Why don't we have better over-the-counter sleep aids?

I go down some rabbit holes and uncover potential issues in how drugs are regulated

In July 2022, about a month after getting COVID-19, I developed the worst insomnia I’ve ever experienced.

I decided to try a number of sleep medications — I figured if there was one that worked really well it would be worth finding it. My endeavor was partially inspired by Luisa Rodriguez’s post on trying all the antidepressants. I also read a couple of books on sleep.

I’ve emerged from all that quite opposed to the use of sleep medications. You can read more details about my experience on my Medium. Here are my takeaways:

• The best first-line treatment for insomnia is Cognitive Behavioral Therapy for Insomnia (CBT-i) (I’ve put a free CBT-i manual on my website).

• Sleep medications can help to resolve treatment resistant insomnia by helping stop various vicious cycles that occur.

• All sleep medications come with significant negative side effects. Most have tolerance, dependence, and withdrawal effects to one degree or another.

• Many sleep medications are probably not safe for long term use.

• Sleep medications should not be used for more than a week or two.

• The new orexin antagonist sleep aids (Belsomra, Dayvigo, and Quviviq) are safer and have a more tolerable side effect profile than older sleep aids.

For more on why sleeping pills are bad, I recommend the book The Sleep Solution by Chris Winter. See also Scott Siskind’s Lorien Psychiatry page on insomnia.

While I am fairly against the use of sleep medications, if one really does want to take one, I noticed that the OTC options really suck. This led me down a rabbit hole where I stumbled upon some issues regarding how medications get to market.

Benadryl and doxylamine

Again, I want to reiterate that CBT-i is the best way to cure insomnia because it directly addresses the most common root causes of chronic insomnia.

However, CBT-i is hard. So, the allure of a magic pill that can help you sleep is extremely strong. If you go into any CVS or Walgreens you’ll see a wall of products containing sleep medications. That wall contains dozens of different products, but only a few actual substances. ~98% of what is sold there is either melatonin, diphenhydramine (Benadryl), or doxylamine succinate (often branded as Unisom). The other 2% of things there ashwaganda or valerian.

Now, melatonin is actually not much of a sleep aid according to randomized controlled trials, it’s more of “circadian rhythm setter that helps you fall asleep slightly faster”. Read Scott Siskund’s excellent article on melatonin if you want the full low-down on that.

The other two substances, Benadryl and doxylamine, are in shockingly wide use. A 2019 study found that “more than a third of older adults (age 50 - 80) use medications or aids to help with sleep—most commonly OTC aids.”

I also saw research that found that many parents give OTC sleep aids to their kids as a quick fix to sleep issues, a practice that doctors do not recommend. (Side note: I would also avoid giving children melatonin! Growing evidence suggests that melatonin can disrupt really important hormonal processes involved in puberty.)

The problems with Benadryl and doxylamine

Anticholinergic side effects & long term safety issues

Both Benadryl and doxylamine bind to the muscarinic acetylcholine receptors very strongly. This induces a large number of unpleasant side effects. Most prominent among these effects is the drying out of mucus membranes, so you may wake up with a sore throat. It’s also well known that anticholinergic activity disrupts cognitive function, causing impaired thinking and poorer memory. I have seen some speculation that these brain-slowing effects help contribute to the sedation of these drugs. This makes a bit of sense - it’s hard to stay awake thinking and/or worrying about something if you can’t think very well. Still, these effects on cognitive function linger the next day, especially with doxylamine, which isn’t good. Even with Benadryl, which has a shorter half-life, studies show impaired psychomotor function the next day. Older patients have fewer liver enzymes, so these drugs will remain for longer in the body, resulting in cognitive problems that have been found in studies. Of course, nearly all sleep medications, even orexin antagonists, result in impairments early the next day.

The more troubling thing is the growing body of work has found that long-term use of anticholinergic drugs is associated with a greater risk of dementia and Alzheimer’s. The hazard ratios are small (1.2 - 2.0) but the effect seems fairly robust (see this meta-analysis).

Tolerance & dependence

One study suggests that tolerance to the sedating effects of Benadryl develops in just four days. (If you’ve ever met someone who takes Benadryl for allergies regularly but doesn’t suffer from sedation, that’s why). What do people do when they develop tolerance to the sedative effects? They up the dose. That leads to more anti-cholinergic side effects. It can also cause dependence (when you lower the dose it becomes harder to sleep).

They might mess with sleep architecture

Many medical experts suspect that first generation antihistamines have negative effects on sleep quality. I could not find any studies on Benadryl and doxylamine in particular, but some evidence exist that other first generation antihistamines reduce REM sleep and increase the time until REM onset. For example, a 1971 EEG study on the first generation antihistamines methapyrilene and scopolamine showed a reduction of REM sleep.¹

We don’t actually know how well they work

These two medications clearly work to some degree. Yet, there isn’t much good scientific work on either compound. A 1983 study found Benadryl had some benefit but only tracked patients for one week. One randomized controlled trial (RCT) on Benadryl showed positive but not statistically significant effects on insomnia parameters (their best p-value appears to be 0.078). The study is also confounded by the inclusion of valerian and hops. A 2008 study reports that Benadryl does not help with sleep quality or total sleep time. A review of three studies on Benadryl concludes there are “limited beneficial effects” and the compound “lacks robust clinical evidence”. I could only find two RCTs on doxylamine. They show some benefit but one lacked objective measurements and the other trial was open-label (not blinded).

Other possible issues

Benadryl is dangerous in overdose and can cause respiratory depression and death. This is in pretty stark contrast to second generation antihistamines.

Another point is that histamine is really important to many functions in the body far beyond its role in the sleep-wake cycle. There may be long-term negative consequences to taking antihistamines that we don’t know about. For instance, research from 2016 suggests that histamine plays an important role in the body’s response to exercise. A 2021 study in Science Advances found that H1 & H2 blockers blunted the body’s response to exercise. Replication and further investigation is needed, but this is a cause for concern.

Why are Benadryl and doxylamine OTC?

“The first-generation antihistamines, especially diphenhydramine, did not pass through the same safety and efficacy standards that are applied to newer second-generation medications.” from Diphenhydramine: Time to Move on?, 2022

If diphenhydramine and doxylamine were brought to the market today, there’s a good chance neither would be authorized for OTC use by the FDA because of all of the safety issues touched on in the last section. In Germany, the Netherlands, and Sweden, Benadryl is only available with a prescription.

It’s a bit interesting to see how both of these substances ended up OTC:

• Benadryl was discovered in 1940 and was approved by the FDA in 1946 for prescription use for allergies. In 1982, for reasons that are mysterious, the FDA made it OTC.

• Doxylamine was discovered in 1948. By the 1950s the FDA had approved it as an OTC allergy drug. It was added to NyQuil in 1966.

Crucially, both of these drugs were originally approved prior to the passing of the 1962 Kefauver-Harris Drug Amendments. Those laws, passed in response to the thalidomide tragedy in Europe, mandated efficacy testing in addition to safety testing before drugs could be sold on the market.

Many medical organizations and doctors have articles online arguing that FDA should remove these drugs from OTC status. While I am generally opposed to stricter regulation of drugs and supplements, I tend to agree with this. Diphenhydramine and doxylamine are dangerous and safer more effective alternatives are available, both for the treatment of allergies and for insomnia.

Are there any candidate drugs that could replace Benadryl and doxylamine as OTC sleep aids? I explore two options below - esmirtazapine and orexin antagonists.

What happened with esmirtazapine development?

Dozens of antihistamines are known (41 are listed on Wikipedia). All antihistamines cause sedation to some extent. However, many newer (“second generation”) cause very low amounts of sedation because they struggle to cross the blood-brain barrier. As far as I can tell there is only one antihistamine known with the following desiderata:

1. A short half life (< 10 hours), to prevent next day sedation.

2. No anticholinergic activity, to prevent cognitive dysfunction and ensure long-term safety.

3. Crosses the blood-brain barrier.

That compound is esmirtazapine.

Esmirtazapine has the added benefit of also being a 5-HT2a receptor antagonist (like trazodone), which is known to have sleep-promoting effects.

Mirtazapine (Remeron) is a racemic mixture of S-mirtazapine and R-mirtazapine. Low dose mirtazapine (3.75 - 10 mg) is one of the most effective sleep aids known based on the data I’ve seen. My understanding is that the brain does develop some tolerance to the anti-histamine effect of mirtazapine within a few days, but that obtaining complete tolerance to the drug is very rare.² I believe this is for two reasons. Firstly, the antihistamine effect is so incredibly potent that it is hard for the body to develop complete tolerance. Secondly, tolerance to 5-HT2a antagonism is not a thing. In fact, 5-HT2a antagonism leads to paradoxical down-regulation, which is actually the reverse of what one would expect to happen during tolerance.

Low dose mirtazapine has one critical flaw though as a sleep aid - its very long half-life of 20 - 40 hours. This results in people being cognitively sluggish during the day. Some people compensate by consuming caffeine so this is not an issue, but it is a major issue overall, in my view. (Note: at the higher doses used for anxiety and depression, the norepinephrine promoting effects of mirtazapine do a really good job compensating for this so sedation is rarely reported as an issue).

Mirtazapine is a mixture of R-mirtazapine and S-mirtazapine. The S-enatiomer of mirtazapine (“esmirtazapine”) has a shorter half life of around 10 hours.³ This obviously makes it very attractive for the treatment of insomnia. In 2016 Organon, the inventor of mirtazapine, conducted a Phase II trial on esmirtazapine to determine the optimal dosage. The trial found good results and Organon claimed there was no evidence of significant next-day sedation or withdrawal effects upon discontinuation. Organon then went on to conduct a Phase III trial which also found good results.

However, Organon ultimately decided to terminate development of esmirtazapine in 2010 “for strategic reasons”. No other information was given on why they abruptly stopped working on it.

A research poster from Schering-Plough Corporation, a subsidiary of Merck, offers a clue as to what happened. It appears Merck conducted a bunch of Phase I trials investigating esmirtazapine. Their research uncovered that the half-life of esmirtazapine is significantly longer in ~15% of patients who have genetics that make them “poor CYP2D6 metabolizers”. In poor metabolizers the overall drug exposure becomes twice as high. Now, mirtazapine as a racemic mixture is metabolized by several liver enzymes - CYP1A2, CYP2D6, and CYP3A4. The relative abundance of these enzymes differs widely between people due to both CYP gene mutations and the number of copies those genes they have.

Looking closer at the results Organon’s Phase III trial, while they claimed that next-day sedation was not an issue, 15% of patients complained of daytime sleepiness vs 3.5% in the placebo group. 15% is exactly the fraction of people that the Merck study identified as having poor metabolism of esmirtazapine. (For reference, the FDA states that 7% of Caucasians and 2% of African Americans are “poor CYP2D6 metabolizers”).

So now the crux of the issue comes into view. The drug works well for 85% of the population, but for 15% it causes next-day sedation which makes it dangerous (for instance if someone hops out of bed and then drives to work while sedated).

Now if there is one thing the FDA cares a lot about when it comes to sleep medications, it’s next day sedation. So, my guess is that both Organon and Merck gave up on developing esmirtazapine because 15% of people will experience sedation that is too great.

The logical thing would be for the FDA to approve esmirtazapine, but only if a psychiatrist obtains a genetic test which shows it is not contraindicated. However, there doesn’t seem to be a standard mechanism for this sort of approval (more on that in the conclusion).

Should orexin antagonists be schedule IV drugs?

Orexin is a neuropeptide that plays an important role in regulating wakefulness and appetite. It is one of the most critical biochemical components of the sleep-wake cycle. Narcoleptics lack neurons which produce orexin, resulting in circadian disruption.

Dual orexin receptor antagonists (DORAs) are a relatively new class of sleep medication which block two different orexin receptors, resulting in a reduction in wakefulness and arousal. Three DORAs have been approved by the FDA - Belsomra (suvorexant), Dayvigo (lemborexant, 2019), and Quviviq (daridorexant, 2022).

I think there is a case that Dayvigo and Quviviq should be available over-the-counter (Belsomra however has a much longer half-life, making it an inferior option).

Personally I have tried both Quviviq and Dayvigo. What is remarkable about these drugs is how incredibly ‘clean’ they are. They have almost no appreciable side effects other than residual next day sedation, which is nearly impossible to remove from any sleep medication. Overall these medications are pretty gentle compared to most other sleep medications. One effect I noticed is a slight boost in mood. Intriguingly, DORAs are being investigated as novel treatments for depression.

However, we have a long way to go before any DORAs become OTC. That’s because so far the DEA has classified all three DORAs on the market as Schedule IV drugs. This is the same scheduling as benzodiazapines, z-drugs like Ambien, and the opioid Tramadol.

The DEA believes that DORAs have enough “abuse potential” to warrant Schedule IV. I find this hard to believe. DORAs have no euphoric, anxiolytic, hallucinogenic, or dissociative effects. But what does the science say? The first studies done on rats found no evidence of abuse potential. A summary of studies says “there was a lack of positive reinforcing effects of the DORAs in self-administration studies in rats, which is in sharp contrast to drugs of the benzodiazepine or Z-drug classes, which have been shown to maintain self-administration in animals”. No evidence for abuse potential was found in monkeys either.

However, there is some evidence of abuse potential among “recreational users of sedative drugs”. A 2016 study found Belsomra had abuse potential among sedative drug users “consistent with schedule IV classification”. In sedative drug users, Quviviq is said by the FDA to have abuse potential significantly higher than placebo but much lower than Ambien.

There is no evidence, however, for abuse potential in people who are not already sedative drug abusers. In phase III trials with over 1,000 patients there was no evidence of abuse uncovered for Belsomra or Quviviq.

The aforementioned studies on sedative drug users were done after the DEA’s 2014 decision to make Belsomra Schedule IV. I am not sure what studies (if any) the DEA based their decision on. The only study from 2014 or earlier I can find indicating abuse potential is a 2014 study on Almorexant, a DORA that was under development by GSK. It appears the DEA said “this is a new class of drugs, and might be dangerous, so let’s make it Schedule IV to be safe”.

As far as I can tell, Belsomra is not scheduled in Australia, Canada, or Japan. The drug company Idorsia has submitted a petition to the DEA asking that DORAs be descheduled.

The abuse potential of Benadryl is fairly well established. I think it is quite plausible that the abuse potential of the three DORAs on the market are less than that of Benadryl.

There is also no evidence of dependence developing to any DORA, which is quite remarkable compared to nearly all other sleep aids.

Given all of this I think there may be a case for making a DORA like Dayvigo OTC while removing Benadryl and doxylamine from OTC status.

Among the three DORAs on the market, Dayvigo has the shortest effective half-life and the least next-day residual effects. The only reason I can think of for being cautious about making these drugs OTC is the potential for drug-drug interactions. All three DORAs on the market are only metabolized by CYP3A4. If another drug inhibits this enzyme, the DORA’s duration of action will increase. On the other hand, Benadryl and doxylamine have potential for interactions as well, and Benadryl is an in inhibitor of CYP2D6.

The tldr takeaways

In summary, I explained why Benadryl and doxylamine are unsafe, have questionable efficacy for insomnia, and probably should not be available OTC. I then discussed two alternatives which I thought might be reasonable to make OTC - esmirtazapine and dual orexin receptor antagonists (DORAs).⁴

It appears esmirtazpine development was halted because about 7-15% of people have genetics that make them poor metabolizers of it. Ideally, the FDA should be able to approve a drug but only on the condition that genetic testing is done beforehand to see if its a good fit.

Right now there are many drugs where the FDA gives dosage guidance based on genetics, but I have not found any psychiatric drugs where the FDA requires genetic testing.⁵

As two examples of how the FDA does things right now, the drug label for the antipsychotic Abilify states that “known CYP2D6 poor metabolizers” should be administered a half dose. Another example is the popular ADHD medication Strattera. In poor CYP2D6 metabolizers, the half-life of Strattera skyrockets from 4-5 hours to 19-20 hours and drug exposure becomes 10x greater. The FDA recommends “dosing adjustment” in “known CYP2D6 poor metabolizers” but does not give any further guidance. In my view, the FDA should recommend or perhaps even mandate genetic testing before Strattera is prescribed.

Regarding DORAs, the DEA’s decision to make them Schedule IV drugs seems questionable, since abuse potential has only been found in sedative drug users. Even in sedative drug users, the abuse potential was much less than ambien or benzodiazapines. It’s quite possible that the abuse potential of Benadryl is higher than any DORA.

Given their efficacy and almost non-existent side-effect profile I think there is a case for making a DORA like Dayvigo OTC. Two very short half-life DORAs are currently under development by Taisho Pharmaceuticals and Janssen. If shown to be effective I see very little reason to not make them OTC.

1. Kales, J., et al. “Are Over-the-Counter Sleep Medications Effective? All-Night EEG Studies.” Current Therapeutic Research, Clinical and Experimental, vol. 13, no. 3, Mar. 1971, pp. 143–51. This paper was really hard to obtain so I uploaded a PDF to my website.

2. Read a fascinating mirtazapine experience on the qwerky science blog here.

3. This is very similar to how S-modafinil has a shorter half-life than R-modafinil, which I wrote about previously.

4. One might think it makes sense to make low-dose trazodone OTC. It is considered very safe. However, there are rare but serious side effects like priapism. Dose calibration is really important for trazodone, so that’s one reason I feel it is best given under the supervision of a psychiatrist. The trazodone metabolite mCPP might be dangerous. If you dig deep into this there is an interesting story about how different CYP genetics effect trazodone’s metabolism. Long story short, about 10% of people have a bad reaction to trazodone due to increased concentrations of the metabolite mCPP in the middle of the night. As with several other drugs mentioned in this post, ideally genetic testing should be used to determine if trazodone is a good fit.

5. There are some non-psychiatric drugs that the FDA says are contra-indicated for people with certain mutations. For instance, valproic acid is contraindicated in people with mutations to the POLG gene. There are also several cancer drugs which are only indicated for tumors that have a certain genetic profile.