Sleep is one of the first casualties of an ICU stay. Between alarms, ventilators, nursing checks, and the underlying illness itself, critically ill patients rarely get the restorative rest their bodies need. Clinicians have long turned to sleep medications to help, but a large new study from Japan reveals that what they reach for has changed dramatically over the past decade, and that the decision to prescribe in the ICU carries consequences that extend well beyond the unit’s doors.
Researchers from Tohoku University, the University of Tokyo, and the Institute of Science Tokyo analyzed data from the Japanese Diagnosis Procedure Combination database, covering more than 1.5 million adult ICU patients admitted between April 2012 and March 2021. The study, published in Australian Critical Care, is the largest nationwide look at ICU sleep medication prescribing patterns in Japan to date.
The old guard steps down
The headline finding is a striking shift in drug class preference. In 2012, benzodiazepines were the dominant sleep agent in Japanese ICUs, prescribed to 22.1% of patients. By 2020, that figure had fallen to 11.1%, a reduction of roughly half.
Benzodiazepines have long been viewed warily in hospital settings. They carry risks of delirium, respiratory depression, and daytime sedation, all particularly dangerous in already compromised ICU patients. The steady decline suggests that awareness of these risks has translated into practice change, though the drugs have not disappeared entirely.
New classes surge in
Into the gap left by benzodiazepines, two newer drug classes have rushed in. Orexin receptor antagonists, which work by blocking the brain’s wake-promoting signaling pathway, were essentially absent from Japanese ICUs in 2012. By 2020, they were the most prescribed sleep medication class in the study, reaching 18.0% of ICU patients. The rise from zero to one in six patients in under a decade represents one of the fastest medication adoption curves documented in critical care.
Melatonin receptor agonists followed a similar trajectory, though from a small baseline. Prescribing rates climbed from 1.3% in 2012 to 10.7% in 2020. These drugs mimic the body’s natural sleep hormone and have a favorable safety profile, making them an attractive option for fragile patients.
Taken together, the data paint a clear picture: Japanese intensivists are moving decisively away from benzodiazepines and toward orexin antagonists and melatonin agonists as their first-line sleep aids.
The ICU-to-discharge pipeline
But the study’s most consequential finding may be what happens when patients leave the ICU. Among patients who received sleep medication during their ICU stay, 47.0% were prescribed a sleep medication at discharge. Among those who did not receive sleep medication in the ICU, the discharge prescribing rate was just 11.3%. The standardized mean difference between the two groups was 85.5%, indicating a very large effect.
In statistical terms, receiving a sleep medication in the ICU was associated with a nearly sevenfold increase in the odds of being discharged with a sleep prescription (odds ratio 6.90, 95% confidence interval 6.59 to 7.23).
This is the central tension of the study’s findings. The shift toward newer drug classes may represent progress in terms of safety and side effect profiles, but the underlying pattern of starting a sleep medication in the ICU and continuing it through discharge has not changed. Patients who begin taking sleep aids in the hospital are far more likely to leave the hospital with them, raising questions about appropriate deprescribing and the risk of long-term dependence.
The study also found that ICU sleep medication users tended to be sicker overall, with higher illness severity scores, longer ICU stays, and more invasive treatments such as mechanical ventilation. This makes clinical sense: the patients who most need sleep support are often the most unwell, but it also creates a selection dynamic that complicates interpretation.
Limitations and cautions
The authors note several important limitations. As a retrospective observational study using administrative claims data, it can identify associations but cannot establish causation. The database lacks information on why specific medications were chosen or discontinued, and it does not capture sleep quality outcomes or delirium rates. The study also cannot determine whether the increase in orexin antagonist and melatonin agonist prescribing reflects intentional deprescribing of benzodiazepines or simply the availability of new options.
Because the data come from a nationwide Japanese database, the findings are directly relevant to clinical practice in Japan. Whether similar patterns hold in other countries, where drug availability and prescribing cultures differ, remains an open question.
The bottom line
Japan’s ICUs are in the midst of a genuine pharmacological shift. Benzodiazepine use has been cut in half, while orexin antagonists and melatonin agonists have emerged as dominant alternatives. This change likely reflects growing attention to the harms of sedation in critically ill patients and the availability of newer agents with more targeted mechanisms.
Yet the study also reveals a persistent structural pattern: starting a sleep medication in the ICU strongly predicts leaving the hospital with one. That finding suggests that alongside the conversation about which drug to choose, there needs to be an equally important conversation about when to stop. For a field rightly focused on the dangers of starting medications in the hospital, the question of how to safely stop them deserves equal attention.
Source: Iwasaki Y, Ohbe H, Kaiho Y, et al. Prescription patterns of sleep medications for critically ill patients: A nationwide retrospective observational cohort study in Japan. Australian Critical Care. 2026;39(4):101648. DOI: 10.1016/j.aucc.2026.101648.