Smoking cessation initiated during hospital stay for patients with coronary artery disease: a randomized controlled trial

Setting and participants

The setting was 4 cardiac units in a large urban hospital in western Canada. We considered eligible patients to be those who were 18 years of age or older, used tobacco in the month before admission, had a minimum projected hospital stay of 36 hours (to allow time for the intervention), were willing to be randomly assigned to an intervention and had telephone access to receive counselling after discharge. We excluded patients who were pregnant, involved in a concurrent trial for tobacco cessation, medically unstable (as determined by a physician), lived in an institution without telephone access, could not speak English or had trouble communicating, had a history of substance abuse or psychiatric disorders, or for whom the patient’s physicians refused to allow participation.

The Conjoint Health Research Ethics board of the University of Calgary provided ethics approval for this randomized clinical trial.

Procedure

Over a 15-month period beginning in December 1999, the research nurse asked all patients admitted to each of the 4 cardiac units if they had used tobacco in the month before admission. She reviewed the medical charts of smokers for eligibility, approached medically stable patients at the bedside, described the study, obtained informed consent and collected baseline data. She opened the randomization envelope and informed the patients of intervention assignment (intensive or minimal). We developed the randomization code using a computer random-number generator to select random permuted blocks of 10. Randomization was stratified by acute MI and CABG to ensure equal numbers in each intervention group.

Minimal intervention

The research nurse advised patients to quit smoking by personalizing the message to each patient’s medical conditions. The nurse reviewed 2 pamphlets (how to quit and where to find help quitting) with the patient, and she put a note in each patient’s chart to ask the attending physician to deliver a scripted nonsmoking message 12 at the bedside during the patient’s hospital stay. Pharmacotherapy was not part of the study. However, pharmacotherapy was introduced as an aid to cessation and was available through the hospital formulary during the patient’s hospital stay if the patient requested it and a physician ordered it. The research nurse explained how to use the medications to patients who were interested.

Intensive intervention

Patients in this group received the minimal intervention plus 45–60 minutes of bedside education and counselling, take-home materials (video, workbook, audiotape) and 7 telephone counselling sessions initiated by the research nurse (2, 7, 14, 21, 30, 45 and 60 days after discharge). Education covered the patient’s personalized risks associated with smoking and the benefits of quitting, as well as withdrawal, weight gain and the benefits of making the home smoke-free. Counselling was based on Marlatt and Gordon’s relapse prevention model. 13 Patients rated their confidence to remain smoke-free in 14 high-risk situations, 14 and they developed behavioural, cognitive and social support strategies to remain smoke-free in situations for which they had less than 70% confidence. Telephone counselling, which was designed to last 5–10 minutes per call, 15 focused on the prevention of relapse (developing cognitive, behavioural and social support strategies for situations identified as high risk for relapse).

Follow-up after discharge

A research assistant who had no prior contact with the participants called each participant at 3, 6 and 12 months after discharge to assess the participant’s use of tobacco in the previous 7 days. The assistant was blinded to the intervention assignment by not having any information about the patients, except for their name, scheduled callback date and telephone number (the intervention was not known). During the 12-month follow-up call, the research assistant asked all patients who reported not using tobacco to provide the name and contact number of a family member or friend to provide proxy corroboration of their tobacco status. The assistant then called the friend or family member and asked if the patient had smoked or used tobacco in the last 7 days.

Measures

We collected baseline sociodemographic data, including age, education, employment status, marital status, ethnic background, sex, weekly alcohol consumption, whether patients lived alone or with a smoker, social support to quit tobacco, and the degree of restrictions on tobacco use at home and work. Medical information included history of acute MI, CABG, percutaneous translumial coronary angioplasty, peripheral vascular disease, chronic obstructive pulmonary disease, diabetes, number of weeks of stable angina, hospital admissions in the previous year, length of the current hospital stay, and discharge diagnosis based on the International Classification of Disease (ninth revision [ICD-9]) codes. 16

We collected information on the history of tobacco use, including the amount and number of years smoked, number of quit attempts, tobacco addictions (5-item modified Fagerstrom Tolerance Questionnaire 17), and previous use of pharmacotherapy, as well as in-hospital withdrawal and difficulty quitting while in hospital. Psychosocial information included single-item screens of the patient’s confidence to quit (0%–100% confidence), intention to quit (1–7, with 7 being full intention), depressed mood (0 [not at all] to 8 [very severely depressed]), and beliefs about whether they could avoid health problems by quitting (1 [very likely] to 4 [very unlikely]), whether tobacco was harming their health, and whether it was worth quitting after 20 years of using tobacco (1 [strongly agree] to [4 strongly disagree] 18).

Primary outcome

Our primary outcome was smoking status at 3, 6 and 12 months after discharge. We measured smoking status using the National Heart, Lung and Blood Institute’s 19 consensus-conference definition of self-reported 7-day point prevalence of abstinence (not even a puff for a minimum of 7 consecutive days before the assessment) and long-term (continuous) abstinence, which was measured by self-reported 7-day point-prevalence at 3, 6 and 12 months.

Although self-reported tobacco status is a valid measure, 20^,21 we used proxy confirmation as a conservative fidelity check of abstinence at 12 months. Proxy corroboration results in similar abstinence rates to carbon monoxide confirmation of tobacco rates. 22^,23 In this trial, it was more feasible to use a proxy measure than to ask patients to return to hospital to provide a sample for biochemical confirmation because patients resided not only in the hospital community but also in outlying rural areas.

Data analyses

We assumed that the 12-month confirmed abstinence would be 30% in the minimal intervention group (based on the Stanford study 10); thus, 136 participants per group were needed to give an 80% probability of detecting a clinically significant 10% increase in abstinence rates with the intensive intervention (α = 0.05, 1 tailed).

We compared the baseline characteristics between the intervention groups using χ² tests for categorical and dichotomous variables and t tests for continuous variables. Although statistical testing for baseline differences is discouraged by some, 24 others maintain that baseline differences can occur between treatment groups, even with adequate randomization and allocation concealment. 25 Thus, the differences should be checked to ensure that the results of this trial and future systematic reviews and meta-analyses that include this trial are not biased. 26

For outcomes related to tobacco status, we used hierarchical logistic regression to test for the main effects of the intervention (intensive v. minimal) and the reason for admission to hospital (CABG v. acute MI) and the interaction between the intervention and the reason for admission. Odds ratios (ORs) with 95% confidence intervals (CIs), derived from the last statistically significant step in the logistic analyses, were used to determine the significance of the main effects and the interaction. We excluded from the anslyses participants who had died. Those who had dropped out or for whom cessation data were not available were considered smokers for the time points at which their data were missing.

Our secondary analyses included a hierarchical logistic regression analysis to test for the effects of the intervention, pharmacotherapy and the interaction between the intervention and pharmacotherapy on 12-month abstinence. We also performed logistic regression analysis to determine predictors of 12-month cessation, while adhering to conventions to prevent over-fitting the prediction model to the sample. 27 The predictors included in the analysis were age, sex, education, admission for CABG or acute MI, history of acute MI, addiction, confidence to quit tobacco, restrictions on indoor tobacco use at home, depression, use of alcohol, and whether the patient had quit tobacco for at least 1 week in the year before hospital admission.

Patient enrollment and flow through the trial

Recruitment began in December 1999, and follow-up ended in March 2003. Of the 10 336 admissions to the 4 cardiac units, 19% were because of acute MI (n = 1921) and 14% were for CABG (n = 1425). Of the patients with acute MI, 34% (n = 654) were smokers, and 15% of the patients admitted for CABG (n = 211) were smokers. The overall smoking prevalence rate was 26% (865/3346). Of the patients admitted because of acute MI or for CABG who smoked, 32% were enrolled (276/865) in the study (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1: Patient enrollment, allocation, intervention completion and follow-up. Note: CABG = coronary artery bypass graft, MI = myocardial infarction.

All of the patients that were randomly assigned to the minimal intervention and all but 1 of the patients assigned to the intensive intervention received the in-hospital intervention. Of the patients in the intensive intervention, 95% (130/137) received the postdischarge intervention. On average, 93% of patients were available per call, 5% were not reached, and 2% were not called because the research nurse was absent, on vacation or missed a call. Patients received an average of 6.56 (standard deviation 0.8) calls. At the end of the 12-month period, 4 (1%) patients had died, 16 (6%) had dropped out of the study and 9 (3%) were not reached for the final follow-up (i.e., phone line busy, no answer, answering machine, messages not returned).

Participant characteristics

Stratified randomization resulted in equal numbers of CABG and acute MI patients assigned to each intervention group (Table 1). Nine patients did not have an ICD-9 discharge diagnosis code of CABG or acute MI, although they had been admitted for acute MI. There were no significant differences between the groups for medical history (Table 1) or baseline variables (Table 2).

Self-reported abstinence at 3, 6 and 12 months

More patients in the intensive intervention than in the minimal intervention reported not smoking at 3 months (p = 0.009), 6 months (p = 0.003) and 12 months (p = 0.007) (Table 3). The odds of quitting were 2 times greater for patients who received the intensive intervention compared with the minimal intervention at 3, 6 and 12 months (Table 3). The reason for admission to hospital (CABG v. acute MI) or the interaction between the intervention and reason for admission did not contribute significantly to the variance in cessation at any follow-up time.

Confirmation of self-reported abstinence

Of the patients who reported not smoking after 12 months, 16% (23/147) did not provide proxy information to allow us to confirm their smoking status. These patients were considered smokers in this analysis. Of the patients who provided proxy information (84%, 124/147), 3 were reported by a friend or family member to have smoked during the past 7 day (2 in the minimal intervention group, 1 in the intensive intervention group). These 3 patients were considered smokers in this analysis.

More patients in the intensive intervention than in the minimal intervention were confirmed nonsmokers at 12 months (p = 0.002) (Table 3). There was no significant effect of reason for admission to hospital (CABG v. acute MI) or significant interaction between the intervention and the reason for admission.

Continuous abstinence at 3, 6 and 12 months

The intervention (p = 0.004) and the reason for hospital admission (CABG v. acute MI) (p = 0.04) were significant predictors of continuous abstinence. The interaction between the intervention and having CABG or acute MI was not significant (p = 0.24). Patients who received the intensive intervention had significantly higher rates of continuous abstinence than those who received the minimal intervention (57% [77/135] v. 39% [54/137]; OR 2.1, 95% CI 1.3–3.4, p = 0.003). Patients admitted to hospital for CABG had significantly higher rates of continuous abstinence than those admitted because of acute MI (57% [50/88] v. 44% [81/184]; OR 1.7, 95%CI 1.0–2.9, p = 0.038;Table 4).

Pharmacotherapy

Although pharmacotherapy was not part of either intervention, it was used by 34% of patients in both groups (minimal intervention: 40/118 patients; intensive intervention: 40/116 patients [data were missing for 19 patients in each group]). Both the intervention and the use of pharmacotherapy were predictors of abstinence (p = 0.001); the interaction between the intervention and the use of pharmacotherapy was not (p = 0.26). Patients who received the intensive intervention had significantly higher odds of being abstinent at 12 months than those who received the minimal intervention (69% [80/116] v. 48% [56/118], OR 2.7, 95% CI 1.5–4.8). In both interventions, patients who used pharmacotherapy had significantly lower odds of being abstinent at 12 months (39% [31/80] v. 68% [105/154], OR 0.3, 95% CI 0.2–0.5;Table 4).

Predictors of 12-month abstinence

Twenty-four patients were missing data for at least 1 of the predictor variables and were excluded from the analyses, leaving a sample size of 248 patients. Significant predictors of the point-prevalence of abstinence at 12 months included receiving the intensive (v. minimal) intervention (OR 2.12, 95% CI 1.2–3.7), having no history (v. having a history) of acute MI before the current admission (OR 2.94, 95% CI 1.3–6.5), having a postsecondary education (v. having a high school education or less) (OR 2.34, 95% CI 1.3–4.1) and having at least some smoking restrictions at home (v. no restrictions) (OR 1.96, 95% CI 1.1–3.5).