Zolpidem Use Increases Fall Risk for Inpatients

Abstract & Commentary

By Jennifer A. Best, MD, FACP, FHM, Assistant Professor, University of Washington School of Medicine, Seattle, WA. Dr. Best reports no financial relationships in this field of study.

Synopsis: Zolpidem, a commonly utilized agent for sleep disturbance, is associated with increased fall risk in hospitalized, non-pregnant, non-critically ill patients. Its use for a given patient and its appropriateness within standard order sets should be carefully considered.

Source: Kolla BP, Lovely JK, Mansukhani MP, Morgenthaler TI. Zolpidem Is Independently Associated With Increased Risk of Inpatient Falls. J Hosp Med 2013;8:1-6.

Sleep disturbance, a symptom frequently experienced by hospitalized patients, is often managed pharmacologically with hypnotic medications. In the United States, zolpidem is utilized most commonly because of its relative safety and lack of significant side effects. Of concern, however, are data suggesting that among the population of ambulatory community dwellers with hip fracture, rates of zolpidem use are higher than among the general population; this may suggest an association between zolpidem and fall risk. Falls in the hospital are associated with multiple adverse patient and system-related outcomes; hence, it would be important to know whether zolpidem use is associated with an increased fall risk amongst hospitalized patients.

In this IRB-approved retrospective, cohort study, Kolla and colleagues at the Mayo Clinic sought to determine whether zolpidem administration is associated with fall risk among inpatients. All subjects were adults, 18 years or older, admitted in 2010 to the Mayo Clinic (a large teaching hospital) in Rochester, Minnesota. Pregnant and critically ill patients in an intensive care unit were excluded. All subjects had an active prescription for zolpidem, either scheduled or “as needed” (PRN), as determined by an internal pharmacy database review. In analyzing outcomes, the cohort of hospitalized patients with a zolpidem prescription who received the medication was compared with a cohort with a prescription who did not receive the medication. Here, investigators intended to control for differential factors that would render a patient eligible (or ineligible) for zolpidem use.

The electronic medical record was reviewed for patient demographics as well as ICD-9 diagnostic codes for a number of conditions known to be associated with increased fall risk: visual impairment, gait instability, cognitive impairment/dementia, delirium and insomnia. Hospital length of stay (LOS), the Charlson comorbidity index (a score based on the number of patient medical conditions) and Hendrich’s fall risk score as calculated by nursing on the admission date were also abstracted.

All falls occurring over the study period were identified from a central institutional reporting system through which all events are analyzed. For each fall event, the patient’s medication administration record was accessed to determine all medications received within the prior 24 hours. Of note, medications in classes previously shown to increase fall risk (antidepressants, antipsychotics, antihistamines, sedative antidepressants, benzodiazepines and opioids) were included.

Statistical analysis consisted of the following: univariate analysis to calculate the odds ratio (OR) of inpatient falls for fixed variables: receipt of zolpidem, male sex, surgical admission, insomnia, visual, gait or cognitive impairment or delirium. Odds ratios were also calculated based on continuous variables: hospital LOS, zolpidem dose, Charlson index and fall risk score. Multivariate analysis was then performed to calculate the OR for fall following adjustment for all significantly associated fixed or continuous variable. Rates of use of other medication classes associated with falls were compared between the cohort who was administered and the cohort in whom it was only prescribed.

Over the study period, 16,320 patients were prescribed zolpidem, with 88% of prescriptions written as PRN. 30.4% (4962) of these patients received a dose of the medication. As compared with those with a prescription only, patients who received zolpidem were significantly older, more commonly male, more likely to have insomnia or delirium, had higher Charlson index scores and were more likely to be surgically admitted. No differences were observed between the two groups in admission fall risk scores, hospital LOS, or rates of visual, gait or cognitive impairment.

Within this study population, 672 falls occurred (among 609 unique patients). The OR for falls for patients administered zolpidem, as compared with those who did not receive it was 4.37 (95% CI = 3.33-5.74, P<0.001). The OR remained increased at 2.5 (95% CI 2.08-3.02; P<0.001) when compared with patients who did not receive zolpidem — with or without a prescription. Based on an absolute risk increase of 1.8%, the number needed to harm was 55.

The following variables were found to significantly increase fall risk: delirium (OR 4.96), zolpidem administration (OR 4.37), insomnia (OR 2.37), Henrich’s fall score (OR 1.36), male sex (OR 1.36), Charlson index (OR 1.29), zolpidem dose (OR 1.21) and age (OR 1.01). After adjustment for these factors, the risk of falls associated with zolpidem administration remained elevated (OR 6.39; 95% CI = 3.07-14.49; P<0.001). Relative to their age or use of other medications increasing fall risk, patients sustaining a fall associated with zolpidem administration did not differ from inpatients sustaining a fall in absence of zolpidem.

These authors conclude that zolpidem use in hospitalized patients is associated with an increased risk of falls, even when this risk is adjusted for other factors which are known to increase fall risk. To date, this is the largest study of its kind and the only one which has utilized multivariate analysis. Several study limitations are apparent. First, results cannot be generalized to pregnant patients or those in the ICU. Second, the study utilized administrative data for identification of comorbidities. Next, investigators did not account for the severity of visual, gait or cognitive impairment which may have ramifications for the results. Furthermore, this study does not account for changes in fall risk over a hospitalization, as it utilizes only a baseline fall risk score. More practically, we are provided no additional data with which to weigh the relative fall risks of alternative hypnotic agents, such as trazodone. It is interesting to consider that although insomnia is the most common indication for a zolpidem prescription, insomnia itself was again shown here to increase fall risk and hence the risk of zolpidem may be additive, particularly in patients for whom the medication is ineffective in inducing sleep.

In summary, hospitalists should be aware of the risks of zolpidem use, and as with any medication, weigh cautiously the risks and benefits of use. As no medication is without some risk, nonpharmacologic therapies for sleep should be utilized and prioritized. Providers should monitor the effects of zolpidem on an ongoing basis so as not to continue the medication in patients gaining no benefit. From a systems standpoint, the decision as to whether zolpidem should be included as standard option within templated admission order sets should be carefully considered. Finally, other contributors to fall risk in the hospital should be identified and remedied.