A total of 11 RCTs were included (n=15,511 participants). Attrition rates varied widely and intention-to-treat analysis was used by all but one trial. The populations appeared to be homogeneous, although insufficient data were reported in the primary trials. Outcome assessor blinding was poorly reported. Trials used varying definitions of the term abstinence. Treatment and control intervention content were reported as poorly matched and only two trials reporting having carried out power calculations.
There was a significant increase in quit rates for participants using interactive online interventions versus any minimal control condition (RR 1.5, 95% CI 1.2 to 1.9; six trials), although there was also significant heterogeneity. Grouping the analyses according to type of minimal control found a significant non-heterogeneous effect for comparison with booklet or email controls (RR 1.8, 95% CI 1.4 to 2.3; three trials), and a non-significant heterogeneous difference versus static website controls (two trials). One trial reporting interactive online intervention versus waiting-list control found a significant benefit in favour of the online group.
Interactive on-line interventions increased abstinence by 17% (95% CI 12 to 21) compared with minimal controls at six months (RR 1.93, 95% CI 1.44 to 2.6; three trials) although these data were based on point prevalence.
Two trials compared interactive online interventions with face-to-face counselling, and two trials compared behavioural treatment alone versus behavioural treatment plus online components. Neither of these groups found any significant differences between the intervention arms.
No significant results for the publication bias tests were reported.
The full paper also reported details of analyses around effect moderators and variables. In summary, there appeared to be no impact according to length of treatment, no clear impact of automated versus non-automated interventions, and no effect of pharmacotherapy use such as nicotine replacement therapy.
Satisfaction was reported in three trials: overall rates were around 90% and tailored interactive interventions were significantly rated more highly than non-tailored static interventions (details not reported).