DEAR DR. ROACH: I have occasional difficulty when going to sleep and/or getting back to sleep. I have tried Ambien, trazodone, clonazepam and Lunesta. All have their pros and cons.

My question regards sleep quality. I know there are several stages of sleep that we should go through and that the brain cleans out junk during sleep. Do these sleep aids still allow a person to go through the needed sleep stages and also allow the brain to clean itself? Are any of these meds better or worse than the others for occasional use? — S.H.

ANSWER: How much time we spend in the different stages of sleep is called sleep architecture. Just being asleep isn’t enough; people need to have enough time in the different phases of sleep for optimal health.

For example, you refer to the body’s waste removal system from the brain, called the glymphatic system. This system removes waste materials from the brain, including amyloid and tau proteins, and it’s predominantly active in the N3 stage of non-rapid-eye-movement sleep (non-REM sleep, also called slow-wave sleep or deep sleep). You’re quite right that many sleep medications affect sleep architecture, which is one major reason why many sleep experts recommend against the habitual use of sleep medications.

All the medicines you’ve used can affect sleep architecture. Ambien (zolpidem) and Lunesta (eszopiclone) are in the Z-drug class, which have small to moderate effects on sleep architecture with modest decreases in both REM and N3 sleep. Many people quickly become accustomed to using these drugs, but the occasional use of them presents a low risk.

Clonazepam is a benzodiazepine, which are among the worst offenders against sleep architecture due to their significant reduction in REM and N3 sleep. A reduction in REM sleep is associated with worsened heart disease and overall mortality, as well as emotional problems and dementia. I strongly recommend against the habitual use of benzodiazepines for sleep.

Trazodone actually increases N3 sleep and has no consistent effect on REM sleep, so it’s commonly been recommended for sleep. Unfortunately, more recent studies have shown that some people can get used to the drug after only a few weeks, so it stops working as well. Furthermore, studies have only shown very small improvements (a few minutes’ worth) in sleep duration. Many people feel very drowsy the following day, and older people are particularly at risk for falls. There are other uncommon but serious side effects, and I’ve mostly stopped prescribing trazodone to my patients.

Whenever possible, I avoid drugs for the treatment of insomnia. Cognitive behavioral therapy for insomnia is, at least, as effective as medication and has none of the medication-associated risks. There are free cognitive behaviorial therapy programs for insomnia (CBT-I) for mobile devices.

For people who don’t do well with CBT-I or people who can’t receive it, the medications that are currently considered the safest are low-dose doxepin and dual orexin receptor antagonists (DORAs), such as suvorexant (Belsomra). Both of these medications have minimal negative effects on sleep architecture, and they increase REM sleep. Unfortunately, DORAs are very expensive (around $500 per month, not often covered by insurance), compared to doxepin (less than $50 per month without insurance).

Dr. Roach regrets that he is unable to answer individual letters, but will incorporate them in the column whenever possible. Readers may email questions to ToYourGoodHealth@med.cornell.edu. (c) 2026 North America Syndicate Inc.

All Rights Reserved