Several people in the comments pointed out existing lower-cost CBT-i apps! This was news to me - I’d searched pretty comprehensively and hadn’t found any besides the VA’s CBT-i Coach, which is not intended for individual use. They were:

1: Night Owl, seems good but only available for iOS

2: Sleepedy is more of a service than an app, and involves consultations and coaches. When I tried to sign up, it made me take an annoying quiz, click through a bunch of testimonials, and then finally gave me a “Schedule your free call today!” page. Still probably less annoying than seeing an in-person therapist, I guess, and $29/month.

3: Dozy, seems potentially good but still in private beta. Will probably launch in a few months; expected $10 - $30 price point.

Someone mentioned the founder of Dozy was an effective altruist and connected to me through the social graph, so I reached out. He says he’s a CS student who dropped out to work on “creat[ing] more accessible & impactful mental health treatments, with insomnia as a starting point”. He writes that he’s looking for potential co-founders, fundraising help, and advisors in the field. If you’re interested, please contact him at sam@dozy.health.

(I know some of you don’t like when I advertise things on here; feel free to tell me in the comments whether or not you think this example was appropriate)

Honorary mention to Say Good Night To Insomnia, which is a CBT-i book rather than an app; it got rave reviews from several commenters. It’s written by a world expert and probably the most trustworthy source here.

Moving on to other comments, Bugmaster asks:

What does the app do, exactly, and how effective is it ? You say it works as well as a therapist, but how well does that work ? It must have went through the FDA approval process, so there must be public records of the studies they ran, right ? On the one hand, I am somewhat skeptical that an app can fully replace personal instruction – not to mention, prescription medicine!

There are a lot of studies here, but I’m going to choose two kind of random ones. Taylor et al show that in-person cognitive behavioral therapy for insomnia has an effect size of 0.98, and the same therapy delivered over the Internet has an effect size of 0.51 (both numbers significantly different from control, not significantly different from each other). Somryst itself has significantly outperformed placebo in several studies. A meta-analysis finds that “Low to moderate grade evidence suggests CBT-I has superior effectiveness to benzodiazepine and non-benzodiazepine drugs in the long term, while very low grade evidence suggests benzodiazepines are more effective in the short term”.

In other words, the therapy works, it works at least as well as medications, and online therapy might or might not work quite as well as in-person but still works pretty well.

According to UpToDate ($) “CBT for insomnia (CBT-I) is preferred as first-line therapy for chronic insomnia in most patients. Several prospective studies have assessed the comparative efficacy of CBT-I, medications, and combination therapy as an initial approach. In short-term trials, CBT-I alone and CBT-I in combination with medications demonstrate relatively equivalent outcomes, and both are superior to medication alone.”

Argos writes:

You blame the US health care system, and the people who chose not to make 10 dollar apps. In that vein, you should probably also blame consumers. It is extremely difficult to sell apps in the range of 10 dollars, especially for health apps. Companies have tried with diabetes apps, digital health companions, etc, and mostly failed. You can offer a trial period, but since the benefit of CBT-i only accrues over time, it does not lead to much higher sales.

And Freddie deBoer adds:

Yes, the app economy writ large has struggled with any price point bigger than $5. It’s a culture thing I guess: once we created the expectation that you pay that or less for an app, people just balk at a higher price even if the app is deserving.

Jon writes:

Interestingly it seems that the FDA is trying to lower the burden of or at least substantially change software certification.

It looks like Somryst was simultaneously assessed through the 510k pathway and some kind of lighter touch test regulatory procedure for apps from trusted vendors. Looks like they’re currently running apps through both processes to compare the results (and presumably see if the new process misses anything disastrous).

https://peartherapeutics.com/pear-therapeutics-obtains-fda-authorization-for-somryst-a-prescription-digital-therapeutic-for-the-treatment-of-adults-with-chronic-insomnia/

https://www.fda.gov/medical-devices/digital-health-center-excellence/digital-health-software-precertification-pre-cert-program

Yeah, this is interesting. I agree that apps are inherently low-risk, and that it makes sense to hold them to slightly lower standards so that people don’t have to jack up the price too much.

But a part of me worries that “FDA-approved” has a certain cultural meaning, and if the FDA approves one category of thing with lower proof than usual, people are going to get confused and think it’s on a firmer evidence base than it is. Given that it’s entirely legal to sell apps without FDA approval, I don’t really know what the FDA thinks it’s gaining here.

See also comments from the clinical director of a digital therapeutics company, the libertarian Cato Institute’s Director of Health Policy Studies, and a former employee of Pear Therapeutics.