Smoking quit rates among patients receiving pharmacotherapy who received general practitioner counselling versus intensive counselling: a retrospective cohort study

Therapeutic modalities for smoking cessation include behavioral and pharmacological therapy. Behavioral treatment options include brief guidance by the general practitioner (GP), intensive counselling workshops (usually moderated by nurses), personal telephone consultations, and web-based programs. A Cochrane review from 2021, that assessed the effect of behavioral interventions to support smoking cessation, concluded that all forms of counselling were effective, with more pronounced effect in the absence of pharmacotherapy [5]. Drug options include varenicline, an α4β2 nicotinic acetylcholine receptor partial agonist, sustained release bupropion, a norepinephrine and dopamine reuptake inhibitor, and nicotine replacement therapy (NRT) [6]. Behavioral treatment is recommended as support therapy for people using pharmacotherapy, [7, 8] as it increases quit rates by 10–20% [9].

There is a paucity of evidence regarding the effectiveness of brief counselling session provided by the patient’s personal GP, compared with intensive workshops. A randomized, controlled trial conducted in the Netherlands, compared these two modalities, both augmented with varenicline, and found similar quit rates with GP counselling, compared to nurse-moderated intensive workshops. The quit rates were 39% vs. 32.2% at 9–26 weeks following the intervention, and 28.8% vs. 25.5% at 9–52 weeks following the intervention, respectively [10]. A study from the US compared counselling by nurses with counselling by GPs and found a non-significant (NS) trend towards better quit rates with nurses (29.1% vs. 18.2%) [11]. A Cochrane review from 2017 compared individual counselling with no counselling (or brief counselling) and also more intensive counselling interventions with less intensive interventions [12]. Results showed that individual counselling is better than minimal contact control and that more intensive counseling is better than less intensive counselling. In both cases, all patients received NRT or bupropion.

In 2010 smoking cessation workshops became available, free of charge, to all smokers in Israel, through all Israeli health maintenance organizations (HMOs) [13]. Varenicline and bupropion were offered at reduced cost to smokers who attended smoking-cessation workshops. Starting in 2015, NRT was also offered at a reduced cost as a second line treatment for smokers who attended the workshops. Starting in 2019, varenicline and bupropion were also available at reduced cost to smokers who were counselled by their GPs [14].

Over the past 10 years the number of patients attending intensive counseling workshops in Israel has increased by 230% from 11,844 persons in 2010 to 27,342 persons in 2017.

Maccabi Healthcare Services (MHS), the second largest HMO in Israel, offers different smoking cessation programs including eight 90-minute meetings in an intensive workshop, usually led by a nurse, personal telephone counselling delivered through six 30-minutes calls, and per-request GP smoking cessation counselling. GPs were invited to a short smoking cessation counselling training, in which they were advised to provide three smoking counselling sessions for each patient; however, it was not mandatory to provide all three. GPs were paid for smoking cessation counselling by the HMO once for each patient, regardless of the number of consultations that were actually done. We therefore assumed that each patient received between 1 to 3 sessions (the estimated duration of a session was 10–15 minutes). All these options can be augmented by pharmacotherapy, often with varenicline, which is the drug of choice because of its good efficacy and safety profiles [15, 16].

In light of the health policy change made in 2019, the aim of this study was to compare the effectiveness of smoking cessation by GP office visits (GP group) with intensive counselling (behavioral support; workshops or personal telephone support) (IC group), in both cases augmented by varenicline. The secondary aim was to compare patients’ satisfaction with the different programs.

We designed a nation-wide retrospective cohort study using the MHS database. We identified patients who filled a prescription for varenicline from October 2019 to September 2020, and divided them to two groups according to the smoking cessation program they used: patients who participated in group workshops or received personal telephone sessions were included in the IC group, while patients who received smoking cessation counselling from their GP were included in the GP group.

Patients in both groups were selected randomly. We sent text messages to all patients 26 to 52 weeks after they first purchased varenicline, with a request to complete an online questionnaire to assess their smoking status. Text messages were sent in October 2020 and again in March 2021. Patients who did not respond were contacted by telephone between November 2020 and April 2021.

The outcome variables were (a) quit rate, using an ordinal scale of 0–3 (0 – does not smoke, 1 – smokes rarely, 2 – smokes every day, but less than before, and 3 – smokes as much as before) and (b) satisfaction with the smoking cessation process using a 5-point Likert scale (1 = very dissatisfied, 5 = very satisfied). The quit rate was later reduced to a binary variable with patients who said they did not smoke at all categorized as non-smokers and the others categorized as smokers. Smoking status variable described the patient’s abstinence status from the pharmacological treatment until the time that the questionnaire was answered. Data for both variables were collected by patient self-report; no biochemical confirmation was obtained.

Behavioural support for smoking cessation, either intensive counselling or GP counselling, served as the exposure variable. Data were extracted from the electronic records and verified using patient self-report.

Confounder variables included sociodemographic data such as gender, age and socioeconomic status (SES); medical data including comorbid diseases such as diabetes mellitus type 2 (DMt2), hypertension, ischemic heart disease (IHD), chronic obstructive pulmonary disease (COPD), a history of malignancy; and smoking-related variables including cigarettes per day, duration of smoking in years and duration of treatment with varenicline in months. All sociodemographic and medical variables were obtained from the electronic medical records. All smoking related variables were obtained by patient self-report.

MHS proactively follows workshop participants with an 80% quit rate at the end of the workshop, and a 45% quit rate 1 year after the workshop. Accordingly, we assumed a quit rate of 45% in the IC group and considered a difference of 5% or less as non-inferior. 301 patients for each group were required for a significance level of 5% with a power of 80% in a 2-sided test. The sample size was calculated using the WinPepi software, v11.65.

Descriptive statistics were used for all variables, with absolute numbers and percentages for categorical variables and mean and standard deviation for continuous variables. The chi-square test was used for univariate analyses of categorical variables. Student’s t-test was used for normally distributed continuous variables and the Mann-Whitney test for continuous variables which did not distribute normally. We performed a multivariate logistic regression analysis including variables with statistically significant associations with the outcome variables in univariate analyses, with the addition of smoking cessation group, age and gender regardless of the results of the univariate analysis. For null findings, we used the Bayes Factor to demonstrate that no association is a real finding and that the evidence supports the null hypothesis [17]. A Bayes Factor of less than 0.3 is considered substantial evidence to support the null hypothesis. We calculated a non-inferiority margin of 0.815, based on an assumption of 5% delta and a quit rate of 45% in the IC group. We used the Statistical Package for Social Sciences (SPSS) software version 27 for data analysis, and BayesFactor package in the R statistical programming environment to calculate the Bayes factor.

Ethical considerations.

All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was given via the online questionnaire or orally during telephone calls. The local ethics committee (IRB) of MHS approved this study (ID 0062–20-MHS).

From October 2019 to September 2020, 3817 patients fulfilled prescriptions for varenicline in MHS, 3066 (80.3%) received intensive behavioural support (by workshop or telephone), and 751 (19.7%) received GP counselling. We randomly sampled 435 patients from each group to participate in our survey. We sent the study questionnaire via text message and later conducted phone interviews with those who did not respond to the text messages. 604 patients (69%) agreed to participate in the study. Three patients were excluded due to missing data on behavioural counselling. Of the 601 patients who were included in the study, 300 were in the IC group and 301 in the GP group. The two groups were similar in gender and SES distribution, but patients in the IC group were older (47.4 vs. 44.7 years, P = 0.007) and smoked more (26.1 vs. 22.6 years, P < 0.001) (Table 1).

The quit rate for the GP group was 36.5% and for the IC group 42.3% (P = 0.147). The crude odds ratio (OR) for success in the GP group was 0.78 (0.56, 1.09) (Table 2). A test of association yielded a Bayes factor of 0.117 in support of the conclusion that there is no relationship between study group and outcome.

In a logistic regression analysis, controlling for age, gender, SES, IHD, COPD, number of cigarettes per day and duration of varenicline consumption, no significant difference in quitting odds was found between the IC group and the GP group. The adjusted OR (aOR) for GP group quitting was 0.79 (0.56, 1.13) with a Bayes Factor of 0.1 (Table 2). Non-inferiority could not be proven and the result is inconclusive.

For the entire study population, significantly higher aORs for quitting were found in the highest SES group with an aOR of 2.06, (1.39,3.07) and with a longer duration of varenicline consumption with an aOR of 1.30 (CI 1.15,1.47). Patients with COPD had an insignificantly higher aOR of 2.46 (0.96,6.30). Age, gender, and number of cigarettes per day were not associated with quit rate.

In the entire study population, satisfaction was significantly higher for patients who stopped smoking (82%) compared with those who continued smoking (35%) (P < 0.001). In the GP group, 68% of the participants were satisfied (57% very satisfied and 11% satisfied), while only 19% were not satisfied or not satisfied at all. In the IC group 64% were satisfied (51% very satisfied and 13% satisfied) and 14% were not satisfied or not satisfied at all. The proportion of participants who were neither satisfied nor dissatisfied was higher in the IC group (23% vs 14%) (P = 0.007).

We conducted a retrospective nation-wide cohort study to assess the difference in quit rates using two counselling styles, brief GP counselling during an office visit or intensive counselling, in both cases augmented by varenicline. We report a non-significant lower quit rate in the GP group compared with the IC group (36.5% vs. 42.3% respectively), with an aOR of 0.79 (0.56,1.13). Non-inferiority could not be proven, and the result was inconclusive. In the entire study group, we found a higher satisfaction rate for quitters compared with patients who continued smoking (82% vs. 35%) and for the GP group compared with the IC (68% vs. 64%).

The strengths of this study are its relatively large sample size, the high response rate (69%) and the heterogeneity of the participants and the large number of health providers personnel (GPs and nurses). Another strength is the observational design that provided real-world results, as the smoking cessation counsellors and the patients were not aware of being observed, they behaved as they usually do. This design eliminated a possible Hawthorne effect.

This study has several limitations. Being an observational study, compared with a randomised controlled trial study, exposes the study to selection bias. Patients who chose to attend an intensive counselling group might have had different characteristics from those who chose GP counselling. In order to overcome this bias, we compared the groups and checked for differences that might have impacted the outcomes, including comorbidities (Table 1). We also compared basic differences between patients who chose to attend the study and those who refused to take part (Table 3). In addition, we did not enquire regarding baseline psychological factors that might have influenced the outcomes. Another limitation is lack of uniformity of GP counselling, which could vary between merely providing a prescription to repeated counselling sessions. Although the study had a relatively large sample size, it was inadequate to prove a significant association (or lack of it) in quitting rates between the two groups. Last, as a retrospective study, the information which was provided in the questionnaire, including self-reported quitting status might have been biased.