Modeling count data in the addiction field: Some simple recommendations

Longitudinal examination of alcohol use motives, item-level protective behavioral strategies, and alcohol-related consequences

BACKGROUND

Alcohol misuse among college students is a public health concern. Protective behavioral strategies (PBS) can be used before, during, after, or instead of drinking to reduce alcohol use and negative consequences, but findings on their utility at the aggregate level are mixed. Although recent work has provided important information on the performance of individual PBS items, it is limited by research designs that are cross-sectional, do not examine consequences, or do not examine other important correlates, such as drinking motives. This study examines both the association between item-level PBS and alcohol-related negative consequences and the moderating effect of drinking motives longitudinally.

METHODS

College students from two universities (n = 200, 62.5% female, Mage = 20.16) completed the Drinking Motives Questionnaire-Revised, Protective Behavioral Strategies Survey, the Rutgers Alcohol Problem Index, and a measure of the quantity of alcohol use at baseline and 3-month follow-up. Generalized linear models were conducted to assess direct effects of item-level PBS on alcohol-related consequences and the moderating effects of drinking motives.

RESULTS

Two PBS items were associated with fewer alcohol-related consequences at follow-up, and two items were associated with greater alcohol-related consequences at follow-up. Drinking motives differentially moderated associations between item-level PBS and alcohol-related consequences for a proportion in the sample. Enhancement motives moderated the greatest number of associations, followed by coping, conformity, and social motives. Certain PBS (e.g., drink slowly, rather than gulp or chug) were moderated by several drinking motives, whereas other PBS items were not moderated by any motives.

CONCLUSION

Consistent with previous research, some item-level PBS were associated longitudinally with increased negative consequences, and some were associated with decreased negative consequences. Drinking motives, particularly enhancement, moderated several item-level PBS and consequence associations, suggesting that reasons for drinking may be important for understanding the associations between PBS strategies and alcohol-related consequences.

The selection of statistical models for reporting count outcomes and intervention effects in brief alcohol intervention trials: A review and recommendations

Understanding the efficacy and relative effectiveness of a brief alcohol intervention (BAI) relies on obtaining a credible intervention effect estimate. Outcomes in BAI trials are often count variables, such as the number of drinks consumed, which may be overdispersed (i.e., greater variability than expected based on a given model) and zero-inflated (i.e., greater probability of zeros than expected based on a given model). Ignoring such distribution characteristics can lead to biased estimates and invalid statistical conclusions. In this critical review, we identified and reviewed 64 articles that reported count outcomes from a systematic review of BAI trials for adolescents and young adults from 2013 to 2018. Given many statistical models to choose from when analyzing count outcomes, we reviewed the models used and reporting practices in the BAI trial literature. A majority (61.3%) of analyses with count outcomes used linear models despite violations of normality assumptions; 75.6% of outcome variables demonstrated clear overdispersion. We provide an overview of available count models (Poisson, negative binomial, zero-inflated or hurdle, and marginalized zero-inflated Poisson regression) and formulate practical guidelines for reporting outcomes of BAIs. We provide a visual step-by-step decision guide for selecting appropriate statistical models and reporting results for count outcomes. We list accessible resources to help researchers select an appropriate model with which to analyze their data. Recent advances in count distribution-based models hold promise for evaluating count outcomes to gauge the efficacy and effectiveness of BAIs and identify critical covariates in alcohol epidemiologic research. We recommend that researchers report the distributional properties of count outcomes, such as the proportion of zero counts, and select an appropriate statistical analysis for count outcomes using the provided decision tree. By following these recommendations, future research may yield more accurate, transparent, and reproducible results.

Primary and pharmaceutical care usage concurrent associations with a severe smoke episode and low ambient air pollution in early life

BACKGROUND

Due to climate change, landscape fires account for an increasing proportion of air pollution emissions, and their impacts on primary and pharmaceutical care are little understood.

OBJECTIVES

To evaluate associations between exposure in two early life periods to severe levels of PM_2.5 from a mine fire, background PM_2.5, and primary and pharmaceutical care.

METHODS

We linked records of births, general practitioner (GP) presentations and prescription dispensing for children born in the Latrobe Valley, Australia, 2012-2014, where a severe mine fire occurred in February-March 2014 in an area with otherwise low levels of ambient PM_2.5. We assigned modelled exposure estimates for fire-related (cumulative over the fire and peak 24-hour average) and annual ambient PM_2.5 to residential address. Associations with GP presentations and dispensing of prescribed medications in the first two years of life (exposure in utero) and in the two years post-fire (exposure in infancy) were estimated using two-pollutant quasi-Poisson regression models.

RESULTS

Exposure in utero to fire-related PM_2.5 was associated with an increase in systemic steroid dispensing (Cumulative: IRR = 1.11, 95%CI = 1.00-1.24 per 240 μg/m³; Peak: IRR = 1.15, 95%CI = 1.00-1.32 per 45 μg/m³), while exposure in infancy was associated with antibiotic dispensing (Cumulative: IRR = 1.05, 95%CI = 1.00-1.09; Peak: IRR = 1.06, 95%CI = 1.00-1.12). Exposure in infancy to ambient PM_2.5, despite relatively low levels from a global perspective (Median = 6.1 μg/m³), was associated with an increase in antibiotics (IRR = 1.10, 95%CI = 1.01-1.19 per 1.4 μg/m³) and in GP presentations (IRR = 1.05, 95%CI = 1.00-1.11), independently from exposure to the fire. We also observed differences in associations between sexes with GP presentations (stronger in girls) and steroid skin cream dispensing (stronger in boys).

DISCUSSION

Severe medium-term concentrations of PM_2.5 were linked with increased pharmaceutical treatment for infections, while chronic low levels were associated with increased prescriptions dispensed for infections and primary care usage. Our findings also indicated differences between sexes.

An examination of the effects of ADHD symptoms and sex on the relation between cannabis protective behavioral strategies and cannabis consequences

The risk for cannabis-related problems is elevated among young adults with attention-deficit/hyperactivity disorder (ADHD) symptoms. It is unknown whether the use of cannabis protective behavioral strategies (PBS) mitigates this risk for college students with ADHD. Prior research finds that college students who use alcohol and report high levels of ADHD symptoms particularly benefit from employing alcohol PBS, and these relations are strongest for male students. Thus, this study examined the moderating effects of ADHD symptoms and sex assigned at birth on the relation between cannabis PBS use and cannabis-related problems among college students who use cannabis. Participants were 384 college students (66.9% female, 57.8% white non-Hispanic, M_age = 19.29) from 12 United States universities who reported past-month cannabis use. Participants completed measures of demographics, ADHD symptoms, past-month cannabis frequency and related problems, and cannabis PBS use via an online survey. There was a significant interaction of ADHD hyperactive/impulsive symptoms, PBS use, and sex on cannabis-related problems, controlling for cannabis use frequency. The strength of the negative association between PBS use and problems depended on level of ADHD symptoms for females, but not for males. However, there were no interactive effects for ADHD inattentive symptoms. These results supplement the literature base on relations between PBS use and ADHD symptoms in college students, extending support for their use to cannabis users. Importantly, promoting PBS use to female college students who are high in hyperactive/impulsive ADHD symptoms is recommended.

Protocol for a mobile laboratory study of co-administration of cannabis concentrates with a standard alcohol dose in humans

Cannabis is commonly used among people who drink alcohol, yet evidence on acute effects of co-use is conflicting. Two important variables that may influence the effects of cannabis and alcohol are cannabinoid content (i.e., the ratio of cannabidiol [CBD] and 9-tetrahydrocannabinol [THC]) as well as the order of use (i.e., cannabis before alcohol vs. alcohol before cannabis). Research is mixed regarding the acute imapct of cannabis on alcohol consumption and intoxication, with some studies suggesting additive effects of alcohol and cannabis, and others demonstrating negligible effects of combining these substances. Further complicating this, high-THC-content cannabis concentrates are increasingly popular on the legal-market, but to our knowledge, no studies have explored concentrate and alcohol co-use. In addition to cannabinoid content, order of use may influence intoxication and other acute effects, but is also understudied. Co-use studies typically administer a fixed dose of alcohol before cannabis, and there is a lack of data on the acute effects of cannabis before alcohol. Thus, there is a need for experimental co-use studies exploring the impact of cannabinoid content (particularly of highly potent cannabis concentrates) and order effects on intoxication. This study uses a federally-compliant mobile laboratory procedure to explore the effects of co-administration of legal-market cannabis concentrates with a moderate alcohol dose (.8g/kg) in a sample of community participants who regularly use alcohol and cannabis. The study will also explore alcohol and cannabis order effects (cannabis before alcohol vs. alcohol before cannabis). Outcomes are objective intoxication (measured using blood cannabinoid level, heart rate, psychomotor performance and breath alcohol level [BrAC]) and subjective intoxication (assessed via self-report measures). Overall, this study may influence harm-reduction recommendations for individuals who drink alcohol and use cannabis.

The secondhand effects of alcohol use and the risks of drinking to cope

Objective:The secondhand effects of alcohol use (SEA) are adverse consequences caused by another's drinking. This study explored the relationship among the experience of SEAs, alcohol use, and alcohol related consequences (ARC). In addition, we examined whether coping (i.e., adaptive, maladaptive, substance use coping, maladaptive coping without substance use items) served as a moderator of SEA effects on alcohol outcomes. Method: 1,168 students completed a survey assessing SEA, alcohol outcomes, ARC, and coping strategies. Results: SEA was significantly positively associated with alcohol use (RRheavydrinking = 1.05, SE = 0.005, p < .01; RRAUDIT = 1.04, SE = .005, p < .01) as well as ARC (RR = 1.06, SE = .005, p < .01). Various forms of maladaptive coping moderated the relationship between SEA and alcohol outcomes. Conclusion: This study provided evidence for a relationship between SEA and more alcohol use and ARC. This relationship was exacerbated by maladaptive coping.

Gambling and gaming during COVID-19: The role of mental health and social motives in gambling and gaming problems

BACKGROUND

The uncertain and stressful global situation caused by the COVID-19 pandemic has the potential to cause anxiety and impact people's mental health. Simultaneously, social distancing policies have isolated people from their normal social interactions. These societal changes have inevitably influenced gambling and gaming practices, and many people may have turned to betting or gaming as a means of social exchange. This study examined the influence of COVID-19 anxiety on increased mental health problems and their relationship to gambling and gaming problems. The mediating role of social motives for gambling and gaming was also investigated.

MATERIAL AND METHODS

Finnish adults (N = 1530; Mage = 46.67; 50.33% male) were recruited from a volunteer participant panel to participate in a survey study. COVID-19 anxiety was assessed using a scale adapted from the Spielberger State-Trait Anxiety Inventory. Mental health problems were evaluated with the five-item Mental Health Inventory. Gambling problems were measured using the Problem Gambling Severity Index, and gaming problems were measured with the Internet Gaming Disorder Test. Social motives for gaming were evaluated using a single-item measure. Analyses used generalised structural equation modelling.

RESULTS

COVID-19 anxiety predicted increased mental health problems, which were associated with heightened gambling and gaming problems. Social motives for gambling and gaming were a significant mediator, suggesting that gambling and gaming problems are emphasized markedly among those who are socially motivated by and involved in such games.

CONCLUSIONS

Mental health problems experienced during the pandemic and the social properties of games might accentuate gambling and gaming problems.

You're Stressing Me Out: The Secondhand Effects of Alcohol

Introduction: Secondhand effects of alcohol use (SEA) are adverse consequences experienced by an individual due to another's drinking (i.e., study/sleep interrupted, being insulted or humiliation, sexual assault or rape). SEA is a serious public health concern among college students due to its serious consequences and high prevalence (60-84%). The present study examined the associations among SEA, personal alcohol use, depression, anxiety, and stress. Methods: 836 students completed an online survey that assessed for study variables. Results: SEA significantly predicted all outcomes with those endorsing SEA reporting higher levels of depression, anxiety, and stress. Moreover, participants who endorsed SEA also reported 30% more frequent binge drinking and 18% more drinks on their heaviest reported drinking day. Conclusions: Findings indicate that the experience of SEA is a risk factor for poor mental and behavioral health outcomes among college students.

Relations among protective behavioral strategies, biological sex, and ADHD symptoms on alcohol use and related problems: Who benefits most, and from what type of strategy?

Attention-deficit/hyperactivity disorder (ADHD) symptoms among college students are associated with high rates of alcohol use and alcohol-related consequences. Use of protective behavioral strategies (PBS) is generally related to lower levels of alcohol use and problems; however, it is unclear how effectively students with ADHD symptoms can implement PBS, and whether certain types of PBS use may yield better outcomes. This study examined relations between PBS type and ADHD symptoms on both alcohol use and consequences, and whether these relations varied by biological sex. Participants were 875 college student drinkers from three universities who completed measures of ADHD symptoms, PBS, past-month alcohol use, and alcohol-related consequences. There were significant moderation effects of ADHD symptoms, such that the relation between PBS use and alcohol use was more pronounced for students high in inattention, and the relation between PBS use and alcohol-related consequences was more pronounced for students high in either inattention or hyperactivity/impulsivity symptoms. These relations were found for both manner of drinking and stopping/limiting drinking PBS, and they tended to be strongest for male students. There were no significant interaction effects that included serious harm reduction PBS; for all students, increased use of this type of PBS was associated with fewer problems. These results suggest that PBS are likely effective for students with ADHD symptoms. Interventions that provide explicit instruction in employing PBS, particularly related to manner of drinking and stopping/limiting drinking strategies, are recommended for students with ADHD symptoms.

Young adult use, dual use, and simultaneous use of alcohol and marijuana: An examination of differences across use status on marijuana use context, rates, and consequences

Objective: Co-use of alcohol and marijuana has increased among college students, though comparisons among simultaneous (i.e., use of both substances such that effects overlap), dual (i.e., use of both substances within a similar time period but without overlapping effects), and marijuana-only use are limited. This study aimed to understand differences between simultaneous, dual, and marijuana-only users on marijuana use rates, consequences, and context of use in a multi-university study. Method: College students (N = 4,764; M_age = 19.9 years) who were mainly female (70.6%) and White (67.9%) completed an online survey. The Marijuana Use Grid captured marijuana use quantity/frequency, and the Brief Marijuana Consequences Questionnaire and the Cannabis Use Disorders Identification Test-Revised assessed problem use. Location, method of consumption, and social context of use also were assessed. Results: Fifty-five percent of the sample endorsed lifetime use of alcohol and marijuana. Of these students, 36.1% endorsed past-month simultaneous use, 10.8% endorsed past-month dual use, and 6.4% endorsed past-month marijuana-only use. Simultaneous users reported more marijuana use and problems than dual users. Marijuana-only users did not differ from simultaneous users on marijuana use indices, though they reported greater use than dual users as well. Simultaneous users used marijuana in plant form, at parties, and with unknown others a greater percentage of the time than dual users, while dual users used edibles and ingested marijuana a greater percentage of the time. Conclusions: Given their greater levels of marijuana use and marijuana-related problems, screening and interventions for simultaneous alcohol-marijuana use are needed in college students. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Identifying an accurate self-reported screening tool for alcohol use disorder: evidence from a Swiss, male population-based assessment

BACKGROUND AND AIMS

Short screenings for alcohol use disorder (AUD) are crucial for public health purposes, but current self-reported measures have several pitfalls and may be unreliable. The main aim of our study was to provide empirical evidence on the psychometric performance of self-reports currently used. Our research questions were: compared with a gold standard clinical interview, how accurate are (1) self-reported AUD, (2) self-reported alcohol use over time and (3) biomarkers of alcohol use among Swiss men? Finally, we aimed to identify an alternative screening tool.

DESIGN

A single-center study with a cross-sectional design and a stratified sample selection.

SETTING

Lausanne University Hospital (Switzerland) from October 2017 to June 2018.

PARTICIPANTS

We selected participants from the French-speaking participants of the ongoing Cohort Study on Substance Use and Risk Factors (n = 233). The sample included young men aged on average 27.0 years.

MEASUREMENTS

We used the Diagnostic Interview for Genetic Studies as the gold standard for DSM-5 AUD. The self-reported measures included 11 criteria for AUD, nine alcohol-related consequences, and previous 12 months' alcohol use. We also assessed biomarkers of chronic excessive drinking (ethyl glucuronide and phosphatidylethanol).

FINDINGS

None of the self-reported measures/biomarkers taken alone displayed both sensitivity and specificity close to 100% with respect to the gold standard (e.g. self-reported AUD: sensitivity = 92.3%, specificity = 45.8%). The best model combined eight self-reported criteria of AUD and four alcohol-related consequences. Using a cut-off of three, this screening tool yielded acceptable sensitivity (83.3%) and specificity (78.7%).

CONCLUSIONS

Neither self-reported alcohol use disorder nor heavy alcohol use appear to be adequate to screen for alcohol use disorder among young men from the Swiss population. The best screening alternative for alcohol use disorder among young Swiss men appears to be a combination of eight symptoms of alcohol use disorder and four alcohol-related consequences.

The Role of Perceived Loneliness in Youth Addictive Behaviors: Cross-National Survey Study

BACKGROUND

In the ever-growing and technologically advancing world, an increasing amount of social interaction takes place through the Web. With this change, loneliness is becoming an unprecedented societal issue, making youth more susceptible to various physical and mental health problems. This societal change also influences the dynamics of addiction.

OBJECTIVE

Employing the cognitive discrepancy loneliness model, this study aimed to provide a social psychological perspective on youth addictions.

METHODS

A comprehensive survey was used to collect data from American (N=1212; mean 20.05, SD 3.19; 608/1212, 50.17% women), South Korean (N=1192; mean 20.61, SD 3.24; 601/1192, 50.42% women), and Finnish (N=1200; mean 21.29, SD 2.85; 600/1200, 50.00% women) youths aged 15 to 25 years. Perceived loneliness was assessed with the 3-item Loneliness Scale. A total of 3 addictive behaviors were measured, including excessive alcohol use, compulsive internet use, and problem gambling. A total of 2 separate models using linear regression analyses were estimated for each country to examine the association between perceived loneliness and addiction.

RESULTS

Loneliness was significantly related to only compulsive internet use among the youth in all 3 countries (P<.001 in the United States, South Korea, and Finland). In the South Korean sample, the association remained significant with excessive alcohol use (P<.001) and problem gambling (P<.001), even after controlling for potentially confounding psychological variables.

CONCLUSIONS

The findings reveal existing differences between youths who spend excessive amounts of time online and those who engage in other types of addictive behaviors. Experiencing loneliness is consistently linked to compulsive internet use across countries, although different underlying factors may explain other forms of addiction. These findings provide a deeper understanding in the mechanisms of youth addiction and can help improve prevention and intervention work, especially in terms of compulsive internet use.

Models for analyzing zero-inflated and overdispersed count data: an application to cigarette and marijuana use

INTRODUCTION

This paper describes different methods for analyzing counts and illustrates their use on cigarette and marijuana smoking data.

METHODS

The Poisson, zero-inflated Poisson (ZIP), hurdle Poisson (HUP), negative binomial (NB), zero-inflated negative binomial (ZINB) and hurdle negative binomial (HUNB) regression models are considered. The different approaches are evaluated in terms of the ability to take into account zero-inflation (extra zeroes) and overdispersion (variance larger than expected) in count outcomes, with emphasis placed on model fit, interpretation, and choosing an appropriate model given the nature of the data. The illustrative data example focuses on cigarette and marijuana smoking reports from a study on smoking habits among youth e-cigarette users with gender, age, and e-cigarette use included as predictors.

RESULTS

Of the 69 subjects available for analysis, 36% and 64% reported smoking no cigarettes and no marijuana, respectively, suggesting both outcomes might be zero-inflated. Both outcomes were also overdispersed with large positive skew. The ZINB and HUNB models fit the cigarette counts best. According to goodness-of-fit statistics, the NB, HUNB, and ZINB models fit the marijuana data well, but the ZINB provided better interpretation.

CONCLUSION

In the absence of zero-inflation, the NB model fits smoking data well, which is typically overdispersed. In the presence of zero-inflation, the ZINB or HUNB model is recommended to account for additional heterogeneity. In addition to model fit and interpretability, choosing between a zero-inflated or hurdle model should ultimately depend on the assumptions regarding the zeros, study design, and the research question being asked.

IMPLICATIONS

Count outcomes are frequent in tobacco research and often have many zeros and exhibit large variance and skew. Analyzing such data based on methods requiring a normally distributed outcome are inappropriate and will likely produce spurious results. This study compares and contrasts appropriate methods for analyzing count data, specifically those with an over-abundance of zeros, and illustrates their use on cigarette and marijuana smoking data. Recommendations are provided.

Modeling count data in the addiction field: Some simple recommendations

Analyzing count data is frequent in addiction studies but may be cumbersome, time-consuming, and cause misleading inference if models are not correctly specified. We compared different statistical models in a simulation study to provide simple, yet valid, recommendations when analyzing count data.We used 2 simulation studies to test the performance of 7 statistical models (classical or quasi-Poisson regression, classical or zero-inflated negative binomial regression, classical or heteroskedasticity-consistent linear regression, and Mann-Whitney test) for predicting the differences between population means for 9 different population distributions (Poisson, negative binomial, zero- and one-inflated Poisson and negative binomial, uniform, left-skewed, and bimodal). We considered a large number of scenarios likely to occur in addiction research: presence of outliers, unbalanced design, and the presence of confounding factors. In unadjusted models, the Mann-Whitney test was the best model, followed closely by the heteroskedasticity-consistent linear regression and quasi-Poisson regression. Poisson regression was by far the worst model. In adjusted models, quasi-Poisson regression was the best model. If the goal is to compare 2 groups with respect to count data, a simple recommendation would be to use quasi-Poisson regression, which was the most generally valid model in our extensive simulations.