We conducted a cluster-randomized, controlled, superiority trial with 1:1 allocation of PCPs who then recruited patients seen in consultation. The trial was conducted in French-speaking Switzerland and the Rhone-Alps region of France between June 2021 and October 2023. The PCPs in Switzerland all worked in small, physician-owned practices (2 to 10 physicians) with medical assistants, but no nurses. Switzerland has universal, mandatory private health insurance with variable deductibles. The PCPs in France worked in multidisciplinary group practices that integrate public health nurses. France has universal public health insurance. The trial was registered prior to inclusion of the first participant (NCT04868474) and the protocol published.27 The trial was approved by the Vaud Ethics Committee (2020–02.898) and received an exemption from the National College of Teachers in General Practice (CNGE) in France (Decision 121222435). We followed CONSORT reporting guidelines.28

We recruited PCPs in private practices with >80 patients per month. PCP-level exclusion criteria were recent training in smoking cessation and plans to retire or relocate within <12 months. After the training program, PCPs were instructed to recruit among patients who were current smokers presenting for routine care. Patient-level inclusion criteria were age ≥18 years old, used tobacco daily (cigarettes, cigars, smokeless tobacco), and considered the recruiting PCP as their primary care doctor. Patient-level exclusion criteria were an acute medical condition that precluded even a brief discussion of smoking cessation, inability to provide informed consent (including due to language barriers as materials were in French), recent enrolment in a smoking cessation trial, and current daily use of a pharmacologic smoking cessation aid including electronic cigarettes. PCPs were encouraged to discuss smoking cessation with their patient during the baseline visit, if time permitted, using either the default choice with DA (intervention), or their usual approach (control). Follow-up visits to discuss smoking cessation were at the discretion of the PCPs, while all follow-up information was collected by telephone by the research team, without attempting to influence patients’ decision to quit smoking. PCPs received compensation for participating in the training, completing follow-up questionnaires, and per patient enrolled. Patients were not compensated.

The general approach has been described elsewhere.13,27 Briefly, the intervention consisted of a half-day training with two parts. First, a 1.5-h presentation focusing on the benefits and use of pharmacological and non-pharmacological smoking/tobacco cessation aids, as well as the concept of presenting tobacco cessation aids as the default choice with an encounter decision aid. The encounter decision aid was designed and tested by the study authors and made available in both paper and electronic forms.22 And second, a demonstration of the default choice technique with a video, followed by role plays with common scenarios from primary care to use default choices and the electronic decision aid.

The control group received the first 45 to 60 min of the intervention, primarily how to use pharmacological and non-pharmacological tobacco cessation aids, but without use of the DA. Both groups were trained to identify current smokers eligible for the trial and complete the informed consent.

The primary outcome was self-reported 7-day, point prevalence smoking abstinence at 6 months recorded during a telephone interview. Patients not responding to multiple telephone calls or email prompts to complete the questionnaire online were contacted by text messaging requesting if they had smoked tobacco in the previous 7 days. Secondary outcomes were the point-prevalence smoking abstinence, quit attempts since the last follow-up, and use of smoking cessation aids since the last follow-up, measured at 3 weeks, 3 months, and 6 months, and the patient-reported CollaboRATE scale (scale 1–10, higher scores = more involvement).29 The CollaboRATE scale contains three questions developed in English and validated in French: how much the provider helped the patient understand, listened to what matters most, and included what matters most to them in choosing what to do next.29 Important changes to the protocol were expanding recruitment into France, stopping offering carbon monoxide testing for those having quit at 6 months follow-up (very few patients willing to attend in person), and ending the trial prematurely due to inadequate recruitment. Despite extending the recruitment period, most PCPs recruited fewer patients than planned and we did not have sufficient resources to recruit additional PCPs.

We hypothesized that in the control group, 10% of current smokers would be recommended a smoking cessation aid and 4% would successfully quit.30 We thought that 10.5% of patients would successfully quit with the default choice approach (odds ratio of 2.8). If each arm had 20 clusters with 20 patients each, we would achieve 81.4% power to detect this difference with an intracluster correlation of 0.03. Assuming 15% drop out in both groups, we aimed to include 46 PCPs recruiting 23 patients each, so 1058 patients in total (that is 529 by group).

Randomization was done by the study statistician (MF) in blocks of two to four PCPs. PCPs from the same practice were randomized together. PCPs were blinded, as all materials mentioned just two versions of a training about smoking cessation, and not the nature of the intervention. Patients were likewise told that the study compared two training programs of their physicians. Outcome assessors were not blinded to study arm when performing follow-ups with PCPs and patients; however, care was taken not to reinforce teachings related to the intervention (i.e., use of the decision aid or default choices). The study statistician was blinded to group assignment.

PCP and patient characteristics were described using proportion or means with standard deviation as appropriate. The primary outcome was analyzed using an intention to treat approach with logistic regression controlling for clustering by the recruiting PCP, with persons lost to follow-up considered as still smoking. All secondary outcomes were analyzed with logistic regression controlling for clustering by the recruiting PCP, calculated with complete data only, assuming missing data were missing completely at random. Data were analyzed using Stata 18 (StataCorp. 2023, Stata Statistical Software: Release 18, College Station, TX).