Substance Abuse
ISSN: 0889-7077 (Print) 1547-0164 (Online) Journal homepage: http://www.tandfonline.com/loi/wsub20
Risk Estimation Modeling and Feasibility Testing for a Mobile eHealth Intervention for Binge Drinking Among Young People: The D-ARIANNA (Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults) Project Giuseppe Carrà MD, MSc, PhD, Cristina Crocamo MSc, Alessandro Schivalocchi MD, Francesco Bartoli MD, PhD, Daniele Carretta MD, Giulia Brambilla MD & Massimo Clerici MD, PhD To cite this article: Giuseppe Carrà MD, MSc, PhD, Cristina Crocamo MSc, Alessandro Schivalocchi MD, Francesco Bartoli MD, PhD, Daniele Carretta MD, Giulia Brambilla MD & Massimo Clerici MD, PhD (2015) Risk Estimation Modeling and Feasibility Testing for a Mobile eHealth Intervention for Binge Drinking Among Young People: The D-ARIANNA (Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults) Project, Substance Abuse, 36:4, 445-452, DOI: 10.1080/08897077.2014.959152 To link to this article: http://dx.doi.org/10.1080/08897077.2014.959152
View supplementary material
Accepted author version posted online: 09 Sep 2014. Published online: 09 Sep 2014.
Submit your article to this journal
Article views: 105
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=wsub20 Download by: [University of London]
Date: 03 December 2015, At: 08:39
SUBSTANCE ABUSE, 36: 445–452, 2015 Copyright Ó Taylor and Francis Group, LLC ISSN: 0889-7077 print / 1547-0164 online DOI: 10.1080/08897077.2014.959152
ORIGINAL RESEARCH
Downloaded by [University of London] at 08:39 03 December 2015
Risk Estimation Modeling and Feasibility Testing for a Mobile eHealth Intervention for Binge Drinking Among Young People: The D-ARIANNA (DigitalAlcohol RIsk Alertness Notifying Network for Adolescents and young adults) Project Giuseppe Carra, MD, MSc, PhD,1 Cristina Crocamo, MSc,2,3 Alessandro Schivalocchi, MD,2 Francesco Bartoli, MD, PhD,2 Daniele Carretta, MD,2 Giulia Brambilla, MD,2 and Massimo Clerici, MD, PhD2 ABSTRACT. Background: Binge drinking is common among young people but often relevant risk factors are not recognized. eHealth apps, attractive for young people, may be useful to enhance awareness of this problem. We aimed at developing a current risk estimation model for binge drinking, incorporated into an eHealth app—D-ARIANNA (Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults)—for young people. Methods: A longitudinal approach with phase 1 (risk estimation), phase 2 (design), and phase 3 (feasibility) was followed. Risk/protective factors identified from the literature were used to develop a current risk estimation model for binge drinking. Relevant odds ratios were subsequently pooled through meta-analytic techniques with a random-effects model, deriving weighted estimates to be introduced in a final model. A set of questions, matching identified risk factors, were nested in a questionnaire and assessed for wording, content, and acceptability in focus groups involving 110 adolescents and young adults. Results: Ten risk factors (5 modifiable) and 2 protective factors showed significant associations with binge drinking and were included in the model. Their weighted coefficients ranged between ¡0.71 (school proficiency) and 1.90 (cannabis use). The model, nested in an eHealth app questionnaire, provides in percent an overall current risk score, accompanied by appropriate images. Factors that mostly contribute are shown in summary messages. Minor changes have been realized after focus groups review. Most of the subjects (74%) regarded the eHealth app as helpful to assess binge drinking risk. Conclusions: We could produce an evidence-based eHealth app for young people, evaluating current risk for binge drinking. Its effectiveness will be tested in a large trial.
Keywords: Adolescents, binge drinking, eHealth, young adults 1 Division of Psychiatry, Faculty of Brain Sciences, University College London, London, UK 2 Department of Surgery and Interdisciplinary Medicine, University of Milano-Bicocca, Milan, Italy 3 Unit of Biostatistics and Clinical Epidemiology, Department of Public Health, Experimental and Forensic Medicine, University of Pavia, Pavia, Italy Correspondence should be addressed to Cristina Crocamo, MSc, Department of Surgery and Interdisciplinary Medicine, University of Milano-Bicocca, Via Cadore 48, 20900 Monza (MB), Italy. E-mail:
[email protected];
[email protected] Color versions of one or more figures in this article can be found online at www.tandfonline.com/wsub.
INTRODUCTION Excessive alcohol consumption among young people, especially binge drinking (4 drinks for women, 5 drinks for men), is a public health concern in the United States, as more than 15% of those aged between 18 and 24 years are engaged in binge drinking, with a male/female ratio of 3:1.1 In the high-risk group with hazardous alcohol use, this age range contributes 40%.2 European Union member states show similar evidence. The European School Survey Project on Alcohol and Other Drugs (ESPAD), which gathers information on substance abuse among students in 35 European countries, shows that in France, UK, Finland,
Downloaded by [University of London] at 08:39 03 December 2015
446
SUBSTANCE ABUSE
Denmark, and Belgium, 45% of males and 36% of females aged between 15 and 16 years consumed 5 or more drinks on a single occasion during the last 30 days.3 Also in culturally distinct southern Europe countries such as Spain, with possibly healthier drinking cultures, more than 15% of young adults experienced binge drinking at least once in the past month.4 Binge drinking is problematic and dangerous, resulting not only in acute impairment: over half of the annual 80,000 deaths caused by excessive alcohol consumption are determined by binge drinking.5 Indeed, adolescents and young adults who engage in binge drinking are more likely to report other health risk behaviors such as riding with a driver who had been drinking, smoking cigarettes, being a victim of violence, attempting suicide, or using illicit drugs.6 Young people’s knowledge and awareness of binge drinking consequences are often low,7 and perception in terms of risky behavior is scant.8 Impaired decision making is actually a common issue for substance users, who show little regard for consequences and often deny or are unaware that they have a problem.9 This seems even more true for college students who binge drink, although they are more likely to report adverse consequences of drinking (e.g., missing classes, spending less time studying, experiencing unplanned and/or unsafe sex, becoming injured, and getting into legal trouble).10 People misusing alcohol tend to choose courses of action leading to immediate rewards even when these are associated with significant, negative consequences.11 Since compromised decisionmaking mechanisms make people unable to anticipate the negative consequences of their misuse and to learn from previous mistakes, its role in substance use has been acknowledged.9 In particular, recent studies showed that binge drinking among college students is predictive of disadvantageous decision making, even though this should be nearly or fully developed in young people.11 eHealth tools are designed also to support behavioral changes, and nowadays 90% of individuals worldwide have access to mobile phone services, including vulnerable populations, such as people with substance use disorders.12 eHealth technology supporting behavioral treatments for substance use disorders encompasses a wide range of delivery formats (e.g., computer-based, smartphones, tablets) and types of intervention (e.g., brief interventions, behavioral therapy, treatment adherence tools) and have been used across various substances of abuse (e.g., opioids, cocaine, alcohol, cannabis),13 for an array of populations (adults, adolescents and young adults, criminal justice populations, postpartum women), and in a number of different settings (addiction specialty treatment programs, schools, emergency rooms, criminal justice settings).14 Accessibility and availability across settings, enhanced patientclinician communication, conveyance of information in an engaging manner, and individualization and tailoring of intervention are all eHealth advantages appropriate for people with addiction problems.15 As nearly 90% of individuals with a drug or alcohol problem do not access treatment,2 technologybased interventions may allow improved perceived privacy and anonymity, coping with stigmatization or embarrassment about drug use, and increasing the number of people receiving treatment for illicit recreational drug use.16 In particular, eHealth tools for prevention programs have shown encouraging results with regard to identification of binge drinking, alcohol use reduction, and behavioral support among young
people.17,18 Relevant trials are being conducted for mobile phone text message interventions, with participants receiving feedback tailored to their individual responses.19 Traditional preventive intervention on binge drinking has shown poor effectiveness among young people.20 Possibly, impaired decision making in young people who binge drink makes this attempt even more difficult. eHealth tools might address these difficulties, taking advantage of young people’s propensity to use, and expertise with, electronic devices (e.g., smartphones). The objective of this study was to develop an evidence-based current risk estimation model for binge drinking and to incorporate it into an eHealth app for adolescents and young adults. This innovative approach could be useful in designing prevention strategies for binge drinking in young people. In this paper, we report on phase 1 (risk estimation), phase 2 (design), and phase 3 (feasibility) of the D-ARIANNA (Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults) study.
PHASE 1—RISK ESTIMATION Risk estimation models have been used to estimate an outcome of interest determined by multiple interacting risk/protective factors, balancing their relative contributions, to enhance individuals’ decision-making abilities.21 Similar to other biomedical fields (e.g., cardiovascular diseases), factors associated with an increased risk of developing an unhealthy behavior such as binge drinking can be modeled. Although the use of only factors that are potentially modifiable might appear reasonable, including all those that improve risk estimation could encourage individuals at high risk to change remaining risk factors. Furthermore, factors included in estimation models need to be weighted, according to their prevalence in relevant populations, to obtain risk estimation functions. In the cardiovascular field, for example, the most common approach is based on proportional hazards model, either Cox (semiparametric) or Weibull (parametric), accounting for variable follow-up times and losses to follow-up.22 However, logistic regression is also used for risk estimation functions, translating coefficients into more intuitive clinical interpretations. Estimation models have been shown to effectively reduce risk factors levels for cardiovascular diseases (e.g., Cooney et al.22), and logistic regressions have been used in models dealing also with different outcomes (e.g., breast cancer,23 postoperative complications,24 stroke25). Model performance is assessed in terms of several domains: discrimination (maximum achievable sensitivity and specificity); calibration (measuring how closely predicted outcomes agree with actual ones in an external data set); ability to detect implicit interactions among different risk factors; generation of confidence intervals; and ease of clinical interpretation.22 In sum, logistic regression is a powerful statistical method of modeling a binomial outcome through the combination of predictor variables. In the regression model, each regression coefficient provides a description of the size of the contribution of the corresponding predictor variable to the outcome. Therefore, combining relevant predictor variables leads to a regression equation and allows calculating the expected probability that an outcome is “presence of condition of interest” for a given combination of predictors, through a composite function that reflects the relationship between the predictor variables and presence of condition of interest. Thus, a logistic regression estimation model, comprehensively
ET AL. CARRA
including significantly associated risk and protective factors and nested in an appropriately drafted questionnaire, is adequate to predict binge drinking risk.
Methods for Identification of Risk Factors for Binge Drinking The identification of risk/protective factors of binge drinking to implement the risk prediction model has been based on a systematic review of the scientific literature. Furthermore, as national young binge drinkers may have different drinking cultures, as compared with their US and Northern Europe peers, we explored correlates of binge drinking from the Italian Institute of Statistics databases.
Downloaded by [University of London] at 08:39 03 December 2015
Search strategy We used PubMed electronic database for search purposes in order to identify relevant studies published up to May 2013. No time limits, nor language restrictions, were applied. Search phrases combined index and free-text search terms, i.e., “Binge Drinking” and “Risk Factors.” Results were filtered to include only articles on adolescents (13–18 years) and young adults (18– 24 years). Furthermore, we hand-searched reference lists of relevant systematic or narrative reviews on binge drinking among young adults or adolescents.
Eligibility criteria We included any observational study based on a cross-sectional, case-control, or prospective design with the following characteristics: (1) estimates of binge drinking prevalence; (2) analysis of variables associated with binge drinking; and (3) samples of young adults or adolescents. If we found the same data published in multiple works, we retained only the study with more complete information to avoid duplicate results. We included only studies published on peer-review journals and dissertations, excluding conference abstracts.
Data collection process We made a preliminary screening based on titles and abstracts. Papers were then retrieved in full text to test the final eligibility according to inclusion criteria. The eligibility assessment was performed by 2 authors independently (G.B. and D.C.). Discordance on the inclusion or the exclusion of articles was analyzed and disagreements resolved by consensus.
Data extraction We built a data extraction template, including for all the eligible studies key items based on year of publication, country, recruited population, sample size, methods to assess risk factors and binge drinking, and main results. Binge drinkers were considered as cases, and subjects who were not engaged in binge drinking as controls. Data on potential risk factors for binge drinking were collected for both cases and controls. We considered for meta-analyses only risk/protective factors for binge drinking with data available at least from 2 different studies. Factors with statistically significant associations—either positive or negative—with binge drinking were included in the final model. Statistical
447
analyses were performed using Stata version 10.0 SE (StataCorp, College Station, TX, USA).
Risk/Protective Factors and Estimation Model for Binge Drinking Our comprehensive approach identified different risk/protective factors of binge drinking. Since several variables may influence the risk of binge drinking, there is the need to develop a model that can take into account multiple components and cumulative and sometimes, synergistic effects, providing a more tailored approach with a set of identified factors. We developed a risk estimation logistic regression model using the following risk factors statistically associated with binge drinking in young people, ranked in order of coefficients magnitude: past 30 days’ cannabis use, interest for clubs and parties, smoking cigarettes, gender, past 2 weeks’ binge, drinking onset at age 17 or younger, peer influence, parental alcohol misuse, age, and impulsivity as measured by SURPS.26 In addition, 2 protective factors were identified and included in the model, i.e., volunteering and school proficiency (see Supplemental Material, eReferences, for studies used for the model). The regression equation needed then to be translated back into a predicted probability value for a given combination of predictors. Thus, each possible combination of predictors has an expected probability calculated from the regression equation. As the model provided 9216 scenarios based on possible combinations of risk factors, distribution data were split into 5 equal groups, thus calculating 4 quintiles.27 Relevant cutoff points, rounded to the nearest integers, were 43, 62, 82, and 93. However, for ease of interpretation, these were amalgamated, deriving a simple percentage scoring system in 3 levels. Based on the coefficients of the model, we identified low (0–43%), moderate (43.1– 82%), and high (82.1–100%) risk levels. Although most of risk factors can be generalizable across different populations, accounting for variability due to local drinking cultures, prevalence data for age groups and gender of young binge drinkers were collected from the Italian Institute of Statistics databases. Thereby, the model can be easily adapted if analogous national data on age and gender can be retrieved. If other investigators wish to explore modification of binge drinking risk by demographic factors, our model, along with the eHealth app, will be available upon request by following the link http://darianna.org/eng/download-filemakereng/. Weighted coefficients and main characteristics for each factor are shown in Table 1. The vast majority of samples of included studies ranging from 76 to 44,610 individuals made up of school, college, and university students from Anglo-Saxon countries. We selected factors on the basis of several features, i.e., the strength of association with binge drinking, consistency among the definitions of the independent variable provided by different studies, and sensitive nature of the topic. Modifiable nature of risk factors was also considered. Several different risk factors, originally identified from the literature review, were not included. We deliberately excluded those that could look inappropriate in terms of sensitiveness, specifically according to the local cultural background (i.e., sexual orientation,28 history of sexual abuse,29 flatshare,30 single-parent family,31 and having a religion32). Factors with paucity of data (i.e., videogames,32 pocket money,33 ethnicity,34 and drunk-driving35) were also excluded, as well as those without specific measures, appropriate to be nested in an easy-touse eHealth app (i.e., alcohol expectancies36 and anxiety and
448
SUBSTANCE ABUSE
TABLE 1 Factors and Weighted Coefficients Contributing to Binge Drinking Risk Estimation Model in Young People: Characteristics of Included Studies Risk/protective factor (weighted coefficient) Cannabis use (past 30 days) (1.908)
Interest for clubs and parties (1.61)
Downloaded by [University of London] at 08:39 03 December 2015
Smoking cigarettes (1.24)
Gender (1.058) Binge (past 2 weeks) (0.944) Drinking onset at age 17 or younger (0.753)
Parental alcohol misuse (0.752) Age (0.647) Peer influence (0.405)
Impulsivityx(0.272) Volunteering (¡0.326) School proficiency (¡0.717)
Source*
Country
Setting
Size
Odds ratio (95% CI)
McGee et al., 2010 Digrande et al., 2000 Wechsler et al., 1995 Vickers et al., 2004 Gallimberti et al., 2011 Sanchez et al., 2011 Wechsler et al., 1995 Wechsler, Fulop et al., 1997 Griffiths et al., 2006 Schorling et al., 1994 Harrison et al., 2008 Wickholm et al., 2003 Jonas et al., 2000 ISTAT National Institute of Statistics, 2012 Wechsler, Davenport et al., 1997 Beets et al., 2009 Cranford et al., 2006 Digrande et al., 2000 Youth Risk Behavior Surveillance System (YRBSS) 2011 Youth Risk Behavior Surveillance System (YRBSS) 2003 Maggs et al., 2011 Sanchez et al., 2011 Wechsler et al., 1995 ISTAT, National Institute of Statistics, 2012 Hanewinkel and Sargent, 2009 Eaton et al., 2004 Harakeh et al., 2012 Gallimberti et al., 2011 Kim et al., 2009
New Zealand Italy USA USA Italy Brazil USA USA China USA USA Sweden UK Italy
University students University students College students University students High school students High school students College students College students University freshmen students College students Young adults High school students Young adults Adolescents and young adults
1356 1911 17,592 412 802 2582 17,592 17,592 2630 3374 5838 6287 14,762 8325
10.34 [6.39–16.73] 9.55 [6.99–13.06] 7.13 [6.36–7.99] 11.05 [5.10–23.96] 5.04 [2.03–12.52] 5.26 [4.40–6.29] 5.38 [5.00–5.80] 4.69 [4.37–5.00] 4.20 [2.50–7.20] 4.11 [3.28–5.15] 2.57 [2.12–3.12] 5.90 [5.00–6.90] 3.54 [3.16–3.97] 2.88 [2.44–3.42]
USA USA USA Italy USA
College students College freshmen College students University students Young adults
17,592 827 4580 1911 14,751
2.84 [2.6–3.1] 1.23 [1.15–1.32] 2.74 [2.34–3.20] 7.62 [5.66–10.26] 2.03 [1.65–2.49]
USA
Young adults
10,387 2.32 [1.89–2.85]
USA Brazil USA Italy
College freshmen High school students College students Adolescents and young adults
200 2582 17,592 8325
2.35 [1.48–3.72] 1.94 [1.55–2.43] 2.15 [1.97–2.34] 1.91 [1.78–2.06]
Germany USA Netherlands Italy China
2708 2004 1742 845 3041
1.53 [1.25–1.88] 1.31 [1.08–1.61] 1.40 [1.02–1.92] 1.25 [1.10–1.43] 4.10 [2.30–7.60]
Castellanos-Ryan et al., 2011 Friedel, L.K. dissertation, 2011 Weitzman and Chen, 2005 Weitzman and Kawachi, 2000 Miller et al., 2007 Engs et al., 1997
UK Netherlands USA USA USA USA
76 210 27,687 17,592 14,114 11,621
1.27 [1.02–1.58] 1.33 [1.15–1.54] 0.69 [0.65–0.73] 0.77 [0.71–0.83] 0.44 [0.39–0.51] 0.46 [0.34–0.63]
Hanewinkel et al., 2012
Italy
2668
0.39 [0.31–0.49]
Donath et al., 2012 Wechsler et al., 1995 Porter and Pryor, 2007
Germany USA USA
Rasic et al., 2011
Canada
Secondary school students High school students Secondary school students Secondary school students 1st- and 2nd-year university students Secondary school students Adolescents College students College students High school students College and university students Secondary and high school students High school students College students College and university students High school students
44,610 0.84 [0.82–0.87] 17,592 0.64 [0.60–0.68] 41,599 0.75 [0.72–0.78] 1615
0.48 [0.38–0.61]
*Full references of studies used to develop risk estimation model are listed in Supplemental Material (eReferences). xAssessed by SURPS (Substance Use Risk Profile Scale).
depression,37,38 although comorbid mental disorders might play an important role39–42). Other factors were omitted for several different reasons, such as equivocal findings (i.e., socioeconomic status32), wide range of legal drinking ages in different countries (i.e., ease of alcohol availability43), heterogeneity of risk factor definitions (i.e., playing sports44 and media influence45), and overlapping/correlation (e.g., impulsivity and engaging in physical fighting46).
PHASE 2—DESIGN Questionnaire Drafting and Incorporation of Risk Estimation Model Into the eHealth App Young people may have specific insights in terms of appearance, and peculiar key questions/impediments/facilitators need to be
Downloaded by [University of London] at 08:39 03 December 2015
ET AL. CARRA
considered. Thus, we designed a specific questionnaire investigating domains derived from identified risk/protective factors (see Appendix). We took into account features that could affect response, in particular order and wording of the questions based on theoretical concepts applied to research in health.47 We utilized closed questions in order to develop easily suitable response codes, providing comprehensive and unambiguous categories and, if necessary, an “other” category if we felt that there might be some unexpected answers. We built short queries, banning negatives, since these are confusing and ambiguous. Wording was based on familiar statements and phrases that young people can understand. Hence, we avoided, when not strictly necessary, an excessively formal lexicon, preferring a smooth and easy-to-understand language. We placed first simple and basic questions in order to retain rapport and goodwill, and those that seemed most sensitive at a later stage. For questions on impulsivity, we used a Likert scale, a quick and popular method that contains a series of “opinion” statements about an issue, as required by SURPS. Finally, the questionnaire was included in an eHealth app (DARIANNA, Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults), estimating user’s percentage risk for binge drinking and providing the above-described classification of different risk levels (from low to high), with userfriendly screens and simplified graphical interfaces. Risk factors that mostly contribute to the overall score are shown in a closing summary message (Supplemental Material, eFigure).
PHASE 3—FEASIBILITY Design A feasibility study was conducted, aimed at verifying users’ ability to comprehend the questionnaire, removing expressions or content that could be experienced as offensive or provocative, and gathering feedback and suggestions about the graphics and usability of the eHealth app. Recruitment took place at schools and in urban locations of Milan, e.g., live music events. Young people with the following characteristics were consecutively recruited: (1) aged between 16 and 24 years and (2) having exceeded with alcohol at least once in the last 6 months (screening question). Informed consent of the subject, as well as the informed permission of the parents as required, was sought. Those who joined the study received an information sheet and signed a written consent. In order to ease the sampling and to minimize embarrassment, the recruitment was conducted by peers similar to the target population introducing the tool and asking the screening question. Eleven focus groups, each made up of 10 participants were organized. The Ethics Committee of University of Milano–Bicocca approved the study.
Results With a response rate of 82% among those approached, the sample included 70 males (64%) and 40 females (36%). The vast majority of them were born in Italy, and 23% had been engaged in binge drinking in the last 2 weeks. All were students at different educational stages. We evaluated acceptability and wording of DARIANNA app, providing each enrollee with a form, immediately after app completion. Ninety-eight percent of subjects considered D-ARIANNA an easy-to-use app, and about 94% of subjects
449
would recommend it to a friend. Nevertheless, 18% felt that some questions in the app sounded provocative or covered sensitive topics (i.e., parental alcohol misuse, school proficiency, past 2 weeks’ binge, past 30 days’ cannabis use, and drinking onset at age of 17 or younger). In addition, 13% of subjects suggested minor wording changes in questions on impulsivity, peer influence, smoking cigarettes, and interest for clubs and parties. We assessed also subjects’ knowledge of binge drinking before and after app self-administration. At the first stage, 72% of subjects did not know what binge drinking was; only 4% knew this behavior, whereas 24% had a grasp about binge drinking as a concept but not about the term. Having completed D-ARIANNA, 91% of subjects were able to provide a correct definition of binge drinking. Furthermore, in the overall evaluation of the eHealth app, most of the subjects found it helpful to assess binge drinking risk (74%), although 24% stated that the questionnaire was not entirely, and 2% not at all, appropriate.
DISCUSSION Main Findings We used risk factor information collected from the literature to predict binge drinking in young people. The resulting model establishes a series of risk and protective factors as highly clinically significant predictors of binge drinking that are even more powerful factors than age. The model confirms most previously established risk factors as independent predictors of binge drinking in young people. Furthermore, taking into account variability due to local drinking cultures, it can be easily adapted in different contexts if prevalence data for age groups and gender of young binge drinkers can be retrieved from national sources. Acceptability and wording of the related questionnaire, as incorporated into the D-ARIANNA eHealth app, even before introducing minor changes, seem excellent according to targeted focus groups in the feasibility study. The vast majority of the sample would recommend it to a friend.
Implications for Prevention The modifiable nature of a number of risk factors challenges possibly compromised decision-making mechanisms in young binge drinkers.11 Cannabis use, interest for clubs and parties, smoking cigarettes, peer influence, as well as identified protective factors (volunteering and school proficiency), all are to some degree manageable conditions that can be targeted by preventive programs. Furthermore, if adopted in clinical, more protected, settings, the risk estimation model can be complemented with additional risk factors that had to be excluded in the digital version because they covered too sensitive topics (i.e., sexual orientation,28 history of sexual abuse,29 flat-share,30 single-parent family,31 and having a religion32) or needed specific measures not appropriate for an eHealth app (i.e., anxiety and depression37). However, nesting the model into D-ARIANNA app, and considering established characteristics of eHealth tools supporting behavioral changes related to substance use,17 it may prove helpful in contributing to improved decision making. Our preliminary findings show that D-ARIANNA dramatically improves knowledge of binge drinking among young people. However, possibly empowering self-awareness on the negative consequences of hazardous drinking, D-ARIANNA may
450
SUBSTANCE ABUSE
convey a preventive message on binge drinking. Hence, the model running on the eHealth app will be tested at the individual level in a large trial run in natural settings, assessing at least its short-term effectiveness. However, like other psychosocial interventions, there may be the need to boost its effect over time (e.g., Eyberg et al.48), but this can be easily addressed, planning repeated administrations, by the digital nature of the eHealth app.
Downloaded by [University of London] at 08:39 03 December 2015
Limitations We acknowledge several limitations. First, our risk estimation model for binge drinking was built on scientific literature evidence; hence, we lose specificity in terms of local sociodemographic predictors, although generalizability is increased as based on the rigor of selected studies. Second, as the ultimate goal of the model is to predict binge drinking risk from a collection of risk factors, we have to acknowledge that some of these may be unknown for some youngsters. Furthermore, our model shows final point estimates of risk in percentage (e.g., 90% “high”). Although this may be useful in decision-making analysis, these numbers by themselves, without confidence intervals, may create a false sense of certainty,49 and need to be considered with caution. Finally, to the best of our knowledge, logistic regression models have been widely used for risk estimation of conditions fitting the classical medical model of disease (e.g., coronary heart disease, breast cancer, postoperative complications stroke). Behavioral patterns such as binge drinking may stem from predictor variables determining implicit interactions and much more complex relationships in the data.
Conclusions Despite these limitations, the eHealth app D-ARIANNA can be helpful in estimating tailored binge drinking risk and improving self-awareness of adolescents and young adults on the negative consequences of hazardous drinking. Translating scientific evidence into preventive actions supported by eHealth tools has been shown to be cost-effective in similar, pioneering studies.18,50,51 Also, substance use professionals and families could use this novel instrument as a first approach for adolescents and young adults about their alcohol-related behaviors, even before they get involved in dangerous use.52 Working with difficult to engage young people experiencing alcohol-related harm may be less difficult if using eHealth tools that fit their lifestyles.
FUNDING This study was supported (co-funding 50%) by a grant from the Cariplo Foundation (D-ARIANNA, Digital-Alcohol RIsk Alertness Notifying Network for Adolescents project; Rif. 2011-1528). Stefano Reato and Luca Cortese from Saysoon and Eikondata software houses developed the eHealth app. The authors have no conflict of interest in relationship to this paper.
AUTHOR CONTRIBUTIONS All authors made substantial contributions to the intellectual content of the paper. Conception and design: Giuseppe Carra and
Cristina Crocamo. Acquisition of data: Alessandro Schivalocchi, Francesco Bartoli, Daniele Carretta, and Giulia Brambilla. Analysis and interpretation of data: Cristina Crocamo and Giuseppe Carra. Drafting of the manuscript: Giuseppe Carra, Alessandro Schivalocchi, and Francesco Bartoli. Critical revision of the manuscript for important intellectual content: Massimo Clerici and Cristina Crocamo. Statistical analysis: Cristina Crocamo and Giuseppe Carra. Supervision: Massimo Clerici and Giuseppe Carra. All authors gave final approval of the submitted manuscript.
SUPPLEMENTAL MATERIAL Supplemental data for this article can be accessed on the publisher’s Web site.
REFERENCES [1] Center for Disease Control and Prevention. CDC Health disparities and inequalities report—United States, 2011. MMWR Morb Mort Wkly Rep. 2011;60(Suppl):101–104. [2] Substance Abuse and Mental Health Service Administration. Results From the 2011 National Survey on Drug Use and Health: Summary of National Findings. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2012. NSDUH Series H-45, HHS Publication No. (SMA) 12–4725. [3] Hibell B, Guttormsson U, Ahlstrom S, et al. The 2007 ESPAD Report. Substance Use Among Students in 35 European Countries. Stockholm: The Swedish Council for Information on Alcohol and other Drugs (CAN); 2009. [4] Soler-Vila H, Galan I, Valencia-Martin JL, Leon-Mu~noz LM, Guallara-Castillon P, Rodriguez-Artalejo F. Binge drinking in Spain, 2008–2010. Alcohol Clin Exp Res. 2013;2:1–10. [5] World Health Organization. Global Status Report on Alcohol and Health. Geneva: WHO Press, World Health Organization; 2011. Available at: http://www.who.int/substance_abuse/publications/glob al_alcohol_report/msbgsruprofiles.pdf. Accessed March 2014. [6] Miller JW, Naimi TS, Brewer RD, Jones SE. Binge drinking and associated health risk behaviors among high school students. Pediatrics. 2007;119:76–85. [7] De Visser RO, Birch JD. My cup runneth over: young people’s lack of knowledge of low-risk drinking guidelines. Drug Alcohol Rev. 2012;31:206–212. [8] Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality. The NSDUH Report: Trends in Adolescent Substance Use and Perception of Risk From Substance Use. Rockville, MD: Substance Abuse and Mental Health Service Administration; 2013. Available at: http://www.samhsa.gov/ data/2k13/NSDUH099a/sr099a-risk-perception-trends.pdf. Accessed March 2014. [9] Bechara A. Risky business: emotion, decision-making and addiction. J Gambl Stud. 2003;19:23–51. [10] Figlock, D. Impaired Decision Making as a Risk Factor for College Student Drinking [PhD dissertation]. Ames, IA: University of Iowa; 2010. Available at: http://ir.uiowa.edu/etd/801. Accessed March 2014. [11] Goudriaan AE, Grekin ER,SherKJ.Decision making and binge drinking: a longitudinal study. Alcohol Clin Exp Res. 2007;31:928–938. [12] McClure EA, Acquavita SP, Harding E, Stitzer ML. Utilization of communication technology by patients enrolled in substance abuse treatment. Drug Alcohol Depend. 2013;129:145–150. [13] Kiluk BD, Carroll KM. New developments in behavioral treatments for substance use disorders. Curr Psychiatry Rep. 2013;15:420.1–9.
Downloaded by [University of London] at 08:39 03 December 2015
ET AL. CARRA [14] Marsch LA, Carroll KM, Kiluk BD. Technology-based interventions for the treatment and recovery management of substance use disorders: a JSAT special issue. J Subst Abuse Treat. 2014;46:1–4. [15] Olmstead TA, Ostrow CD, Carroll KM. Cost-effectiveness of computer assisted training in cognitive-behavioral therapy as an adjunct to standard care for addiction. Drug Alcohol Depend. 2010;110:200–207. [16] Wood SK, Eckley L, Hughes K, et al. Computer-based programmes for the prevention and management of illicit recreational drug use: a systematic review. Addict Behav. 2014;39:30–38. [17] Fraeyman J, Van Royen P, Vriesacker B, De Mey L, Van Hal G. How is an electronic screening and brief intervention tool on alcohol use received in a student population? A qualitative and quantitative evaluation. J Med Internet Res. 2012;14:e56. doi: 10.2196/jmir.1869. [18] Kypri K, Hallett J, Howat P, et al. Randomized controlled trial of proactive Web-based alcohol screening and brief intervention for university students. Arch Intern Med. 2009;169:1508–1514. [19] Suffoletto B, Callaway CW, Kristan J, Monti P, Clark DB. Mobile phone text message intervention to reduce binge drinking among young adults: study protocol for a randomized controlledtrial. Trials. 2013;14:93.1–8. [20] Ferri M, Allara E, Bo A, Gasparrini A, Faggiano F. Media Campaigns for the prevention of illicit drug use in young people. Cochrane Database Syst Rev. 2013;(6):CD009287. doi: 10.1002/ 14651858.CD009287.pub2. [21] D’Agostino RB, Grundy S, Sullivan LM, Wilson P, for the CHD risk prediction group. Validation of the Framingham coronary heart disease prediction scores results of a multiple ethnic groups investigation. JAMA. 2001;286:180–187. [22] Cooney MT, Dudina A, D’Agostino R, Graham IM. Cardiovascular risk-estimation systems in primary prevention: do they differ? Do they make a difference? Can we see the future? Circulation. 2010;122:300–310. [23] Ayer T, Chhatwal J, Alagoz O, Kahn CE Jr, Woods RW, Burnside ES. Comparison of logistic regression and artificial neural network models in breast cancer risk estimation. Radiographics. 2010;30:13–22. [24] Toner CC, Broomhead CJ, Littlejohn IH, et al. Prediction of postoperative nausea and vomiting using a logistic regression model. Br J Anaesth. 1996;76:347–351. [25] Aviv RI, d’Esterre CD, Murphy BD, et al. Hemorrhagic transformation of ischemic stroke: prediction with CT perfusion. Radiology. 2009;250:867–877 [26] Woicik PA, Stewart SH, Pihl RO, Conrod PJ. The Substance Use Risk Profile Scale: a scale measuring traits linked to reinforcementspecific substance use profiles. Addict Behav. 2009;34:1042–1055. [27] Altman DG, Bland JM. Statistics notes: quartiles, quintiles, centiles, and other quantiles. BMJ 1994;309:996–996. [28] Jasinski JL, Ford JA. Sexual orientation and alcohol use among college students: the influence of drinking motives and social norms. J Alcohol Drug Educ. 2007;51:63–82. [29] Mouilso ER, Fischer S, Calhoun KS. A prospective study of sexual assault and alcohol use among first-year college women. Violence Vict. 2012;27:78–94. [30] Kypri K, Paschall MJ, Langley J, Baxter J, Cashell-Smith M, Bourdeau B. Drinking and alcohol-related harm among New Zealand University students: findings from a national Web-based survey. Alcohol Clin Exp Res. 2009;33:307–314. [31] Fisher LB, Miles IW, Austin SB, Camargo CA Jr, Colditz GA. Predictors of initiation of alcohol use among US adolescents findings from a prospective cohort study. Arch Pediatr Adolesc Med. 2007;161:959–966. [32] Sanchez ZM, Martins SS, Opaleye ES, Moura YG, Locatelli DP, Noto AR. Social factors associated to binge drinking: a cross-sectional survey among Brazilian students in private high schools. BMC Public Health. 2011;11:201. [33] Bellis MA, Hughes K, Morleo M, et al. Predictors of risky alcohol consumption in schoolchildren and their implications for preventing alcohol-related harm. Subst Abuse Treat Prev Policy. 2007;2:15:1–10.
451
[34] Elton-Marshall T, Leatherdale ST, Burkhalter R. Tobacco, alcohol and illicit drug use among Aboriginal youth living off-reserve: results from the Youth Smoking Survey. CMAJ. 2011;183: E480–E486. [35] Tin ST, Ameratunga S, Robinson E, Crengle S, Schaaf D, Watson P. Drink driving and the patterns and context of drinking among New Zealand adolescents. Acta Pædiatr. 2008;97:1433–1437. [36] Eaton DK, Forthofer MS, Zapata LB, et al. Factors related to alcohol use among 6th through 10th graders: the Sarasota County Demonstration Project. J Sch Health. 2004;74:95–104. [37] Oei TP, Morawska A. A cognitive model of binge drinking: the influence of alcohol expectancies and drinking refusal self-efficacy. Addict Behav. 2004;29:159–179. [38] Mitchell AJ, Coyne JC. Do ultra-short screening instruments accurately detect depression in primary care? A pooled analysis and meta-analysis of 22 studies. Br J Gen Pract. 2007;57: 144–151. [39] Carra G, Johnson S. Variations in rates of comorbid substance use in psychosis between mental health settings and geographical areas in the UK. A systematic review. Soc Psychiatry Psychiatr Epidemiol. 2009;44:429–447. [40] Carra G, Johnson S, Bebbington P, et al. The lifetime and pastyear prevalence of dual diagnosis in people with schizophrenia across Europe: findings from the European Schizophrenia Cohort (EuroSC). Eur Arch Psychiatry Clin Neurosci. 2012;262: 607–616. [41] Bartoli F, Carra G, Brambilla G, et al. Association between depression and non-fatal overdoses among drug users: a systematic review and meta-analysis. Drug Alcohol Depend. 2014;134:12–21. [42] Carra G, Bartoli F, Crocamo C, Brady KT, Clerici M. Attempted suicide in people with co-occurring bipolar and substance use disorders: systematic review and meta-analysis. J Affect Disord. 2014;167:125–135. doi: 10.1016/j.jad.2014.05.066. [43] Weitzman ER, Chen YY. Risk modifying effect of social capital on measures of heavy alcohol consumption, alcohol abuse, harms, and secondhand effects: national survey findings. J Epidemiol Community Health. 2005;59:303–309. [44] Tao FB, Xu ML, Kim SD, Sun Y, Su PY, Huang K. Physical activity might not be the protective factor for health risk behaviours and psychopathological symptoms in adolescents. J Paediatr Child Health. 2007;43:762–767. [45] Hanewinkel R, Sargent JD, Poelen EA, et al. Alcohol consumption in movies and adolescent binge drinking in 6 European countries. Pediatrics. 2012;129:709–720. [46] Swahn MH, Simon TR, Hammig BJ, Guerrero JL. Alcohol-consumption behaviors and risk for physical fighting and injuries among adolescent drinkers. Addict Behav. 2004;29:959–963. [47] Bowling A. The tools of quantitative research. In: Bowling A, ed. Research Methods in Health. 2nd ed. Philadelphia, PA: Open University Press, McGraw-Hill Education; 2002:273–309. [48] Eyberg SM, Edwards D, Boggs SR, Foote R. Maintaining the treatment effects of parent training: the role of booster sessions and other maintenance strategies. Clin Psychol Sci Pract. 1998;5:544–554. [49] Schwartz LM, Woloshin S, Welch HG. Risk communication in clinical practice: putting cancer in context. J Natl Cancer Inst Monogr. 1999;25:124–133. [50] McCambridge J, Bendtsen M, Karisson N, White IR, Nilsen P, Bendtsen P. Alcohol assessment and feedback by email for university students: main findings from a randomised controlled trial. Br J Psychiatry. 2013;203:334–340. [51] McTavish FM, Chih MY, Shah D, Gustafson DH. How patients recovering from alcoholism use a smartphone intervention. J Dual Diagn. 2012;8:294–304. [52] Carra G, Clerici M. The Italian Association on Addiction Psychiatry (SIPDip), formerly The Italian Association on Abuse and Addictive Behaviours. Addiction. 2003;98:1039–1042.
452
SUBSTANCE ABUSE
APPENDIX
Downloaded by [University of London] at 08:39 03 December 2015
D-ARIANNA (Digital-Alcohol RIsk Alertness Notifying Network for Adolescents and young adults)
Questionnaire Q1. You are a. Male b. Female Q2. How old are you? a. 16 b. 17 c. 18 d. 19 e. 20 f. 21 g. 22 h. 23 i. 24 Q3. What’s your favorite thing to do on night out? a. Cinema b. Clubs, parties c. Pub d. Other Q4. Do you spend time volunteering? a. No b. Yes Q5. Do you study? a. No b. Yes, in high school c. Yes, in college/university Q6. What’s your grade point average? a. I don’t study anymore b. E-F c. D d. C e. B f. A Q7. Do you smoke? a. No b. Yes Q8. Have you used cannabis in the past 30 days? a. No b. Yes Q9. How many of your friends drink too much? a. Only a few b. Most of them
Q10. In the last two weeks, have you ever had 4 (or 5)* drinks in a row? *5 if male; 4 if female a. No b. Yes Q11. Had you ever drunk more than a few sips of alcohol before you were 17? a. No b. Yes Q12. Does anyone in your family drink too much? a. No b. Yes
Answer according to how much you agree or disagree Q13. Would you say about yourself: “I often don’t think things through before I speak.” a. Disagree strongly b. Disagree somewhat c. Agree somewhat d. Agree strongly Q14. Would you say about yourself: “I often involve myself in situations that I later regret being involved in.” a. Disagree strongly b. Disagree somewhat c. Agree somewhat d. Agree strongly Q15. Would you say about yourself: “I usually act without stopping to think.” a. Disagree strongly b. Disagree somewhat c. Agree somewhat d. Agree strongly Q16. Would you say about yourself: “Generally, I am an impulsive person.” a. Disagree strongly b. Disagree somewhat c. Agree somewhat d. Agree strongly
