This observational, cross-sectional comparative study using semi-structured interviews examined GP programmes delivered by different medical schools in six high-income countries.

Australia, Canada, England, Ireland, Norway and Iceland were targeted as countries having similar health systems and medical programmes to that of New Zealand. Approach to key informants in departments of general practice for interviews was pragmatic, based on our existing knowledge of, and sometimes established contacts in, the institution and the populations they serve. We obtained contact details for key participants, and wherever practical, we conducted in-person visits. Medical schools were selected for maximum diversity, rural versus urban, and traditional programmes versus newly established schools.

Teaching heads in the relevant general practice departments were identified by personal connections and networks of the researchers, or by online searches, and invited to participate by email. Contacted participants were asked to pass the email to colleagues if they were inappropriately identified as the lead academic. Participant Information Sheets were provided. Informed consent was obtained orally, recorded and transcribed. While key informants were not anonymous to the interviewing researcher, they were not identified in the database, and the universities were assigned numerical codes. However, given the specific characteristics of some of the programmes, it is possible that some of the universities and the teaching heads may be identifiable. No personal or sensitive information about the key informants was obtained.

Ethical approval was obtained from the University of Auckland Human Participants Ethics Committee on 6/05/2024 for three years.

Appropriate key people were interviewed from 44 universities. Participants underwent single semi-structured interviews about the structure and nature of their programmes conducted by three experienced GP academics (FG, RR, SG) and one nurse practitioner (EW) between April and November 2024. These were face-to-face wherever possible, otherwise via zoom. Contemporaneous notes were taken, interviews audiotaped with oral consent recorded, audio-files AI-transcribed and data entered into a spreadsheet.

Data were initially examined using the SPICES framework described by Harden which assists with understanding the educational strategies used in medical education and when considering curriculum (re)design.¹⁹ The SPICES framework components are Student-centred versus teacher-centred; Problem-based versus information gathering; Integrated versus discipline-based; Community-based versus hospital-based; Electives versus standard programme, and Systematic versus apprenticeship-based or opportunistic. The SPICES framework has been used extensively in medical education to examine where curricula sit along the six educational continua and has strong face and construct validity as an educational framework. SPICES is considered methodologically appropriate for descriptive and comparative studies, but it is not a validated measurement instrument.

We selected this framework to enable us to standardise, audit and compare against elements of curriculum design. Ultimately, we had sparce data identifying student or teacher-centred features. Further, the systematic approach of SPICES relies on students keeping a logbook against a required list of patients with specific diseases seen and skills mastered. While this may be appropriate in medical speciality training in hospitals, in general practice patients present with undifferentiated concerns, complaints and symptoms, not diagnoses. The key element for students to learn is the broad scope of patient care in the context of people’s lives, rather than diagnosis and treatment of specific diseases, and hence for this discipline, the opportunistic model of seeing ‘whomever walks through the door’ is more suitable than the systematic approach.

We therefore created a modified SPICES framework called EPIC-GP, whereby Electives (additional GP exposure) versus standard curriculum programme, Problem-based versus information-gathering, Integrated versus discipline-based, and Community-based versus hospital-based categories remain, but Student-centred versus teacher-centred and Systematic versus apprenticeship-based or opportunistic are replaced with GP-focused university-teaching (students have early and regular engagement with GPs on campus or online versus no exposure to GPs or information given on campus is from the GP perspective) and Professional education (interprofessional education (IPE) activities within the curriculum or students taught by other primary care professionals, versus taught by doctors only) – Figure 1. In short, we adapted SPICES to better reflect the educational context of general practice, and our resulting EPIC-GP framework was used to support systematic comparison.

Data collected from the semi-structured interviews included the date when the medical school was established, nature of the medical programmes including undergraduate or graduate, student intake numbers and details of the GP component in the curriculum, as well as the challenges, innovative aspects, solutions developed, and perspectives from key informants about what has worked/been successful, and what has not (Supplement 1). Qualitative findings are to be published elsewhere.

Interview transcripts and contemporaneous notes were reviewed in full and analysed using the EPIC GP framework. One researcher (RJ) undertook the primary coding by systematically mapping interview data to each EPIC GP domain (Electives, Problem based learning, Integrated curriculum, Community based exposure, GP focused university teaching, and Professional/ interprofessional education). A second researcher (FG) provided iterative review, feedback, and interpretive oversight throughout the analytic process.

For each domain, evidence from the interviews was synthesised to assign an overall score on a five-point ordinal scale, based on the extent to which the programme demonstrated the characteristics defined in the EPIC GP matrix. Scoring was undertaken iteratively, with provisional scores refined through repeated reference to the source data. The final coding decisions and domain scores were reviewed and discussed with FG and discrepancies or uncertainties were resolved through interpretive discussion to achieve conceptual consensus.

EPIC-GP scores were summed and correlation statistics calculated between EPIC-GP variables, type of programme and year of establishment using Stata 19. Cluster analysis was then undertaken on the data. Correlation and cluster analyses were undertaken as exploratory, descriptive tools to identify patterns and potential typologies across programmes, rather than to test predefined hypotheses.

A hierarchical cluster analysis using Ward’s minimum linkage variance method in Stata 19 was performed and a cluster dendrogram generated to determine an appropriate number of clusters. Descriptive statistics were used to describe the three clusters and subsequent proposed typologies of medical schools. Factor analysis on EPIC-GP was not carried out due to the underpowered sample.

Our pragmatic sample consists of 44 medical schools in the high-income countries of Australia, Canada, England, Ireland, Norway and Iceland. Fifty-seven were approached: ten did not reply, one declined and two were unable to schedule a suitable time. The programmes range from very long-standing to recently established programmes, with annual student intakes ranging from <50 to >400 (Table 1) and from traditional lecture-based preclinical followed by clinical years, through to innovative integrated, problem-based, or spiral curricula.

The EPIC-GP matrix with examples of high and low scores from responses are shown in Table 2. The range in EPIC-GP scores for each university can be seen in Figure 2.

Correlation analysis shows moderately strong correlation between integrated and GP-focused university teaching, and moderate correlation between integrated teaching and community-based GP exposure and attachments, between problem-based learning and community-based attachments, and between GP-focused university teaching and community-based attachments. There is weaker correlation between problem based learning and integrated teaching, and problem-based learning and GP-focused university teaching. Elective and other practitioners’ teaching are not correlated with other variables (Appendix Table A). There is little to no correlation with whether the medical programme is graduate or undergraduate or the year the medical school was established.

Exploratory cluster analysis suggested three potential typologies of medical school approaches to general practice teaching (see Figure 3 and Appendix Table B). The first cluster is medical schools with low-medium scores across the majority of EPIC-GP variables, which we designate hospital-focused teaching and learning programmes. The second cluster is medical schools with high scores across the majority of EPIC-GP variables with the exception of other practitioners teaching and electives, designated GP-focused programmes. The third cluster is medical schools with medium to high scores across the majority of EPIC-GP variables, including teaching and learning with and by other health professionals, but with the exception of electives. We designate these ‘community- or primary health care-focused programmes’.

This study has a number of strengths. It draws on a large and diverse international sample, incorporating perspectives from leaders of general practice curricula at 44 medical schools across six high-income countries. Access to senior academic informants enabled insight into programme-level curriculum design, strategic priorities, and educational constraints rather than individual teaching experiences. The comparative cross-national approach allowed identification of common patterns and innovations across differing institutional and national contexts. Use of semi-structured interviews provided rich descriptive data, while the systematic and team-based analytic approaches enhanced credibility and transparency.

However, there are several limitations. Medical schools were selected through pragmatic sampling, which may have favoured institutions with existing engagement in curriculum innovation or stronger academic general practice profiles. Data were based on self-reported programme descriptions rather than direct observation or documentary analysis, introducing potential reporting bias. As a cross-sectional study, findings represent a snapshot in time and do not capture curriculum evolution nor longitudinal outcomes. The study did not examine student learning outcomes nor career decisions, limiting conclusions about the effectiveness of identified approaches. Finally, because the sample was restricted to high-income countries with broadly comparable health systems, transferability to other contexts may be limited.

This study did not assess student career intentions, workforce outcomes, or educational effectiveness, and therefore cannot establish links between programme design and subsequent career choice. Our observations related to potential workforce implications are therefore interpretive and informed by existing literature rather than the primary data.

This study provides a descriptive, international comparison of how undergraduate general practice curricula are structured and delivered across medical schools in high-income countries. The findings demonstrate that there is no single dominant model for integrating general practice within medical programmes; instead, institutions employ a range of approaches that balance practice-based learning, campus-based teaching, simulation, longitudinal experiences, and digital modalities. For curriculum planners, these findings offer a menu of feasible strategies that have been adopted in comparable contexts facing similar placement capacity constraints.

For the University of Auckland, the study has practical implications for curriculum redesign. The international examples identified in this study have directly informed our local consideration of how general practice exposure can be redistributed across the programme, including earlier engagement with general practice, greater use of GP-led campus teaching, and selective replacement of in-practice time with simulation and digital learning where appropriate. Importantly, the study supports the view that meaningful exposure to general practice does not rely solely on traditional block placements, which may be increasingly unsustainable.

More broadly, the findings highlight the importance of making general practice visible as an academic discipline within medical schools. Many programmes described embedding GPs as teachers across clinical, communication and case-based learning, suggesting that the way general practice is positioned within the formal and hidden curriculum warrants careful consideration by medical educators.

Thanks to all the educational leaders who freely gave their time to be interviewed about their medical programmes. This work was supported by the John and Jill Richards Professorship Fund under Grant number MAL 020/24.

The authors declare that there is no conflict of interest.