Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

RESEARCH REPORT doi:10.1111/add.

15472

Trends in cannabis use among adults with children in the


home in the United States, 2004–2017: impact of state-
level legalization for recreational and medical use

Renee D. Goodwin1,2 , June H. Kim3, Keely Cheslack-Postava4, Andrea H. Weinberger5,6,


Melody Wu1,2, Katarzyna Wyka7 & Meyer Kattan8,9
Department of Epidemiology and Biostatistics, Graduate School of Public Health and Health Policy, The City University of New York, New York, NY, USA,1 Department
of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY, USA,2 Institute for Implementation Science in Population Health, The City
University of New York, New York, NY, USA,3 Department of Psychiatry, Columbia University, New York, NY, USA,4 Ferkauf Graduate School of Psychology,
Yeshiva University, Bronx, NY, USA,5 Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY, USA,6 Center for Systems
and Community Design, Graduate School of Public Health and Health Policy, The City University of New York, New York, NY, USA,7 Department of Pediatrics, New
York-Presbyterian/Columbia University Medical Center, New York, NY, USA8 and Division of Pediatric Pulmonology, New York-Presbyterian/Morgan Stanley
Children’s Hospital, New York, NY, USA9

ABSTRACT

Background and Aims Cannabis use among parents may be increasing with legalization, but perception of associated
risk has declined. The study investigated the association between cannabis legalization and cannabis use among adults
with children in the home over time in the United States (US). Design A difference-in-difference approach was applied
to public and restricted-use data from the 2004–2017 National Survey on Drug Use and Health (NSDUH), an annual
cross-sectional survey. Setting A representative sample of the United States. Participants/Cases Respondents ages
18+ with children living in the home drawn from the NSDUH (n = 287,624), which is administered to non-institutional-
ized civilians in the 50 states and District of Columbia. Measurements Exposures were year and state-level cannabis pol-
icy in state of residence annually. Outcomes were past-30-day cannabis use and daily cannabis use. Sociodemographic
variables included age, gender, marital status, annual family income, race/ethnicity, educational attainment, and strength
of state-level tobacco control. Findings In 2017, past-month cannabis use (11.9%, 9.3%, and 6.1%) and daily cannabis
use (4.2%, 3.2%, and 2.3%) were more common in states with recreational marijuana laws (RML), followed by states with
medical marijuana laws (MML) and without legal cannabis use, respectively. RML and MML were associated with signifi-
cantly higher prevalence of past-month cannabis use (adjusted odds ratio [AOR] = 1.28, 95% confidence interval
[CI] = 1.12–1.46; AOR = 1.12, 95% CI = 1.03–1.22) and daily cannabis use (AOR = 1.25, 95% CI = 1.03–1.51;
AOR = 1.16, 95% CI = 1.02–1.32), respectively. The impact of MML was particularly salient among adults ages 50+ and
the highest income and education subgroups. Conclusions Among adults with children living in the home, cannabis
use appears to be more common in US states with legalized cannabis use compared with states with no legal cannabis
use. Recreational legalization appears to increase use among adults with children in the home broadly across nearly all
sociodemographic groups, whereas the effect of legalization for medical use is heterogeneous by age and socioeconomic
status.

Keywords Cannabis, epidemiology, legalization, marijuana, NSDUH, substance use.

Correspondence to: Renee D. Goodwin, Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, Rm 1030D,
New York, NY 10032. E-mail: rdg66@columbia.edu
Submitted 9 June 2020; initial review completed 1 September 2020; final version accepted 24 February 2021

INTRODUCTION primary source of child secondhand smoke (SHS) exposure


[2,3], and youth exposure to secondhand tobacco smoke
More than one in three (34.4%) children in the United (STS) has well-documented health consequences [4].
States (US) live in a household with at least one adult Extensive public health and clinical intervention efforts
cigarette smoker [1]. Residential parental smoking is the have led to lowered cigarette smoking among adults overall

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


Cannabis use in adults with children at home 2771

in the US, and increased enactment of smoke-free laws has estimated the relationship between MML/RML adoption
reduced exposure to STS among adults and youth in both and the odds of past-month current and daily cannabis
public and private spaces [4–6]. Although cigarette use among adults with children in the home in the US.
smoking is decreasing, cannabis use is simultaneously in-
creasing among adults in the US [7–9], with recent data METHODS
suggesting that increases in cannabis use are more rapid
in states that have legalized cannabis for recreational (rec- Data and population
reational marijuana laws, RML) and/or medical (medical Public and restricted-access data from the 2004 to 2017
marijuana laws, MML) use [10], especially among individ- National Survey on Drug Use and Health (NSDUH) were
uals already using cannabis [11]. Child exposure to sec- used. The NSDUH is an annual cross-sectional, nationally
ondhand cannabis smoke (SCS) also appears to be representative survey on substance use and mental health
associated with potential harms to respiratory health in the US. Weighted interview response rates for 2004 to
[12,13]. 2017 ranged from 66.5% to 77.0%. Data are collected an-
With legalization, perception of risk associated with nually from male and female civilian non-institutionalized
cannabis use has declined rapidly [8,14,15]. Therefore, in persons ages 12 and older in each of the 50 US states and
contrast with cigarette use, cannabis use among parents the District of Columbia. The analysis was not
may be increasing, whereas awareness or concern over pre-registered on a publicly available platform, and the re-
risks associated with exposure to SCS possibly remains lim- sults should be considered exploratory.
ited. It is not known whether legalization for medical and/ Information about state of residence was used to
or recreational use has impacted cannabis use among determine state MML and RML status and these data,
adults with children in the home. Increased exposure could which are restricted-use variables, were accessed through
theoretically thwart the public health gains achieved in re- the Research Data Center. Data collection for the NSDUH
ducing child exposure to SHS. In addition, parental canna- was approved by the National Center for Health Statistics
bis use is associated with increased risk of cannabis use (NCHS) Research Ethics Review Board (ERB). Analysis of
among offspring [16]. de-identified data from the survey is exempt from federal
If legalization for medical or recreational purposes has regulations for the protection of human research partici-
led to increases in cannabis use among adults living in pants. Analysis of restricted data through the NCHS
the home with children, increased attention to public Research Data Center was also approved by the NCHS
health education programs addressing potential risks of ERB. Our sample was restricted to adult respondents (ages
SCS exposure and parental cannabis use to youth should 18 and older) who reported having children ages
be considered. Interventions and programming in commu- <18 years living in the home (n = 287,624), and a sub-
nity and clinical settings can provide advice on how to min- sample of the 2017 NSDUH respondents with the same
imize youth exposure to SCS, as has been done with adult characteristics (n = 22,308). Data were pooled for survey
cigarette use and youth exposure to STS. Further, recent years 2004 to 2017 (long format).
data demonstrated differences in the prevalence of canna-
bis use among parents with children in the home by demo-
Measures
graphics [9]. It is not known whether changes in cannabis
use among adults with children in the house by cannabis Current cannabis use
legalization in their state of residence differ among demo-
graphic subgroups. Understanding these patterns and Respondents were asked about their past-month use of
trends can facilitate targeting of vulnerable populations cannabis or hashish. Past-month cannabis use was
and intervention resources. classified as use of cannabis on at least 1 day in the past
The goal of the current study was to address these gaps 30. Respondents were classified as daily cannabis users if
in knowledge of this timely and urgent public health issue they reported using on 30 of the past 30 days.
by investigating cannabis use among adults with children
Sociodemographic characteristics
living in the home in the US. We hypothesize that the prev-
alence of past-month current and daily cannabis use is To control for individual- and state-level factors that may
higher among those residing in states in which cannabis confound or moderate the association between cannabis
use is legal than in states without any cannabis legaliza- legalization and current cannabis use, estimates were
tion. First, the study estimated the prevalence of adjusted for the following covariates; individual age (18–
past-month current and daily cannabis use among adults 25, 26–34, 35–49, ≥50 years old), gender (male, female),
with children living in the same home in the US, by state- marital status (married, widowed/divorced/separated,
level cannabis legalization status in 2017, overall and by never married), annual family income (<$20,000,
sociodemographic characteristics. Second, the study $20, 000–$74,999, ≥$75,000), race/ethnicity (non-

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


2772 Renee D. Goodwin et al.

Hispanic White, non-Hispanic Black, Hispanic, non-His- whose legalization status remained stable. This is done by
panic Other), educational attainment (less than high school, including fixed effects for state of residence and for calendar
high school or equivalent, some college, college graduate or year along with time-varying indicators for the presence of
above), and strength of state-level tobacco control. MML and RML. Hence, separately for each cannabis out-
come, two-way fixed effect models that included calendar
Cannabis legalization year and state of residence, time-varying indicators for
Cannabis legalization was defined by the presence of MML and RML (unadjusted model), as well as
enacted MML or RML. The time-varying indicators for sociodemographic covariates and state-level tobacco con-
MML and RML were “0” in years before enactment, and trol (adjusted model) were estimated. DiD estimates
“1” for all subsequent years and for MML and RML, respec- (expressed as adjusted odds ratios [AORs]) are only reflec-
tively (states can be “1” for MML and “1” for RML within tive of states that have changed status within the study pe-
the same year). Comparison states were states without riod. Therefore, only states that have a status change for
cannabis legalization during the study period (AL, GA, MML (from “0” indicating no legalization to “1” indicating
IA, ID, IN, KS, KY, MS, NC, NE, OK, SC, SD, TN, TX, UT, legalization) are included among the MML “treated” group
VA, WI, and WY). (AZ, AK, CT, DC, DE, FL, IL, LA, MA, MD, MI, MN, MO, ND,
NH, NJ, NM, NY, OH, OK, PA, RI, UT, VT, and WV). Simi-
larly, only states that have a status change for RML (from
Statistical analysis
“0” to “1”) are included among the RML “treated” group
First, among adults with children in the home, we esti- (AK, CA, CO, DC, MA, ME, NV, OR, and WA). All other
mated the prevalence of past-month cannabis and states contribute as “controls” and are used to estimate
past-month daily cannabis use by the cannabis legalization the “counterfactual” trend that treatment states would
status of their state of residence in 2017. The cannabis le- have demonstrated had they not been “treated” (i.e.
galization groups include states with RML and MML in changes in the outcomes over the same time period in
2017 (AK, CA, CO, DC, MA, ME, NV, OR, and WA), states states that did not pass MML or RML). The AOR indicates
with only MML in 2017 (AR, AZ, CT, DE, FL, HI, IL, LA, the change in the odds of cannabis use over time in states
MD, MI, MN, MO, MT, ND, NH, NJ, NM, NY, OH, PA, RI, where MML/RML were enacted compared to states with-
VT, and WV), and states with no legalization in 2017 out MML/RML enactment.
(AL, GA, IA, ID, IN, KS, KY, MS, NC, NE, OK, SC, SD, TN, Fourth, heterogeneity in DiD estimates across
TX, UT, VA, WI, and WY). sociodemographic strata was explored by including inter-
Second, we attempted to visually inspect the action terms between the sociodemographic factor of inter-
parallel trends assumption, which assumes that the pre- est and time-varying indicators for RML and MML (i.e. two
intervention trends in exposed and unexposed groups are separate two-way interactions, with no interaction be-
parallel. To this end, we first categorized states into four tween MML and RML). Interactions between MML/RML
fixed groups based on their RML/MML status in 2018, and each sociodemographic factor were explored one fac-
which is the most recent year of NSDUH data availability tor at a time (e.g. age*MML + age*RML), and adjusted in-
(having passed MML early in the study period, i.e. 2004– teraction models included the other un-interacted,
2011, or post 2011; having ever passed RML; no MML/ sociodemographic factors. Stratum-specific DiD estimates
RML in the study period). States that had passed MML be- were obtained using the “effects” command. The P value
fore versus after 2011 were categorized separately because for the interaction term (denoted P_int in the tables) indi-
of observed heterogeneity based on duration of passage cates whether the strength of association varies between
[17]. Next, for every calendar year between 2004 (the first a specific sociodemographic group versus the reference
available year) and 2013 (the first year in which RML was group. All analyses were conducted using SAS-callable
passed); the prevalence of each outcome (i.e. past-month SUDAAN version 11.0.1 and incorporated survey weights
cannabis use, past-month daily cannabis use) was esti- for all analyses. P values <0.05 were considered statisti-
mated (see Supporting information Figs. S1a and S1b). cally significant.
Third, we assessed the impact of both medical and rec-
reational cannabis legalization on current cannabis use RESULTS
among adult respondents living with children in the house-
Prevalence of current cannabis use in 2017
hold using difference-in-difference (DiD) models using lo-
gistic regression for binary outcomes. The DiD method The prevalence of past-month cannabis use among adults
estimates the effect of legalization by comparing changes with children living in the household varied by cannabis le-
in the outcome (i.e. cannabis use, daily cannabis use) from galization status in 2017 (Table 1a), such that past-month
before to after passage of a law in states passing laws com- prevalence was highest among adults residing in states
pared to the changes over the same time period in states with RML (11.9%), followed by states with MML (9.4%)

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


Cannabis use in adults with children at home 2773

Table 1 (a) Prevalence of past-month cannabis use among adults with children <18 years in the household by cannabis legalization
status and sociodemographic characteristics, NSDUH 2017.

Cannabis legalization status

a a a
No MML (n = 7500) MML (n = 10,800) RML (n = 3900)

Overall 600 (6.11%) 1300 (9.44%) 600 (11.91%)


Age
18–25 300 (13.24%) 500 (19.74%) 300 (19.87%)
26–34 200 (7.47%) 300 (10.86%) 100 (15.27%)
35–49 100 (4.53%) 400 (7.59%) 200 (10.01%)
≥50 <100 (2.13%) <100 (4.44%) <100 (6.42%)
Gender
Male 300 (7.88%) 600 (11.67%) 300 (14.69%)
Female 300 (4.69%) 600 (7.62%) 300 (9.65%)
Marital status
Married 200 (3.37%) 400 (5.87%) 200 (8.61%)
Widowed/divorced/separated <100 (6.77%) 100 (8.10%) <100 (10.47%)
Never married 400 (13.07%) 800 (17.77%) 400 (20.92%)
Income
<$20,000 100 (8.79%) 300 (14.11%) 100 (16.63%)
$20,000–$74,000 200 (7.79%) 400 (10.04%) 200 (13.48%)
≥$75,000 200 (4.61%) 500 (8.07%) 300 (10.32%)
Race/Ethnicity
Non-Hispanic White 300 (6.26%) 700 (9.91%) 300 (15.61%)
Non-Hispanic Black 100 (8.27%) 200 (12.77%) <100 (20.23%)
Hispanic <100 (3.92%) 200 (7.02%) 200 (8.43%)
Non-Hispanic Other <100 (5.91%) 100 (5.63%) <100 (7.21%)
Education
Less than high school 100 (8.25%) 200 (11.17%) 100 (10.05%)
High school or equivalent 200 (7.44%) 400 (11.09%) 200 (13.46%)
Some college 200 (7.44%) 500 (10.68%) 200 (15.18%)
College graduate or above <100 (2.51%) 200 (6.24%) 100 (9.26%)

a
MML = medical marijuana laws; NSDUH = National Survey on Drug Use and Health; RML = recreational marijuana laws. Unweighted sample sizes must be
rounded to the nearest 100 to use NSDUH restricted-use data in the Research Data Center setting.

and then states with neither MML nor RML (6.1%). The associated with increases in past-month cannabis use
prevalence of past-month cannabis use decreased with in- among adults with children in the household, compared
creasing age, was higher among males, those who never with states that did not have legalized marijuana during
married, those who reported a family income less than the study period.
$20,000 and non-Hispanic Black individuals. The association between MML and past-month
Similarly, the prevalence of past-month daily cannabis cannabis use significantly varied by age (F(3) = 8.43,
use also varied by cannabis legalization status (Table 1b), P < 0.001), marital status (F(2) = 4.62, P = 0.010), in-
with the highest prevalence of daily cannabis use reported come (F(2) = 3.23, P = 0.040), race/ethnicity
among adults residing in states with RML (4.2%), followed (F(3) = 7.85, P < 0.001) and education (F(3) = 11.94,
by states with MML (3.2%) and then states with neither P < 0.001), such that the association was strongest
MML nor RML (2.4%). The prevalence of daily cannabis among those older than 50 years (AOR = 1.75, 95%
use was higher among younger age groups, males, those CI = 1.38–2.22), those who were widowed/divorced/sepa-
who never married, those who reported a family income rated (AOR = 1.22, 95% CI = 1.05–1.43) followed by those
less than $20,000 and non-Hispanic Black individuals. who were married (AOR = 1.19, 95% CI = 1.07–1.32),
those who reported a family income greater than
$75,000 (AOR = 1.19, 95% CI = 1.08–1.31), non-
Difference-in-difference estimates for past-month any
Hispanic White individuals (AOR = 1.19, 95% CI = 1.08–
cannabis use (Table 2a)
1.30) and those with a college degree or higher
Overall, both medical cannabis (AOR = 1.12, 95% confi- (AOR = 1.42, 95% CI = 1.23–1.65). The association be-
dence interval [CI] = 1.03–1.22) and recreational canna- tween RML and past-month cannabis use significantly var-
bis legalization (AOR = 1.28, 95% CI = 1.12–1.46) were ied by age (F(3) = 3.92, P = 0.008), such that the strongest

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


2774 Renee D. Goodwin et al.

Table 1 (b) Prevalence of past-month daily cannabis use among adults with children <18 years in the household by cannabis legalization
status and sociodemographic characteristics, NSDUH 2017.

Cannabis legalization status

a a a
No MML (n = 7500) MML (n = 10,800) RML (n = 3900)

Overall 200 (2.35%) 500 (3.22%) 200 (4.21%)


Age
18–25 100 (4.81%) 200 (6.69%) 100 (8.26%)
26–34 <100 (3.21%) 100 (4.79%) <100 (6.10%)
35–49 <100 (1.55%) 100 (2.27%) <100 (3.60%)
≥50 <100 (1.02%) <100 (0.99%) <100 (0.39%)
Gender
Male 100 (3.07%) 300 (4.40%) 100 (5.47%)
Female <100 (1.77%) 200 (2.25%) 105 (3.19%)
Marital status
Married <100 (1.15%) 100 (1.80%) <100 (2.93%)
Widowed/divorced/separated <100 (3.35%) <100 (1.99%) <100 (2.24%)
Never married 100 (4.97%) 300 (6.86%) 100 (8.32%)
Income
<$20,000 <100 (3.63%) 100 (5.43%) <100 (7.20%)
$20,000–$74,000 <100 (2.80%) 200 (4.11%) <100 (4.82%)
≥$75,000 <100 (1.81%) 200 (2.29%) 100 (3.36%)
Race/Ethnicity
Non-Hispanic White 100 (2.26%) 300 (3.38%) 100 (5.13%)
Non-Hispanic Black <100 (3.41%) <100 (4.36%) <100 (7.35%)
Hispanic <100 (1.63%) <100 (2.39%) <100 (3.35%)
Non-Hispanic Other <100 (2.61%) <100 (1.90%) <100 (2.62%)
Education
Less than high school <100 (2.97%) <100 (4.74%) <100 (4.87%)
High school or equivalent <100 (3.28%) 200 (4.29%) <100 (5.89%)
Some college <100 (2.84%) 200 (3.98%) <100 (5.22%)
College graduate or above <100 (0.72%) <100 (1.01%) <100 (2.02%)

a
MML = medical marijuana laws; NSDUH = National Survey on Drug Use and Health; RML = recreational marijuana laws. Unweighted sample sizes must be
rounded to the nearest 100 to use NSDUH restricted-use data in the Research Data Center setting.

association was observed among those ages 26 to 34 DISCUSSION


(AOR = 1.40, 95% CI = 1.15–1.71).
The present study has three main findings. First, the odds
of cannabis use among adults with children in the home
Difference-in-difference estimates for past-month daily
are higher in states with RML, followed by MML, and low-
cannabis use (Table 2b)
est in states without MML or RML. Second, RML adoption
Similarly, both medical (AOR = 1.16, 95% CI = 1.02–1.32) appears to increase the odds of past-30-day cannabis use
and recreational cannabis legalization (AOR = 1.25, 95% and daily cannabis use among adults with children in the
CI = 1.03–1.51) were associated with increases in daily home broadly across nearly all sociodemographic groups.
cannabis use relative to states that did not have legalized Third, MML adoption is associated with increased odds of
marijuana during the study period. For medical cannabis, past-30-day cannabis use and daily cannabis use among
the association significantly varied by age (F(3) = 6.95, adults who are older, have higher incomes, and have
P < 0.001), race/ethnicity (F(3) = 2.76, P = 0.041) and higher formal educational attainment.
education (F(3) = 5.71, P = 0.001), such that the strongest Overall, MML and RML are associated with an increase
associations were observed for those older than 50 years in cannabis use among adults with children living in the
(AOR = 2.09, 95% CI = 1.36–3.24), non-Hispanic White home. This is consistent with previous findings reporting
individuals (AOR = 1.24, 95% CI = 1.07–1.44) and those associations between MML and increased cannabis use
with a college degree or higher (AOR = 1.81, 95% outcomes in US adults overall and within certain age cate-
CI = 1.35–2.44). For RML, the association did not differ gories (e.g. ages 26+) [9,10,18,19]. In 2017, the preva-
by any sociodemographic subgroup. lence of cannabis use among adults living with children

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


Cannabis use in adults with children at home 2775

Table 2 (a) Difference-in-difference estimates for past-month cannabis use among adults with children <18 years in the household,
NSDUH 2004–2017.

Adjusted DiD estimate

MML (AOR) 95% CI P_int RML (AOR) 95% CI P_int

Overall 1.12 (1.03, 1.22) 1.28 (1.12, 1.46)


Age
18–25 1.01 (0.92, 1.10) <0.001 1.01 (0.85, 1.20) 0.425
26–34 1.06 (0.95, 1.19) <0.001 1.40 (1.15, 1.71) 0.563
35–49 1.14 (1.03, 1.27) 0.001 1.39 (1.14, 1.70) 0.579
≥50 1.75 (1.38, 2.22) ref 1.22 (0.80, 1.87) ref
F(3) = 8.432, P < 0.001 F(3) = 3.919, P = 0.008
Gender
Male 1.09 (0.99, 1.19) 0.099 1.21 (1.02, 1.43) 0.214
Female 1.17 (1.06, 1.28) ref 1.37 (1.17, 1.60) ref
F(1) = 2.718, P = 0.099 F(1) = 1.547, P = 0.214
Marital status
Married 1.19 (1.07, 1.32) 0.010 1.46 (1.23, 1.74) 0.043
Widowed/divorced/separated 1.22 (1.05, 1.43) 0.041 1.05 (0.76, 1.45) 0.584
Never married 1.05 (0.96, 1.14) ref 1.16 (0.97, 1.38) ref
F(2) = 4.620, P = 0.010 F(2) = 2.895, P = 0.055
Income
<$20,000 1.07 (0.96, 1.20) 0.080 1.40 (1.08, 1.82) 0.334
$20,000–$74,999 1.04 (0.94, 1.16) 0.015 1.31 (1.06, 1.62) 0.585
≥$75,000 1.19 (1.08, 1.31) ref 1.21 (1.02, 1.45) ref
F(2) = 3.227, P = 0.040 F(2) = 0.490, P = 0.613
Race/Ethnicity
Non-Hispanic White 1.19 (1.08, 1.30) ref 1.23 (1.05, 1.44) ref
Non-Hispanic Black 1.14 (1.01, 1.39) 0.584 1.40 (1.00, 1.96) 0.467
Hispanic 0.89 (0.77, 1.03) <0.001 1.33 (1.06, 1.67) 0.580
Non-Hispanic Other 0.87 (0.72, 1.06) 0.001 1.48 (1.08, 2.02) 0.255
F(3) = 7.851, P < 0.001 F(3) = 0.552, P = 0.647
Education
Less than high school 0.88 (0.78, 0.99) <0.001 1.34 (1.01, 1.78) 0.599
High school or equivalent 1.12 (1.01, 1.24) 0.001 1.04 (0.80, 1.30) 0.035
Some college 1.12 (1.01, 1.24) 0.002 1.26 (1.04, 1.53) 0.322
College graduate or above 1.42 (1.23, 1.65) ref 1.47 (1.15, 1.89) ref
F(3) = 11.937, P < 0.001 F(3) = 1.646, P = 0.177

AOR = adjusted odds ratio; CI = confidence interval; DiD = difference-in-difference; MML = medical marijuana laws; NSDUH = National Survey on Drug Use
and Health; P_int = P value for interaction; RML = recreational marijuana laws. Adjusted ORs are from logistic regression models containing state and year
fixed effects, MML/RML, the sociodemographic factor of interest and their interactions, while simultaneously controlling for all other sociodemographic factors
in the table. The comparison group for all estimates are states that did not enact marijuana laws during the study period; P_int indicates a statistically signif-
icant difference.

in states with RML was higher than the prevalence among RML increased use broadly among adults across
the general adult population across the US [20]. sociodemographic groups. As states shift toward legalizing
The impact of legalization on cannabis use among cannabis for recreational use, a major public health prior-
adults with children living in the home varied by ity should be to ensure that all adults, especially those in
sociodemographic characteristics in several respects. Nota- vulnerable groups, have access and exposure to education
bly, RML had a more uniform impact on adults with chil- and treatment regarding the potential risks associated with
dren living in the household, whereas MML had a youth exposure to parental cannabis use.
heterogeneous impact across certain groups including Although much of the tobacco control and harm re-
age, income, and education. Our results suggest that duction efforts protecting youth from exposure to SHS
MML-associated increased use was particularly salient has focused on cigarette smoking outside/not in the pres-
among adults over the age of 50, both married or ence of children, the potential risks of exposure to SCS
widowed/divorced/separated and among those with and education on the merits of protecting youth from expo-
higher income and educational attainment. In contrast, sure to SCS have received little attention in the cannabis

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


2776 Renee D. Goodwin et al.

Table 2 (b) Difference-in-difference estimates for past-month daily cannabis use among adults with children <18 years in the household,
NSDUH 2004–2017.

Adjusted DiD estimate

MML (AOR) 95% CI P_int RML (AOR) 95% CI P_int

Overall 1.16 (1.02, 1.32) 1.25 (1.03, 1.51)


Age
18–25 0.97 (0.84, 1.11) <0.001 1.06 (0.82, 1.39) 0.633
26–34 1.14 (0.96, 1.36) 0.007 1.38 (1.04, 1.82) 0.278
35–49 1.27 (1.06, 1.53) 0.028 1.41 (1.03, 1.92) 0.222
≥50 2.09 (1.36, 3.23) ref 0.88 (0.41, 1.86) ref
F(3) = 6.953, P < 0.001 F(3) = 1.326, P = 0.264
Gender
Male 1.13 (0.99, 1.30) 0.404 1.13 (0.89, 1.42) 0.083
Female 1.21 (1.02, 1.42) ref 1.46 (1.13, 1.89) ref
F(3) = 6.953, P < 0.001 F(3) = 1.326, P = 0.264
Marital status
Married 1.25 (1.05, 1.49) 0.075 1.54 (1.18, 2.02) 0.019
Widowed/divorced/separated 1.31 (0.99, 1.74) 0.175 1.10 (0.67, 1.80) 0.88
Never married 1.08 (0.94, 1.24) ref 1.06 (0.83, 1.35) ref
F(2) = 2.237, P = 0.107 F(2) = 2.895, P = 0.055
Income
<$20,000 1.15 (0.96, 1.38) 0.605 1.40 (1.00, 1.96) 0.544
$20,000–$74,999 1.10 (0.93, 1.30) 0.286 1.16 (0.87, 1.54) 0.668
≥$75,000 1.21 (1.03, 1.42) ref 1.25 (0.96, 1.61) ref
F(2) = 0.570, P = 0.565 F(2) = 0.408, P = 0.665
Race/Ethnicity
Non-Hispanic White 1.24 (1.07, 1.44) ref 1.18 (0.94, 1.48) ref
Non-Hispanic Black 1.08 (0.89, 1.33) 0.197 1.47 (0.88, 2.44) 0.416
Hispanic 1.04 (0.81, 1.32) 0.137 1.31 (0.92, 1.86) 0.586
Non-Hispanic Other 0.81 (0.58, 1.13) 0.009 1.45 (0.90, 2.35) 0.424
F(3) = 2.764, P = 0.041 F(3) = 0.412, P = 0.744
Education
Less than high school 0.98 (0.82, 1.18) <0.001 1.33 (0.95, 1.88) 0.942
High school or equivalent 1.08 (0.92, 1.27) 0.001 1.09 (0.81, 1.47) 0.482
Some college 1.22 (1.03, 1.45) 0.013 1.26 (0.96, 1.66) 0.928
College graduate or above 1.81 (1.35, 2.44) ref 1.30 (0.83, 2.04) ref
F(3) = 5.711, P = 0.001 F(3) = 0.380, P = 0.767

AOR = adjusted odds ratio; CI = confidence interval; DiD = difference-in-difference; MML = medical marijuana laws; NSDUH = National Survey on Drug Use
and Health; P_int: P value for interaction; RML = recreational marijuana laws. Adjusted ORs are from logistic regression models containing state and year
fixed effects, MML/RML, the sociodemographic factor of interest and their interactions, while simultaneously controlling for all other sociodemographic factors
in the table. The comparison group for all estimates are states that did not enact marijuana laws during the study period; P_int indicates a statistically signif-
icant difference.

legalization effort. Specifically, it is not clear to what degree parents who use cannabis are at increased risk of cannabis
analogous approaches are being taken to prevent SCS ex- use themselves [16], estimating the degree to which can-
posure in children. nabis use has increased among adults with children in
In contrast to cigarettes, there are no public health or the home, whether the distribution of cannabis use differs
clinical guidelines for parents designed to address or edu- among vulnerable sociodemographic groups, and whether
cate about best practices for use of cannabis toward legalization of cannabis for recreational and/or medical use
avoiding or reducing harmful exposures of SCS to chil- has impacted changes in cannabis use are critical to inform
dren’s health. Furthermore, because cannabis is legal for the allocation of resources for prevention and intervention
use only in private dwellings in most states with RML, pro- efforts to protect children in this new era of increasing can-
grams that advocate using combustible products outside nabis laws.
and keeping a “smoke-free home” are not a viable solution Limitations must be considered in interpreting these re-
for reducing SCS. Given that SCS exposure has a number of sults. First, social desirability may differentially affect re-
undesirable health risks for youth, and adolescents with sponses in states based on legalization, thereby leading to

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


Cannabis use in adults with children at home 2777

higher reporting of cannabis use in states where cannabis Declaration of interests


is legal and artificially suppressing reports of use in states
None.
where cannabis remains illegal. Second, the DiD estimates
assume that there are no unmeasured factors that differen- Acknowledgements
tially affected cannabis use outcomes in states that enacted
This work was supported by the National Institutes of
MML/RML during the study period versus comparison
Health/National Heart, Lung and Blood Institute grant
states. Nonetheless, the higher cross-sectional prevalence
R21-HL149773 and the National Institute on Drug Abuse
of use among adults with children in the home in states
grant R01-DA20892 to Goodwin. The National Institutes
with these laws suggests that state marijuana legalization
of Health had no further role in study design; in the collec-
status is an important indicator of potential youth SCS ex-
tion, analysis, and interpretation of data; in the writing of
posure. Third, it is difficult to draw conclusions about the
the report; or in the decision to submit the paper for
impact of cannabis use on children’s health because this
publication.
outcome is not one of the outcomes of the current study
and there are no data on age of the children, which means
Author contributions
that it is not possible to examine the nature of potential im-
pact and level of exposure. Next steps in this line of research Renee Goodwin: Conceptualization; funding acquisition;
should focus on examining these outcomes, especially investigation; methodology. June Kim: Formal analysis; in-
among children who are physically vulnerable (e.g. infants, vestigation. Keely Cheslack-Postava: Investigation; meth-
children with respiratory conditions like asthma) to SCS odology. Andrea Weinberger: Investigation. Melody Wu:
exposure. Investigation. Katarzyna Wyka: Investigation; methodol-
The current data are also unable to address ogy. Meyer Kattan: Investigation; methodology.
completely the degree to which cannabis use among
adults affects children’s exposure to SCS in the home for References
at least two reasons: we do not have information on loca-
1. King K., Martynenko M., Bergman M., Liu Y., Winickoff J.,
tion of smoking, whether it occurs in the house and/or in
Weitzman M. Family composition and children’s exposure to
the proximity of children; and we do not have informa- adult smokers in their homes. Pediatrics 2009; 123:
tion about the quantity of use because even daily canna- e559–e564.
bis use is likely to produce a different quantity than 2. Dwyer T., Ponsonby A. L., Couper D. Tobacco smoke exposure
non-daily use. Further, we do not have information about at one month of age and subsequent risk of SIDS--a prospec-
tive study. Am J Epidemiol 1999; 149: 593–602.
the methods of cannabis use among parents, which
3. Matt G. E., Quintana P. J., Hovell M. F., Bernert J. T., Song S.,
would affect the level of possible SCS exposure. Recent ev- Novianti N., et al. Households contaminated by environmen-
idence indicates most cannabis use occurs via smoking tal tobacco smoke: sources of infant exposures. Tob Control
whereas edible forms account for less than 10% of use 2004; 13: 29–37.
among US adults [21,22]. Future research is needed to 4. U.S. Department of Health and Human Services. Surgeon Gen-
obtain specific estimates of the degree to which dose eral’s Report—The Health Consequences of Involuntary Exposure
to Tobacco Smoke. Atlanta, GA: U.S. Department of Health
and length of exposure are changing among children in and Human Services, Centers for Disease Control and Preven-
the home. Furthermore, exposure of children to SCS at tion, National Center for Chronic Disease Prevention and
a certain age could be more detrimental than at other Health Promotion, Office on Smoking and Health; 2006.
ages because of the “critical period” effect, but the 5. Collins B. N., Nair U. S., Hovell M. F., DiSantis K. I., Jaffe K.,
NSDUH does not specify age of children in the household. Tolley N. M., et al. Reducing underserved children’s exposure
to tobacco smoke: a randomized counseling trial with mater-
Research using biomarkers of cannabis and tobacco expo-
nal smokers. Am J Prev Med 2015; 49: 534–44.
sure in children of different ages and in households 6. Levy D., Chaloupka F., Gitchell J. The effects of tobacco control
implementing varying degrees of protection (e.g. house- policies on smoking rates. J Public Health Manag Pract 2004;
hold smoking bans) is warranted to advance the field. 10: 338–53.
Our findings suggest that both MML and RML are as- 7. Hasin D. S., Saha T. D., Kerridge B. T., Goldstein R. B., Chou
S. P., Zhang H., et al. Prevalence of marijuana use disorders
sociated with increases in cannabis use among adults
in the United States between 2001-2002 and 2012-2013.
with children in the home. We found that RML have a JAMA Psychiat 2015; 72: 1235–42.
broader impact whereas MML are associated with in- 8. Compton W. M., Han B., Jones C. M., Blanco C., Hughes A.
creases among adults who are older, with higher educa- Marijuana use and use disorders in adults in the USA,
tion, and income levels. As an increasing number of 2002-14: analysis of annual cross-sectional surveys. Lancet
Psychiatry 2016; 3: 954–64.
states legalize cannabis for medical or recreational use,
9. Goodwin R. D., Pacek L. R., Copeland J., Moeller S. J., Dierker
educational efforts to prevent unintended consequences L., Weinberger A., et al. Trends in daily cannabis use among
in terms of risks to children via exposure to parental can- cigarette smokers: United States, 2002-2014. Am J Public
nabis use may be essential. Health 2018; 108: 137–42.

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


2778 Renee D. Goodwin et al.

10. Hasin D., Sarvet A., Cerdá M., Keyes K., Stohl M., Galea S., 19. Cerda M., Mauro C., Hamilton A., Levy N. S., Santaella-
et al. US adult illicit cannabis use, cannabis use disorder, and Tenorio J., Hasin D., et al. Association between recreational
medical marijuana laws. JAMA Psychiat 2017; 74: 579–88. marijuana legalization in the United States and changes in
11. Rusby J., Westling E., Crowley R., Light J. Legalization of recre- marijuana use and cannabis use disorder from 2008 to
ational marijuana and community sales policy in Oregon: 2016. JAMA Psychiat 2019.
impact on adolescent willingness and intent to use, parent 20. Weinberger A. H., Zhu J., Levin J., Barrington-Trimis J. L.,
use, and adolescent use. Psychol Addict Behav 2018; 32: Copeland J., Wyka K., et al. Cannabis use among US adults
84–92. with anxiety from 2008 to 2017: the role of state-level can-
12. Volkow N. D., Baler R. D., Compton W. M., Weiss S. R. Adverse nabis legalization. Drug Alcohol Depend 2020; 214: 108163.
health effects of marijuana use. N Engl J Med 2014; 370: 21. Schauer G. L., King B. A., Bunnell R. E., Promoff G., McAfee
2219–27. T. A. Toking, vaping, and eating for health or fun: mari-
13. Volkow N. D., Swanson J. M., Evins A. E., DeLisi L. E., Meier juana use patterns in adults, U.S., 2014. Am J Prev Med
M. H., Gonzalez R., et al. Effects of cannabis use on human be- 2016; 50: 1–8.
havior, including cognition, motivation, and psychosis: a 22. Borodovsky J. T., Crosier B. S., Lee D. C., Sargent J. D., Budney
review. JAMA Psychiat 2016; 73: 292–7. A. J. Smoking, vaping, eating: is legalization impacting the
14. Pacek L. R., Mauro P. M., Martins S. S. Perceived risk of regular way people use cannabis? Int J Drug Policy 2016; 36: 141–7.
cannabis use in the United States from 2002 to 2012: differ-
ences by sex, age, and race/ethnicity. Drug Alcohol Depend
2015; 149: 232–44.
Supporting Information
15. Pacek L. R., Weinberger A. H., Zhu J., Goodwin R. D. Rapid in-
crease in the prevalence of cannabis use among people with Additional supporting information may be found online in
depression in the United States, 2005–17: the role of
the Supporting Information section at the end of the
differentially changing risk perceptions. Addiction 2019 Epub
ahead of print. article.
16. Madras B. K., Han B., Compton W. M., Jones C. M., Lopez E. I.,
McCance-Katz. Associations of parental marijuana use with Figure S1 (a) Trends in the prevalence of past-month can-
offspring marijuana, tobacco, and alcohol use and opioid mis- nabis use among adults with children living in the house-
use. JAMA Netw Open 2019; 2: e1916015. hold, 2004–2013 National Survey on Drug Use and
17. Williams A. R., Olfson M., Kim J. H., Martins S. S., Kleber H. D.
Health. (b) Trends in the prevalence of past-month daily
Older, less regulated medical marijuana programs have much
greater enrollment rates than newer ‘medicalized’ programs. cannabis use among adults with children living in the
Health Aff (Millwood) 2016; 35: 480–8. household, 2004–2013 National Survey on Drug Use
18. Martins S. S., Mauro C. M., Santaella-Tenorio J., Kim J. H., and Health.
Cerda M., Keyes K. M., et al. State-level medical marijuana
laws, marijuana use and perceived availability of marijuana
among the general U.S. population. Drug Alcohol Depend
2016; 169: 26–32.

© 2021 Society for the Study of Addiction Addiction, 116, 2770–2778


This document is a scanned copy of a printed document. No warranty is given about the
accuracy of the copy. Users should refer to the original published version of the material.

You might also like