Structured Microskill Stepwise Progression for Undergraduate Tooth Extraction: A Randomized Mixed-Methods Study Authors: Zongxi Wu, Mingyue Wang, Xueting Luo, Xiangqi Liu, Le Yang, Siyong Gao, Sien Zhang, Guangsen Zheng Published: 2026 ABSTRACT AbstractObjectivesTo evaluate whether a structured microskill stepwise progression model improves procedural competence and reduces performance variability in undergraduate tooth extraction training compared with traditional apprenticeship-based teaching.MethodsThis randomized mixed-methods pilot study allocated 40 fifth-year dental students (1:1) to traditional apprenticeship teaching or a microskill stepwise progression model during an 8-week oral surgery rotation. The intervention divided tooth extraction into nine stages, requiring five targeted observations and five successful supervised performances before progression. End-of-rotation competence was assessed using a modified Direct Observation of Procedural Skills (DOPS) scale by two blinded examiners. Between-group differences and score variability were analysed using Welch’s t test and the F-test. Questionnaires and qualitative feedback explored learner and instructor perceptions.ResultsBaseline theoretical examination scores were comparable between groups. Inter-rater reliability for the total modified DOPS scores was good (ICC = 0.86; 95% CI, 0.74-0.93; P < .001). The intervention group achieved higher modified DOPS scores than the control group (94.60 ± 3.97 vs 88.25 ± 6.41; mean difference, 6.35; 95% CI, 2.94-9.76; P < .001; Cohen’s d = 1.19) and lower score variability (F = 2.616; P = .042), with a narrower score range (87-100 vs 75-96). Qualitative findings suggested clearer progression and more consistent competence development. Questionnaire findings were exploratory; four item-level differences remained significant after Benjamini–Hochberg correction.ConclusionsThe microskill stepwise model was associated with improved tooth extraction performance and reduced variability in assessed procedural competence compared with traditional teaching. This mastery-oriented approach may support more structured supervision and more consistent educational readiness in oral surgery training.Clinical relevanceThis framework helps educators organize undergraduate tooth extraction training using procedure-specific stages, supervised-performance thresholds, and documented progression criteria. It may promote consistent standards and transparent entrustment decisions, but should be adapted to local case availability and supervisor workload.Key words: Clinical competence, Dental education, Direct Observation of Procedural Skills, Tooth extraction, Undergraduate dental students Introduction Ensuring that dental graduates can safely perform common invasive procedures is a central component of oral health workforce readiness and patient safety.1 Tooth extraction remains one of the most frequently performed procedures in general dental practice, but it requires the integration of anatomical knowledge, case selection, local anaesthesia, instrument handling, surgical judgment, doctor-patient communication, and complication management.2,3 For undergraduate dental students, the transition from preclinical learning to supervised clinical practice is therefore particularly challenging. In China, dental students typically complete systematic theoretical coursework in oral and maxillofacial surgery during the fourth year and enter formal clinical clerkships in the fifth year. During this transition, students are expected to apply previously acquired theoretical knowledge to real patient care within a relatively short period. Without appropriate instructional scaffolding, this rapid knowledge-to-practice transfer may impose excessive cognitive load on novices and compromise both learning efficiency and clinical confidence.4 Traditional clinical training in tooth extraction is commonly based on an apprenticeship model,5 in which students first observe complete procedures and then gradually attempt selected clinical cases under supervision. Although this model provides exposure to authentic clinical practice, it often lacks explicit progression criteria and objective evidence for entrustment. Supervisors may rely heavily on subjective impressions when deciding whether a student is ready to perform a procedure or manage a more difficult case. Such informal judgments, although grounded in clinical experience, can be influenced by variability in teaching style, case availability, supervisor expectations, and students’ early performance.6 As a result, students may receive unequal learning opportunities, and some may be exposed to cases that exceed their current level of competence. For novice learners, premature engagement in complex procedures can increase cognitive burden, reduce procedural fluency, require supervisor takeover, and negatively affect task-specific self-efficacy.7 A further limitation of unstructured apprenticeship-based training is the potential widening of competence differences within the same cohort. This phenomenon is conceptually consistent with the ‘Matthew Effect’, whereby small initial differences in performance or learning opportunities may be amplified over time, resulting in cumulative advantage or disadvantage.8 Students who perform well in early procedures may gain more trust, confidence, and opportunities for further practice, whereas students who experience early difficulty may become more hesitant and receive fewer operative opportunities. This cumulative advantage or disadvantage may contribute to substantial variability in clinical competence at the end of a rotation. From the perspective of undergraduate dental education, reducing such performance variability is important because graduation readiness requires not only high average competence but also a consistent minimum level of procedural ability across trainees. Simulation-based9 and structured teaching approaches have been increasingly incorporated into dental education to support the acquisition of complex clinical skills.10 Existing studies suggest that tooth-extraction model training can improve students’ perceived readiness for clinical practice and comfort with specific technical procedures. However, improvements in learner confidence and anxiety have been inconsistent, indicating that technical preparedness and affective outcomes may respond differently to educational interventions. Randomized studies in dental and oral and maxillofacial surgery education have examined specific instructional components, such as direct expert feedback and individualized or unsupervised video feedback, for teaching procedural skills.11 Nevertheless, much of the broader evidence remains derived from cohort comparisons, pre–post studies, or questionnaire-based evaluations, and randomized evidence specifically evaluating structured, stepwise progression for undergraduate oral surgery procedures remains limited. In particular, it remains unclear whether predefined progression criteria and staged supervised practice can improve objectively assessed competence in tooth extraction. The present study addresses this gap by evaluating the effectiveness of a structured progressive teaching model for undergraduate tooth extraction using a randomized controlled design. Competency-Based Medical Education and mastery learning provide useful principles for addressing these challenges. Both approaches emphasize explicit learning outcomes, progressive skill acquisition, repeated practice, feedback, and advancement based on demonstrated competence rather than time or exposure alone.12 However, evidence remains limited regarding how these principles can be operationalized in authentic undergraduate dental clinics for high-risk invasive procedures such as tooth extraction. In particular, there is a need for practical models that deconstruct complex dental procedures into teachable microskills, define transparent criteria for progression, support supervisor entrustment decisions, and protect patient safety while allowing students to develop autonomy. To address this gap, we developed a structured microskill stepwise progression model for undergraduate tooth extraction training. In this model, the tooth extraction process was deconstructed by senior oral and maxillofacial surgery educators into nine sequential stages of increasing complexity. Students were required to complete targeted observation and supervised practice at each stage before progressing to the next level. Specifically, progression required five successful supervised performances at the current stage and five targeted observations of the subsequent stage. Students who did not meet the progression criteria received additional reinforcement training. This design was intended to provide objective evidence for staged entrustment, reduce premature exposure to overly complex cases, and promote gradual development of procedural confidence and competence in real clinical settings. This randomized mixed-methods study aimed to evaluate whether the microskill stepwise progression model could improve undergraduate dental students’ tooth extraction competence compared with traditional apprenticeship-based training. We also examined whether the model reduced variability in performance, enhanced students’ perceived readiness and confidence, and provided a feasible framework for standardizing clinical training in invasive dental procedures. We hypothesized that a structured, mastery-oriented progression pathway would improve procedural performance and reduce competence heterogeneity among undergraduate dental trainees. Methods Study design and participants This prospective, randomized, mixed-methods pilot study was conducted among fifth-year undergraduate dental students enrolled in their first oral and maxillofacial surgery clinical rotation at the Hospital of Stomatology, Sun Yat-sen University. Before enrolment, all participants had completed the same undergraduate theoretical course in oral and maxillofacial surgery. No participant had received formal clinical training in tooth extraction or performed tooth extraction independently before enrolment. A total of 40 students from the Class of 2021 were included and allocated in a 1:1 ratio to either the traditional apprenticeship-based teaching group, referred to as the control group, or the microskill stepwise progression model group, referred to as the experimental group. Each group comprised 20 students. As this was a pilot educational intervention study conducted within one clinical rotation, no formal a priori sample size calculation was performed. The sample size was determined by the number of eligible students in the rotation. To characterize the detectable effect size of the pilot sample, a supplementary posthoc power analysis was conducted based on the primary outcome, namely the end-of-rotation clinical skill assessment score. The effect size was calculated as Cohen’s d using the observed mean difference between groups and the pooled standard deviation. With a two-sided alpha level of 0.05 and the actual sample size of 20 students per group, the achieved statistical power was estimated using G*Power version 3.1.9.7. In addition, a sensitivity analysis was performed to determine the minimum detectable effect size that could be identified with 80% power under the same alpha level and allocation ratio. Before the start of the clinical rotation, students were randomly allocated to the two groups in a 1:1 ratio using a lottery method. The allocation sequence was generated by an independent administrative staff member who was not involved in teaching, assessment, or data analysis. Specifically, 20 identical lots labelled ‘intervention’ and 20 identical lots labelled ‘control’ were prepared, folded to be indistinguishable, thoroughly mixed, and drawn sequentially to create the allocation order. The assignments were then placed in sequentially numbered, opaque, sealed envelopes that were identical in appearance. Each envelope was opened only after the corresponding student had been enrolled and baseline information had been recorded. The random allocation was completed before clinical teaching commenced, and the independent staff member managed the allocation process to minimize the possibility that group assignment could be predicted or influenced. Owing to the nature of the educational intervention, blinding of students and instructors was not feasible. However, the end-of-rotation clinical assessments were conducted by independent examiners who were not involved in the teaching process and were blinded to group allocation. To reduce potential variation related to individual teaching styles, clinical teaching for both groups was delivered by the same two senior attending physicians with experience in oral and maxillofacial surgery education. Before the study, the instructors participated in calibration sessions to standardize teaching procedures, assessment expectations, and progression criteria for each study arm. Measures were taken to avoid cross-contamination between the two teaching approaches, including separate teaching records and group-specific progression documentation. Before clinical onboarding, all participating students had taken a standardized closed-book final examination after completing the required theoretical course in oral and maxillofacial surgery. Baseline theoretical examination scores were compared between the two groups to assess comparability in preintervention theoretical knowledge. The mean theoretical examination score was 84.45 ± 1.72 in the experimental group and 83.25 ± 1.50 in the control group (Figure 1). There was no statistically significant difference between the groups, suggesting similar baseline theoretical knowledge before entry into the clinical environment. Comparison of baseline theoretical examination scores between the two cohorts (ns, P > .05). This study was reported according to the CONSORT statement for randomized controlled trials. Ethical approval was obtained from the Institutional Ethics Committee of the Hospital of Stomatology, Sun Yat-sen University (Approval No. KQEC-2025-143-01). All participating students received written and verbal information about the study purpose, procedures, voluntary participation, and their right to withdraw at any time without academic penalty. Written informed consent was obtained from all student participants before enrolment. All clinical procedures were performed in the routine outpatient teaching setting under direct supervision by licensed oral and maxillofacial surgeons. Patients were informed that undergraduate students might participate in their care under supervision, and routine clinical informed consent for treatment was obtained before procedures. The study was conducted in accordance with the principles of the Declaration of Helsinki. Educational context and intervention The study was conducted during an 8-week clinical rotation in oral and maxillofacial surgery. Both groups received teaching in the same clinical department, were supervised by the same instructors, and encountered patients within the same outpatient service. The main difference between the two groups was the structure used to determine students’ progression from observation to independent performance of increasingly complex tooth extraction procedures under supervision. Control group: Traditional apprenticeship-based teaching Students in the control group received traditional apprenticeship-based clinical teaching. During the initial phase of the rotation, students observed complete patient encounters and tooth extraction procedures demonstrated by the attending instructors. After this observation period, students were gradually allowed to participate in patient management and to perform tooth extractions under supervision, beginning with relatively simple cases. In this model, progression to more complex procedures was determined by the supervising instructors based on their clinical judgement. Decisions regarding whether a student was ready to attempt a more difficult case were made according to the instructor’s assessment of the student’s prior performance, perceived readiness, and the complexity of the current clinical case. No predefined numerical progression threshold was used in the control group. Because the control group received traditional apprenticeship-based training without a structured stage-based logbook, the total number of observed cases, hands-on procedures, and case-complexity distribution were not prospectively quantified for each control-group student. Therefore, individual-level clinical exposure could not be formally compared between the two groups. Experimental group: Microskill stepwise progression model Students in the experimental group received training using the microskill stepwise progression model. The nine-stage microskill progression framework was developed by a three-member institutional teaching panel consisting of senior oral and maxillofacial surgery educators with experience in undergraduate clinical supervision. The panel deconstructed the tooth extraction procedure into nine stages of increasing procedural complexity according to expected technical difficulty, patient safety considerations, and the level of student participation required. The nine stages were operationally defined according to specific tooth-extraction categories and case-complexity characteristics that are commonly encountered in undergraduate oral surgery training. To reduce reliance on local teaching experience alone and to clarify the clinical basis of the framework, the nine-stage progression was cross-checked against the published Chinese expert consensus on standardized clinical applications of minimally invasive tooth extraction techniques.13 Recommendations from the consensus regarding preoperative assessment, patient communication, radiographic evaluation, aseptic preparation, local anaesthesia, surgical manipulation, tooth removal, postoperative management, complication prevention, and follow-up were mapped onto corresponding teachable microskills. Each stage was therefore intended to represent a discrete cognitive or procedural component, progressing from observation and preoperative assessment to supervised procedural participation and integrated procedural performance. The nine-stage progression framework is shown in Table 1. Table 1Detailed deconstruction of the nine-stage microskill stepwise progression model (MSPM).Table 1 dummy alt textStageProcedural content and requirements1Comprehensive clinical observation (mandatory observation of at least five complete tooth extraction procedures).2Nonsurgical clinical skills (patient history taking, clinical documentation/charting, and delivery of postoperative instructions).3Basic perioperative skills (local infiltration and regional nerve block anaesthesia, incision and mucoperiosteal flap elevation, irrigation and curettage of the extraction socket, alveolar bone compression/contouring, suturing, and suture removal).4Uncomplicated cases (extraction of mobile teeth).5Moderate complexity cases I (routine extraction of vertically positioned maxillary third molars and therapeutic extractions for orthodontic purposes).6Moderate complexity cases II (routine extraction of vertically positioned mandibular third molars and simple extraction of retained roots/residual crowns without surgical sectioning).7Surgical extractions requiring sectioning (use of motorized rotary instruments for tooth or root sectioning, eg, for retained roots/crowns of first and second molars).8Complex impacted extractions (mid-to-deep level impactions, including partially erupted, mesioangular, horizontal, or distoangular impacted third molars).9Highly complex cases (extraction of full bony impactions or supernumerary teeth). Detailed deconstruction of the nine-stage microskill stepwise progression model (MSPM). The model incorporated predefined progression criteria informed by principles of deliberate practice, mastery-oriented learning, workplace-based assessment, and entrustment-based progression, which emphasize repeated supervised practice, feedback, direct observation, and progression based on demonstrated readiness rather than exposure alone.14 From Stage 2 to Stage 9, students were required to complete five targeted observations and five successful supervised hands-on performances at the current stage before being allowed to progress to the next stage. The numerical threshold was used as a pragmatic minimum exposure benchmark for this pilot study rather than as a literature-validated competency cut-off. Each clinical attempt was directly observed by the supervising instructor and recorded as either ‘Pass’ or ‘Fail’ according to predefined criteria. A ‘Fail’ was recorded if the instructor had to take over the procedure to ensure completion or patient safety, or if the student’s performance created a potential safety concern. Failed attempts were not counted towards the progression threshold. Students who did not meet the progression requirement continued deliberate practice at the same stage until five successful performances were documented. Progression to cases of higher complexity was permitted only after the required benchmark had been achieved. Although the control group did not receive stage-based training during the rotation, the same nine-stage case-complexity framework was used at the end of the rotation as a reference for students to self-report the highest level of case complexity they had performed under supervision. These data were used only as a descriptive measure of self-reported clinical exposure rather than as an objectively verified measure of actual exposure or competence. Because the control group did not use a structured stage-based logbook, students’ reported highest stage could not be independently confirmed through prospective case records, supervisor verification, or chart review. To minimize cross-contamination, the intervention-specific teaching materials, progression logbooks, and advancement criteria were restricted to the experimental group, and teaching records were maintained separately. Instructors were asked to adhere to the assigned teaching approach for each group. However, because both groups were taught in the same clinic by the same instructors, complete prevention of cross-contamination was not possible. Assessment tools and data collection At the end of the 8-week clinical rotation, educational outcomes were evaluated using both quantitative and qualitative approaches. Quantitative outcomes included objective procedural performance assessed by a Modified Direct Observation of Procedural Skills (modified DOPS) scale15 and students’ self-reported perceptions measured using a questionnaire. Qualitative data were collected through open-ended student feedback, instructor interviews, and written reflections. Development and content validation of the modified DOPS tool The modified DOPS scale was adapted from the original DOPS framework for undergraduate tooth extraction training in oral and maxillofacial surgery. The adaptation was guided by the learning objectives of the clinical rotation, departmental teaching requirements, and the expected procedural competencies for undergraduate dental students. The scale retained core domains of procedural assessment, including patient communication, preoperative preparation, aseptic technique, local anaesthesia, instrument handling, procedural execution, complication awareness, and postoperative instructions. An additional domain related to the safe use of motorized rotary instruments was incorporated to reflect contemporary clinical practice. The scale used a 100-point scoring system, and the assessment form is provided in Supplementary Table 1. To evaluate content validity, the modified DOPS tool was reviewed by an expert panel consisting of 6 members, including 3 oral and maxillofacial surgery specialists, 2 experienced undergraduate clinical supervisors, and 1 dental education assessment expert. All panel members had more than 10 years of clinical teaching experience. Each expert independently rated the relevance of each item using a 4-point scale, where 1 = not relevant, 2 = somewhat relevant, 3 = quite relevant, and 4 = highly relevant. The item-level content validity index, I-CVI, was calculated as the proportion of experts rating the item as 3 or 4. The scale-level content validity index, S-CVI/Ave, was calculated as the average I-CVI across all items. Expert comments were used to revise item wording, scoring anchors, and procedural descriptors to improve clarity and applicability to undergraduate tooth extraction training. Objective clinical skills assessment End-of-rotation procedural competence was assessed using the modified DOPS scale during supervised tooth extraction procedures in the clinical setting. To improve comparability of the end-of-rotation DOPS assessment, all students were assessed during supervised extraction of vertically impacted mandibular third molars. Impaction angulation was confirmed using preoperative panoramic radiographs and/or CBCT images according to Winter’s classification. Cases with acute infection, poor patient cooperation, or other conditions that could substantially interfere with procedural assessment were not selected for the DOPS evaluation. Preoperative imaging was reviewed as part of routine clinical case selection to avoid cases with obviously unusual anatomy or anticipated extreme surgical difficulty. Because residual variation in difficulty may remain even among vertical impactions, impaction depth and ramus relationship were further classified according to the Pell and Gregory classification. The distribution of Pell and Gregory classifications was compared between groups to evaluate residual differences in assessment-case difficulty. Two independent examiners who were not involved in the preceding teaching activities assessed each student’s performance. The examiners were blinded to students’ group allocation. Before the formal assessment, the two examiners reviewed the modified DOPS criteria and discussed the scoring anchors to promote consistency in rating. Each student’s final modified DOPS score was calculated as the mean of the scores assigned independently by the two examiners. Subjective psychological characteristics questionnaire Students completed an anonymous questionnaire at the end of the rotation. The questionnaire was designed to explore students’ self-reported clinical skill mastery, psychological self-efficacy, perceptions of the teaching process, and perceptions of the clinical learning environment. Students were informed that their responses would be used only for research purposes and would not affect their academic assessment or clinical rotation grade. The questionnaire was developed by the research team based on the study objectives and reviewed by three dental education experts for content relevance, clarity, and appropriateness to the undergraduate clinical training context. It comprised 15 items rated on a 5-point Likert scale, with higher scores indicating stronger agreement or more positive perceptions. The items were grouped a priori into four domains: Clinical Skills Mastery, psychological self-efficacy, teaching process evaluation, and clinical environment and overall evaluation. The questionnaire is provided in Supplementary Table 2. Because the questionnaire was developed for this study, expert review was used to support content relevance and clarity only. No formal psychometric validation, such as exploratory or confirmatory factor analysis, test–retest reliability assessment, or criterion validity testing, was performed. Therefore, the questionnaire was used as an exploratory instrument rather than as a fully validated scale. Qualitative data collection: Semistructured interviews and written reflections Qualitative data were collected to explore students’ and instructors’ perceptions of the teaching models and to help interpret the quantitative findings. Data sources included open-ended questionnaire responses from students, semistructured interviews with the two instructors, and written instructor reflections. For student feedback, open-ended questions were appended to the questionnaire. Students were invited to describe their experiences during clinical tooth extraction training, including perceived confidence, difficulties encountered during patient care, experiences of supervision, feelings of safety or frustration, and perceptions of progression through different levels of case complexity. For instructor feedback, semistructured interviews and written reflections were collected from the two senior physicians who delivered the clinical teaching. The interview guide focused on three areas: instructors’ perceptions of entrustment decisions, perceived changes in students’ procedural competence and variability, and the feasibility of implementing a structured progression model in a high-volume outpatient clinical setting. All qualitative data were anonymized before analysis. Interview notes and written reflections were reviewed alongside open-ended student responses. Representative quotations were translated from Chinese into English and checked by bilingual members of the research team to preserve the intended meaning. Statistical analysis Quantitative data were analysed using GraphPad Prism software. Continuous variables, including baseline theoretical examination scores and end-of-rotation modified DOPS scores, were summarized as mean ± standard deviation. Normality was assessed using the Shapiro–Wilk test and visual inspection of distributions. Between-group comparisons of continuous variables were performed using Welch’s t test, which does not assume equal variances. Score dispersion in end-of-rotation modified DOPS performance was compared between the two groups using the Brown–Forsythe test. To quantify the magnitude of between-group differences in variability, the control-to-experimental standard deviation ratio and the corresponding variance ratio were calculated. Their 95% confidence intervals were estimated using stratified bootstrap resampling with 10,000 repetitions. Questionnaire responses were analysed at the item level using Mann–Whitney U tests. Because the questionnaire contained 15 Likert-scale items and each item was tested separately, P values were adjusted for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure. Both unadjusted and FDR-adjusted P values were reported, and statistical significance was defined as an FDR-adjusted P value <.05. Inter-rater reliability of the modified DOPS total score was assessed using a two-way random-effects intraclass correlation coefficient for absolute agreement. Because the final score was based on the mean of two examiner ratings, the average-measure ICC was reported. Internal consistency of the student questionnaire was assessed using Cronbach’s alpha for the overall questionnaire and its predefined domains. Cronbach’s alpha was used as a descriptive indicator of internal consistency and was not considered sufficient to establish the psychometric validity of the questionnaire. Qualitative data were analysed using thematic analysis. The qualitative component was designed to provide explanatory and contextual insight into the quantitative findings rather than to develop a standalone qualitative theory. All available open-ended student responses, instructor interview notes, and written instructor reflections were included in the analysis. Two researchers independently reviewed the qualitative data and generated initial codes. Coding results were compared, and discrepancies were resolved through discussion; when consensus could not be reached, a third researcher was consulted. Related codes were then grouped into candidate themes, which were reviewed and refined through iterative comparison with the original text and the full dataset. Student and instructor data were compared to support triangulation across participant perspectives. Representative quotations were selected to illustrate the main themes. Results Objective clinical skills assessment Expert review of the modified DOPS scale indicated acceptable content validity. The I-CVI values ranged from 0.83 to 1.00, and the S-CVI/Ave was 0.92. Minor revisions were made based on expert comments, mainly involving clarification of wording, scoring anchors, and procedural safety descriptors. Detailed item-level CVI results are presented in Supplementary Table 3. Inter-rater reliability for the total modified DOPS score was good. Using a two-way random-effects model with absolute agreement and average measures, the ICC was 0.86 (95% CI, 0.74-0.93; P < .001). Detailed inter-rater reliability results are presented in Table 2. Table 2Inter-rater reliability of modified DOPS scores.Table 2 dummy alt textAssessment itemICC modelICC95% CIP valueInterpretationTotal modified DOPS scoreTwo-way random-effects, absolute agreement, average measures0.860.74-0.93<.001GoodICC values were interpreted as follows: <0.50, poor; 0.50 to 0.75, moderate; 0.75 to 0.90, good; >0.90, excellent reliability.CI, confidence interval; DOPS, Direct Observation of Procedural Skills; ICC, intraclass correlation coefficient. Inter-rater reliability of modified DOPS scores. ICC values were interpreted as follows: <0.50, poor; 0.50 to 0.75, moderate; 0.75 to 0.90, good; >0.90, excellent reliability. CI, confidence interval; DOPS, Direct Observation of Procedural Skills; ICC, intraclass correlation coefficient. The distribution of Pell and Gregory depth position did not differ significantly between the experimental and control groups, with most cases classified as Position A or B. Similarly, the distribution of ramus relationship classes was comparable between groups. Fisher–Freeman–Halton exact test showed no significant between-group differences in either depth position or ramus relationship classification (Supplementary Table 4). At the end of the rotation, the experimental group achieved significantly higher modified DOPS scores than the control group. The mean modified DOPS score was 94.60 ± 3.97 in the experimental group and 88.25 ± 6.41 in the control group. Welch’s independent-samples t test showed a significant between-group difference, with a mean difference of 6.35 points (95% CI, 2.91-9.79; P < .001; Cohen’s d = 1.19) (Figure 2). The experimental group also showed lower score dispersion than the control group. Scores ranged from 87 to 100 in the experimental group and from 75 to 96 in the control group. The control-to-experimental SD ratio was 1.62 (bootstrap 95% CI, 1.13-2.31), and the corresponding variance ratio was 2.62 (bootstrap 95% CI, 1.27-5.35). The Brown–Forsythe test indicated a statistically significant between-group difference in score dispersion, F(1, 38) = 4.75, P = .036. This finding should be interpreted cautiously given the modest sample size and the bounded nature of DOPS scores. Nevertheless, it suggests that the structured progression model may be associated with more consistent attainment of procedural competence. As shown in Figure 2, the experimental group had a narrower score range and a higher minimum score than the control group. Comparison of end-of-rotation objective clinical skills (modified DOPS) scores between the two groups (Welch’s independent-samples t test, *** P < .001). A sensitivity analysis indicated that, with 20 students per group, a two-sided alpha level of 0.05, and 80% power, the study could detect an effect size of approximately Cohen’s d = 0.91. The posthoc power calculated from the observed effect size was approximately 0.956. These analyses were conducted as supplementary estimates of the detectable effect size and should not be interpreted as a substitute for an a priori sample size calculation. Subjective psychological characteristics questionnaire Because the questionnaire was developed for this study and had not undergone comprehensive psychometric validation, the questionnaire findings were interpreted as exploratory. The overall Cronbach’s alpha for the 15-item questionnaire was high at 0.936. Domain-specific Cronbach’s alpha values were 0.914 for clinical skills mastery, 0.948 for psychological self-efficacy, 0.947 for teaching process evaluation, and 0.617 for clinical environment and overall evaluation. The lower alpha value for the clinical environment and overall evaluation domain indicated limited internal consistency; therefore, findings related to this domain were interpreted cautiously. This may partly reflect the fact that the domain contained only two items. Item-level internal consistency analysis showed that corrected item-total correlations ranged from 0.424 to 0.840. No item had a corrected item-total correlation below 0.30. Deletion of individual items did not result in a meaningful improvement in the overall Cronbach’s alpha. Therefore, all 15 items were retained for exploratory analysis. These internal consistency results were not considered sufficient to establish the psychometric validity of the questionnaire. At the end of the clinical rotation, students’ self-reported highest case-complexity stage observed or performed under supervision was analysed as an exploratory indicator of perceived clinical exposure. The nine stages were operationally defined according to specific and commonly encountered tooth-extraction categories and case-complexity characteristics, rather than abstract or subjective descriptors. Therefore, students in both groups were expected to be able to map their clinical experience to the predefined stage descriptions. In this exploratory self-reported analysis, the experimental group reported a higher maximum case-complexity stage than the control group (median 7 vs 6, P < .05). As shown in Figure 3, the control group reported stages ranging from 4 to 8, with an interquartile range of 5.0 to 6.75, whereas the experimental group reported stages ranging from 5 to 8, with an interquartile range of 6.0 to 7.0. Nevertheless, because the experimental group used the stage framework prospectively during training whereas the control group reported this information retrospectively, this finding may have been affected by recall bias, expectation bias, and differential familiarity with the framework. Therefore, it should be interpreted cautiously as a self-reported exploratory indicator and should not be considered objective evidence of clinical exposure or procedural competence. Exploratory comparison of self-reported highest case-complexity stages between the two groups. The stages were based on predefined tooth-extraction categories and case-complexity characteristics. *Exploratory Mann–Whitney U test, P < .05. As exploratory student-reported outcomes, feedback on the 15 questionnaire items was further analysed using Mann–Whitney U tests. The items were grouped into four predefined domains: Clinical skills mastery, psychological self-efficacy, teaching process evaluation, and clinical environment and overall evaluation (Table 3). To account for multiple item-level comparisons, P values were adjusted using the Benjamini–Hochberg false discovery rate procedure. Six items showed nominally significant between-group differences before correction; however, four items remained statistically significant after FDR correction. Table 3Comparison of subjective psychological characteristics questionnaire between the two groups.Table 3 dummy alt textDimensions/ItemsControlMedian (IQR)ExperimentalMedian (IQR)Mann–Whitney U valueUnadjusted P valueFDR-adjusted P valueClinical skills masteryTranslation of theory to practice4.0 (4.0-4.75)4.0 (4.0-5.0)182.585.627Independence in basic procedures4.0 (4.0-4.0)5.0 (4.0-5.0)110.5.006**.045*Competence in simple extractions4.0 (4.0-4.75)5.0 (5.0-5.0)95.001***.015*Surgical planning for complex cases4.0 (3.0-4.0)4.0 (3.0-5.0)155.5.223.257Recognition and prevention of complications4.0 (3.0-4.0)4.0 (4.0-5.0)130.044*.110Psychological self-efficacyConfidence during independent practice4.0 (4.0-4.0)4.0 (4.0-5.0)148.5.123.213Overcoming fear of failure4.0 (4.0-4.0)4.0 (4.0-5.0)158.5.214.257Stability of skill improvement4.0 (3.25-4.75)4.0 (4.0-5.0)154.218.257Constructive help-seeking under difficulty4.0 (4.0-5.0)5.0 (4.0-5.0)147.5.128.213Teaching process evaluationRationality of teaching pace4.0 (4.0-5.0)5.0 (4.0-5.0)143.103.213Sufficiency of basic skills training4.0 (4.0-5.0)5.0 (4.0-5.0)150.205.257Initiative in clinical learning4.0 (4.0-4.75)5.0 (4.0-5.0)113.013*.049*Objectivity of assessment criteria4.0 (4.0-5.0)5.0 (4.25-5.0)110.010*.049*Clinical environment and overall evaluationPerceived level of patient trust4.0 (3.0-4.0)4.0 (3.0-4.75)185.673.673Utility for future independent practice4.0 (4.0-5.0)5.0 (4.0-5.0)128.040*.110Data are presented as median (IQR). Differences between the two groups were analysed using the Mann–Whitney U test. The 15 item-level P values were adjusted for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure. Statistical significance was defined as FDR-adjusted P < .05(*P< .05, **P< .01, ***P< .001).IQR, interquartile range. Comparison of subjective psychological characteristics questionnaire between the two groups. Data are presented as median (IQR). Differences between the two groups were analysed using the Mann–Whitney U test. The 15 item-level P values were adjusted for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure. Statistical significance was defined as FDR-adjusted P < .05(*P< .05, **P< .01, ***P< .001). IQR, interquartile range. In the clinical skills mastery domain, the experimental group reported higher perceived procedural competence. After FDR correction, significant differences were retained for ‘Independence in basic procedures’ median: 5.0 vs 4.0; unadjusted P = .006, FDR-adjusted P = .045 and ‘Competence in simple extractions’ median: 5.0 vs 4.0; unadjusted P = .001, FDR-adjusted P = .015. Although ‘Recognition and prevention of complications’ was nominally significant before correction unadjusted P = .044, this difference did not remain significant after FDR correction FDR-adjusted P = .110. In the teaching process evaluation domain, two items remained significant after FDR correction: ‘Objectivity of assessment criteria’ median: 5.0 vs 4.0; unadjusted P = .010, FDR-adjusted P = .049 and ‘Initiative in clinical learning’ median: 5.0 vs 4.0; unadjusted P = .013, FDR-adjusted P = .049. In contrast, the item ‘Utility for future independent practice’ in the clinical environment and overall evaluation domain was only nominally significant before correction unadjusted P = .040 and was no longer significant after FDR adjustment FDR-adjusted P = .110. For the remaining 11 items, including all four items in the psychological self-efficacy domain, no statistically significant between-group differences were observed after FDR correction. Median scores for these items were generally 4.0 or above in both groups, indicating relatively positive perceptions in both instructional groups. A domain-level psychological self-efficacy score was calculated as the mean of the four items. Although the score was higher in the experimental group than in the control group, the difference was not statistically significant (4.35 ± 0.59 vs 3.96 ± 0.87; Mann–Whitney U = 147.0, P = .14). However, scores clustered towards the upper end, particularly in the experimental group, where 7 students (35.0%) achieved the maximum score and 9 students (45.0%) had near-maximum scores, suggesting a potential ceiling effect (Supplementary Table 5). Thematic analysis of qualitative feedback Qualitative feedback from instructor interviews and students’ written reflections was analysed thematically. The quotations presented below were selected as representative excerpts from the broader qualitative dataset rather than as the complete dataset. Four main themes were identified regarding the implementation and perceived effects of the microskill stepwise progression model. Theme 1: Reducing performance disparity and supporting more consistent competence development Qualitative findings were consistent with the quantitative observation of reduced variability in DOPS scores. Instructors reported that the stepwise progression model helped establish a clearer minimum competency standard and reduced the gap in procedural ability among students. ‘Previously, the gap between students was enormous; some could only extract mobile teeth, while others could handle complex teeth extractions. However, through this Micro-skill Stepwise Progression, the discrepancy among students has significantly narrowed. Now, the vast majority of them are competent in performing complex teeth extractions’. (Instructor 1) Students also reported that dividing complex procedures into progressive microskills made advanced clinical tasks feel more manageable. ‘When I first observed the extraction of a mesioangularly impacted tooth, I thought it required years of hands-on practice to develop the right feel for it. I never expected that the training under this Micro-skill Stepwise Progression model would enable me to independently extract a mesioangularly impacted tooth within just 8 weeks’. (Student 1, Experimental Group) Theme 2: Building confidence through verified skill progression Although the quantitative psychological self-efficacy domain did not show significant between-group differences, qualitative feedback suggested that students in the experimental group developed confidence based on verified procedural milestones rather than general reassurance alone. Instructors observed that structured progression appeared to support both clinical confidence and patient communication. ‘In the past, because their foundational skills were weak, students often felt intimidated by complex cases. Now that students have explicit progression targets to clear, both their communication with patients and their clinical operations have become noticeably more confident’. (Instructor 2) One student described how prior progression through comparable cases helped them remain confident when facing patient doubt. ‘A patient once doubted me, saying, “You’re just an intern, you’ve surely never extracted a tooth this difficult.” But I felt very grounded internally because, under this progression system, I had already done plenty of cases just as hard as theirs. Ultimately, I successfully completed the extraction, and the patient was highly satisfied’. (Student 2, Experimental Group) Theme 3: The value of progression constraints as a safety mechanism Students also described the nonbypassable progression requirements as a safety mechanism. Although these requirements sometimes slowed advancement, they were perceived as helping prevent premature exposure to more complex procedures before adequate foundational competence had been achieved. ‘During the sixth stage, my first few cases went quite smoothly. However, on my fifth case, I encountered a mandibular third molar with highly curved roots. No matter what I did, I couldn’t extract it, and I ultimately had to seek the instructor’s help. Afterward, the instructor detailed the critical techniques for extracting that specific tooth. It made me realize that, were it not for this progression system, I might have rashly advanced to the next stage without having laid a solid foundation’. (Student 3, Experimental Group) Theme 4: Implementation burden and rigidity in authentic clinical settings Despite the perceived educational benefits, students also identified practical challenges in implementing the model in a high-volume outpatient environment. These challenges were mainly related to unpredictable patient case-mix and the administrative burden of verifying progression through multiple stages. ‘Sometimes I felt I had completely mastered a certain stage, but the difficulty of the cases coming into the clinic was beyond our control. Because I hadn’t accumulated the required 5 cases, the instructor wouldn’t let me extract a slightly more difficult tooth. It felt like my progress was being dragged down’. (Student 4, Experimental Group) ‘I think dividing it into 9 stages is too fragmented. For example, Stage 2 (writing medical records) and Stage 3 (administering anaesthesia) could easily be merged. Having to wait for the instructor’s verification every single time feels extremely cumbersome, especially when there is a high volume of patients’. (Student 5, Experimental Group) Discussion This randomized mixed-methods study evaluated a structured microskill stepwise progression model for undergraduate tooth extraction training. Compared with traditional apprenticeship-based training, the model was associated with higher objective procedural performance and lower variability in DOPS scores at the end of the rotation. The intervention should be interpreted as a multicomponent structured training model, in which targeted observation, supervised performance thresholds, documentation, and stepwise progression were integral components. Thus, the observed differences cannot be attributed to microskill decomposition alone. Rather, these findings suggest that an integrated approach combining task decomposition, structured exposure, supervised practice, and predefined progression criteria may provide a feasible approach for standardizing undergraduate training in invasive dental procedures. Importantly, the model appeared to improve not only average competence but also the consistency of competence within the student cohort, which is particularly relevant to graduation readiness and patient safety in undergraduate dental education. One of the most important quantitative findings was the reduced variance in DOPS scores among students in the intervention group. In clinical education, the mean performance score alone may not fully reflect the quality of a training program. For procedures such as tooth extraction, ensuring that all students reach a minimum level of safe procedural competence is as important as improving overall performance. In traditional apprenticeship-based training, entrustment decisions are often influenced by supervisors’ clinical impressions, case availability, and students’ early performance. Although such judgments are valuable, they may lead to heterogeneous learning opportunities. The wider distribution and lower-score tail observed in the control group may therefore reflect differences in clinical exposure, baseline confidence, case complexity, or opportunities for hands-on participation. One possible interpretation is that students who perform well early may receive more operative exposure and progressively more challenging cases, whereas those who struggle initially may become more hesitant and receive fewer opportunities for independent practice. However, because individual-level case exposure, hands-on experience, and case-complexity distribution were not systematically tracked, this cumulative advantage or disadvantage explanation remains speculative and cannot be confirmed from the present data. The integrated stepwise training model may have contributed to lower performance variability by introducing a structured safety-net mechanism. All students were required to complete targeted observation and supervised practice before advancing to the next procedural stage. This requirement created a shared minimum standard for progression and reduced reliance on purely subjective entrustment. From the perspective of Cognitive Load Theory, tooth extraction imposes a substantial intrinsic cognitive load on novice learners because it requires simultaneous integration of anatomical knowledge, tactile perception, force control, communication, and clinical judgment.7 By dividing the procedure into nine sequential stages, the model allowed students to focus their attention on a limited set of task-specific skills at each stage. Such progressive segmentation may reduce unnecessary cognitive burden during early learning and allow cognitive resources to be gradually reallocated as students achieve greater fluency. Previous studies in procedural education have supported the use of part-task training, simulation, and mastery learning for complex skills such as laparoscopy, endoscopy, and airway management.4,16,17 However, much of this evidence has been generated in simulated or postgraduate training environments. The present study extends these principles to an authentic undergraduate dental clinic, where training must balance student autonomy, supervisor workload, patient safety, and case availability. The findings indicate that a mastery-oriented progression pathway can be implemented in real clinical tooth extraction training and may help supervisors make more transparent and evidence-informed entrustment decisions. The exploratory questionnaire findings and qualitative data provide supportive insight into the possible educational mechanisms of the intervention, but they should be interpreted cautiously. The item-level questionnaire analysis involved multiple comparisons. Before correction, six questionnaire items showed nominally significant between-group differences; after Benjamini–Hochberg false discovery rate correction, four items remained statistically significant: independence in basic procedures, competence in simple extractions, initiative in clinical learning, and objectivity of assessment criteria. In contrast, the differences in recognition and prevention of complications and utility for future independent practice did not remain statistically significant after correction. Therefore, these findings should be regarded as supportive and hypothesis-generating rather than confirmatory. Although the intervention group showed better objective performance, no statistically significant difference was observed between groups in the psychological self-efficacy domain of the questionnaire. This finding should be interpreted cautiously. Self-efficacy scores among senior dental students may be affected by ceiling effects and self-assessment bias.18,19 Students entering clinical clerkships may report relatively high confidence despite limited operative experience, and self-perceived confidence does not always correspond to actual clinical competence. Therefore, the absence of a significant between-group difference in general self-efficacy does not necessarily indicate that the intervention had no psychological impact. The additional distributional analysis of the psychological self-efficacy domain showed upper-end clustering of scores, particularly in the intervention group. Although the between-group difference was not statistically significant, 35.0% of students in the intervention group achieved the maximum possible score and 45.0% achieved near-maximum scores, suggesting a potential ceiling effect that may have limited the questionnaire’s sensitivity to detect further improvements in self-reported psychological self-efficacy. The qualitative data further suggested that the nature and source of confidence differed between groups. Students in the traditional training group may have derived confidence from general academic achievement, observation experience, or verbal reassurance, whereas students in the intervention group often linked their confidence to repeated successful completion of specific procedural steps. According to self-efficacy theory, mastery experiences are among the strongest sources of durable self-efficacy.20 Therefore, although the quantitative questionnaire data did not demonstrate a statistically significant difference in the psychological self-efficacy domain, the qualitative findings suggest that the structured progression model may have supported more task-specific and performance-grounded confidence. This distinction is important in invasive dental procedures, where overconfidence without sufficient competence may compromise patient safety, whereas confidence supported by repeated supervised success may facilitate safer clinical autonomy. The intervention also highlighted an important practical tension in competency-based dental education. While predefined progression criteria can enhance standardization and safety, they may also increase administrative workload and reduce flexibility. Some students reported that the staged requirements were time-consuming, fragmented, or dependent on the availability of suitable cases. These concerns are important because undergraduate dental clinics operate under real-world constraints, including limited chair time, variable patient flow, and competing demands on clinical supervisors.21 A rigid progression system may be particularly burdensome for fast learners who demonstrate competence earlier than expected. Therefore, future versions of the model should consider adaptive progression criteria. For example, students who demonstrate consistently excellent performance across repeated attempts could be allowed to advance earlier, whereas students requiring additional support could receive targeted remediation. Combining highly overlapping lower-level steps may also improve feasibility without compromising educational safety. Although this study focused on tooth extraction, the broader principles of task segmentation, supervised progression, and staged entrustment may be relevant to other complex dental procedures. However, such transferability should be regarded as hypothesis-generating rather than established. The present model was developed and evaluated only for tooth extraction, and both the nine-stage framework and modified DOPS tool were procedure-specific. Other dental procedures differ in psychomotor demands, cognitive load, risk profile, case availability, and assessment requirements. Therefore, any extension of this model to procedures such as local anaesthesia, periodontal instrumentation, endodontic access preparation, surgical flap design, or implant-related training22, 23, 24 would require procedure-specific task analysis, expert consensus on progression criteria, validation of assessment tools, and prospective empirical evaluation. This pilot study has several limitations. First, it was conducted at a single institution with a relatively small sample, and no formal a priori sample size calculation was performed. The study may therefore have been underpowered to detect small-to-moderate effects, and the generalizability of the findings is limited. Although the groups were comparable in theoretical examination scores, potentially relevant factors such as manual dexterity, anxiety, motivation, informal clinical exposure, and baseline procedural skill were not assessed. Second, the nine-stage framework was developed by a teaching panel from a single institution and was not externally validated through a Delphi process or independent educational consensus. Individual clinical exposure, particularly in the control group, was not fully quantified. The self-reported highest clinical stage was also subject to recall and expectation bias, and differential familiarity with the framework may have affected reporting. It should therefore be regarded as an exploratory indicator of perceived clinical exposure rather than an objective measure of exposure or competence. Because the intervention combined several components, their individual effects could not be disentangled, and cross-contamination between groups could not be completely excluded because they trained in the same clinical setting with overlapping instructors. Third, patient-level outcomes were not systematically collected. Procedure duration, complications, supervisor takeover, postoperative pain, bleeding, infection, patient satisfaction, and adverse events were not prospectively recorded. Accordingly, the findings provide preliminary evidence of educational performance and variability in assessed competence, but not of improved patient safety or clinical outcomes. Finally, the intervention was evaluated during only one clinical rotation, so long-term retention was not assessed. The modified DOPS tool and study-specific questionnaire require further psychometric validation; the questionnaire was expert-reviewed but not comprehensively validated, and one domain showed relatively low internal consistency (Cronbach’s α = 0.617). Self-reported questionnaire findings may also have been affected by response bias and ceiling effects. Larger multicentre studies with longer follow-up, externally validated frameworks and assessment tools, broader baseline assessment, and prospective patient-level outcome collection are needed. Despite these limitations, this study provides preliminary evidence that an integrated structured microskill stepwise progression model was associated with higher assessed procedural competence and lower performance variability in undergraduate tooth extraction training. The findings support the value of combining competency-based education, mastery learning, and staged entrustment in authentic dental clinical settings. Rather than replacing clinical supervisors’ professional judgment, the model may complement it by providing transparent progression criteria and objective evidence of student readiness. For undergraduate dental education, such approaches may help graduates achieve a more consistent level of assessed competence in common invasive procedures. These educational benefits may have potential patient-safety implications, but such implications require direct evaluation using patient-level outcome measures. Conclusion In conclusion, this pilot randomized study provides preliminary evidence that the microskill stepwise progression model may be associated with higher assessed tooth extraction performance and lower variability in procedural competence among undergraduate dental students. However, given the small single-centre sample and absence of a formal a priori sample size calculation, these findings should be interpreted as exploratory rather than confirmatory. In future implementation, the model should be applied adaptively rather than rigidly, allowing faster progression for students with consistently high performance and targeted remediation for those requiring additional support. Larger, adequately powered multicentre studies with longer follow-up and patient-level outcome measures are needed to confirm the educational effectiveness of this approach and determine whether it translates into measurable patient-level benefits. Author contributions Conceptualization: Wu, Wang. Methodology: Wu, Luo. Data curation: Wu, Wang. Formal analysis: Wu, Wang. Investigation: Wu, Luo, Liu. Resources: Gao, Zhang, Zheng. Validation: Yang, Liu. Visualization: Wu, Yang. Writing – original draft preparation: Wu, Wang, Luo. Writing – review and editing: Gao, Zhang, Zheng. Supervision: Gao, Zhang, Zheng. Project administration: Gao, Zhang, Zheng. Funding acquisition: Zhang. Funding This work was supported by the Guangdong Natural Science Foundation (Grant No.2023A1515011763 and No.2025A1515010170). Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. Footnotes Supplementary material associated with this article can be found in the online version at doi:10.1016/j.identj.2026.111097. Contributor Information Siyong Gao, Email: gaosy7@mail.sysu.edu.cn. Sien Zhang, Email: zhangsen5@mail.sysu.edu.cn. Guangsen Zheng, Email: zhenggs@mail.sysu.edu.cn. Supplementary materials REFERENCES Associated Data Supplementary Materials