insomnia clockA patient presents to the emergency department complaining of increasing insomnia due to anxiety. She states that she is not actively suicidal nor homicidal but she has trouble “turning off her brain” at night in order to sleep and her insomnia is worsening her anxiety. She has a history of morbid obesity and smokes 1 pack of cigarettes per day. In order to help you consider writing her a prescription for 5 mg of zolpidem as you presume it to be a benign way to deal with her current sleep disorder. But what does the evidence say about these drugs and the risks of harm?


Anxiolytics and hypnotics are commonly prescribed drugs used for various purposes from insomnia to anxiety. For the purposes of discussion, the drugs can be broken into the standard benzodiazepines (diazepam, lorazepam) and the newer class of benzo-like Z-drugs (zolpidem, zaleplon) primarily utilized as sleep agents. As the prescriptions written have increased, a concomitant recognition of increasing untoward effects of the drugs has caused alarm. There have been individual studies demonstrating an association between these drugs and risks of cancer, dementia, and premature mortality. Most of these studies however have been small and did not control for the multitude of potential confounding variables.

Article Citation

Weich S, Pearce HL, et al. Effect of anxiolytic and hypnotic drug prescriptions on mortality hazards: retrospective cohort study. BMJ 2014;348:g1996. BMJ full article (free). Pubmed

Study Objectives

Determine the association between anxiolytic and hypnotic drugs with the risk of premature mortality

Study Methods 

  • Retrospective matched cohort study utilizing the General Practice Research Database (GRPD)
  • Large primary care database comprising 70 million patient years in 630 practices
  • Eligible patients
    • Age > 16 years with 12 months of records
    • Study drug: defined as Z-drugs and Benzodiazepines
    • January 1998 to December 2001
    • Received a minimum of 2 prescriptions for the study drug (criteria included to presumable exclude those who received but did not fill/take the prescription)
    • Matched 2:1 with controls
  • Outcomes
    • All cause mortality
  • Covariates
    • Adjusted for potential confounders including sex, age, sleep disorder, anxiety, medical comorbidity, and socioeconomic status


  • 34,727 patients vs. 69,418 controls
  • Study Drugs Prescribed
    • Benzodiazepines 63.7%
    • Z-drugs 23.0%
    • Prescribed more than one drug type 33.5% 
  • Patients
    • Study population more likely than controls to smoke, have physical morbidity, and exhibit sleep/anxiety/psychiatric disorders
  • Mortality
    • Hazard Ratio: 3.46 (95% CI 3.34-3.59) for mortality from exposure to any study drug
    • Hazard Ratio: 3.32 (95% CI 3.19-3.45) after adjusting for potential confounders
    • Hazard Ratio: 4.51 (95% CI 4.22-4.82) associated with dose response, as based on duration of daily usage of medication
    • The study used three models to assess mortality and possible confounders, all of which demonstrated mortality associations with prescriptions of studied drugs


This retrospective study of a large primary care database found a positive association between prescribing benzodiazepines and Z-drugs with all-cause mortality. This is consistent with multiple previous small studies which suggested the association. It is interesting to note, however, that at baseline, individuals exposed to the study drug had higher rates of medical and psychiatric comorbidities, as would be expected from the population primarily responsible for the consumption of these drugs. However, when the authors adjusted for confounders such as the aforementioned comorbidities, an association with all-cause mortality was still found despite of the statistical adjustment.

This is intriguing but far from definitive. The retrospective statistical adjustment is helpful to account from confounders but cannot entirely exclude the role that comorbidity plays in all cause mortality. While it may be true that the Z-drugs and benzos are solely responsible for the increase in all-cause mortality, it may also be true that mortality is due primarily to the underlying illness and not the drugs.  The prescribing of drug may in fact be a surrogate for underlying comorbidity, not the triggering event that leads to all cause mortality.

Practical Take Home Points

  1. Benzodiazepines and Z-drugs have increased their market share and penetrance into the patient population. Though these drugs are typically presumed to be benign, a clear association with untoward events has been discovered and should be taken into consideration when prescribing these drugs.
  2. Though the underlying association may in fact be due to comorbidity factors, individuals with the highest rates of medical and psychiatric disease are the greatest consumers of these drugs. Extreme caution should be utilized when deciding to prescribe in this vulnerable cohort.

Future Directions

It is important that this study, along with previous smaller studies, be considered when assessing the potential risk/benefit ratio of a benzodiazepine/Z-drug prescription. These studies CAN NOT state with certainty that the greater mortality is due to the exposure to hypnotics/anxiolytics alone. That being said, these findings of association are concerning, especially given the dose-response association. Continued vigilant post-marketing surveillance, reporting of untoward events, and a large prospective cohort study are required to determine if we are harming individuals by prescribing these drugs.

Image credit

William Paolo, MD

William Paolo, MD

Residency Program Director
Assistant Professor of Emergency Medicine
SUNY-Upstate Medical Center