Skip to main content
. 2019 Dec 2;180(2):206–214. doi: 10.1001/jamainternmed.2019.5176

Table 3. Generalized Estimating Equation for Biochemically Validated Abstinence at 6- and 12-Month Follow-up Visits Among 1571 Participantsa.

Variable 6 mo 12 mo
aRR (95% CI) P Value aRR (95% CI) P Value
Study group
Intervention 3.21 (1.74-5.93) <.001 2.23 (1.25-3.97) .004
Control 1 [Reference] NA 1 [Reference] NA
Age 1.01 (0.98-1.05) .74 1.00 (0.99-1.02) .59
Sex
Male 0.36 (0.15-0.84) .02 0.92 (0.32-2.62) .87
Female 1 [Reference] NA 1 [Reference] NA
Marital status
Single 0.61 (0.27-1.34) .20 0.54 (0.25-1.19) .11
Divorced, separated, or widowed 0.73 (0.26-2.04) .54 0.73 (0.28-1.88) .50
Married or cohabiting 1 [Reference] NA 1 [Reference] NA
Employment status
Unemployed or retired 1.24 (0.49-3.17) .64 1.37 (0.66-2.83) .38
Employed 1 [Reference] NA 1 [Reference] NA
Educational level
Primary or below 0.57 (0.17-1.84) .33 0.93 (0.26-3.37) .92
Secondary 0.61 (0.23-1.58) .29 0.65 (0.23-1.80) .39
Tertiary 1 [Reference] NA 1 [Reference] NA
Monthly household income, US $
<1275 0.72 (0.23-2.23) .56 1.08 (0.40-2.91) .88
1275-3825 1.02 (0.44-2.37) .96 1.40 (0.57-3.40) .44
≥3825 1 [Reference] NA 1 [Reference] NA
Smoking-related chronic disease
Yes 1.15 (0.58-2.25) .88 1.20 (0.62-2.34) .57
No 1 [Reference] NA 1 [Reference] NA
Health utility score by the SF-6Db 0.19 (0.01-3.68) .27 0.08 (0.01-1.69) .09
Self-efficacy against tobacco by the SEQ-12c 1.02 (0.97-1.07) .11 1.02 (0.99-1.05) .13
Nicotine dependence by the Heaviness of Smoking Indexd
Moderate to heavy 0.31 (0.15-0.61) .001 0.38 (0.20-0.72) .002
Light 1 [Reference] NA 1 [Reference] NA

Abbreviations: aRR, adjusted relative risk; NA, not applicable; SEQ-12, Smoking Self-Efficacy Questionnaire; SF-6D, Short-Form Six-Dimension.

a

The generalized estimating equation model was built to derive RRs based on the goodness of fit. The variables that were statistically significant in the logistic regression and that have been shown to be associated with smoking cessation were selected in the model, and then other variables were put in the model 1 by 1. The model with the least quasi-Akaike information criterion was selected as the best model.

b

The SF-6D is composed of 6 multilevel dimensions. The SF-6D scores were weighted from a sample of the general population, which ranged from 0 to 1.

c

On a 12-item 5-point Likert-type scale in the SEQ-12, responses ranged from “not at all sure” to “absolutely sure.” A summary score of the SEQ-12 ranged from 12 to 60, with higher scores indicating higher self-efficacy.

d

The Heaviness of Smoking Index,31 a 2-item index from multiple-choice response options (0-3), was determined by assessing cigarettes smoked per day and latency to smoke after waking; the higher the indexes, the greater the smoking nicotine dependence.