This was an educational improvement project completed between 2014 and 2024 at Stanford University School of Medicine within the ACGME-approved Pediatric Anesthesiology Fellowship. The fellows practice at Lucile Packard Children’s Hospital Stanford, a level-one trauma center with 20 operating rooms. The training occurs over a single year, with approximately 8 fellows in postgraduate year five or greater.

During the 2021‐2022 year, an andragogic curriculum was integrated into the fellowship program. Before this new curriculum, fellow education consisted of after-hours didactics, one simulation session, and an apprenticeship model of clinical care. The new curricular framework relied on Knowles’ Andragogic Model combined with Kolb’s Experiential Learning Cycle [8,9]. These models were preferred for their established application in health professions education and direct alignment with the characteristics of the fellowship learner population. Fellows enter the program as fully qualified physicians with professional identities, diverse prior clinical experiences, and intrinsic motivation to achieve subspeciality competence. Knowles’ Andragogic Model was chosen for its emphasis on the role of the independent self-directed learner, prioritized problem-centered approaches, and application of learning into daily life [8]. Kolb’s Experiential Learning Cycle was selected as a complementary framework since it provides a mechanistic account of how learning occurs within the experiential environment [9,10]. While Knowles’ principles guided the overall curriculum design and ensured that it was self-directed, problem-centered, and responsive to fellows’ prior experience, Kolb’s cycle was used to structure the learning process itself. Specifically, the curriculum was designed for concrete clinical experiences to be systematically followed by structured reflection, didactic conceptualization, and opportunities for active experimentation, rather than solely relying on experience.

Curricular components with various delivery platforms were implemented that prioritized the unique characteristics of fellowship-level learners and the structured process by which direct clinical experiences are converted into meaningful learning. A 12-component syllabus that addressed clinical and nonclinical ACGME requirements was developed, which included experiential, self-directed, problem-based, and asynchronous remote learning platforms (Table 1) [11]. Within these 12 learning modalities and trainings, each curricular component was curated around chapters in a foundational pediatric anesthesiology textbook and core knowledge requirements dictated by ACGME pediatric anesthesiology milestones [12]. Learning modalities within the new curriculum included directed readings, didactic lectures, quality improvement training, problem-based learning discussions, simulation, workshops, professional and behavioral health series, doctoring with care series, oral board preparation series, DEI training, journal clubs, and self-assessment in addition to daily clinical care responsibilities (Multimedia Appendix 1). Andragogic principles such as self-concept, need to know, problem-centered orientation, prior experience, readiness to learn, and internal motivation were applied to the learning modalities within the new curriculum (Table 1). The principle of problem-centered orientation was embedded through case-based conferences, simulations, and workshops. Internal motivation was applied to coaching conversations, and reflections focused on growth and ownership of a quality improvement training module and a quality improvement project of their choosing. Didactic sessions and directed readings highlighted the need to know and readiness to learn principles with explicit framing of clinical context. Additionally, the stages within Kolb’s Experiential Learning Cycle (concrete experience, reflective observation, abstract conceptualization, and active experimentation) were used within the curricular components to ensure that clinical experiences were followed by thoughtful reflection, feedback, and conceptualization (Table 1).

The primary outcome assessed fellows’ perception of educational content. This was selected because it demonstrated a broad representation of nonclinical components highlighted by ACGME guidelines. Secondary outcomes explored DEI training and programmatic resources. The DEI category was selected since it assessed reflective practice and inclusivity, and the resources category assessed the program’s educational infrastructure.

Outcomes were measured with the ACGME trainee survey, which assesses education quality and compliance with ACGME requirements [13]. Sections within the ACGME survey are standardized into preset categories. The educational content subsections include a broad range of questions on well-being, burnout, and scholarly activities. The resources subsections assess protected time for educational and professional activities and overall institutional support. The DEI subsections include questions regarding inclusivity, diversity of patient populations, and engagement of a program’s DEI efforts. Survey responses are based on a 5-point Likert scale and are annually completed by the fellows. ACGME fellow survey results from 2014 to 2024 were collected from the educational content and resources categories on the survey. Data were collected from 2019 to 2024 regarding DEI training, a new category introduced in 2019. Each category contained several subsection scores, which were averaged by the ACGME to provide the overall category mean.

To analyze the primary outcome, a statistical process control p-chart monitored a change in fellows’ responses to the educational content subsection of the ACGME survey. The Western Electric rules assessed special cause variation [14]. Rule one in Western Electric states that special cause variation occurs when a data point falls outside the 3 SD upper control limit from the center line. This occurred in the 2021‐2022 academic year, leading to the establishment of a new center line. Mean outcomes were assessed using censored regression modeling if there was a ceiling effect or a t test if no ceiling effect. A P value less than .05 was considered significant. All analyses were performed using Excel (Microsoft Corporation) and R statistical software (version 4.5.1; R Foundation for Statistical Computing; censReg package) [15].

This project received a waiver of consent by the Stanford University Institutional Review Board (eProtocol No. 87506). Data were deidentified by the ACGME before reporting the results to the program. There was no compensation given to participants.

Of the 60 pediatric anesthesiology fellows, 58 completed the ACGME survey from 2014 to 2024, with an overall response rate of 97%. During the data collection period, 56% (n=34) of the fellows identified as female and 44% (n=26) as male.

Special cause variation was noted in the statistical process control p-chart for mean scores for educational content during the 2021‐2022 academic year, which coincides with the implementation of the new curriculum. Given a ceiling effect, censored regression was used to estimate the mean difference. A mean difference of 0.89 (SE 0.27) from 2014 to 2021 was observed compared to 2022‐2024 (P<.001).

The mean difference between DEI training scores did not exhibit a ceiling effect; therefore, a t test was used to analyze the results. For DEI training (n=36), we demonstrated a difference of 0.52 (SE 0.2; P=.03). Fellows’ perceptions of program resources (n=58) had a ceiling effect, and no estimated difference was observed (mean difference −0.13, SE 0.24; P=.98).