Can this be done by tommorrow at the due date and time?????????

708

T here is a wealth of literature highlighting the negative physical (eg, type II diabetes, car- diovascular problems) and psychosocial (eg,

depression, low self-worth) consequences of ado- lescent obesity.1-3 However, less attention has been given to the role adolescent weight status plays in future health-risk behaviors, such as problematic substance use. With adolescent overweight and obesity rates remaining high (33.6% overweight, 18.4% obese 12-19 years),4 and substance use more prevalent in young adulthood than any other developmental period,5 identification of adolescent weight status as a predictor of future problematic substance use behavior is likely to have a signifi- cant impact on research and clinical work aimed to reduce multiple health risks in the transition from adolescence to adulthood.

Adolescence is a crucial period for prevention ef- forts aimed to reduce problematic substance use in young adulthood. According to the National Survey of Drug Use and Health,5 young adults have the highest rates of current tobacco use (39.5% overall including 33.5% cigarette use) and illicit drug use (21.4%), with 19.0% using marijuana in the past month. Binge drinking has been reported for 39.8% and heavy alcohol use for 12.1% of 18- to 25-year-

olds. In the past 30 years, many epidemiological longitudinal studies have identified several key risk factors for problematic substance use, includ- ing regular cigarette smoking, binge drinking, and marijuana use, in adolescence and young adult- hood. Temperament,6 behavioral disinhibition,7 ex- ternalizing behaviors,8 poor parental monitoring,9 lack of parental support,10 negative peer interac- tions,11 and affiliation with deviant peers12 have been well-established as critical factors involved in the development of problematic substance use.13-15 Considering the array of risk factors in adolescence contributing to future problematic substance use, it is likely that other health-risk conditions, such as overweight or obesity status, are linked to prob- lematic substance use behavior.

Little is currently known about the relationship between adolescent weight status and future prob- lematic substance use; however, use of an adoles- cent developmental framework is likely to increase our understanding of why this relationship may be a significant one to address. One explanation may be that a shared underlying factor like impulsivity may explain co-occurring obesity and problematic substance use. As children learn to self-regulate behaviors, those who have difficulties with self- control are more likely to over-consume energy- dense food contributing to obesity risk16,17 and en- gage in antisocial behaviors leading to substance abuse and dependence.18,19 Although a shared underlying factor explanation is plausible, under- standing adolescent behavior without considering the social context is incomplete.

H. Isabella Lanza, Research Associate and Christine E. Grel- la, Professor-in-Residence, Semel Institute for Neuroscience and Human Behavior, University of California, Los Angeles. Paul J. Chung, Associate Professor, Department of Pediatrics, University of California, Los Angeles. Correspondence Dr Lanza: [email protected]

Does Adolescent Weight Status Predict Problematic Substance Use Patterns?

H. Isabella Lanza, PhD; Christine E. Grella, PhD; Paul J. Chung, MD

Objectives: To identify underlying pat- terns of cigarette smoking, alcohol use, and marijuana use in young adulthood, and ascertain whether adolescent over- weight or obesity status predicts prob- lematic substance use patterns. Methods: The study included 15,119 participants from the National Longitudinal Study of Adolescent Health (Add Health) at Wave 1 (11-19 years) and Wave 3 (18-26 years). Latent class analysis was conducted. Re- sults: Participants were classified into a Low Substance Use (35%), Regular Smok-

ers (12%), High-risk Alcohol use (33%), or High Substance Use (20%) class. Over- weight/obese adolescents had a greater likelihood of belonging to the Regular Smokers class. Conclusions: Overweight/ obese adolescents are at higher risk of en- gaging in regular cigarette smoking with- out problematic alcohol or marijuana use.

Key words: adolescence, alcohol, ciga- rette smoking, marijuana, obesity, young adulthood

Am J Health Behav. 2014;38(5):708-716 DOI: http://dx.doi.org/10.5993/AJHB.38.5.8

Lanza et al

Am J Health Behav.™ 2014;38(5):708-716 709 DOI: http://dx.doi.org/10.5993/AJHB.38.5.8

Critical to the discussion on risk-taking behav- iors is the knowledge that social standing among peers is a prominent goal for most adolescents. Taking into consideration the important of self- regulation for risk-taking outcomes, Steinberg’s social neuroscience perspective on adolescent risk-taking20 posits that increases in risk-taking are a result of heightened sensitivity to the social context and its rewards (ie, peer acceptance), as well as slower-developing self-regulatory processes linked to rational decision-making. Both failure to be accepted by peers and desire for higher social status may increase vulnerability to risk behav- iors, such as substance use. Earlier work by Tajfel and Turner21,22 highlights why overweight and obese adolescents may be more likely to experience a negative peer context, which increases vulner- ability to later risk-taking. As adolescents derive their self-concept from the social group(s) to which they belong, social status is often achieved by be- having in ways that are normative for the group. Adolescents who do not fit the group norm, such as those who are different in physical appearance (eg, obese adolescents), are less likely to be ac- cepted by peers.23,24 Overweight and obese adoles- cents are indeed at higher risk for peer alienation and victimization than normal-weight peers.25-27 Those deviating from the group norm may try to overcome their poor social status by engaging in behaviors (eg, substance use) that will increase status among certain social groups, like deviant peers.28 They also may engage in risky behaviors, like substance use, as a way to cope with the nega- tive feelings stemming from poor social status.29,30

It appears that overweight and obese adolescents may be experiencing a social context and lack of self-regulation that increases their risk of engag- ing in problematic substance use as they transi- tion into adulthood. Prior cross-sectional stud- ies have provided tentative evidence that higher weight status in adolescence is related to problem- atic cigarette smoking and alcohol use. A positive relationship between cigarette smoking and body mass index (BMI) has been reported among ear- ly adolescent Danish boys.31 In a study of Portu- guese adolescents, obese girls and boys were more likely to report daily alcohol consumption and fre- quent drunkenness compared to non-obese ado- lescents.32 A study of Taiwanese adolescents also found that girls and boys with higher BMI were more likely than counterparts to report regular al- cohol use and cigarette smoking, but not other il- licit drugs.33 Furthermore, findings using a large sample of US adolescents reported that obese girls, but not boys, were more likely to use alcohol and smoke cigarettes, but not marijuana.34

Findings from longitudinal studies examining adolescent weight status and substance use are more inconclusive and have mainly focused on to- bacco use. A study of early adolescents in the US indicated that obese girls were less likely to initiate tobacco use 2 years later compared to non-obese

girls.35 On the other hand, another US study and a Swedish study found that smoking initiation was more likely among obese girls compared to non- obese girls.36,37 No significant relationships were found among boys. A recent study using 2 samples of US adolescents indicated that BMI did not pre- dict alcohol or other substance use 2 years later.38 Similarly, an epidemiological study of rural US adolescents did not find a significant association between obesity trajectory and substance use.39 The inconsistent results from these longitudinal studies call attention to the need for additional prospective and more comprehensive research.

The current study sought to clarify past find- ings on adolescent weight status and risk of prob- lematic substance use by utilizing a large sample of US adolescents to identify whether overweight or obesity status in adolescence is a predictor of unique patterns of problematic substance use in young adulthood. To achieve this end, a person- centered approach, latent class analysis (LCA), was utilized to identify unique patterns of prob- lematic substance use by considering responses to multiple items on cigarette smoking, alcohol use, and marijuana use simultaneously. Prediction of adolescent weight status to problematic substance use patterns was then assessed.

METHODS Participants

The current study used data from The Nation- al Longitudinal Study of Adolescent Health (Add Health), consisting of a nationally representative sample of adolescents in grades 7-12 in the US during the 1994-95 school year. Participants were enrolled in 80 high schools and 52 middle schools. During the initial wave of the study, 20,745 ado- lescents in grades 7-12 (11-19 years of age) par- ticipated in both a school survey and in-home in- terview between April and December 1995. Written informed consent was obtained from both parent and adolescent. Participants were eligible to par- ticipate in the second wave of data collection about one year later (April-August 1996). A third wave of in-home interviews occurred from July 2001 to April 2002, which included 15,197 young adults aged 18-26 years. Information was collected on mental and physical health, health-risk behav- iors, and contextual factors related to family, peer, school, and neighborhood.

The analytic sample for the current study includ- ed 15,119 of the 15,197 participants interviewed at Wave 3 (young adulthood) during 2001-02; we ex- cluded 78 persons for whom no substance use data were available. Of these 15,119 young adults, 53% were women; 54% White, 21% Black, 15% Latino; 7% Asian; 3% other. The average age of participants at Wave 1 was 16.10 years ± 1.72 and at Wave 3 was 22.47 years ± 1.76. Wave 1 (adolescent) BMI% was available for 97% of the analytic sample, with 25% meeting overweight (14%) or obesity (11%) status. In terms of socio-economic variables, 93% of partic-

Does Adolescent Weight Status Predict Problematic Substance Use Patterns?

710

ipants in the sample reported mother education at Wave 1. Of available data, 20% of mothers received less than a high school education; 34% were high school graduates; 19% completed some college; and 28% completed college or beyond. Also, 75% of parents of participants in the analytic sample com- pleted an in-home interview at Wave 1, including annual household income. Of available data, 29% reported less than $25,000, 33% between $25,000 and $49,999; 23% between $50,000 and $74,999; and 8% between $75,000 and $99,999.

Measures Individual characteristics. Adolescents were

asked to report their sex (1 = female, 0 = male) and race/ethnicity at Wave 1. Ethnicity variables for African-American, Asian, Latino, and white were dummy coded (eg, 1 = African-American, 0 = non-African-American). Adolescents also re- ported on residential mother’s education status at Wave 1 (1 = less than a high school education; 2 = high school grad; 3 = completed some college; 4 = completed college or beyond), which was used as a proxy for SES, given that household income was only available for 75% of participants who had completed parent interviews at Wave 1.

Weight status. At Wave 1 (11-19 years old), ado- lescents self-reported their height and weight. Self- reported height and weight data have been found to be reliable for 96% of adolescents in the Add Health sample.40 Height and weight were used to calculate age- and sex-specific BMI [weight(lbs)/[height(in)2] x 703] percentiles using the Centers for Disease Prevention (CDC) 2000 growth charts.41 Obesity was defined as having a BMI percentile at or above the 95th percentile, and overweight defined as hav- ing a BMI percentile at or above the 85th percentile and below the 95th percentile. Adolescents either meeting overweight or obesity status were com- bined to create an overweight/obesity indicator (1 = overweight or obese, 0 = non-overweight or obese).

Substance use. At Wave 3 (18-26 years), young adults responses to multiple questions on cigarette smoking, alcohol use, and marijuana use were in- cluded in analyses. Five items related to cigarette smoking were selected, which included whether par- ticipants had ever: (1) tried cigarette smoking, even just 1 or 2 puffs (0 = no, 1 = yes); (2) smoked an en- tire cigarette (0 = no, 1 = yes); (3) smoked cigarettes regularly, that is, at least 1 cigarette every day for 30 days (0 = no, 1 = yes); and (4) smoked at all in the past 30 days (0 = no, 1 = yes). They were also asked (5) how many cigarettes smoked per day in past 30 days, which was recoded into a categorical item (0 = none, 1 = 1, 2 = 2-9, 3 = 10-20, 4 = 20+ cigarettes).

Six items on alcohol use were included in analy- ses. Participants were asked: (1) whether they had drank more than 2 or 3 times since June 1995 (Wave 1) (0 = no, 1 = yes); (2) days consumed al- cohol in the past year (0 = none, 1 = couple times a year, 2 = couple times a month, 3 = 1-2 times a week, 4 = 3-7 times a week); (3) days consumed

5 or more drinks in past 12 months (0 = none, 1 = couple times a year, 2 = once a month or less, 3 = couple times a month/week, 4 = 3-7 times a week); (4) days consumed 5 or more drinks in the last 2 weeks (0 = none, 1 = once, 2 = 2-9, 3 = 10+ times); (5) whether they had been drunk in the past year (0 = none, 1 = couple times a year, 2 = once a month or less, 3 = couple times a month/ week, 4 = 3-7 times a week); and (6) whether they had driven while drinking since June 1995 (Wave 1) (0 = no, 1 = yes).

Three items related to marijuana use were also selected for analyses. Participants were asked: (1) whether they had used marijuana since June 1995 (Wave 1) (0 = no, 1 = yes); (2) whether they had used marijuana in the past year (0 = no, 1 = yes); and (3) number of times marijuana consumed in the past 30 days, which was recoded into a categorical item (0 = none, 1 = once, 2 = 2-9, 3 = 10+ times).

Planned Analyses Latent class analysis (LCA) is used to identify

underlying patterns among observed categorical indicators (eg, substance use behaviors) and clas- sify individuals who respond similarly into latent classes.42-44 LCA is an iterative process using full information maximum likelihood estimation. Us- ing Mplus version 7,45 model-building steps were taken to select the best-fitting class model of sub- stance use in young adulthood and then ascertain whether adolescent weight status and other in- dividual characteristics predicted membership in particular substance use classes.

Statistical indices, parameter estimates, and practical implications are used to determine the best-fitting model.43,44,46 The unconditional model is first specified (ie, 1-class model), which is then used as a comparison for an increasing number of class- es until the models specified no longer converge or have useful application. Statistical indices, like the Bayesian Information Criterion (BIC)47 and the Lo- Mendell-Rubin likelihood ratio test (LMR LRT),48 as well as interpretability of classes are key in deter- mining model selection. Item-response probabilities refer to the likelihood that an individual in a given latent class will endorse a particular item response. They are used to confirm that individuals in each latent class have similar response patterns to the observed indicators and that class response pat- terns are distinct from each other. After selecting the best-fitting model, covariates are added to the model. Logistic regression coefficients are estimated by setting the beta parameter to 0 for the reference class; thus, providing an estimation of log-odds that indicate an endorsement of a covariate for a certain class relative to the reference class.

Ethnicity, sex, and a proxy for socioeconomic status (maternal education) were included in the covariates analysis in addition to weight status, as each is strongly tied to disparities in obesity prevalence and substance use risk. Among adoles- cents, African Americans and Latinos have higher

Lanza et al

Am J Health Behav.™ 2014;38(5):708-716 711 DOI: http://dx.doi.org/10.5993/AJHB.38.5.8

overweight and obesity prevalence than Whites, and boys are more likely to be overweight or obese compared to girls.4 In young adulthood, men are more likely to use marijuana, cigarette smoking is more prevalent among Whites than African Ameri- cans, and Whites and Latinos report more binge drinking than African Americans.5 Generally, SES indicators like parental education and household income have shown that lower SES is associated with higher weight status,49 although associations with substance use are mixed.5

RESULTS Descriptive Data

Table 1 compares the average rate of overweight/ obese adolescents’ substance use in young adult- hood to non-overweight/obese adolescents. Across all cigarette smoking indicators, overweight/obese adolescents had higher rates compared to non- overweight/obese adolescents. In contrast, over- weight/obese adolescents had lower rates of high- risk alcohol use compared to non-overweight/ obese adolescence. No significant differences were found between groups for marijuana use.

Latent Class Analysis Model selection. LCA was conducted to identify

latent classes of cigarette smoking, alcohol use, and marijuana use in young adulthood. Table 2 presents the statistical fit indices for 5 classes (the 6-class model did not converge). Model selection is generally based on a scree-like test, in which bet- ter fitting models are represented where the indices begin to level off.50 Although the 5-class model had

the lowest values in fit criteria, indices began to level off significantly after the 3-class model; con- sequently, the 3-, 4-, and 5-class models were fur- ther explored before selecting a best-fitting model. Examination of parameter estimates identified the 4-class model as best-fitting the data in terms of classifying the underlying heterogeneity of ciga- rette, alcohol, and marijuana use in the sample. Class probabilities, homogeneity of item-response within classes, and distinct item-response patterns across classes were assessed to determine which model was most interpretable.43,46 The 3-class mod- el was able to identify a low and a high substance use class, but the third class lacked homogeneity. Two classes in the 5-class model lacked homogene- ity and there was a lack of distinctiveness between 2 classes. Figure 1 illustrates the item-response probabilities of each class of the 4-class model. Classes appear homogenous and distinct, and class sizes were found to be substantial as well.

Identified classes. Considering item-response probabilities for each class (Figure 1), classes were identified based on endorsement of substance use behaviors. A little over one-third of the participants (34.7%) reported very low substance use behavior (Class 1; Low Substance Use). Of these, close to 80% stated that they had never smoked an en- tire cigarette, over 95% stated they had not been drunk or binged on alcohol, and almost 90% had never tried marijuana.

Participants in the Regular Smokers class (Class 2; 12.4%) showed high endorsement of cigarette use, with 96% reporting having been a regular smoker at some point (at least 1 cigarette a day for

Table 1 Cigarette, Alcohol, and Marijuana Use in Young Adulthood: Overweight/Obese vs. Non-overweight/Obese Adolescents

Overweight/Obese Non-overweight/Obese Cigarette % Tried cigarette Smoked entire cigarette*** Smoked regularly*** Smoked past 30 days*** No. cigarettes per day (more than 1)***

74 61 41 35 32

73 59 38 32 29

Alcohol % Drank since W1** Alcohol in past year (multiple times per month or week)*** Binge drink in past year (multiple times per month or week) * Binge drink in past 2 weeks* Drunk past year (multiple times per month or week*** Drunk driving***

76 40 20 31 16 21

78 46 22 33 19 24

Marijuana % Marijuana since W1 Marijuana past year No. of times used marijuana in past year(more than once)

44 31 18

45 32 17

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

Does Adolescent Weight Status Predict Problematic Substance Use Patterns?

712

past 30 days) and all reporting smoking in the past 30 days. Over one-third (38%) smoked 2-9 ciga- rettes and just over half (52%) smoked 10 or more cigarettes on days smoked in the past 30 days. Similar to the Low Substance Use class, problem- atic alcohol use was very low. Also, although 40% had tried marijuana since the start of the study, only one-fourth had used marijuana in the past year.

The High-risk Alcohol Use class (Class 3; 32.6%) was characterized by problematic alcohol use but less risky cigarette and marijuana use. Approxi-

mately 70% stated they had used alcohol multi- ple times a month or week in the past year, about 80% indicated they had engaged in binge drinking or been drunk in the past year, and 50% stated they have binged in the last 2 weeks. Also, 35% reported drunk driving. In regards to cigarette and marijuana use, few had ever been regular smokers (16%), and almost none had smoked in the past 30 days. Although 39% reported using marijuana in the past year, less than one-fourth reported using more than once in the last year.

Last, the High Substance Use class (Class 4;

Table 2 Latent Class Model Fit Indices (N = 15,119)

Classes Log

Likelihood Free

Parameters BIC Adjusted

BIC AIC LMR LRT p- value for k-1

1 -174995.39 30 350279.49 350184.15 350050.78 N/A 2 -146775.79 61 294138.63 293944.78 293673.58 .000 3 -134435.31 92 269756.00 269463.63 269054.62 .000 4 -128469.95 123 258123.61 257732.72 257185.89 .000 5 -123837.44 154 249156.94 248667.54 247982.89 .000

Note. AIC = Akaike information criterion; BIC = Bayesian information criterion; LMR LRT = Lo-Mendell-Rubin likelihood ratio test.

Figure 1 Substance Use Classes in Young Adulthood: Item-response Probabilities

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Tr ie

d s

m o

ki ng

W ho

le c

ig ar

et te

E ve

r r eg

ul ar

s m

o ke

r S

m o

ke d

p as

t 3 0

d ay

s N

o c

ig ar

et te

s d

ay A

lc o

ho l s

in ce

W 1

D ay

s al

co ho

l p as

t y ea

r B

in g

e d

rin k

p as

t y ea

r B

in g

e d

rin k

p as

t 2 w

ks D

ru nk

p as

t y ea

r D

riv e

d ru

nk M

ar iju

an a

si nc

e W

1 M

ar iju

an a

p as

t y ea

r N

o m

ar iju

an a

p as

t …

Tr ie

d s

m o

ki ng

W ho

le c

ig ar

et te

E ve

r r eg

ul ar

s m

o ke

r S

m o

ke d

p as

t 3 0

d ay

s N

o c

ig ar

et te

s d

ay A

lc o

ho l s

in ce

W 1

D ay

s al

co ho

l p as

t y ea

r B

in g

e d

rin k

p as

t y ea

r B

in g

e d

rin k

p as

t 2 w

ks D

ru nk

p as

t y ea

r D

riv e

d ru

nk M

ar iju

an a

si nc

e W

1 M

ar iju

an a

p as

t y ea

r N

o m

ar iju

an a

p as

t …

Tr ie

d s

m o

ki ng

W ho

le c

ig ar

et te

E ve

r r eg

ul ar

s m

o ke

r S

m o

ke d

p as

t 3 0

d ay

s N

o c

ig ar

et te

s d

ay A

lc o

ho l s

in ce

W 1

D ay

s al

co ho

l p as

t y ea

r B

in g

e d

rin k

p as

t y ea

r B

in g

e d

rin k

p as

t 2 w

ks D

ru nk

p as

t y ea

r D

riv e

d ru

nk M

ar iju

an a

si nc

e W

1 M

ar iju

an a

p as

t y ea

r N

o m

ar iju

an a

p as

t …

Tr ie

d s

m o

ki ng

W ho

le c

ig ar

et te

E ve

r r eg

ul ar

s m

o ke

r S

m o

ke d

p as

t 3 0

d ay

s N

o c

ig ar

et te

s d

ay A

lc o

ho l s

in ce

W 1

D ay

s al

co ho

l p as

t y ea

r B

in g

e d

rin k

p as

t y ea

r B

in g

e d

rin k

p as

t 2 w

ks D

ru nk

p as

t y ea

r D

riv e

d ru

nk M

ar iju

an a

si nc

e W

1 M

ar iju

an a

p as

t y ea

r N

o m

ar iju

an a

p as

t …

P r o b a b i l i t y

Item

3-7 times wk/20+

Couple times mo or wk/10+

Once a mo/2-9

Yes/Once/Couple times yr

No/None

Class 1 Low Substance Use

34.7%

Class 2 Regular Smokers

12.4%

Class 3 High-risk Alcohol Use

32.6%

Class 4 High Substance Use

20.2%

 

Lanza et al

Am J Health Behav.™ 2014;38(5):708-716 713 DOI: http://dx.doi.org/10.5993/AJHB.38.5.8

20.2%) represents participants with the highest levels of problematic substance use. Almost all (99%) reported to be regular smokers who smoked in the last 30 days, with over half smoking 10+ cigarettes on days smoked. Over 85% reported drinking multiple times per month or week in the past year, and the majority reported binge drinking and being drunk in the past year (93%) and binge drinking in the past 2 weeks (74%). Over half (55%) stated they had driven drunk. Two-thirds (69%) re- ported marijuana use in the last year, with 45% using more than once in that period.

Predictors of class membership. Covariates were added to the LCA model to determine wheth- er adolescent weight status predicted particular substance use classes. A covariate model includ- ing weight status (overweight or obese), sex, race/ ethnicity (African American, Asian, Latino, White), and maternal education was estimated simultane- ously using latent multinomial logistic regression (Table 3). Using the Low Substance Use class as a reference class, adolescents meeting overweight or obesity status had a greater likelihood of being classified into the Regular Smokers class (β =.37, OR = 1.45, p < .001) than the Low Substance Use class. Similarly, overweight or obese adolescents were more likely to belong to the Regular Smokers class than the High Substance Use (β =.35, OR = 1.42, p < .001) and the High-risk Alcohol Use (β =.43, OR = 1.54, p < .001) classes.

As for other individual characteristics (see Table 3 for statistical values), women were less likely to be classified into any problematic substance use class compared to the Low Substance Use class. African

Americans, Asians, and Latinos were less likely to be classified into the High Substance Use or Regu- lar Smokers classes than the Low Substance Use class, and African Americans and Asians also were less likely to belong to the High-risk Alcohol Use class compared to the Low Substance Use class. On the other hand, Whites were more likely to be- long to the High-risk Alcohol Use class compared to the Low Substance Use class. Adolescents re- siding with mothers with higher education status were more likely to be classified into the High Sub- stance Use and High-risk Alcohol Use classes than the Low Substance Use Class, but were less likely to belong to the Regular Smokers class than the Low Substance Use class. The covariate analysis also was conducted with annual household in- come, with results unchanged. Annual household income was not included in the final model because a significant proportion of the sample (25%) did not have household income information, as this infor- mation was collected with parent-report at Wave 1.

DISCUSSION Sole consideration of average group differences

of substance use rates give an incomplete picture of the potential risks overweight or obese adoles- cents face in young adulthood. The average group differences shown in Table 1 suggest that cigarette smoking and alcohol use rates only vary slightly be- tween overweight/obese and non-overweight/obese groups. Statistical, but not clinically meaningful differences, may contribute to a false assumption that weight status is not an important predictor of substance use. However, with the utilization of

Table 3 Estimated Odds Ratios (OR) of Class Membership in Relation to Obesity,

Sex, Race/Ethnicity, and Education Indices ased on a Multinomial Latent Class Regression Model

Reference class: Low Substance Use (Class 1) vs Regular Smokers

(Class 2) vs High-risk Alcohol

Use (Class 3) vs High Substance Use

(Class 4) β(SE) OR(95%CI) β(SE) OR(95%CI) β(SE) OR(95%CI)

Overweight/ obese vs non-overweight/obese .37(.07)*** 1.45(1.26-1.66) -.06(.05) .95(.85-1.04) .02(.06) 1.02(.91-1.15) Female vs Male -.20(.06)** .82(.73-.92) -.67(.05)*** .51(.46-.56) -.95 (.06)*** .39(.34-.44) African American vs. non-AA -1.23(.18)*** .29(.21-.42) -1.03(.15)*** .36(.27-.48) -1.75 (.17)*** .17(.12-.24) Asian/Pacific Islander vs non-API -.96(.20)** .38(.26-.57) -.58(.16)*** .56(.41-.77) -.97(.18)*** .38(.27-.54) Latino vs non-Latino -1.19(.18)*** .31(.21-.43) -.11(.16) .90(.65-1.23) -1.08 (.17)*** .34(.24-.47) White vs non-White -.16(.17) .86(.61-1.19) .29(.15)* 1.34(1.003-1.79) .24(.16) 1.27(.93-1.74) Higher vs lower mother education .-.16(.03)*** .85(.80-.90) .20(.02)*** 1.22(1.17-1.27) .09(.02)*** 1.10(1.05-1.14) *** p < .001; ** p < .01; * p < .05

Does Adolescent Weight Status Predict Problematic Substance Use Patterns?

714

a person-centered approach, not only were unique patterns of substance use behavior in young adult- hood identified, but a more comprehensive assess- ment of adolescent weight status as a risk factor for problematic substance use was achieved.

Ultimately, the key relationship between high- er weight status and problematic substance use points to regular cigarette smoking. Overweight or obese adolescents had a greater likelihood of belonging to the Regular Smokers class in young adulthood than any other class. On the other hand, overweight or obesity status in adolescence did not predict greater likelihood of belonging to the High-risk Alcohol Use of High Substance Use classes; consequently, higher weight status does not appear to be a risk factor for problematic al- cohol or marijuana use. Also, given that over- weight or obese adolescents did not have a lower likelihood of belonging to the Low Substance Use class vs problematic substance use classes, higher weight status does not appear to lower the likeli- hood of problematic substance use behavior.

Although the current study clearly indicates that overweight or obese adolescents have a greater likelihood of being a regular smoker in the absence of problematic alcohol or marijuana use compared to non-overweight/obese adolescents, past longi- tudinal studies have reported mixed findings re- garding the role of higher weight status on ciga- rette smoking.35-38 Consideration of alcohol use and marijuana use alongside cigarette smoking, and assessment of cigarette smoking behaviors other than smoking initiation, which was a focus of previous studies, increases understanding of the relationship between weight status and cigarette smoking. One of the main questions arising from the current findings pertains to why higher weight status is related specifically to regular cigarette smoking but not other forms of problematic sub- stance use. Although there are many hypotheses linking higher weight status to cigarette smoking, to date, only a handful of studies present evidence on the potential pathways by which higher weight status in adolescence is linked to cigarette smok- ing. For instance, overweight and obese individu- als may initiate cigarette smoking because they perceive it as an effective weight loss strategy.51,52 Also, perception of being overweight or obese and focus on body size are related to smoking initia- tion, particularly among adolescent girls.37,53 In addition to social factors, biological ones also may inform the relationship between higher weight sta- tus and regular cigarette smoking. Food and drugs are known to activate the same neurological path- ways containing dopaminergic receptors linked to reward sites in the brain54,55 that may explain their co-occurrence. For instance, recent studies examining the pathway from cigarette smoking to obesity from adolescence to young adulthood sug- gest that decreased cigarette smoking is associ- ated with increased weight status.56,57 The negative relationship between decreased smoking and in-

creased weight status suggest smoking and eating have similar neurophysiological response systems, as noted by other studies.58,59 For this reason, bi- directional relationships between weight status and cigarette smoking during the transition from adolescent to young adulthood should be explored more fully in future studies.

Other findings from the current study revealed that ethnic and sex differences in substance use membership were generally in line with previous findings. Similar to prevalence rates reported in the National Survey on Drug Use and Health,5 re- sults showed that African-Americans, Asians, and Latinos were at lower risk of problematic cigarette, alcohol, and marijuana use, and males were at higher risk of belonging to problematic substance use classes. Future research should seek to as- sess whether ethnicity and/or sex moderates the relationship between higher weight status and sub- stance use, as this is yet unclear. Given that ethnic and sex disparities exist among substance use, as well as obesity status,4 these differences may influ- ence the degree to which risk behaviors are engaged in by overweight or obese adolescents from specific ethnic and/or sex groups. For instance, if the objec- tive to engage in regular cigarette smoking for over- weight or obese adolescents is to lose weight, this relationship may be stronger among white females than other groups, as they are most often subjected to social stigma and other negative social conse- quences resulting from higher weight status.60

Of course, several limitations need to be con- sidered. First, the current study relied on self- reported height and weight to measure BMI% for adolescents. Although measured BMI data is fa- vored over self-report, relatively minor differences in reliability between self-reported and measured BMI data have been reported in adolescence40,61 and numerous studies have successfully used ADD Health self-reported BMI data.62-64 Second, it is important to note that weight status was only accounted for at one point in time during adoles- cence. It is unknown whether adolescents meeting overweight or obesity status at Wave 1 met over- weight or obesity status across most or all years of adolescence. Similarly, we expect that some ado- lescents meeting overweight or obese status at dif- ferent points in time during adolescence were not accounted for in the analyses. BMI data are avail- able one year after Wave 1 (Wave 2) and 60% of participants meeting overweight or obesity status in Wave 1 also met overweight or obesity status in Wave 2. Third, the study was unable to examine socio-contextual and neighborhood variables that may inform how higher weight status contributes to cigarette smoking risk, such as low peer status or limited neighborhood resources for maintaining healthy weight. Fourth, the variation of age in the sample at each assessment point (eg, 18-26 years at Wave 3 increases the difficulty in interpreting findings as a result of developmental milestones, such as transition to college when high-risk alco-

Lanza et al

Am J Health Behav.™ 2014;38(5):708-716 715 DOI: http://dx.doi.org/10.5993/AJHB.38.5.8

hol use increases substantially. Acknowledging these limitations, the current

study still has important implications for public health efforts aimed to mitigate health-risk behav- iors among young adults. Specifically, emphasis should be placed on decreasing the risk of ciga- rette smoking in overweight or obese adolescents. Although more research is needed to identify the processes by which adolescent weight status influ- ences cigarette smoking risk, physicians and other health professionals should address smoking risk with overweight or obese adolescents and probe for psychosocial or physiological stressors that may initiate high-risk smoking activity. Adolescents are exposed to many psychosocial stressors as they transition into young adulthood (eg, college, em- ployment, leaving home). Taking on the responsi- bilities of their own health behaviors may be too difficult for some adolescents already negotiating with significant developmental changes. Over- weight and obese adolescents may find themselves dealing with these types of stressors; on top of this they likely may be experiencing social exclusion and victimization. The culmination of risks may contribute to an overweight/obese adolescent’s de- cision to engage in regular smoking behavior as a weight management strategy and/or to decrease anxiety. They may be especially willing to engage in regular cigarette smoking if they believe smok- ing has similar effects on their physiological state as eating does (eg, feeling calm, pleasure). How- ever, without knowing the health service needs of overweight and obese adolescents, as well as other socio-contextual factors that largely influence obe- sity risk (physical activity resources, access to and knowledge of healthy nutrition), intervention ef- forts targeting smoking among overweight or obese youth will not be as effective. Empirical studies that can integrate multiple lines of research on obesity and substance use risk are warranted to answer questions about preventing and treating health risks in this unique population.

Human Subjects Statement The following study has been approved by the

University of California, Los Angeles Institutional Review Board (#10-001106).

Conflict of Interest Statement None of the authors have conflict of interests

pertaining to this study.

Acknowledgement This study is supported by the National Institute

of Drug Abuse (R03DA033497 and T32DA007272).

References 1. Needham BL, Epel ES, Adler NE, Kiefe C. Trajectories of

change in obesity and symptoms of depression: the CAR- DIA study. Am J Public Health. 2010;100(6):1040-1046.

2. American Diabetes Association. Type 2 diabetes in chil- dren and adolescents. Pediatrics. 2000;105(3):671-680.

3. Sweeting H, Wright C, Minnis H. Psychosocial correlates of adolescent obesity, ‘slimming down’ and ‘becoming obese’. J Adolesc Health. 2005;37(5):409.e9-409.e17.

4. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010. JAMA. 2012;307(5):483- 490.

5. SAMHSA, Substance Abuse and Mental Health Services Administration. Results from the 2011 National Survey on Drug Use and Health: Summary of National Find- ings. NSDUHSeries H-44, HHS Publication No. (SMA) 12-4713. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2012.

6. Dick DM, Aliev F, Latendresse SJ, et al. Adolescent alco- hol use is predicted by childhood temperament factors before age 5, with mediation through personality and peers. Alcohol Clin Exp Res. 2013;37(12):2108-2117.

7. Tarter RE, Kirisci L, Mezzich A, et al. Neurobehavioral dis- inhibition in childhood predicts early age at onset of sub- stance use disorder. Am J Psychiatry. 2003;160(6):1078- 1085.

8. Glantz MD, Leshner AI. Drug abuse and developmental psychopathology. Dev. Psychopathol. 2000;12(04):795- 814.

9. Griffin KW, Botvin GJ, Scheier LM, et al. Parenting prac- tices as predictors of substance use, delinquency, and aggression among urban minority youth: moderating ef- fects of family structure and gender. Psychol Addict Be- hav. 2000;14(2):174.

10. Stice E, Barrera M. A longitudinal examination of the re- ciprocal relations between perceived parenting and ado- lescents’ substance use and externalizing behaviors. Dev Psychol. 1995;31(2):322.

11. Kaplow JB, Curran PJ, Dodge KA. Child, parent, and peer predictors of early-onset substance use: a mul- tisite longitudinal study. J Abnorm Child Psychol. 2002;30(3):199-216.

12. Dishion TJ, Owen LD. A longitudinal analysis of friend- ships and substance use: bidirectional influence from adolescence to adulthood. Dev Psychol. 2002;38(4):480- 491.

13. Dodge KA, Greenberg MT, Malone PS. Testing an ideal- ized dynamic cascade model of the development of seri- ous violence in adolescence. Child Dev. 2008;79(6):1907- 1927.

14. Hawkins JD, Catalano RF, Miller JY. Risk and protective factors for alcohol and other drug problems in adoles- cence and early adulthood: implications for substance abuse prevention. Psychol Bull. 1992;112(1): 64-105.

15. Kilpatrick DG, Acierno R, Saunders B, et al. Risk fac- tors for adolescent substance abuse and dependence: data from a national sample. J Consult Clin Psychol. 2000;68(1):19-30.

16. Anzman SL, Birch LL. Low inhibitory control and restric- tive feeding practices predict weight outcomes. J Pediatr. 2009;155(5):651-656.

17. Cutler DM, Glaeser EL, Shapiro JM. Why have Ameri- cans become more obese? J Econ Perspect. 2003;17:93- 118.

18. Gottfredson MR, Hirschi T. A general theory of crime. Stanford, CA: Stanford University Press; 1990.

19. Pratt TC, Cullen FT. The empirical status of Gottfreds- sons and Hirschi’s General Theory of Crime: a Meta anal- ysis. Criminology. 2000;38(3):931-964.

20. Steinberg L. A social neuroscience perspective on adoles- cent risk-taking. Dev Rev. 2008;28(1):78-106.

21. Tajfel H, Turner JC. An integrative theory of intergroup conflict. In Austin WG Worchel S., (Eds). The Social Psy- chology of Intergroup Relations. Monterey, CA: Brooks- Cole; 1979.

22. Tajfel H, Turner JC. The social identity theory of inter- group behavior. In Worchel S. Austin LS (Eds). Psychol-

Does Adolescent Weight Status Predict Problematic Substance Use Patterns?

716

ogy of Intergroup Relations. Chicago: Nelson-Hall; 1986. 23. Sentse M, Scholte R, Salmivalli C, Voeten M. Person-

group dissimilarity in involvement in bullying and its relation with social status. J Abnorm Child Psychol. 2007;35(6):1009-1019.

24. Van Vilerberghe L, Braet C. Dysfunctional schemas and psychopathology in referred obese adolescents. Clinical Psychology and Psychotherapy. 2007;14(5):342-351.

25. Falkner NH, Neumark-Sztainer D, Story M, et al. Social, educational, and psychological correlates of weight sta- tus in adolescents. Obes Res, 2001;9(1):32-42.

26. Pearce MJ, Boergers J, Prinstein MJ. Adolescent obesity, overt and relational peer victimization, and romantic re- lationships. Obes Res. 2002;10(5):386-393.

27. Strauss RS, Pollack HA. Social marginalization of overweight children. Arch Pediatr Adolesc Med. 2003;157(8):746-752.

28. Dishion TJ, Medici SN. An ecological analysis of monthly “bursts” in early adolescent substance use. Appl Dev Sci, 2000;4(2):89-97.

29. Abrantes AM, Brown SA, Tomlinson KL. Psychiatric comorbidity among inpatient substance abusing ado- lescents. J Child Adolesc Subst Abuse. 2003;13(2):83- 101.

30. Bizarri JV, Rucci P, Sbrana A, et al. Reasons for sub- stance use and vulnerability factors in patients with sub- stance use disorder and anxiety or mood disorder. Addict Behav. 2006;32(2):384-391.

31. Dhariwal M, Rasmussen M, Holstein BE. Body mass in- dex and smoking: cross-sectional study of a represen- tative sample of adolescents in Denmark. Int J Public Health. 2010;55(4):307-314.

32. Fonseca H, Matos MG, Guerra A, Pedro JG. Are over- weight and obese adolescents different from their peers? Pediatr Obes. 2009;4(3):66-174.

33. Liu T, Yen J, Ko C, et al. Associations between substance use and body mass index: moderating effects of sociode- mographic characteristics among Taiwanese adoles- cents. Kaohsiung J Med Sci. 2010;26(6):281-289.

34. Farhat T, Iannotti RJ, Morton-Simons BG. Overweight, obesity, youth, and health-risk behaviors. Am J Prev Med. 2010;38(3):258-267.

35. Austin SB, Gortmaker SL. Dieting and smoking initiation in early adolescent girls and boys: a prospective study. Am J Public Health. 2001;91(3):446-450.

36. Caria MP, Bellocco R, Zambon A, et al. Overweight and perception of overweight as predictors of smokeless to- bacco use and of cigarette smoking in a cohort of Swed- ish adolescents. Addiction. 2009;104(4):661-668.

37. Cawley J, Markowitz S, Tauras J. Lighting up and slim- ming down: the effects of body weight and cigarette prices on adolescent smoking initiation. J Health Econ. 2004;23(2):293-311.

38. Pasch KE, Velazquez CE, Cance JD, et al. Youth sub- stance use and body composition: does risk in one area predict risk in the other? J Youth Adolesc. 2012;41(1):14- 26.

39. Mustillo S, Worthman C, Erkanli A, et al. Obesity and psychiatric disorder: developmental trajectories. Pediat- rics. 2003;111(4):851-859.

40. Goodwin E, Hinden BR, Khandewal S. Accuracy of teen and parental reports of obesity and body mass index. Pediatrics. 2000;106(1):52-58.

41. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, et al. CDC Growth Charts: United States Advance Data from Vi- tal and Health Statistics. No. 314. Atlanta, GA: National Center for Health Statistics; 2000.

42. Hagenaars JA, McCutcheon AL. Applied Latent Class Analysis. Cambridge: Cambridge University Press; 2002.

43. Lanza ST, Collins LM, Lemmon D, Schafer JL. PROC

LCA: a SAS procedure for latent class analysis. Struct Equ Modeling. 2007;14(4):671-694.

44. Nylund K, Bellmore A, Nishina A, Graham S. Sub- types, severity, and structural stability of peer victim- ization: what does latent class analysis say? Child Dev. 2007;78(6):1706-1722.

45. Muthén LK, Muthén BO. Mplus User’s Guide (7th ed.). Los Angeles, CA: Muthén & Muthén; 1998-2012.

46. Collins LM, Lanza ST. Latent Class and Latent Transition Analysis: For Applications in the Social, Behavioral, and Health Sciences. Hoboken, NJ: Wiley; 2009.

47. Schwartz G. Estimating the dimension of the model. Ann Stat. 1978;6(2):461-464.

48. Lo Y, Mendell NR, Rubin DB. Testing the num- ber of components in a normal mixture. Biometrika. 2001;88(3):767-778.

49. Wang Y. Cross-national comparison of childhood obesity: the epidemic and the relationship between obesity and socioeconomic status. Int J Epidemiol. 2001;30(5):1129- 1136.

50. Giang MT, Graham S. Using latent class analysis to iden- tify aggressors and victims of peer harassment. Aggress Behav. 2008;34(2):203-213.

51. Lowry R, Galuska DA, Fulton JE, et al. Weight man- agement goals and practices among U.S. high school students: associations with physical activity, diet, and smoking. J Adolesc Health. 2002;31(2):133-144.

52. Plotnikoff RC, Bercovitz K, Rhodes RE, et al. Testing a conceptual model related to weight perceptions, physical activity and smoking in adolescents. Health Educ Res. 2007;22(2):192-202.

53. French SA, Perry CL, Leon GR, Fulkerson JA. Weight concerns, dieting behavior, and smoking initiation among adolescents: a prospective study. Am J Public Health. 1994;84(11):1818-1820.

54. Epstein LH, Jaroni JL, Paluch RA, et al. Dopamine trans- porter genotype as a risk factor for obesity in African- American smokers. Obes Res. 2002;10(12):1232-1240.

55. Volkow ND, Wise RA. How can drug addiction help us understand obesity? Nat Neurosci. 2005;8(5):555-560.

56. Huang DY, Lanza HI, Anglin MD. Association between adolescent substance use and obesity in young adult- hood: a group-based dual trajectory analysis. Addict Be- hav. 2013;38(11):2653-2660.

57. Kvaavik E, Tell GS, Klepp KI. Predictors and tracking of body mass index from adolescence into adulthood: fol- low-up of 18 to 20 years in the Oslo Youth Study. Arch Pediatr Adolesc Med. 2013;157(12):1212-1218.

58. Mineur YS, Abizaid A, Rao Y, et al. Nicotine decreases food intake through activation of POMC neurons. Sci- ence. 2011;332(6035):1330-1332.

59. Seeley RJ, Sandoval DA. Neuroscience: weight loss through smoking. Nature. 2011;475(7355):176-177.

60. Strauss RS, Pollack HA. Social marginalization of overweight children. Arch Pediatr Adolesc Med. 2003;157(8):746-752.

61. Field AE, Aneja P, Rosner B. The validity of self-reported weight change among adolescents and young adults. Obesity. 2007;15(9):2357-2364.

62. Harris KM, Perreira K M, Lee D. Obesity in the transi- tion to adulthood: predictions across race/ethnicity, im- migrant generation, and sex. Arch Pediatr Adolesc Med. 2009;163(11):1022-1028.

63. Merten MJ. Weight status continuity and change from adolescence to young adulthood: examining disease and health risk conditions. Obesity. 2010;18(7):1423-1428.

64. Tanner-Smith EE. Negotiating the early developing body: pubertal timing, body weight, and adolescent girls’ sub- stance use. J Youth Adolesc. 2010;39(12):1402-1416.

Copyright of American Journal of Health Behavior is the property of PNG Publications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.