Methods

Design

This was an open-label, multicentre, pragmatic superiority trial in which participants were randomised 1 : 1 to non-use or use of a nicotine patch for 4 weeks prior to quit day. Participants used standard pharmacotherapy and behavioural support for stopping smoking thereafter. The primary outcome was prolonged biochemically validated abstinence, measured 6 months after quitting. The protocol was published26 and implemented with one change, which related to participants who had moved from the area during follow-up and were not available to attend in person for carbon monoxide (CO) measurement to confirm abstinence. We, therefore, asked such participants to provide a saliva sample and measured cotinine or anabasine concentrations to confirm abstinence from smoking. Anabasine is a tobacco-specific alkaloid that will indicate smoking abstinence in a person who is using NRT or e-cigarettes. Cotinine is a metabolite of nicotine with a much longer half-life, making it a good indicator of nicotine consumption within the last week, and will be in low concentrations in people who do not smoke or use nicotine.

In the original proposal to the National Institute for Health Research (NIHR), we did not suggest collecting genetic samples or assessing the impact of quitting on weight gain. However, these additional non-trial-related hypotheses were added to the published protocol.26 In the event, it turned out to be impractical to collect genetic samples and this ambition was abandoned. The observational analyses of the impact of quitting and failed quit attempts on weight gain will be reported separately.

Participants and settings

In three recruitment centres, Birmingham, Bristol and Nottingham, general practitioners (GPs) spoke to, wrote to, e-mailed or texted patients listed as smokers on the electronic health record. GPs encouraged their patients who wished to quit smoking with the help of pharmacotherapy and behavioural support to enrol in the trial. The final centre, London, was an existing NHS smoking cessation clinic, which asked participants seeking treatment if they would also like to take part in the trial. Thus, in all centres, patients were recruited from those who sought treatment at their own instigation, by referral from their GP or by advertising in various online and print media.

Potential participants telephoned the research team to learn more about the trial and were screened for eligibility and entered into our online trial database. If a potential participant appeared to be eligible and wanted to participate, we made an appointment and sent out the participant information sheet. At this initial appointment, we again described the trial and obtained witnessed consent with a signature and confirmed eligibility. The researcher, typically a research nurse, saw participants at their own general practice or at the London centre. We included people who were:

• regular smokers of cigarettes, cigars and/or roll-up tobacco cigarettes, with or without marijuana, aged ≥ 18 years
• suitable for preloading in the judgement of the researcher (see below)
• seeking support to stop smoking from the NHS Stop Smoking Service
• willing to set a quit day in 4 weeks
• able to understand and willing to adhere to study procedures.

We excluded:

• pregnant or breastfeeding women
• people with extensive dermatitis or another skin disorder that precluded patch use
• people who had acute coronary syndrome or who had had a stroke in the past 3 weeks
• people with an active phaeochromocytoma or uncontrolled hyperthyroidism that would increase the risk of arrhythmias from the nicotine patch.

We judged if people would be suitable for preloading by assessing whether or not they were addicted to smoking. This was based on a clinical judgement of the following factors, with no specific cut-off points used:

• time to first cigarette in the morning, with earlier use reflecting a higher level of addiction
• number of cigarettes smoked per day, with a greater number reflecting a higher level of addiction
• higher levels of exhaled CO, which reflect a higher level of addiction
• failure of previous quit attempts despite use of appropriate pharmacotherapy.

Interventions

Intervention group

In the intervention group, participants wore a 21-mg nicotine patch daily for an intended 4 weeks prior to quit day. The intention was that this would be worn for 24 hours per day. However, any participants who had experienced previous AEs at night from 24-hour patch use were advised to wear the patch during waking hours only.

We asked participants to smoke as normal and not reduce nicotine consumption. Reducing consumption would probably lower the nicotine concentration in the blood, which could have resulted in cigarettes being more rewarding and, thus, have undermined the supposed benefits of preloading.27

We referred patients to the NHS Stop Smoking Service to set a quit day, obtain behavioural support and receive a prescription of pharmacotherapy to support cessation. Although we aimed for this quit day to be 4 weeks from the start of preloading, in keeping with the pragmatic design, we asked participants and NHS clinics to set this date within a window from 3 to 5 weeks after starting preloading. Preloading was allowed to continue for up to 8 weeks in exceptional circumstances. Preloading was also allowed to restart once, if necessary, for example if this was interrupted because a participant was admitted to hospital. In such cases, participants aimed to complete 3–5 weeks of preloading from the date of recommencement. We dispensed NRT in two lots, one of 2 weeks and a second, 1 week later, of 3 weeks. We made arrangements to dispense additional lots of NRT if preloading was extended.

The researchers explained the rationale of preloading and prompted action planning to maximise adherence to the patches, that is, they talked through participants’ daily routine and assisted them to use it and environmental cues to minimise the chance of forgetting to use the patches. They provided reassurance about concerns that participants may have about side effects and how to manage them. A booklet containing this information was supplied to participants. We aimed to see participants 1 week after commencing preloading. At this visit, we asked participants about their experiences and their understanding of the necessity of preloading and again addressed these beliefs as well as concerns about preloading.

We offered participants lower-strength patches from the beginning if they reported experiencing adverse reactions to the 21-mg patch or if, during the treatment course, they experienced symptoms of nicotine overdose such as nausea, salivation and/or pounding heartbeat or other symptoms that they believed were due to the patch strength. We stopped preloading if a participant requested it, if it was not possible to alleviate AEs by reducing the dose or if an intervening health state or contraindication to preloading emerged.

Control group

The research team were keen to offer participants in the control group a placebo, but the NIHR Health Technology Assessment (HTA) Board would not allow this. We were concerned that offering no treatment would lead to disengagement and, to counteract this, we developed a behavioural intervention that could seem plausible to participants but had no evidence that it would increase abstinence. We asked participants to consider their smoking pattern and the triggers for particular cigarettes (e.g. the first cigarette of the day, the first after dinner) and to plan ways to reduce these cues. Such a process is standard in smoking cessation support, so participants in the intervention group were likely to engage in this after preloading and in preparation for quitting. The control group received a booklet outlining this process, which was designed to be comparable to the booklet supplied to the intervention group. As in the intervention group, participants in the control group were referred to the NHS Stop Smoking Service to commence a quit attempt between 3 and 5 weeks after enrolment.

Standard smoking cessation treatment common to both arms

In both arms, at the first and second contacts, we facilitated contact between participants and the local NHS Stop Smoking Service and wrote a referral letter to encourage the NHS Stop Smoking Service to work with us on encouraging participants to continue preloading. In particular, we asked the NHS Stop Smoking Service advisors to ignore the presence or absence of preloading when discussing and recommending the use of pharmacotherapy to support the quit attempt. We especially asked advisors to feel free to use varenicline, which starts prior to quit day and, hence, would be used concurrently with nicotine preloading. This is because we were keen that the treatment provided by the NHS Stop Smoking Services did not differ between trial arms. NICE guidance has specifically recommended against this combination of NRT and varenicline,28 which advisors wrongly assumed was because of safety concerns about concurrent use. We therefore addressed this in the referral letter, by telephone and in face-to-face discussions with the NHS Stop Smoking Service.

The NHS trains Stop Smoking Service advisors to give weekly behavioural support starting 1–2 weeks prior to quit day and continuing until at least 4 weeks after quit day. This support addresses issues such as planning for the quit day, the ‘not a puff’ rule and how to deal with difficult situations, such as others smoking around the person who has quit. It also provides monitoring of behaviour and validation of abstinence through CO testing. The support is largely withdrawal-orientated therapy.29

Our protocol26 allowed all other medication to be used concurrently with preloading.

Outcomes

We followed up participants on five occasions for outcome assessment. These were named in relation to quit day and are shown in Table 1.

The first follow-up appointment was a face-to-face appointment, when possible, and occurred 1 week after baseline and 3 weeks before quit day. One week after quit day we telephoned participants to collect data on AEs, adherence to preloading and use of, and adherence to, other smoking cessation pharmacotherapy. We obtained data on smoking cessation from the NHS Stop Smoking Service at 4 weeks or, failing this, from a telephone call to the participant. At 6 and 12 months, we telephoned participants to obtain data on smoking status and health service use. We invited participants who claimed to be abstinent for at least 1 week to a meeting where we asked them to provide an exhaled CO concentration reading. Participants were compensated £15 for their time for attending this meeting. Therefore, if they attended a meeting at both 6 and 12 months’ follow-up, participants were compensated a total of £30.

Primary outcome

The primary outcome was 6-month prolonged abstinence, defined according to the Russell Standard criteria.30 This allows a grace period of 2 weeks following quit day when lapses do not count against abstinence. Thereafter, we counted a person as abstinent if they smoked fewer than five cigarettes or cigars up to the 6-month assessment and were biochemically confirmed abstinent by an exhaled CO concentration of < 10 parts per million (p.p.m.).

Secondary outcomes

The secondary outcomes were 12-month Russell Standard abstinence and biochemically confirmed 7-day point prevalence abstinence at 4 weeks, 6 months and 12 months.

Adverse events

We defined AEs as those that occurred after baseline and until 1 week after quit day, thereby covering the period of preloading and an additional week for AEs to emerge. This excluded planned events, such as scheduled surgery. In the protocol,26 we deemed that we would report only AEs of moderate severity or above, that is, those events that hindered or prevented the person going about their normal activities. This is because these events matter more to patients and because NRT has been used for many years and so many common minor AEs are known already.

We assessed the AEs arising from preloading in two ways. First, we elicited AEs from participants in both arms at the contacts at –3 weeks (1 week after baseline) and +1 week (5 weeks after baseline and 1 week after quit day). Spontaneously occurring AEs were coded at the level of preferred term and system organ class using the MedDRA (Medical Dictionary for Regulatory Activities) coding system.31 However, spontaneously occurring reports may introduce bias. It seemed illogical to participants in the control group and to treating staff to discuss AEs when the participant was taking no medication at all. Therefore, second, at 1 week after baseline we also gave all participants a questionnaire concerning symptoms of nicotine overdose in the previous 24 hours (such as nausea and excessive salivation).

We defined serious adverse events (SAEs) as those leading to hospitalisation, death, permanent disability or congenital abnormality and reported these separately. We obtained the medical history of participants experiencing SAEs and these were classified by an independent assessment panel blinded to allocation as unrelated or unlikely to be, possibly, probably or definitely related to the use of nicotine patches.

Sample size

We determined the sample size from plausible estimates of 6-month abstinence. We estimated that 15% of participants in the control group would attain 6-month abstinence based on data from similar trials.17,32,33 We felt that a RR of 1.4 was plausible based on our meta-analysis12 and would make preloading valuable to NHS Stop Smoking Service commissioners.12 This gave us a sample size of 893 per group, or 1786 in total, to achieve 90% power (Table 2).

Randomisation

The research team randomised participants at the baseline visit following consent, assessment of eligibility and enrolment. An independent statistician used Stata^® 14.2 (StataCorp LP, College Station, TX, USA) to generate a randomisation list, stratified by treatment centre and using randomly permuted blocks of varying sizes, using a 1 : 1 ratio. This sequence was incorporated into an online database and the sequence remained concealed from all research staff.

Blinding

This was an open-label trial so participants, research staff and NHS Stop Smoking Service personnel knew to which group participants were assigned. Blinded follow-up was impossible as all staff had been involved in recruitment.

Statistical analysis

The analysis of all cessation outcomes was performed using the intention-to-treat (ITT) principle as outlined in the Russell Standard.30 Thus, everyone was included in the denominator and presumed to be smoking if this information remained unknown. This presumption was shown to be true in one trial that went to extraordinary lengths to establish the true status of those who did not respond to standard follow-up.30 For the primary analysis, we calculated adjusted odds ratios (ORs) using multivariable logistic regression in Stata 14.2. We also calculated the percentage achieving abstinence, the risk difference (RD), RRs and 95% CIs using the post-estimation adjrr procedure in Stata 14.2. As the randomisation was stratified by centre, we adjusted for this covariate in the analysis and this was considered the primary analysis. In sensitivity analyses, we adjusted for two well-known predictors of abstinence to improve precision: the longest previous abstinence and the degree of addiction measured by the strength of urges to smoke at baseline.34–36 Finally, because we envisaged that the use of NRT would deter NHS Stop Smoking advisors from prescribing varenicline, which is more effective than other pharmacotherapy,6 we also adjusted for varenicline use.

We had pre-planned subgroup analyses that were assessed by including multiplicative interaction terms in the calculations described in the previous paragraph. The presumed mechanism of action of preloading suggests that preloading could be more effective in people who are more addicted to nicotine. This was assessed by investigating whether or not outcomes varied with baseline Fagerström Test for Cigarette Dependence (FTCD) score or exhaled CO concentration readings and is reported in Chapter 3. In another analysis, we examined whether or not the effect of preloading was less pronounced in people who used varenicline in the pre-quit period. This is because varenicline includes a period of use of at least 1 week prior to quit day and this appears to have similar effects to preloading.21

The population in whom we analysed the occurrence of AEs was all those who provided data on such events. The analysis of spontaneously elicited AEs was confined to events of moderate severity and above. We anticipated that most minor AEs would relate to well-established adverse reactions to nicotine patches; confining the analysis to AEs of moderate severity and above may identify novel adverse reactions to preloading specifically that were of more concern to patients. The analysis used analogous statistical models to those applied for the primary and secondary outcomes.

Design features classified by PRECIS-2

Trials vary on an explanatory to pragmatic continuum; this can be assessed and represented by a tool called PRECIS-2 (PRagmatic Explanatory Continuum Indicator Summary).37 The aim of this trial was to examine the effectiveness of preloading in everyday clinical practice. In the UK, this was in the context of the NHS Stop Smoking Service. However, the aim was also to examine AEs and safety, which meant that these were assessed by researchers trained to assess these outcomes appropriately. Furthermore, NHS Stop Smoking Services do not routinely follow up people who try to stop smoking beyond 4 weeks, which meant that we instituted special follow-up measures to assess smoking abstinence. It was because of these aims that the trial design scored less than the maximum for pragmatic design features on the PRECIS-2 wheel (Figure 2).

FIGURE 2 The PRECIS-2 domains. Created using http://precis-2.org.38 (Figure not reproduced here; see the original article.)