Soria and colleagues used trained physicians to give motivational interviewing (MI). Physicians recruited and randomised using sealed envelopes. The distribution of patients favours MI with 114 people in the MI group and 86 in the brief advice group. The P-value for the χ2 for goodness of fit to the binomial distribution is 0.048. Patients in the MI group were more likely to intend to stop smoking or consider it than those in the brief advice group measured by stage of change (P = 0.036). Neither of these values prove that randomisation was subverted, but sealed envelopes are notorious for this. Were checks made?
Five people in the MI group received bupropion, but none in the advice group. Bupropion doubles the likelihood of cessation.1 The authors used logistic regression to potentially adjust for confounders if significant, but this leads to important confounding. Epidemiologists recommend adjusting for a range of potential confounders regardless of their statistical significance.2
The outcome assessment makes interpretation difficult. The outcome is point prevalence abstinence for an undefined period measured by the physician giving treatment. MI patients had up to three sessions with the outcome assessor to motivate cessation, while those in the brief advice arm had one. Might those in the MI arm have felt pressure to declare abstinence when it was not fully achieved? Most smokers have an exhaled carbon monoxide in the non-smoker range by overnight abstinence.3 Additionally, most who are point prevalent abstinent do not achieve lifetime abstinence, which is the outcome linked to health benefits. If the authors have the data, it would be preferable (and in line with recommendations)4 to report sustained abstinence for 6 months between 6 and 12 months, as around 30–40% of these patients will achieve lifetime abstinence.
Soria et al compare the odds ratio for MI in this study (6.25) to the odds ratio from the meta-analysis in the Cochrane review (1.56) of individual behavioural support for smoking cessation.5 These are not comparable interventions. MI, like brief advice, primarily motivates patients to attempt to stop smoking. Behavioural support assists people who have already stopped smoking. In the UK, we have a network of smoking cessation services, but the rates of advice to stop smoking given by GPs are low and more than 90% of quit attempts do not use this support.6 The test for MI is whether teaching GPs these skills could change this.
- © British Journal of General Practice, 2007.
