Prospective longitudinal relations among frequent social media use, nicotine vaping and experiencing internalizing mental health problems.

AIMS: To compare four a priori rival mediated pathways of frequent social media use, electronic nicotine delivery systems (ENDS) use and internalizing mental health (MH) problems across five waves of nationally representative data. DESIGN, SETTING AND PARTICIPANTS: This was a longitudinal study using data drawn from waves 2-5 (October 2014-November 2019) of the Population Assessment of Tobacco and Health Study, a nationally representative cohort study spanning approximately 5 years, conducted in the United States. The analytical sample of participants included those who were aged 12-14 years at wave 2 and who provided data in subsequent waves until wave 4.5 (n = 4627, 69.7% were White and 51.4% were male). MEASUREMENTS: Frequent social media use (several times a day), ENDS use (past 30-day use) and internalizing MH problems (endorsed symptoms on four items in the past year) were dichotomized for analysis. FINDINGS: The weighted proportions of the three key variables increased over time. From wave 2 to wave 5, frequent social media use grew from 56.9 to 77.2%; internalizing MH problems from 18.9 to 29.0%; and ENDS use from 1.4 to 11.4%. In weighted logistic regressions using generalized linear mixed models with random effects, there was a significant within-person association between frequent social media use at time t and greater ENDS use at t + 1 [adjusted odds ratio (aOR) = 1.87; 95% confidence interval (CI) = 1.47, 2.37] and worsened internalizing MH problems at t + 1 (aOR = 1.19; 95% CI = 1.04, 1.37). A model-based causal mediation analysis and marginal structural models were fitted to estimate the average causal mediation effect. Among all four examined mediation pathways throughout the three constructs, partial mediation was observed, and all the pathways were significant for both boys and girls. Sex differences did not emerge in the examined prospective mediated pathways. CONCLUSIONS: Among youth in the United States, frequent social media use appears to mediate the prospective association between experiencing internalizing mental health problems and using electronic nicotine delivery systems.

Frequent Social Media Use and Its Prospective Association With Mental Health Problems in a Representative Panel Sample of US Adolescents.

PURPOSE: This study examined the relationship between frequent social media use and subsequent mental health in a representative sample of US adolescents. Also investigated were sex differences in multiyear growth trajectories of mental health problem internalization relative to social media use. METHODS: Four waves (2013-2018) of nationally representative, longitudinal Population Assessment of Tobacco and Health data were analyzed. A total of 5,114 US adolescents aged 12-14 years at baseline had repeated data across all waves. Statistical analysis involved testing a series of sequential-weighted single-group and multi-group latent growth curve models using R version 3.6.2. RESULTS: Of the 5,114 respondents, 2,491 were girls (48.7%). The percentage of frequent social media use was 26.4% at Wave 1 and 69.1% at Wave 4 for boys compared to 38.3% and 80.6% for girls (p < .001). Boys showed an improving (-0.218, p = .005) but girls showed a deteriorating linear trend (0.229, p = .028) for mental health at the full multigroup latent growth curve model. Social media use accounted for mental health conditions across Waves 1-3 for boys (ps<.01) but only at Wave 1 for girls (p = .035). With the addition of the social media use variable alone, model fit dramatically improved, and residual variances in growth patterns (i.e., random effect) became nonsignificant for boys. Substantial sex differences existed in baseline status, directionality, and shape of mental health growth trajectories as well as interplay of social media use with other factors. DISCUSSION: Findings of the study suggest that frequent social media use is associated with poorer subsequent mental health for adolescents.

Identifying emerging predictors for adolescent electronic nicotine delivery systems use: A machine learning analysis of the Population Assessment of Tobacco and Health Study.

Intervention strategies to prevent adolescents from using electronic nicotine delivery systems (ENDS) should be based on robust predictors of ENDS use that may differ from predictors of conventional cigarette use. Literature points to the need for uncovering emerging predictors of ENDS use. This study identified emerging predictors of adolescent ENDS use using machine learning (ML) techniques. We analyzed nationally representative multi-wave longitudinal survey data (2013-2018) drawn from the Population Assessment of Tobacco and Health Study. A sample of adolescents (12-17 years) who never used any tobacco products at baseline and completed Wave 2 (n = 7958), Wave 3 (n = 6260) and Wave 4 (n = 4544) were analyzed. We developed a supervised ML prediction model using the penalized logistic regression to assess self-reported past-month ENDS use (i.e., current use) at Waves 2-4 based on the variables measured at the previous wave. We then extracted important predictors from each model. The penalized logistic regression models showed suitable capability to discriminate between ENDS uses and non-uses at each wave based on the area under the receiver operating characteristic curve and the area under the precision-recall curve. Interestingly, social media use emerged as an important variable in predicting adolescent ENDS use. ML models appear to be a promising method to identify unique population-level predictors for U.S. adolescent ENDS use behaviors. More research is warranted to investigate emerging predictors of ENDS use and experimentally examine the mechanism by which these emerging predictors affect ENDS use behavior across different spectrum of populations.

Predicting opioid misuse at the population level is different from identifying opioid misuse in individual patients.

Tumin and Bhalla mentioned challenges associated with the use of population-based survey and machine learning (ML) results on adolescent opioid misuse to clinical settings. In a clinical setting, medical providers do know patient's identity. So, it is not surprising that drug misuse is rarely identified through patient's self-report especially if it involves illicit drug. Even though self-report is susceptible to bias, it is a valid and affordable tool to gather data on illicit drug use at the population level. Use of audio computer-assisted self-interviewing (ACASI) and computer-assisted personal interviewing (CAPI) in NSDUH provides the respondent with a highly private and confidential mode for responding to questions, which helps increase the level of honest reporting of illicit drug use and other sensitive behaviors. As acknowledged in the paper, opioid misuse should not be inferred at the individual level from our ML models. Such interpretations may lead to ecological fallacy. Predicting opioid misuse at the population level is different from identifying opioid misuse in individual patients. Nonetheless, we believe that coordinated multisectoral collaborations that leverage the expertise and resources of both public health and clinical sectors would offer a promising model for addressing the opioid crisis.

Using susceptibility measures to prospectively predict ever use of electronic cigarettes among adolescents.

Preventing adolescents from using e-cigarettes is crucial given that e-cigarette use can lead to conventional cigarette smoking. In order to inform prevention efforts, the present study examined the role of susceptibility measures as well as psychosocial, behavioral, and environmental factors in prospectively predicting ever use of electronic cigarettes among adolescents. We analyzed Wave 1 and Wave 2 of the Population Assessment of Tobacco and Health (PATH), nationally representative longitudinal panel datasets. Nicotine naïve adolescents, ages 12-17 at baseline (N = 7933) were included in the study sample. Multivariable logistic regression was conducted to examine the determinants of adolescents' ever use of e-cigarettes. Overall, 12.3% (n = 983) of adolescents who were naïve to nicotine products at Wave 1 became ever users of e-cigarettes at Wave 2. Susceptibility to e-cigarette use at Wave 1 was a significant predictor of ever use at Wave 2 (adjusted odds ratio = 2.27; 95% CI = 1.92, 2.68). Adolescents who were not susceptible to e-cigarette use at Wave 1 but became ever users at Wave 2 were more likely to show a higher level of alcohol use, marijuana use, other substance use, have modified family, be exposed to secondhand tobacco smoke, and have a higher level of psychological problems. The specificity of susceptibility measure was 73.2% (5080/6936) and sensitivity was 57.3% (563/983). The findings of the present study appear to support the predictive validity of the susceptibility to e-cigarette use measure as a significant predictor of future e-cigarette use.

Using machine learning to predict opioid misuse among U.S. adolescents.

This study evaluated prediction performance of three different machine learning (ML) techniques in predicting opioid misuse among U.S. adolescents. Data were drawn from the 2015-2017 National Survey on Drug Use and Health (N = 41,579 adolescents, ages 12-17 years) and analyzed in 2019. Prediction models were developed using three ML algorithms, including artificial neural networks, distributed random forest, and gradient boosting machine. The performance of the ML prediction models was compared with performance of the penalized logistic regression. The area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC) were used as metrics of prediction performance. We used the AUPRC as the primary measure of prediction performance given that it is considered more informative for assessing binary classifiers on imbalanced outcome variable than AUROC. The overall rate of opioid misuse among U.S. adolescents was 3.7% (n = 1521). Prediction performance was similar across the four models (AUROC values range from 0.809 to 0.815). In terms of the AUPRC, the distributed random forest showed the best performance in prediction (0.172) followed by penalized logistic regression (0.162), gradient boosting machine (0.160), and artificial neural networks (0.157). Findings suggest that machine learning techniques can be a promising technique especially in the prediction of outcomes with rare cases (i.e., when the binary outcome variable is heavily lopsided) such as adolescent opioid misuse.

A prospective longitudinal relation between elevated use of electronic devices and use of electronic nicotine delivery systems.

BACKGROUND: In recent years, there has been a rapid increase in the use of both electronic nicotine delivery systems (ENDS) and electronic devices among U.S. youth. Informed by the Diffusion of Innovations Theory (DIT), it was hypothesized that elevated use of electronic devices (EUED) prospectively would predict ENDS use among youth. METHODS: Data were drawn from the Population Assessment of Tobacco and Health (PATH) Study, a longitudinal cohort study in a nationally representative sample. Participants who were 12-17 years old, and naïve to both conventional cigarettes and ENDS at baseline (N = 11,325) were sampled. A total of 8723 respondents had matched data from Wave 1 to Wave 2 and 6051 respondents had matched data for all the three waves. Multivariable sequential logistic regressions were conducted to examine determinants of ENDS use in later waves using R version 3.5.2. RESULTS: Among youth who were naïve to both ENDS and conventional cigarettes at baseline, those with EUED were more likely to initiate ENDS use in later years than those without EUED even after controlling for exposure to ENDS advertisements and other well-established covariates of ENDS use. Daily (adjusted odds ratio [AOR] ranges from 2.76 to 3.56) and weekly (AOR ranges from 2.16 to 2.65) social networking service (SNS) users were more likely to initiate ENDS use than non-users of SNS in the adjusted models. CONCLUSIONS: The findings support the hypothesis that EUED prospectively predicts ENDS use among youth. The use of DIT framework helps understand the link between EUED and ENDS use.

Association between psychological distress and elevated use of electronic devices among U.S. adolescents: Results from the youth risk behavior surveillance 2009-2017.

OBJECTIVE: The purpose of this study was to provide updated information about the prevalence and temporal trends of elevated use of electronic devices (EUED) in leisure time (i.e., 3 h or more on an average school day) in nationally representative samples of U.S. adolescents in recent years and to determine whether there is a significant association between EUED and psychological distress. METHODS: We used the national Youth Risk Behavior Survey (YRBS) data from 2009, 2011, 2013, 2015, and 2017 (N = 75,807). Propensity score matching was used to reduce selection bias due to potential confounding factors with EUED. Ordinal logistic regression analyses were performed for the matched samples to predict the association between EUED and psychological distress. RESULTS: The prevalence of EUED in U.S. youth has substantially increased from 24.9% in 2009 to 43.1% in 2017 (p < .001). Boys had higher rates of EUED than girls only in 2009 and 2011 but not in 2013, 2015, and 2017. A significant association between EUED and psychological distress was identified throughout all the five survey years. The odds of having a higher level of psychological distress increased approximately 1.5 times among youth with EUED than those without. CONCLUSIONS: The prevalence of U.S. youth with psychological distress and EUED has increased simultaneously in the past several years. Future longitudinal studies are warranted to examine causal and/or reciprocal relationship between the two.