To address the overarching aim of evaluating the effectiveness and feasibility of MI training in primary care, this study used a mixed methods design composed of three components: (1) Evaluation of GP learning outcomes: A before-and-after study assessed changes in GPs’ and GP trainees’ knowledge and confidence related to MI through pre- and posttests; (2) Assessment of patient clinical outcomes: A quasi-experimental, nonrandomized 2-group design compared biomarker changes in patients with diabetes from intervention (MI-trained) and control (routine care) sites; and (3) Qualitative exploration of GP experiences: A descriptive qualitative study using semistructured interviews explored GPs’ experiences in applying MI in routine consultations after training and their experiences in the MI education. GP interviewees were selected based on their confidence in applying MI in routine consultation after their completion of the MI courses. This integrated structure is justified by the implementation nature of the study, aiming to evaluate not only educational effects but also practical application and outcome relevance. The integration is intentional and appropriate for a pragmatic, real-world educational intervention. Improving GP knowledge without understanding its translation into patient care would offer limited insights. Conversely, changes in patient biomarkers without a provider perspective could misattribute causality. Including qualitative data allowed us to contextualize the quantitative results, explain variability in outcomes, and identify barriers and facilitators to implementation.

The sample size was calculated separately for the two primary components of the study: (1) the before-and-after evaluation of GPs’ knowledge and (2) the comparison of patient biomarker outcomes between the intervention and control groups. For the knowledge assessment, the sample size was calculated using the G*Power (version 3.1.9.7; Heinrich-Heine-Universität Düsseldorf) based on the proportion of GP trainees enrolled in diabetes CME [44]. Assuming a power of 80%, a significance level of 0.05, and an effect size of 0.6, a total of 24 participants was sufficient to detect a significant difference in knowledge scores before and after completing the CME courses. For the patient outcome component, the sample size estimation was based on previously published research in which patients with diabetes received MI-guided behavioral change counseling [26]. To detect a 10% effect size with an alpha of .05 and a power of 80%, a total of 241 participants was required. This meant that 121 patients were needed for each of the study and control groups.

Figure 1 shows the flow of participant recruitment for the study. The study was conducted from January 2024 to January 2025.

An interactive 4-hour workshop on MI was offered to GPs and GP trainees who worked at the primary care unit of the study group. The instructors (IT and PN) attended several MI web-based courses provided by William Miller, Stephen Rollnick, and Theresa Moyers. The GPs and GP trainees voluntarily attended the workshop. The content of the workshop included human behavior, introduction to MI, roles of GPs in MI technique, MI spirit, 4 processes of MI, brief MI for busy GPs caring for people with diabetes, MI vs stage of change, and case studies using MI in each stage of change.

An additional educational course on MI was offered to GPs and GP trainees to self-study at their convenience. A structured web-based course on MI and stage of change was designed and developed to provide the foundation for GPs using MI in their routine consultations. The web-based course was developed and designed by incorporating several steps to create effective web-based learning activities, as recommended in a previous study [45]. The intervention, designed to provide GPs with knowledge of MI, was offered for a period of study. The content was designed as text-based plus interaction with text, that is, knowledge-based quizzes, self-practice, and self-reflection activities. The course comprised approximately 4 hours of learning activities for individuals to complete. The course content was developed in a manner similar to the workshop. Notably, participants could access other forms of MI education during the study period. The content of this course could be accessed through our medical school web-based platform free of charge.

A paper version of brief MI guides as a prompt tool for GPs to help people with diabetes change their behavior was available and placed in the GP consultation room at the primary care unit. These guides were developed based on “A Taste of MI” Exercise from the Motivational Interviewing–Foundational web-based course by William Miller, Stephen Rollnick, and Theresa Moyers (see the English translated version of brief MI guides in Multimedia Appendix 1).

Semistructured interviews were also conducted. The GPs and GP trainees who completed the pretest were invited to participate in semistructured interviews. A small sample of 10 GPs and GP trainees was sought for convenience. A total of 5 GPs or GP trainees from each of the following groups were sought: GPs and GP trainees who reported having less confidence in applying MI in their routine consultation and those who reported having confidence in applying MI in their routine consultation. The interviews were conducted in September 2024 via Zoom (Zoom Communications) and were recorded for documentation. Each session lasted approximately 15 minutes. The audio recordings were transferred to a computer for repeated listening and detailed content review. The recordings were transcribed into Microsoft Word documents, and they were confirmed with participants to ensure accuracy and for further analysis.

IBM SPSS 19 for Windows (version 20.0) was used for statistical analysis, and a pairwise deletion strategy was applied to handle missing data. Paired-sample t tests were applied to assess within-group differences between pre- and postintervention measurements, and independent-sample t tests were used to compare outcomes between the study and control groups. Baseline characteristics were descriptively compared across study and control groups; however, formal statistical adjustments for potential confounders (eg, baseline differences in patient demographics or clinical characteristics) were not performed. This decision was based on the study’s design focus on population-level comparisons and the available sample size, which limited the feasibility of multivariable analyses. Descriptive statistics were used to describe demographic data. The participants’ responses on Likert scales to a 10-item questionnaire on their confidence in applying MI for diabetes management were tallied, creating a confidence score range of 10 to 50. A paired-samples t test was used to compare the mean differences in the knowledge scores between the two tests.

Interview transcripts were analyzed using conventional content analysis. An inductive coding approach was applied, whereby initial codes were generated from repeated reading of transcripts. These codes were then grouped into categories, and overarching themes were derived by identifying recurring patterns and relationships among the data. Coding and analysis were conducted manually by a single trained coder (IT). Although intercoder reliability was not assessed, the coding process was reviewed by members of the research team to enhance consistency and credibility. The codes were then organized into themes, identifying the relationships and connections between them.

The quantitative and qualitative findings were synthesized during the interpretation phase to provide a comprehensive understanding of the intervention’s impact. Quantitative data provided objective evidence of improvements in knowledge and patient outcomes, while qualitative insights offered contextual understanding of how GPs applied MI, the challenges encountered, and their suggestions for improvement. This integration supported triangulation of results and enhanced the depth and relevance of the study’s conclusions.

This study was approved by the Human Research Ethics Committee of Khon Kaen University (project HE631031). GPs and GP trainees were recruited through the researcher’s assistant, who invited volunteers to participate. Before completing the questionnaires, participants were informed that participation was voluntary and that they could drop out of the study at any time. They were informed that their opinions were important for enhancing diabetes practice in primary care and were therefore encouraged to express them. All GPs and GP trainees voluntarily agreed to participate without receiving compensation. The participants’ privacy and identity were protected, and confidentiality was assured. Informed consent was obtained from each participant who participated in the semi-structured interviews. The study objectives were explained to the participants and the study was conducted according to the academic ethical code.

For patients, individual consent was not required because all clinical outcome data were extracted from anonymized, routinely collected health records, in accordance with institutional ethical guidelines. No experimental procedures or additional data collection were imposed on patients. After the study concluded, the brief MI materials and web-based MI course were shared and offered to clinicians in the control group as part of standard CPD dissemination. This approach ensured that no participants were denied care and aligns with accepted ethical principles for real-world implementation research.

The MI workshop was attended by 32 GPs and GP trainees out of a total 35 participants (91%). A total of 2 participants out of 38 had previous training in MI (n=2, 5%). While both GPs and GP trainees participated in the educational interventions (n=32), only 28 GP trainees (87%) completed the pre- and posttests to assess their knowledge, and later 10 GP trainees provided feedback through semistructured interviews, whereas no GPs participated in these assessments (see Table 1).

Data from the web-based learning management system (LMS; (updated on January 21, 2025) showed that there were 35 GPs and GP trainees (100%) enrolled in the structured web-based MI courses, 14/35 (40%) completed the courses, 5/35 (14%) had not yet started the lesson, and 16/35 (46%) had started but had not yet completed. The course had 21 lessons. Participants’ enrollment time in the course was a median of 3 minutes (range 0-416 min), with a mean of 68.54 (SD 108.82). The number of completed lessons by participants was a median of 2 lessons (range 0-21), with a mean of 6.46 (SD 8.79).

A paired-sample t test was conducted to evaluate the impact of the MI course on learners’ knowledge. There was a statistically significant improvement in knowledge test scores from Time 1 (mean 11.46, SD 3.48) to Time 2 (mean 15.04, SD 2.35), t₂₈=7.74; P<.001 (2-tailed). The mean increase in knowledge test score was 3.57 (SD 2.44), 95% CI 2.62-4.52. Notably, overlapping pre- and postintervention CIs do not invalidate the finding, as this was a within-subjects comparison. The eta-squared statistic indicated a large effect size (eta-squared=0.85). Table 2 shows participants’ confidence levels in applying MI techniques after completing the course. The highest confidence was reported for the focusing process (mean 3.86, SD 0.65) and using affirmations (mean 3.79, SD 0.63). Other skills, such as using open-ended questions, planning, and inviting change talks, scored slightly lower but consistently reflected moderate to high confidence levels, indicating the course’s effectiveness in enhancing practical MI skills.

Table 3 provides biomarker outcomes for patients with diabetes over 6 months after GP trainees completed the MI workshop. In the study group, there were significant improvements, including reductions in HbA_1c (mean difference 0.35, 95% CI 0.03-0.66; P=.03), body weight (mean difference 1.11, 95% CI 0.27-1.95; P=.01), triglyceride (mean difference 14.18, 95% CI 1.69-26.67; P=.02), systolic blood pressure (mean difference 5.14, 95% CI 1.06-9.03; P=.01) and diastolic blood pressure (mean difference 3.82, 95% CI 1.66-5.99; P=.001). Between-group analysis showed that the study group had significantly better outcomes for HbA_1c (mean difference −0.50, 95% CI −0.91 to −0.09; P=.02) and diastolic blood pressure levels (mean difference −5.96, 95% CI −8.66 to −3.25; P<.001) compared to the control group.

Table 4 summarizes the participants’ expectations before enrolling in the MI workshop and how these expectations were met after the workshop. Participants reported improvement in skills and knowledge about MI, with mean scores increasing from 3.96 to 4.07 and from 3.93 to 4.14, respectively. In addition, the learning methods, venue, and learning period received positive ratings postcourse, highlighting the course’s effectiveness in meeting expectations.

Table 5 summarizes the interviews with participating GP trainees regarding their experience in implementing MI during their routine consultation. The GP trainees found MI useful for engaging patients with diabetes, especially in brief MI formats.

All participants reported that MI techniques were used in patient interactions, particularly in cases of diabetes, hypertension, and dyslipidemia. The engaging and evoking processes were most frequently used, while planning and responding to sustain talk were less commonly applied.

However, many participants noted difficulty in applying the evoking and planning processes.

Challenges also included time constraints and unfamiliarity with the MI approach.

Despite these obstacles, participants valued the patient-centered nature of MI.

This study evaluated the implementation of a combined educational intervention on MI for GPs and GP trainees involved in type 2 diabetes care. The findings revealed significant improvements in participants’ MI knowledge, confidence, and related patient outcomes. Furthermore, participants reported positive experiences with the training and perceived enhancements in their communication skills and patient care practices.

All GPs and GP trainees in the study group completed the interactive MI workshop, but only GP trainees participated in knowledge pre- and posttests, and the latter half of GP trainees completed the structured web-based courses. The study demonstrated that MI education provided for GPs and GP trainees in interactive workshop formats significantly improved their knowledge, and together with additional self-learning through structured web-based courses and brief MI guides, had a measurable impact on the patient outcomes. In terms of patient outcomes, the MI-trained group showed improvements in key health metrics, such as HbA_1c and diastolic blood pressure, compared to the control group, emphasizing the practical benefits of MI training in clinical settings. Semistructured interviews revealed that GP trainees valued MI for its patient-centered approach but faced challenges related to its time demands and their own proficiency in its processes. However, this study was conducted in a single study area; therefore, the results should be interpreted with caution.

This study contributes to the literature on CME and the application of behavioral science in clinical practice. It addresses the gap between theoretical knowledge of MI and its practical application by GPs and trainees in routine care. Through the design of a structured, multifaceted CME intervention—comprising an interactive workshop, web-based learning, and practical tools—this study illustrates how behavioral science principles can be effectively translated into real-world practice. The findings may also guide the development of future training programs to enhance patient-centered communication among healthcare providers.

GP trainees who underwent MI training demonstrated a significant improvement in their knowledge scores. The educational program successfully met its objective of enhancing knowledge among GP participants, aligning with previous studies that highlight the potential of MI education to improve both knowledge and confidence in delivering behavior-change interventions [17,31].

Knowledge assessments were conducted before and immediately after the workshop, indicating an immediate knowledge gain. This finding is consistent with previous studies on the effectiveness of MI training in primary care, which have demonstrated improvements in both knowledge acquisition and skill application [17,46,47]. Given that knowledge was assessed immediately after the workshop, the observed improvement could be attributed primarily to the interactive workshop component. However, following the workshop, participants also had access to web-based resources and MI guides. The combination of these educational methods aligns with existing evidence demonstrating that multifaceted educational interventions effectively enhance not only knowledge but also clinical practices and patient outcomes [34,35].

The educational content covered MI processes and their practical application through case studies in accordance with the frameworks established by Miller and Rollnick [11]. Notably, all participants who took the knowledge pre- and posttests had no previous MI training, which aligned with their self-reported learning expectations from the pretraining questionnaire. This targeted approach contributed to the successful development and implementation of an effective educational program as it addressed learners’ specific needs [44].

Participants’ feedback indicated that fundamental MI processes such as engaging, evoking, and planning were effectively taught. However, more advanced skills, particularly responding to change talk, remain challenging, an issue frequently noted in MI training [30,48]. Participants also reported that this was their first exposure to MI, and that the combination of workshop-based learning, structured web-based modules, and brief MI guides was sufficient for acquiring foundational MI knowledge.

In addition to knowledge acquisition, the training significantly enhanced participants’ confidence in applying MI techniques, particularly in the practical aspects of brief MI. As shown in Table 3, participants reported the highest confidence in performing the focusing process and using affirmations, both essential elements of patient-centered communication. Other MI skills—including open-ended questioning, planning, and inviting change talks—also received moderate to high confidence ratings. These findings support the effectiveness of the combined educational interventions in building participants’ self-efficacy for using MI in clinical practice. The pattern of reported confidence aligns with the stages of skill development, where foundational processes such as engaging and focusing are typically mastered earlier, while more advanced tasks like evoking and planning require further practice and reinforcement. This confidence boost is crucial, as self-efficacy is strongly linked to successful behavioral change counseling and continued use of MI techniques in routine care.

Patient outcomes in this study were analyzed at the unit level. Patients with diabetes treated by MI-trained GPs and GP trainees demonstrated significant improvements in HbA_1c levels, body weight, triglyceride levels, systolic blood pressure, and diastolic blood pressure when comparing pre- and posteducational interventions and MI implementation. However, only HbA_1c levels and diastolic blood pressure showed statistically significant differences between the study and control groups after the educational interventions and MI implementation. Notably, although the study was powered to detect a 10% change in key clinical outcomes, the observed changes in most patient biomarkers—including HbA_1c, blood pressure, and lipid levels—were smaller than this threshold. While some changes reached statistical significance, the study may have been underpowered to detect modest effects, and tests of significance should be interpreted with caution. These findings suggest that knowledge gained from a single training session, supplemented by self-learning through the provided resources, can be effectively applied in clinical practice, leading to measurable improvements in diabetes patient outcomes.

Although few studies have examined MI delivered by GPs for diabetes patient outcomes, this study aligns with previous research demonstrating MI’s mixed but promising efficacy in GP-led diabetes management [17]. Similar to the current findings, previous studies have reported improvements in HbA_1c levels between the study and control groups [49,50], and reductions in diastolic blood pressure [49].

This study found improvements in systolic blood pressure within both groups, but no statistically significant difference between them, which is consistent with previous findings [51]. A similar pattern was observed for triglyceride levels [51]. Notably, the study also found a positive effect on body weight in both groups; however, the control group showed a statistically greater reduction than the study group. In contrast, a previous study reported no significant impact on body weight [52]. Despite the variability in effects of MI on patient outcomes, these findings contribute to a growing body of evidence supporting the potential of MI when implemented by GPs in diabetes care.

Semistructured interviews with GP trainees in this study reinforced the quantitative findings, as participants reported that MI facilitated better patient engagement and empowered individuals to make sustainable behavioral changes. Although many GP trainees did not complete all 4 MI processes with patients, the majority implemented brief MI guides during diabetes consultations. Participants most frequently applied the engaging and evoking processes, which may help explain the observed improvements in HbA_1c and diastolic blood pressure—both outcomes known to be influenced by enhanced patient motivation and communication. Even limited MI skills, such as empathic listening and goal-focused dialog, have been perceived as effective in fostering patient engagement, supporting the feasibility of brief MI formats in high-volume primary care settings. However, some clinical metrics, such as triglyceride levels and systolic blood pressure, have shown mixed results, highlighting the need for further research on MI uptake and its long-term impact, particularly regarding the sustained effects of brief interventions. This variability aligns with a systematic review that identified mixed yet promising results in MI interventions, underscoring the importance of more consistent training and implementation to improve outcomes [17].

While the observed improvements in HbA_1c and diastolic blood pressure among patients in the MI-trained group are promising, it is important to interpret these findings with caution. HbA_1c is a valid and widely used clinical outcome; however, it can be influenced by a variety of factors beyond the educational intervention, such as medication adherence, lifestyle changes, appointment attendance, or changes in health service delivery. Since we did not measure intermediate outcomes such as patient adherence, satisfaction, or visit frequency, the attribution of clinical improvements solely to MI training may be limited. Future studies should incorporate additional process measures to more fully explore the mechanisms by which MI training may impact patient outcomes. We also acknowledge limitations related to the interpretation of lipid profile outcomes. Variability in LDL and total cholesterol levels across groups may reflect unmeasured factors such as baseline statin use, medication adherence, and dietary or genetic differences. These variables were not adjusted for in this analysis. We recommend that future research include medication adherence monitoring and baseline lipid-lowering therapy documentation to better account for such confounders in lipid-related outcomes.

In addition to statistical significance, the improvements observed in certain outcomes appear to be clinically meaningful. Patients managed by MI-trained GPs and GP trainees demonstrated significant reductions in HbA_1c and diastolic blood pressure compared to the control group. These changes are important in diabetes management, as even modest reductions in HbA_1c and blood pressure have been linked to decreased risk of microvascular and cardiovascular complications. While improvements in other parameters, such as systolic blood pressure and triglycerides, were mixed or nonsignificant between groups, the observed HbA_1c and diastolic blood pressure reductions suggest that gains in MI-related knowledge and confidence may have translated into tangible patient benefits. This underscores the potential for brief MI training, when implemented in routine primary care, to yield clinically relevant outcomes.

MI requires complex skills [53]. Effective MI implementation necessitates comprehensive training programs that integrate didactic presentations, experiential exercises, and role-playing to develop both basic and advanced MI skills. This study used these combined educational methods. In addition, these educational methods encourage practitioners to progress toward becoming trainers themselves [54]. Evidence suggests that practical applications and expert feedback are crucial for developing MI proficiency [46]. Participants in this study expressed a strong need for coaching or observational support from experts who could demonstrate MI in real consultations, aligning with research indicating that mentorship significantly enhances MI skills [55].

The GPs and GP trainees in this study were trained in the full MI process for comprehensive consultations. However, time constraints in Thai primary care settings make full-session MI difficult to implement. This issue has been noted in other health care settings, where time limitations hinder MI adoption [17]. To address this, brief MI guides were introduced, enabling GPs and GP trainees to apply MI techniques to patients with diabetes within typical consultation durations. Previous studies have suggested that brief MI can be an effective alternative in high-volume practices while maintaining its patient-centered benefits [30]. Participants from this study who had foundational MI knowledge from the workshop found brief MI guides particularly useful for integrating MI into clinical practice, aligning with research that highlights the importance of initial training before adopting streamlined approaches [26].

Qualitative data revealed mixed experiences with the implementation of MI. While GP trainees valued MI’s structured, patient-centered approach, they faced barriers such as time constraints, difficulty in mastering advanced techniques, and limited follow-up opportunities due to rotational training schedules. The engaging and evoking phases were widely used, but many GP trainees struggled with the planning phase, a challenge documented in previous MI research [17]. Participants also noted that MI’s emphasis on empathic listening and reflective questioning extended consultation times, making it difficult to sustain in high-volume clinical settings [30]. Although brief MI guides were designed to fit within standard consultation durations, some GP trainees reported that a lack of confidence in advanced MI techniques and incomplete mastery of MI skills could inadvertently prolong consultations. This aligns with findings from previous studies, which highlight that practitioners in the early MI training phases may take longer to implement MI effectively [47,56].

Despite these challenges, all participants acknowledged the benefits of MI, particularly brief MI, which they found more manageable within their workloads. Most reported using brief MI in nearly all encounters with patients with diabetes, while recognizing the need for continued practice to enhance their confidence and further refine their skills.

Our findings align with previous literature that supports the potential of MI to improve clinical outcomes in patients with type 2 diabetes when delivered by primary care providers [17,49,50]. Similar to previous studies, we observed a significant reduction in HbA_1c and diastolic blood pressure following MI training [17,49], reinforcing the value of even brief MI interventions when embedded in routine consultations. However, as reported in earlier work [17,30,48], our participants also faced challenges in mastering advanced MI processes such as evoking and planning. This underscores the need for structured, staged training that starts with foundational MI concepts and provides opportunities for progressive skill development.

In line with recommendations from systematic reviews on effective MI training [13,53,55], our study supports the implementation of multifaceted educational strategies. These include combining in-person workshops with asynchronous web-based learning, distributing practical tools (eg, brief MI guides), and offering ongoing peer coaching or expert feedback. Such approaches help accommodate varying learning styles and time constraints, particularly in high-volume clinical settings. Based on our findings, we recommend that MI training programs for GPs incorporate brief MI frameworks early in the training, followed by reinforcement through mentorship, peer observation, or reflective practice sessions to sustain skill application over time. Tailoring training to local practice constraints—such as patient volume, consultation duration, and provider rotation—can enhance feasibility and long-term adoption.

This study has several strengths and limitations. The study design combining quantitative and qualitative approaches enriched the understanding of MI’s implementation in real-world practice and its impact on GPs and their patients, offering both measurable outcomes and contextual insights. The study’s focus on routine practice settings enhances the relevance and applicability of the findings to primary care environments. In addition, the pre- and posttraining assessments provided robust evidence of the effectiveness of training in improving knowledge of GPs.

While this study incorporated a control group to compare patient outcomes, we acknowledge that the nonrandomized design limits the ability to draw strong causal inferences regarding the effectiveness of the MI intervention. The control group consisted of primary care units in another province where GPs did not receive MI training and thus served as a comparator for usual care. Although this allowed for between-group comparisons, differences in context, provider characteristics, or patient demographics may have introduced selection bias or residual confounding. To minimize these biases, control sites were selected based on similarities in service structure, population size, and staffing (ie, family doctors in community-based settings). In addition, the use of population-level data from primary care catchment areas helped reduce provider-specific effects and reflect routine care delivery more accurately. However, unmeasured differences—such as provider motivation, local health system factors, or patient case mix—may have influenced the results. Therefore, findings should be interpreted with caution and viewed as indicative of association rather than definitive evidence of causality. Future studies employing randomized controlled designs or matched comparison groups with standardized training exposures and comparing the observed changes with broader regional trends in diabetes outcomes are recommended to strengthen causal conclusions and improve generalizability.

This study has many limitations. First, the study did not include a control group among GPs or GP trainees for the educational intervention, limiting our ability to attribute changes in learning outcomes solely to the MI training. Comparisons were made only within the intervention group before and after the training. Second, the sample size of GP trainees who completed both pre- and postassessments was modest, which may affect statistical power and limit the robustness of subgroup analyses. Third, the use of self-reported measures for MI confidence and qualitative interviews introduces potential bias, as participants may have overestimated their skills or provided socially desirable responses. Fourth, the relatively short follow-up period of 6 months for assessing clinical outcomes such as HbA_1c and blood pressure. While this timeframe allowed us to evaluate initial effects of MI training and implementation, it may be insufficient to capture longer-term behavioral change and sustained clinical benefits. In addition, while the study demonstrated statistically significant improvements in certain diabetes outcomes (eg, HbA_1c and diastolic blood pressure), the magnitude of these changes was modest. This may be due in part to the relatively short duration and intensity of the MI training provided. A single 4-hour workshop, even when supplemented by optional web-based modules, may not be sufficient to fully develop and sustain advanced MI competencies. Furthermore, qualitative findings indicated that GPs and trainees faced several implementation barriers, including time constraints, limited patient continuity, and lack of confidence in using more advanced MI processes such as evoking and planning. These challenges likely impacted the consistency and depth of MI delivery. Fifth, despite having a control group of patients with diabetes, the study sample consisted of a homogeneous group of GPs, GP trainees, and patients from a single area, potentially limiting the generalizability of the findings to broader populations. Another limitation is that no formal statistical adjustments were made for potential confounders, such as baseline differences in patient characteristics between the study and control groups. Although descriptive comparisons were conducted, the absence of adjusted analyses means residual confounding cannot be ruled out. Future studies using larger sample sizes and multivariable analytical approaches are warranted to strengthen the robustness of causal inferences. Another limitation is the absence of data on intermediate behavioral indicators—such as patient adherence to lifestyle recommendations, medication compliance, frequency of follow-up visits, or satisfaction with care—which may mediate or moderate the observed effects on HbA_1c and blood pressure. Including such measures in future studies would provide a more comprehensive understanding of the impact and mechanisms of MI in real-world primary care settings. Sixth, a further limitation relates to the qualitative analysis process. Thematic derivation was based on conventional content analysis using an inductive coding approach, whereby initial codes were generated from repeated readings of the transcripts and subsequently organized into categories and overarching themes. However, coding was undertaken by a single trained coder, and intercoder reliability was not formally assessed. Although the coding process followed established qualitative research methods and was reviewed for internal consistency by members of the research team, the absence of formal intercoder reliability testing may have introduced subjectivity in theme interpretation. Finally, GP trainees’ inability to follow-up with the same patients owing to rotational schedules reduced the opportunity to apply MI processes comprehensively.

MI may assist in addressing behavioral and relational barriers to diabetes care—such as patient ambivalence, low motivation, or resistance to change—we acknowledge that MI alone cannot resolve the full range of complex challenges associated with chronic disease management. Structural issues, resource constraints, and system-level determinants also contribute significantly to outcomes. Thus, MI should be viewed as a single component of a broader multifaceted strategy aimed at enhancing communication, empowering patients, and supporting behavioral change.

While MI is often perceived as time-consuming, the use of brief MI, as emphasized in our training, can actually help structure consultations more efficiently. Several participants reported that once familiar with brief MI, they were able to guide patient discussions more purposefully, avoid redundant advice-giving, and promote greater patient involvement in setting goals. Over time, this may reduce the burden on clinicians by improving communication flow and minimizing repeated discussions of unresolved issues in future visits. Thus, brief MI may mitigate, rather than exacerbate, the challenges posed by time-limited practice environments. The development of brief MI protocols tailored to high-volume clinical settings could enhance feasibility without compromising effectiveness. MI’s emphasis on patient autonomy and behavioral change aligns with modern health care priorities, making it a valuable tool in chronic disease management. To support GPs in integrating MI into their daily practices, it is essential to provide continued education, peer coaching, and expert supervision. Policymakers should consider supporting MI training as part of CPD programs, given its demonstrated benefits in improving patient outcomes.

Further studies should explore the key components of brief MI that can retain the effectiveness of full-process MI while remaining practical in busy clinical environments. In addition, future studies should include longer follow-up periods to assess the sustained impact of MI on both GPs’ skills retention and patient outcomes over time. Research should also focus on optimizing MI training approaches by identifying effective educational methods suited to GPs, as well as strategies to increase uptake and sustained implementation of MI in routine practice. The potential of digital tools, such as mobile apps or web-based coaching platforms, to support MI training and delivery warrants further exploration to enhance accessibility and long-term integration into primary care.

This study underscores the transformative potential of MI in primary care, particularly in diabetes management. MI training not only enhanced GP trainees’ knowledge, but also translated into tangible improvements in patient health metrics. Brief MI is an alternative if one cannot perform a full session of MI and better suits the busy schedule of GPs. However, challenges such as time demands, skill mastery, and patient continuity must be addressed to maximize its usability. By refining training methods, supporting GPs in their implementation efforts, and exploring innovative applications of MI, health care systems can better harness its benefits to improve patient outcomes and advance the quality of care.