A multi-level analysis of non-significant counseling effects in a randomized smoking cessation trial

Abstract

Aims: To determine, in the context of a trial in which counseling did not improve smoking cessation outcomes, whether this was due to a failure of the conceptual theory identifying treatment targets or the action theory specifying interventions.

Design: Data from a randomized clinical trial of smoking cessation counseling and bupropion SR were submitted to multi-level modeling to test whether counseling influenced real-time reports of cognitions, emotions and behaviors, and whether these targets predicted abstinence.

Setting: Center for Tobacco Research and Intervention, Madison, WI.

Participants: A total of 403 adult, daily smokers without contraindications to bupropion SR use. Participants were assigned randomly to receive individual counseling or no counseling and a 9-week course of bupropion SR or placebo pill. Cessation counseling was delivered in eight 10-minute sessions focused on bolstering social support, motivation, problem-solving and coping skills.

Measurements: Pre- and post-quit ecological momentary assessments of smoking behavior, smoking triggers, active prevention and coping strategies, motivation to quit, difficulty quitting and reactions to initial lapses.

Findings: Counseling prompted avoidance of access to cigarettes, improved quitting self-efficacy, reduced perceived difficulty of quitting over time and protected against guilt and demoralization following lapses. Results also supported the importance of limiting cigarette access, receiving social support, strong motivation and confidence and easing withdrawal distress during cessation efforts. Quitting self-efficacy and perceived difficulty quitting may partially mediate counseling effects on abstinence.

Conclusions: Smoking cessation counseling may work by supporting confidence about quitting and reducing perceived difficulty quitting. Counseling did not affect other targets that protect against relapse.

Figure 1

Enrollment and retention flow diagram, collapsed over medication conditions. Note that the number retained for analyses is the total pool of participants who completed the quit day visit and provided at least some post-quit data. The number of subjects included in specific multilevel models was sometimes smaller than this total sample size due to missing mediator data. Abstinence outcomes were not missing for any subject, as intent-to-treat analyses were conducted.

Figure 2

Study timeline. Target days of visits (relative to the day 0, the target quit day), are shown at bottom on the timeline. Intervals between visits varied slightly across subjects. Screening visits included a group orientation and an individual visit including a physical exam. Electronic Diary (ED) training was conducted at the next visit and the 2-day ED training period (noted as T) followed. The ED assessment period was parsed into a 1-week pre-treatment baseline period, a 1-week treatment run-up period, and a 4-week post-quit mediator assessment period, as shown in the lowest level of the figure. Counseling sessions began 1-week pre-treatment and were offered at this and the following 7 visits. Medication treatment began 1-week pre-quit (at 150 mg) per day, increased to 300 mg per day on day −4, and continued until day 56. The abstinence outcome used in analyses (noted as 7-day Abst in the figure) was intent-to-treat complete abstinence (not even a puff) in the week preceding the final visit confirmed by CO (collected at all visits) and not disconfirmed by serum cotinine testing.

Figure 3

Estimated post-quit probabilities of behaviors reported in random prompts as a function of counseling condition (solid line indicates no counseling, dashed line indicates counseling). Panel A shows the probability of spending time where: smoking was permitted, cigarettes were easily accessible, and someone was smoking. Panel B shows the probabilities of: attempting to prevent urges to smoke, attempting to cope with urges, attempting to cope with stressful events, and receiving social support.

Figure 4

Estimated post-quit evening report ratings as a function of counseling condition (solid line indicates no counseling, dashed line indicates counseling). Panel A shows ratings of confidence related to quitting, motivation to quit, and willingness to work hard at quitting. Panel B shows perceived difficulty quitting and being bothered by withdrawal.

Figure 5

Estimate means or probabilities of slip report responses as a function of counseling condition (no counseling shown by solid lines, counseling shown by dashed lines). Panels A and B depict mean ratings of confidence related to quitting, feelings of guilt about slipping, and feeling like giving up after a slip as a function of days from the quit day and the number of slips reported, respectively. Panels C and D depict the probability of attempting to abstain prior to a lapse and of making plans to avoid future slips following a lapse as a function of days post-quit and number of slips, respectively.