Introduction

In Europe, tobacco consumption among women remains the main preventable cause of an increasing prevalence of tobacco-related cancer and cancer-related mortality1,2,3,4. The preventive measures aiming to reduce the burden of preventable cancer cases include smoking cessation5 and cancer screening programs6. The Dutch cervical cancer screening program takes place in general practice. For Dutch inhabitants, general practice, where support to stop smoking is available7, is the primary and most contacted health care service, but opportunities to provide stop smoking advice and support are often underutilized8,9.

We hypothesized that women who smoke could benefit from a stop smoking intervention combined with routine cervical cancer screening. The smear test visit can serve as a teachable moment, creating a link between a woman’s behavior (smoking) and the reason for the visit (cancer prevention)10. In addition, this approach could routinely reach in general practice (younger) women who smoke, among whom the prevalence of tobacco-related disease is still low11. Smoking is a risk factor for persistent cervical high-risk human papillomavirus (hrHPV) infection, carcinoma in situ, and cervical cancer12,13,14,15.

In a cluster randomized controlled trial (cRCT), we studied the effect of a brief stop smoking strategy based on the Ask-Advise-Connect (AAC) method16 on smoking cessation outcomes in women who participate in cervical cancer screening17,18. In Dutch general practice, practice assistants (PAs) routinely perform cervical smears. Parallel to the effect study, we conducted a process evaluation (PE)17.

The Smoking cessation strategy in the national cervical cancer screening program (SUCCESS) cRCT results suggested that brief stop smoking advice after the cervical smear might encourage women who smoke to attempt quitting and seek cessation counseling, but a significant effect could not be demonstrated18.

The brief stop smoking strategy is considered a complex intervention19,20. When studying complex interventions, a PE adds to the study of the effect an evaluation of factors that could influence the intervention delivery and its implementation19,20. Several issues are potentially relevant. For example, stop smoking care is not part of PAs’ daily working routine. Also, the smear is a delicate moment, and women have reported being ambivalent towards stop smoking advice after the smear test in a qualitative study on prospective acceptability21. Lastly, the organization of care in general practices throughout the Netherlands is heterogeneous. It is important to assess whether such an opportunistic approach could affect future screening participation rates22.

A mixed-methods approach allows an in-depth and representative PE, including the perspectives and experiences of the individuals who received and delivered the intervention, while taking into account the context in which the intervention was delivered19,20. In this PE, we aimed to study the acceptability, feasibility, and implementation of a stop smoking strategy delivered by PAs to women who smoke and visit their general practice for cervical cancer screening.

Results

Reach

Among the women enrolled in the SUCCESS study, on average, 16.9% smoked (Table 1). Compared to the nonsmoking women, the smokers were slightly younger and tended to have a lower level of education (Supplementary Table 1). The percentage of women with comorbidities was very low (Table 2). At follow-up, the women smokers more often reported having had an abnormal smear test result (hrHPV-positive) compared to the women who did not smoke (Table 2). However, around 10–15% of the women reported not remembering their smear test result, which may hamper the interpretation of this comparison. All women reported participating in cervical cancer screening mainly for early detection or to be sure that they did not have a cervical abnormality (cancer/pre-cancer). Among the women smokers, a perceived increased risk of developing cervical cancer or having had an aberrant smear in the past were reasons for participating in cervical screening (Table 2).

Table 1 Reach: prevalence of smoking in all women participants of the SUCCESS study
Full size table
Table 2 Descriptive variables of women included in the SUCCESS study collected at 6 months follow-up
Full size table

Effectiveness

Future cervical cancer screening intention measured at follow-up was high and did not significantly differ between the women smokers from both study arms (intervention 98.1%; control 98.9%) or between the women who do (98.5%) and the women who do not smoke (98.3%) (Table 3; Supplementary Tables 2 and 3).

Table 3 Cervical cancer screening intention
Full size table

Implementation

Brief stop smoking advice after the cervical smear was appreciated (greatly or a little) by 47.4% of the women who smoke from the intervention group, while another 12.4% did not (at all) appreciate the strategy, and 25.5% had no opinion or did not know (for 15%, the data on acceptability were missing).

In total, 39 women smokers in the intervention arm were connected to smoking cessation counseling within their general practice. They greatly appreciated the support and experienced a high willingness of the healthcare provider to provide support (Supplementary Table 4). Reasons for declining an appointment for cessation counseling after the smear most often were not (yet) willing to quit smoking or not willing to receive support for stopping smoking (Supplementary Table 4).

Maintenance

In the intervention practices, around 21%–25% of PAs will continue to ask for smoking status, and up to around 32%–45% aim to connect women with cessation counseling (Table 4).

Table 4 Characteristics of individual PAs at baseline and follow-up, best performing vs. least performing intervention practices
Full size table

At follow-up, 70.8% of the women smokers from the intervention group who completed the follow-up questionnaire (vs. 49.1% of the controls) were significantly likelier to report that they would contact their general practice for quitting smoking (Supplementary Table 4), odds ratio [OR] 2.14 (1.35–3.39), P = 0.03 (ICC = 0.059). In addition, women from the intervention group were significantly likelier to believe they would succeed in stopping smoking with help from their general practice (75.8% vs. 60.5% of controls), OR 1.79 (1.14–2.81), P = 0.02 (ICC = 0.057).

Of the 31 intervention practices included in this PE, 15 were categorized as best performing practices and 16 as least performing practices (Fig. 1). In the best performing practices, at least 60% of the women smokers experienced the SUCCESS stop smoking strategy as acceptable, the number of unexposed participants was low (max. 12.5%), and random effects on the primary outcome for the effect study (quit attempt) were positive for 12 (out of the 15 best performing) practices. In the 16 least performing practices, at least 30% of the women smokers did not experience the strategy as acceptable, and at least 25% of participants were unexposed to the strategy, together with a stronger negative random effect.

Fig. 1: Selection and classification of intervention practices by study performance.
figure 1

a Assigned number of participating practice, i.e. cluster. GREEN = best performing clusters, based on acceptability, exposure, and random effect. ORANGE = least performing clusters. b Random effect of cluster on primary outcome within the effect study. AAC Ask-Advise-Connect, acc. acceptability

Full size image

Table 5 presents the practice characteristics. The best performing practices tended to be a little smaller (i.e., fewer registered patients), with fewer smear tests per year and PAs who perform the smear, but more time per smear consultation. The PAs from the best performing practices tended to have slightly more work experience and be more trained to advise patients to stop smoking (Table 5).

Table 5 Characteristics of best vs. least performing intervention practices
Full size table

The follow-up measurements among the PAs are shown in Table 4. The PAs from the best practices were slightly more exposed to the AAC steps. The PAs perceived stop smoking advice after the smear as their role, making a positive impact. However, they appeared to be more neutral about whether it was their task. Asking for smoking status and connecting women with cessation counseling were perceived as feasible, and this was found a little more among the PAs from the best performing practices.

Qualitative results

Three themes were identified regarding the implementation, reach, and maintenance of the SUCCESS stop smoking strategy, covering the barriers and facilitators and differences between the best and least performing practices. Theme 1: The practice—Teamwork and the individual PA, Theme 2: The strategy—Ask & connect and acceptability, and Theme 3: Context—Low prevalence of smoking and COVID-19. Table 6 contains illustrative quotations from Themes 1–3. Supplementary Table 5 details which constructs differed between the best performing and least performing practices. Supplementary Tables 6 and 7 detail the barriers and facilitators for each RE-AIM construct.

Table 6 Theme 1–3: Illustrative quotations
Full size table

Theme 1: The practice—Teamwork and the individual PA

In all the best performing and in many least performing practices, PAs initiated participation in the study, with support from the general practitioner (GP), which encouraged the PAs to feel responsible for delivering the strategy (Table 6, Supplementary Table 5). However, in some least performing practices, the management or the general practitioner imposed participation in the study. In these practices, the PAs did not feel ownership, often because insufficient time or support was allocated to them to deliver the strategy (Table 6, Supplementary Table 5). A close-knit team with short communication lines and a clear division of tasks facilitated implementation, whereas staff changes or non-participation of colleague PAs were experienced as barriers (Table 6). In all the best performing practices, all PAs in the team participated in the study. In some larger healthcare centers, we observed that it was more challenging to include all PAs and ensure they were all adequately trained. A greater number of larger centers were among the least performing practices.

Individual PAs could make a positive difference in the implementation; this was found in larger centers or least performing practices. Usually, one PA was the contact point and took responsibility for devising tasks, giving reminders to deliver the strategy after the smear (Table 6). Additionally, if individual PAs were involved in stop smoking care during other healthcare visits (such as for cardiovascular risk management), this enhanced their self-efficacy and perceived role or task in providing brief advice after the smear (Table 6). This was observed more frequently, but not exclusively, in practices that prioritized prevention or stop smoking care (represented more among the best performing practices). In a few cases, participation in the study encouraged PAs to give stop smoking advice during other healthcare visits. The PA’s enhanced exposure to stop smoking care functioned as a facilitator for maintenance. Similarly, practices or individual PAs who perceived prevention or stopping smoking as their role or task were likelier to proceed with the strategy (Supplementary Table 5). In summary, implementation was influenced by support from the GP, teamwork, and training of PAs, together with PAs’ perceptions of the role or task.

Theme 2: The strategy—Ask & connect and acceptability

Ask & connect

Although the PAs were instructed to deliver the stop smoking strategy after the cervical smear, they implemented it either before or after the smear. For both approaches, PAs aimed to consider the needs of the women, primarily to reassure them and reduce stress (Table 6). This did not clearly differ between the best and least performing practices. Most PAs asked about smoking status (Ask) and either offered an appointment for cessation support or informed women about the possibility of support (Connect). The PAs typically explained why they asked about smoking during the appointment (Table 6). However, most PAs reported skipping the Advise step due to time constraints, lack of skills, or not perceiving it as their task. This was observed more often among the least performing practices (Supplementary Table 5). Some PAs, especially those in the best practices, used example sentences or motivational interviewing techniques during this step. In several best practices, PAs went the extra mile with a more in-depth assessment of the motivation to stop or the use of illustrations during the consultation. Nearly all PAs believed that the Ask step could be easily integrated into the standard questionnaire for cervical screening, thereby facilitating maintenance (Table 6).

Acceptability

Most women considered the smear consultation to be an appropriate moment to talk about smoking. Most women felt that the PA met their needs, although some felt confronted or surprised. Nonetheless, certain women still found the moment to be a trigger to consider cessation and especially appreciated the offer of cessation support in the practice. The PA’s attitude, work experience, or explanation of why smoking was addressed positively impacted acceptability (Table 6). Conversely, a pushy attitude, lack of trust, or tension about the smear itself had a negative impact. A clear preference among women for receiving advice before or after the cervical smear was not observed.

Nearly all PAs reported positive or neutral reactions from women to the strategy (Table 6). However, many PAs discontinued the conversation if they sensed that women were not motivated to quit or not receptive to advice. This was more often observed in the least performing practices. Most, but not all, PAs found the smear consultation to be a legitimate moment to provide brief stop smoking advice (Table 6). Taken together, delivering the Ask and Connect steps seems feasible. Stop smoking advice after the cervical smear is acceptable, according to women and PAs, taking several requirements into consideration.

Theme 3: Context—Low prevalence of smoking and COVID-19

The prevalence of smoking among women visiting their practice for cervical cancer screening was perceived as low by PAs from nearly all practices. For some PAs, this meant that they were not frequently exposed to providing stop smoking advice. This could, in turn, affect the perceived complexity of the intervention, their skills, self-efficacy, or their motivation to (continue to) deliver the strategy (Table 6). Conversely, the low exposure also made the strategy seem relatively simple to integrate into the standard way of working without negatively affecting the consultation time. Additionally, PAs sometimes found it more challenging to provide advice to women who did not smoke daily, as they perceived them to be less motivated or receptive to advice. This was observed a little more frequently in the least performing practices (Supplementary Table 5).

The COVID-19 pandemic restrictions negatively impacted recruitment of women to the study and limited the time and possibility to provide advice after the smear (Table 6). The Dutch government temporarily stopped the national cervical cancer program from March to June 2020. Practices had to discontinue their study participation and found it challenging to resume the strategy afterwards. This applied more often, but not solely, to the least performing practices (Supplementary Table 5). In summary, both the relatively low prevalence of smokers and the COVID-19 pandemic regulations influenced the implementation of brief stop smoking advice after the cervical smear.

Discussion

This comprehensive PE was conducted in parallel with the study of the effect of a brief stop smoking strategy after cervical cancer screening delivered by PAs in Dutch general practice. In total, 47.4% of women smokers experienced the strategy as acceptable, and 25.5% were neutral about it. Acceptability was facilitated by the PA’s attitude, meeting a woman’s needs, explaining the link between smoking and cervical cancer, and offering cessation counseling. The strategy did not influence the intention for future cervical cancer screening.

According to the PAs, asking for smoking status and connecting women to cessation counseling at cervical screening were assessed as feasible. The addition of a question on smoking status in the standard questionnaire for cervical cancer screening could facilitate the integration of this approach into usual care. The low prevalence of smoking and the COVID-19 pandemic negatively influenced the implementation, reach, and maintenance of the strategy. Effective implementation of the strategy was mainly determined by factors at the general practice level and of the individual PA. Implementation facilitators that should be stimulated were training PAs in the AAC steps, deploying PAs in stop smoking care during other consultations, and supportive collaboration with the team and the GP.

This PE has several strengths. The perspectives and experiences of the women who participated in the study and the PAs from the intervention practices were combined. The data of the women who do not smoke, such as future cervical screening intention, could be compared with the findings of women who smoke. To the best of our knowledge, this is the first thorough PE of a behavior change intervention coupled with a cancer screening program10. From the three previous studies that introduced a smoking cessation intervention at cervical screening23,24,25, only Hall’s pilot study assessed feasibility and acceptability24. The main focus of this study was to assess implementation of the strategy. The comparison of best with least performing practices, together with retrospective acceptability, enabled the identification of factors that stimulate effective delivery of stop smoking advice by PAs after smear tests.

This PE also has important limitations. A significant difference between the intervention and control groups could not be demonstrated in the cRCT18. Therefore, this PE could not explore why the SUCCESS stop smoking strategy did or did not work. For years, the workload in Dutch general practices has been high, with many practices facing shortages of staff, including PAs26. During the recruitment of general practices for the SUCCESS study, it is likely that practices with personnel shortages and high workloads did not consider study participation. This might have affected the representativeness of our results for practices facing these challenges. Furthermore, for the in-depth interviews, we sampled both women smokers who did and did not find the strategy acceptable. Nonetheless, it could be that the perspective of women who did not appreciate it was underrepresented.

All measurements were patient-/participant-reported, which might have led to social desirability bias in answering the questionnaires. It was not possible to obtain an objective assessment of how PAs delivered the strategy (for example, via audio or video recordings), as this would have been too intrusive for the women undergoing a cervical smear. The strategy effect on primary and secondary outcomes was published in the effect study paper, together with several measurements collected for the PE, such as the uptake of cessation counseling18. For the effectiveness construct, we did not assess how the strategy stimulated women who smoke to attempt to quit. An exploration of the factors related to the strategy and the context that stimulated women to attempt to quit will be published separately17. The study was conducted in a Dutch primary care context. Therefore, generalizability of the findings could be limited for other primary care or general practice contexts, especially in which PAs or allied health care personnel have another role.

In the cRCT, 73.3% of women who smoke from the intervention group were effectively exposed to the strategy18, indicating that implementation was not delivered as intended for all participants (26.7%). Practice-level barriers, individual PA factors, and COVID-19 restrictions likely contributed to the implementation gap.

Training of PAs and implementing stop smoking care into the PA’s daily routine emerged as a key facilitator for implementation. The higher uptake of cessation counseling in the intervention group (17.2% vs. 2.8%) further underscores the importance of skilled delivery and experience18.

However, PA performance varied, particularly in the least performing practices. While PAs found the Ask and Connect steps feasible, they were less confident in providing advice. This aligns with previous findings that medical assistants are less likely to assess willingness to quit compared to nurses27. Several studies support the need for adequate training and physician support27,28,29. If these requirements are met, PAs can be equally effective in providing stop smoking advice compared to nurses and improving referral rates for smoking cessation28,29,30.

Our quantitative findings showed that around 50% of women reported appreciating the brief stop smoking advice after the smear, while 25.5% were neutral about it, and 12.5% did not appreciate the advice. The interviews with women who smoke added more nuance to these findings. These women believed that it was the PA’s role to address smoking, that confronting advice could also stimulate a quit attempt, and even those who felt confronted appreciated the possibility of being connected to cessation counseling. The prospective acceptability31, which was studied before the trial started, overlapped with the retrospective acceptability21. Stevens et al. reported that people are willing to receive lifestyle advice at cervical, breast, and bowel cancer screening32. Likewise, Hall et al.’s exploratory trial reported that women did not seem deterred from participating in cervical screening after receiving stop smoking advice24. Additionally, in this study, neutrality towards stop smoking advice after the smear did not seem to negatively influence future cervical cancer screening intention.

Our findings indicate that women who smoke and received brief advice after the smear reported an increased perceived risk for cervical cancer as a reason to participate in cervical screening. This creates an opportunity to use follow-up consultations as another teachable moment with women who smoke and who have an aberrant smear test result, as they might be more receptive to advice from the GP or gynecologist33. Smoking status has been associated with lower initial cancer screening participation rates as part of more risky health behavior34. Concerns have been raised that women who smoke might be further discouraged from participating in screening if they might receive stop smoking advice22. A Dutch review on the reasons for (non)adherence to cancer screening reported that direct involvement of the GP results in higher attendance rates of high-risk low-attendance risk groups35. From a preventive perspective, we believe these women would benefit if they were invited to participate in cervical screening by their GP, together with advice and support to stop smoking.

In conclusion, this PE showed that integrating brief stop smoking advice by the PA into routine cervical cancer screening is both feasible and acceptable. Training of PAs, embedding smoking cessation into their daily working routine, and efficient communication and support stimulate implementation. Ensuring an accessible referral pathway for cessation counseling is essential, as acceptability increased when women were offered tailored information and direct connections to cessation support. Importantly, brief advice did not reduce intentions for future cervical cancer screening.

Methods

SUCCESS study design

The SUCCESS study was a type 1 hybrid study consisting of a two-arm cluster randomized trial with a parallel PE conducted in Dutch general practice. The Dutch Ministry of Health, Welfare, and Sport approved the trial following an advisory report from the Health Council (Nr. 2018/17) and granted a license to conduct the study under the Population Screening Act22. The Medical Ethics Committee of the Academic Medical Center, Amsterdam University Medical Centers, the Netherlands, approved the study protocol (2017_263). As a trial cannot fall under both the Population Screening Act and the Medical Research Involving Human Subjects Act, there was no official monitoring from the Medical Ethics Committee. However, standard procedures and regulations on data management by the Medical Ethics Committee of the Academic Medical Center, Amsterdam University Medical Centers, were followed. This study was conducted in accordance with the Declaration of Helsinki.

Participating practices were randomized 1:1 to either the intervention or control condition. In the intervention practices, the SUCCESS study stop smoking strategy based on the AAC method16 was intended to be delivered by PAs to women attending cervical screening directly after their smear had been taken. The results of the SUCCESS cRCT suggested that brief stop smoking advice after a cervical smear might encourage women who smoke to attempt quitting and seek cessation counseling, but a significant effect could not be demonstrated18. Undertaking a serious quit attempt during the 6-month follow-up (primary outcome) did not significantly differ between the women who smoked from the intervention (39.8%) and control groups (36.0%), OR 1.18 (95% confidence interval [CI]: 0.80–1.72, P = 0.41)18. The direction of effects was, however, in favor of the intervention group, and uptake of cessation counseling was higher in the intervention (14.7%) than in the control group (2.8%)18.

The current study consisted of a PE that ran in parallel to the trial using a mixed-methods sequential explanatory design, following recommendations from the Medical Research Council on complex interventions36. A full description of the trial and PE methods was published previously17 (see also the TIDieR checklist in the additional files). No methodological adaptations were made during the COVID-19 period.

This PE study aimed to assess the acceptability and feasibility of the strategy and identify barriers and facilitators to its implementation among women and PAs who participated in the trial, formulating the following research questions: 1) What are the reach, effectiveness, adoption, implementation, and maintenance of the SUCCESS study stop smoking strategy? 2) What barriers and facilitators influence the reach, effectiveness, and implementation of the stop smoking strategy? 3) How do these barriers and facilitators explain differences in reach, effectiveness, and implementation between practices?

Quantitative data were collected based on the RE-AIM (Reach, Effectiveness, Adoption, Implementation, and Maintenance) framework37. The RE-AIM framework facilitates a broad evaluation via assessment of the reach among individual women who smoke, the effectiveness of the intervention, and the adoption, implementation, and maintenance of the strategy in participating general practices37. Qualitative data were collected using the Consolidated Framework for Implementation Research (CFIR)38,39. CFIR constructs can be used for a more thorough understanding of the implementation process based on five different domains associated with effective implementation (e.g., factors at the level of the individual or the organization)38,39. The study protocol details the RE-AIM and CFIR constructs17.

Study participants and recruitment

In participating general practices (the clusters), all women who scheduled a consultation to have their smear taken in the national cervical cancer screening program were invited to participate in a questionnaire study about lifestyle. Participants were eligible for the study when both of the following inclusion criteria were met: i) giving written informed consent for participation in a questionnaire study about lifestyle and ii) undergoing a cervical smear at a practice participating in the study on an invitation from the national cervical screening program organization. Participants were ineligible for the study if they visited the general practice for a cervical smear without an invitation from the national cervical screening program organization. Eligible participants received both verbal and written information about the purpose of the study. We aimed to include women regardless of their motivation to stop smoking or heaviness of smoking and to not influence genuine participant behavior. Therefore, both women who smoke and those who do not smoke were recruited, and details in the informed consent information regarding the stop smoking strategy were omitted to minimize selection bias. Invitations for the cervical smear are sent by the national screening organization; therefore, the number of eligible participants could not be assessed. Due to time constraints, the practices could not register how many women visited the practices for cervical screening or were invited for study participation. At the study commencement, the practices reported performing around 50–300 smears per year, depending on practice size. During the study period, smoking prevalence was around 20% among Dutch women aged 30–60 years. Written consent was obtained prior to participation. In total, 480 women smokers were included at baseline (266 in the intervention; 214 in the control arm in 75 general practices)18. From the 75 practices, 2,358 women nonsmokers were recruited in the SUCCESS study at baseline (1,271 in the intervention group; 1087 in the control group). From the women nonsmokers, a random selection of 802 women (413 in the intervention, 389 in the control arm) was selected for follow-up measurements for the PE.

Procedure SUCCESS stop smoking strategy

In the intervention clusters, the SUCCESS study’s stop smoking strategy was delivered by PAs to all women who underwent a smear for cervical cancer screening. The strategy consisted of the following steps: (1) Ask about smoking status, (2) Advise: provide brief stop smoking advice, and (3) Connect: actively offer an appointment for stop smoking support within the practice. Cessation counseling was not part of the strategy. PAs from intervention practices received training from a certified trainer to deliver the stop smoking strategy in a two-hour single session before the study commenced in their practice, focusing on interviewing techniques. During the training, the PAs were instructed to deliver the strategy after the smear was taken and to create a teachable moment by linking a woman’s behavior (smoking) to the reason for her visit (cancer prevention)17. See Supplementary Table 8 for more details on the content of the PA training. The strategy was not delivered in the control practices, where women only had their smear taken following care as usual.

Quantitative methods and analysis

At baseline, all study participants (both women smokers and nonsmokers) in both study arms completed a baseline lifestyle questionnaire. Demographic variables and behavioral characteristics were collected, including smoking status, physical activity, alcohol consumption, age, educational level, and behavior change intentions40 (Reach).

At the 2-week follow-up, smokers from the intervention group received a questionnaire to assess the acceptability of the stop smoking strategy31,40 (Supplementary Table 9)) (Implementation).

At 6 months, the 480 women smokers at baseline who participated in the trial were sent a questionnaire to collect outcome measurements for the PE: satisfaction with stop smoking counseling41 (Implementation) and intention to consult the general practice for support for a future quit attempt41 (Maintenance) (Supplementary Table 9).

In both women smokers and nonsmokers (as defined at baseline), we measured cervical screening intention at the 6-month follow-up (Supplementary Table 2, supplementary data)32 (Effectiveness), comorbidity, health perception42 (Supplementary Table 9), the result from the cervical smear taken at baseline, and reasons to participate in cervical cancer screening17 (Supplementary Table 2) (Reach). Questionnaires were sent by secure e-mail or post. To assess the smear test results, a simplified measurement was chosen. Women were asked how they remembered their smear test results, categorizing them as either good or not-good, if they tested HPV positive, and if abnormal cells were found. It was not possible to obtain the clinical smear test results because these are collected by the national cancer screening organization. To assess Reach, it was deemed too sensitive to ask women to report their exact medical smear test results.

Intervention practices and PAs

PAs from intervention practices completed a baseline questionnaire and filled out a practice plan prior to study commencement (Implementation). The questionnaire assessed sociodemographic variables, tasks in stop smoking care, and attitudes and experiences towards smoking cessation. The practice plan assessed practice characteristics and the within-practice organization of stop smoking care39. After study completion, a brief follow-up questionnaire was used to assess the perceived role, task, acceptability, feasibility, and maintenance of the strategy37,39,41 (Supplementary Table 9).

Analysis

The baseline characteristics of women smokers have been presented in the effect study paper, compared to the baseline characteristics of women nonsmokers to assess reach. The effect of the stop smoking strategy on smoking cessation outcomes was analyzed in the effect study paper (primary outcome: undertaking a serious quit attempt during 6 months’ follow-up)18. The effect study paper also addressed the adoption of the strategy by the intervention practices (number of intervention practices that started providing the intervention and number of practices dropping out)18. In the current study, we only assessed cervical cancer screening intention to address Effectiveness (as a potential unintended effect of the strategy) with qualitative methods solely17.

For baseline variables, we computed means with standard deviations (continuous variables) and counts with percentages (categorical variables) using IBM SPSS Statistics 28. Similarly, descriptive analyses were used for the data collected at follow-up from the individual study participants and PAs. To enhance interpretability, cervical cancer screening intention, acceptability of post-smear stop smoking advice, and intention to seek cessation support were dichotomized. To explore baseline differences between women who smoke and who do not smoke, we compared the distribution of categorical variables using the Pearson chi-squared test and compared the means of continuous variables using a two-tailed Student t test. The same method was applied to explore differences between women who smoke from both study groups (intervention vs. control) or differences between women who smoke and who do not smoke in measurements collected at the 6-month follow-up. This concerns the variables measured at 6 months used to describe and explore the Reach (comorbidity, health perception, lifestyle changes, smear test results, and reasons to participate in cervical screening). We did not perform analyses to account for cluster variability for these constructs, as these are used to explore Reach and not the stop smoking strategy effectiveness. Due to an (expected) difference in response percentages between women who do smoke and those who do not smoke (more effort was put into collecting follow-up questionnaires from women who smoke for the trial), we present the valid percentages to compare these groups, presenting the number of complete cases alongside each measurement.

To analyze the potential effectiveness of the stop smoking strategy on cervical cancer screening intention, mixed-effects logistic regression models were used, incorporating a fixed effect for the screening intention and random effects for the practices (i.e., clusters). The analysis was performed incorporating the imputation of missing values using MI (i.e., assuming data were missing at random). Multivariate imputation by chained equations with the mice package in R37 was employed to impute missing responses within practices using two-level MI methods. This involved m = 10 imputed datasets and 30 iterations for each dataset. The results from the imputed datasets were pooled using Rubin’s rules to estimate the strategy’s effect and reported as ORs with 95% CIs for both primary and secondary outcomes. First, a comparison between the women who smoke from both study arms was performed. Similarly, cervical cancer screening intention was compared between smokers and nonsmokers, with a random effect for smoking status. Age, education, and ethnicity were incorporated into the multivariate imputation to account for baseline imbalances between women who smoke and those who do not smoke. A random effect for practices was additionally used when comparing smokers and nonsmokers within the treatment arms. The results are presented as ORs with 95% CIs based on the fixed effects. The same approach was used to assess the difference between treatment arms among women who smoke on intention to seek cessation support for the Maintenance construct. Analyses were performed using R (version 4.2.1; R Foundation for Statistical Computing) with the lme4 and ggeffects packages.

Qualitative methods and analysis

To explore participant attitudes and experiences with the strategy, a subset of women who smoke and PAs from intervention practices were interviewed in depth after the study completion. Interview guides were developed using CFIR and findings from qualitative explorations conducted prior to the trial17,21,39,43. To obtain representative data, women who smoke were purposively sampled based on age, having performed a quit attempt, and acceptability with the strategy (as measured at 2 weeks’ follow-up). Practices were sampled based on their performance in the effect study. In total, 56 women and PAs from 24 practices were invited via e-mail or telephone to participate in an in-depth interview.

In-depth interviews were conducted with 28 women who smoke and with PAs from 22 participating practices (the interviewed women were from the practices of interviewed PAs). The interviews were conducted between February 2020 and April 2022, either face-to-face or via telephone (depending on COVID-19 pandemic regulations). The interviews were performed by MM, a female PhD student and GP trainee who is experienced in qualitative methods, and by three different female Master’s medical students, who were each supervised during the first two interviews by MM. The interviews lasted 30–45 minutes. No relationship was established between the interviewers and women who smoke, but in the context of the randomized trial, a relationship (via e-mail for trial progression purposes) was established between MM and a proportion of the PAs before the interviews took place. The function of the interviewer was shared with the interviewees.

Audio recordings of the interviews were transcribed verbatim, reviewed for accuracy, and imported into MAXQDA 2022. We used an iterative process of data collection and analysis. Coding was both inductive and deductive, starting with open coding and gradually developing a code tree based on the RE-AIM and CFIR constructs, adding the interaction between smoker and PA, as well as the acceptability of the intervention. Coding was performed by the three women Master’s students, by MM, and by MC and KvA (senior female researchers with extensive experience in qualitative methods). The framework method44 enabled a comparison between women and practices and the generation of themes. Codes, categories, and themes were discussed between the students, MM, MC, and KvA until consensus was reached.

Integrating qualitative and quantitative data

After separate analyses, the qualitative and quantitative data were integrated. The intervention practices were categorized into best and least performing clusters. Best and least performing clusters were first defined using acceptability. Best performing clusters were defined as practices with ≥60% of what was assessed as positive acceptability (defined as the percentage of women who smoke per cluster who answered ”Appreciated a little” or “Greatly appreciated” to How did you appreciate the stop smoking advice after the cervical smear? measured at 2 weeks’ follow-up.) Least performing practices were defined as practices with ≥30% negative acceptability (and who did not achieve ≥60% positive acceptability), defined as the percentage of women who smoke per cluster who answered “Didn’t appreciate at all” or “Didn’t appreciate very much” to How did you appreciate the stop smoking advice after the cervical smear? After this categorization, the percentage of women unexposed to the strategy and the direction and strength of the random effect of each cluster for the effect study were included (Implementation). In total, 31 of the 37 intervention practices were categorized. We used this categorization to analyze the in-depth interviews.

Finally, to answer our research questions, we assessed the barriers and facilitators to the RE-AIM and CFIR constructs. Then, we analyzed which constructs related to the implementation differed between the best and least performing practices. Finally, three themes were identified that were most important to the implementation process but also related to Reach and Maintenance. These three themes cover the identified barriers and facilitators. We report the results in accordance with the 32-item checklist of the Consolidated Criteria for Reporting Qualitative Research (COREQ) (see additional files)45.