Emergency Department Visits Attributed to Adverse Events Involving Benzodiazepines, 2016–2017

Abstract

Introduction:

Characterization of emergency department visits attributed to adverse events involving benzodiazepines can be used to guide preventive interventions. This study describes U.S. emergency department visits attributed to adverse events involving benzodiazepines by intent, patient characteristics, and clinical manifestations.

Methods:

Data from the 2016–2017 National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance project were analyzed in 2019 to calculate estimated annual numbers and rates of emergency department visits attributed to adverse events involving benzodiazepines, by intent of benzodiazepine use.

Results:

Based on 6,148 cases, there were an estimated 212,770 (95% CI=167,163, 258,377) emergency department visits annually attributed to adverse events involving benzodiazepines. More than half were visits involving nonmedical use of benzodiazepines (119,008; 55.9%, 95% CI=50.0%, 61.9%), one third were visits involving self-harm with benzodiazepines (64,721; 30.4%, 95% CI=25.6%, 35.2%), and a smaller proportion of visits involved therapeutic use of benzodiazepines (29,041; 13.6%, 95% CI=11.4%, 15.9%). The estimated population rate of visits was highest for non-medical use of benzodiazepines by patients aged 15–34 years (7.4 visits per 10,000 people). Among visits involving nonmedical use of benzodiazepines, 54.8% (95% CI=49.8%, 59.8%) were made by patients aged 15–34 years, 82.7% (95% CI=80.1%, 85.4%) involved concurrent use of other substances (illicit drugs, alcohol, prescription opioids, and/or other pharmaceuticals), and 24.2% (95% CI=17.7%, 30.6%) involved cardiorespiratory arrest or unresponsiveness.

Conclusions:

These findings support recommendations to assess for and address substance use disorder before initiating or continuing benzodiazepines and reinforce the need for validated self-harm risk assessment tools for clinicians.

INTRODUCTION

Benzodiazepines have a wide range of indications for use including treatment of anxiety, insomnia, seizures, and acute alcohol withdrawal and for peri-procedural sedation. The risk of severe adverse events at therapeutic doses is generally considered low. However, product labeling, including a boxed warning,¹ cautions that co-administration with another central nervous system depressant, such as opioids or alcohol, can cause severe respiratory depression.^2,3 Prolonged use of benzodiazepines increases risk of physical dependence,⁴ and use by older adults has been associated with increased risk of cognitive impairment, falls, and fractures.⁵

Benzodiazepines have also been identified as one of the classes of medications most frequently involved in harms from nonmedical use of substances in the U.S. Use of benzodiazepines increased from 2.0% in 1999–2000 to 4.2% in 2013–2014,⁶ with 13.5 million adults filling a prescription for benzodiazepines in 2013.⁷ From 2004 to 2011, benzodiazepines were involved in nearly as many emergency department (ED) visits for misuse and abuse of medications as opioid pain relievers,^8–10 and by 2016, benzodiazepines were implicated in more ED visits for nonmedical use (47%) than prescription opioids (36%).¹¹

To help target interventions to reduce harm involving benzodiazepines, this study uses nationally representative public health surveillance data to estimate numbers of ED visits attributed to adverse events by patient characteristics and intent of use and to identify concurrently involved substances and clinical manifestations.

METHODS

Study Sample

The authors used data from the National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance (NEISS–CADES) project, a joint collaboration of the Centers for Disease Control and Prevention (CDC), the U.S. Consumer Product Safety Commission, and the U.S. Food and Drug Administration. This active public health surveillance system is based on a nationally representative probability sample of 60 hospitals in the U.S. and its territories with at least 6 beds and a 24-hour ED.¹² Data collection for the NEISS–CADES project is considered a public health surveillance activity by federal human subjects oversight bodies and does not require human subject review or IRB.¹³

As described previously,^14,15 abstractors review medical records to identify ED visits attributed to adverse events involving medications used for any reason. Adverse events include adverse reactions to therapeutic doses, administration errors by adults, unsupervised ingestions by children, overdoses of any kind, and secondary injuries (e.g., choking on a pill) attributed to medication. Up to 4 implicated medications, free-text narratives of the event (including clinical manifestations, relevant preceding events, illicit drug use, or alcohol consumption), clinician diagnoses, laboratory testing, treatment administered, and discharge disposition are recorded. Clinical manifestations and medications are coded based on the Medical Dictionary for Regulatory Activities, version 9.1, and a modified version of the Veterans Administration National Drug File, respectively.

Measures

Cases in which a benzodiazepine was implicated from January 1, 2016 through December 31, 2017 were selected. If benzodiazepine use was based only on positive toxicology testing without supporting clinical documentation, the case was excluded. Clinicians’ assessment of patients’ intent of benzodiazepine use was classified as therapeutic (used as directed or unintentional errors or ingestions), self-harm (used intentionally to injure oneself), or nonmedical use. Nonmedical use included the following: (1) current abuse (clinician diagnosis of abuse or documentation of recreational use; e.g., to get high), (2) therapeutic misuse (documented therapeutic intent, but not used as directed; e.g., taking someone else’s benzodiazepine to self-treat anxiety), or (3) overdoses without documentation of therapeutic intent, self-harm, abuse, or misuse (e.g., patients who have documented overdoses but are unable or unwilling to describe the event). Although concern has been raised that the term abuse may contribute to stigma,^16,17 it is employed here because the term remains commonly used by clinicians in medical documentation.

Statistical Analysis

Cases were weighted based on the inverse probability of selection, adjusted for nonresponse and hospital nonparticipation, and post-stratified to adjust for changes in the total number of hospital ED visits each year. National estimates of ED visits with corresponding 95% CIs were calculated using the SURVEY-MEANS procedure in SAS, version 9.4 to account for the sample weights and complex sample design. NEISS–CADES estimates for the 2-year period (2016–2017) were divided by 2 to obtain the average annual estimates. Estimates based on small numbers of cases (<20) or total estimates <1,200 are considered statistically unstable and are not shown. Estimates with coefficients of variation >30% are considered statistically unstable and are noted. To estimate rates of ED visits by age, the authors used intercensal estimates from the U.S. Census Bureau.¹⁸ Data were analyzed in 2019.

RESULTS

Based on 6,148 NEISS–CADES surveillance cases, there were an estimated 212,770 (95% CI=167,163, 258,377) ED visits attributed to adverse events involving benzodiazepines in the U.S. per year in 2016–2017, accounting for 9.9% (95% CI=8.2%, 11.5%) of the total number of estimated ED visits attributed to medication adverse events. More than half were visits involving nonmedical use of benzodiazepines (119,008, 95% CI=87,010, 151,005; 55.9%, 95% CI=50.0%, 61.9%), nearly equally divided between visits with documented abuse (26.1%) and overdoses without documentation of therapeutic intent, self-harm, abuse, or misuse (25.1%) (Table 1) (Appendix Table 1, available online). One third were visits for self-harm using benzodiazepines (30.4%, 95% CI=25.6%, 35.2%; 64,721, 95% CI=49,939, 79,503), and a smaller proportion of visits was due to adverse events involving therapeutic use of benzodiazepines (29,041, 95% CI=22,403, 35,679; 13.6%; 95% CI=11.4%, 15.9%).

Table 1.

Emergency Department Visits Attributed to Adverse Events Involving Benzodiazepines, by Patient Characteristics, 2016–2017

Patient characteristics	Nonmedical use of benzodiazepinesa			Self-harm using benzodiazepines			Therapeutic use of benzodiazepinesb
	Cases n	Annual estimate		Cases n	Annual estimate		Cases n	Annual estimate
		n	% (95% CI)		n	% (95% CI)		n	% (95% CI)
Patient age, yearsc
<5	0	—	—	0	—	—	106	2,617d	9.0 (6.9, 11.1)
5–14	50	1,215e	1.0e (0.3, 1.7)	36	697e	1.1e (0.4, 1.8)	30	—	—
15–24	902	30,929	26.0 (20.6, 31.4)	377	13,675	21.1 (17.9, 24.4)	61	1,761	6.1 (3.9, 8.2)
25–34	1,004	34,239	28.8 (24.8, 32.8)	339	12,481	19.3 (16.7, 21.9)	102	4,001	13.8 (10.7, 16.8)
35–44	562	18,076	15.2 (12.8, 17.5)	335	13,231	20.4 (17.8, 23.1)	84	2,893	10.0 (6.6, 13.3)
45–54	512	17,278	14.5 (12.5, 16.5)	320	11,938	18.4 (16.9, 20.0)	133	4,569	15.7 (12.7, 18.8)
55–64	381	12,740	10.7 (8.7, 12.7)	223	8,258	12.8 (10.4, 15.1)	138	5,173	17.8 (14.1, 21.5)
65–74	104	3,640	3.1 (2.0, 4.1)	76	3,016	4.7 (2.4, 6.9)	112	3,653	12.6 (9.0, 16.2)
≥75	22	838e	0.7 (0.3, 1.1)	33	1,426e	2.2 (0.9, 3.5)	104	4,059	14.0 (9.9, 18.1)
Patient sex
Female	1,459	50,877	42.8 (40.4, 45.1)	1,131	42,598	65.8 (62.7, 69.0)	499	16,272	56.0 (50.8, 61.3)
Male	2,080	68,131	57.2 (54.9, 59.6)	608	22,123	34.2 (31.0, 37.3)	371	12,769	44.0 (38.7, 49.2)
Dispositionf
Hospitalized	1,310	45,521	38.3 (27.1, 49.4)	1,468	52,929	81.8 (76.4, 87.1)	299	9,577	33.0 (26.7, 39.3)
Not hospitalized	2,229	73,487	61.7 (50.6, 72.9)	271	11,792	18.2 (12.9, 23.6)	571	19,464	67.0 (60.7, 73.3)
Total	3,539	119,008	100.0 (N/A)	1,739	64,721	100.0 (N/A)	870	29,041	100.0 (N/A)

Note: Data are from the National Electronic Injury Surveillance System-Cooperative Adverse Drug Event Surveillance project, Centers for Disease Control and Prevention. Estimates based on <20 cases or total estimates of <1,200 are considered statistically unstable and are not shown (—).

^aIncludes abuse, therapeutic misuse, and overdoses without documentation of therapeutic intent, self-harm, abuse, or misuse.

^bincludes therapeutic adverse drug events (e.g., adverse effects, allergic reactions, medication errors, and unsupervised ingestions by children).

^cAge missingfor 2 cases of nonmedical use.

^dIncludes 2,477 estimated emergency department visits for unsupervised ingestions.

^eCoefficient of variation >30%.

^fHospitalized includes inpatient admissions, observation admissions, and transfers to other hospitals. Not hospitalized includes patients treated and released and those who left against medical advice.

N/A, not applicable.

Most visits involving nonmedical use of benzodiazepines involved male patients (57.2%), whereas most visits involving self-harm (65.8%) and therapeutic use of benzodiazepines (56.0%) involved female patients. More than four fifths (81.8%) of estimated self-harm ED visits involving benzodiazepines resulted in hospitalization. The estimated annual number of hospitalizations involving nonmedical use of benzodiazepines (45,521, 95% CI=30,065, 60,978) was almost 5 times higher than the estimated number of hospitalizations for therapeutic use of benzodiazepines (9,577, 95% CI=6,297, 12,857), although hospitalization proportions were similar (38.3% vs 33.0%).

More than half of the visits involving nonmedical use of benzodiazepines were made by young patients aged 15–34 years (54.8%, 95% CI=49.8%, 59.8%). In this age group, the annual estimated number of visits involving nonmedical use (65,167, 95% CI=45,039, 85,296) was almost 11-fold higher than the estimated number of visits involving therapeutic use of benzodiazepines (5,762, 95% CI=4,243, 7,280). Among patients aged ≥65 years, half of the estimated visits involved therapeutic use (7,711, 95% CI=4,889, 10,535; 46.4%, 95% CI=40.4%, 52.3%). Among children aged <5 years, almost all ED visits involved unsupervised ingestions of benzodiazepines (2,477, 95% CI=1,741, 3,214; 94.7%, 95% CI=90.1%, 99.2%).

Population rates of estimated ED visits attributed to benzodiazepine adverse events varied by patient age and intent of benzodiazepine use (Figure 1). Visits involving nonmedical use of benzodiazepines by teenagers and young adults (aged 15–34 years) had the highest overall estimated population rate of ED visits per 10,000 people (7.4, 95% CI=5.1, 9.7). By contrast, among this age group, harms involving therapeutic benzodiazepine use led to only 0.6 (95% CI=0.5, 0.8) estimated visits per 10,000 people. The estimated population rates of visits involving nonmedical use of benzodiazepines declined after young adulthood, and rates of visits involving self-harm declined after middle age, whereas rates of visits involving therapeutic use increased with age to 1.9 visits per 10,000 people among those aged ≥75 years (95% CI=1.2, 4.1).

Figure 1.

Estimated U.S. population rates of emergency department visits attributed to adverse events involving benzodiazepines, by age, 2016–2017. Estimates are from the National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance project, Centers for Disease Control and Prevention. There were <20 cases or a national estimate of <1,200 emergency department visits (considered statistically unstable) for nonmedical use and self-harm by children aged <5 years and therapeutic use by children aged 5–14 years. The coefficient of variation was >30% for nonmedical use and self-harm visits involving children aged 5–14 years and for visits involving nonmedical use and self-harm for adults aged ≥75 years.

More than 80% of annual ED visits involving nonmedical use and self-harm with benzodiazepines also involved concurrent use of illicit drugs, alcohol, prescription opioids, or other pharmaceuticals (81.6%, 95% CI=79.3%, 84.0%; 149,992, 95% CI=114,680, 185,304). Among ED visits involving nonmedical use, a single additional class of substances was documented to be involved along with benzodiazepines in almost half of the visits (47.7%), and multiple additional classes of substances were documented to be involved in one third of the visits (35.0%) (Table 2). Similarly, among ED visits involving self-harm, 46.3% involved a single class of additional substances, and 33.3% involved multiple other classes of substances along with benzodiazepines. ED visits attributed to adverse events involving therapeutic use of benzodiazepines were more than twice as likely to be attributed to benzodiazepines alone (42.8%) than ED visits for non-medical use (17.2%) or ED visits for self-harm involving benzodiazepines (20.4%).

Table 2.

Emergency Department Visits Attributed to Adverse Events Involving Benzodiazepines, by Concurrent Substance, 2016–2017

Concurrent substance	Nonmedical use of benzodiazepines, annual estimate		Self-harm using benzodiazepines, annual estimate		Therapeutic use of benzodiazepines,a annual estimate
	n	% (95% CI)	n	% (95% CI)	n	% (95% CI)
Benzodiazepines onlyb	20,523	17.2 (14.6, 19.9)	13,214	20.4 (17.6, 23.2)	12,441	42.8 (37.1, 48.6)
Benzodiazepines and 1 other class of substance	56,839	47.7 (45.3, 50.3)	29,955	46.3 (43.6, 49.0)	13,288	45.8 (40.4, 51.1)
Illicit drugs onlyc	24,078	20.2 (17.3, 23.1)	4,689	7.2 (5.7, 8.8)	—	—
Alcohol only	14,096	11.8 (9.2, 14.4)	8,384	13.0 (10.6, 15.4)	1,004	3.5 (2.2, 4.7)
Prescription opioids only	12,217	10.3 (7.6, 13.0)	3,274	5.1 (4.0, 6.2)	3,902	13.4 (10.1, 16.8)
Nonopioid medications only	6,448	5.4 (4.1, 6.7)	13,608	21.0 (18.5, 23.5)	8,267	28.5 (23.5, 33.5)
Benzodiazepines and multiple classes of substances	41,646	35.0 (31.2, 38.8)	21,552	33.3 (29.9, 36.7)	3,312	11.4 (8.3, 14.5)
Nonopioid medication(s) and (illicit drugs or alcohol)	17,295	14.5 (12.3, 16.8)	13,051	20.2 (17.6, 22.7)	838	2.9 (1.5, 4.3)
Prescription opioid(s) and (illicit drugs or alcohol)	9,358	7.9 (5.6, 10.1)	1,658	2.6 (1.5, 3.6)	—	—
Prescription opioid(s) and nonopioid medication(s) and/or (illicit drugs or alcohol)	8,027	6.7 (5.3, 8.2)	4,850	7.4 (6.1, 8.9)	2,291	7.9 (4.9, 10.9)
Illicit drugs and alcohol	6,966	5.9 (4.6, 7.2)	1,995	3.1 (1.5, 4.6)	—	—
Total	119,008	100.0 (N/A)	64,721	100.0 (N/A)	29,041	100.0 (N/A)

Note: Data are from the National Electronic Injury Surveillance System-Cooperative Adverse Drug Event Surveillance project, Centers for Disease Control and Prevention. Estimates based on <20 cases or total estimates of <1,200 are considered statistically unstable and are not shown (—).

^aIncludes 2,477 estimated emergency department visits for unsupervised ingestions by children aged <5 years.

^bWithout documentation of involvement of other substances. Two different benzodiazepines were involved in 8 cases involving therapeutic use, 16 cases involving self-harm, and 24 cases involving nonmedical use.

^cincludes unspecified opioids and unspecified amphetamines. Does not include prescription opioids, nonopioid medications, or alcohol.

N/A, not applicable.

Other medications were more frequently co-implicated with benzodiazepines in ED visits involving therapeutic use (53.2%) and self-harm (49.7%) compared with ED visits involving nonmedical use of benzodiazepines (33.8%) (Table 3). Prescription opioids were the most commonly co-implicated medication class with benzodiazepines in ED visits involving nonmedical use (24.9%) and therapeutic use of benzodiazepines (21.9%). Antidepressants (15.6%) and prescription opioids (15.1%) were more frequently co-implicated with benzodiazepines than other medications in ED visits involving self-harm. Nearly half (47.8%) of visits involving nonmedical use of benzodiazepines occurred in the presence of illicit drugs, mostly commonly marijuana (22.1%), cocaine (13.2%), and heroin (13.0%). Illicit drugs were documented in a quarter (26.1%) of ED visits for self-harm involving benzodiazepines and were rarely documented in ED visits involving therapeutic use of benzodiazepines (2.8%).

Table 3.

Concurrent Substances Commonly Involved in Emergency Department Visits Attributed to Adverse Events Involving Benzodiazepines, 2016–2017

Concurrent substance	Nonmedical use of benzodiazepines, annual estimate		Self-harm using benzodiazepines, annual estimate		Therapeutic use of benzodiazepines, annual estimate
	n	% (95% CI)	n	% (95% CI)	n	% (95% CI)
Benzodiazepines and other medicationsa	40,191	33.8 (28.8, 38.7)	32,171	49.7 (46.7, 52.7)	15,458	53.2 (47.9, 58.5)
Prescription opioids	29,602	24.9 (19.8, 30.0)	9,781	15.1 (12.8, 17.4)	6,353	21.9 (17.3, 26.4)
Antidepressants	4,511	3.8 (2.4, 5.2)	10,081	15.6 (13.5, 17.6)	2,652	9.1 (7.0, 11.3)
Sedative hypnotics (nonbenzodiazepine)	3,713	3.1 (2.4, 3.8)	4,296	6.6 (5.2, 8.1)	1,108	3.8 (2.0, 5.7)
Antipsychotics	3,624	3.0 (2.5, 3.6)	5,404	8.3 (6.5, 10.2)	2,759	9.5 (5.9, 13.1)
Skeletal Muscle Relaxants	3,286	2.8 (1.6, 4.0)	2,636	4.1 (2.8, 5.3)	1,475d	5.1 (2.5, 7.6)
Anticonvulsants	3,032	2.5 (1.6, 3.5)	4,022	6.2 (4.5, 7.9)	2,058	7.1 (4.7, 9.5)
Amphetamine related stimulants	1,840	1.5 (1.1, 2.0)	1,197	1.8 (1.1, 2.6)	—	—
Other medications	18,342	15.4 (13.5, 17.3)	15,307	23.7 (21.6, 25.7)	4,719	16.2 (11.8, 20.7)
Benzodiazepines in the presence of alcoholb	31,453	26.4 (23.8, 29.1)	19,542	30.2 (26.9, 33.6)	1,600	5.5 (3.9, 7.1)
Benzodiazepines in the presence of illicit drugsc	56,843	47.8 (43.2, 52.3)	16,918	26.1(22.2, 30.1)	826	2.8 (1.3, 4.4)
Marijuana	26,314	22.1 (19.4, 24.9)	9,607	14.8 (11.5, 18.2)	709	2.4 (0.9, 3.9)
Cocaine	15,715	13.2 (10.0, 16.4)	4,124	6.4 (4.5, 8.3)	—	—
Heroin	15,469	13.0 (9.7, 16.3)	1,451	2.2 (1.0, 3.4)	—	—
Unspecified opioid	9,740	8.2 (6.3, 10.1)	2,411	3.7 (2.3, 5.1)	—	—
Methamphetamine	5,386	4.5 (2.4, 6.7)	1,176	1.8 (0.9, 2.7)	—	—
Unspecified amphetamine	4,434	3.7 (2.4, 5.1)	1,999	3.1 (1.9, 4.2)	—	—
Illicit fentanyl	1,466d	1.2d (0.3, 2.2)	—	—	—	—
Other illicit drugs	3,767	3.2 (2.6, 3.7)	—	—	—	—

Note: Estimates may add to more than 100% because multiple substances may be involved in a single emergency department visit. Data are from the National Electronic Injury Surveillance System–Cooperative Adverse Drug Event Surveillance project, Centers for Disease Control and Prevention. Estimates based on <20 cases or total estimates of <1,200 are considered statistically unstable and are not shown (—).

^aAlcohol and illicit drugs may also be involved.

^bPrescription medications and illicit drugs may also be involved.

^cPrescription medications and alcohol may also be involved.

^dCoefficient of variation >30%.

Clinical manifestations documented in ED visits attributed to adverse events involving benzodiazepines varied with intent and concurrent use of other substances (Appendix Table 2, available online). Cardiorespiratory arrest or unresponsiveness was documented in nearly one quarter (24.2%, 95% CI=17.7%, 30.6%) of ED visits involving nonmedical use of benzodiazepines, one eighth (13.2%) of visits involving therapeutic use, and one tenth (9.9%) involving self-harm. Among patients with documented cardiorespiratory arrest or unresponsiveness, 73.7% involved nonmedical use of benzodiazepines, 16.5% involved self-harm, and 9.8% involved therapeutic use (Appendix Table 3, available online). Cardiorespiratory arrest or unresponsiveness was documented in 30.9% of visits involving nonmedical use of illicit drugs with benzodiazepines and 29.6% of visits involving nonmedical use of prescription opioids with benzodiazepines, compared with 13.7% of visits involving nonmedical use of benzodiazepines alone (Appendix Table 4, available online).

Alprazolam was the most commonly implicated active ingredient in ED visits attributed to adverse events involving benzodiazepines (51.2%, 95% CI=45.6%, 56.7%), ranging from 31.2% of visits involving therapeutic use to 61.9% of visits involving nonmedical use (Appendix Table 5, available online).

DISCUSSION

From 2016 to 2017, there were an estimated 212,770 ED visits annually for adverse events involving benzodiazepines in the U.S., of which nearly 7 of 8 involved nonmedical use (56.0%) or self-harm (30.4%). Among visits for nonmedical use of benzodiazepines, about one fifth involved benzodiazepines alone, whereas the remainder involved other medications or occurred in the presence of other substances (illicit drugs or alcohol). With such high frequency of involvement of other medications and substances, the next step in reducing ED visits involving benzodiazepines may be identifying and addressing underlying substance use disorders and self-harm risk in patients before prescribing benzodiazepines.

Avoiding the prescription of benzodiazepines and opioids for the same patients, when possible, as recommended by product labeling, the CDC Guideline for Prescribing Opioids for Chronic Pain, and other guidelines,^3,19–22 is certainly a first step, but alone will not be sufficient to prevent most ED visits involving benzodiazepines. Benzodiazepines have been found to be the most frequently implicated pharmaceutical in ED visits for nonmedical use,¹¹ and opioids have been identified in more than 70% of deaths that involved benzodiazepines.^23,24 In this study of morbidity involving non-medical use of benzodiazepines, one quarter (24.9%) of visits involved co-implication of prescription opioids and benzodiazepines, one quarter (26.4%) involved alcohol and benzodiazepines, and almost half (47.8%) involved illicit drugs and benzodiazepines. Thus, concurrent misuse of substances other than prescription opioids is contributing to more ED visits than prescription opioids in combination with benzodiazepines.

Applying additional principles from the CDC Guideline for Prescribing Opioids for Chronic Pain may be appropriate for guiding benzodiazepine prescribing. The CDC Guideline, and now some health department guidelines for prescribing benzodiazepines,^25,26 recommends prescribers review a patient’s prescriptions for controlled substances using a prescription drug monitoring program and use urine testing for prescribed medications and illicit drugs during the course of opioid treatment.³ Identifying concurrent prescriptions for controlled substances in addition to opioids when patients are receiving benzodiazepines, and, in certain circumstances, laboratory testing²⁶ for other substances during the course of benzodiazepine treatment, may help reduce the risk of patient harm from drug overdoses involving benzodiazepines in combination with other drugs.²⁷

Screening patients for substance use (illicit drugs, prescription drugs, and alcohol²⁸) before prescribing benzodiazepines could also reduce harm from patients combining benzodiazepines with other substances. Polysubstance use, drug binging, heroin injection, and history of overdose are associated with nonmedical use of benzodiazepines.²⁹ Benzodiazepines may be used to enhance the high of opioids, to come down from other drugs such as crack or cocaine, to self-manage anxiety, and to self-treat withdrawal symptoms from opioids.²⁹ This study’s findings from data collected in 2016–2017 are remarkably consistent with Drug Abuse Warning Network data collected from 2004 to 2008, which also reported that nearly 80% of ED visits involving nonmedical use of benzodiazepines involved other drugs or alcohol, and peak incidence was in adolescents and young adults.¹⁰ Thus, involvement of benzodiazepines in polysubstance overdoses is not a new problem,^30–32 but an ongoing one that may require additional approaches to address.

Identifying self-harm risk before prescribing benzodiazepines could be an opportunity to reduce the 30% of benzodiazepine-related ED visits involving self-harm attempts, as some experts similarly have suggested screening for self-harm risk before prescribing opioids.³³ Unfortunately, the diagnostic accuracy of screening tests for self-harm risk in the general population is not high enough for screening to be recommended by the U.S. Preventive Services Task Force.³⁴ The frequent involvement of benzodiazepines in ED visits for self-harm attempts is not a new finding,³⁵ but the continued frequency of benzodiazepine involvement reinforces the need for validated tools that could be used for identifying self-harm risk among patients prescribed benzodiazepines.

Education on the consequences of nonmedical use of benzodiazepines may need to be targeted toward teens, young adults, and prescribers. In this study, young people aged 15–34 years were most impacted by adverse events involving nonmedical use of benzodiazepines (54.8% of ED visits, with an estimated rate of 7.4 ED visits per 10,000 people). The means of obtaining benzodiazepines (e.g., prescribed, from relatives or friends, or illicitly purchased) was not able to be assessed in this study, but based on a survey of young adults in New York City, benzodiazepines were described as readily available and inexpensive to obtain.²⁹ Therefore, it is also important for healthcare prescribers to ask about and caution patients against using benzodiazepines obtained from other sources, particularly in combination with opioid analgesics or other central nervous system depressants. There is also evidence that school- and family-based programs may have positive long-term effects in preventing alcohol and drug misuse,^36,37 and teaching children and adolescents coping and problem solving skills can reduce both suicide and substance use later in life.³⁸

Both older patients and prescribers should receive education on the safe use of benzodiazepines. Annually, an estimated 16,632 ED visits were made by patients aged ≥65 years. Although the American Geriatrics Society recommends avoiding benzodiazepines in older adults due to increased sensitivity, benzodiazepines remain widely used in this age group.⁵ Another estimated 2,477 ED visits occurred in children aged <5 years for unsupervised ingestion of benzodiazepines. Adults who have children at home should be reminded about safe storage and consequences of unintentional ingestion.^39,40

Given the nearly 120,000 annual estimated ED visits involving nonmedical use and nearly 65,000 annual estimated visits for self-harm involving benzodiazepines, the ED setting can provide an opportunity for secondary prevention, for example, linking ED patients to substance abuse treatment services before discharge.⁴¹ Though more than 80% of ED patients treated for self-harm involving benzodiazepines were hospitalized, those who are treated and released may need psychiatric and social assessment and close follow-up, as the first episode of self-harm is a strong predictor for subsequent suicide and death.^35,42–44 Patients whose ED visits were attributed to nonmedical use rather than self-harm may also benefit from mental health assessment and treatment,⁴⁵ as illicit drug use and medication overdoses are often associated with mental illnesses.⁴⁶

Limitations

When patients use more than 1 substance it can be complicated to attribute symptoms to specific substances; nonetheless, clinicians make such determinations to guide treatment. This public health surveillance system is based on the documentation of treating clinicians and likely underestimates the number of urgent, unplanned visits for adverse events involving benzodiazepine use. The NEISS–CADES surveillance system collects information about ED visits attributed only to acute adverse events; therefore, totals do not include patients visiting other healthcare settings (e.g., urgent care clinics) or choosing to not seek medical attention for less severe symptoms (e. g., dizziness or somnolence); however, there might be repeated visits made by the same patient. Totals do not include patients with harms that were not acute effects of active benzodiazepine use (e.g., withdrawal, patients seeking substance use disorder treatment, and violence-related injuries) and patients who may have died in or en route to the ED. Totals do not include an additional 1,529 cases in which benzodiazepines were identified by a positive laboratory test but were not implicated in the adverse event by the treating clinician. In addition, although the documentation of the intent of drug use by treating clinicians and subsequent classification (nonmedical use, self-harm, or therapeutic) has not been validated, classification based on clinician reporting is used widely in regulatory and research activities^47–49; nonetheless, misclassification may occur. Patients may misreport intentionality (e.g., patients may report taking additional benzodiazepines for intractable insomnia, whereas their actual intent may be recreational use or self-harm), clinicians may not fully document intentionality or details surrounding the event, and patients may arrive to the ED unresponsive and unable to report intent of drug use. Finally, use of all substances may not be reported, and substances may be misclassified due to cross-reactivity in laboratory testing or may not be detected by some screening panels.

CONCLUSIONS

Benzodiazepines are implicated frequently in ED visits for adverse events, particularly involving nonmedical use and self-harm, with highest rates among teens and young adults. The findings that nearly 7 of 8 estimated ED visits attributed to adverse events involved nonmedical use of benzodiazepines or self-harm and more than 80% of these visits involve concurrent use of other substances, support recommendations to assess for and address substance use disorders before initiating or continuing benzodiazepines, and reinforce the need for validated self-harm risk assessment tools for clinicians.