Older adults’ marijuana use, injuries, and emergency department visits

ABSTRACT

Background: Despite increasing marijuana use among older adults, little research has been done on marijuana’s effects on their healthcare use. Objectives: To examine whether (1) marijuana use is associated with the likelihood of emergency department (ED) visits through increased injury risk, and (2) marijuana use patterns are associated with injury risk and ED visits among older adults. Methods: Using the 2012–2013 National Epidemiologic Survey on Alcohol and Related Problems (N = 14,715 aged 50+; 6,379 men and 8,336 women), descriptive statistics were used to compare past-year marijuana users to nonusers on sociodemographic and physical/mental health characteristics. Structural equation modeling was used to simultaneously estimate whether injury mediates past-year marijuana use’s association with ED visits. Logistic regression analysis was used to examine associations of marijuana use patterns with injury and ED visits. Results: Past-year injury rates were 18.9% for nonusers and 28.8% for users (3.9% of the 50+ age group), and past-year ED visit rates were 23.5% for nonusers and 30.9% for users (p < .001). Logistic regression models showed that marijuana use was associated with injury (OR = 1.48, 95% CI = 1.18–1.85), and injury was associated with ED visits (OR = 6.14, 95% CI = 1.70–1.93). Mediation analysis found significant indirect effect (z = 2.86, p = .004) and direct effect not significantly differing from zero (OR = 1.16, 95% CI = 0.90–1.50), indicating that marijuana use increases the likelihood of ED visits through increased injury risk. Marijuana use patterns were not associated with injuries or ED visits. Conclusions: Healthcare providers should screen for marijuana and other substance use among older adults and provide education about associated injury risks.

KEYWORDS:

• Marijuana
• older adults
• injury
• ED visits

Introduction

After alcohol and tobacco, marijuana is the most commonly used psychoactive substance in the United States (1). Recent epidemiological data show that marijuana use has increased significantly over the past decade among all demographic subgroups, with especially notable increases among middle-aged and older adults (2–4). A study based on the 2012–2013 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC-III) found that 3.9% of the 50+ age group reported past-year marijuana use (5). The increasing marijuana use rate among older adults has been attributed to the aging baby boomers who have had greater exposure to marijuana; more permissive attitudes toward its recreational use than preceding generations; and marijuana’s increased availability of and accessibility to marijuana owing to legalization of its medicinal and/or recreational use (6–8). Gallup polls show that in 2003 and 2005, 29% of 55+ year olds supported legalizing marijuana; by 2016, the figure has risen to 45% (9). At the beginning of 2017, 28 states and the District of Columbia had laws legalizing medical and/or recreational marijuana use.

A majority of marijuana users aged 50+, including medical marijuana users, are long-term users who initiated use early in life; about one in six users had past-year marijuana use disorder; almost 90% used alcohol; and nearly a quarter also used other illicit drugs (5). Another study noted that one-third to more than one-half of registered medical marijuana participants in eight states were 50+ years old (10). Along with symptom relief (e.g., from chronic pain, anxiety, and other health/mental health conditions), baby-boomers who use marijuana claim they do so for relaxation/stress reduction, social/recreational reasons, and enhancement/expansion (e.g., heightened awareness, concentration, creativity) (11). A majority of older users perceive no or only slight health-related or other harm from frequent (e.g., several times a week) marijuana use (6,12). However, research findings show that both short-term marijuana use and early onset and chronic/long-term marijuana use are associated with more adverse health, mental health, and cognitive health outcomes than benefits (13,14).

Previous research also shows significant associations between marijuana use and all types of injury among younger and middle-aged groups. Adjusting for age, alcohol use, and other potential confounders, both self-reported and/or toxicology-determined acute and/or chronic marijuana use have been found to be independently associated with an increased likelihood of fatal and nonfatal motor vehicle collision/injuries, assault injuries that resulted in hospitalization, and other types of unintentional and intentional injuries in both nonclinical and clinical samples (15–17). Evidence from multiple studies led the World Health Organization (WHO) to posit that marijuana use has a small causal impact on traffic injury; however, WHO summarized that evidence on marijuana’s effect on nontraffic injuries is less conclusive (18). Although the exact mechanisms of injury in marijuana use are not known, marijuana-induced agitation/aggression, psychosis, anxiety, and other acute intoxicating effects on consciousness, cognition, perception, affect and behaviors, and other psychophysiological functions and responses, especially when combined with alcohol and other illicit drug use, may increase vulnerability to injury (18,19). Emergency department (ED) patients who admitted to alcohol and marijuana use also tended to use other hazardous substances and admitted to participating in high-risk behaviors (20).

In late life, continued marijuana use, with or without other substance use, may have particularly adverse consequences, precipitating or worsening various health conditions and increasing vulnerability to all types of injuries, which in turn can lead to more ED and other healthcare utilization compared to those who do not use marijuana. A recent population-based study found that of past-year substance users in the 50–64 and the 65+ age groups, 14.5% and 6.2%, respectively, self-reported driving under the influence (DUI), and that after adjusting for health and mental health conditions, frequency of alcohol use, and other substance use, marijuana use was a significant predictor of DUI reports (odds ratio [OR] = 3.12, 95% confidence interval [CI] = 2.54–3.82, p < .001 for the 50–64 age group; OR = 4.94, 95% CI = 2.12–11.52, p < .001 for the 65+ age group) (21). A study of older trauma patients’ (aged 55+) ED outcomes also found that those who screened positive for marijuana (or cocaine) had significantly higher odds of ICU admission and surgery than those with negative screens (22).

Overall, however, previous research on the relationship between marijuana and injury has focused on younger age groups with scant attention to older adults’ marijuana use and its relationship to injuries and healthcare utilization. The purposes of the present study were to examine whether: (1) past-year marijuana users aged 50+ differed from their non-using age peers with respect to physical and mental health status and past-year prevalence of injury and ED visits; and (2) marijuana use was associated with ED visits with injury as a mediator. Our literature review led to the following hypotheses: (Hypothesis 1) older adults’ marijuana use will be significantly associated with a higher rate of injury, which in turn will be significantly related to ED visits, and (Hypothesis 2) among older adult marijuana users, younger age at first use and greater frequency and quantity of use will be associated with a higher likelihood of injury and ED visits. In the absence of research on marijuana use’s effects on health and healthcare service use in late life, this study is the first to examine marijuana use’s association with injury and ED visits among the 50+ age group.

Methods

Data and sample

Data came from the 2012–2013 NESARC- III, a national probability sample survey of the U.S. civilian noninstitutionalized population aged 18+ years (N = 36,309), sponsored by the U.S. National Institute on Alcohol Abuse and Alcoholism (NIAAA). Using computer-assisted personal interviewing, data were collected on alcohol and other substance use, substance use disorders, and related physical and mental disabilities. The semi-structured diagnostic interview used to collect information was NIAAA’s Alcohol Use Disorder and Associated Disabilities Interview Schedule (AUDADIS-5). In NESARC-III’s multistage probability sampling, primary sampling units were individual counties or, in some small rural counties, combined contiguous counties; secondary sampling units were groups of Census-defined blocks; and tertiary sampling units were households within sampled secondary sampling units, from which eligible adult respondents were randomly selected, with Hispanic, Black, and Asian individuals oversampled (23). The sample for this study consisted of 14,715 individuals aged 50+, after excluding 23 respondents for whom marijuana use/nonuse data were missing. The authors received NIAAA’s approval to use the public domain, NESARC-III data.

Measures

Marijuana use and use disorder

We focused on past-year marijuana use (yes or no). Marijuana use disorder was identified based on criteria found in the Diagnostic and Statistical Manual of Mental Disorders, fifth ed (DSM-5). For marijuana users, information is also included about their initiation and peak use ages, duration of peak use period, number of joints usually smoked in a day during the peak use period and in the past 12 months, frequency of use, and medical marijuana recommendation status. In NESARC, frequency of use was coded as every day; nearly every day; 3–4 times a week; 1–2 times a week; 2–3 times a month; once a month; 7–11 times in the last year; 3–6 times in the last year; 2 times in the last year; or once in the last year. In this study, we combined these 10 categories into three: (1) every day/nearly every day; (2) 1–4 times a week to 1–3 times a month; and (3) 1–11 times in the past year.

Injury status

Respondents were asked about the number of injuries they had in the past 12 months that “caused seeking medical help or cutting down usual activities for more than half a day.” Injury status in this study was coded as 1 (had one or more injury incidents) versus 0 (no injury).

ED visits

Respondents were asked about the number of times they were treated in an ED in the past 12 months. ED visit status was coded as 1 (had one or more ED visits) versus 0 (none).

Covariates

(1) Sociodemographic characteristics included age, gender, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic Asian/Pacific Islander [Asian American hereafter], and American Indian/Alaska Native [American Indian hereafter]), and Medicaid coverage as a proxy for low-income status. (2) Health status was represented by the number of diagnosed chronic medical conditions (arthritis, cancer, diabetes, hypertension, heart disease, stroke, liver disease, and lung disease). (3) Past-year DSM-5 other substance use disorders included alcohol use disorder, other (than marijuana) drug use disorder (sedatives/tranquilizers, painkillers, cocaine/crack, stimulants, club drugs, hallucinogens/psychedelics, inhalants/solvents, heroin, and other drugs/medicines, including antidepressants, antipsychotic drugs, steroids, and any other medicines or drugs used for nonmedical reasons), and nicotine use disorder. (4) Past-year mental disorder included any of the following DSM-5 diagnoses: major depressive disorder; any anxiety disorder (specific phobia, social phobia, panic disorder, agoraphobia, or generalized anxiety disorder); post-traumatic stress disorder (PTSD); and bipolar 1 disorder or manic/hypomanic episode. Other demographic characteristics (marital status, education, and employment) and self-rated health are presented for sample description purposes only.

Analysis

All descriptive analyses were performed with Stata/MP 14’s svy function to account for NESARC-III’s multi-stage probability sampling design. Stata’s subpop command was used for all subsample analyses (e.g., those aged 50+, past-year marijuana users) to ensure that variance estimates incorporate the full sampling design. All estimates presented in this study are weighted except sample sizes. We used χ² and t tests to examine whether past-year marijuana users and nonusers differed on sociodemographic characteristics, health status, substance use and mental disorders, injury, and ED visit status. Univariate frequencies were used to describe past-year marijuana users’ ages at first use and peak use, duration of peak use, and number of joints smoked during peak use and in the past 12 months. Hypothesis 1 (association of past-year marijuana use with the risk of any injury and ED visit) was tested with structural equation modeling (SEM) using Mplus 7.4 (24) using the same survey design features described in the Stata models above to simultaneously estimate whether injury mediates past-year marijuana use’s association with ED visits (indirect effect). We tested the indirect effect in a path model following MacKinnon’s guidelines (25), which specify two critical precursor conditions for mediation: (a) a relationship between the independent variable (marijuana use) and the mediator (injury) and (b) a relationship between the mediator (injury) and the outcome (ED visit). The indirect effect was evaluated using counterfactually based hypothesis tests for evaluating mediation effects (26). The counterfactual-based hypothesis test (26) represents a previously unavailable method for evaluating indirect effects for binary outcomes by evaluating the difference between observed outcomes and outcomes that would have been observed under opposite conditions (e.g., the predicted ED visit outcome had a marijuana user not been a marijuana user).

The mediation model included all covariates described in the measures section. These covariates were also included in the models evaluating precursor conditions. Hypothesis 2 (associations of injuries and ED visits with marijuana initiation age and marijuana use frequency and quantity among past-year marijuana users) was tested using multivariable binary logistic regression analysis. Results are presented using ORs and 95% CIs. The significance level set for all statistical tests was p < .05. Asian Americans were excluded in testing Hypothesis 2 due to their small sample size (n = 4).

Results

Sample characteristics

Table 1 shows that 3.9% of the respondents aged 50+ (5.6% of the 50–64 year olds and 1.3% of the 65+ year olds) reported using marijuana in the past 12 months. Compared to nonusers, past-year users were about six years younger and more likely to be male and non-Hispanic Black or American Indian but less likely to be Hispanic or Asian. Users were also less likely to be married or widowed and more likely to be divorced or never married. Despite their younger age, users were less likely to have a college degree and to be employed, and almost 16% (compared to 8% of nonusers) had Medicaid coverage. With respect to health status, users did not differ from nonusers in the number of chronic medical conditions, but they rated their health lower. Almost one-third of users, compared to less than 20% of nonusers had a mental disorder/s. Significant differences were also found in substance use disorders. Of the marijuana users, 29.3% had alcohol use disorder, 47.5% had nicotine use disorder, and 23.3% had a drug use disorder, including 17.5% who had marijuana use disorder, while among those who did not use marijuana, 5.4% had alcohol use disorder, 14.3% had nicotine use disorder, and 0.9% had any drug use disorder.

Table 1. Sample sociodemographic, physical and mental health, and emergency department (ED) visit characteristics by past-year marijuana use status.

	Nonusers^1 96.13% (n = 14,020)	Users^2 3.87% (n = 695)	p
Age (yrs; M,SE)^3	63.8 (0.13)	57.6 (0.33)	<.001
Age group (%)			<.001
50–64 yrs	58.56	86.24
65+ yrs	41.44	13.76
Male (%)	45.89	67.59	<.001
Race/ethnicity (%)			<.001
Non-Hispanic White	75.13	73.24
Non-Hispanic Black	9.81	15.60
Hispanic	9.13	6.31
Non-Hispanic Asian	4.63	0.79
American Indian	1.31	4.06
Marital status (%)			<.001
Married/cohabiting	64.20	43.19
Widowed	12.91	6.28
Divorced/separated	16.70	37.75
Never married	6.19	12.78
College degree (%)	29.75	20.78	<.001
Medicaid coverage (%)	8.03	15.85	<.001
No. of chronic medical conditions^4 (M, SE)	2.04 (0.04)	1.95 (0.20)	.669
Self-rated health (1 = poor, 5 = excellent; M, SE)	3.32 (0.02)	3.04 (0.06)	<.001
Major depressive disorder (%)	8.09	17.33	<.001
Anxiety disorder^5 (%)	11.91	22.98	<.001
Post-traumatic stress disorder (%)	3.20	9.44	<.001
Bipolar, manic, or hypomanic disorder (%)	1.15	3.82	<.001
Any of the above past-year mental disorders (%)	18.86	33.23	<.001
Alcohol use disorder (%)	5.41	29.26	<.001
Nicotine use disorder (%)	14.34	47.47	<.001
Marijuana use disorder (%)	n/a	17.54
Other drug use disorder (%)	0.88	8.40	<.001
Past-year any injury^6 (%)	18.94	28.84	<.001
No. of injuries among the injured^7 (M, SE)	1.53 (0.03)	1.97 (0.25)	.088
Past-year ED visit (%)	23.50	30.86	<.001
ED visit only	11.47	16.48	<.001
Both ED visit and hospitalization	12.03	14.38	.169
No. of ED visits among visitors^8 (M, SE)	1.70 (0.03)	1.83 (0.14)	.380

¹ Never used marijuana (74.42%) or used it prior to the past 12 months only (21.71%).

² Used marijuana in the past 12 months.

³ M = Mean; SE = Standard error.

⁴ Includes arthritis, cancer, diabetes, hypertension, heart disease, stroke, liver disease, and lung disease.

⁵ Includes specific phobia, social phobia, panic disorder, agoraphobia, and generalized anxiety disorder.

⁶ Any injury in past 12 months that caused the sample person to seek medical help or cut down usual activities for more than half a day.

⁷ n = 2,731 for the paired comparison of nonusers vs. users.

⁸ n = 3,653 for the paired comparison of nonusers vs. users.

Compared to 18.9% of nonusers, 28.8% of marijuana users had an injury (p < .001) in the past 12 months that caused seeking medical help or limiting usual activities for more than half a day. Of the injured, the number of times injuries occurred did not differ between nonusers and users (p = .088). Marijuana users also had higher rates of ED visits than nonusers (30.9% and 23.5%, respectively, p < .001). Of those who visited an ED, approximately half of each group was hospitalized (p = .357). Among those who visited an ED and/or had a hospital admission, users and nonusers did not differ on the number of ED visits (p = .380) or hospital admissions (p = .963).

Marijuana use patterns among past-year users

Table 2 shows that on average, past-year marijuana users initiated use in their late teens. Their age at first use and starting age of peak use did not differ. The duration of their peak use period averaged more than 13 years. Though a third of past-year users appeared to be occasional users (i.e., used marijuana less than 12 times in the past 12 months), 26.1% reported having used it every day or nearly every day. Data also show that 12.7% of marijuana users (12.3% of the 50–64 year olds and 15.8% of the 65+ year olds) reported marijuana use for medical purposes and more than half (53.1%) of these self-reported medical users used it every day or nearly every day.

Table 2. Marijuana use patterns of past-year marijuana users.

Measures for marijuana use pattern
Age first used (yrs; M,SE)^1	19.0 (0.39)
Age when began using the most (yrs; M, SE)^1	19.0 (0.39)
No. of joints usually smoked in a day when using the most (M, SE)^1	3.19 (0.20)
Duration (weeks) used the most (M, SE)^1	691.83 (30.98)
Frequency of use in past 12 months (%)
Every day/nearly every day	26.12
1–4 times a week to 1–3 times a month	39.96
1–11 times past year	33.92
No. of joints usually smoked in a day in past 12 months (M, SE)^1	1.67 (0.09)
Past-year medical marijuana user (%)	12.74

Due to missing data, sample sizes varied between 691 and 676

for the examined variables.

¹ M = Mean; SE = Standard error.

Additional analyses found that marijuana use patterns did not differ between those with and without injuries or between those who had and did not have ED visits, except that more of those with than without injury (19.2% vs. 10.1%) used marijuana medically (along with nonmedical use) (p = .002). Additional analyses also found that medical users did not differ from nonmedical users in age, race/ethnicity, or education, but they were more likely to have Medicaid coverage (25.8% vs. 14.4%, p = .011) and had more medical conditions (1.56 [SE = 0.15] vs. 1.10 [SE = 0.07], p = .006) and a higher rate of mental disorders (43.1% vs. 31.8%, p < .030). Though not statistically different, medical users also had a higher rate of past-year marijuana use disorder (22.5% vs. 16.8%, p = .335). Medical users were more likely to report an injury (43.4% vs. 26.7%, p = .002), but they were not more likely to have had ED visits and hospital admissions.

Association of marijuana use, injuries, and ED visits: Structural equation modeling results

Preliminary logistic regression models established that both precursor conditions to mediation were present in the data: marijuana use was associated with injuries (OR = 1.48, 95% CI = 1.18–1.85, z = 3.45, p < .001), and injuries were associated with ED visits (OR = 6.14, 95% CI = 5.48–6.68, z = 31.18, p < .001).¹ Table 3 presents results for the equations in the mediation model. The direct effect of marijuana use on ED visits in in the model did not significantly differ from zero (OR = 1.16, 95% CI = 0.90–1.50, z = 1.15, p = .249), indicating no direct relationship between marijuana use and ED visits. However, the indirect effect, which tests the mediation hypothesis, was significant (z = 2.86, p = .004), supporting Hypothesis 1 that marijuana use increases the likelihood of ED visits through the mediator, increased injury risk.

Table 3. Effects of marijuana use on emergency department (ED) visits through injury: Model parameters from mediation model using structural equation modeling.

Model parameter	OR (95% CI)^1	z	p
ED visit
Injury	6.12 (5.46–6.86)	31.12	<.001
Marijuana use	1.16 (0.90–1.50)	1.15	.249
Age	1.01 (1.00–1.02)	3.50	<.001
Male	0.95 (0.85–1.05)	−1.03	.302
Non-Hispanic black	1.61 (1.42–1.83)	7.37	<.001
Hispanic	1.08 (0.92–1.27)	0.96	.338
Non-Hispanic Asian	1.02 (0.80–1.29)	0.14	.889
American Indian	1.05 (0.70–1.59)	0.25	.802
Medicaid	1.59 (1.36–1.86)	5.80	<.001
No. of chronic medical conditions	1.50 (1.44–1.57)	18.75	<.001
Alcohol use disorder	1.19 (0.95–1.47)	1.54	.124
Nicotine use disorder	1.05 (0.92–1.19)	0.74	.459
Other drug use disorder	1.41 (0.92–2.17)	1.56	.119
Any mental disorder	1.22 (1.07–1.39)	3.07	.002
Injury
Marijuana use	1.48 (1.18–1.85)	3.45	.001
Age	0.99 (0.99–1.00)	−2.94	.003
Male	0.99 (0.90–1.10)	−0.11	.911
Non-Hispanic black	0.77 (0.67–0.89)	−3.62	<.001
Hispanic	0.83 (0.71–0.97)	−2.32	.020
Non-Hispanic Asian	0.60 (0.46–0.80)	−3.55	<.001
American Indian	1.31 (0.91–1.90)	1.44	.151
Medicaid	1.00 (0.85–1.18)	0.02	.984
No. of chronic medical conditions	1.20 (1.15–1.25)	8.60	<.001
Alcohol use disorder	1.15 (0.94–1.39)	1.36	.176
Nicotine use disorder	1.01 (0.89–1.15)	0.22	.825
Other drug use disorder	1.48 (1.03–2.11)	2.14	.033
Any mental disorder	1.43 (1.26–1.61)	5.68	<.001

Reference group for non-Hispanic Black, Hispanic, non-Hispanic Asian, and American Indian

was non-Hispanic White.

¹ OR = Odds ratios; CI = Confidence intervals.

Table 3 also shows that of the covariates, older age and belonging to any of the three minority groups (as opposed to being non-Hispanic White) were associated with a lower likelihood of having any injury, but more chronic medical conditions, having other drug use disorder, and having any mental disorder increased the likelihood. Older age, being non-Hispanic Black, and having Medicaid coverage, more chronic medical conditions, and any mental disorder were associated with an increased likelihood of an ED visit.

Association of marijuana use patterns with injuries and ED visits among past-year marijuana users

Table 4 shows that our multivariable binary logistic regression analyses found no significant association between marijuana use patterns (initiation age and frequency and amount of use) and injury or ED visit, thus failing to support Hypothesis 2. In these analyses, the only significant correlate for both injury and ED visits was the number of chronic medical conditions.

Table 4. Logistic regression results: Association of marijuana use patterns with injuries and emergency department (ED) visits among past-year marijuana users.

	Injury vs. no injury OR (95% CI)^1	ED visit vs. no ED visit OR (95% CI)^1
Age	0.98 (0.94–1.02)	0.96 (0.92–0.99)*
Male	1.12 (0.73–1.71)	0.87 (0.56–1.37)
Race/ethnicity
(Non-Hispanic White)
Non-Hispanic Black	0.70 (0.42–1.18)	1.86 (1.14–3.05)
Hispanic	1.13 (0.56–2.26)	0.97 (0.43–2.19)
American Indian	0.81 (0.29–2.25)	0.33 (0.10–1.05)
Medicaid	1.21 (0.67–2.17)	1.29 (0.77–2.16)
No. of chronic medical conditions	1.27 (1.07–1.50)**	1.52 (1.23–1.87)***
Alcohol use disorder	1.60 (0.95–2.68)	0.90 (0.52–1.57)
Nicotine use disorder	1.11 (0.64–1.94)	1.08 (0.66–1.76)
Other drug use disorder	1.12 (0.53–2.36)	1.78 (0.82–3.87)
Any mental disorder	1.59 (0.93–2.73)	1.51 (0.95–2.41)
Age at first marijuana use	1.01 (0.97–1.04)	1.02 (0.99–1.06)
Frequency of marijuana use in past 12 months
(1 to 11 times past year)
Every day/nearly every day	0.69 (0.41–1.17)	0.89 (0.48–1.64)
1–4 times a week to 1–3 times a month	1.10 (0.70–1.72)	0.79 (0.45–1.40)
No. of joints usually smoked in a day in past 12 months	1.02 (0.89–1.16)	0.98 (0.86–1.12)
	N = 668; Model F (15,97) = 2.22,design df = 111, p = .010	N = 668; Model F (15,97) = 3.08,design df = 111, p < .001

¹ OR = Odds ratios; CI = Confidence intervals.

***p < .001; **p < .01; *p < .05.

Discussion

Using data from a U.S. epidemiologic survey, we examined whether marijuana use and marijuana use patterns among individuals aged 50+ were associated with injury and ED visits. Compared to nonusers, past-year users (3.9% of the age group) were younger and more likely to be Black, American Indian, divorced, or never married. Based on education level, employment status, and Medicaid coverage, users had lower socioeconomic status than nonusers. Marijuana users also had higher rates of other substance use and mental disorders than nonusers. Despite their younger age, marijuana users did not differ from nonusers in number of chronic medical conditions, but they had higher rates of injuries and ED visits. The findings also show that on average, these marijuana users began using during their teen years and that their initiation and peak use age were the same, suggesting that use peaked shortly after initiation.

This study’s key finding is that marijuana use was associated with a higher rate of injury, which in turn led to a higher rate of ED visits among the 50+ age group, even when other potential risk factors for injury and/or ED visits in late life, including chronic medical conditions and other substance use and mental disorders were included as covariates. Our findings show that marijuana use has effects independent of other substance use disorders. However, among marijuana users, marijuana use frequency and quantity were not associated with injury or ED visits, suggesting that marijuana use, regardless of frequency and amount, is a risk factor for injury and higher healthcare utilization in this age group.

The exact mechanisms of injury due to marijuana use are not known; however, marijuana-induced agitation/aggression, psychosis, anxiety, and other intoxicating effects on consciousness, cognition, perception, affect, motor coordination, behaviors, and other psychophysiological functions and responses are likely to increase vulnerability to injury (18). With steadily increased tetrahydrocannabinol (THC) potency (27), older adults may be even more susceptible to these adverse effects. A national survey in Spain found that adjusting for sociodemographic factors and alcohol, tobacco and other drug use, the association between marijuana use in the past 12 months and prevalence of non-traffic injuries was greater in the 35–64 age group than in the 15–34 age group (17). Contrary to our findings of the lack of association between injury and frequency and quantity of marijuana use, this Spanish study found that weekly marijuana use in the 35–64 age group was especially strongly associated with “knocks and bumps.” More research is needed to examine the effects of age, health conditions, THC potency, and other confounding factors on the association between marijuana use patterns and different types of injury. Marijuana combined with alcohol and other illicit drug use is likely to increase vulnerability to injury in late life (17,18). As discussed, regardless of age, ED patients who admitted to alcohol and marijuana use also tend to use other hazardous substances and participate in high-risk behaviors (20) that could result in injury.

There are some study limitations. First, since NESARC-III data do not distinguish between traffic and nontraffic injuries, injury-specific analysis could not be done. Second, the small sample size of older marijuana users prohibited analyses of subgroups (e.g., those with marijuana use disorder, within genders). Third, because data on marijuana use frequency and quantity were self-reported, social desirability and recall bias may have affected validity of the data. Fourth, and perhaps most importantly, in this cross-sectional study, causation cannot be determined. Ideally, in a mediation model, it would be clear that marijuana use preceded injury; however, since marijuana use is typically a reoccurring phenomenon in contrast to injuries, which are rare events, we may assume that marijuana use was generally a precursor to injuries.

Despite these limitations, the study has important clinical and research implications. First, our findings of higher rates of injury and ED visits among older marijuana users than nonusers should be considered along with previous studies’ findings. For example, older marijuana users perceive little risk associated with use, and a significant number of older, regular marijuana users use it to self-medicate a wide variety of somatic and psychiatric conditions without a physician’s recommendation (5,11,28). This growing body of research underscores the need for primary care physicians to screen older adults, especially those who present with injuries, for marijuana and other substance use and provide education about injury prevention and about these substances’ adverse physical and mental health effects. Second, older-adult ED patients who screen positive for marijuana and other substance use and disorders should be referred for further diagnostic evaluation and to treatment as needed. Given that nearly one-quarter of medical marijuana users had a marijuana use disorder in the past year, the ethics of recommending or supporting marijuana use for older individuals, especially those who present with marijuana-related problems, should be addressed, especially when other remedies for physical and mental health problems are available. Third, more research is needed to examine associations of marijuana use and use patterns with injury and healthcare service use in late life. In particular, researchers should investigate marijuana’s effects on both traffic and nontraffic injuries (e.g., falls, nonsuicidal self-injury, suicide attempts) among older adults given the personal, social, and economic costs of these injuries in late life. Fourth, future research also needs to examine the confounding effect of alcohol and drug (other than marijuana) use that did not meet diagnostic criteria for use disorder on the association between marijuana use and injury and ED visits. Even if diagnostic criteria were not met, alcohol and/or illicit drug use may enhance marijuana’s effects on injury and ED visits. Fifth, because older adults are the most frequent users of healthcare services, longitudinal research should also examine the effects of long-term marijuana use on physical and mental health status and service use in late life.

Declaration of interest

The authors report no relevant financial conflicts.

Acknowledgments

This manuscript was prepared using a limited access data set obtained from the National Institute on Alcohol Abuse and Alcoholism and does not reflect the opinions or views of NIAAA or the U.S. Government. The authors also acknowledge the University of Texas at Austin Addiction Research Institute for support of this work.

Funding

No grant funding was received.

Additional information

Funding

No grant funding was received.

Notes on contributors

Namkee G. Choi

All four authors conceptualized the paper, contributed to the literature review, wrote the manuscript, and approved the final version. The first and second authors conducted the statistical analysis.

Notes

¹ Coefficients in the preliminary models were consistent with the model parameters in Table 3 in the direction of effect and in terms of statistical significance.