The Effect of Interactivity, Tailoring, and Use Intensity on the Effectiveness of an Internet-Based Smoking Cessation Intervention Over a 12-Month Period: Randomized Controlled Trial

Background: eHealth approaches show promising results for smoking cessation (SC). They can improve quit rates, but rigorous research is sparse regarding their effectiveness and the effects of their interactivity, tailoring, and use intensity. Objective: We examined the effectiveness of Techniker Krankenkasse Smoking Cessation Coaching (TK-SCC), an internet-based, tailored, and interactive SC intervention. Our hypotheses were as follows: hypothesis 1, in the intervention group (IG; access to TK-SCC), a clinically relevant number of participants will be abstinent at the 12-month follow-up (T3); hypothesis 2, the number of abstinent participants will be significantly greater in the IG than the control group (CG) at T3; and hypothesis 3, in the IG, more intense use of TK-SCC will be positively associated with abstinence. Methods: Individuals who smoke were randomized into the IG (563/1115, 50.49%) or CG (552/1115, 49.51%), which received a noninteractive, nontailored, and information-only web-based intervention. Data were collected before the intervention, at the postintervention time point (T1), at the 4-month follow-up (T2), and at T3. We tested hypothesis 1 through equivalence tests between the IG's success rate and success rates of comparable effective interventions reported in 2 current meta-analyses. For hypothesis 2, we conducted binary logistic regressions. For hypothesis 3, we assigned the IG participants to 1 of 4 user types and used binary logistic regressions with user types as the independent variable and smoking abstinence as the dependent variable. Results: In the IG, 11.5% (65/563) and 11.9% (67/563) of participants were smoke free at T1 and T3, respectively. These values were statistically equivalent to the effects in the 2 meta-analyses, which reported 9% (z score=0.64, P=.74) and 10.9% (z score=-0.71, P=.24) success rates, respectively. In the CG, 6.2% (34/552) of the participants were smoke free at T1, which increased up to 8.2% (45/552) at T3. The difference between the IG and CG was statistically significant only at T1 (odds ratio [OR] 2.0, 99% CI 1.1 to 3.6; P=.002), whereas the effect was nonsignificant following alpha error corrections at T3 (OR 1.6, 99% CI 0.9 to 2.7; P=.02). In the IG, constant users of the program became smoke free significantly more often than rare users of the program (T1: OR 15.0, 99% CI 6.1 to 36.9; P<.001; T3: OR 6.5, 99% CI 2.8 to 15.5; P<.001). Conclusions: TK-SCC is effective for SC. However, its superiority compared with a minimal SC intervention could not be confirmed in the long term. Insufficient implementation of the techniques used and cotreatment bias could explain this outcome. Higher use intensity of TK-SCC was positively related to abstinence. Therefore, additional efforts to motivate users to adhere to intervention use as intended could improve the intervention's effectiveness.