Physical activity as an aid to smoking cessation during pregnancy (LEAP) trial: study protocol for a randomized controlled trial

The aim of this paper is to describe the protocol for an RCT designed to evaluate an intervention for increasing PA as an aid for smoking cessation during pregnancy.

The London Exercise And Pregnant smokers (LEAP) study is a pragmatic two-arm RCT into which pregnant women who smoke are recruited from antenatal clinics in National Health Service (NHS) trust hospitals (see Figure1 for a flowchart of the LEAP study design). The study will compare the effectiveness on smoking cessation of individual behavioral support for smoking cessation plus a PA intervention, relative to individual behavioral support alone at the end of pregnancy.

Ethical approval for the study was given by Wandsworth Research Ethics Committee and research governance approval was obtained from each of the 13 hospital trusts recruiting to the study. Each participant is to provide written informed consent.

Exercise trials for smoking cessation often target those who report smoking 5 to 10 or more cigarettes a day. However, many pregnant smokers under-report the number of cigarettes they smoke[29], therefore, women are eligible if they report currently smoking at least one cigarette a day and report smoking at least five cigarettes a day before pregnancy. Participants are eligible if they are aged 16 to 50 years, 10–24 weeks pregnant (subject to confirmation that they have had a scan to show a viable pregnancy), prepared to quit smoking 1 week after enrolment, and confirm that they are able to walk continuously for at least 15 minutes. Women are excluded if they are unable to complete self-administered questionnaires in English (because of lack of resources for translators) or if they report having any medical condition that might be exacerbated by exercise. There are no documented contraindications to moderate-intensity exercise, but if the women have been advised by their doctor or midwife not to take exercise during pregnancy, if they have any complications during their pregnancy, or if they have any cautions for taking exercise[24, 30], a consultant obstetrician and gynecologist at their hospital will be consulted to check that it is safe for them to participate. Participants joining the trial are monitored at each treatment session for cautions to exercise and for adverse events. Those with drug or alcohol dependence will also be excluded as the intervention described here is not sufficiently comprehensive to address the problems of women who have drug or alcohol dependence.

Although NRT is licensed for use in pregnancy, there is mixed evidence on its effectiveness in helping women quit smoking during pregnancy[15], and the most robust trial to date has found no evidence that it is effective[31]. In addition, many pregnant smokers prefer not to use NRT[16]. Allowing participants in the trial to use NRT might create confounding, and therefore, women who indicate that they wish to use NRT at the outset will be excluded. Our pilot work shows that pregnant smokers are willing to attempt to quit without using NRT as part of a research study[28]. Following the guidelines[17], those women who are unable to stop smoking after their quit day, and who express a clear wish to receive NRT, will be prescribed NRT by their general practitioner (GP). Use of NRT will be monitored throughout the intervention period with the following questions:

In addition, at the end of pregnancy follow-up participants will be asked the following questions:

The subjects’ GPs, midwives, and obstetricians will be informed of their patient’s participation in the trial.

A Cochrane review suggests that approximately 9% of women who are still smoking at the time of their first antenatal visit will stop smoking with usual care through to the end of their pregnancy, and a further 6% will stop as a result of a smoking cessation program using individual behavioral support[18]. Thus, in our control group we expect a smoking cessation rate of around 15% at the end of pregnancy. Combining our pilot studies, 25% (8/32) of participants in the treatment group sustained continuous smoking abstinence to the end of pregnancy. Therefore, in the trial we conservatively estimate an abstinence rate of 23% at the end of pregnancy in the treatment group, which would be similar to the effect shown for NRT with non-pregnant smokers[14]. We aim to recruit 433 women to each arm to detect the above absolute difference (8%) in smoking cessation rates between the groups at end of pregnancy with a two-sided significance level of 5% and power of 83%. This calculation is based on a χ² test with Yates correction.

Recruitment for the trial is taking place over a 44-month period, between April 2009 and November 2012, in 13 NHS hospital trusts in London and the south-east of England. During the trial, smoking status for all pregnant women will be recorded in the hospital computerized patient administration system at the first antenatal booking visit, which is typically at 9–14 weeks of gestation. At this time, the midwife will inform all women recorded as smokers that it is the hospital’s policy to telephone them to offer smoking-cessation support. This support would usually be offered by the primary care trust (PCT) stop-smoking advisor, but during the period of recruitment to the study, a trial researcher will telephone the women. Those who are interested in receiving help with quitting will be invited to join the trial or be referred to the PCT. Information for the first antenatal appointment will be accompanied by a flyer inviting women who smoke to join the study. Those women expressing interest in volunteering will be screened for eligibility by a telephone interview, and eligible women will be sent a patient information sheet. As an incentive for recruitment, and to increase attendance rates, all women attending at least 80% of their treatment sessions will be entered into an annual lottery with three annual prizes of £100 shopping vouchers.

At the first visit, eligible and consenting patients will be individually randomized to one of the two treatment arms based on a computer-generated pseudo-random code using random permuted blocks of randomly varying size, created by the Nottingham University Clinical Trials Unit (CTU) and held on a secure server. The randomization will be stratified by center, and block randomization will be used to ensure good balance between intervention and control within centers and thus minimize the risk of any ‘center effect’ influencing the outcome. Access to the sequence will be confined to the CTU data manager. Allocation to treatment arms will be in the ratio 1:1, and investigators will access the treatment allocation for each participant by means of a remote, secure internet-based randomization system developed and maintained by the CTU. Allocation will be concealed from the patient until they have completed all baseline assessments. The sequence of treatment allocations will be concealed until interventions have all been assigned and recruitment, data collection, and laboratory analyses are complete.

The interventions (Table1, Table2) follow CONSORT guidelines for non-pharmacologic interventions[32, 33]. The same therapists, including research midwives, nurses, and psychologists, will deliver both the control and exercise interventions. Two therapists operate at three of the recruitment sites, and the remaining 10 sites have been assigned a single therapist. Delivery of the interventions will be standardized by training and by the therapists following the supplied manuals (see Additional file1, see Additional file2). The initial competence of the therapists will be assessed by the trial manager observing role-play scenarios during training. The fidelity of the interventions will be checked during the first 6 months by regular observations (at least five intervention sessions) by the trial manager. All sessions will be face-to-face and one-to-one, and will be delivered in a private room at the hospital or in a community health center. Social cognitive (learning) theory[34] is the theoretical basis for the interventions. This theory recognizes the interplay of individual factors (for example, self-efficacy to quit smoking or increase PA) and social/environmental factors (for example, social support) on health behavior change. For each session that they attend, the women are paid £7 for their travel expenses.

At baseline, demographic characteristics will be recorded, including age, marital status, number of children, highest educational qualification, ethnicity, occupation, weeks of gestation, and history of premature births. Smoking history will also be recorded, including cigarettes smoked per day (now and before pregnancy), urge to smoke[45, 46], tobacco withdrawal symptoms[45, 46], and Fagerström Test for Nicotine Dependence score[47], as well as the partner's smoking status. Depression will be assessed with the 10-item Edinburgh Postnatal Depression Scale (EPDS)[48]. Physical activity levels in the previous week will be assessed for both groups using the 7-day Physical Activity Interview[49]. The woman’s confidence about taking up regular PA[50] and stopping smoking[51] will also be reported.

From a health-service perspective, we will examine the cost-effectiveness of smoking -cessation support plus exercise relative to smoking-cessation support alone. We will document resources consumed that are related to each intervention. Relevant resources include: cost of personnel, materials, space, equipment, and administrative overheads. Data collection methods include: (i) accounting for staff time using time and effort reports, (ii) accounting for computer time, mailing, and program costs using an accounting system that has been created to facilitate real-time aggregation of these costs, (iii) and using interviews with staff to determine the amount of time they devote to tasks related to the program. The benefits of smoking cessation in pregnancy will be modeled and not measured. Long-term cessation leads to long-term benefits to the mother, which must be discounted appropriately. In ‘normal quitters’ who sustain abstinence for about 6 months, meta-analyses show that half will relapse, while half will become lifetime abstainers and enjoy these benefits[62, 63]. Based on this, we might expect that most women who stop smoking in pregnancy, usually for at least 6 months, would maintain lifetime abstinence. However, a much higher proportion of women who stop smoking return to it after pregnancy than would be expected in these models, diminishing the benefits to the woman herself from smoking cessation. However, the benefits of smoking cessation in pregnancy also extend to the fetus, which will be less likely to be born pre-term[18]. This leads to reduced NHS costs and a lower likelihood of complications, such as cerebral palsy and similar health problems. Avoidance of these latter health problems leads to gains in quality-adjusted life years for the baby that must also be included in the utility measure of the decision analytic model. Thus we will use an existing method that has been used for cost-effectiveness modeling for the National Institute for Clinical Excellence (NICE), updating utility estimates to include the benefits to the fetus. The evidence for this model in measuring the benefits to the fetus and child will be derived from a systematic review of the effects of smoking cessation in pregnancy on the consequences for the fetus and child that is currently being conducted by a team at the University of Nottingham, led by Professor Tim Coleman (co-investigator). The analysis will allow us to calculate incremental cost-effectiveness ratios for smoking cessation support plus exercise compared with smoking cessation support alone in terms of cost of treatments per smoker who quits. We will also be able to examine ratios for subgroups; for example, according to nicotine dependence or age at baseline.

We will conduct a descriptive comparison of the baseline characteristics of the two treatment groups. Analysis will be on an intention-to-treat (ITT) basis. Participants who, for any reason, have missing outcome data on the primary outcome or any secondary smoking outcomes, will be assumed to have resumed smoking[55]. Our primary outcome measure, continuous abstinence from smoking from quit date to end of pregnancy, will be compared between treatment groups using logistic regression, adjusted for recruitment center only, with statistical significance determined by the likelihood-ratio test and with the estimate of effect given as the odds ratio and 95% confidence interval. Our primary analysis will not adjust for any further variables because effect estimates can be sensitive to decisions concerning what variables should be adjusted for and how these are specified. However, we will conduct secondary analyses adjusting for the following variables that are predicted to be related to the outcome: education, nicotine dependence, age, and depression. We will analyze other binary smoking outcomes in a similar way.

We will compare secondary outcomes, including urge to smoke, withdrawal symptoms, self-confidence, and PA, in the first week of abstinence, and the same variables plus maternal gestation weight and depression, over subsequent time points, using mixed effects modeling to allow for repeated measures, with adjustment for center. To deal with non-normally distributed variables we will use transformations to normality, residual bootstrapping, or dichotomizing. Differences between groups in perinatal outcomes, including birth weight, gestation, mode of delivery, and complications, will be analyzed by linear or logistic regression, with adjustment for center. For fetal outcomes, the primary analysis will be of singleton births and we will carry out a sensitivity analysis including multiple births, allowing for the clustering of outcomes. If we observe differences between the two groups in use of NRT, or use of behavioral support outside of the intervention sessions, then we will conduct a sensitivity analysis to examine the effect of controlling for any differences between groups in these variables.

We will determine the quantity and distributions of missing data. We will carry out a complete case analysis, and we will compare this with an analysis using multiple imputation to deal with missing values, which assumes data are missing at random, describing any differences in terms of the biases in the data. The exception is smoking outcomes, where those with missing data will be assumed to have resumed smoking[55].

If the intervention is effective, we will use mediation analysis to examine whether there is evidence that the change in PA levels is the likely causal factor in determining smoking abstinence. We will examine the association between treatment group and change in PA levels, and the association between level of PA and abstinence. Finally, we will include level of PA in a logistic regression model of the association between treatment group and abstinence, with mediation assessed using MacKinnon’s causal steps criteria[64].

Since trial enrollment began on 1 April 2009, the following important changes to the methods have been made.

1)For the follow-up at end of pregnancy, the valid period for assessment was originally defined as 38 weeks gestation to 2 weeks after the birth. Because there were a number of women who could not be contacted to be followed up during this time frame, the valid period was extended to 36 weeks gestation to 4 weeks after the birth (approved by the research ethics committee on 18 May 2010). However, because there were still a few women being followed up later than 4 weeks after the birth, the valid period was further revised to 36 weeks gestation to 10 weeks after the birth (approved 31 Jan 2012). The aim remains to attempt to follow-up as many women as possible within 2 weeks of the birth.

2)To give the women an incentive to complete the follow-ups at end of pregnancy and 6 months after the birth, all women who complete these follow-ups will be given a £10 shopping voucher for each of the follow-up sessions attended (approved 18 May 2010).

3)Originally, to be eligible women had to report smoking at least 10 cigarettes a day before their pregnancy. We found that a good number of women reported smoking five to nine cigarettes a day at this time. Therefore, we extended the eligibility criteria to include women smoking at least five cigarettes a day before pregnancy (approved 15 September 2010). These women are still likely to be dependent on smoking, as there is evidence that women who say they were smoking five to nine cigarettes before pregnancy are back to smoking 14 cigarettes a day at 18 months post-partum[65].

4)Initially, women had to be between 12 and 24 weeks' gestation to be eligible for the trial. However, since the trial began, most of the hospital trusts began offering earlier antenatal booking appointments (before 12 weeks gestation), and because we wished to recruit the women as early as possible in pregnancy, we revised this to 10 to 24 weeks gestation (approved 31 Jan 2012).