Prospective associations between cannabis use and negative and positive health and social measures among emerging adults

Abstract

Background:

In light of on-going policy changes related to cannabis use in the United States, it is important to examine possible associations between cannabis use and subsequent behaviors of public health interest. This study identified prospective associations between cannabis use during first-year post high-school and a wide range of positive and negative health and social measures one year later.

Methods:

Data were from Waves 4 (Time 1; 1st year after high-school) and 5 (Time 2; one year later) of the NEXT Generation Health Study, a national sample of emerging adults in the United States (n = 1915; mean age = 20.2; 61% female). Multinomial logistic regressions adjusting for pertinent covariates were conducted to examine odds of substance use, nutrition, physical activity, sedentary behaviors, school performance, family relations, mental health, driving behaviors and health perceptions at Time 2.

Results:

Compared with non-use, frequent use (20+ times in the past year) at Time 1 was associated with Time 2 negative health and social measures, including risky driving behaviors (AOR = 1.78, CI-1.45–2.19), depressive symptoms (AOR = 1.68, CI-1.43–1.98), unhealthy weight control behaviors (AOR = 1.55, CI-1.31–1.84), psycho-somatic symptoms (AOR = 1.55, CI-1.30–1.83), and low school achievement (AOR = 1.46, CI-1.23–1.75). Frequent users relative to non-users had a lower probability of being overweight and obese (AOR = 0.75, CI-0.60–0.92).

Regarding positive measures frequent users relative to non-users had a higher probability of meeting recommendation of physical activity (AOR = 1.28, CI-1.09–1.51), but a lower probability of consuming fruits and vegetables (AOR = 0.82, CI-0.70–0.96) or attending college/university (AOR = 0.57, CI-0.44–0.75).

Findings:

on occasional cannabis use (1–19 times in the past year) were more similar to frequent cannabis use for negative than positive health and social measures.

Conclusion:

Results demonstrate complex prospective patterns in which significant prospective associations with most adverse measures were found for both occasional and frequent users, and with few significant associations of positive health measures mostly among occasional cannabis users.

Introduction

Cannabis is the most common illicit drug used among adolescents and emerging adults (Cerdá, Wall, Keyes, Galea, & Hasin, 2012; Conway et al., 2013; Leite et al., 2015; Lanza, Vasilenko, Dziak, & Butera, 2015; Pacek, Mauro, & Martins, 2015; Wua, Zhua, & Swartza, 2016). Despite a general decline since the peak in the 1970’s (Lanza, et al., 2015), an increase among those 18–25 years old was reported for the period 2005–2013 based on the National Surveys on Drug Use (Wua et al., 2016). Prevalence of marijuana use among 18–25 age years old in 2015 was 52% for lifetime, 32% in the past year, and 20% in the past month (NIDA, 2016). In light of national trends toward legalization, it is concerning that negative consequences of cannabis use may increase following changes in perceived harmfulness and actual use of cannabis (Cerdá et al., 2012; Pacek et al., 2015; Monte, Zane, & Heard, 2015).

Recent studies have documented emerging evidence that policies related to cannabis, including legalization, may contribute to perceptions and actual use of cannabis. Prior to recent policy changes in the United States, an international comparison of adolescents in the United States, Canada, and the Netherlands showed that prevalence of cannabis use did not differ by countries with different cannabis policies (Simons-Morton, Pickett, Boyce, ter Bogt, & Vollebergh, 2010). A more recent study conducted in the United States found that states that legalized medical cannabis had higher rates of illegal cannabis use and abuse (Cerdá et al., 2012). In Colorado, cannabis legalization was linked to a decline in perceptions of the risks of cannabis use among teens and adults (Schuermeyer et al., 2014). National data showed a similar pattern, with reductions in the perceived risk of regular cannabis use from 51.3% in 2002 to 40.3% in 2012 (Pacek et al., 2015). Recent study suggested that enforcement of medical marijuana laws is associated with decline probability of obesity (Sabia, Swigert, & Young, 2017). These state- and national-level trends call for empirical research that further our understanding of both negative and positive consequences of cannabis use and the potential benefits and versus risks of legalization (Monte et al., 2015).

Negative health and social measures for cannabis use

Negative health outcomes for cannabis use have been shown in past research for psychological, social, and physical variables (Arria, Caldeira, Bugbee, Vincent, & O’Grady, 2016; Barnwell, Earleywine, & Wilcox, 2006; Fergusson & Boden, 2008; Looby & Earleywine, 2007; Simpson, Janssen, Boyce, & Pickett, 2006; Troup, Andrzejewski, Braunwalder, & Torrence, 2016), although an examination of a wide range of these outcomes in a single study is rare. Relative to non-users, cannabis users reported more depressive symptoms (Looby & Earleywine, 2007; Troup et al., 2016), more psycho-somatic symptoms (Simpson et al., 2006; Osborn et al., 2015), and lower levels of subjective well-being (Barnwell et al., 2006). Among first year college students, heavy and moderate cannabis users, relative to non-users showed poorer physical and mental health outcomes, injuries, illness, emotional problems, and psychological distress (Arria et al., 2016).

The literature on the associations between cannabis use and weight-related behaviors is less consistent and more controversial. In a longitudinal study, young adult cannabis users in Australia were less likely to be overweight and obese (Hayatbakhsh et al., 2010). Similarly, in a sample of Inuit, past-year cannabis use was associated with a lower likelihood of obesity (Ngueta, Bélanger, Laouan-Sidi, & Lucas, 2015). Among U.S. adults, cannabis use was associated with lower fasting insulin levels, lower insulin resistance, lower BMI, and smaller waist circumferences (Penner, Buettner, & Mittleman, 2013) and as note before, enforcement of medical marijuana laws found to be associated with decline probability of obesity (Sabia et al., 2017). However, other studies suggested that cannabis use may be associated with increased appetite, “munchies”, overeating and weight gain (Greydanus, Hawver, & Greydanus, 2013; Sidney, 2016). Recent research suggest some confounders in previous studies as acute versus chronic use to be involve in the associations between weight and cannabis use (Sansone & Sansone, 2014). In addition, limited research has been conducted on cannabis use and sedentary behaviors, with one study suggesting a positive association between computer use and ever cannabis use among male adolescents (Lesjak & Stanojevic-Jerkovici, 2015).

Associations with negative social measures for cannabis use have also been documented in past studies. According to the “a-motivational syndrome” theory, cannabis use may underlie symptoms of fatigue and apathy, leading to difficulties with successful progress through life (Barnwell et al., 2006). Indeed, cannabis use was inversely associated with family satisfaction (Jessor, Chase, & Donovan, 1980), and initiation was associated with poorer academic performance (Fergusson & Boden, 2008) and increased rates of school dropout (Lynskey & Hall, 2000). Moreover, Marie and Zölitz (2017) found that the academic performance of students who are no longer legally permitted to buy cannabis increases substantially. In one study, frequent cannabis use, relative to both infrequent use and nonuse, was associated with lower likelihood of earning a bachelor’s degree (Maggs et al., 2015), highlighting the importance of differentiating between infrequent versus frequent cannabis use in relation to academic achievement measures. In Homel, Thompson and Leadbeater (2014) frequent users had the lowest high school grades and the most conduct problems and were least likely to enroll in postsecondary education. Occasional users did not differ from abstainers on high school grades or conduct problems and were no less likely than abstainers to enroll in postsecondary education. Cannabis use is also associated with impaired driving (Li, Simons-Morton, Gee, & Hingson, 2016; Ronen et al., 2010; Robertson, Woods-Fry, & Morris, 2016; Vaca, Li, Hingson, & Simons-Morton, 2016).

Positive health and social measures for cannabis use

There is limited research on associations with positive health of cannabis use, and the findings are mixed. Much of the research has focused on cannabis as a medical treatment to increase appetite (Ko, Bober, Mindra, & Moreau, 2016) or decrease pain (Whiting et al., 2015), decrease PTSD symptoms (Yarnell, 2014), or treat symptoms of ADHD (Milz & Grotenhermen, 2015). Wilens et al. (2007) stated that those with ADHD reported self-medication as their primary reason for cannabis use, while other research, consistent with findings from an online chats study, people reported using cannabis without a prescription because the drug was therapeutic for their ADHD (Mitchell, Sweitzer, Tunno, Kollins, & McClernon, 2016). Few studies that examined happiness as part of quality of life and life satisfaction and found cannabis use does not seem to enhance quality of life (Ventegodt & Merrick, 2003; Fischer, Clavarino, Plotnikova, & Najman, 2015), motivation, happiness, or life satisfaction (Looby & Earleywine, 2007). In one longitudinal study, found that chronic marijuana use in adolescence and emerging adulthood had little effect on life satisfaction in the mid–30 s (White, Bechtold, Loeber, & Pardini, 2015). Not surprisingly, the primary motivation for cannabis use among young people is the pleasure of getting high (Rella, 2015).

Research findings on possible associations between cannabis use and fruits and vegetables intake were not consistent. An early study found that adolescent marijuana abusers reported eating more snack foods and less fruit and vegetables than other groups (Farrow, Rees, & Worthington-Roberts, 1987). A study with high school students found no significant associations between cannabis use and fruit and vegetable intake (Arcan, Kubik, Fulkerson, Hannan, & Story, 2011). The findings on physical activity are also mixed. A study of school-aged adolescents from Slovenia reported no association between cannabis use and physical activity (Lesjak & Stanojevic-Jerkovici, 2015). Another study with French university students suggested cannabis might enhance sports performance, especially for participants who practiced more extreme sports as windsurfing, skiing or snowboarding (Lorente, Peretti-Watel & Grelot, 2005 in Gillman, Hutchison, & Bryan, 2015). In contrast, chronic use of cannabis has been associated with reduced physical activity (Greydanus et al., 2013), lower fitness measured by a maximal effort fitness test and increased heart rate at less than maximal exercise levels (Sidney, 2002).

Purpose of the study

The available literature on prospective relationships between cannabis use and health/social measures is limited both by the restricted range of outcomes examined (much less on positive measures) and by the insufficient differentiation by type of users (e.g., occasional vs. frequent users). Utilizing a large and recent national sample of U.S. emerging adults, the purpose of this study was to examine prospective effects of cannabis use frequency at Time 1 on a wide range of health and social behaviors at Time 2. We hypothesized that relative to non-users, occasional or frequent cannabis users would report higher probabilities of negative health or social (sedentary behavior, over-weight/obese weight status, unhealthy weight control behaviors, psycho-somatic symptoms, depressive symptoms, distracted driving, risky driving, riding with an impaired driver, and school achievement) and lower probabilities of positive health or social measures (physical activity, fruit and vegetable intake, optimism, happiness and self-perceived good health, perceived satisfaction from family relationships, enrollment in post-secondary education, educational aspirations and high grade point average).The primary contributions of this study are in comparing prospective associations for never and occasional cannabis use with those for frequent cannabis use, and examining a variety of negative and positive health measures within the same cohort.

Methods

Procedure and sample description

Data were from the NEXT Generation Health Study, a longitudinal study of a nationally representative U.S. sample (Simons-Morton, Haynie et al., 2016; Simons-Morton et al., 2017). This national cohort was assessed annually from the 10th grade (Wave 1) through two years past high school graduation (Wave 5). Using multistage sampling primary-sampling units consisted of school districts or groups of school districts stratified across the nine U.S. census divisions. Within this sampling framework, 81 schools out of 137 (58.4%) randomly selected schools agreed to participate starting in the 2009–2010 school year (Conway et al., 2013). Within each participating school, 10th grade classes were randomly selected to participate. Participant assent and parental consent were obtained, upon turning 18, participant consent was obtained. This study used data from web-based assessments during the first year (Time 1, 2013; n = 2177) and second year after high school (Time 2, 2014; n = 2141). The protocol was approved by Institutional Review Board of the Eunice Kennedy Shriver National Institute of Child Health and Human Development. For the purposes of this research, 1915 participates (38.8% male) with data on cannabis use at Time 1 and participated in Time 2 were retained in the analysis. Participants mean age was 20.2 (± 0.54) at Time 2. For this study, we used data on cannabis use from waves 4 and 5 only, timepoints where the prevalence of cannabis use was sufficiently frequent to examine.

Description of the variables

Cannabis Use (Currie et al., 2004; Time 1 and 2)

Frequency of cannabis use in the last 12 months was assessed using an item taken from the Health Behaviour in School-aged Children study. Response options included 7 choices ranging from never to 40 times or more. Referencing past studies that used this item to distinguish frequent/regular from occasional/some cannabis use among young adults (LaBrie, Hummer, & Lac, 2011; Schulenberg et al., 2005), three cannabis groups were created: (1) never used in the past year (69.2%), (2) occasional use (those who reported using once to 19 times in the past year; 20.2%), and (3) frequent use (those who reported 20 times or more in the past year; 10.7%). In this paper, experimenters (frequency of once or twice use in the past year) were included in the occasional use group as they were not significantly different from the non-use group in most of the measures and to avoid small size groups and strengthen statistical power. Control variables- Demographic variables included the following: gender, race/ethnicity (non-Hispanic white, non-Hispanic African American, Hispanic and Other), family affluence (Currie et al., 2004; low, medium, high), and family structure (biological parents, biological and step parent, single parent, or other). Long-term illness (Time 1, Currie et al., 2008) was assessed by a question asking about physician-diagnosed long-term illness, disability, or medical condition and prescription medication use- dichotomized to yes/no. Self-reported 30-day alcohol and tobacco use (Time 1, Currie et al., 2004) were dichotomized to never versus ever.

Negative health measures (assessed at Time 1 and Time 2)

Sedentary behaviors (Currie et al., 2004) Participates were asked about how many hours a day do they usually play games on a computer or game console (weekday and weekend) and separately, watch network TV shows, cable, webisodes, or videos (weekday and weekend) in their free time. Response ranged from none at all to about 7 or more hours a day. Responses were summed (range 4–36) and then dichotomized by the median (12) to none versus a half an hour a day or more (4 items scale).

Overweight and obese (Centers for Disease Control and Prevention, 2010) Participants self- reported their height and weight. BMI was calculated and weight status categories were created based 2000 CDC criteria guidelines as appropriate for the respondents age (National Center for Health Statistics, 2015). The variable was dichotomized to overweight/obese versus normal weight. Underweight is a different health issue than overweight, and would be inappropriate to include with the normal weight group. (Farhat, Iannotti, & Simons-Morton, 2010). Thus the category of underweight was treated as missing and excluded from the analysis (10% of the sample).

Unhealthy weight control behaviors (Neumark-Sztainer et al., 2012) Participants were asked about engaging in the weight loss/weight control behaviors (yes/no): fasted, ate very little food, ate very little food specifically because they planned to drink alcohol afterwards, and skipped meals. Items were summed (range 0–4) then dichotomized by the median (0) as 0- never versus 1- ever (4 items).

Psycho-somatic symptoms (Currie et al., 2004) Participants were asked how often they have had the following in the last 6 months: headache, stomach-ache, back ache, feeling low, irritability or bad temper, feeling nervous, difficulties in getting to sleep and feeling dizzy. Responses ranged from rarely or never to about every day on a 5 point scale. Items summed (range 8–40) and dichotomized by the median (13) to 0- less symptoms, 1- more symptoms (8 items).

Depressive symptoms scale (Irwin et al., 2010) study participates completed the PROMIS scale, including 7 days reports of- felt like you couldn’t do anything right; felt everything in their life went wrong; felt unhappy; felt lonely; felt sad; felt alone; thought that their life was bad; could not stop feeling sad. Responses ranged from never to almost always in 5 values scale. Items summed (range 8–40) and dichotomized by the median (13) to 0- less symptoms, 1- more symptoms (8 items).

Driving measures- All driving measures were first filtered by a question about how many days in the last 30 days the participant had driven a motor vehicle. All participants whom wrote more than 0 entered the analysis (88.3% of the sample).

Distracting secondary tasks scale (Simons-Morton, Li, Ehsani, & Vaca, 2016) Participants recorded the number of days they engaged in 14 tasks that took their eyes away from the road while driving: (e.g. made a call on a phone, read/sent a text message, looked in the mirror to fix hair or put on makeup). Items summed (range 0–420) and then dichotomized by the median (34) to less (34 day or less) versus more (35 days or more) distracted driving (14 items).

Risky driving behaviors scale (Simons-Morton, Haynie et al., 2016) Participants recorded number of days they engaged in 23 risky driving behaviors (e.g. exceeded the speed limit in residential or school zones, purposely tailgated or followed another vehicle very closely, made an illegal U-turn; Changed lanes without signaling). Items summed (range 0–690) and then dichotomized by the median (26) to less (26 days or less) versus more (27 days or more) risky driving (23 items scale).

Riding with cannabis-impaired driver (CDC, 2010) Participants were asked how many times they rode in a car or other vehicle driven by someone else who had been smoking cannabis. Responses ranged from 0 to 6 times or more. Variable was dichotomized to 0 = never and 1 = 1 time or more.

Low academic achievement Participants reported grades from the last semester they were in school on a scale ranging from A to D and No grade/Don’t know. Responses dichotomized by the median grade which was a B into 0- higher grades (B to A), 1- lower grades (B- to D), and no grade/don’t know recoded as missing).

Positive health measures

Physical activity- Participants reported the number of days out of the last 7 they were physically active, and how many days they engaged in vigorous activity, (CDC, 2010; Li et al., 2016) A dichotomous variable was calculated representing meeting or not meeting the recommendations by Centers for Disease Control and Prevention (2008) guidelines of MVPA 60- minutes per day/5 days per week.

Fruit and vegetable intake (CDC, 2010; Vereecken, De Henauw, & Maes, 2005) Participants completed questions regarding their fruit and vegetable intake as part of a nutrition screener. Fruit and vegetable intake frequency was calculated by summing responses to fruit, 100% fruit juice, green vegetables, orange vegetables, and beans. Responses ranged from never to 4 or more times per day. Responses were summed (range 4–28) and then dichotomized by the median (10) to 0 = less fruits and vegetables, 1 = more fruits and vegetables (4 items).

Optimism scale (Scheier, Carver, & Bridges, 1994). Participants indicated the extent of their agreement (strongly agree to strongly disagree, 1–5) with 10 items regarding their level of optimism (e.g. In uncertain times, I usually expect the best; I’m always optimistic about my future; I don’t get upset too easily; I rarely count on good things happening to me). 3 items were recoded so that a higher response indicated greater optimism. Variables were summed (range 19–56) then dichotomized by the median (40) to 0 = less optimistic, and 1 = more optimistic (10 items).

Happiness (Levin & Currie, 2014) Participants circled the number on a scale of 1 to 10 that best described how happy they are with their life, with 10 being very happy to 1 being very unhappy. Scale dichotomized at the median (8) as 0 = less happy and 1 = happier.

Positive health perception (Currie et al., 2004) Participants endorsed the statement that best describes their perception of their health. Responses ranged from excellent to poor. Variable was dichotomized to 0 = fair and poor perception of health, 1 = excellent and good perception of health.

Satisfaction with family relationships (Currie et al., 2004) Participants circled the number on a scale from 1 to 10 that best describes their feelings of satisfaction with their family relationships, with 10 indicating very good relationships in our family to 1 indicating very bad relationships. Scale dichotomized by the median (2) to 0 = worse relationships, 1 = better relationships.

Attending school (Neumark-Sztainer, Story, Hannan, & Croll, 2002) Participants indicated if they were not attending school, attending high school, technical, community college, or a 4-year college or university. 2 variables were made. One contrasted those in any secondary education (=1) with those not attending school or still in high school (=0). The second contrasted those attending a 4-year college or university (=1) with those not attending college or university (=0).

Educational Aspirations- Participants were asked what is the most schooling they think they will complete. Responses were: Not finish high school; Graduate from high school; Go to trade, technical or vocational school; Complete an associate degree or 2 years or less of college; Graduate from a 4-year college or university; Go to graduate or professional school. Order scale cut by the median (4). Variable was dichotomized to complete 2 years in college or less (=0) versus completing college, university or professional school (=1).

High academic achievement (A- or A)- Using the same grade point average scale described above (ranged from A to D and No grade/Don’t know), we coded high grade (A and −A) to indicate high academic achievement (=1) versus as those reported B+ to D. No grade/Don’t know were coded as missing. We included this measure in the analysis as high grades consider being positive measure of success in academic performance.

Data analysis

Analyses were conducted in the SPSS 21, accounting for the complex survey design, including clustering, stratification, and weighting. Sample weights calculated for waves 4–5 were used to account for attrition from the original sample. Initial analyses included frequencies of cannabis use by each control variable at Time 1. To provide a consistent analytic framework, the health measures were dichotomized. Chi-square tests of independence were used to assess unadjusted associations between cannabis use and each outcome. Then, multinomial logistic regressions run for each outcome. For each outcome, three models were constructed: Model 1 included cannabis use (referent = non-use), gender, ethnicity, family affluence, parent education, family structure; Model 2 added alcohol and tobacco; Model 3 added health status and the Time 1 variable for the outcome (i.e., in the model examining Time 2 depressive symptoms, Time 1 depressive symptoms was added). Additional analysis was conducted using the multinomial logistic regressions to indicate P values for differences between AOR of frequent group to occasional group.

Results

Of the 2785 participants in the total sample, 657 (23.5%) participants were lost to follow up due to missing data on cannabis yearly use at Time 1 and 643 (23.1%) were missing in time 2. The missing data from both waves was 870 (31.2%) and the final study sample included 1915 participants. Table 1 shows the characteristics of cannabis use groups in time 1- never, occasional and frequent use by socio-demographic characteristics, health status and substance use. Overall at Time 1, 69.2% reported no use, 20.2% reported occasional use (1–19 times), and 10.7% reported frequent use (20 or more times) in the past year.

Table 1.

Characteristics of cannabis use groups by socio-demographic characteristics, health status and substance use weighted for survey design (Cluster, Weight T1 + T2, Strata).^a

Variables	Never use in T1	Occasional use T1	Frequent use T1	All
Socio-demographic charcteristics
N	1325	386	204	1915
% (Weighted)	69.2%	20.2%	10.7%	100%
Gender
Male	36.0%	40.3%	55.9%	38.8%
Female	64.0%	59.7%	44.1%	61.2%
Ethnicity
White	33.5%	30.6%	37.5%	33.3%
Black/African American	26.6%	31.5%	29.5%	27.9%
Hispanic	34.1%	34.0%	29.7%	33.6%
Others	5.8%	3.9%	3.3%	5.2%
SES- FAS
0-Low	34.2%	31.4%	27.9%	33.0%
1-Medium	45.7%	47.4%	44.5%	45.9%
2-High	20.1%	21.2%	27.6%	21.1%
Health long term illness T1
Any report of health issue	15.9%	17.7%	17.8%	16.4%
No report of health issue	84.1%	82.3%	82.2%	83.6%
Smoking cigarettes in the last 30 days T1
0-Never	89.6%	72.7%	59.3%	83.2%
1-Ever	10.4%	27.3%	40.7%	16.8%
Ever drink alcohol in the last 30 days T1
0-Never	65.5%	29.1%	16.3%	53.5%
1-Ever	34.5%	70.9%	83.7%	46.5%

^aSample descriptive for participants in both waves.

Table 2 shows cross-tabulation of Time 1 cannabis use and Time 2 negative and positive health and social measures. Relative to the nonuse group, greater percentages of those in the occasional and frequent use groups reported negative health and social measures one year later. For example, frequent negative psycho-somatic symptoms were reported among 60.5% of frequent cannabis use group, 56.0% of occasional cannabis use group and 50.8% of cannabis non-use group. Conversely, the patterns on positive measures were less consistent. For example, greater happiness was reported among 58.0% of cannabis nonusers, 49.5% of cannabis occasional users, and 51.5% of cannabis frequent users.

Table 2.

Cross-tabulation frequencies of cannabis use at Time 1 by negative and positive health and social measures at Time 2, weighted for survey design (Cluster, Weight, Strata)

Negative health and social measures at T2			Total	aNever use T1	aOccasionally use T1	aFrequent use T1	aPerson chi-square X2
Negativehealth measures T2	More hours of sedentary behaviors- Computer & TV		53.5%	53.2%	55.5%	51.7%	P = 0.072
	Being overweight and obese (BMI categories)		37.4%	38.4%	35.7%	34.3%	P = 0.006
	More unhealthy weight control behaviors		35.5%	34.8%	35.5%	40.0%	P = 0.003
	Frequent negative psycho-somatic symptoms		52.8%	50.8%	56.0%	60.5%	P = 0.000
	More depressive symptoms		49.2%	46.5%	54.4%	58.2%	P = 0.000
Negative social measures T2	More distracting secondary tasks (only among drivers)		49.9%	47.2%	56.0%	56.9%	P = 0.000
	More risky driving behaviors (only among drivers)		49.2%	44.7%	58.5%	62.3%	P = 0.000
	More riding with marijuana impaired driver (only among drivers)		19.8%	8.8%	36.2%	64.0%	P = 0.000
	Low academic achievement- grade point by median		60.1%	57.4%	63.7%	72.1%	P = 0.000
Positive health and social measures at T2
Positive health measures T2	Meeting physical activity recommendations		28.3%	28.6%	24.1%	34.6%	P = 0.000
	Higher fruit & vegetable intake scale		49.5%	50.3%	48.6%	45.1%	P = 0.003
	More optimism		49.4%	49.8%	48.2%	48.8%	NS
	Greater Happiness (8 and above)		55.7%	58.0%	49.5%	51.5%	P = 0.000
	Higher self-perceived good health		68.5%	67.2%	73.8%	66.8%	P = 0.000
Positive social measures T2	High satisfaction from family relationships		60.7%	61.9%	56.9%	59.8%	P = 0.000
	Attending school	1- Any post-secondary school	28.1%	28.9%	24.5%	29.4%	P = 0.000
		2- University or college	41.8%	41.0%	49.3%	33.4%
	High educational aspirations		66.4%	66.2%	70.4%	59.9%	P = 0.000
	High academic achievement		22.6%	23.7%	22.1%	15.6%	P = 0.000

^aCrosstabulation frequencies and Person chi-square X² for difference between groups.

Table 3 shows adjusted odds ratios for associations between cannabis use at Time 1 and negative health and social measures at Time 2. Significant results in the hypothesis direction are in bold. Significant results oppose to the hypothesis direction are in Italic.

Table 3.

Adjusted odds ratios for associations between cannabis use at Time 1 and negative health and social measures at Time 2 (Weighted for survey design) significant in hypothesis direction (values provided in bold) /counter to hypothesis direction (values provided n italic).

Negative health and social measures (scales median split) at T2	aModel 1 (Sociodemographic)			Model 2 (SD) drink and smoke			Model 3 (SD, drink and smoke) health & previous wave
	AOR Never use (Ref)	AOR Occasional CI (Low-High)	AOR Frequent CI (Low-High)	AOR Never use (Ref)	AOR Occasional CI (Low-High)	AOR Frequent CI (Low-High)	AOR Never use (Ref)	AOR Occasional CI (Low-High)	AOR Frequent CI (Low-High)
Negative health measures T2
Sedentary behaviors Computer & TV (Ref = less)	1.00	1.03 (0.93–1.14)	0.78*** (0.68–0.89)	1.00	1.16 (1.03–1.29)	0.89** (0.77–1.04)	1.00	1.18 (1.05–1.33)	0.90** (0.77–1.06)
Overweight and obese (Ref = normal)	1.00	0.89 (0.80–0.99)	0.88 (0.76–1.01)	1.00	0.92 (0.82–1.03)	0.91 (0.78–1.06)	1.00	1.10 (0.94–1.28)	0.75** (0.60–0.92)
Unhealthy weight control behaviors (Ref = = less)	1.00	1.14 (1.02–1.27)	1.66*** (1.44–1.92)	1.00	1.18 (1.06–1.33)	1.79*** (1.54–2.09)	1.00	1.16 (1.02–1.31)	1.55** (1.31–1.84)
Psycho-somatic symptoms (Ref = less)	1.00	1.37 (1.23–1.52)	2.05*** (1.77–2.37)	1.00	1.24 (1.11–1.39)	1.74*** (1.49–2.03)	1.00	1.15 (1.05–1.35)	1.55** (1.30–1.83)
Depressive symptoms (Ref = less)	1.00	1.42 (1.28–1.57)	1.96*** (1.71–2.26)	1.00	1.32 (1.18–1.48)	1.86*** (1.59–2.16)	1.00	1.20 (1.06–1.35)	1.68*** (1.43–1.98)
Negative social measures T2
Distracting secondary tasks (Ref = = less)	1.00	1.34 (1.18–1.51)	1.71** (1.46–2.01)	1.00	1.15 (1.01–1.31)	1.40* (1.18–1.67)	–	–	–
Risky driving behaviors (Ref = less)	1.00	1.86 (1.65–2.11)	2.32* (1.98–2.73)	1.00	1.61 (1.42–1.84)	1.87 (1.58–2.22)	1.00	1.83 (1.57–2.14)	1.78 (1.45–2.19)
Riding with cannabis impaired-driver (Ref = never)	1.00	6.05 (5.22–7.01)	19.44*** (16.22–23.30)	1.00	4.66 (3.99–5.44)	14.63*** (12.08–17.73)	–	–	–
Low academic achievement (Ref = higher grades)	1.00	1.27 (1.14–1.42)	1.76*** (1.52–2.04)	1.00	1.22 (1.09–1.36)	1.67*** (1.43–1.96)	1.00	1.22 (1.07–1.39)	1.46 (1.23–1.75)

Model 1- Adjusted odd ratios- Controls for sociodemographic characteristics- (Gender, Ethnicity, FAS- Family Affluent Scale, Family-Structure).

Model 2- Adjusted odd ratios- Controls for (sociodemographic characteristics) drink and smoke.

Model 3- Adjusted odd ratios- Controls for (sociodemographic characteristics, drink, smoke) health and previous wave.

Weighted for sampled school clusters, participant weight, Strata.

- Empty cells- no same measure in previous year.

^*p < 0.05,

^**p < 0.01,

^***p < 0.001- for indicate P values for differences between AOR of frequent group to occasional group.

^aMultinomial logistic regressions for each measure independently both for occasional vs. non-use, and frequent vs. non-use in the same model.

Across Models 1–3, the findings contrasting occasional and frequent users to nonusers (except for sedentary behaviors and overweight/obese) were largely consistent. In Model 3, which controls for Time 1 value of each outcome, plus socio-demographics, alcohol, tobacco, health status, compared to non-use, those who reported frequent cannabis use had higher odds of being in the more versus less at-risk group of the following negative health behaviors one year later: driving risky behaviors (AOR = 1.78, CI = 1.45–2.19), and depressive symptoms (AOR = 1.68, CI = 1.43–1.98), unhealthy weight control behaviors (AOR = 1.55, CI = 1.31–1.84), psycho-somatic symptoms (AOR = 1.55, CI = 1.30–1.83) and low school achievement (AOR = 1.46, CI = 1.23–1.75). Each of these outcomes was also significant for occasional vs non-use, and most of the measures showed significant differences between the frequent and the occasional groups. Regarding weight status, only those who reported frequent use compared to non-use had lower odds of being in the overweight/obese group compared to the normal weight group in the fully adjusted model (AOR = 0.75, CI = 0.60–0.92). Occasional users reported more sedentary behaviors than non-users (AOR = 1.18, CI = 1.05–1.33) one year later. Riding with cannabis impaired driver (AOR = 14.59, CI = 12.04–17.68) and distracting secondary tasks (AOR = 1.40, CI = 1.18–1.66), measures that were not included in time 1.

The results of the multinomial regressions for the positive health and social measures are found in Table 4. In Model 3, relative to the non-use group, frequent cannabis use was significantly negatively related to fruits and vegetables intake (AOR = 0.82, CI-0.70–0.96) and attending college or university (AOR = 0.57, CI-0.44–0.75). Occasional cannabis group was negatively related to physical activity (AOR = 0.78, CI-0.69–0.90) and family relationship (AOR = 0.76, CI-0.67–0.86). However, compared to non-use group, and significantly different from the occasional group, those in the frequent use group were more likely to meet the recommendations for physical activity (AOR = 1.28, CI-1.09–1.51). Relative to those in the non-use group and the frequent group, those in the occasional use group were more likely to report self-perceived good health (AOR = 1.49, CI-1.30–1.71), college or university attending (AOR = 1.43, CI-1.18–1.73), and educational aspirations (AOR = 1.56, CI-1.35–1.80) one year later.

Table 4.

Adjusted odds ratios for associations between cannabis use at Time 1 and positive health and social measures at Time 2 (Weighted for survey design) significant in hypothesis direction (values provided in bold) /counter to hypothesis direction (values provided in italic).

Positive health and social measures (scales median split) at T2		aModel 1 (Sociodemographic)			Model 2 (SD) drink and smoke			Model 3 (SD, drink, smoke) health & previous wave
		AOR Never (Ref)	AOR Occasional CI (Low-high)	AOR Frequent CI (Low-High)	AOR Never (Ref)	AOR Occasional CI (Low-High)	AOR Frequent CI (Low-High)	AOR Never (Ref)	AOR Occasional use CI (Low-High)	AOR Frequent CI (Low-High)
Positive health measures T2
Physical activity meet the recommendation (Ref = did not meet the recommendation)		1.00	0.71 (0.63–0.80)	1.17*** (1.01–1.35)	1.00	0.71 (0.63–0.81)	1.19*** (1.01–1.39)	1.00	0.78 (0.69–0.90)	1.28*** (1.09–1.51)
Fruits & vegetables intake (Ref = less fruits & vegetables)		1.00	0.94 (0.85–1.04)	0.81 (0.71–0.93)	1.00	0.93 (0.83–1.04)	0.80 (0.69–0.93)	1.00	0.95 (0.85–1.07)	0.82 (0.70–0.96)
Optimism (Ref = less optimistic)		1.00	1.04 (0.94–1.15)	1.06 (0.93–1.22)	1.00	0.96 (0.86–1.07)	0.94 (0.81–1.09)	–	–	–
Happiness (Ref = unhappy)		1.00	0.68 (0.61–0.75)	0.73 (0.63–0.84)	1.00	0.70 (0.62–0.78)	0.79 (0.68–0.91)	–	–	–
self-perceived good health (Ref = fair, poor)		1.00	1.41 (1.25–1.59)	0.74*** (0.64–0.86)	1.00	1.42 (1.26–1.61)	0.75*** (0.64–0.88)	1.00	1.49 (1.30–1.71)	1.04*** (0.87–1.23)
Positive social measures T2
Family relationships (Ref = worse relationship)		1.00	0.79 (0.71–0.88)	0.93* (0.81–1.07)	1.00	0.76 (0.68–0.85)	0.88 (0.76–1.03(	1.00	0.76 (0.67–0.86)	1.15*** (0.98–1.36)
Attending school	1. Any school (Ref = not attending)	1.00	0.99 (0.86–1.14)	0.89 (0.75–1.05)	1.00	1.09 (0.94–1.26)	1.02 (0.85–1.22)	1.00	0.98 (0.83–1.15)	1.04 (0.85–1.28)
	2. College/Uni. (Ref = not attending)	1.00	1.42 (1.25–1.61)	0.63*** (0.53–0.74)	1.00	1.58 (1.38–1.82)	0.77*** (0.64–0.92)	1.00	1.43 (1.18–1.73)	0.57*** (0.44–0.75)
Educational Aspirations Most schooling you think you’ll complete (Ref = lower)		1.00	1.19 (1.07–1.33)	0.75*** (0.65–0.86)	1.00	1.29 (1.15–1.45)	0.88*** (0.76–1.03)	1.00	1.56 (1.35–1.80)	0.99*** (0.82–1.20)
High academic achievement (Ref = lower grades)		1.00	0.96 (0.85–1.09)	0.63*** (0.53–0.75)	1.00	1.06 (0.93–1.21)	0.68*** (0.56–0.82)	1.00	1.16 (0.99–1.35)	0.90* (0.72–1.13)

Model 1- Adjusted odd ratios- Controls for sociodemographic characteristics- (Gender, Ethnicity, FAS- Family Affluent Scale, Family-Structure).

Model 2- Adjusted odd ratios- Controls for (sociodemographic characteristics) drink and smoke.

Model 3- Adjusted odd ratios- Controls for (sociodemographic characteristics, drink, smoke) health and previous wave.

- Empty cells- no same measure in previous year.

^*p < 0.05,

^**p < 0.01,

^***p < 0.001- for indicate P values for differences between AOR of frequent group to occasional group.

^aMultinomial logistic regressions for each measure independently both for occasional vs. non-use, and frequent vs. non-use in the same model.

Discussion

This study examined prospective associations between cannabis use and multiple health and social measures among generally healthy emerging adults one year later. The results indicate that frequent use of cannabis (20 times or more a year) and occasional cannabis use (1–19 times a year) in the previous year were related to negative social and health measures one year later. Cannabis use was prospectively associated with unhealthy weight control behaviors, higher psycho-somatic symptoms, higher depressive symptoms, higher distractive secondary tasks in driving, higher risky driving behaviors, higher riding with cannabis impaired driver, low school achievement and lower happiness compare to non-use group. In our study, not only did we replicate previous research indicating that frequent use was associated with elevations in these negative outcomes (for example, psycho-somatic symptoms as in Simpson et al., 2006; depressive symptoms as in Looby & Earleywine, 2007; low school achievement as in Lynskey & Hall, 2000 and else), but we also found that occasional use was associated with multiple negative measures. Most of the negative measures were consistent with the hypotheses suggesting adverse health and social outcomes in frequent and occasional use groups compared with the never use group. The findings further indicate significantly worse outcomes for the frequent use group compared to the occasional use group.

Past research (Jessor et al., 1980; Lynskey & Hall, 2000) suggested that early cannabis use is linked to a-motivation and apathy which could lead to poorer academic outcomes, yet this contention has been questioned (Fergusson & Boden, 2008; Pacek et al., 2015; Schuermeyer et al., 2014). In a prior study, Maggs et al. (2015) found that infrequent cannabis users were as likely as non-users to complete school. Similarly, we found that frequent cannabis use was negatively associated with attending college/university. However, occasional use was associated with greater likelihood of college or university attendance and better educational aspirations. One of the explanations for that could be that our occasional use group included many participants who used only once or twice during the past year and that did not interfere with their studying or educational inspiration.

Uniquely, this study examined potential positive outcomes of cannabis use in a population-based sample. Cannabis was found to be associated with depressive symptoms (Looby & Earleywine, 2007; Troup et al., 2016), anxiety problems, and psychological distress (Arria et al., 2016). However, the absence of negative symptoms do not necessarily equate to high levels of life satisfaction. Notably, we examined how cannabis use was associated with subsequent optimism and happiness, and found that both frequent and occasional cannabis use were inversely associated with happiness. These results are consistent with a previous study showing lower happiness and life satisfaction among cannabis use groups (Ventegodt & Merrick, 2003), and extended it by showing that similar findings were not present for optimism. Our findings add to Ventegodt and Merrick by comparing also occasional users and not ever vs. never only. Looby and Earleywine (2007) findings were restricted to comparisons between dependent and non-dependent daily cannabis users and found lower motivation, happiness, or life satisfaction among dependent users. Our findings suggest that greater cannabis use, regardless of frequency, was associated with not only more depressive symptoms, but also reduced happiness. Fischer et al. (2015) found that probability for being unhappy is higher for early onset (15 years old or less) cannabis use, but not between more and less frequent users. It is unclear how cannabis use is linked to future happiness, but other socialization processes or direct consequences of use not examined in this study may be a factor.

In terms of weight status, frequent cannabis use was negatively associated with overweight/obese status for frequent but not occasional users. Our results are consistent with previous findings reported by Ngueta et al. (2015) using an ever versus never cannabis use dichotomy, and found lower BMI in the ever group, in contrast with some other publication which describe cannabis use as associated with overeating and weight gain (Greydanus et al., 2013; Sindey, 2016). Penner et al. (2013) found lower BMI in current use (at least once in the last 30 days) and in past use (at least once but not in the past 30 days) compare to the non-use group. In a study that looked at five different frequencies of use, Hayatbakhsh et al. (2010) found that regardless of how frequently participants used, all using groups were less likely to be in overweight or obese compare to the non-use group. Our finding regarding occasional use differed perhaps because it included some participants with very low levels of use. Possible biological explanation of lower insulin resistance was offered by Penner et al. (2013).

In terms of the probability of meeting the physical activity guidelines, our finding differed in direction of association for frequent (positively associated) and occasional cannabis use (negatively associated). Frequent group finding were consistent with Gillman et al. (2015) that showed frequent cannabis use was associated with athletic performance, although their study measured a very different aspects of physical activity than ours. Their results suggest that athletes might use more cannabis than others to improve sports performance. Our occasional cannabis group finding was consistent with Greydanus et al. (2013) and Sindey (2002) that showed cannabis to be associated with reduced physical activity and lower fitness. A possible explanation for that might be the a-motivation cannabis use is associated with, including a-motivation for physical activity. Similarly, as with our finding on frequent use, other research found cannabis use was not associated with watching TV more than 2 h a day (Lesjak & Stanojevic-Jerkovici, 2015), but occasional use was associated with slightly more sedentary behaviors. Further research is needed to understand this complexity of findings examining matching measures both for physical activity and for sedentary behaviors.

Consistent with Farrow et al. (1987), frequent cannabis use was associated with low fruits and vegetables intake. Our findings update and extend Farrow et al. by demonstrating that this association was robust even after controlling for other substance use including alcohol. Findings suggest that cannabis use, as a risk behavior, tend to be inversely associated with health promoting behaviors and positively associated with other health risk behaviors, as suggested by the Health Promotion Model (HPM; Pender, Murdaugh, & Parsons, 2006) that describes the propensity of a person to adopt or reject a healthy lifestyle. Our findings that those in the frequent and occasional cannabis groups were more likely to use weight control behaviors also support the HPM.

Summarizing the findings we see occasional and frequent cannabis use were prospectively associated with adverse negative measures the year after. In contrast, findings on positive outcomes were less consistent. Frequent cannabis use was associated with two positive results-lower probability for overweight and obese and higher probability of physical activity, and occasional cannabis use was associated with college/university attending, educational aspiration and self-perceived good health. This study is unique in that we were able to adopt a consistent definition of occasional vs. frequent cannabis use to evaluate the prospective associations between cannabis use and a wide range of health and social measures during a critical developmental stage. Additional study strengths include the use of a large, national sample, statistical control for multiple pertinent covariates, as well as the inclusion of positive measures. Study limitations include limited measures of cannabis use that did not fully detail extent of use and the lack of statistical control for other illicit drugs (< 3%) or consideration of state-level cannabis legal status (as our study design did not include state-level random sampling). Future research should consider extending the period of study to address how cannabis use trajectory is associated with both positive and negative outcomes. A fixed effects approach should be considered in future studies in order to provide consistent estimates of within-participants changes in usage over time.

Conclusion

In a national sample of U.S. emerging adults, frequent cannabis users generally showed poorly negative and positive measures while occasional users showed poorly negative measures but well on some positive measures. Distinguishing between frequent to occasional use is important when considering the associations with overweight and obese, physical activity, college/university attending and educational aspiration. Since legal status regarding marijuana is dynamically changing, it is important to inform the public about the potential dangers of using the drug. More research is needed to clarify the nature of the association between cannabis and physical activity or college/university attending. This paper results highlight the adverse negative health and social outcomes for occasional use and for frequent use, a finding that may inform policy regarding cannabis use. Although positive certain measures might be considered useful to specific sub-populations with limitation of frequency, more research should be done to clarify that conclusion.