Smoking Cessation
Predictors and moderators of outcome in different brief interventions for smoking cessation in general medical practice

https://doi.org/10.1016/j.pec.2009.07.005Get rights and content

Abstract

Objective

To explore demographic-, health-, and smoking-related predictors and moderators of outcome in smokers who participated in two different brief smoking cessation interventions.

Methods

Data were acquired using a quasi-randomized controlled trial that tested the efficacy of computer-generated tailored letters and physician-delivered brief advice against assessment only. Daily smokers (n = 1499) were recruited from 34 general medical practices. We used Generalized Estimating Equation analyses to investigate the relationship between 6-month prolonged smoking abstinence assessed at 12-, 18-, and 24-month follow-ups and potential predictors and moderators.

Results

Female gender (OR = 1.49, 95% CI = 1.01–2.19), higher level of education (OR = 1.82, 95% CI = 1.18–2.82), intention to quit smoking (OR = 1.66, 95% CI = 1.16–2.38), and smoking cessation self-efficacy (OR = 1.30, 95% CI = 1.03–1.64) were positively, nicotine dependence (OR = 0.84, 95% CI = 0.76–0.94) and the presence of a smoking partner (OR = 0.60, 95% CI = 0.42–0.85) were negatively associated with smoking abstinence. Compared to assessment only, physician advice was less effective for people without an intention to quit smoking and for unemployed.

Conclusion

Smoking cessation interventions might be improved by tailoring them to demographic- and smoking-related variables which were identified as predictors in this study.

Practice implications

The results suggest that tailored letters are a more universally applicable brief intervention in general medical practice than physician advice.

Introduction

Brief smoking cessation interventions in primary care medicine are promising to substantially reduce the incidence of smoking-related disease [1]. In combination with a systematic screening program, they can reach the majority of the smoking population within 1 year [2]. Brief advice delivered by physicians resulted in a small but significant effect on smoking cessation rates [3]. However, a lack of time and training, inadequate reimbursement, and insufficient patient motivation to change have all been reported as major barriers to the success of smoking cessation interventions by healthcare professionals [4], [5], [6].

Computer-generated smoking cessation interventions [7] are time-saving alternatives to interpersonal counseling by the primary care physician and may be crucial for implementing smoking cessation interventions in primary care. Although computer-generated tailored self-help materials proved to be more effective than no intervention or untailored materials [8], [9], studies that tested the efficacy of tailored self-help materials in primary medical care settings revealed mixed results [10], [11], [12]. The first direct comparison of physician-delivered counseling and computer-tailored self-help materials in primary care medicine was the study “Proactive interventions for smoking cessation in general medical practices” (ProGP) [13]. It tested the efficacy of (1) computer-generated tailored letters (TL) and (2) brief advice delivered by general practitioners (PA) against (3) an assessment-only condition (AO). Analyses including the 6-, 12-, 18-, and 24-month follow-up assessments confirmed statistically significant effects of both interventions compared to the assessment-only group on 7-day (TL: OR = 1.7, 95% CI: 1.2–2.4; PA: OR = 1.5, 95% CI: 1.1–2.2) and 6-month abstinence rates (TL: OR = 1.7, 95% CI: 1.1–2.7; PA: OR = 1.8, 95% CI: 1.1–2.9). Assuming patients lost to follow-up to be smokers, 6-month abstinence rates at the final follow-up were 10.2 % in the tailored letters group, 9.7 % with physician advice, and 6.7 % in the assessment-only group.

Findings from controlled studies like the ProGP study should be considered a preliminary step in understanding efficacious interventions. Additional investigations can identify those for whom treatment may work best or can investigate the mechanisms by which a treatment may achieve its aims [14]. Predictors specify which individuals are likely to have better outcomes irrespective of the treatment they receive. Knowledge of predictors can be used as basis for tailoring intervention materials [15]. Moderators specify which individuals are likely to have a better response to a specific treatment. Following the definitions by [14], a predictor of treatment outcome is a fixed (e.g., sex) or variable (e.g., age, weight) factor that precedes outcome. A moderator is a fixed or variable factor that precedes outcome, is unrelated to the treatment received, and has an interactive effect with treatment on outcome.

Previous studies compared the predictive validity of variables derived from different theoretical approaches [16], [17], [18] or tested multivariate predictors of smoking cessation within population surveys [19], [20]. However, only few analyses have investigated a comprehensive set of predictors or moderators in the context of behavioral interventions for smoking cessation.

Results from a multivariate analysis in a combined data set from five trials in which expert system interventions were provided revealed, that a longer duration of a past quit attempt, a lower number of cigarettes smoked per day, later stage of change and more years of education were positively associated with 6-month abstinence at follow-up assessments [15]. In another study that tested the effectiveness of a computer-tailored smoking cessation program vs. no intervention, having made a quit attempt in the previous year, stage of change, and time to first cigarette in the morning were all predictive of 4-week smoking abstinence at follow-up [21]. In terms of moderators, the program was more effective among smokers in precontemplation or contemplation stages of change than among smokers in the preparation stage, and the program was more effective in smokers who said that quitting would be easy than in those who said that it would be difficult. The demographic variables of age, sex and school education did not predict or moderate program effectiveness according to multivariate analyses.

Our study explored a comprehensive set of demographic-, health-, and smoking-related variables as possible predictors and moderators of smoking cessation within the ProGP data set. This is the first study to test moderators and predictors using data from a trial in which two brief intervention approaches were directly compared to an assessment-only group. The results of our study may guide future intervention planning by specifying those patients for whom a certain intervention is most appropriate. Our work could be used as a basis for further improving tailored interventions.

Section snippets

Sample

For this study, we used data from the project ProGP. This project tested the efficacy of brief interventions delivered by a physician or by computer-generated tailored letters compared to an assessment-only group in 34 randomly selected general practices in a defined region of Northern Germany [13]. In the general practices, study nurses screened every consecutive patient for age and smoking status over a period of 3 weeks. Following screening, patients fulfilling the inclusion criteria (daily

Baseline characteristics

The baseline characteristics of the three study groups and the total sample are shown in Table 1. There were no significant differences between the study groups with respect to these baseline characteristics, except for the number of children under the age of 15 living at home (χ2 = 7.7, p = .02) and alcohol consumption measured by the AUDIT-C (F = 4.7, p = .01). The percentage of children under the age of 15 who were living at home was higher in the intervention groups than in the assessment-only

Discussion

This exploratory study aimed to test the differential effectiveness of a study conducted within the ProGP project in order to guide future intervention planning. We analyzed potential predictors and moderators of smoking abstinence in a sample of 1499 adults who smoked daily and who participated in one of two brief interventions for smoking cessation or in an assessment-only group.

The smoking-related variables FTND-score, intention to quit smoking, and smoking cessation self-efficacy proved to

References (55)

  • J. Cornuz et al.

    Physicians’ attitudes towards prevention: importance of intervention-specific barriers and physicians’ health habits

    Fam Pract

    (2000)
  • D. Twardella et al.

    Lack of training as a central barrier to the promotion of smoking cessation: a survey among general practitioners in Germany

    Eur J Public Health

    (2005)
  • F. Vogt et al.

    General practitioners’ and family physicians’ negative beliefs and attitudes towards discussing smoking cessation with patients: a systematic review

    Addiction

    (2005)
  • T. Lancaster et al.

    Self-help interventions for smoking cessation. The cochrane library

    (2000)
  • S.M. Noar et al.

    Does tailoring matter? Meta-analytic review of tailored print health behavior change interventions

    Psych Bull

    (2007)
  • P. Aveyard et al.

    A controlled trial of an expert system and self-help manual intervention based on the stages of change versus standard self-help materials in smoking cessation

    Addiction

    (2003)
  • Lennox AS, Osman LM, Reiter E, Robertson R, Friend J, McCann I, Skatun D, Donnan PT. Cost effectiveness of computer...
  • V.J. Strecher et al.

    The effects of computer-tailored smoking cessation messages in family practice settings

    J Fam Pract

    (1994)
  • C. Meyer et al.

    Proactive interventions for smoking cessation in general medical practice: a quasi-randomized controlled trial to examine the efficacy of computer-tailored letters and physician-delievered brief advice

    Addiction

    (2008)
  • H.C. Kraemer et al.

    Mediators and moderators of treatment effects in randomized clinical trials

    Arch Gen Psychiatry

    (2006)
  • W.F. Velicer et al.

    Demographic variables, smoking variables, and outcome across five studies

    Health Psychol

    (2007)
  • D.B. Abrams et al.

    Stages of change versus addiction: a replication and extension

    Nicotine Tob Res

    (2000)
  • A.J. Farkas et al.

    Addiction versus stages of change models in predicting smoking cessation

    Addiction

    (1996)
  • M. Kleinjan et al.

    Associations between the transtheoretical processes of change, nicotine dependence and adolescent smokers’ transition through the stages of change

    Addiction

    (2008)
  • N. Hymowitz et al.

    Predictors of smoking cessation in a cohort of adult smokers followed for five years

    Tob Control

    (1997)
  • R. West et al.

    Smoking cessation and smoking patterns in the general population: a 1-year follow-up

    Addiction

    (2001)
  • J.F. Etter et al.

    Effectiveness of a computer-tailored smoking cessation program. A randomized trial

    Arch Intern Med

    (2001)
  • Cited by (35)

    • Evidence from interventions based on theoretical models for lifestyle modification (physical activity, diet, alcohol and tobacco use) in primary care settings: A systematic review

      2015, Preventive Medicine
      Citation Excerpt :

      Long-term effectiveness was evaluated in six (Cabezas et al., 2011; Meyer et al., 2008; Haug et al., 2010; Salmela et al., 2009; Van Sluijs et al., 2004; Logue et al., 2005; Prochaska et al., 2005) studies; in the remaining 11 studies, follow-up lasted longer than 3 months but less than 12. Methodological quality was high in six (35.3%) of the 17 projects (Young et al., 2008; Cabezas et al., 2011; Perkins-Porras et al., 2005; Van Sluijs et al., 2004; Logue et al., 2005; Lai et al., 2010), average in 10 (58.8%) (Pinto et al., 2001; Lennox et al., 2001; Pieterse et al., 2001; Meyer et al., 2008; Haug et al., 2010; Salmela et al., 2009; Steptoe et al., 2000, 2001, 2003a,b; Prochaska et al., 2005; Williams et al., 2004; Kinmonth et al., 2008; Hardeman et al., 2009, 2011a,b; Simmons et al., 2010; Blanchard et al., 2007; Fortier et al., 2007, 2011a,b; Borrelli et al., 2005, 2010) and low in one study (5.9%) (Märki et al., 2006). The 17 included studies are summarized in Table 2.

    View all citing articles on Scopus
    View full text