Sex differences in Cannabis Use Disorder and associated psychosocial problems among US adults, 2012-2013

Abstract

While men show greater prevalence of cannabis use disorder (CUD) than women, whether cannabis use frequency drives this difference is unknown, and little is known about sex differences in problems associated with CUD. We therefore assessed the association of CUD with sex, adjusted for frequency of use, and compared the association of psychosocial and health-related problems with CUD between men and women. We included US adults age ≥18 who reported past-year cannabis use in the 2012-2013 National Epidemiologic Survey on Alcohol and Related Conditions-III (n=3,701). Cannabis use frequency, DSM-5 CUD and problems (interpersonal, financial, legal, health-related) were assessed. Associations between psychosocial problems, sex and DSM-5 CUD were assessed using prevalence differences (PD) and 95% confidence intervals (CI) from logistic regression models, controlling for demographics and cannabis use frequency, and effect modification by sex was assessed. We found that the prevalence of CUD among men versus women was not significantly greater after adjusting for use frequency. Women had significantly higher prevalence of interpersonal, financial and health-related problems than men, adjusting for frequency of use. Women showed significantly greater association of CUD with interpersonal problems with a boss or co-workers (p<0.05) and a neighbor, relative or friend (p<0.05) compared to men. Lack of sex differences in CUD after adjusting for frequency of use suggests use frequency may be an important target of CUD prevention efforts. CUD showed stronger associations for interpersonal problems among women than men, suggesting the need for particular emphasis on treating interpersonal problems related to cannabis use among women.

INTRODUCTION

Cannabis use disorder (CUD) is prevalent among people who use cannabis and may be associated with adverse medical and social consequences. A recent meta-analysis indicated that approximately 20% of cannabis users and 33% of frequent users developed CUD.¹ Adult cannabis use and CUD have increased in the past two decades, with over 4.2 million US adults meeting CUD criteria in 2019.^2–6 Increasing rates of CUD are cause for concern because CUD is associated with mental illnesses,^7,8 e.g., substance-induced psychosis and schizophrenia,^9–12 cannabis-related healthcare utilization,¹³ emergency department utilization,¹⁴ severe functional impairment,¹⁵ and psychosocial problems.¹⁴ Thus, gaining a better understanding of the subgroups at particular risk for CUD-related psychosocial and health problems is an important public health priority.

Prior studies indicate potential sex differences in the epidemiology of cannabis use, CUD and CUD-related problems. Although there is mixed evidence regarding sex differences in cannabis use trends over time,^3,16–18 rates of CUD are consistently higher among adult men,^7,8,15,17,19 and men have an increased probability of transitioning from use to CUD.^20,21 In contrast, women have been shown to transition to CUD more quickly after cannabis initiation.^7,8,18 Transition differences may be due to increased cannabis use frequency and may contribute to sex differences in CUD-related problems. For example, one study suggested that only in women, CUD is associated with lower ability to meet physical roles, worse bodily pain and worse general health.²² This study also revealed a dose-dependent relationship between cannabis use frequency and quality of life, and sex differences in this relationship, as among women each joint smoked was associated with greater reductions in mental health scores than men.²² Another study⁷ found that among those with CUD, women were more likely to be unemployed, widowed/divorced, and have mood and anxiety disorders. These studies highlight important differences in characteristics of men and women with CUD and potential differences in CUD-associated problems, but neither study adjusted for cannabis use frequency. As frequent cannabis use is associated with a greater likelihood of developing CUD¹ and experiencing CUD-related problems,²³ further research evaluating sex differences in CUD-associated problems adjusting for cannabis use frequency is necessary.

Our previous work found that CUD was associated with psychosocial and health-related problems, but did not examine sex differences in CUD-associated problems.¹⁴ Given national increases in the prevalence of CUD, determining sex differences in such problems is important to inform prevention and intervention strategies. This study leveraged nationally representative data on US adults who reported past-year cannabis use from the 2012-2013 National Epidemiologic Survey on Alcohol and Related Conditions-III (NESARC-III) to investigate sex differences in: 1) the frequency of cannabis use; 2) CUD prevalence adjusting for cannabis use frequency; 3) the burden of psychosocial and health problems adjusting for cannabis use frequency and CUD; and 4) the burden of psychosocial and health-related problems associated with CUD. This study builds on previous work in this nationally representative sample¹⁴ by restricting our sample to past-year cannabis users at risk for CUD, adjusting for cannabis use frequency, and exploring important sex differences in psychosocial and health-related problems.

METHODS

Study Design and Study Population

We used data from the 2012-2013 NESARC-III, a nationally representative cross-sectional household survey of US adults (ages 18±).²⁴ Non-institutionalized adults (N=36,309) were selected using a multi-stage probability sampling approach, and Hispanic, Black, and Asian respondents were oversampled.²⁵ Household, person-level, and overall response rates were 72.0%, 84.0% and 60.1%, respectively.²⁴ Sample weights adjusted for oversampling and non-response.^24,25 Data were collected 04/2012-06/2013 by trained interviewers who conducted face-to-face interviews using computer-assisted personal interviews.²⁵ Participants provided informed consent and received financial compensation for participation.²⁵ NESARC-III protocols and consent procedures were approved by Institutional Review Boards at the National Institutes of Health and Westat. The analytic sample was limited to participants who reported any past-year cannabis use (n=3,701), since cannabis use is a necessary condition for developing CUD.

Frequency of Past-year Cannabis Use

Frequency of use was assessed by asking “how often did you use cannabis in the past 12 months.” Responses included using everyday, 5-6 times/week, 3–4 times/week, 1–2 times/week, 2–3 times/month, once a month, 7–11 times/year, 3–6 times/year, 2 times/year, and once in the last year. Frequency of use was considered an outcome in analysis of sex differences, and a predictor in associations of CUD and problems with sex. In the model with frequency of use as an outcome, responses were converted to number of days used in the past year for easier interpretation. We estimated the number of days used by multiplying: number of times per day by 365; per week by 52; per month by 12; and per year by 1. For categories with ranges of responses, the average of the endpoints was rounded to the nearest integer (e.g., 5-6 times a week was ((5*52)+(6*52))/2=286 days, and 2-3 times a month was ((2*12)+(3*12))/2=30 days).

Past-Year CUD

CUD was assessed using the Alcohol Use Disorder and Associated Disabilities Interview Schedule 5^th Edition (AUDADIS-5), a structured diagnostic interview using the Diagnostic and Statistical Manual of Mental Disorders 5^th edition (DSM-5) criteria to assess substance use disorders and other psychiatric disorders. Respondents were positive if they met 2± of the eleven DSM-5 CUD criteria in the past 12 months. Past-year DSM-5 CUD showed moderate test-retest reliability in the general population (kappa=0.41); with 2.86 weeks between test and re-test on average (range:1-10 weeks).²⁶ The procedural validity of the AUDADIS-5 was appraised using clinician-administered, semi-structured interviews with the Psychiatric Research Interview for Substance and Mental DSM-5 (PRISM-5) in a general population, and showed moderate concordance for past-year CUD (kappa=0.60).²⁷ The PRISM-5 is a good comparator because the test-retest reliability of DSM-5 CUD in the PRISM-5 indicated substantial reliability in a general population (kappa=0.74), with 5.27 days between test and re-test on average (range:1-21 days).²⁸ Past-year CUD was considered an outcome in analysis of sex differences and a predictor of psychosocial and health problems.

Outcomes: Past-year Psychosocial and Health-related Problems

Past-year psychosocial problems (i.e., interpersonal, financial, legal) and health-related problems were assessed in separate modules from substance use questions.¹⁴ Respondents were asked if they had a series of experiences in the past year (yes/no). Interpersonal problems included: having trouble with a boss or co-worker; having problems with a neighbor, relative or friend; or breaking off a major relationship. Financial problems included: being fired or laid off from a job, being unemployed and looking for a job for more than a month, declaring bankruptcy, experiencing so much debt they could not repay it, or being homeless. Legal problems included having serious trouble with the police/law. Health-related problems included hospitalization, emergency department care, serious injury, sleep problems, and suicide attempt. Hospitalization was positive if respondents reported at least one overnight stay in a hospital, not due to childbirth. Emergency treatment was positive if respondents reported medical care in an emergency department at least once. Serious injury was positive if respondents reported at least one injury that caused them to seek medical help or cut down usual activities for more than half a day. Experiencing sleep problems was positive if respondents reported problems falling or staying asleep. Finally, respondents were categorized as experiencing a current suicide attempt if the age of their first or most recent suicidal attempt was their current age.

Covariates

Sociodemographic covariates included age, sex (male, female), education (less than high school education, high school, or more than high school), and race/ethnicity (non-Hispanic white, non-Hispanic Black, Hispanic, Other [non-Hispanic American Indian/ Alaska Native/ Asian/ Native Hawaiian/ Pacific Islander/ Mixed race]). Substance use covariates included past-year alcohol use, past-year illicit drug use (i.e., cocaine, club drugs, hallucinogens, inhalants, heroin and other drugs, and non-medical use of prescription sedatives, opioids, and stimulants), past-year DSM-5 alcohol use disorder (AUD), and past-year DSM-5 drug use disorder (DUD) (i.e., disorder involving any of the drugs listed above). Sex was included as a potential moderator of the relationship between CUD and psychosocial and health-related problems.

Statistical Analysis

We applied sample weights and described sample characteristics using weighted prevalences. We used a Poisson regression model to determine the association between sex and past-year cannabis use frequency (days). We used a series of logistic regression models to determine the association between sex (men vs. women) and past-year CUD, the association between psychosocial and health-related problems and sex and CUD, and whether the associations between these problems and CUD differed by sex. We back-transformed model estimates to the prevalence scale, and association was reported as the prevalence difference (PD), for ease of interpretation and to limit unobserved heterogeneity bias with odds ratios.²⁹ All tests were 2-tailed, with significance set at p<.05, as indicated by 95% confidence intervals not including 0.0. Analyses were conducted using SUDAAN 11.0.1 to adjust for complex sampling by incorporating sampling weights.

We first compared the average frequency of past-year cannabis use between men and women, adjusting for age, race/ethnicity, and education. We then compared past-year CUD between men and women, adjusting for age, race/ethnicity, and education, and past-year cannabis use frequency. The reported PDs represent the prevalence of CUD among men minus the prevalence of CUD among women. Next we examined the association between each psychosocial/ health-related problem and sex, adjusted for age, race/ethnicity, and education level, cannabis use frequency and CUD. The reported PD represents the prevalence of the problem among women minus the prevalence of the problem among men. Similarly, we assessed association of each psychosocial/ health-related problem with CUD, with the PD representing prevalence of the problem among those with CUD minus among those without CUD. To determine if these associations differed by sex, we incorporated an interaction term (CUD*sex) in regression models, and associations (PDs) were contrasted for women versus men. In sensitivity analyses we repeated our analyses with two additional models additionally adjusting for: (1) past-year alcohol and drug use, and (2) past-year AUD and DUD. Previous literature indicates that co-occurring AUD and CUD, or CUD and other DUDs may contribute to greater problems than either disorder alone,^14,30,31 making it important to additionally adjust for past-year AUD and DUDs that may compound the problems associated with CUD.

RESULTS

Sample Description

This sample of adults with any past-year cannabis use (n=3,701) was predominantly male (62%), ages 18-29 (48%), non-Hispanic white (65%), and had more than a high school education (58%) (Table 1). Approximately one quarter had past-year DSM-5 CUD (n=972, 27%), 47% had past-year DSM-5 AUD, and 9.7% had past-year DSM-5 DUD. Participants with CUD were younger (58% ages 18-29), more likely to be non-Hispanic Black (21%), have AUD (59%) or DUD (15%), and less likely to have more than a high school education (53%) compared with the overall sample and participants without CUD. Approximately one third of the sample reported cannabis use every day or nearly every day in the past year (29%). This differed by CUD status: 54% of people with CUD reported cannabis use every day or nearly every day in the past year, compared with 20% of people without CUD. The majority of participants without CUD used cannabis less than once per week (62%). Prevalence of the psychosocial problems in the overall sample, and among people with and without CUD are reported in Supplemental Table 1.

Table 1:

Sample Description, overall and by DSM-5 cannabis use disorder (CUD), among NESARC-III past year cannabis users (N=3,701)

			CUD
	Whole sample (N=3,701)		Yes (n=972; 26.7% [SE=0.95])		No (n=2,729; 73.3% [SE=0.95])
	n	% (SE)	n	% (SE)	n	% (SE)
Sociodemographic characteristics
Sex
Men	2181	62.24 (1.19)	620	66.10 (2.07)	1561	60.83 (1.26)
Women	1520	37.76 (1.19)	352	33.90 (2.07)	1168	39.17 (1.26)
Age
18-29	1638	48.31 (1.22)	525	58.43 (2.06)	1113	44.62 (1.47)
30-44	1100	27.38 (0.89)	281	25.60 (1.85)	819	28.03 (1.13)
45-64	879	21.88 (0.87)	154	14.45 (1.39)	725	24.59 (1.16)
65+	84	2.43 (0.41)	12	1.51 (0.57)	72	2.76 (0.46)
Race/Ethnicity
Non-Hispanic White	1922	65.40 (1.31)	442	58.21 (1.92)	1480	68.02 (1.46)
Non-Hispanic Black	976	15.74 (1.06)	315	20.77 (1.86)	661	13.90 (1.06)
Hispanic	612	13.06 (0.80)	167	14.93 (1.20)	445	12.39 (0.89)
Other	191	5.80 (0.63)	48	6.09 (1.08)	143	5.69 (0.64)
Education level
Less than high school	557	13.29 (0.68)	172	16.15 (1.31)	385	12.24 (0.83)
High school	1030	28.22 (0.93)	291	30.70 (1.63)	739	27.32 (1.09)
More than high school	2114	58.49 (1.09)	509	53.15 (1.89)	1605	60.44 (1.33)
Frequency past year cannabis use (number of days a )
Every day (365)	732	17.44 (0.85)	336	32.63 (2.01)	396	11.91 (0.77)
Nearly every day: 5-6 times a week (286)	429	11.89 (0.73)	195	21.26 (1.98)	234	8.48 (0.64)
3-4 times a week (182)	366	9.29 (0.62)	146	15.00 (1.27)	220	7.21 (0.64)
1-2 times a week (78)	377	10.15 (0.61)	94	10.14 (1.03)	283	10.16 (0.71)
2-3 times a month (30)	372	10.01 (0.57)	69	7.37 (1.06)	303	10.97 (0.69)
Once a month (12)	303	8.30 (0.54)	43	4.56 (0.84)	260	9.66 (0.74)
7-11 times in last year (9)	210	5.91 (0.52)	38	3.49 (0.69)	172	6.79 (0.65)
3-6 times in last year (5)	362	10.79 (0.72)	25	3.03 (0.67)	337	13.62 (0.97)
2 times in last year (2)	307	8.90 (0.51)	14	1.18 (0.37)	293	11.71 (0.66)
Once in last year (1)	231	6.97 (0.51)	8	0.70 (0.29)	223	9.25 (0.69)
Missing	12	0.35 (0.12)	4	0.65 (0.36)	8	0.24 (0.10)
Other substance use/disorder
Past year alcohol use	3460	94.45 (0.47)	907	93.87 (0.81)	2553	94.66 (0.57)
Past year other illicit drug useb	1149	32.66 (1.11)	351	38.40 (2.18)	798	30.57 (1.19)
Past year alcohol use disorder	1695	47.45 (1.20)	594	59.42 (1.68)	1101	43.09 (1.41)
Past year illicit drug use disorderb	347	9.70 (0.62)	144	15.30 (1.47)	203	7.66 (0.64)

SE=Standard Error

^acalculated by multiplying number of times: per day by 365; per week by 52; per month by 12; and per year by 1. For categories with ranges of response, the average of the endpoints was used. For example, 5-6 times a week was ((5*52)+(6*52))/2. There were 12 people who were missing data on frequency of cannabis use (n=3,689).

^bIllicit drug use includes sedatives/tranquilizers, opioid pain relievers, cocaine, stimulants, club drugs, hallucinogens, inhalants, heroin, and other drug

Association between Sex and Frequency of Cannabis Use and CUD

The average frequency of past-year cannabis use was higher among men than women (mean ratio=1.16, 95% CI=1.05, 1.28) (Table 2a). On average, men used cannabis on 135.3 days (SE=4.04), whereas women used cannabis on 116.4 days (SE=5.08).

Table 2a:

Association of frequency of cannabis use with sex, among past year cannabis users, NESARC-III (N=3,689)

	Average number of days used (SE)	Mean Ratio (95% CI) a
Men	135.3 (4.04)	1.16 (1.05, 1.28)
Women	116.4 (5.08)	reference

Note:. CI = confidence interval

^aadjusted for age, race, education

The prevalence of CUD was significantly higher among men than women (28.5% v 23.8%, respectively; PD=4.7%, 95% CI: 1.29%, 8.12%) after adjusting for age, race, and education (Table 2b, Model 1). After adjusting for past-year cannabis use frequency, the sex difference narrowed (27.3% v 25.4%, respectively), and was no longer significant (PD=1.9%, 95% CI: −1.35%, 5.10%; [Table 2b, Model 2]).

Table 2b:

Association of CUD with sex, among past year cannabis users, NESARC-III (N=3,701)

	Model 1a		Model 2b
	% CUD (SE)	PD (%) (95% CI)	% CUD (SE)	PD (%) (95% CI)
Men	28.5 (1.24)	4.7 (1.29, 8.12)	27.3 (1.16)	1.87 (−1.35, 5.10)
Women	23.8 (1.34)	reference	25.4 (1.37)	reference

CUD = cannabis use disorder; PD=Prevalence Difference, percent prevalence of outcome (CUD) among men minus percent prevalence of outcome among women; CI = confidence interval.

^aadjusted for age, race, education

^badjusted for age, race, education, past year frequency of cannabis use

Association of Sex with Psychosocial Problems

Compared to men, women had a significantly higher prevalence of problems after adjusting for age, race/ethnicity and education (Table 3, Model 1). Men were the reference category so positive numbers indicate higher prevalence among women. Women showed significantly higher prevalence of all interpersonal problems (range of PDs=4.9% to 6.6%), and some financial problems including unrepayable debt, and bankruptcy (range of PDs=1.4% to 8.2%). Additionally, women had higher prevalence of hospitalization, emergency treatment, and sleep problems (range of PDs=3.1% to 9.5%). These relationships were similar after additionally adjusting for past-year cannabis use frequency (Table 3, Model 2).

Table 3:

Association of sex with psychosocial problems, past year cannabis users, NESARC-III

	Model 1a			Model 2b			Model 3c
	Men	Women		Men	Women		Men	Women
Past Year Problems	% (SE)	% (SE)	PD (%) (95% CI)	% (SE)	% (SE)	PD (%) (95% CI)	% (SE)	% (SE)	PD (%) (95% CI)
Interpersonal Problems
Trouble with boss or co-workers	15.55 (0.93)	21.90 (1.21)	6.35 (3.50, 9.20)	15.45 (0.93)	21.99 (1.25)	6.54 (3.63, 9.45)	15.41 (0.93)	22.07 (1.26)	6.66 (3.73, 9.59)
Problems with neighbor, relative, or friend	13.86 (0.89)	20.48 (1.30)	6.63 (3.73, 9.52)	13.69 (0.89)	21.04 (1.35)	7.35 (4.33, 10.36)	13.64 (0.88)	21.16 (1.35)	7.53 (4.53, 10.52)
Broke up major relationship	14.57 (0.89)	19.50 (1.28)	4.93 (2.05, 7.81)	14.45 (0.90)	19.71 (1.22)	5.26 (2.47, 8.04)	14.40 (0.89)	19.81 (1.23)	5.42 (2.61, 8.23)
Financial and Legal Problems
Fired or laid off	15.28 (1.02)	15.78 (1.21)	0.50 (−2.63, 3.63)	15.09 (1.03)	15.84 (1.25)	0.76 (−2.46, 3.98)	15.08 (1.03)	15.85 (1.25)	0.77 (−2.45, 3.99)
Unemployed	31.59 (1.24)	31.47 (1.48)	−0.12 (−4.03, 3.79)	31.26 (1.24)	31.87 (1.48)	0.60 (−3.35, 4.56)	31.22 (1.25)	31.94 (1.48)	0.72 (−3.25, 4.70)
Homeless	5.07 (0.52)	5.18 (0.79)	0.11 (−1.66, 1.88)	4.97 (0.50)	5.40 (0.82)	0.44 (−1.36, 2.23)	4.96 (0.50)	5.43 (0.82)	0.47 (−1.32, 2.26)
Declared bankruptcy	0.56 (0.16)	1.95 (0.41)	1.40 (0.54, 2.25)	0.55 (0.15)	2.04 (0.43)	1.49 (0.61, 2.38)	0.54 (0.15)	2.05 (0.44)	1.51 (0.62, 2.40)
So much debt couldn’t repay	22.28 (1.11)	30.45 (1.85)	8.17 (4.09, 12.24)	22.00 (1.12)	30.71 (1.85)	8.71 (4.56, 12.86)	21.96 (1.12)	30.78 (1.85)	8.82 (4.69, 12.95)
Trouble with law/police	6.93 (0.66)	5.26 (0.73)	−1.66 (−3.38, 0.05)	6.76 (0.65)	5.54 (0.78)	−1.22 (−3.01, 0.56)	6.74 (0.64)	5.57 (0.78)	−1.17 (−2.94, 0.60)
Health-related Problems
Hospitalized	8.30 (0.71)	11.36 (0.81)	3.06 (1.09, 5.03)	8.26 (0.71)	11.37 (0.81)	3.11 (1.16, 5.07)	8.26 (0.71)	11.38 (0.81)	3.13 (1.17, 5.08)
Emergency treatment	25.74 (1.29)	33.46 (1.40)	7.72 (4.26, 11.18)	25.60 (1.30)	33.70 (1.41)	8.11 (4.60, 11.61)	25.56 (1.31)	33.78 (1.42)	8.23 (4.66, 11.79)
Suicide attempt at current age	3.09 (0.48)	4.27 (0.66)	1.18 (−0.65, 3.02)	3.10 (0.50)	4.22 (0.65)	1.12 (−0.73, 2.97)	3.10 (0.49)	4.25 (0.66)	1.15 (−0.72, 3.02)
Serious injury	26.67 (1.22)	26.28 (1.56)	−0.39 (−4.35, 3.58)	26.56 (1.21)	26.48 (1.56)	−0.08 (−4.03, 3.87)	26.52 (1.21)	26.55 (1.58)	0.02 (−3.96, 4.00)
Sleep problems	28.87 (1.42)	38.40 (1.70)	9.53 (5.23, 13.83)	28.64 (1.42)	38.32 (1.71)	9.68 (5.38, 13.98)	28.57 (1.41)	38.45 (1.71)	9.88 (5.60, 14.16)

CI = confidence interval; PD = prevalence difference, prevalence of outcome (consequence) among men minus prevalence of outcome among women.

^aadjusted for age, race/ethnicity, and education level

^badjusted for age, race/ethnicity, education level, and frequency of cannabis use in the past year

^cadjusted for age, race/ethnicity, education level, frequency of cannabis use in the past year, and past year cannabis use disorder

Note: Participants with missing responses for an outcome were excluded for that analysis (range of missingness: 0 [0.0%] – 24 [0.63%]) in Models 1 and 2. Participants missing for frequency of cannabis use (12, 0.35%) were excluded from model 2. Bolded values are significant (p<.05).

Association of CUD with psychosocial problems

CUD was associated with a higher prevalence of all interpersonal problems (range of PDs=5.8%-11.4%), financial and legal problems (unemployment, homelessness, unrepayable debt, and trouble with the police/law; range of PDs=4.2% to 9.2%), and health-related problems (emergency treatment, serious injury and sleep problems; range of PDs=5.7%-12.5%) (Supplemental table 2, Model 1). This remained mostly consistent after additionally adjusting for past-year cannabis use frequency (Supplemental table 2, Model 2).

Association of CUD with Psychosocial Problems by Sex

The overall association between CUD, psychosocial and health-related problems was similar among men and women (Table 4a). Many of the problems associated with CUD did not differ by sex, including: a major relationship break-up; unemployment; homelessness; unrepayable debt; trouble with the police/law; emergency treatment; and sleep problems. However, CUD showed significantly greater association with women than men for trouble with a boss or coworkers (Difference in PD with men as reference=8%, 95% CI: 0.47%, 15.48%), and problems with a neighbor, relative, or friend (Difference in PD=9.6%, 95% CI: 2.32%, 16.84%). After adjusting for past-year cannabis use frequency (Table 4b), the association between CUD and trouble with boss or coworkers and problems with a neighbor, relative or friend remained significantly stronger among women, but CUD was no longer associated with homelessness among men or women, or with a major relationship break-up, trouble with the police/law, or sleep problems among women.

Table 4a:

Association^a of CUD with consequences, by sex, among past year cannabis users, NESARC-III

	Men			Women			Difference between men and women
	CUD = Yes	CUD = No		CUD = Yes	CUD = No
Past Year Problems	% (SE)	% (SE)	PD (%) (95% CI)	% (SE)	% (SE)	PD (%) (95% CI)	Difference in PD (%) (95% CI)
Interpersonal Problems
Trouble with boss or co-workers	17.60 (1.97)	14.69 (1.00)	2.92 (−1.37, 7.21)	30.13 (3.01)	19.24 (1.34)	10.89 (4.24, 17.55)	7.97 (0.47, 15.48)
Problems with neighbor, relative, or friend	19.51 (1.77)	11.61 (1.00)	7.91 (3.99, 11.83)	33.81 (3.07)	16.33 (1.35)	17.49 (10.96, 24.01)	9.58 (2.32, 16.84)
Broke up major relationship	21.53 (2.03)	11.60 (1.04)	9.93 (5.21, 14.66)	24.71 (2.82)	17.80 (1.24)	6.92 (1.36, 12.48)	−3.02 (−9.80, 3.76)
Financial and Legal Problems
Fired or laid off	15.58 (1.86)	15.15 (1.17)	0.43 (−3.77, 4.63)	19.76 (2.57)	14.45 (1.42)	5.31 (−0.60, 11.21)	4.88 (−2.14, 11.89)
Unemployed	37.34 (2.20)	29.15 (1.41)	8.19 (3.34, 13.03)	39.78 (2.90)	28.71 (1.63)	11.07 (4.80, 17.34)	2.88 (−5.00, 10.77)
Homeless	7.47 (1.28)	4.07 (0.58)	3.40 (0.48, 6.32)	9.45 (2.48)	3.72 (0.65)	5.73 (0.64, 10.82)	2.32 (−3.16, 7.81)
Declared bankruptcy	1.11 (0.44)	0.35 (0.13)	0.75 (−0.13, 1.64)	2.98 (1.03)	1.66 (0.43)	1.33 (−0.85, 3.50)	0.58 (−1.43, 2.58)
So much debt couldn’t repay	26.76 (2.03)	20.57 (1.33)	6.19 (1.36, 11.03)	39.26 (3.32)	27.74 (1.89)	11.51 (5.04, 17.99)	5.32 (−2.57, 13.21)
Trouble with law/police	10.74 (1.43)	5.12 (0.69)	5.62 (2.50, 8.74)	8.69 (1.61)	3.96 (0.70)	4.73 (1.54, 7.92)	−0.89 (−5.47, 3.68)
Health-related Problems
Hospitalized	9.72 (1.51)	7.76 (0.72)	1.96 (−1.13, 5.06)	11.98 (1.88)	11.18 (0.99)	0.80 (−3.63, 5.23)	−1.16 (−6.68, 4.36)
Emergency treatment	30.02 (2.11)	24.00 (1.45)	6.02 (1.44, 10.59)	40.27 (3.16)	31.33 (1.67)	8.93 (1.54, 16.33)	2.92 (−6.15, 11.98)
Suicide attempt at current age	3.15 (1.22)	3.06 (0.62)	0.09 (−2.91, 3.09)	4.98 (1.75)	4.02 (0.79)	0.96 (−3.21, 5.13)	0.87 (−4.19, 5.92)
Serious injury	32.52 (2.46)	24.36 (1.24)	8.16 (3.14, 13.17)	27.23 (2.95)	26.00 (1.72)	1.23 (−5.11, 7.56)	−6.93 (−15.11, 1.26)
Sleep problems	39.89 (2.17)	24.68 (1.64)	15.21 (10.07, 20.34)	44.15 (2.81)	36.67 (1.99)	7.48 (0.94, 14.02)	−7.73 (−16.76, 1.30)

CUD = cannabis use disorder; CI = confidence interval; PD = prevalence difference, prevalence of outcome (consequence) among those with CUD minus prevalence of outcome among those without CUD.

^aadjusted for age, race/ethnicity, education level

Note: Participants with missing responses for an outcome were excluded for that analysis (range of missingness: 0 [0.0%] – 24 [0.63%]).

Table 4b:

Association^a of CUD with consequences, by sex, among past year cannabis users, adjusted for frequency of cannabis use, NESARC-III

	Men			Women
	CUD = Yes	CUD = No		CUD = Yes	CUD = No		Difference between men and women
Past Year Problems	% (SE)	% (SE)	PD (%) (95% CI)	% (SE)	% (SE)	PD (%) (95% CI)	Difference in PD (%) (95% CI)
Interpersonal Problems
Trouble with boss or co-workers	17.87 (2.00)	14.55 (0.97)	3.32 (−0.89, 7.53)	30.55 (3.04)	19.08 (1.39)	11.46 (4.64, 18.28)	8.15 (0.63, 15.66)
Problems with neighbor, relative, or friend	18.32 (1.73)	11.85 (1.02)	6.47 (2.60, 10.35)	32.44 (3.04)	16.98 (1.48)	15.46 (8.68, 22.23)	8.98 (1.77, 16.20)
Broke up major relationship	21.04 (2.28)	11.70 (1.08)	9.34 (3.97, 14.70)	23.77 (3.06)	18.12 (1.22)	5.65 (−0.79, 12.10)	−3.69 (−10.52, 3.15)
Financial and Legal Problems
Fired or laid off	14.42 (1.92)	15.40 (1.20)	−0.98 (−5.40, 3.43)	18.39 (2.59)	14.92 (1.51)	3.47 (−2.62, 9.57)	4.46 (−2.32, 11.23)
Unemployed	35.15 (2.23)	29.66 (1.41)	5.49 (0.58, 10.40)	37.68 (3.07)	29.71 (1.70)	7.97 (1.06, 14.88)	2.48 (−5.47, 10.43)
Homeless	6.37 (1.21)	4.29 (0.63)	2.08 (−0.85, 5.01)	8.17 (2.19)	4.15 (0.73)	4.02 (−0.60, 8.64)	1.94 (−2.95, 6.84)
Declared bankruptcy	0.97 (0.39)	0.37 (0.14)	0.61 (−0.20, 1.41)	2.67 (0.94)	1.79 (0.47)	0.88 (−1.16, 2.93)	0.28 (−1.60, 2.16)
So much debt couldn’t repay	26.06 (2.17)	20.54 (1.29)	5.52 (0.65, 10.40)	38.82 (3.34)	27.99 (1.90)	10.83 (4.22, 17.44)	5.31 (−2.58, 13.20)
Trouble with law/police	9.01 (1.22)	5.48 (0.75)	3.53 (0.72, 6.34)	7.41 (1.36)	4.54 (0.84)	2.87 (−0.07, 5.82)	−0.65 (−4.66, 3.35)
Health-related Problems
Hospitalized	9.19 (1.48)	7.89 (0.73)	1.31 (−1.77, 4.39)	11.40 (1.83)	11.34 (1.00)	0.06 (−4.29, 4.41)	−1.25 (−6.57, 4.08)
Emergency treatment	29.54 (2.15)	24.08 (1.47)	5.46 (0.76, 10.16)	40.06 (3.27)	31.51 (1.70)	8.54 (0.89, 16.19)	3.08 (−5.92, 12.09)
Suicide attempt, current age	3.33 (1.38)	3.05 (0.61)	0.28 (−3.00, 3.55)	6.14 (2.27)	3.82 (0.68)	2.31 (−2.59, 7.22)	2.04 (−3.73, 7.80)
Serious injury	32.51 (2.53)	24.31 (1.27)	8.20 (2.83, 13.56)	27.49 (3.02)	26.04 (1.69)	1.45 (−4.93, 7.84)	−6.74 (−15.00, 1.51)
Sleep problems	39.91 (2.35)	24.58 (1.61)	15.33 (9.98, 20.69)	43.79 (3.09)	36.36 (2.07)	7.43 (−0.03, 14.90)	−7.90 (−17.16, 1.36)

CUD = cannabis use disorder; CI = confidence interval

^aadjusted for age, race/ethnicity, education level, frequency of past year cannabis use

^bPD = prevalence difference, prevalence of outcome (consequence) among those with CUD minus prevalence of outcome among those without CUD

Note: Participants with missing responses for an outcome were excluded for that analysis (range of missingness: 0 [0.0%] – 24 [0.63%]). Participants missing for frequency of cannabis use (12, 0.35%) were excluded as well.

Sensitivity Analyses

We conducted sensitivity analyses for the associations between sex, cannabis use frequency, CUD, and psychosocial problems, additionally adjusting for past-year alcohol and drug use or AUD and DUD. Results were generally the same as main analyses. The average frequency of cannabis use remained higher among men than women (Supplemental Table 3a), and there continued to be no significant sex difference in prevalence of CUD after adjusting for past-year cannabis use frequency (Supplemental Table 3b). Women continued to show significantly higher prevalence of interpersonal, financial, and health-related problems after adjusting for past-year alcohol and drug use (Supplemental Table 4, Model 1) or past-year AUD and DUD (Supplemental Table 4, Model 2). CUD also remained significantly associated with psychosocial and health-related problems (Supplemental Table 5). Finally, women with CUD continued to have a greater association with interpersonal problems than men, after adjusting for past-year alcohol and drug use (Supplemental Table 6a). After adjusting for past-year AUD and DUD (Supplemental Table 6b) the association between CUD and trouble with boss or coworkers was no longer significantly higher among women.

DISCUSSION

Using nationally representative data of US adults who used cannabis in the past year, our study found that men had a greater prevalence of CUD than women, but this difference was partly explained by sex differences in frequency of cannabis use. The prevalence of psychosocial and health-related problems was elevated among both men and women with CUD, but women had a greater likelihood of reporting trouble with boss or co-workers, and problems with a neighbor, relative or friend. These findings suggest that men and women with similar cannabis use frequencies have a similar likelihood of CUD, but women were significantly more likely to report interpersonal problems associated with CUD.

Similar to previous literature, the majority of people who used cannabis in the past year, and people with CUD in the past year, were men.^3,8,18,19,32 However, sex differences were no longer significant after adjusting for frequency of use, suggesting the gender gap in CUD and the higher likelihood of transition from cannabis use to CUD among men^20,21 may be due to higher frequency of cannabis use among men.⁷ Evidence that women progress from cannabis initiation to CUD more quickly than men (i.e., telescoping)^7,8,33 may also be due to rapid escalation in frequency of use, especially as women are more likely to experience cannabis withdrawal symptoms.^34,35 These results are consistent with a previous study showing that the association between cannabis use frequency and CUD did not vary by sex.³² Because our results indicate cannabis use frequency may help explain sex differences in the prevalence of CUD, frequency of cannabis use may be an important target for CUD prevention efforts among men and women who use cannabis. High-frequency cannabis use is associated with a greater risk of developing CUD,¹ therefore CUD prevention efforts may include online or in-person treatment interventions such as motivational interviewing, cognitive behavioral therapy, or personalized normative feedback assessing baseline cannabis use and aim to reduce past-month cannabis use frequency.^36,37 Such programs have contributed to reductions in past-month cannabis use frequency among cannabis users.^36,37

The increased likelihood of interpersonal problems among women with CUD is consistent with previous literature showing women with CUD were more likely to be unemployed, have mood or anxiety disorders, and mental health decrements.^7,22 The elevated prevalence of problems with a boss or co-worker among women with CUD may contribute to unemployment. Furthermore, increased interpersonal problems among women with CUD may associated with CUD-related mood and anxiety disorders.^30,38–42 Women are also more likely to experience cannabis withdrawal symptoms,^34,35 which overlap with depression and anxiety symptoms^43–45 and may contribute to self-medication with cannabis and increased use to maintain symptom relief.^46,47 Sex differences in mood and anxiety disorders and cannabis withdrawal may provide context for the higher likelihood of interpersonal problems among women with CUD. These results were also robust to additional sensitivity analyses adjusting for past-year alcohol and drug use, but differential associations between women and men were attenuated after adjusting for AUD and DUD. Efforts to prevent CUD among women should include screening for cannabis withdrawal symptoms that may promote more frequent cannabis use for self-medication and increase the risk for developing CUD, health education efforts regarding risks associated with increased frequency of cannabis use, and helping clients or patients learn more effective stress-coping strategies.^46,48 These results also suggest the need for sex-specific treatment for women with CUD,⁴⁹ with an emphasis on strengthening social support networks to reduce the risk of interpersonal problems.

A previous study showed no sex differences in cannabis-related problems after adjusting for frequency and quantity of past-year cannabis use.³² The lack of sex differences may be because the previous study defined cannabis-related problems using subscales of CUD DSM-5 criteria, rather than independent psychosocial or health-related problems. We found no sex differences in likelihood of CUD after adjusting for frequency of use, so we would expect cannabis-related problems defined using DSM-5 criteria would also not differ.

Study limitations are noted. First, the cross-sectional design of the NESARC-III does not allow us to infer the direction of the association between CUD and psychosocial and health-related problems. Although individuals may use cannabis as self-medication to cope with psychosocial or health-related problems,⁴⁶ our findings are consistent with longitudinal studies examining the adverse consequences of cannabis exposure.⁵⁰ Future studies should leverage longitudinal data to understand the direction of association. Second, certain DSM-5 CUD criteria (e.g., social problems or neglecting professional responsibilities)⁴³ may be highly correlated with interpersonal and financial problems evaluated. However, our analysis extended the literature by evaluating sex differences in problems associated with CUD. Future studies may consider evaluating associational differences accounting for which DSM-5 criteria were met. Third, CUD may be underreported due to social desirability or recall bias, particularly in the context of the 2012-2013 cannabis legal landscape when states first began to legalize recreational cannabis use.⁵¹ Women may also be more likely to under-report CUD, particularly when pregnant, due to fear of legal repercussions of substance use.⁵² If there was differential misclassification of exposure among women, whereby women with CUD were misclassified as unexposed, then the reported relationship between CUD and psychosocial problems among women may be underestimated. Fourth, similar to other nationally representative surveys, the NESARC-III excludes people who are institutionalized (e.g., incarcerated) or homeless. Substance use disorders are associated with increased risk of institutionalization and homelessness and there are potential sex differences in this relationship,^53–58 suggesting our estimates of the association between CUD, homelessness, and trouble with the police/law, and sex differences in these associations may be conservative.

CONCLUSION

Our study of past-year cannabis users found that sex differences in the likelihood of CUD may be explained by frequency of use, but women showed greater prevalence of interpersonal, financial, and health-related problems. As changes in cannabis legalization policies have been associated with both increased frequent cannabis use and CUD among US adults,⁵⁹ our findings may suggest that frequency of use is an important target for CUD prevention strategies. Although CUD was associated with interpersonal, financial, legal, and health-related problems among both men and women, women with CUD had greater prevalence of reporting interpersonal problems than men. Differences in interpersonal problems may be due to the worse mental health outcomes among women with CUD, suggesting that CUD treatment for women should focus on increasing support among interpersonal relationships.