Table 2

Effect of intervention on biochemically verified smoking cessation*

 Intervention
(BA-YBE)		Control
(VBA)
% (number)Percentage point difference
(95% CI)		Intracluster correlation coefficient
(95% CI)		Model 1: Relative risk†
Model 2: Adj. relative risk‡
Model 3: Adj. relative risk§
(95% CI)
Primary outcome: abstinence for 6 months¶2.6 (16/611)0.5 (3/602)2.12 (0.74 to 3.51)0.014 (0.000 to 0.033)Model 1: 5.32 (1.43 to 19.74) **
Model 2: 5.10 (1.46 to 17.84) **
Model 3: 4.54 (1.21 to 17.01) ††
Secondary outcome: point prevalence at 6 months‡‡3.1 (19/611)0.8 (5/602)2.28 (0.72 to 3.83)0.014 (0.000 to 0.034)Model 1: 3.77 (1.31 to 10.81) **
Model 2: 3.71 (1.25 to 11.02) §§
Model 3: 2.87 (0.92 to 8.93) ¶¶

  • *Outcome measures were biochemically verified using saliva cotinine; all participants reporting no tobacco use at final follow-up were asked to provide a saliva samples. Failure to provide biochemical verification was low (1.2% (14/1213); there were no differences by treatment group) with those not providing saliva samples or lost to follow-up counted as treatment failures.

  • †Results from conditional/cluster-specific approach with mixed effects log-binomial regression are presented, using intention to treat analysis based on all participants (N=1213). Results were confirmed using a marginal/population approach which yielded comparable estimates (primary outcome: 5.33, 95% CI 1.30 to 21.90, p=0.020; secondary outcome: 3.76, 95% CI 1.31 to 10.82, p=0.014).

  • ‡Results from model 1 were adjusted for chance imbalances, including all variables in table 1. Data missing at random were imputed using standard multivariate imputations with chained equations (MICE) in STATA (‘mi chained’ command with burn in of 100 iterations and 20 imputations) which is more suitable for mixed (dichotomous and continuous) variables,32 and this analysis is therefore based on all participants (N=1213).

  • §Results from model 1 were adjusted for chance imbalances, including all variables in table 1. Only participants with complete data were included in the analysis (N=1009).

  • ¶The primary outcome conforms to the Russell Standard guidelines, that is, self-reported abstinence for at least 6 months (with no more than five occasions of tobacco use) and no tobacco use in the last week, as indicated at final follow-up.

  • **p=0.01.

  • ††p=0.03.

  • ‡‡The secondary outcome was point prevalence at 6 months defined as complete abstinence in the week prior to follow-up.

  • §§p=0.02.

  • ¶¶p=0.07.

  • BA-YBE, brief advice, yogic breathing exercises; VBA, very brief advice.