Brief intervention during hospital admission to help patients to give up smoking after myocardial infarction and bypass surgery: randomised controlled trial

Participants

The study took place in 17 hospitals in England. Patients admitted after myocardial infarction or for coronary bypass surgery were screened for eligibility. Participants were current smokers or those who had recently stopped smoking. All patients had recovered enough to receive the intervention, had no gross memory impairment, were under 76 years of age, could read English, had not smoked at all since admission to hospital, and were motivated to stop smoking permanently.

Sample size

With usual care we would expect 30-40% of participants not to start smoking again for at least 12 months. A sample of 244 participants in each group would give a 90% probability of detecting a 15% increase in abstinence rates and 80% probability of detecting a 13% difference (α=0.05, two tailed test). We estimated that up to 10% of the sample would be lost to follow up because of death or moving to an unknown address, and therefore we aimed to recruit 540 participants.

Procedures

The cardiac rehabilitation nurses recruited eligible patients on wards. Participants signed the consent form, filled in the study questionnaire, and were randomised to the intervention or control group on a 1:1 ratio by nurses opening a serially numbered, opaque, sealed envelope designating the patient's allocation. The randomisation was in batches of 20 with envelopes provided to each nurse.

All participants in the control group received verbal advice to remain abstinent and a British Heart Foundation booklet Smoking and your Heart. Participants in the intervention group also received a carbon monoxide reading to show the health benefits of quitting; a booklet on smoking and cardiac recovery that challenged the belief that smoking alleviates stress and provided advice on avoiding relapse; a written quiz on the contents of the booklet, which was subsequently discussed with the nurse, to improve retention and understanding of the information; and an offer to be put in contact with another cardiac patient who had also recently stopped smoking to provide mutual support (a “buddy”). There is evidence that such support can be effective.12 Participants in the intervention group also had a declaration of commitment not to smoke that they had signed as a motivational reminder. A sticker was placed on their notes as a reminder to the staff to reinforce the intervention at future contacts.

The cardiac nurses recorded which elements of the intervention they implemented and the duration of the intervention. Before the study started, six nurses in five hospitals piloted the intervention and necessary changes were made. The pilot work suggested that the intervention took about 20 minutes of nurses' time.

Outcome measures

We defined participants as being continuously abstinent if they reported that they had smoked no more than five cigarettes (or “roll-ups,” cigars, or pipes) since recruitment and had not smoked at all in the past week. They also had to have an expired carbon monoxide reading <10 ppm and, at 12 months, a salivary cotinine concentration <20 ng/ml. We defined point prevalence abstinence as a self report of not having smoked at all for the past week and an expired carbon monoxide reading <10 ppm or, for the 12 month outcome, a salivary cotinine concentration <20 ng/ml.

Nurses recorded their own compliance with individual study procedures by ticking the intervention elements listed on the form. They also noted the duration of the intervention. Patients recorded their recall of the intervention by ticking items on a self completed questionnaire at the six week follow up.

Follow up

Data at the six week follow up were collected by nurses at the time of the routine hospital visit after discharge, or patients were visited at home. Carbon monoxide concentration was recorded in all participants who claimed to have stopped smoking and also in those in the intervention group who had not stopped smoking. This was to motivate abstinence, if nurses thought it was appropriate and worth while.

At 12 months participants were telephoned. If they said they had stopped smoking a home visit was arranged to collect a carbon monoxide reading and a saliva sample for cotinine analysis. If a home visit was not possible, a saliva collection kit was sent to the participant instead.

Data analysis

We used logistic regression analyses to assess the relation between baseline measures and outcomes. We assessed the effect of the intervention with χ² tests and 95% confidence intervals and multiple logistic regressions to control for possible confounding variables. All P values are two tailed.

Results

Recruitment to the study was slower than expected so the recruitment period had to be extended to 30 months. There were no significant differences between the intervention and control groups in any of the baseline characteristics (table 1). Most participants were heavy smokers who smoked their first cigarette each morning within 30 minutes of waking up.

Follow up and sample attrition

By six weeks 65 patients (12%) were lost to follow up, 14 of whom had died. By 12 months 60 participants were lost to follow up, 26 of whom who had died (12 in addition to 14 who died by six weeks), and nine had moved away to unknown addresses. There were no significant differences in numbers lost to follow up (26 and 34, respectively) or numbers who died or moved away (15 and 20, respectively) between the two groups. Participants who had died or moved to unknown addresses were removed from the sample; those not available for follow up for other reasons were counted as smokers. Our final sample for analysis was 526 at six weeks and 505 at 12 months.

We measured cotinine concentration in saliva samples of 217 out of the 229 participants who reported that they had stopped smoking at 12 months. In five cases analysis was not possible or appropriate (for example, when the person was using nicotine replacement therapy) so we used a carbon monoxide reading to validate abstinence. In seven cases, neither measure of validation was available, and we classified participants as smokers.

Smoking cessation rates

Table 2 shows validated abstinence rates at six weeks and 12 months. There were no significant differences between the two groups.

Adherence to the intervention procedures

Table 3 shows that almost every patient in the intervention group was given a carbon monoxide reading and the study booklet. Less than 70% (178) signed a commitment card or were told about the opportunity to be paired with a “buddy.” In only 7% (33) did nurses note that the “buddy” offer was accepted, and in only 4% (12) was the actual pairing with another patient recorded as having been organised. There was a significant difference in the mean number of elements implemented by each study site (F_16,257=4.2, P<0.0005).

We used a set of logistic regressions (one for each measure of abstinence) to examine the relation between abstinence and the number of elements of the intervention implemented. The regressions were non-significant for abstinence at six weeks, but significant for abstinence at 12 months (odds ratio 1.4, 1.1 to 1.9, P=0.003, for continuous abstinence and 1.3, 1.0 to 1.6, P=0.003, for point prevalent abstinence). The only individual item that predicted continuous abstinence at 12 months was signing the commitment card; 43% of those who signed the card remained abstinent compared with 22% of those who did not (1.6, P=0.002).

The intervention took an average of 34 minutes to implement (range 4-85 minutes), including obtaining informed consent and filling in the study questionnaire. A logistic regression showed no linear relation between the duration of the intervention and abstinence rates.

Table 4 shows the proportion of participants from each group who reported remembering each intervention element. Though the people in the control group did not receive the intervention, a substantial proportion of them “recalled” various procedures. Almost half of all participants “recalled” a non-existent control item (advice to drink water). The recall of advice did not differ in smokers and abstainers. Encouragingly, most members of both groups found the advice they were given to be helpful, with the intervention group finding the advice significantly more helpful than the control group. Only a few patients resented the advice.

Predictors of outcome

Of the baseline variables listed in table 1, older age, having a partner, having educational qualifications, smoking fewer cigarettes per day, waiting longer before smoking the first cigarette of the day, and wanting to remain abstinent were all positive predictors of at least one measure of abstinence. Patients who had had a myocardial infarction were much more likely to be abstinent at 12 months compared with patients who had undergone bypass surgery (191/280 v 5/29).

We analysed all these variables together with group allocation and study with a multivariate logistic regression equation. Table 5 shows the five variables that remained in the equation as predictors of outcome.