Changes in Tobacco Dependence and Association With Onset and Progression of Use by Product Type From Waves 1 to 3 of the Population Assessment of Tobacco and Health (PATH) Study

Psychometric properties of measures for electronic cigarette dependence among former-smoking electronic cigarette users

BACKGROUND

Measures designed to assess electronic cigarette (ECIG) dependence are largely adapted from those developed for cigarette smoking, and have been evaluated mainly among dual ECIG-cigarette users. The present study evaluated the psychometric properties of four dependence measures among former-smoking ECIG users.

METHODS

Respondents (N = 187) were daily ECIG users who had formerly smoked cigarettes regularly. As part of an online survey, they completed four dependence measures: Penn State Electronic Cigarette Dependence Index (PSECDI); E-Cigarette Dependence Scale (EDS-4); Glover Nilsson Behavioral Questionnaire (GNBQ); and Diagnostic and Statistical Manual 5th Edition (DSM-5).

RESULTS

Confirmatory factor analysis revealed a single factor structure for all four measures, and fit indices were adequate for all models. Most inter-item correlations (mean r's = 0.21 to 0.45) and all item-total correlations (mean r's = 0.53 to 0.80) were significant for all four measures. Internal consistency was highest for the GNBQ (α = 0.86) followed by the EDS-4 (α = 0.81), PSECDI (α = 0.76), and DSM-5 (α = 0.75). Age of ECIG initiation was correlated negatively with scores on the EDS-4 and DSM-5, while number of ECIG quit attempts was correlated positively with scores on all measures except the GNBQ. Convergent validity was highest for comparisons between the EDS-4 and PSECDI (r = 0.75) though all comparisons were significant (p's < 0.001).

CONCLUSIONS

All four measures demonstrated acceptable reliability and validity in this sample of former-smoking ECIG users.