a larger component.

According to the authors of the umbrella review, what we really need is a paradigm shift in research. The treatments just don’t work that well. Unfortunately, this isn’t for a lack of trying. Thousands of clinical trials have been conducted and millions of dollars have been spent researching mental disorders. It’s a hard problem to solve. Therapy and medications can help, and it turns out that therapists are more effective when they’re less confident. But therapy and medication both have downsides. Around twenty percent of patients who use cognitive behaviour therapy have negative side effects, for instance, and the negative effects of SSRIs and other medications are well known.

We need to hold two thoughts in our head at once. Since mental healthcare works for some people, improving access to it will help. But when the odds of treatment success are so low, even perfect access still won’t be that good. For that, we need better treatments.

---

Quick Hits

• Irish Committee to Recommend Laws for Assisted Dying. Brilliant work by my Irish colleagues.

• My IVF Years

• When Therapy Makes It Worse

• A Boring Capital for a Young Democracy. Just the Way Residents Like It. I lived in a town near Belmopan, the capital of Belize, and I can confirm that it’s boring. I don’t know if the restaurant is still there, but it had excellent Chinese food.

• Blueprint for More and Better Housing. Canadian housing policy is increasingly moving in the right direction.

• Judge Denies MAID Stay on Montreal Centre. Let’s do this nationally next.

1. When the SMD is zero, the NNT becomes infinite, so there’s a large discontinuity. Because of this, some statisticians recommend against using NNT for meta-analysis. But they also made a table showing the conversion, thereby demonstrating that harm reduction works in many situations.