The Safety of Benzodiazepines and Opioids in the Geriatric Population: An Analysis of Patient Safety Events Reported by Hospitals and Ambulatory Surgical Facilities

Introduction

Since the introduction of benzodiazepines in 1960 with the approval of chlordiazepoxide, this class of medications has become one of the most widely prescribed in the United States.1 Although benzodiazepines are frequently prescribed to patients age 65 years and older (hereafter referred to as older adults), the American Geriatrics Society (AGS), which is a nonprofit, nationwide society of geriatrics healthcare professionals dedicated to improving the health,  independence, and quality of life of older people, has explicitly designated this class as potentially inappropriate for older adults in the AGS Beers Criteria, which is a regularly updated list of medications that should be avoided in older adults.² Older adults have both an increased sensitivity to and a decreased metabolism of benzodiazepines, and in general, all drugs in this class may increase the risk of cognitive impairment, delirium, falls, fractures, and motor vehicle crashes in this patient population.² The concomitant use of benzodiazepines and opioids is also designated as potentially inappropriate for older adults due to the increased risk of overdose.² 

Despite the risks of prescribing benzodiazepines and opioids together in older adults, these vulnerable patients continue to receive these two classes of medications together in both the inpatient and outpatient settings. Older patients who are hospitalized may be at a heightened risk because they often receive these medications intravenously during or after a procedure, meaning the effects of the combination may be more rapid and potent. The Pennsylvania Patient Safety Reporting System (PA-PSRS)* received several reports of serious events in which the patient may have experienced the effects of benzodiazepine and opioid toxicity following a procedure, which in some reports led to death, although sometimes the root cause of these events was only discovered after further investigation. In this analysis, we specifically searched PA-PSRS for reports of patient safety events related to the concomitant use of benzodiazepines and opioids in older adults, because this combination is not recommended for these patients. The goal of the analysis was to gain a better understanding of patient safety events potentially resulting from the combined use of these medications in older adults.

Methods

The Patient Safety Authority has been collecting reports of patient safety events in PA-PSRS since 2004, and this database is one of the largest of its kind in the world. Pennsylvania healthcare facilities submit reports of patient safety events to the database that include answers to both structured fields—including event date, patient age, patient sex, care area, and facility type—and free-text fields, such as event detail, which allows the reporter to provide a narrative of the patient safety event in their own words. Reports may vary in length and quality based on not only the specifics of the event, but also on what the reporter deems important or relevant. Reporters are often bedside providers involved directly in an event, so time constraints may also impact reporting.

We queried PA-PSRS on April 5, 2021, for reports of patient safety events that occurred between January 1 and December 31, 2020, which was the most recent full year period reported to PA-PSRS. We limited the search to reports involving patients who were 65 years or older. We searched free-text narrative fields (i.e., event detail, event comments, event recommendation, and event subtype other) and medication name fields for two groups of keywords, looking for at least one keyword from each group. The first group of keywords was intended to identify reports mentioning that an opioid pain medication or opioid antagonist was used and included words such as “opioid,” “opiate,” and “narcotic,” as well as brand and generic names of all currently available opioid pain medications and opioid antagonist medications. The second group of keywords was intended to identify reports mentioning that a benzodiazepine or benzodiazepine antagonist was used and included the keyword “benzo” as well as brand and generic names of all currently available benzodiazepine and benzodiazepine antagonist medications. In both groups, if there was a common root for several medications, that was searched instead of the individual names; for example, we searched for the term “codone” to capture reports mentioning either hydrocodone or oxycodone, and we searched for the term “zolam” to capture reports mentioning either alprazolam or midazolam. 

We conducted a preliminary review of each report to identify events for inclusion. A report was included in the analysis if the patient received both an opioid medication and a benzodiazepine medication, as indicated by either direct mention or because a reversal agent was mentioned, and the patient subsequently experienced an adverse drug reaction (ADR) that may have been caused by either or both medications. For the purposes of this analysis, an ADR was defined as a considerably harmful or unpleasant reaction, resulting from the use of a medication administered as part of prescribed therapy, which required specific treatment, a change in the dosage regimen, or withdrawal of the medication.4 Reports were included regardless of whether the ADR may have been unpredictable or was the result of a medication error. Reports were excluded if these medications were mentioned but the patient did not experience an ADR or if the patient took their own home medications in these classes and experienced an ADR. Duplicate reports were also excluded. We performed a descriptive analysis to identify trends in patient age and sex, event classification and harm score, event type and subtype, and care area group. We then performed a qualitative analysis of each report, with a focus on free-text narrative fields and medication name fields. We coded each report for the medications involved, whether reversal agents were used, and the type of reaction the patient experienced.

Results

Descriptive AnalysisThe query identified 190 reports in the PA-PSRS database that occurred between January 1 and December 31, 2020. We excluded 110 reports from the analysis based on the inclusion and exclusion criteria specified in the methods; the most common reason for excluding a report was that a benzodiazepine and/or an opioid were mentioned, but no ADR was described in the report. The final dataset included the remaining 80 reports, which were submitted by 39 facilities, including 35 acute care hospitals, three ambulatory surgical facilities, and one long-term acute care facility.

Reports were split evenly between males (50.0%; 40 of 80) and females (50.0%; 40 of 80). Keeping in mind that the search was limited to patients aged 65 years and older, patients described in reports ranged in age from 65 to 97 years, with a median patient age of 74 years (25th percentile = 69 years; 75th percentile = 81 years). 

The reporting facility assigned an event classification (incident† versus serious event‡ ) and harm score to each report. cidents comprised 85.0% (68 of 80) of reports and had harm scores of A§  (n=2), C# (n=11), and D¶ (n=55). The remaining 15.0% (12 of 80) of reports were classified as serious events and had harm scores of E** (n=11) and F^†† (n=1). For comparison, in 2020, just 3.0% (8,368 of 278,548) of reports across the entire PA-PSRS acute care database were classified as serious events.⁵

The reports investigated in this study had a significantly higher proportion of serious events than the proportion in all reports in 2020 in the PA-PSRS database (15.0% versus 3.0% respectively, p < .001, Fisher’s exact test).The most common event types, which were assigned by the facility, were ADRs^‡‡ (52.5%; 42 of 80) and complications of a procedure, treatment, or test (23.8%; 19 of 80). Other event types were medication errors (7.5%; 6 of 80); other/miscellaneous (6.3%; 5 of 80); falls (5.0%; 4 of 80); errors related to a procedure, treatment, or test (3.8%; 3 of 80); and skin integrity (1.3%; 1 of 80). Among ADRs, reported subtypes were other (52.4%; 22 of 42), mental status change (35.7%; 15 of 42), and hypotension (11.9%; 5 of 42). Among complications of a procedure, treatment, or test, the most frequently reported subtypes were cardiopulmonary arrest outside of the intensive care unit (47.4%; 9 of 19) and anesthesia event (36.8%; 7 of 19).