Hide table of contents

Introduction & overview

Hi all! I'm Sam, a US-based web developer & designer. My apologies for the clickbait title, but some basic research has led me to think that insomnia is a neglected, burdensome problem on a global scale with a tractable solution. Specifically, based on assumptions laid out below, I think insomnia has a burden of ~130 million QALYs (Quality-Adjusted Life Years) annually (compare to malaria's 55 million for scale). Furthermore, I think a (technically simple) implementation of a proven therapy intervention can drastically reduce the above number. I'm building a project to attempt the very thing (www.dozy.health) and am looking for some EA aligned team members.

Insomnia annual QALY loss estimation

Estimates on the prevalence of insomnia vary widely, according to definition and methodology. According to one 2008 paper, numbers range from 10% to 40%, and "Given all the information available, the prevalence of insomnia symptoms may be estimated at 30% and specific insomnia disorders at 5–10%".

A 2018 paper estimated the QALY burden of insomnia at 5.6 million QALYs per year. This estimate only counts quality of life effects, not length of life, so the real number may well be larger than this. Its methods are not extremely robust but they end up with an insomnia prevalence of 28%, which is close to the number from the previous paper.

If we assume the QALY burden of insomnia is similar among global populations (may or may not be true), then we can expand the US number to ballpark the global annual burden of insomnia at 130,000,000 QALYs lost per year.

QALY burden in perspective

A Gates Foundation-funded report in 2015 estimated the global DALY burden of malaria at 55M QALYs per year. All cardiovascular diseases together were estimated at 347M/year, depressive disorders at 54M/year, drowning at 17M/year, and interpersonal violence at 21M/year. By these estimates, insomnia has a comparably serious burden to some major issues facing humanity.

The report doesn't have insomnia as one of their 315 evaluated conditions, so I can't get a direct number for accuracy, but I can provide some estimates. For example, the report says that depressive disorders have a burden of 55M QALYs/year. If we divide that down to the US population, that's 2.3M QALYs/year. This is more conservative than the 2018 insomnia paper, which estimates depression's impact at 4M/year in the US. Same order of magnitude, but a bit less than twice what the Gates report estimated. Their anxiety burden estimate differs by a similar amount.

If we construct an (informal, statistically invalid) confidence interval for this, it might range from 30M QALYs/year to 200M QALYs/year, again depending on definitions and methods.

Tractability of insomnia

Cognitive Behavioral Therapy for Insomnia (CBT-i) is a therapy treatment that takes 4-10 weeks, has 2-3x the impact of sleeping pills, and whose benefits are essentially permanent. We have a handful of meta-analyses to show that Cognitive Behavioral Therapy for Insomnia (CBT-i) is effective for nearly all sleep indicators - it speeds sleep onset, reduces nightly wakefulness, improves quality, reduces daytime symptoms. It works even in cases where the insomnia is comorbid with other medical or psychiatric issues.

The problem with CBT-i is it's currently inaccessible. In the entire United States, where ~60M people deal with insomnia annually, there's a total of ~250 certified providers of CBT-i. To make matters worse, many of them are researchers, or not accepting new patients, and neither patients nor doctors are aware of the treatment in the first place. This is especially shocking considering it's usually the first-line treatment recommended by relevant governing bodies. The problem is similar in other developed countries, and like other mental health treatments, completely lacking in developing countries.

Researchers who've studied the issue are excited by the potential of digital CBT-i (dCBT-i), but the issue encountered with existing solutions is a lack of personalization. It's like trying to get surgery done on yourself by taking a class on how to do surgery. Researchers suggest that further personalization options and custom treatment could result in much better retention and outcomes.

Such a digital solution would be extremely scalable, with a primarily automated treatment system supported by human attention to answer questions and reduce churn. In principle, the treatment could even be delivered over SMS.

Long story short, despite its non-communicable nature, I (and researchers in the field, plus my sleep therapist advisors) believe it's possible to dramatically reduce the DALY burden of insomnia by creating effective tools and rapidly improving them with RCT-type experiments.

Introduction to Dozy (an attempted solution)

Hence, why I'm working on my current project, Dozy. Using my existing design & development expertise, I'm creating an app that does exactly what I talked about above - automated, personalized treatment, ideally like a sleep therapist in your pocket (with human support).

Have only run a few people through my prototype so far but initial results are mostly on par with the meta-analysis on human therapists. One user went from moderate severity insomnia to no clinically significant insomnia within 6 weeks (sleep efficiency 60% > 94%, sleep duration 5.6 hours > 6.6 hours).

I'm not profit driven on this, so am looking for alternative legal structures for maximum positive impact (such as a steward ownership structure). Have been working on it unpaid full-time since May/June 2019, living extremely cheaply.

Some asks

First, is this estimate / are these assumptions reasonable to you? I don't have much formal EA research experience, so would love to know if I'm missing anything obvious.

Second, I'm actively looking for cofounders on this project to help me speed up development & iteration. If you're a mobile developer who's impact-motivated (and possibly interested in sleep science), or know anyone who fits that description, please reach out - am ready to offer an equal equity split (with cliff etc) to the right person(s).

Finally, legal structures - is a startup the best structure to scale quickly & thoroughly address the problem? Are there charity or non-profit structures that would work? I'd like to be paid eventually but will donate most of what I make anyway (already took the Founders Pledge, will likely raise my pledge later), so personal financial return is not a primary concern.


Prevalence of insomnia and insomnia burden:

2008, Insomnia: Prevalence, Impact, Pathogenesis, Differential Diagnosis, and Evaluation. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2504337/

2018, Insomnia and Impaired Quality of Life in the United States (no public full text, message me and I can send it to you). https://www.ncbi.nlm.nih.gov/pubmed/30256547

2015, Global, regional, and national disability-adjusted life-years (DALYs) for 315 diseases and injuries and healthy life expectancy (HALE), 1990–2015: a systematic analysis for the Global Burden of Disease Study 2015 https://www.ncbi.nlm.nih.gov/pubmed/27733283

Efficacy of CBT-i:

2018, Cognitive and behavioral therapies in the treatment of insomnia: A meta-analysis. https://www.sciencedirect.com/science/article/abs/pii/S1087079217300345?via%3Dihub

2015, Cognitive Behavioral Therapy for Chronic Insomnia: A Systematic Review and Meta-analysis. https://www.ncbi.nlm.nih.gov/pubmed/26054060

2015, Cognitive Behavioral Therapy for Insomnia Comorbid With Psychiatric and Medical Conditions: A Meta-analysis. https://www.ncbi.nlm.nih.gov/pubmed/26147487

2019, Digital Delivery of Cognitive Behavioral Therapy for Insomnia. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6546653/





More posts like this

Sorted by Click to highlight new comments since: Today at 11:21 AM

I have written up the instructions for CBT-i here for those interested - https://johnhalstead.org/index.php/2020/10/11/how-to-cure-your-insomnia/

Hi, thanks for this. I am personally very enthusiastic about CBT-i as it worked for me and has done for a couple of friends and the evidence seems to be reasonably good. One of the attractions of CBT-i for me was that you I don't really think you need a trained therapist or an app, you can just follow instructions in a book about sleep hygiene and sleep restriction. I did use an online course, but I don't think it was really necessary. The instructions are quite straightforward -

  • have a set waking up time. Spend the time in bed that you wish to spend sleeping (e.g. 7 hours) and then taper down time in bed by a half an hour each week until you reach the time you actually spend sleeping (e.g. 4 hours) and then ramp back up half an hour a week once the association between bed and stress is broken. If you can't sleep for an hour get out of bed.
  • Sleep hygiene also seems quite straightforward - cut out caffeine, booze and nicotine, dim the lights, don't look at your phone for tv two hours before bed, exercise. Just use bed for sex and sleep - don't read in bed.

I think you could just get this by reading the second half of Colin Espie's short book Overcoming Insomnia. Given this, what value do you think an app adds?

My favoured EA insomnia solution would be to publicise the Espie book far and wide and to get doctors to do what is required by their guidelines and recommend CBT-i, rather than just prescribing sleeping pills.

Perhaps an app is an efficient way to popularize the ideas from the book? Many people don't commonly read non-fiction.

This feels more like something that could be conveyed on 1 side of A4. Could someone create a webpage with the evidence on CBT-i and instructions on how to do it?

Interesting, makes sense! I like that suggestion.

Thanks for your reply! I agree, CBT-i is super promising and I'm all about finding effective ways to spread it.

Almost got confused as to which Overcoming Insomnia book before I added the author title - first result is a $50 therapist manual lol.

I think it's possible to treat yourself this way if you're decently analytical (as I suspect most visitors in this forum to be), but quite difficult to stick to otherwise. E.g. I can't really imagine telling my aunt with insomnia to read the book and apply the techniques to herself, both from a comprehension perspective and a habit maintenance perspective. It's kinda similar to how there's a lot of high-quality books out there on fitness, but classes or personal trainers provide more value to most people.

Regarding the CBT-i guidelines, different techniques are more or less useful in different circumstances. For instance, what you describe in the first bullet point is known as Sleep Compression Therapy - a useful technique, but it takes longer and has less empirical support than Sleep Restriction Therapy (SRT, where you immediately restrict time in bed based on a baseline of time sleeping). The slower Compression edition is helpful in cases where patients have trouble with the sudden schedule shift, so it's quite useful, but not always ideal.

Similarly, sleep hygiene education is actually significantly less impactful than other core CBT-i techniques such as Stimulus Control Therapy (SCT), which is all about turning the bedroom into a strong environmental trigger for sleep. And sometimes the above doesn't click without additional help from relaxation training (usually Progressive Muscle Relaxation) or some cognitive interventions (such as Paradoxical Intention Therapy).

It looks like Espie addresses much of these in his book, which is great, but it's harder to distill into core techniques. If one did distill into core techniques, sleep hygiene probably wouldn't be mentioned due to that being the first advice anyone with insomnia gets. The sleep therapists I've talked to usually emphasize SRT, SCT, and relaxation training where necessary (covering other bases like hygiene over time).

tl;dr: There are many potentially helpful techniques and a lot of information to consider, but it's cognitively demanding to do so and manage your own treatment. The app will take away that cognitive strain by customizing treatment and offering behavioral support along the way, so the user only sees the information they need and can reach out for support when necessary. We've all seen research on how much convenience affects behavior - I want to take this therapy and make it extremely convenient, thus improving treatment outcomes and reaching a larger audience.

Does that make sense / illustrate the value I think the app provides?

(am definitely on board with convincing more doctors to push CBT-i instead of sleeping pills - the book would be a far better option than the pills (though with more cognitive effort required)).

Really appreciate the work that you are putting into this app, and this write-up. I'm excited by your app, and hope that it will help a lot of people to solve their sleeping problems! John Halstead also wrote a post on CBT-i a while ago, and while I assume that you've reached it independently, it's great to see attempt at real-world solutions and impact assessments.

There are two points that I think are missing from your analysis. First, regarding Tractability, I'm curious as to what would cause people with insomnia to seek help and find the CBT app. That is, even if CBT is very effective, it might still be very hard to reach people and to put the treatment in practice.

Second, I'd like to see an assessment of the marginal contribution for Slumber over existing efforts. There seem to be other apps for CBT-i. 

Thanks again! I've suggested Slumber for a friend to try out :)

Hi EdoArad!

Thanks for your comment. Unfortunately there are a lot of unknowns with both of these questions but I can lay out some context of the current environment.

As things stand, a relative minority of people seek medical treatment for their insomnia. E.g. while estimates put the prevalence of insomnia itself between 10-40%, the actual number who seeks treatment and gets diagnosed is around 5-10% (mentioned in the first citation above). There may be several factors for this, but the one I've heard most commonly from user interviews is people (reasonably) don't want to go on sleeping pills, and that's what they'll get 90% of the time in most medical establishments.

So honestly, the current rate of people seeking treatment through official medical channels is bad, though seeking informal solutions is quite common and going up. The biggest example of this is the "sleep stories" and other sleep-focused features in Calm and Headspace, which have reportedly expanded to be a giant chunk of their userbase (no numbers released unfortunately). My hypothesis is that having an effective, well-reviewed app out there will keep that barrier to entry low and get a much larger percentage of the insomniac population to attempt a fix.

The other factor I'm shooting for re: awareness is word of mouth. (Nearly) very insomniac I've spoken to knows multiple others, and the users I've treated so far have both spread the word about Slumber without me asking them to. The kind of improvements that are possible in fixing insomnia have potential to create a lot of superpromoter users, who (maybe coupled with some kind of referral incentive) can spread the word more effectively than any other channel.

Regarding marginal contribution: The major existing player (Sleepio) has several key flaws that I believe keep it from being the solution we need.

1. As mentioned above, Sleepio functions more as an online class on how to administer yourself CBT-i than a therapist who administers it for you. You watch lectures, take home homework, and try to figure things out yourself for the most part (though I believe they do have a human chat available for questions). This approach has a very high churn rate (how many people start an online class and never finish?) and around 70% of the efficacy of in-person therapy. For example, on this study on dCBT-i, the churn rate was 35% (https://www.researchgate.net/publication/280584339_Predictors_of_dropout_from_internet-based_self-help_cognitive_behavioral_therapy_for_insomnia). In others, it was as high as 49%.

I'm still experimenting with the format, but my hypothesis is that flipping the perspective - creating an app that functions as a therapist (with custom treatments & decision trees), plus social measures for churn - will produce a solution that approaches or matches that of in-person therapy (churn rate of ~12%).

2. The price. Sleepio doesn't show the price of their program without digging, but a quick Google search suggests it's currently around $400. They don't seem to take insurance. This puts it out of financial reach of many, and I'd guess it gives them a *very* healthy profit margin - especially considering the percentage of users who never complete the program, and the amount that's completely automated. It's my goal with Slumber to make something substantially more affordable, probably at least half the cost for more developed countries and less for places with less purchasing power.

You'd think Sleepio would have greater market penetration than they do - the company got started in 2012 and has raised $15M+, but their product still looks like it was made in Flash player, their mobile app is broken for what seems like half of their users, and society as a whole doesn't seem more familiar with CBT-i as a result of their efforts and funds.

3. I've interviewed a few people who've gone through the Sleepio program, and my sleep therapist advisors tell me they get many patients who've tried it. Common threads seem to be it's hard to finish, it didn't work for them, or (to one person) the content was presented condescendingly, and they didn't have the flexibility to change their sleep schedule or know when they'd be off restricted sleep. So have heard a number of complains. Additionally, Sleepio doesn't have financial incentive to help a user once they've bought in - it's just the upfront fee, with no money back guarantees or anything of the sort. Those people who it didn't work for still made Sleepio money. From a profit perspective, the company seems well off.

There are few other options in this field worth mentioning. A company called Pear Sleep is developing an app from a pharma view, with the goal of getting it prescribed in doctors' offices. This is good in terms of institutional acceptance but the product is inaccessible as a regular consumer - another barrier to entry. The freely available online resources are garbage - it's necessary to dig into the literature to find guides on treatment.

Even if everything I build is only as good as what's already out there, drawing more people into treatment and offering an alternative to sleeping pills is worth a lot, given the health cost of sleeping pills (50k ER visits in the US, in one year, from one brand) and the relative lack of improvement they bring (avg 11 mins reduced sleep onset, or ~1/3 to 1/2 as effective as CBT-i). Solving the problems of awareness, growth, and scaling would make a big dent in the QALY burden.

I think Slumber can address the above by being

  • A therapist in your pocket with custom, flexible treatment plans via decision trees & human support
  • Mobile-first (Sleepio has an iOS app but it's at 2.6 stars)
  • Affordable (sub $100 at least) with subscription model, or with money-back guarantee

Hope this answers your questions, or at least the state of my current answers to them. I'm aiming to have more solid evidence re: efficacy and churn within the next few months, which should help me assess marginal impact with more confidence.

(Nearly) every insomniac I’ve spoken to knows multiple others

Just want to highlight a potential selection effect: If these people spontaneously tell you that they're insomniacs, they're the type of people who will tell other people about their insomnia, and thus get to know multiple others. There might also be silent insomniacs, who don't tell people they're insomniacs and don't know any others. You're less likely to speak with those, so it would be hard to tell how common they are.

I wrote up some thoughts on CBT-I and the evidence base behind it here.

Since this insomnia is apparently a high-impact topic, I might as well share some anecdotes from my own battle with sleep difficulties.

I've had some success with behavioral solutions to insomnia ("don't use screens after 11 PM" type stuff). But the problem with behavioral solutions, in my view, is that they are too brittle. Life always happens and your habit breaks at some point. So in the spirit of Nassim Nicholas Taleb's comments on fragility, I've instead recently focused on finding "robust" or "antifragile" solutions to the problem of getting enough sleep. These tend to be technological. Right now I'm stacking a bunch of different technologies for better sleep:

  • Ebb forehead cooler device
  • Weighted blanket
  • f.lux
  • White noise machine
  • Eye mask
  • Glycine
  • Airway expansion. Note: I haven't gotten a sleep study, and I doubt I would strictly meet the criteria for sleep apnea diagnosis, but I still seem to be benefiting a lot from this.
  • Lying on an acupressure mat. Note: I think the most common explanations for why acupuncture works are pseudoscience. I recommend this book.
  • If I have to get up in the middle of the night, I wear orange glasses to block blue light. I also colored the night lights in our house with a red marker so they emit less blue light.

It might sound like a lot, but the nightly overhead of maintaining this is not high--less than 1% of the time I spend asleep. In aggregate this all seems to improve my sleep considerably in a way that doesn't depend on fragile behavioral interventions. (Some of the most valuable-seeming additions have been pretty recent, so we'll see how things work long term.)

Note: I suspect my sleep problems are more "physiological" than "psychological" in nature. CBT-i might work better for someone whose problems are more psychological.

Hi. Did you hear about "Dream" by Panda company?
Dreem has developed a combined hardware/software solution to
It is a wearable headband that is an FDA class II medical device and is equipped with five EEG sensors, one pulse oximeter, and one accelerometer that all work together to measure sleep activity and various stages of sleep. The device also delivers sound vibrations through bone conduction to ease patients to sleep, and a digital app guides patients through a coaching program inspired by cognitive behavioral therapy for insomnia (CBTI)

The company has raised $60 million from investors so I'm sure you will find confederates


For what it's worth, I tried CBT-I and it wasn't successful for me, then tried ACT for insomnia which solved the issue. Specifically via a book called "The Sleep Book" by Guy Meadows. Highly recommended.

Thanks for sharing - can I ask in what form you tried CBT-i? (e.g. with an in-person therapist, online course, book etc).

Have run into ACT for insomnia in an article or two, but haven't met anyone like you who's used it. Have downloaded the book and will give it a read!

Tried in an online course. Hope the book helps

This analysis accurately frames my starting point in working on Dozy. It convinced me to commit full-time to the project on the merits of its potential impact. Some of the awareness the post generated led to me raising a small pre-seed round from EAs, which has been invaluable. The problem size and treatment claims have been largely substantiated by studies that came out since, but I underestimated the challenges of getting users through the complete treatment program. Also, there are a few direct-to-consumer CBT-i options coming out as of now, so the counterfactual impact is more limited than I initially suggested.

That said, I still think startups / new projects represent one of the best ways to rapidly explore solutions in problem spaces like this.

Re: followup work, I'd love to see more problem analyses - especially in the mental health space - as it provides inspiration and launch support for new projects. Large-scale interventions in sleep health could be incredibly worthwhile if we can find them - insomnia continues to get more prevalent, along with other mental health related suffering. I don't know how to reverse this trend and am looking for ideas.  :)