Adolescent Behavioral Cancer Prevention in the United States: Creating a Composite Variable and Ranking States’ Performance

Abstract

Preventive behaviors established during adolescence can reduce cancer throughout the life span. Understanding the combinations of multiple behaviors, and how these behaviors vary across states, is important for identifying where additional interventions are needed. Using data on 2011-2015 vaccination, energy balance, and substance use from national surveys, we created state-level composite scores for adolescent cancer prevention. Hierarchical Bayesian linear mixed models were used to predict estimates for states with no data on select behaviors. We used a Monte Carlo procedure with 100,000 simulations to generate states’ ranks and 95% confidence intervals. Across states, hepatitis B vaccination was 84.3% to 97.1%, and human papillomavirus vaccination was 41.8% to 78.0% for girls and 19.0% to 59.3% for boys. For energy balance, 20.2% to 34.6% of adolescents met guidelines for physical activity, 4.1% to 15.8% for fruit and vegetable consumption, and 66.4% to 82.0% for healthy weight. For substance use, 82.5% to 93.5% reported abstaining from binge alcohol use, 84.3% to 95.4% from cigarette smoking, and 62.9% to 92.8% from marijuana use. (1) Rhode Island, (2) Colorado, (4) Hawaii and New Hampshire (tied), and (5) Vermont performed the best for adolescent cancer prevention, and (47) Missouri, (48) Arkansas, Mississippi, and South Carolina (tied), and (51) Kentucky performed the worst. However, 95% CIs around ranks often overlapped, indicating lack of statistical differences. Adolescent cancer prevention behaviors clustered into a composite index. States varied on their performance on this index, especially for states at the high and low extremes, but most states did not differ statistically. These findings can inform decision makers about where and how to intervene to improve cancer prevention among adolescents.

Keywords

adolescent health behavioral index energy balance performance rankings substance use vaccination

Cancer is the second most common cause of death in the United States (Xu, Kochanek, Murphy, & Tejada-Vera, 2016); however, up to 50% of cancer deaths (Islami et al., 2018; Song & Giovannucci, 2016) and 30% of cancer cases (Islami et al., 2018) can be prevented through behavioral changes, such as never smoking (or smoking cessation) and engaging in adequate physical activity (National Cancer Institute, 2015).

Lifelong patterns of these behaviors often emerge in adolescence or earlier (Chen & Jacobson, 2012; Fuemmeler, Pendzich, & Tercyak, 2009; Wright, Parker, Lamont, & Craft, 2001). For example, 88% of adult smokers initiated tobacco use before age 18 years (U.S. Department of Health and Human Services, 2012b), and 80% of obese male and 92% of obese female adolescents (age 16-17 years) are obese at age 38 years (Wang, Chyen, Lee, & Lowry, 2008). In addition, health behaviors overlap such that people who engage in one prevention (or risk) behavior are more likely to engage in other prevention (or risk) behaviors (Berrigan, Dodd, Troiano, Krebs-Smith, & Barbash, 2003; Hale & Viner, 2016; Kabat, Matthews, Kamensky, Hollenbeck, & Rohan, 2015). For example, smokers are less likely to consume adequate fruits and vegetables (F&V) (Muff et al., 2010) and are more likely to misuse alcohol (Meader et al., 2016) than nonsmokers. Intervening to promote multiple prevention behaviors in this sensitive developmental period could accrue many benefits to preventing cancer (and other chronic diseases) among adolescents as they age.

Composite measures that synthesize multiple indicators can provide insight into the overlap of health behaviors and create a user friendly measure of overall health. The popularity of programs such as the County Health Rankings (Remington, Catlin, & Gennuso, 2015) indicate the value of ranking such composite measures across geographies to motivate public health initiatives (Peppard, Kindig, Dranger, Jovaag, & Remington, 2008; Rohan, Booske, & Remington, 2009). However, performance rankings are usually presented without error measures, obscuring their underlying statistical uncertainty (e.g., error introduced through survey procedures; Groves, 2006; Schmidt & Hunter, 1996) (Arndt, Acion, Caspers, & Blood, 2013; Remington, 2015; Zhang et al., 2014). Researchers have debated whether performance rankings are “good enough” to be presented without error (Arndt, 2015; Gerzoff & Williamson, 2001; Remington, 2015), especially given how digestible they are to the public and policymakers, but few studies have empirically assessed the implications of statistical error for geographic performance rankings (Arndt, Acion, Caspers, & Diallo, 2011; Barker et al., 2005; Zhang et al., 2014).

In this study, we used data from nationally representative sources to (1) develop a composite state-level measure of adolescent cancer prevention behaviors and (2) rank and statistically compare states’ performance on this measure. Behaviors included in the current study are some of the most important cancer prevention behaviors that occur or emerge during adolescence: vaccination, energy balance, and substance use. For vaccination, we analyzed hepatitis B (HepB) vaccination, which prevents liver cancer (Centers for Disease Control and Prevention [CDC], 2011), and human papillomavirus (HPV) vaccination, which prevents cervical and other cancers (President’s Cancer Panel, 2014). For energy balance, we analyzed adequate physical activity, F&V consumption, and healthy weight, which are protective against several types of cancer, including colon, breast, and endometrial (Doubeni et al., 2012; Fuemmeler et al., 2009; National Cancer Institute, 2015; Song & Giovannucci, 2016; Willer, 2003). For substance use, we analyzed recent use of alcohol (i.e., binge drinking), cigarettes, and marijuana, which have been linked to lung, liver, and colorectal cancers (Hashibe et al., 2005; National Cancer Institute, 2015; Pelucchi, Tramacere, Boffetta, Negri, & Vecchia, 2011; Song & Giovannucci, 2016; U.S. Department of Health and Human Services, 2014). The findings from this analysis can inform policymakers and researchers about potential research studies and interventions to promote cancer prevention behaviors among adolescents (and where to deliver them) as well as the importance of error measures for geographic performance rankings.

Materials and Methods

Data Sources and Measures

State-level estimates of the prevalence of cancer-related vaccination, energy balance, and substance use behaviors came from the National Immunization Survey (NIS)-Teen, the Youth Risk Behavior Surveillance Survey (YRBSS), and the National Survey on Drug Use and Health (NSDUH). Survey items appear in Supplemental Table S1 (available in the online version of this article). Weighted prevalence estimates and standard errors for each measure were generated using survey procedures in SAS version 9.3 (Cary, NC), incorporating each survey’s sampling weights and accounting for a complex survey design (CDC, 2016a, 2016b; Abuse and Mental Health Services Administration, 2017). Since these estimates are derived from publicly available data sources and have been published before, we present the raw state-level prevalence estimates in Supplemental Table S2 (available in the online version of this article).

Description of National Surveys on Adolescent Cancer Prevention Behaviors

NIS-Teen is an annual, population-based telephone survey of parents of adolescents (age 13-17 years) managed by the CDC (2016a). NIS-Teen staff solicit parental report of vaccination, which they then verify with health care provider records. Annually, NIS-Teen collects vaccination data for approximately 20,000 adolescents across 50 states and the District of Columbia (hereafter referred to collectively as “states”). Additional details on NIS-Teen procedures are available (CDC, 2016a).

YRBSS is an in-school survey of students (grades 9 to 12) designed to monitor health behaviors among adolescents (CDC, 2016b). CDC coordinates the survey, and state health or education departments administer it biennially. Not all states contributed YRBSS data to the current study: Some states did not participate in YRBSS at all or during select years, some did not release their data due to low response rates, and some used survey instruments that excluded relevant measures. Thus, only 43 states had complete data on all YRBSS measures in the current analysis. Three states (District of Columbia, Minnesota, Oregon, and Washington) had no data for any of the energy balance and substance use measures, one state (Hawaii) was missing data for F&V consumption and marijuana use, and three states (Delaware, Maine, and New York) were missing data only for F&V consumption. Additional details on YRBSS procedures are available (CDC, 2016b).

NSDUH is an in-home survey of people aged 12 years and older administered annually by the Substance Abuse and Mental Health Services Administration (2017) to estimate use of alcohol, cigarettes, and other drugs. Additional details on NSDUH procedures are available (Substance Abuse and Mental Health Services Administration, 2017).

Vaccination Measures: NIS-Teen

We gathered provider-verified receipt of HepB (3 doses) and HPV (1+ dose; females and males separately) vaccines for 103,729 adolescents in NIS-Teen 2011-2015, estimating 5-year vaccination coverage for each state (Waldrop, Moss, Liu, & Zhu, 2017).

Energy Balance Measures: YRBSS

Energy balance measures included 531,777 adolescents (Grades 9-12) and indicated whether adolescents met national guidelines for each measure (60+ minutes of physical activity per day (U.S. Department of Health and Human Services, 2016), at least 2 servings of fruit and 3 servings of vegetables per day (U.S. Department of Health and Human Services, 2015), and weight below the 85th percentile of age- and sex-specific body mass index (CDC, 2015); we estimated the prevalence of each outcome in each state across the 3 survey years (2011, 2013, and 2015) (Moss, Liu, & Zhu, 2017).

Substance Use Measures: NSDUH and YRBSS

From NSDUH, we gathered self-reported abstention from binge drinking, cigarette smoking, and marijuana use in the past month among 69,200 adolescents aged 14 to 17 years in NSDUH 2011-2015, estimating the 5-year prevalence of each behavior for each state. (In 2015, NSDUH changed the definition of binge drinking among females (Substance Abuse and Mental Health Services Administration, 2017); thus, we only included data for binge drinking from NSDUH 2011-2014 in the current analysis.)

From YRBSS, we gathered self-reported abstention from binge drinking, cigarette smoking, and marijuana use in the past month from 450,050 adolescents aged 14 to 17 years and Grades 9 to 12 in YRBSS 2011, 2013, and 2015, estimating the prevalence of each behavior for each state.

Studies monitoring substance use among adolescents use a variety of items and time periods (Bauman & Phongsavan, 1999; Mulye et al., 2009; Weinberg, Rahdert, Colliver, & Glantz, 1998); however, we focused on the past-month abstention variables to make our findings consistent with reports from YRBSS and NSDUH.

Statistical Analysis

Missing Data

For states that were missing YRBSS data on energy balance or substance use measures (k = 4-8), we predicted prevalence estimates using state-level hierarchical Bayesian linear mixed models, an approach commonly used in small area estimation (Liu & Gilary, 2014; Rao & Isabel, 2015). First, we gathered 28 potential state-level covariates relevant to adolescent health (e.g., percent of children enrolled in public school with a disability and without health insurance) from the American Community Survey 5-year average (2011-2015) and Census 2010 (U.S. Census Bureau, 2018). Binge drinking and cigarette smoking prevalence estimates from NSDUH were also included in the pool of covariates to predict these behaviors in YRBSS, and binge drinking, cigarette smoking, and marijuana use prevalence estimates from NSDUH were included in the pool of covariates to predict marijuana use in YRBSS. Then, we used backward selection of covariates to identify the set of predictors that was significantly associated with observed estimates and used Bayesian modeling to predict prevalence estimates for states with no data utilizing information from other states with similar covariates. Models were validated by comparing the observed and model-predicted prevalence estimates for the states with observed data in YRBSS. For states that had YRBSS data on these measures, we retained the observed prevalence estimates.

Adolescent Cancer Prevention Behavior Composite Index

To create a composite index summarizing the vaccination (n = 3), energy balance (n = 3), and substance use (n = 6) measures, we standardized each measure to a mean of 0 and standard deviation of 1, and then we created a weighted average of the standardized scores. Each of the vaccination and energy balance measures was assigned an equal weight of 1/9. Because the estimates of substance use varied systematically between YRBSS and NSDUH (perhaps due to differences in survey administration) (U.S. Department of Health and Human Services, 2012a), and because the true prevalence of each behavior likely falls between the estimates from the two surveys, we assigned each of the substance use measures a weight of 1/18 (creating averages of each substance use indicator across surveys; separate estimates from YRBSS and NSDUH appear in Supplemental Table S3, available in the online version of this article). Finally, we summed the weighted score of each measure for an overall composite index score. We generated the rank by ordering states’ composite index scores from the lowest (rank = 51, i.e., poorest performance on adolescent cancer prevention behaviors) to the highest (rank = 1, i.e., best performance on adolescent cancer prevention behaviors).

To generate 95% confidence intervals (CIs) for the performance rankings on the composite index for each state, we conducted a simulation analysis using a Monte Carlo method with 100,000 simulations (Zhang et al., 2014). For each simulation, we estimated the “true” prevalence of each behavioral measure given the observed (or predicted) estimate and its standard error, assuming a truncated normal distribution, recalculating the composite index score for each simulation as described above (i.e., creating a weighted average of the standardized scores). A truncated normal distribution (bounded between 0 and 100) was assumed to prevent a simulated estimate outside the possible range—that is, simulated estimates that indicated less than 0% or greater than 100% of adolescents engaged in a behavior were not allowed. We calculated the ranks for states’ composite index scores separately for each simulation, and we used each state’s median rank from across the simulations as its “simulated rank.” The 95% CIs for the ranks were derived from the distribution of ranks across the simulations; to adjust for multiple testing, the simultaneous CIs are presented (Zhang et al., 2014), though individual CIs are available.

Analysis

We calculated the Pearson’s correlation coefficients for each pair of prevalence estimates, and we calculated the Spearman’s rank correlation coefficients for each pair of raw ranks of individual behavioral measures. We generated a choropleth of the adolescent cancer prevention behavior composite index scores to illustrate regional differences in performance. In addition, we compared the means of the overall United States versus those states whose scores on the cancer prevention index indicated a particularly high (in the top five ranked states) or low (in the bottom five ranked states) performance. We calculated absolute and relative differences between the mean prevalence of each behavior for the United States versus each category of states. Finally, we generated Spearman’s correlation coefficients for ranks on the cancer prevention index with ranks for the component behaviors to identify the behaviors most closely linked with performance on the index.

As a preliminary investigation of the face validity of the constructed index, we calculated (1) Pearson’s correlation coefficient for each state’s composite index score with age-adjusted 2010-2014 incidence rate (per 100,000 people) for all cancers and (2) Spearman’s correlation coefficient for each state’s simulated rank with rank of 2010-2014 incidence rate for all cancers; we repeated these analyses to calculate the correlations for the composite index score with mortality rate and for the simulated rank with mortality rank. Cancer incidence and mortality rates came from the State Cancer Profiles website (www.statecancerprofiles.cancer.gov).

As a sensitivity analysis, we repeated these procedures to develop a cancer prevention index score and rank using the indicators described above, except for HepB vaccination (which is typically administered before adolescence) and abstention from marijuana use (whose link to cancer risk is still under study). To measure the similarity of the results for the main analysis compared with the sensitivity analysis, we generated a Spearman’s correlation coefficient for states’ ranks using these two methods.

Analyses used a two-sided p value of .05 and were conducted in SAS version 9.3.

Results

Across states, 84.3% to 97.1% of adolescents had received HepB vaccination, and 41.8% to 78.0% of girls and 19.0% to 59.3% of boys had initiated HPV vaccination (Supplemental Table S2). For energy balance, 20.2% to 34.6% of adolescents met guidelines for physical activity, 4.1% to 15.8% for F&V consumption, and 66.4% to 82.0% for healthy weight status. For substance use, 82.5% to 93.5% of adolescents abstained from binge drinking in the past month, 84.3% to 95.4% from cigarette smoking, and 62.9% to 92.8% from marijuana use.

Many of the states’ prevalence estimates were correlated (Table 1, below diagonal). For example, abstention from cigarette smoking was positively correlated with girls’ HPV vaccination (Pearson’s r = 0.29), boys’ HPV vaccination (r = 0.33), healthy weight status (r = 0.41), and abstention from binge drinking (r = 0.47) but negatively correlated with physical activity (r = −0.41) (all p < .05). Similarly, many of the raw states’ ranks for each behavior were correlated (Table 1, above diagonal).

Table 1.

Correlation Matrix Among States’ Prevalence Estimates and Raw Ranks of Adolescents Engaging in Cancer Prevention Behaviors.

Cancer prevention behavior	1	2	3	4	5	6	7	8	9
1. HepB vaccination	—	0.19	0.27	−0.22	−0.16	−0.11	0.09	0.00	−0.09
2. HPV vaccination (girls)	0.24	—	0.81***	−0.20	0.16	0.24	−0.01	0.30*	−0.50***
3. HPV vaccination (boys)	0.26	0.83***	—	−0.13	0.23	0.17	0.01	0.34*	−0.55***
4. Daily physical activity	−0.15	−0.18	−0.05	—	−0.18	−0.14	−0.26	−0.39**	0.18
5. High F&V consumption	−0.14	0.13	0.15	−0.18	—	0.35*	0.01	0.22	−0.23
6. Healthy weight status	−0.13	0.19	0.11	−0.15	0.32*	—	−0.31*	0.41**	−0.28
7. Abstention from recent binge drinking	0.02	−0.06	0.00	−0.31*	0.04	−0.12	—	0.38**	0.11
8. Abstention from cigarette smoking	−0.02	0.29*	0.33*	−0.41**	0.14	0.41**	0.47***	—	−0.26
9. Abstention from marijuana use	−0.05	−0.44**	−0.59***	−0.06	0.18	−0.01	0.12	−0.06	—

Note. Estimates below the diagonal indicate Pearson’s correlation coefficients for states’ prevalence estimates. Estimates above the diagonal indicate Spearman’s ranks correlation coefficients for states’ raw ranks. Estimates of HepB and HPV vaccine coverage came from the National Immunization Survey-Teen, 2011-2015; estimates of daily physical activity, high F&V consumption, and healthy weight status came from the Youth Risk Behavior Surveillance Survey, 2011-2015; and estimates of abstention from past-month binge drinking, cigarette smoking, and marijuana use came from the Youth Risk Behavior Surveillance Survey and the National Survey on Drug Use and Health, 2011-2015 (averaged across surveys). Missing data were imputed using state-level covariates relevant to adolescent health. HepB = hepatitis B vaccine coverage; HPV = human papillomavirus vaccine coverage; F&V = fruits and vegetables.

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

Composite index scores ranged from −4.07 in Kentucky (worst performance) to 5.31 in Rhode Island (best performance) (Table 2; Figure 1). Thus, the simulated ranks were from 1 (95% CI = [1, 2]) in Rhode Island to 51 (95% CI = [46, 51]) in Kentucky. Other highly ranked states included Colorado (2, 95% CI = [1, 3]), Hawaii (4, 95% CI = [2, 19]), New Hampshire (4, 95% CI = [3, 8]), and Vermont (5, 95% CI = [3, 8]). Compared with the entire United States, highly ranked states tended to have higher HPV vaccination among girls (mean for the United States 57.5%; mean for the highly ranked states, 66.3%; relative difference, 15.4%), higher HPV vaccination among boys (mean for the United States, 30.7%; mean for the highly ranked states, 42.7%; relative difference: 39.1%), and higher F&V consumption (mean for the United States, 8.6%; mean for the highly ranked states, 12.2%; relative difference, 41.8%) (Table 3). Other poorly ranked states included Missouri (47, 95% CI = [39, 51]), Arkansas (48, 95% CI = [42, 51]), Mississippi (48, 95% CI = [42, 51]), and South Carolina (48, 95% CI = [42, 51]). Compared with the entire United States, these states tended to have lower HPV vaccination among girls (relative difference, −15.3%), lower HPV vaccination among boys (relative difference, −29.2%), and lower F&V consumption (relative difference, −12.2%) (Table 3). Ranks on the cancer prevention index correlated most strongly with states’ ranks for HPV vaccination among boys (r = 0.64), HPV vaccination among girls (r = 0.59), and smoking (r = 0.55). Substantial overlap in the 95% CIs around ranks emerged (Figure 2), indicating relatively low precision and precluding detection of statistical differences among ranks, especially for states ranked in the middle of the distribution.

Table 2.

Scores, Simulated Ranks, and 95% Confidence Intervals for a State-Level Adolescent Cancer Prevention Behavior Composite Index (Higher Scores and Lower Ranks Indicate Greater Cancer Prevention).

State	Composite index score	Simulated rank	95% CI
Alabama	−2.22	45	[39, 50]
Alaska	−0.49	38	[31, 44]
Arizona	−0.18	30	[18, 38]
Arkansas	−3.37	48	[42, 51]
California	3.06	10	[4, 22]
Colorado	3.23	2	[1, 3]
Connecticut	2.59	6	[3, 9]
Delaware	1.32	12	[6, 23]
District of Columbia	−1.43	17	[2, 51]
Florida	2.04	16	[10, 24]
Georgia	0.62	22	[13, 32]
Hawaii	4.26	4	[2, 19]
Idaho	−0.14	33	[22, 39]
Illinois	0.58	25	[17, 33]
Indiana	−2.33	46	[40, 50]
Iowa	0.71	15	[7, 28]
Kansas	−1.84	45	[39, 50]
Kentucky	−4.07	51	[46, 51]
Louisiana	−1.47	33	[20, 41]
Maine	1.25	16	[6, 34]
Maryland	−0.13	32	[22, 38]
Massachusetts	0.50	11	[7, 18]
Michigan	0.42	24	[16, 33]
Minnesota	1.63	22	[8, 37]
Mississippi	−2.75	48	[42, 51]
Missouri	−3.25	47	[39, 51]
Montana	−1.95	27	[18, 35]
Nebraska	3.27	9	[5, 15]
Nevada	0.98	24	[14, 35]
New Hampshire	1.34	4	[3, 8]
New Jersey	0.42	12	[6, 22]
New Mexico	0.45	13	[7, 20]
New York	0.85	18	[10, 31]
North Carolina	0.53	26	[17, 34]
North Dakota	0.16	23	[14, 33]
Ohio	−2.09	39	[33, 47]
Oklahoma	1.89	10	[5, 18]
Oregon	−1.14	22	[9, 38]
Pennsylvania	0.32	27	[16, 36]
Rhode Island	5.31	1	[1, 2]
South Carolina	−2.57	48	[42, 51]
South Dakota	−0.97	32	[20, 39]
Tennessee	−1.11	40	[35, 46]
Texas	−1.63	41	[36, 47]
Utah	2.55	41	[34, 48]
Vermont	1.01	5	[3, 8]
Virginia	0.52	37	[30, 43]
Washington	−1.47	31	[14, 43]
West Virginia	−3.16	41	[34, 48]
Wisconsin	−0.76	31	[19, 38]
Wyoming	−1.26	20	[13, 30]

Note. Estimates of hepatitis B (HepB) and human papillomavirus (HPV) vaccine coverage came from the National Immunization Survey-Teen, 2011-2015; estimates of daily physical activity, high F&V consumption, and healthy weight status came from the Youth Risk Behavior Surveillance Survey, 2011-2015; and estimates of abstention from past-month binge drinking, cigarette smoking, and marijuana use came from the Youth Risk Behavior Surveillance Survey and the National Survey on Drug Use and Health, 2011-2015 (averaged across surveys). Simulated ranks and 95% confidence intervals (CIs) were derived from a Monte Carlo procedure with 100,000 simulations. Missing data were imputed using state-level covariates relevant to adolescent health.

Table 3.

Comparing Prevalence of Adolescent Cancer Prevention Behaviors for All of the United States (U.S.) Versus Highly and Poorly Ranked States.

Category	HepB^a	HPV (Girls)^a	HPV (Boys)^a	Daily Physical Activity^b	High F&V Consumption^b	Healthy Weight Status^b	Abstention From Binge Drinking^c	Abstention From Cigarette Smoking^c	Abstention From Marijuana Use^c
All United States
Mean	92.2	57.5	30.7	25.6	8.6	72.2	86.8	89.1	85.2
Range	(84.3-97.1)	(41.8-78.0)	(19.0-59.3)	(20.2-34.6)	(4.1-15.8)	(66.4-82.0)	(82.5-93.5)	(84.3-95.4)	(62.9-92.8)
Highly ranked states^d
Mean	94.2	66.3	42.7	24.1	12.2	75.3	86.2	89.7	82.9
Range	(92.3-96.6)	(58.7-78.0)	(36.7-59.3)	(21.1-29.2)	(9.3-15.8)	(72.3-82.0)	(84.6-88.3)	(87.9-92.7)	(79.0-92.7)
Comparing mean prevalence of cancer prevention behaviors for high-ranking states versus all United States
Absolute difference	2.1	8.8	12.0	−1.5	3.6	3.1	−0.6	0.6	−2.3
Relative difference	2.2	15.4	39.1	−6.0	41.8	4.3	−0.7	0.7	−2.7
Poorly ranked states^e
Mean	92.0	48.7	21.7	25.0	7.6	68.1	86.8	86.8	87.1
Range	(88.0-94.7)	(44.5-50.9)	(19.2-23.7)	(21.5-28.1)	(6.8-9.0)	(66.4-71.6)	(84.9-88.3)	(84.7-88.2)	(85.9-87.8)
Comparing mean prevalence of cancer prevention behaviors for low-ranking states versus all United States
Absolute difference	−0.2	−8.8	−9.0	−0.6	−1.1	−4.0	0.0	−2.3	1.9
Relative difference	−0.2	−15.3	−29.2	−2.4	−12.2	−5.6	0.0	−2.6	2.3

Note. Prevalence estimates indicate the percentage of adolescents engaging in a given behavior in each state. Estimates of HepB and HPV vaccine coverage came from the National Immunization Survey-Teen, 2011-2015; estimates of daily physical activity, high F&V consumption, and healthy weight status came from the Youth Risk Behavior Surveillance Survey, 2011-2015; and estimates of abstention from past-month binge drinking, cigarette smoking, and marijuana use came from the Youth Risk Behavior Surveillance Survey and the National Survey on Drug Use and Health, 2011-2015 (averaged across surveys). HepB = hepatitis B vaccine coverage; HPV = human papillomavirus vaccine coverage; F&V = fruits and vegetables.

Prevalence estimates for HepB and HPV vaccine coverage are for adolescents aged 13 to 17 years. ^bPrevalence estimates for energy balance measures are for adolescents in Grades 9 to 12. ^cPrevalence estimates for abstention from substance use are for adolescents in Grades 9 to 12 and aged 14 to 17 years. ^dHighly ranked states were Colorado, Hawaii, New Hampshire, Rhode Island, and Vermont. ^ePoorly ranked states were Arkansas, Kentucky, Mississippi, Missouri, and South Carolina.

Low correlations were observed for state-level composite index scores with cancer incidence rates (r = −0.18) and for simulated ranks with ranks of cancer incidence (r = −0.07). However, moderate correlations were observed for state-level composite index scores with cancer mortality rates (r = −0.61) and for simulated ranks with ranks of cancer mortality (r = −0.48). These correlations indicate that as scores and ranks for adolescent cancer prevention increased, cross-sectional cancer mortality rates and ranks (but not cancer incidence rates and ranks) decreased.

In the sensitivity analysis excluding HepB vaccination and abstention from marijuana use from the index, we found generally similar state ranks (Supplemental Table S4, available in the online version of this article). The Spearman’s correlation coefficient for the state ranks across these two methods was 0.82. Descriptively, the states with the greatest deviation in rank from the main versus sensitivity analyses were those with imputed values for the energy balance or substance use measures, perhaps due to the greater variability (i.e., standard errors) in those estimates as a result of the imputation process.

Discussion

In an analysis synthesizing data from several population-based surveys on adolescent cancer prevention behaviors, we found high prevalence of HepB vaccination, healthy weight status, and abstention from recent binge drinking, cigarette smoking, and marijuana use, with lower prevalence of HPV vaccination, physical activity, and F&V consumption. States’ prevalences and ranks demonstrated generally weak to moderate correlations across behaviors (except for HPV vaccination among girls and boys, which were strongly correlated). A composite index indicated substantial variability in performance across states, with Rhode Island ranked first (score = 5.31; rank = 1, 95% CI = [1, 2]) and Kentucky last (score = −4.07; rank = 5, 95% CI = [46, 51]). HPV vaccination, F&V consumption, and smoking appeared to be influential in determining whether states were ranked highly or poorly. Cross-sectionally, states’ scores and ranks were closely linked to cancer mortality levels, indicating preliminary face validity of the adolescent cancer prevention behavior composite index. (Interestingly, states’ scores and ranks were not associated with cancer incidence; this pattern could have emerged due to differences in access to and quality of care, e.g., through health care policies that affect clinical issues around prevention as well as mortality-reducing cancer treatment, which may not have as much influence on incidence.) These findings have important implications for research as well as public health programming.

In terms of research implications, these findings highlight the importance of measures of uncertainty accompanying performance rankings (Arndt, 2015; Arndt et al., 2011; Arndt et al., 2013; Zhang et al., 2014). Statistical rankings are rarely discrete, error-free indicators, and 95% CIs can communicate uncertainty about rankings to lay audiences, policymakers, and other researchers. At least two sources of uncertainty drove the 95% CIs around states’ ranks for adolescent cancer prevention. The first source of uncertainty was related to statistical and survey considerations, for example, the District of Columbia had a very wide 95% CI [2, 51] partly due to (1) its small population (and, therefore, sample) of adolescents and (2) imputed estimates (with relatively wide standard errors) for some energy balance and substance use measures from YRBSS (CDC, 2016b). The second source of uncertainty was a true variation in performance, for example, despite its small population, Rhode Island had a narrow 95% CI [1, 2] because (1) the 95% CI could not extend beyond the minimum bound of 1 and (2) its composite index score (5.31) was more than two points higher than the next highest ranked state, Colorado (3.23). In contrast, states in the middle of the index score distribution were not close to the bounds of 1 to 51, and they had scores clustered more closely together; for example, Illinois was ranked 25th (95% CI = [17, 33]), and its composite index score was within a half point of 13 other states’ scores.

While debate persists as to whether performance ranks can be presented without measures of uncertainty (Arndt, 2015; Remington, 2015), our study adds to a body of literature (Barker et al., 2005; Zhang et al., 2014) illustrating that error-free rankings obscure marked variability in health indicators, perhaps leading to overinterpretation of results. This issue could be especially salient for rankings of smaller geographic units (e.g., counties) and for summary measures (e.g., indices such as ours and others’ (Remington et al., 2015). Efforts to improve the precision of questionnaire items (through survey developmental procedures such as cognitive interviewing [Willis, 2004] and methodological procedures such as large sample sizes [Groves, 2006; McColl et al., 2001]) can help reduce uncertainty and shrink CIs both for prevalence estimates and rankings. Alternatively, the substantial overlap of CIs for these rankings could reflect genuine equivalence among states, suggesting that effort should be focused on improving adolescent cancer prevention primarily in the lowest ranked states to improve their performance up to the level of the rest of the country.

Additionally, these findings emphasize the importance of additional research on the development of behavioral indices. We originally put forth an index based on nine adolescent cancer prevention behaviors, but the sensitivity analysis indicated that a more parsimonious index (excluding HepB vaccination and marijuana use) produced very similar results to the main index (Spearman’s r = 0.82 for the two rankings). An index summarizing behavioral cancer prevention should be stable enough to facilitate comparison across years, but it should also be flexible and responsive to changes in (1) the scientific consensus about the strength of the pathogenic link to cancer and (2) public health relevance of indicators. For example, compared with HPV vaccination or tobacco use, the link between marijuana use and cancer is less well understood; resources devoted to reducing marijuana use for the purpose of cancer prevention might therefore be better spent elsewhere. In addition, HepB vaccination is relatively high nationally (all states >80% uptake), resulting in relatively little variation across states; therefore, the additional benefit of increasing HepB vaccination for cancer prevention is likely less than the benefits we might anticipate from public health efforts focused on other behaviors. Future research efforts should focus on identifying which behaviors are most useful and actionable for cancer prevention efforts, with the goal of developing a parsimonious yet stable index. A related research concern is around measurement of cancer risk behaviors among adolescents: As the scientific understanding of the cancer risk of lifelong patterns of behavior (e.g., alcohol use patterns; marijuana use) develops, items in national surveys of adolescent health behaviors should be developed to facilitate surveillance of these exposures.

In terms of programming implications, state or local public health programs can undertake efforts to improve these adolescent cancer prevention behaviors (National Cancer Institute, 2015). Although many adolescents move out of their home state when they become adults (Mulder & Clark, 2000) and the geographic context of cancer prevention may change over the lifetime, establishing healthy behaviors during this sensitive developmental period is important for lifelong trajectories of risk (Chen & Jacobson, 2012; Fuemmeler et al., 2009; Wright et al., 2001). Descriptively, the states that performed most poorly on the index were primarily located in the Southeast/along the Mississippi River (Quintile 1 in Figure 1), but states that performed highly were scattered geographically (Quintile 5 in Figure 1). Regional differences in the index could be attributable to factors such as demographics, policies, or programs. Poor-performing states could adopt policies or programs that are already implemented in high-performing states to improve adolescent cancer prevention. Identifying the relevant factors underlying these behaviors was beyond the scope of the current analysis, although previous studies have examined this question (Botello-Harbaum et al., 2009; Brener, Wheeler, Wolfe, Vernon-Smiley, & Caldart-Olson, 2007; Dwyer-Lindgren et al., 2017; Moss, Reiter, Truong, Rimer, & Brewer, 2016; Pronk et al., 2004; Reyna & Farley, 2006). Public health departments could prioritize intervening on certain behaviors based on either strength of the link to cancer (Danaei et al., 2005; Song & Giovannucci, 2016) or their state’s relative performance on that outcome (Supplemental Table S2). State-level performance for HPV vaccination (among boys and girls) and smoking were most closely correlated with ranks on the index; these behaviors may therefore be priority targets due to both their influence on the index as well as their importance for public health. Alternatively, departments could develop multitarget programs to affect multiple health behaviors simultaneously (Noar, Chabot, & Zimmerman, 2008; Prochaska, Spring, & Nigg, 2008), for example, a health education program promoting “healthy lifestyles” (Emmons, Linnan, Shadel, Marcus, & Abrams, 1999; de Vries, Kremers, Smeets, Brug, & Eijmael, 2008) or school policies to affect students’ behaviors (Brener et al., 2007). Understanding why some states perform better on adolescent cancer prevention than others requires additional research, including epidemiological and intervention studies.

Study strengths include the use of several high-quality survey data sets (CDC, 2016a, 2016b; Substance Abuse and Mental Health Services Administration, 2017) to measure adolescent cancer prevention behaviors with diverse and well-validated measures. For example, NIS-Teen calculates vaccination status based on adolescents’ medical records (CDC, 2016a), while the prevalence estimates for substance use were derived from self-report measures from questionnaires delivered in schools (YRBSS [CDC, 2016b]) and in homes (NSDUH [Substance Abuse and Mental Health Services Administration, 2017]). These surveys drew on large, nationally representative samples (e.g., more than 500,000 adolescents participated in YRBSS between 2011 and 2015 [CDC, 2016b]), maximizing the generalizability of our findings. In addition, we leveraged rigorous and flexible statistical methods to conduct this analysis. For example, the Monte Carlo procedures with 100,000 simulations to calculate 95% CIs around ranks are well studied (Gerzoff & Williamson, 2001; Zhang et al., 2014) and adaptable for other research questions (National Cancer Institute, 2017).

Study limitations include the use of self-reported measures of adolescent cancer prevention behaviors in YRBSS (CDC, 2016b) and NSDUH (Substance Abuse and Mental Health Services Administration, 2017), which could be biased (Brener, Billy, & Grady, 2003; Groves, 2006). This issue could be especially pertinent for substance use behaviors, which are subject to concerns about social desirability and legality (Brener et al., 2003; Williams & Nowatzki, 2005). However, previous analyses of the psychometric properties of adolescent substance use in these surveys have indicated adequate reliability and validity (Needle, McCubbin, Lorence, & Hochhauser, 1983; U.S. Department of Health and Human Services, 2012a). In addition, we used survey data from 2011-2015 to aggregate generous sample sizes; analyses of single-year survey data would necessarily use smaller samples, resulting in greater uncertainty and wider 95% CIs. Some states were missing data for certain behavioral indicators in YRBSS (CDC, 2016b); however, we attenuated this limitation by predicting missing prevalence estimates. Finally, we limited our analysis to nine cancer prevention behaviors due to data availability, although other indicators could have been included (National Cancer Institute, 2015), including sun safety (Islami et al., 2018). Examining different or additional cancer prevention behaviors, or using different weighting schemes for the index, would produce different composite index scores and rankings.

In conclusion, we combined several indicators of adolescent cancer prevention from high-quality, national surveys to create composite index scores and ranks for states in the United States, finding that Rhode Island had the best overall performance (score = 5.31; rank = 1, 95% CI = [1, 2] and Kentucky had the worst (score = −4.07; rank = 51, 95% CI = [46, 51]). Additional research can use these composite index scores and ranks in epidemiological analyses to explain state-level variation in health outcomes. However, outputs of this line of research should include measures of uncertainty alongside performance rankings of health indicators across geographic areas to avoid overinterpretation, particularly for substate units and summary measures.

Supplemental Material

HEB839111_Supplemental_Material – Supplemental material for Adolescent Behavioral Cancer Prevention in the United States: Creating a Composite Variable and Ranking States’ Performance

Supplemental material, HEB839111_Supplemental_Material for Adolescent Behavioral Cancer Prevention in the United States: Creating a Composite Variable and Ranking States’ Performance by Jennifer L. Moss, Benmei Liu and Li Zhu in Health Education & Behavior

Footnotes

Acknowledgements

The authors wish to acknowledge the assistance of the Substance Abuse and Mental Health Services Administration that conducted the National Survey on Drug Use and Health (NSDUH) surveys and shared their data. In addition, the authors wish to acknowledge the assistance of the Centers for Disease Control and Prevention (CDC) and the state health agencies that conducted the Youth Risk Behavior Surveillance Survey (YRBSS) surveys and shared their data. This article was prepared or accomplished by the authors in their personal capacity.

Authors’ Note

The authors completed this work as part of active duty at the National Cancer Institute. The opinions expressed in this article are the authors’ own and do not reflect the view of the National Institutes of Health, the Department of Health and Human Services, or the U.S. government.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental Tables S1-S4 are available in the online version of this article at .

ORCID iD

Jennifer L. Moss

References

Arndt

(2015). Just how useful are health rankings? Health services research, 50, 1403-1406. doi:10.1111/1475-6773.12350

Arndt

Acion

Caspers

Blood

(2013). How reliable are county and regional health rankings? Prevention Science, 14, 497-502. doi:10.1007/s11121-012-0320-3

Arndt

Acion

Caspers

Diallo

(2011). Assessing community variation and randomness in public health indicators. Population Health Metrics, 9, 3. doi:10.1186/1478-7954-9-3

Barker

L. E.

Smith

P. J.

Gerzoff

R. B.

Luman

E. T.

McCauley

M. M.

Strine

T. W.

(2005). Ranking states’ immunization coverage: An example from the National Immunization Survey. Statistics in Medicine, 24, 605-613. doi:10.1002/sim.2039

Bauman

Phongsavan

(1999). Epidemiology of substance use in adolescence: Prevalence, trends and policy implications. Journal of Drug and Alcohol Dependence, 55, 187-207.

Berrigan

Dodd

Troiano

R. P.

Krebs-Smith

S. M.

Barbash

R. B.

(2003). Patterns of health behavior in U.S. adults. Preventive Medicine, 36, 615-623. doi:10.1016/S0091-7435(02)00067-1

Botello-Harbaum

M. T.

Haynie

D. L.

Iannotti

R. J.

Wang

Gase

Simons-Morton

(2009). Tobacco control policy and adolescent cigarette smoking status in the United States. Nicotine & Tobacco Research, 11, 875-885. doi:10.1093/ntr/ntp081

Brener

N. D.

Billy

J. O. G.

Grady

W. R.

(2003). Assessment of factors affecting the validity of self-reported health-risk behavior among adolescents: Evidence from the scientific literature. Journal of Adolescent Health, 33, 436-457.

Brener

N. D.

Wheeler

Wolfe

L. C.

Vernon-Smiley

Caldart-Olson

(2007). Health services: Results from the School Health Policies and Programs Study 2006. Journal of School Health, 77, 464-485. doi:10.1111/j.1746-1561.2007.00230.x

10.

Centers for Disease Control and Prevention. (2011). Epidemiology and prevention of vaccine-preventable diseases (13th ed.). Washington, DC: Public Health Foundation.

11.

Centers for Disease Control and Prevention. (2015). About child & teen BMI. Retrieved from https://www.cdc.gov/healthyweight/assessing/bmi/childrens_bmi/about_childrens_bmi.html

12.

Centers for Disease Control and Prevention. (2016a). National Immunization Survey: Datasets and related documentation for the National Immunization Survey–Teen. Retrieved form https://www.cdc.gov/nchs/nis/data_files_teen.htm

13.

Centers for Disease Control and Prevention. (2016b). Youth risk behavior surveillance system (YRBSS) overview. Retrieved from https://www.cdc.gov/healthyyouth/data/yrbs/overview.htm

14.

Chen

Jacobson

K. C.

(2012). Developmental trajectories of substance use from early adolescence to young adulthood: gender and racial/ethnic differences. Journal of Adolescent Health, 50, 154-163.

15.

Danaei

Vander Hoorn

Lopez

A. D.

Murray

C. J. L.

Ezzati

, & Comparative Risk Assessment Collaborating Group. (2005). Causes of cancer in the world: Comparative risk assessment of nine behavioural and environmental risk factors. Lancet, 366, 1784-1793.

16.

de Vries

Kremers

Smeets

Brug

Eijmael

. (2008). The effectiveness of tailored feedback and action plans in an intervention addressing multiple health behaviors. American Journal of Health Promotion, 22, 417-424.

17.

Doubeni

C. A.

Major

J. M.

Laiyemo

A. O.

Schootman

Zauber

A. G.

Hollenbeck

A. R.

. . . Allison

(2012). Contribution of behavioral risk factors and obesity to socioeconomic differences in colorectal cancer incidence. Journal of the National Cancer Institute, 104, 1353-1362. doi:10.1093/jnci/djs346

18.

Dwyer-Lindgren

Bertozzi-Villa

Stubbs

R. W.

Morozoff

Mackenbach

J. P.

van Lenthe

F. J.

. . . Murray

C. J. L.

(2017). Inequalities in life expectancy among US counties, 1980 to 2014: Temporal trends and key drivers. JAMA Internal Medicine, 177, 1003-1011. doi:10.1001/jamainternmed.2017.0918

19.

Emmons

K. M.

Linnan

L. A.

Shadel

W. G.

Marcus

Abrams

D. B.

(1999). The Working Healthy Project: A worksite health-promotion trial targeting physical activity, diet, and smoking. Journal of Occupational and Environmental Medicine, 41, 545-555.

20.

Fuemmeler

B. F.

Pendzich

M. K.

Tercyak

K. P.

(2009). Weight, dietary behavior, and physical activity in childhood and adolescence: Implications for adult cancer risk. Obesity Facts, 2, 179-186. doi:10.1159/000220605

21.

Gerzoff

R. B.

Williamson

G. D.

(2001). Who’s number one? The impact of variability on rankings based on public health indicators. Public Health Reports, 116, 158-164.

22.

Groves

R. M.

(2006). Nonresponse rates and nonresponse bias in household surveys. Public Opinion Quarterly, 70, 646-675.

23.

Hale

D. R.

Viner

R. M.

(2016). The correlates and course of multiple health risk behaviour in adolescence. BMC Public Health, 16, 458.

24.

Hashibe

Straif

Tashkin

D. P.

Morgenstern

Greenland

Zhang

Z.-F.

(2005). Epidemiologic review of marijuana use and cancer risk. Alcohol, 35, 265-275.

25.

Islami

Goding Sauer

Miller

K. D.

Siegel

R. L.

Fedewa

S. A.

Jacobs

E. J.

. . . Jemal

(2018). Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. Cancer: A Cancer Journal for Clinicians, 68, 31-54.

26.

Kabat

G. C.

Matthews

C. E.

Kamensky

Hollenbeck

A. R.

Rohan

T. E.

(2015). Adherence to cancer prevention guidelines and cancer incidence, cancer mortality, and total mortality: A prospective cohort study. American Journal of Clinical Nutrition, 101, 558-569. doi:10.3945/ajcn.114.094854

27.

Liu

Gilary

(2014). Small area estimation for the tobacco-use supplement to the Current Population Survey. In JSM proceedings, survey research methods section (pp. 2542-2552). Alexandria, VA: American Statistical Association.

28.

McColl

Jacoby

Thomas

Soutter

Bamford

Steen

. . . Bond

(2001). Design and use of questionnaires: A review of best practice applicable to surveys of health service staff and patients. Health Technology Assessment, 5, 1-256.

29.

Meader

King

Moe-Byrne

Wright

Graham

Petticrew

. . . Sowden

A. J.

(2016). A systematic review on the clustering and co-occurrence of multiple risk behaviours. BMC Public Health, 16, 657. doi:10.1186/s12889-016-3373-6

30.

Moss

J. L.

Liu

Zhu

(2017). Comparing percentages and ranks of adolescent weight-related outcomes among U.S. states: Implications for intervention development. Preventive Medicine, 105, 109-115. doi:10.1016/j.ypmed.2017.09.006

31.

Moss

J. L.

Reiter

P. L.

Truong

Y. K.

Rimer

B. K.

Brewer

N. T.

(2016). School entry requirements and coverage of nontargeted adolescent vaccines. Pediatrics, 138, e20161414.

32.

Muff

Dragano

Jöckel

K.-H.

Moebus

Möhlenkamp

Erbel

. . . Siegrist

(2010). Is the co-occurrence of smoking and poor consumption of fruits and vegetables confounded by socioeconomic conditions? International Journal of Public Health, 55, 339-346.

33.

Mulder

C. H.

Clark

W. A.

(2000). Leaving home and leaving the state: Evidence from the United States. Population, Space and Place, 6, 423-437.

34.

Mulye

T. P.

Park

M. J.

Nelson

C. D.

Adams

S. H.

Irwin

C. E.

Jr. Brindis

C. D.

(2009). Trends in adolescent and young adult health in the United States. Journal of Adolescent Health, 45, 8-24.

35.

National Cancer Institute. (2015). Risk factors for cancer. Retrieved from https://www.cancer.gov/about-cancer/causes-prevention/risk

36.

National Cancer Institute. (2017). Composite indices of cancer burden in CI*Rank. Retrieved from https://surveillance.cancer.gov/cirank/documentation/composite-indices.html

37.

Needle

McCubbin

Lorence

Hochhauser

(1983). Reliability and validity of adolescent self-reported drug use in a family-based study: A methodological report. International Journal of the Addictions, 18, 901-912.

38.

Noar

S. M.

Chabot

Zimmerman

R. S.

(2008). Applying health behavior theory to multiple behavior change: Considerations and approaches. Preventive Medicine, 46, 275-280. 10.1016/j.ypmed.2007.08.001

39.

Pelucchi

Tramacere

Boffetta

Negri

Vecchia

C. L.

(2011). Alcohol consumption and cancer risk. Nutrition and Cancer, 63, 983-990.

40.

Peppard

P. E.

Kindig

D. A.

Dranger

Jovaag

Remington

P. L.

(2008). Ranking community health status to stimulate discussion of local public health issues: The Wisconsin County Health Rankings. American Journal of Public Health, 98, 209-212. doi:10.2105/AJPH.2006.092981

41.

President’s Cancer Panel. (2014). Accelerating HPV vaccine uptake: Urgency for action to prevent cancer. Retrieved from https://deainfo.nci.nih.gov/advisory/pcp/annualreports/HPV/index.htm

42.

Prochaska

J. J.

Spring

Nigg

C. R.

(2008). Multiple health behavior change research: An introduction and overview. Preventive Medicine, 46, 181-188. doi:10.1016/j.ypmed.2008.02.001

43.

Pronk

N. P.

Anderson

L. H.

Crain

A. L.

Martinson

B. C.

O’Connor

P. J.

Sherwood

N. E.

Whitebird

R. R.

(2004). Meeting recommendations for multiple healthy lifestyle factors: Prevalence, clustering, and predictors among adolescent, adult, and senior health plan members. American Journal of Preventive Medicine, 27(2 Suppl.), 25-33. doi:10.1016/j.amepre.2004.04.022

44.

Rao

J. N.

Isabel

(2015). Small-area estimation. Hoboken, NJ: Wiley Online Library.

45.

Remington

P. L.

(2015). County Health Rankings and the Cult of the Imperfect. Health Services Research, 50, 1407-1412. doi:10.1111/1475-6773.12351

46.

Remington

P. L.

Catlin

B. B.

Gennuso

K. P.

(2015). The County Health Rankings: Rationale and methods. Population Health Metrics, 13, 11. doi:10.1186/s12963-015-0044-2

47.

Reyna

V. F.

Farley

(2006). Risk and rationality in adolescent decision making: Implications for theory, practice, and public policy. Psychological Science in the Public Interest, 7, 1-44. doi:10.1111/j.1529-1006.2006.00026.x

48.

Rohan

A. M.

Booske

B. C.

Remington

P. L.

(2009). Using the Wisconsin County Health Rankings to catalyze community health improvement. Journal of Public Health Management and Practice: JPHMP, 15, 24-32. doi:10.1097/PHH.0b013e3181903bf8

49.

Schmidt

F. L.

Hunter

J. E.

(1996). Measurement error in psychological research: Lessons from 26 research scenarios. Psychological Methods, 1, 199-223.

50.

Song

Giovannucci

(2016). Preventable incidence and mortality of carcinoma associated with lifestyle factors among White adults in the United States. JAMA Oncology,2 1154-1161. doi:10.1001/jamaoncol.2016.0843

51.

Substance Abuse and Mental Health Services Administration. (2017). National Survey on Drug Use and Health. Retrieved from https://nsduhweb.rti.org/respweb/homepage.cfm

52.

U.S. Census Bureau. (2018). American Community Survey (ACS). Retrieved from https://www.census.gov/programs-surveys/acs/

53.

U.S. Department of Health and Human Services (2012a). Comparing and evaluating youth substance use estimates from the National Survey on Drug Use and Health and other surveys. Retrieved from https://www.samhsa.gov/data/sites/default/files/NSDUH-M9-Youth-2012/NSDUH-M9-Youth-2012.pdf

54.

U.S. Department of Health and Human Services. (2012b). Preventing tobacco use among youth and young adults: A report of the surgeon general. Atlanta, GA: Author.

55.

U.S. Department of Health and Human Services. (2014). The health consequences of smoking: 50 years of progress: A report of the surgeon general. Atlanta, GA: Author.

56.

U.S. Department of Health and Human Services. (2015). Dietary guidelines for Americans, 2015-2020. Retrieved from https://health.gov/dietaryguidelines/2015/resources/2015-2020_dietary_guidelines.pdf

57.

U.S. Department of Health and Human Services. (2016). Physical activity guidelines for Americans: Active children and adolescents. Retrieved from https://health.gov/paguidelines/guidelines/children.aspx

58.

Waldrop

A. R.

Moss

J. L.

Liu

Zhu

(2017). Ranking states on coverage of cancer-preventing vaccines among adolescents: The influence of imprecision. Public Health Reports, 132, 627-636. doi:10.1177/0033354917727274

59.

Wang

L. Y.

Chyen

Lee

Lowry

(2008). The association between body mass index in adolescence and obesity in adulthood. Journal of Adolescent Health, 42, 512-518.

60.

Weinberg

N. Z.

Rahdert

Colliver

J. D.

Glantz

M. D.

(1998). Adolescent substance abuse: A review of the past 10 years. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 252-261.

61.

Willer

(2003). Reduction of the individual cancer risk by physical exercise. Onkologie, 26, 283-289. doi:10.1159/000071626

62.

Williams

R. J.

Nowatzki

(2005). Validity of adolescent self-report of substance use. Substance Use & Misuse, 40, 299-311.

63.

Willis

G. B.

(2004). Cognitive interviewing: A tool for improving questionnaire design. Thousand Oaks, CA: Sage.

64.

Wright

C. M.

Parker

Lamont

Craft

A. W.

(2001). Implications of childhood obesity for adult health: Findings from thousand families cohort study. BMJ, 323, 1280-1284.

65.

Kochanek

K. D.

Murphy

S. L.

Tejada-Vera

(2016). Deaths: Final data for 2014. National Vital Statistics Reports, 65, 1-122.

66.

Zhang

Luo

Zhu

Stinchcomb

D. G.

Campbell

Carter

. . . Feuer

E. J.

(2014). Confidence intervals for ranks of age-adjusted rates across states or counties. Statistics in Medicine, 33, 1853-1866. doi:10.1002/sim.6071

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.62 MB