- Top of page
- Supporting Information
Smoking is a major public health problem and the leading preventable cause of premature death; 23% of all UK deaths in middle age are attributed to smoking . Despite this 22% of the population in the UK still smoke . Although clinical specialist services offering individual face-to-face support are provided nationwide by the National Health Service (NHS), only around 5% of smokers use them [3, 4]. There is an urgent need to develop self-help interventions that will reach a wider range of smokers who are not able or willing to engage with intensive clinical services.
Traditional self-help materials can be offered to any smoker, but are limited by generic information that does not meet the needs of a diverse population of smokers . Computer-based systems, however, can generate highly tailored advice reports, which apply the basic principles of behavioural interventions used in clinical practice , but can be inexpensively produced and delivered on a larger scale, with the potential to reach most smokers. Previous trials have produced mixed results, but, overall, demonstrate a small, but useful, effect of individually tailored self-help materials on smoking cessation .
Tailored self-help materials can address other important issues that can lead to suboptimal quit rates. Written advice, often written at a level beyond the literacy skills of many smokers , should be tailored to take account of the education and reading level of the individual. Additionally, with approximately 60–70% of smokers having no serious intentions to attempt to quit in the next year [9-11], further tailoring of messages to match and try to enhance the level of the smoker's motivation to quit is essential. Multiple mailings of tailored advice has also been shown to be more effective than a single mailing in promoting cognitive change in smokers with low readiness to quit , and stored data can be used at a later date in combination with new information provided by the smoker to update subsequent advice reports. We applied these principles to adapt an existing computer-based system, originally developed to generate individually tailored advice reports as an adjunct to telephone counselling , and developed a stand-alone brief intervention for use in primary care. We identified smokers from general practitioner (GP) lists, and recruited them proactively  to a randomised controlled trial designed to evaluate the effect of adding advice reports, tailored to relevant participant characteristics, including motivation to quit and levels of reading ability, to a generic self-help booklet, on smoking abstinence for at least three months assessed six months later.
- Top of page
- Supporting Information
ESCAPE aimed to assess the effectiveness of a brief intervention consisting of tailored advice reports when sent to a representative sample of smokers across the UK. Quit rates for the primary outcome of three months of prolonged abstinence were not significantly different between study groups. Thus, the intervention showed no effect. Quit rates in a number of different outcome measures of abstinence also showed no significant intervention effect. However, all outcome measures showed a non-significant trend toward more abstinence in the intervention group.
This trial had been designed to detect an OR of 1.42, consistent with previous findings for similar interventions . The achieved sample size of 6697 allowed for 74% power only to detect such an effect of tailored material. If an OR of 1.20, as observed in this study, represents the true effect, the sample would have needed to exceed 30 000 participants. The CI around the estimate of effect size suggests that it is unlikely that the intervention has a large effect (OR 95% CI upper limit = 1.54). However, it does not provide enough evidence to rule out a small positive effect. The real intervention effect is more likely to be around OR = 1.20 than to be entirely absent. This is equivalent to an absolute difference in quit rates of 0.5%. The potential relevance of the intervention would depend on the minimum effect that could be considered relevant given the ease with which the intervention could be implemented. Some clinicians might find this small number of additional quitters in their practice worthwhile. However, the high scalability of this intervention implies a potentially valuable effect on public health; for example, if the intervention reached 100 000 smokers, we would expect 500 additional quitters.
For our primary outcome we used three months of prolonged abstinence, defined as a period of sustained abstinence during a time window immediately preceding the follow-up, and beginning at any time between the baseline screening and the follow-up. This was appropriate for a ‘cessation induction trial’, designed to increase the probability of quit attempts during a period , and can capture the delayed treatment effects that can occur using tailored communications, particularly in studies including early stage quitters [5, 23]. However, a single primary outcome may not make the best use of the data, thus it is important to include a range of measures which assess other possible benefits of treatment. Prompting quitting activity in smokers not motivated to quit is a valuable effect of an intervention, as this activity has the potential to eventually end with cessation. Quit attempts and the longest duration of abstinence during the follow-up can predict successful future cessation . Post hoc descriptive analysis of self-reported quit attempts in non-quitters showed increases in the intervention over the control group in both the number and length of quit attempts, suggesting that the tailored intervention was stimulating cessation behaviour.
This intervention was designed to reach a wider population of smokers. A strength of this study is that by using proactive recruitment, and employing less stringent selection criteria with few exclusion categories, we enrolled a ‘broad and representative’ sample , which was largely representative of the smoking population of the UK in terms of motivation and socioeconomic status . Further, by recruiting a diverse population of smokers, including low socioeconomic groups and those not motivated to quit, we increased the variation and external validity of our sample . We would therefore expect lower quit rates than trials recruiting a more motivated population . Our results are largely consistent with similar general practice-based studies of tailoring in the UK [28, 29] which enrolled a high proportion of smokers not ready to quit, and reported low overall cessation rates with small increases in the tailored letter group.
A novel feature of our intervention was that the advice reports were adapted to fit the reading ability of the recipient. While it is acknowledged that the best expert systems should be grounded in a strong theoretical basis , research also points to the need to tailor communications to ‘broad spectrum’ general recipient characteristics  of which ability to read and process information is one. Our results reflected the trend for the use of self-help materials to be less successful in lower educated groups . However, a non-significant trend for our adapted intervention to be more effective in this subgroup suggests that these harder to reach groups may be receptive to help and encouragement to quit, and demonstrate the need for further work in this area.
It is not possible, in a pragmatic trial such as this, to separate out all of the components of the intervention to measure the effect of each. Thus, possible limitations of the study are that other factors could account for the modest increase in quit rates in the intervention group. These confounding factors include the additional contact received by the intervention group in the form of an accompanying GP letter with the tailored advice report and additional assessment. Another possibility is that, as participants could not be blinded to condition, the pre-trial information provided had a positive effect on quitting in smokers randomised to the intervention group. However, this is unlikely as minimum information was given to the participants regarding the nature of the intervention. An alternative explanation for the lack of a significant result is that, as all participants received the baseline questionnaire as part of the trial assessment, completing this questionnaire alone may have prompted a change in behaviour and prompted quitting in both groups .
Our detailed, but brief, self-help intervention produced a modest increase in quit attempts and a small number of extra quitters. While our study was not powered to detect such a small effect, from a public health perspective these small increases are important and could make a valuable contribution to lowering smoking prevalence.
Clinical trial registration
Current Controlled Trials ISRCTN05385712.