Evaluation of the Psychometric Properties of Dependence Measures for Exclusive Electronic Cigarette Users.

INTRODUCTION

Extant electronic cigarette (ECIG) dependence measures are largely adapted from those designed for cigarette smoking, though few have been evaluated for their psychometric properties.

AIMS AND

METHODS

Never-smoking ECIG users (N = 134) participating in an online survey completed four dependence measures: Penn state electronic cigarette dependence index (PSECDI), e-cigarette dependence scale (EDS-4), diagnostic and statistical manual for tobacco use disorder (DSM-5), and Glover Nilsson behavioral questionnaire (GNBQ). They also reported on their ECIG use characteristics (eg, behaviors and reasons).

RESULTS

Internal consistency was highest for the EDS-4 (Cronbach's α = 0.88) followed by the GNBQ (α = 0.75), PSECDI (α = 0.72), and DSM (α = 0.71). Confirmatory factor analyses revealed a single-factor structure for the PSECDI, EDS-4, and GNBQ. For the DSM-5, however, two items did not load significantly (ECIG use interferes with responsibilities; reduce/give up activities because of ECIG use). Significant correlations were observed between all measures and the number of ECIG use days/week and/or years using ECIGs, as well as between DSM-5 scores and the number of ECIG quit attempts and initiation age. Endorsement of using ECIGs because "I like flavors" was correlated positively with DSM-5 and GNBQ scores.

CONCLUSIONS

All dependence measures evaluated herein demonstrated adequate reliability and construct validity. Future work should focus on determining which aspects of dependence are those that are unique to ECIG use, and subsequently developing a more comprehensive measure of ECIG dependence.

IMPLICATIONS: The measures assessed herein-PSECDI, EDS-4, DSM-5, and GNBQ-demonstrated adequate to good reliability and construct validity among a sample of never-smoking ECIG users. The dependence domains covered across measures were related yet distinct. Findings demonstrate the need for future evaluation of these different domains to determine which are the most salient characteristics of ECIG dependence.