Trends in Cannabis and Cigarette Use Among Parents With Children at Home: 2002 to 2015

Data were drawn from the public-use data files from the National Survey on Drug Use and Health (NSDUH) from the years 2002 to 2015. NSDUH provided annual cross-sectional national data on the use of tobacco, other substances, and mental health in the United States, as described elsewhere. The weighted overall response rates for NSDUH during these years ranged from 55.2% (2015) to 71.9% (2002). The median response rate was 65.9%. Analyses were restricted to adult respondents (aged ≥18 years) with children aged <18 years living in the home (N = 173 082). Of these respondents, 2.2% were excluded because they could not be classified per our definitions of current smokers or current nonsmokers (see below), resulting in a total study population of N = 169 259.

Respondents were categorized as current cigarette smokers if they reported the following: (1) smoking all or part of a cigarette in the past 30 days and (2) lifetime use of ≥100 cigarettes. Otherwise, respondents were classified as nonsmokers. Respondents (n = 3573) who reported cigarette smoking in the past 30 days but <100 cigarettes used in their lifetime or who reported smoking in the past but were missing information on use of cigarettes in the past 30 days (n = 250) were excluded from analyses.

Current cigarette smokers were categorized as daily smokers if they reported using cigarettes 30 out of the past 30 days and as nondaily smokers otherwise.

Respondents were classified as current cannabis users if they reported using marijuana or hashish at least 1 out of the past 30 days; current daily users were defined as those who reported using cannabis on 30 of the past 30 days.

The following demographic covariates were incorporated to address potential confounding and to explore moderation of the association between current smoking and current cannabis use among parents with children in the home: respondent sex, age, highest level of education, current marital status, current household income, and race. Categorical measures were used for each covariate as shown in the footnote to Table 1.

The annual prevalence of current cigarette smoking and current and daily cannabis use among adult parents with children in the home was calculated for each year from 2002 to 2015. Separate logistic regression models were fit for each by using a continuous term for calendar year to assess overall linear time trends in prevalence over the study period. Models were unadjusted and adjusted for demographic covariates. This adjustment was repeated for each combination of cigarette and cannabis use.

Logistic regression models were fit to determine if current cigarette smoking was associated with current or current daily cannabis use, unadjusted and adjusted for demographic covariates and calendar year (categorical). We used additional models to examine current cigarette smokers as daily versus nondaily. Moderation of the association between current cannabis use with current cigarette smoking was assessed by adding smoking × demographic covariate product terms to the model. Stratum-specific adjusted risk differences (aRDs) for cannabis use in current smokers versus nonsmokers were calculated by using the average marginal predictions from logistic regression models. Comparisons between strata were performed by using the difference in difference P values.

The annual prevalence of current cigarette smoking, current cannabis use, and current daily cannabis use among adult parents with children in the home was calculated for each year from 2002 to 2015. Separate logistic regression models were fit for each by using a continuous term for calendar year to assess overall linear time trends in prevalence over the study period. Models were unadjusted and adjusted for demographic covariates. This process was repeated for each combination of cigarette smoking and cannabis use.

Finally, the annual prevalence of current any cannabis use and current daily cannabis use was calculated by current cigarette smoking status. Variation in time trends in cannabis use by cigarette smoking status was assessed by using logistic regression models with a continuous term for calendar year and year × smoking interaction, unadjusted and adjusted for covariates. The aRDs for cannabis use between 2002 and 2015 were calculated by current smoking status by using the average marginal predictions from the logistic regression models.

All analyses were conducted by using SAS-callable SUDAAN version 11.0.1 (RTI International, Research Triangle Park, NC) and incorporated survey weights to account for the NSDUH complex sampling design. All results, other than raw counts, were adjusted for sampling weights.

From 2002 to 2015, there was a significant decreasing trend in the prevalence of current smoking from 27.6% in 2002 to 20.2% in 2015 (P < .0001; see Fig 1). In contrast, there were significant increasing trends in the prevalence of current cannabis use from 4.9% to 6.8% (P < .0001) and daily cannabis use from 0.7% to 1.6% (P < .0001). Overall, there was a decrease in the prevalence of either cigarette or cannabis use in homes with children from 29.7% in 2002 to 23.5% in 2015. All trends were statistically significant when unadjusted and when adjusted for demographic covariates.

In 2015, the adjusted odds ratio (aOR) for current cannabis use among current cigarette smokers versus nonsmokers was 3.88 (3.16–4.75). The corresponding aORs for current cannabis use among daily cigarette smokers and nondaily cigarette smokers versus nonsmokers were 3.99 (3.19–5.00) and 3.69 (2.90–4.70) (see Table 1). Consistent patterns were observed in the data from all years combined (see Supplemental Table 4).

In 2015, the aOR for daily cannabis use in current smokers versus nonsmokers was 3.70 (2.46–5.55). Daily cannabis use was more common among nondaily (aOR = 2.68 [1.63–4.40]) and daily (aOR = 4.28 [2.71–6.76]) cigarette smokers relative to nonsmokers (Table 1). Similar associations were observed when data from all years were combined (see Supplemental Table 4).

Between the years 2002 and 2015 combined, current cannabis use was significantly more common among cigarette smokers compared with nonsmokers with a child in the home (13.1% vs 2.6%, aRD = 7.9%; see Table 2). The stratum-specific aRD of current cannabis use between current cigarette smokers versus nonsmokers was statistically significant at the P < .0001 level within each demographic group.

The risk differences for current cannabis use associated with current cigarette smoking varied significantly across strata defined by sex, age, marital status, education, and race.

The increased current prevalence of cannabis use associated with smoking was greater among men (aRD = 10.2%) than women (aRD = 6.3%) and among the younger people with children in the home (aRD = 11.2%, 18–25-year-olds) compared with those 50 and older (aRD = 3.8%). In terms of education, the largest difference in risk for current cannabis use between current smokers and nonsmokers occurred among college graduates (aRD = 9.4%). In terms of race, the increase in risk of current cannabis use in current smokers versus nonsmokers was highest among African American respondents (aRD = 10.4%). Although the relationship between current cannabis use and smoking was significant at each income stratum, the magnitude of the difference was similar for all levels of income.

In 2015, nearly 1 in 5 current cigarette smokers used cannabis in the past month (17.4%, up from 11.0% in 2002; see Table 3). Meanwhile, 4.0% of current nonsmokers used cannabis in the past month in 2015 versus 2.4% in 2002. Tests for trend revealed that the prevalence of current cannabis use increased both among current cigarette smokers (aOR = 1.04 [1.03–1.05]) and nonsmokers (aOR = 1.05 [1.04–1.07]) with children in the home from 2002 to 2015; there was no difference in the rate of this increase between the groups (P = .21). However, the aRD for current cannabis use in 2015 vs 2002 was higher among current smokers (aRD = 4.9%) than among current nonsmokers (aRD = 1.8%; P value for difference in differences < .0001).

The prevalence of daily cannabis use was higher in 2015 than in 2002 among current cigarette smokers (4.6% vs 1.7%) and among current nonsmokers (0.76% vs 0.28%; see Table 3). Tests for trend revealed that these increases were significant and did not differ by smoking status (see Table 3). However, the aRD for daily use in 2015 vs 2002 was higher among current smokers (aRD = 1.9%) than among current nonsmokers (aRD = 0.6%; P value for difference in differences < .0001). From 2002 to 2015, the percentage of parents with children under age 18 in the home who used neither cigarettes nor cannabis in the past month increased. The prevalence of smoking only decreased, whereas the prevalence of cannabis use alone or in combination with cigarette smoking increased (see Fig 2).

Recent data suggest that children’s exposure to SCS carries health risks similar to, or worse than, exposure to STS.⁶ There is scant research on the effects of exposure to both cannabis and tobacco on children’s health and development. However, recent studies suggest that children exposed to both SCS and STS may be at heightened risk for cognitive and emotional problems.^16,17 To understand the prevalence of joint exposure to SCS and STS, we estimated changes in the prevalence of cannabis use among current cigarette smokers and nonsmokers living in homes with children from 2002 to 2015, investigated the relationship between cigarette smoking and cannabis use among adults living in homes with children aged <18 in the United States, and identified high-risk groups. The results of this study support the public health gains in reducing overall child STS but raise other public health concerns about child exposure to SCS and especially high risk for combined exposures to SCS and STS in certain subpopulations.

Despite the declining prevalence of cigarette smoking in households with children over this period, an estimated 15.2 to 25.0 million children were living with a parent who was a current cigarette smoker in 2015.¹⁸ Moreover, cannabis use increased among parents with children in the home from 2002 to 2015. This occurred among both cigarette smokers and nonsmokers. Although the net effect shows an overall decline in use of either cigarettes or cannabis in homes with children from 29.7% in 2002 to 23.5% in 2015, there was an increase in the percent of homes with both. Therefore, the increase in cannabis use appears to be compromising progress of tobacco control, especially for vulnerable groups. An estimated 5.3 to 8.0 million children in the United States were living with a parent who was a current cannabis user in 2015. In addition, the prevalence of cannabis use is nearly 4 times as high among cigarette smokers and is increasing rapidly in this group, suggesting that the potential for dual STS-SCS exposure may be even greater now than in the past due to children living with cigarette smokers.

Our findings that both any current cannabis use and daily cannabis use are more common among tobacco smokers compared with nonsmokers are consistent with previous studies.^19,–21 This is a concern for 2 reasons. First, there is the potential for synergistic and/or at least additive effects of exposure to common toxins found in both tobacco and cannabis smoke, increasing risks associated with child exposure in the home. Second, the combination of unique toxins between tobacco and cannabis smoke exposure increases the number of potential risks and consequences for exposed children. Although there is still room for improving STS reduction efforts in homes, public health campaigns have had notable success in reducing smoking and the exposure of nonsmokers to STS in public and work places.²² There also has been growing success of promoting smoke-free homes to protect children from exposure to STS.^23,–25 Our results reveal a potential need to translate these efforts to address child exposure to SCS in the home. To the extent that cannabis use is an obstacle to smoking cessation,^26,27 going forward, protecting children from any type of secondhand smoke exposure may require assessment of both tobacco and cannabis smoking among parents and more intensive and innovative approaches to smoking cessation for those parents.

Several demographic differences were identified in the relationship between cannabis use and cigarette smoking among parents. The prevalence of cannabis use was highest among young parents who smoke cigarettes, with 1 in 5 ages 18 to 25 using cannabis versus 7% of those over 50, and relative to only 1 in 20 parents ages 18 to 25 who were non–cigarette smokers having used cannabis. This suggests that young children may be at particularly high risk for exposure because they spend more time in the home and in closer proximity to a parent than older children and teenagers who may be at school, out with friends, or working for much of the day. Cannabis use was most common in the lowest income group for both cigarette smokers and nonsmokers. This finding highlights another public health priority to address this disparity: ensuring that vulnerable groups have access and exposure to education and treatment designed to protect children from exposure.

We are unable with the current data to address completely the degree to which use among parents affects children’s exposure in the home for at least 2 reasons: (1) we do not have information on location of smoking, whether it occurs in the house and/or in the proximity of children; (2) we do not have information about the quantity of smoke (1 cannabis cigarette per day versus 2 packs of cigarettes), which is relevant because even daily cannabis use is likely to produce a different quantity than cigarettes. Unlike cigarettes, it remains illegal in most places to smoke cannabis outdoors and in a range of public areas. Therefore, there is reason to believe that cannabis use is even more likely to occur in the home than cigarette smoking given their differences in legal status. In addition, although we did not have information about the methods of cannabis use among parents, which would affect the level of possible SCS exposure, recent evidence indicates the vast majority of cannabis use occurs via smoking, whereas edible forms account for <10% of use among US adults.^28,29 Future researchers need to obtain specific estimates of the degree to which level of exposure is changing among children in the home. Researchers using biomarkers of cannabis and tobacco exposure in children of different ages and in households implementing varying degrees of protection (eg, household smoking bans) are needed to advance the field.

Despite the decline in cigarette use among parents with children in the home in the United States, we suggest that cannabis use is increasing among parents with children aged <18 years in the home, particularly among parents who also smoke cigarettes. Public health efforts that have shown success in decreasing exposure to STS in the home may be complicated by increased use of other smoked products, such as cannabis. Parents may benefit from education about protecting children from marijuana products, paraphernalia, waste, and smoke.

aOR

adjusted odds ratio

aRD

adjusted risk difference

NSDUH

National Survey on Drug Use and Health

SCS

secondhand cannabis smoke

STS

secondhand tobacco smoke