Body Mass Index and Tobacco-Product Use among U.S. Youth: Findings from Wave 1 (2013–2014) of the Population Assessment of Tobacco and Health (PATH) Study

Abstract

Introduction

Tobacco and obesity are leading contributors to mortality in the United States. Due to emerging changes in youth tobacco use, further examination of co-occurrence of these issues is warranted.

Methods

This study examined associations between body mass index (BMI) and tobacco-product use and whether these varied by gender in a nationally representative sample of 12,416 Wave 1 (2013–2014) U.S. youth (12–17 years) from the Population Assessment of Tobacco and Health Study. Multivariable logistic regression analyses examined the odds of past 30-day tobacco-product use according to BMI. BMI was analyzed categorically using the Centers for Disease Control and Prevention (CDC) BMI-for-age weight status categories (underweight/healthy weight, overweight, and obese) and as a continuous variable.

Results

Youth classified as overweight or obese were not more likely to use any tobacco, cigarettes, e-cigarettes, any cigar, or hookah. However, youth who were obese were more likely to use smokeless tobacco (Adjusted Odds Ratio (AOR) =1.68, 95% Confidence Interval (CI): 1.01, 2.81). There were no significant gender interactions for these associations. When BMI was analyzed continuously, a 5-unit and 10-unit increase was significantly associated with using any tobacco, cigarettes, any cigar, and smokeless tobacco. This linear association was supported by similar results observed for a log-transformed BMI variable.

Conclusions

Findings suggest a continuum between weight increase and tobacco-product use among American youth. Clinicians should consider screening for tobacco use among youth who gain weight within any weight class, not just those considered overweight or obese.

1. INTRODUCTION

Adolescence is generally characterized by increased engagement in risk-taking behaviors including substance use, and those who are overweight or obese are particularly vulnerable to such behaviors (Farhat, Iannotti, & Simons-Morton, 2010). Although cigarette use among youth has dropped to historically low prevalence (Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2017), use of other tobacco products such as e-cigarettes is becoming increasingly common (Kasza, et al., 2017). Similar to obesity’s pervasive negative consequences into adulthood, youth who use tobacco are at increased risk of nicotine dependence and subsequent health problems later in life (Dierker, Swendsen, Rose, He, & Merikangas, 2012; National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012). As tobacco and obesity are the leading contributors to adult mortality in the United States and remain high priority public health concerns, it is critical to examine use of tobacco products among this high-risk population of youth.

Few nationally representative studies have examined current tobacco use (defined by use in the past 30 days) among youth by weight classes, and results are both mixed and limited in key ways. While some studies have found that youth who were obese were more likely to currently use cigarettes in comparison to healthy weight youth (Farhat, et al., 2010; Ratcliff, Jenkins, Reiter-Purtill, Noll, & Zeller, 2011; Zeller, Reiter-Purtill, Peugh, Wu, & Becnel, 2015), others have found no significant associations between obesity and current cigarette or current cigar use (Neumark-Sztainer, et al., 1997; Ratcliff, et al., 2011). Some studies report gender-specific differences, in that females who were overweight or obese were more likely to use cigarettes and smokeless tobacco than females who were a healthy weight, while no such associations were found for males who were obese (Farhat, et al., 2010; Ratcliff, et al., 2011). Further, existing studies suffer from limited assessments of tobacco products as well as inconsistent comparisons of weight categories (e.g., healthy, overweight, and obese) as specified by the CDC recommendations for definitions of these categories.

Therefore, the present study examined (1) associations between body mass index and past 30-day tobacco use across 12 products and (2) the moderating effects of gender on these associations using Wave 1 youth data from the nationally representative Population Assessment of Tobacco and Health (PATH) Study.

2. METHODS

2.1 Study Design and Participants

The PATH Study is an ongoing, nationally representative longitudinal study of 45,971 U.S. adults (18 years and older) and youth (12–17 years) designed to examine tobacco use and health. This paper reports Wave 1 (September 2013-December 2014) data from 12,416 youth participants. Recruitment employed an address-based, area-probability design that oversampled adult tobacco users, young adults, and African-American adults (weighted response rate among youth: 78.4%). An in-person screener was used to select youth and adults from households. The weighting procedures adjusted for oversampling and nonresponse, allowing estimates to be representative of the non-institutionalized, civilian U.S. population. Data were collected using Audio-Computer Assisted Self-Interviews administered in English or Spanish. Further information on the PATH Study design and methods has been published elsewhere (Hyland, et al., 2016) and available at https://doi.org/10.3886/Series606. This study was conducted by Westat and approved by Westat’s Institutional Review Board.

2.2 Measures

2.2.1 Body Mass Index (BMI)

BMI was calculated from respondents’ self-reported height in inches and weight in pounds (weight (lbs.)/[height (in.)] ² × 703). Based on age and gender using the 2000 CDC growth charts (Kuczmarski, et al., 2002), BMI percentiles were calculated. The following CDC recommended BMI-for-age weight status categories were examined: underweight/healthy weight (<5^th percentile to <85^th percentile), overweight (85^th to <95^th percentile), and obese ( 95^th percentile). Given the relatively low percentage of underweight respondents and the virtually identical results observed when treating the underweight group separately, we collapsed the underweight and healthy weight groups for this report.

2.2.2 Tobacco Use

Self-reported ever use of the following tobacco products were assessed: cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, smokeless tobacco (i.e. loose snus, moist snuff, dip, spit, or chewing tobacco), snus pouches, kreteks, bidis, and dissolvable tobacco. A brief description and pictures of each product (except cigarettes) were shown to participants when asked about the products. The rationale to not include a picture and description of a cigarette in the questionnaire is based on the premise that most respondents know what a cigarette is and what it looks like (Hyland, et al., 2016). Additional questions were also asked of cigar users to determine cigar type. Participants who reported ever use of tobacco products were asked when they last used, and those who had used a product in the past 30 days were considered past 30-day users.

‘Any tobacco use’ was defined as having used any tobacco product in the past 30 days, ‘any cigar use’ was defined as having used traditional cigars, cigarillos, or filtered cigars in the past 30 days, and ‘smokeless including snus’ was defined as having used smokeless tobacco or snus pouches in the past 30 days.

2.2.3 Covariates

Past-year substance use was assessed for each of the following: alcohol, marijuana (including blunts), misuse of prescription drugs (i.e. RitalinR or AdderallR; painkillers, sedatives, or tranquilizers), cocaine or crack, stimulants (i.e. methamphetamine or speed), heroin, inhalants, solvents, and hallucinogens. Sensation seeking, a risk factor for substance use (Hoyle, Stephenson, Palmgreen, Lorch, & Donohew, 2002), was assessed based on three modified items from the Brief Sensation Seeking Scale: 1) “I like to do frightening things”, 2) “I like new and exciting experiences even if I have to break the rules”, and 3) “I prefer friends who are exciting and unpredictable”. Response options for each item were summed to create overall (range: 0–12) and mean scores. The scale was found to be internally consistent among youth in the PATH Study (Cronbach’s α=0.76).

Past-month mental health problems were assessed using the internalizing and externalizing problem subscales of the Global Appraisal of Individual Needs - Short Screener (GAIN-SS), modified for the PATH Study (Dennis, Chan, & Funk, 2006) and designed to identify individuals at risk for mental health disorders using a continuous measure of severity. The subscales have been previously validated (Garner, Belur, & Dennis, 2013) and recommended for use in epidemiological samples by the PhenX Toolkit (Hamilton, et al., 2011). Symptoms endorsed in the past month were summed for the internalizing and externalizing problems subscales (complete data for subscale components were required) and categorized into low (0–2 symptoms), moderate (3–7 symptoms), or high (8–11 symptoms) severity (Conway, et al., 2017). The scale was also found to be internally consistent among youth in the PATH Study (Cronbach’s α = 0.77).

Information was also collected on socio-demographics, including age, gender, and race/ethnicity, and parental education.

2.3 Analytic Approach

Distributions of past 30-day tobacco use according to BMI-for-age percentile weight status categories were examined. Multivariable logistic regression analyses evaluated the associations between BMI-for-age percentile weight status categories and tobacco use, as well as continuous BMI and tobacco use, adjusting for socio-demographics, substance use, mental health problems, and sensation seeking.

Moderation by gender was assessed by including interaction terms with BMI-for-age percentile categories in the final models. Additionally, to examine non-linear associations between continuous BMI and tobacco use, analyses were repeated using natural log-transformed BMI, as well as quadratic and cubic terms for BMI, in the final models. All analyses were also repeated excluding “biologically implausible” values of BMI (WHO Expert Committee, 1995). Estimates were weighted to represent the U.S. youth population; variances and confidence intervals (CIs) were estimated using the balanced repeated replication (BRR) method (McCarthy, 1969) with Fay’s adjustment set to 0.3 to increase estimate stability (Judkins, 1990). A djusted odds ratios (AORs) and 95% CIs were calculated for all regression analyses. Two-sided p-values of <.05 were considered statistically significant. Estimates based on fewer than 50 observations in the denominator or the relative standard error greater than 0.30 were suppressed (Klein, Proctor, Boudreault, & Turczyn, 2002). All analyses were conducted using SAS Survey Procedures, version 9.4 (SAS Institute Inc., Cary, NC).

3. RESULTS

Based on the BMI-for-age percentile weight status, 16.3% of youth were overweight and 14.6% of youth were obese. The mean BMI was 22.4 (median: 21.3, SE: 0.06; range: 5.3–86.2) among youth (Table 1).

Table 1.

Tobacco-Product Use by Body Mass Index among U.S. Youth (12–17 years), PATH Study, 2013–2014.

	Body Mass Index (BMI) Percentiles																	Body Mass Index (BMI) Weighted Mean (SE) b: 22.4 (0.06) Range: 5.3–86.2
Past 30- Day Tobacco Use	Underweight/Healthy Weight 69.1%b (Less than the 5th percentile to less than 85th percentile)					Overweight 16.3%b (85th to less than the 95th percentile)						Obese 14.6%b (Equal to or greater than the 95th percentile)						Continuous 5-unit increase			Continuous 10-unit increase			Continuous Natural Log-Transformed
	na	%b	SEb	AORc	95% CIc	na	%b	SEb	AORc	95% CIc		na	%b	SEb	AORc	95% CIc		AORc	95% CIc		AORc	95% CIc		AORc	95% CIc
Any Tobacco	705	8.95	0.39	referent		197	9.79	0.82	1.18	0.94	1.49	168	9.44	0.82	1.15	0.89	1.48	1.13	1.04	1.22	1.27	1.08	1.48	1.97	1.29	3.01
Cigarettes	391	4.69	0.26	referent		99	4.55	0.42	0.99	0.80	1.24	97	5.11	0.58	1.21	0.89	1.64	1.10	1.00	1.21	1.22	1.01	1.47	1.72	1.00	2.95
E-cigarettes	249	3.07	0.23	referent		71	3.45	0.45	1.18	0.88	1.59	63	3.46	0.52	1.23	0.86	1.75	1.10	0.98	1.23	1.21	0.96	1.52	1.67	0.92	3.05
Any cigar	205	2.53	0.19	referent		59	2.80	0.42	1.11	0.77	1.59	56	2.95	0.41	1.17	0.84	1.62	1.12	1.01	1.24	1.26	1.03	1.53	1.91	1.08	3.36
Hookah	148	1.75	0.19	referent		38	1.94	0.38	1.22	0.79	1.89	27	1.47	0.30	0.98	0.64	1.50	1.06	0.93	1.21	1.12	0.86	1.46	1.49	0.75	2.97
Smokeless (incl. snus)	114	1.50	0.16	referent		30	1.56	0.34	1.06	0.60	1.86	41	2.34	0.48	1.68	1.01	2.81	1.22	1.04	1.43	1.48	1.07	2.05	3.10	1.27	7.53
Other Tobaccod	40	0.47	0.09	referent		†	†		†			†	†		†			0.82	0.61	1.10	0.66	0.37	1.20	0.44	0.11	1.71

^aRepresents unweighted sample size (numbers may not sum to the total due to missing data)

^bPercentages (%) and standard errors (SE) are weighted to represent the US youth population

^cAdjusted Odds Ratios (AORs) and 95% confidence intervals (CIs) from multivariable logistic regression analyses adjusted for age, gender, race/ethnicity, parental education, past year alcohol or drug use, past month mental health problems, and sensation seeking

^dOther tobacco includes pipe, dissolvable tobacco, bidis, and kreteks

^†Estimates with a denominator <50 or relative standard error >30% were suppressed; including individual estimates for pipe tobacco, bidis, kreteks, and dissolvable tobacco

Statistically significant associations at p<.05 indicated in bold text

Nearly 9% of youth used any tobacco product in the past 30 days, with the most commonly used tobacco products being cigarettes (4.6%) and e-cigarettes (3.1%) (results not included in table). Other population characteristics and past 30-day prevalence of tobacco-product use have been reported elsewhere (Kasza, et al., 2017).

3.1 BMI association with tobacco use

Table 1 presents the distributions of past 30-day tobacco-product use by weight categories. Except for smokeless tobacco (including snus pouches), multivariable logistic regression models showed no significant associations with tobacco-product use when comparing respondents in the overweight group to those in the underweight/healthy weight group. Youth who were obese were 1.7 times (95% CI: 1.01, 2.81) more likely to use smokeless tobacco. No significant gender interactions were detected for any of the models; thus, stratified results were not presented in the table.

When BMI was analyzed continuously, a 5- and 10-unit increase in BMI was significantly associated with any tobacco (5-unit: AOR=1.13, 95% CI: 1.04, 1.22; 10-unit: AOR=1.27, 95% CI: 1.08, 1.48), cigarette (5-unit: AOR= 1.10, 95% CI: 1.00, 1.21; 10-unit: AOR=1.22, 95% CI: 1.01, 1.47), any cigar (5-unit: AOR=1.12, 95% CI: 1.01, 1.24; 10-unit: AOR=1.26, 95% CI: 1.03, 1.53) and smokeless tobacco (including snus) (5-unit: AOR= 1.22, 95% CI: 1.04, 1.43; 10-unit: AOR=1.48, 95% CI: 1.07, 2.05) use, but not with e-cigarette or hookah use. The quadratic and cubic terms for BMI were not significant. Additionally, results for the log-transformed BMI were similar to those for the 5- and 10-unit BMI models, with the strongest odds found for smokeless tobacco (AOR=3.10, 95% CI: 1.27, 7.53). Excluding “biologically implausible” values (WHO Expert Committee, 1995) had little effect on the results and, as such, all BMI values were retained in the final analyses.

4. DISCUSSION

In this nationally representative sample of U.S. youth, weight categories were not significantly associated with tobacco use. However, those who were obese were more likely to use smokeless tobacco in the past 30 days in comparison to their underweight/healthy weight counterparts. When BMI was analyzed as a continuous variable, significant positive associations were found for any tobacco, cigarettes, any cigar, and smokeless tobacco. Associations between BMI and tobacco use appeared to be linear at the 5- or 10-unit increase, and this linearity was supported by the findings for the log-transformed BMI. These findings suggest that a continuum of weight rather than weight classes may be associated with tobacco-product use among youth. Further, effective nicotine delivery systems such as cigarettes, cigars, and particularly smokeless tobacco may be appealing to youth experiencing weight increases.

While this study did not show many significant associations within weight classes, we observed small linear increases in tobacco-product use prevalence across the classes. However, the relatively low prevalence of past-30-day tobacco use (<10%) may have limited statistical power to detect associations with small effect sizes. Significant associations between continuous BMI and cigarettes, any cigar, and smokeless tobacco suggest that clinicians should consider monitoring changes in weight among youth, aside from weight categories, as a potential risk factor for tobacco use. Although we did not find significant gender interactions, other studies have found that females with a negative body image may be particularly at risk, as these youth are more likely to use tobacco as an appetite suppressant for weight loss (National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012). Other underlying factors that may influence the association between BMI and tobacco use among youth include low self-esteem and negative self-image (Crocker, et al., 2001), as well as negative affect reduction (National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012). Future research should consider how changes in weight and tobacco use among youth may be related to these potential mechanisms.

The findings from this study also point to the potential appeal of effective nicotine delivery systems among youth who are have higher BMIs. While other studies have associated BMI with the use of cigarettes and smokeless tobacco (Farhat, et al., 2010; Ratcliff, et al., 2011; Zeller, et al., 2015), this study expands to cigars as well. Another striking observation was the association between the obese weight category and smokeless tobacco use, as well as the strong association between continuous BMI and smokeless tobacco. Smokeless tobacco is indeed an effective method of nicotine delivery (Fant, Henningfield, Nelson, & Pickworth, 1999), and in combination, it is possible that its oral administration is appealing for youth who are looking to cut down on food intake. Finally, the association could be explained by a high co-occurrence in rural areas, where prevalence of smokeless tobacco and obesity are high (Befort, Nazir, & Perri, 2012; National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012; Vander Weg, Cunningham, Howren, & Cai, 2011). A federal campaign is already underway to help prevent smokeless tobacco use among rural youth (Food and Drug Administration, 2016). Local community prevention efforts that combine messaging for multiple healthy lifestyle choices can also be effective (Ashe, Graff, & Spector, 2011), but future research is needed to determine how tobacco prevention efforts may impact weight change among youth.

This study had several limitations. In contrast with prior research, this study did not find significant gender interactions; this may have been due to limited statistical power or misclassification due to self-reported height and weight among adolescents (Sherry, Jefferds, & Grummer-Strawn, 2007). Nor did this study control for other potential confounders including body image, physical activity, and diet, as these were not assessed in the PATH Study. Youth are known to use tobacco for several reasons, including to curb appetite (National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012). As specific reasons for tobacco use related to weight loss were not assessed in the PATH Study, we were unable to test this hypothesis directly or adjust for this as a potential confounder in the models. This study was also cross-sectional, and therefore we could not evaluate the temporal order of associations. However, given that few longitudinal studies have examined how body weight and tobacco use are related among youth (National Center for Chronic Disease Prevention and Health Promotion & Office on Smoking and Health, 2012), data from this study and follow-up waves of the PATH Study will allow us to continue to investigate this important question.

5. CONCLUSIONS

This study demonstrated that (1) youth who were obese were more likely to use smokeless tobacco in the past 30 days and (2) increases in weight were associated with past 30-changes among youth for future prevention of tobacco-product use, and researchers may further explore how tobacco products that most effectively deliver nicotine are associated with weight of American adolescents.

Highlights.

• Youth who were obese were more likely to use smokeless tobacco in the past 30 days.

• BMI increase was associated with past 30-day smokeless, cigarette, and cigar use.

• Clinicians may consider weight increase among youth for prevention of tobacco use.