Elsevier

Drug and Alcohol Dependence

Volume 193, 1 December 2018, Pages 1-6
Drug and Alcohol Dependence

Full length article
Development and validation of a virtual agent to screen tobacco and alcohol use disorders

https://doi.org/10.1016/j.drugalcdep.2018.08.025Get rights and content

Highlights

Abstract

Background

Substance use disorders are under-detected and not systematically diagnosed or screened for by primary care. In this study, we present the acceptability and validity of an Embodied Conversational Agent (ECA) designed to screen tobacco and alcohol use disorder, in individuals who did not seek medical help for these disorders.

Methods

Individuals were included from June 2016 to May 2017 in the Outpatient Sleep Clinic of the University Hospital of Bordeaux. DSM-5 diagnoses of tobacco and alcohol use disorders were assessed by human interviewers. The ECA interview integrated items from the Cigarette Dependence Scale-5 (CDS-5) for tobacco use disorder screening, and the “Cut Down, Annoyed, Guilty, Eye-opener” (CAGE) questionnaire for alcohol use disorder screening. Paper version of CDS-5 and CAGE questionnaires and acceptability questionnaire was also self-administered.

Results

Of the 139 participants in the study (mean age 43.0 [SD = 13.7] years), 71 were women, and 68 were men. The ECA was well accepted by the patients. Paper self-administered CDS-5 and CAGE scores had a strong agreement with the ECA (p < 0.0001). The Receiver Operating Characteristic (ROC) analysis of the ECA interview showed AUC of 0.97 (95% CI, 0.93–1.0) and 0.84 (95% CI, 0.69–0.98) for CDS-5 and CAGE respectively with p-value <0.0001.

Conclusions

This ECA was acceptable and valid to screen tobacco or alcohol use disorder among patients not requesting treatment for addiction. The ECA could be used in hospitals and potentially in primary care settings to help clinicians to better screen their patients for alcohol and tobacco use disorders.

Introduction

Tobacco use disorder estimate is 22% in worldwide general population (World Health Organization, 2016), and that of alcohol use disorder is 4.1% (World Health Organization, 2014). However, the proportion of individuals who have currently been engaged in alcohol or tobacco use disorder treatment is low, probably under 10% (Hasin and Grant, 2015; World Health Organization, 2017). Better screening of individuals for tobacco or alcohol use disorder to provide the appropriate care is of major interest. Substance use disorders are under-detected and not systematically diagnosed or screened in primary care (Tai et al., 2012). As many countries, French health authorities have recommended that health professionals must be involved in tobacco cessation promotion and that primary care physicians should systematically screen for tobacco, alcohol and other substance addictions (Haute Autorité de Santé, 2016). However, physicians reported lack of time as a major barrier to performing such a screening (Harris et al., 2016).

Self-administered questionnaires have been developed for the screening of tobacco and alcohol use disorders. The most used questionnaire is the 6-item Fagerström Test for Nicotine Dependence (FTND) and its simplified 2-item version, the Heaviness of smoking index (HIS) (Etter et al., 1999; Heatherton et al., 1991, 1989). Both have a good validity but a low internal consistency (Underner et al., 2012). The Cigarette Dependence Scale (CDS) was developed as a new self-administered measure of cigarette addiction and has been validated (Etter et al., 2003). Two versions of 12 and 5 items are proposed, and the two have acceptable internal consistency, better than the FTND (Etter, 2008).

For alcohol use disorder screening the most used questionnaires were the “Cut Down, Annoyed, Guilty, Eye-opener” (CAGE) questionnaire (Ewing, 1984) and the Alcohol Use Disorders Identification Test (AUDIT) (Saunders et al., 1993). The 10-item AUDIT presents moderate stability in clinical settings (Sahker et al., 2017). Both tests are effective, but the CAGE is shorter than the AUDIT (Bradley et al., 1998) which targets preferentially heavy drinkers (Seppa et al., 1998).

Although these self-administered questionnaires allow a quick screening for tobacco and alcohol addiction, their implementation is not that easy. They are not very attractive to patients that do not use them when made available in a waiting room (Audran, 2016). The physician will need to take time to encourage patients to fill them or to administer the questionnaire with the patient. Even when used in autonomy by the patient, the physician will need to score each questionnaire to determine what to do, which also takes time.

Recently, computerization of standardized procedures has been proposed to improve implementation of screening tests (Harris et al., 2016) and to allow efficient transmission of data to the individual’s electronic health records. Some computerized-interviews for the screening of problematic substance use, such as the ACASI ASSIST have been developed and have been shown to be acceptable, valid and time-efficient (Harris et al., 2016; Kumar et al., 2016; McNeely et al., 2016). However, the dissemination of use of computerized tools remains an important issue in the field of mental and addictive disorders. Empathic interactions with human-machine interfaces are promising approaches to increase usage of computerized diagnostic tools in patients with mental and addictive disorders. Within the development of these technologies coupled with virtual reality, embodied conversational agents (ECA) (i.e., virtual agents) displayed on computer screen to create an interactive dialogue animation (Cassell et al., 1994) have received an increased interest in the health-related field (Insel, 2017; Nochomovitz and Sharma, 2018). ECAs come from affective computing and provide a strong means of human-system interaction. To date, use of ECA has not been applied to the addiction field. The added value of ECAs over computerized questionnaires is that ECAs have different gestures, facial and verbal expressions and have the potential for non-verbal expressions that is very appealing to human motivation (Philip et al., 2014). They also use specific interactive scenarios and non-verbal communication to appear empathic to the patient and to reinforce adherence to clinical evaluation. Indeed, they represent good candidates to improve and to standardize screening of substance use disorders. We initially designed ECAs to explore excessive daytime sleepiness (Philip et al., 2014) and to diagnose depression (Micoulaud-Franchi et al., 2016a, b; Philip et al., 2017). These ECAs have shown very promising results in term of sensitivity/specificity but also in term of acceptability by patients.

In line with these developments, we developed a new ECA (called Jeanne) able to perform a face-to-face interview to screen tobacco and alcohol use disorders. This paper aims to present the acceptability and the validity of this ECA Jeanne in patients who did not seek medical help for a substance use disorder. We hypothesized that ECA Jeanne would 1) be acceptable to screen for alcohol and tobacco use disorders, 2) present high concurrent validity compared to DSM-5 alcohol and tobacco use disorder diagnosis.

Section snippets

Participants

Participants were recruited from June 2016 to May 2017 among individuals who attended the outpatient sleep disorder clinic at the University Hospital of Bordeaux, France for evaluation of sleep complaints. All consecutive adults were asked to participate in the study after completion of their sleep-related evaluation unless they had insufficient capacity to understand and to answer the questionnaires or had visual/auditory deficits that could interfere with the use of the ECA. The local ethics

Sample selection

Fig. 1 describes the sample selection flowchart. During the time of the study, 646 patients attended the sleep clinic, and 508 were eligible. The majority declined participation due to lack of time. Finally, 139 individuals gave their consent to participate.

Sample characteristic

Table 1 displays the characteristics of the 139 participants. Among them, 62 reported current tobacco use (in the last 12 months) and 27 current heavy alcohol use. The prevalence of current DSM-5 tobacco use disorder was 30.9% (n = 43), and

Discussion

This study aimed to describe the validity and acceptability of an ECA to screen for alcohol and tobacco use disorders in individuals attending a sleep disorder clinic for sleep complaints.

The prevalence of tobacco and alcohol use disorders in our sample (30.9% and 8.6% respectively) were comparable to the prevalence in French general population (Basstianic et al., 2013; Beck and Richard, 2014).

The ECA showed good acceptability for both alcohol and tobacco interviews. This is in line with

Contributors

Marc Auriacombe and Pierre Philip had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analyses.

Pierre Philip was the overall principal investigator of the study, obtained funding and access to participants. Marc Auriacombe was co-investigator and obtained funding. Fuschia Serre and Jean-Arthur Micoulaud-Franchi developed the study design and methods. Sarah Moriceau, Stephanie Bioulac, Pierre Philip and Jean-Arthur

Role of funding sources

Funding for this study was provided by EquipEX Phenovirt (French government Investissement d’avenir Grant). The funding sponsors had no role in the design or conduct of the study, nor in the collection, analysis, interpretation of the data, or in the preparation, review, or approval of this manuscript. The researchers confirm their independence from funders and sponsors.

Conflict of interest

No conflict declared.

Acknowledgments

The authors express their gratitude to all participants for their contribution

References (40)

  • F. Audran

    Intérêt de l’utilisation d’auto-questionnaires en salle d’attente de médecine générale. Université d’Angers.

    (2016)
  • Haute Autorité de Santé

    Actualisation du référentiel de pratiques de l’examen périodique de santé: Dépistage du tabagisme et prévention des maladies liéées au tabac

    (2016)
  • K.L. Barry et al.

    Computerized administration of alcoholism screening tests in a primary care setting

    J. Am. Board Fam. Pract.

    (1990)
  • T. Basstianic et al.

    Drogues et addictions, données essentielles. Observatoire Français des drogues et des toxicomanies

    (2013)
  • K.A. Bradley et al.

    Screening for problem drinking: comparison of CAGE and AUDIT

    J. Gen. Intern. Med.

    (1998)
  • R.G. Carey et al.

    A patient survey system to measure quality improvement: questionnaire reliability and validity

    Med. Care

    (1993)
  • J. Cassell et al.

    Animated conversation: rule-based generation of facial expression, gesture and spoken intonation for multiple conversational agents

    SIGGRAPH

    (1994)
  • J.F. Etter et al.

    Validity of the fagerstrom test for nicotine dependence and of the heaviness of smoking index among relatively light smokers

    Addiction

    (1999)
  • J.F. Etter et al.

    A self-administered questionnaire to measure dependence on cigarettes: the cigarette dependence scale

    Neuropsychopharmacology

    (2003)
  • J.A. Ewing

    Detecting alcoholism. The CAGE questionnaire

    JAMA

    (1984)
  • Cited by (35)

    View all citing articles on Scopus
    View full text