This trial has demonstrated several methodological strengths. This is the first and largest study to examine the effect of using introductory taster sessions to encourage people to attend a smoking cessation service. We recruited 4383 people from 18 regions across England. The collection of anonymised data of those
invited but who did not agree to take part in the study allowed us establish that the representativeness of the sample was acceptable.
Another major strength of the study was that for the main outcome we were able to collect complete data from SSS records. Although we were reliant on the accuracy of reporting from the SSSs, this represents a significant improvement on the self-report outcome reported by Murrayet al.,29which relied
on self-reporting and, in addition, had a low follow-up completion rate of 48%. We demonstrated a discrepancy between recalled attendance and validated attendance. Thus, our study represents a much more accurate outcome, not affected by poor recall or misunderstanding of what constituted attendance at the SSS, and not over-reported because of the social desirability response bias of the intervention group, which would involve wanting to report higher attendance.
A further important strength was that smoking outcome data were collected, allowing us to define the ultimate efficacy in terms of abstinence, rather than only at the attendance point. Although abstinence could be estimated based on quit rates within SSSs, different definitions of abstinence have been used, which do not allow comparison with other smoking cessation studies. Furthermore, it allowed us to show that smokers recruited proactively to the SSS can have outcomes as successful as self-referred clients. By randomising at the level of participant rather than by practice, there was a slight risk of contamination by communication between patients at the same practice allocated to different intervention groups. We built into the trial safeguards against contamination by ensuring that only one person from the same household received a screening questionnaire and by monitoring attendance at the taster sessions, to ensure that anyone attending who had not received an invitation was recorded and checked against participants in the control group. We also kept a record of attendance at the taster sessions and measured the amount of contamination at follow-up by asking participants who had not attended a taster session whether or not they personally knew or had spoken to anyone else who had been invited or attended to a taster session. The results suggested that few control participants had access to information from the taster sessions and, thus, there was no contamination bias.
Finally, an assessment of the fidelity of the delivery of taster sessions to the protocol was embedded into the trial. In general, adherence to the protocol was high, although variable. But adherence to the protocol was not related to the main outcome of attendance at the SSS or to abstinence from smoking. The first part of the intervention consisting of the personal risk letter was tailored to the individual’s personal characteristics by computer and, therefore, the format was standard for all participants, and all advisors had received standardised training to deliver the taster sessions. Thus, we can be sure that the outcome is a result of the intervention and not a result of other non-specified variables.
There were also some limitations to the study, in addition to the difficulty of dismantling the two components of the intervention, which has been already discussed.
Although the recruited SSSs were spread across England, were located in both high and low areas of deprivation and represented both large and small organisations, we did include only a sample of areas. Eighteen of the 151 SSSs were included, and these participating SSSs may not necessarily be representative of all SSSs in England. It is likely that those agreeing to participate were the more organised and
enterprising ones. That is not to say, however, that the less organised SSSs would be less successful in encouraging uptake were they to introduce these procedures and offer introductory sessions to encourage attendance. Recruitment took place at a time when SSSs were undergoing changes in commissioning and we do not know how this affected the decision to take part or not.
In addition, although our proactive recruitment strategy was a strength of the study, the recruitment rate was low, at 4%. As a result, we recruited only a small proportion of smokers in each area. Comparison of some demographics confirmed that the sample was reasonably representative of smokers in terms of gender and deprivation, but we failed to recruit sufficient younger smokers. It is important to attract this
population as attendance at the SSS at present tends to be concentrated in older age groups. There was a large difference in response from different SSS areas. Although we aimed to target areas with a high number of ethnic minorities, the response rate in these areas was particularly low and, thus, in areas with a very low response rate the generalisability of the results may be reduced.
We had initially hypothesised that the intervention’s effectiveness might differ by socioeconomic group and analysis of the interaction between the intervention and deprivation was prespecified. Before any analysis began, we extended our planned investigation to investigate interactions of intervention with age group and gender also. These were the only prespecified analyses of interactions.
Observational data were used for the primary outcome rather than subjective self-reported data. However, the collection of these outcome data depended on accurate recording and submission by local SSS
collaborators, and on the collaborators being blind to group assignment. There may have been some cases in which loss of blinding occurred, and SSS advisors knew to which group the client had been assigned, leading to bias in treatment.
Finally, assumptions were made for the long-term CEA. To maintain consistency with the Markov model we adopted to generate lifetime cost savings from the interventions, participants’smoking status was assumed to stay unchanged after the trial period. Specifically, we assumed that those participants who managed to quit at the end of the trial will stay abstinent from smoking for the rest of their life, and smokers who have not given up will carry on smoking until they die. However, in real life, many quitters may relapse to smoking again and some smokers may quit smoking without any aid at some point. Therefore, further research with longer follow-up periods or models that allow for smoking relapse and spontaneous abstinence are recommended.