There has been increased focus on mental health as an area where effective interventions could create a lot of good. Much of this work has focused on scaling interventions such as psychiatric medications and psychotherapy. Often this work thinks creatively about how to promote mental health in under-resourced regions, for example through peer-led psychotherapy interventions. But my feeling is that there is a lot of room for thinking more broadly about what kinds of interventions might be useful.

To some this will suggest distinctively 21st-century interventions, such as the use of AI as a kind of adjunctive psychotherapy. However, this post is about a distinctively older form of technology that is I think inexpensive and underused, both in the United States and internationally. This is the use of books as adjuncts or supplements for traditional psychotherapy, or what is often called bibliotherapy.

I have written on bibliotherapy in various forums, including Psychology Today and my Substack. I am posting on it here because my sense is that this is a rich area for people in EA – it is an area where there are many low-hanging fruit, and where relatively small interventions can yield substantive benefits. I am by training a philosopher and a clinical social worker, and many of the particular interests and skills of people within EA are less familiar to me, so I would very much welcome input on how to frame these issues in a way that will be useful to the EA community (or whether, indeed, that is something worth doing at all).

This post four parts: (i) a brief summary of the evidence for the effectiveness of bibliotherapy, (ii) an estimate of what a large scale bibliotherapy intervention might cost, (iii) the potential effects of such an intervention and (iv) responses to some questions/objections.

The Evidence

One of the better attested results in the literature is that giving a person a book on a mental health topic – Feeling Good, the best-selling primer on applying Cognitive Behavioral Therapy, is a paradigm of the kind of book to be thinking of – is as effective as having them meet regularly with a therapist. 

A 1995 meta-analysis, for example, shows significant effect sizes across a broad range of mental health concerns and crucially finds "There was no significant differences between the effects of bibliotherapy and therapist-administered treatments, as well as no significant erosion of effect sizes at follow-up."

This finding has been pretty robust – there are studies of bibliotherapy for not just for depression and anxiety but for sexual dysfunction and disordered eating as well, as well as bibliotherapy for Chinese-speakers, older adults, and many other populations. To my knowledge every major study tends to find that books perform more or less as well as therapists (and, btw, that therapists and books both have meaningful positive effects).

To my mind this is one of the most important findings in 20th-century psychology, and most people don't know about it. It's an interesting theoretical question how bibliotherapy works – what exactly is the mechanism, why does the therapist contribution seem to be epiphenomenal, etc. – but the main point for this post is that it does.

The Cost

What would it cost to implement a large-scale bibliotherapy intervention? Let's consider a back-of-the-envelope calculation for addressing depression in the United States.

The National Institutes for Mental Health reports that, in 2021, 21.0 million American adults experienced at least one depressive episode. This number is drawn from the National Survey of Drug Use and Health, and more recent years indicate similar or increased numbers. Nonetheless, 21.0 million can serve as a reasonable baseline.

What is the cost of intervening on this population. Well, on Amazon, a paperback copy of Feeling Good costs $14.93. If we wanted to implement bibliotherapy at scale, there would be additional costs (such as shipping) as well as countervailing discounts (such as bulk orders). But $15/per person is a reasonable estimate for providing bibliotherapy to a given individual in the United States.

How much then would it cost to provide bibliotherapy to every person in the United States with depression? There are 21 million people, and a book costs $15, so the cost is roughly 21 million x $15 = 315 million USD.

That is a lot of money. It is useful, however, to contrast it with the cost of some other major mental health interventions. This is roughly on the order of magnitude of running a nationwide suicide hotline. It is estimated by that to a suicide hotline cost approximately $82 per call (this is call center costs only, excluding referrals to emergency services). Estimates vary, but there are plausibly at least 5 millions calls to suicide hotlines per year. So the operational cost of suicide hotlines in the United States is at least 410 million USD per year.

Suicide hotlines are of course very important and this is not an argument against them. Rather it is an argument that even this very extreme form of bibliotherapy – simply sending a book to everyone with depression – is well within the range of costs for mental health interventions that we are doing already. 

The Effects

What would the effects of such an intervention be? This part of the case becomes necessarily more speculative, but there is a reasonable case that this intervention would (i) have a meaningful impact on depression and (ii) it would, in virtue of (i), have a substantive effect on overall well-being – indeed, more than an unconditional cash transfer.

The case for (i) involves a few steps. First, as indicated above, multiple studies show that bibliotherapy has a significant effect on various mental health conditions. For depression specifically, a 2004 meta-analysis indicates an effect size of d = 0.77, and other studies have tended to find similar values.

What does that mean in terms of actual impact on depression? Since d is defined in terms of standard deviations, this means a 0.77 standard deviation improvement in depression. What exactly this looks like will depend on which depression scale we use. If we use the PHQ-9 – not the definitive depression scale but one that many will be familiar with from primary care and other settings – this translates to about 3 to 5 point decrease in PHQ-9 score (since the standard devation of the PHQ-9 appears to range from around 4 to 7 depending on the population in question). 

What effect would this have on overall well-being? I'm not aware of a specific answer to this question, and this gets us into some philosophical questions about the relationship between mental health and well-being. Studies do confirm the intuitive result that your mental health makes a big difference to your overall well-being. At first approximation, let's assume that a 1 standard deviation change in mental health is associated with a 0.5 standard deviation change in mental health. (If that is too high, then all the results below can be scaled down accordingly – as we'll see, the impact is so large that it remains significant under much weaker assumptions).

On this assumption, we might expect a 0.77 SD improvement in mental health to yield a 0.39 SD improvement in overall well-being. What this looks like will depend on our measure for well-being. If we use a standard 10-point scale of subjective well-being, which is known to have an SD of around 2, this will be about an 0.77 point change in reported life satisfaction. Very roughly, someone who rated their life around a 4/10 would, under this intervention, be expected to rate it around a 5/10.

That is a huge change. It is in the neighborhood of the changes that been reported, for example, in basic income experiments. One notable experiment in Germany provides a basic income to individuals and found an 0.417 SD improvement in life-satisfaction. But this program send 1200 Euros per month to individuals for 36 months, at a cost of approximately $50000 per person. In contrast, as we have said, the present intervention costs approximately $15 per person.  Which is to say that the bibliotherapy intervention is roughly 3000x more effective, on a per dollar basis.

There are enough conjectures here that we should have low confidence in any particular number. In particular, this reasoning relies on an assumption about the relationship between mental health and well-being, thought this assumption is I think plausible and directionally correct. But this considerations tell in favor of the view that a large-scale bibliotherapy intervention could make a meaningful difference in the lives on many people, and that it could do so at a level of cost-effectiveness that is basically unprecedented in this cause area.

Questions/Objections

Wait, how would you know who does or doesn't have depression?

You don't, and indeed shouldn't – that is private information. The program sketched above is one that cannot be implemented, and shouldn't be for general ethical reasons. It should really be thought of as a thought experiment: if this could be done, it would make a massive difference. So we should think about more selective and ethically constrained versions of doing the same thing.

What would this look like? At first pass, one intervention would allow people to opt in. Anyone who self-identifies as needing bibliotherapy would be sent a free book. This would lead to some false positives, as well as some opportunities for graft (e.g. reselling books), but in general it does not seem that these would swamp the significant benefits outlined above.

What if people don't read the books?

This is a fair concern. Bibliotherapy is a low-monitoring intervention generally, but this would be super low monitoring. Many people would probably just throw the books away, or be puzzled. So this would significantly reduce the impact of the program.

A couple of points, however.

First, the books will be so effective for the people who do read the books that the low usage will be worth it. If for example 10% of people actually read the books, then this would still be a hugely impactful program. The strategy of just giving people things and letting them use or not use them as they will is familiar in other areas of development. People are after all generally interested in using resources that will help them, so simply providing those resources unconditionally is often the most reasonable strategy.

Second, the more real-world version of this program – the opt in version – would be subject to this problem to a significantly reduced degree, on the supposition that people are more like to read books that they asked for.

How does this compare to other interventions, especially other mental health interventions?

That's a great question. The comparison to basic income interventions above is just one case. There are lots of interventions, and it is reasonable to ask how bibliotherapy compares to these.

There is a large philosophical question here – how do we compare mental health interventions (such as psychotherapy or bibliotherapy) to non-mental health interventions (such as bednets or unconditional cash transfers)? The Happier Lives Institute has done a lot of work on this question. There are also empirical questions, which are briefly but not fully addressed in the cost section above.

In lieu of an answer, here is a conjecture: a targeted bibliotherapy program would be at least as effective as any other mental health intervention. The reasoning behind this is the extremely low-cost of bibliotherapy  compared to other mental health interventions (which often involve training and other resource-intensive activities), and its demonstrated high effectiveness. But, as I say, this remains a conjecture.

What about people who can't read? What about people outside of the US?

These are obviously separate questions. Let us take them in turn.

First, literacy. It is easy to exaggerate the scale of the problem. The global adult literacy rate is about 86%, and even the region with the lowest literacy rate (sub-Saharan Africa) has an adult literacy rate of about 68%. While many people cannot read, and that is a major development concern, bibliotherapy would still be an appropriate intervention for a signficant majority of the adult population anywhere in the world.

It is true that bibliotherapy might take different forms in different places. Feeling Good has a distinctively developed-world frame of reference, and it might well be that other texts would be more appropriate for other places. But literacy per se does not pose an obstacle to the widescale dissemination of bibliotherapy, even in places with relatively low literacy rates.

That brings us to the question of place. My initial example focuses on the United States, partly because we have good data and relatively efficient modes of book delivery (such as Amazon). But ultimately I think the case for bibliotherapy is strongest and most urgent in other places.

Consider, for example, India. The WHO estimates that 1 in 20 people in India (roughly 70 million people) require treatment for depression. Yet India's mental health facilities are significantly under-resourced: there are 7.5 psychiatrists for every million people in India, as opposed to well over 100 psychiatrists per million in the United States. There are amazing NGO's, such as Sangath, working to address these facts, but bibliotherapy is potentially a meaningful addition to this work. (The adult literacy rate in India, incidentally, is estimated to be around 75%). Similar considerations apply to many countries in South Asia and sub-Saharan Africa. Though my thought experiment concerned the US, I think it is these countries where bibliotherapy could have the most significant impact.

(Image Credit: Eugenio Mazzone, Unsplash)