Sleep medications linked to reduced deep sleep and disrupted memory

By Vijay Kumar MalesuReviewed by Lauren HardakerSep 8 2025

Long-term use of benzodiazepines and Z-drugs may help older adults fall asleep, but new research shows they alter sleep architecture and brain activity in ways that could undermine memory and next-day functioning.

Study: Effect of chronic benzodiazepine and benzodiazepine receptor agonist use on sleep architecture and brain oscillations in older adults with chronic insomnia. Image credit: Ground Picture/Shutterstock.com

In a recent study published in Sleep, a group of researchers tested whether chronic benzodiazepine (BZD) and benzodiazepine receptor agonist (BZRA) use alters sleep architecture and non-rapid eye movement (NREM) oscillations in older adults with insomnia.

Background 

Insomnia affects approximately one-third of older adults, and benzodiazepines (BZDs) and benzodiazepine receptor agonists (BZRAs, or “Z-drugs”) are frequently prescribed. These medications can reduce sleep onset latency and improve sleep continuity, but their long-term use is associated with risks such as falls, cognitive impairment, and residual daytime sedation.

Slow-wave sleep and its associated neural oscillations are central to memory consolidation and cognitive function. Chronic sedative use may disrupt these processes, highlighting the need for further research on how specific drugs and doses affect sleep architecture and brain activity in aging populations.

About the study

Older adults aged 55-80 years were classified as good sleepers (GS), individuals with insomnia disorder (INS), or individuals with insomnia plus chronic sedative use (MED). All provided consent and completed polysomnography (PSG). Participants in MED continued their prescribed medication (BZDs or BZRAs). Sleep architecture was scored using American Academy of Sleep Medicine (AASM) rules.

An electroencephalogram (EEG) was recorded from 13 scalp electrodes. Derived indices included total sleep period (TSP), total sleep time (TST), sleep onset latency (SOL), time in bed (TIB), sleep efficiency (SE), arousal density, and sleep fragmentation index (SFI). Spectral power was estimated with fast Fourier transformation and Welch’s method and summarized as relative power in slow oscillation (SO), delta, theta, alpha, sigma, and beta bands. Spindles were detected with an algorithm; slow spindles at frontal (Fz) and fast spindles at parietal (Pz).

Slow oscillations were detected using a finite impulse response (FIR) filter; coupling between SO and spindles was quantified with phase-amplitude coupling (PAC), modulation index (MI), and preferred coupling phase (CP). Group differences were analyzed using analysis of variance (ANOVA) and non-parametric tests with false discovery rate control. Exploratory analyses related to medication dose (diazepam equivalents) and duration to sleep metrics were performed, and BZD versus BZRA users were compared.

Study results

The study analyzed 101 participants (mean age 66 years; 73% women): GS n=28, INS n=26, and MED n=47 taking nightly sedative-hypnotics for more than three nights per week. Compared with GS, both INS and MED showed lower sleep efficiency, more wake after sleep onset, and altered stage distribution.

Chronic medication use was linked to lighter, less restorative sleep. The MED group spent more time in stage N1 (the lightest stage of NREM sleep) and less in stage N3 (deep slow-wave NREM sleep) than both GS and INS, with stage N2 (intermediate NREM sleep) also prolonged versus INS. Rapid eye movement (REM) sleep duration did not differ across groups, although REM latency (trend; not significant after multiple-comparison correction) tended to be longer in MED.

Fragmentation patterns differed across groups. Arousal density was higher in INS than in GS and MED, while the SFI was higher in MED than in INS, indicating more frequent transitions to lighter sleep stages.

In the spectral analysis, the MED group showed lower NREM theta power compared with GS, and lower REM theta power than GS and INS. During NREM sleep, frontal sigma power was higher and frontal low-beta power lower in MED relative to GS. The delta-to-beta ratio, an index of cortical arousal, was reduced in MED compared with both GS and INS. Analysis of discrete oscillations revealed selective changes: frontal spindle density was higher in MED than in INS, while SO density and amplitude did not differ across groups.

Critically, the temporal alignment of rhythms was altered in MED. MI was lower in MED compared with GS, and the preferred CP occurred later in MED compared with GS and INS, a profile unfavorable for memory consolidation. Within MED, higher diazepam-equivalent dose per use correlated with longer SOL and longer latencies to stages N2 and N3, and with higher NREM sigma and beta power. Longer cumulative duration of use was related to shorter parietal spindle duration.

Despite lower SE, TST was higher in MED than in GS, consistent with more time in bed. However, restorative deep sleep was reduced. After adjusting for dose, outcomes did not differ between BZDs and BZRAs.

Age was regressed out of all sleep measures, and insomnia severity scores were higher in INS and MED than in GS, with MED slightly lower than INS. No group differences were observed for SO delta or alpha power; the coupling metrics did not scale with dose. Together, the macro- and micro-level signatures indicate that chronic sedative-hypnotic use is associated with poorer sleep regulation in late life and may help to explain, though the study did not directly test, reports of next-day grogginess and cognitive decline by disrupted oscillatory coupling.

Conclusions

Medicine-assisted sleep is not the same as restorative sleep. In older adults with insomnia, chronic BZD and BZRA use is associated with more light sleep, less deep sleep, reduced theta activity, altered spindle dynamics, and weaker, delayed coupling between slow oscillations and spindles, mechanisms linked to memory.

These objective changes may help explain reports of next-day grogginess and cognitive decline in long-term users and support cautious, dose-aware prescribing. Expanding access to cognitive behavioral therapy for insomnia and developing alternatives that preserve brain rhythms could improve sleep and protect cognition today.

Download your PDF copy now!

Bottom of Form