Abstract

Background:

As the field of primary care research continues to grow, it is increasingly important to address the concerns of our trainees. Trainees are central to workforce development and represent the future of the field. Identifying the specific needs and barriers they face in pursuing primary care research is essential to advancing the discipline.

Methods:

In this mixed-methods approach, we analyzed responses to an 33-item cross-sectional survey via REDCap. We performed quantitative analysis using RStudio for Windows (version 2025.05.1) and manually coded overarching themes in Microsoft Excel using an inductive thematic analysis approach.

Findings:

Sixty-nine survey responses were included in the quantitative analysis. A majority of responses were from allopathic medical (MD) students, representing 28.08% of respondents (n = 18), followed by medical residents (n = 17; 24.64%). We received responses from four countries: the United States (n = 45; 65.20%), Netherlands (n = 12; 17.40%), Canada (n = 11; 15.90%), and Uganda (n = 1; 1.40%). We analyzed 66 quotes from 29 participants using an inductive thematic approach, and uncovered nine overarching themes: (1) Guidance and Mentorship, (2) Networking, (3) Training, (4) Time, (5) Funding, (6) Resources, (7) Support, (8) Sustainability, and (9) Institutional Limitations.

Conclusion:

Primary care research trainees face complex challenges such as limited time, funding, mentorship, and research skills, compounded by clinical demands and institutional barriers. Solutions include protected research time, structured mentorship, networking, and equitable institutional support. Future research should identify trainees’ priorities and develop actionable strategies to support primary care research trainees.

Background

Global demand for primary care is accelerating, but investment in an inclusive, well-prepared primary care research workforce has not kept up. At the core of any research workforce are its trainees. Although trainees constitute a smaller proportion of the current research workforce, they represent its future in full. Yet, the specific needs and barriers of trainees pursuing primary care research are underexplored, as existing literature often focuses primarily on family medicine physicians and general practitioners (1–4). In the context of this study, we define a trainee as an individual currently engaged in formal education or postgraduate training, including students, residents, fellows, and other early-career learners (5). Recent initiatives have emphasized workforce inclusion, personnel diversity, trainee engagement, and accessible funding. These initiatives are critical to expanding the primary care research workforce and producing evidence that meaningfully improves the health and lives of patients and communities worldwide.

Persistent workforce shortages further compound the issue of limited production of quality primary care-orientated research (6–11). There is a dire need for primary care services in nearly every community internationally (12), creating a divide between time devoted to research and clinical care for primary care professionals (13). Despite clear potential to address gaps in protected time, funding and grant-writing support, mentorship, research staff, and access to tools and libraries, infrastructure to support primary care research professionals and trainees remains limited. These systemic challenges contribute to a self-fulfilling constraint that primary care research lacks adequate support and recognition (3), protected time (14), and funding (15–17). These concerns are not unfounded and have large impacts on both primary care trainees and established researchers, as well as the primary care field itself (18). Despite limited empirical literature on this topic, our perspectives as trainees and near-peers suggest that research-oriented trainees may view primary care as less conducive to research careers and instead pursue more traditionally research-intensive specialties (e.g., internal medicine subspecialties, surgery, neurology) (19, 20).

In addition to systemic challenges, it is important to understand additional reasons why trainees may not choose to pursue primary care research. To shift the narrative, we must examine why trainees are not pursuing primary care research, why the field fails to attract them, and how it can be transformed into a more compelling and viable option. This study aims to answer these questions through the assessment of training needs, available resources, and barriers encountered by existing trainees (clinical and non-clinical) in their pursuit of a career in primary care research. Through the identification of key areas for improvement, including skill development, access to research infrastructure and funding, and training, we aim to elucidate existing challenges and offer viable solutions.

Materials and methodsStudy design

This cross-sectional study was conducted using an online survey between November 2024 and June 2025. The survey, administered via the REDCap platform, was in English and consisted of 33 items, including demographics. The survey featured both closed-choice and optional open-text questions, capturing quantitative and qualitative data. Survey items were developed inductively by the research team, based on lived experience as trainees and near-peers, and were informed by existing literature in non-trainee populations (21). We did not provide incentives or compensation for participation. The “Assessing the Needs of Trainees” section of the survey (Items 15-33) can be referenced in the Supplementary Appendix Table A. This exempt study was registered and approved by the Institutional Review Board (IRB) at the University of Missouri, Columbia, US (MU#2122129), with additional forms approved regarding international participation.

Recruitment

The study sample included primary care trainees who were recruited using convenience sampling strategies. We recruited through professional primary care-based organizations and societies, including the NAPCRG (previously North American Primary Care Research Group), Association of Departments of Family Medicine (ADFM), Society of Teachers of Family Medicine (STFM), and Australian Association for Academic Primary Care (AAAPC). The survey was distributed via email and online community forums. At the 2024 NAPCRG annual conference, we recruited trainees to complete the online survey. After the conference, we emailed survey invitations using an internal NAPCRG list of trainees. To expand outreach, we asked faculty and principal investigators affiliated with NAPCRG to disseminate the survey to their students, programs, and institutions. Participants self-screened by identifying as trainees engaged in or planning to engage in primary care research. In this study, the definition of a trainee was purposely open-ended to capture a larger pool of trainee perspectives that encompasses both clinical (MD, MD/Ph.D, DNP, etc.) and research-focused (Ph.D, MA, etc.) trainees. Trainees self-reported their specific training under “Currently Enrolled Program” within demographics (see Table 1); however, no additional stratification or data collection occurred regarding the trainees’ year in program or training.

Self-reported demographicsN [n % (n/69)]Currently enrolled program18 (26.1%) MD
17 (24.6%) medical residency
13 (18.8%) Ph.D.
5 (7.3%) MD/Ph.D.
4 (5.8%) master’s
1 (1.5%) bachelor’s
1 (1.5%) DNP
10 (14.5%) other
6 (8.7%) other: post-doc
2 (2.9%) other: fellow
1 (1.5%) other: GP trainee and Ph.D. candidate
1 (1.5%) other: medical resident and Ph.D. studentCurrent country of study45 (65.2%) United States (US)
12 (17.4%) Netherlands
11 (15.9%) Canada
1 (1.4%) UgandaInternational student status*8 (11.6%)Underrepresented in medicine (URiM)**16 (23.2%)First-generation college student20 (29.0%)Low-income or low-SES background19 (27.5%)Race47 (68.1%) White or Caucasian
8 (11.6%) Asian
4 (5.8%) Black or African-American
5 (7.3%) other/not described
2 (2.9%) prefer not to discloseHispanic ethnicity3 (4.4%) Hispanic or Latino
1 (1.4%) Prefer not to discloseGender identity56 (81.2%) women
11 (15.9%) men
1 (1.4%) not described here
1 (1.4%) prefer not to discloseAge range38 (55.1%) 20–30 years of age
24 (34.8%) 31–40 years of age
5 (7.3%) 41–50 years of age
1 (1.4%) 51–60 years of age
1 (1.4%) decline to answer

Self-reported demographics.

*International student status is an international student who is enrolled in an educational institution outside their home country.

**Underrepresented in medicine (URiM) is defined by the American Association of Medical Colleges as “racial and ethnic populations that are underrepresented in the medical profession relative to their numbers in the general population.”

AnalysisQuantitative analysis

Author AW performed quantitative analysis using RStudio for Windows (version 2025.05.1). The goal of this analysis was to explore factors that may influence trainees’ ability to do primary care research. As such, we have employed an exploratory data analysis approach to generate insights on patterns and relationships between the data. The outcomes of interest were trainees’ research skill confidence, training needs, access to resources, and challenges faced in primary care research. We included the following independent variables in our analysis: type of training program, country of study, gender, Under Represented in Medicine (URiM) as defined by the Association of American Medical Colleges (AAMC)’s 2024 definition (22), low-socioeconomic status (SES), international student status, and first-generation college student status. We used five-point Likert scale questions (from 0, not confident, to 4, very confident) to operationalize research skill confidence across seven domains. We calculated mean scores for each domain. Participants ranked the quality of mentorship on a five-point Likert scale (from 0, very poor, to 4, excellent). We operationalized all other variables of interest using discrete and multiple response formats. We calculated variable frequencies for initial descriptive analysis and analyzed response patterns for each question, using individual-level counts of selections.

We used Wilcoxon rank sum (Mann-Whitney U) to test mean confidence and quality scores, as well as counts of selections for multiple-select questions, against the independent variables. This non-parametric test was selected because Likert-scale data may not be normally distributed and Wilcoxon rank sum is robust to violations of normality assumptions. For bivariate analysis, we conducted Chi-square tests of each selection option in the multiple-select questions and the independent variables. When cell counts were sparse (frequencies < 5), a Fisher’s exact test was used to obtain exact p-values for more reliable inferences. Results were deemed statistically significant at the p < 0.05 level.

Inductive thematic analysis

We performed inductive thematic analysis manually using Microsoft Excel to identify overarching themes (23, 24). First, we categorized open-text responses by question. We analyzed the text verbatim, except for clarifying shorthand or to correct misspellings; these clarifications are indicated in brackets. Two coders (KTB, MB) analyzed the open-text responses independently using inductive codes derived from the literature and from our collective experience as trainees. Both coders are members of the NAPCRG Trainee Committee and are medical students in the United States (US). We calculated Cohen’s kappa to assess intercoder reliability (κ = 0.59; SD = 0.26); however, given the small sample of quotes, the statistic is highly sensitive to minor disagreements and may underestimate true agreement. Thus, we also calculated raw percent agreement to contextualize coder consistency. The two coders (KTB, MB) had an average of 90.22% raw percent agreement (SD = 6.05%), using the ReCal2 online tool (25). When consensus could not be reached, we applied the independent inductive codes from a third coder (KN).

FindingsQuantitative

Of the 99 responses received, 69 responses were included in the analysis after removing incomplete (n = 24) and ineligible (n = 6, self-identified as non-trainee) responses. The majority of participants were allopathic medical (MD) students, representing 28.1% of respondents (n = 18), followed by medical residents (n = 17; 24.6%). Participants spanned across four countries, including the US (n = 45; 65.2%), Netherlands (n = 12; 17.4%), Canada (n = 11; 15.9%), and Uganda (n = 1; 1.4%). We include demographic information on race, Hispanic ethnicity, gender identity, and age to showcase the overall diversity of survey respondents alongside the geographic and training diversity described above. The full list of self-reported demographic data is reported in Table 1.

Mentorship

All trainees reported at least one mentorship need (Figure 1 and Table 2). Fifty-seven participants (n = 57, 82.6%) said they have a research mentor who is formally involved in their training, such as an advisor, primary investigator, committee member, or supervisor. Additionally, one participant indicated that they did not have a formal research mentor; however, they went on to rate mentorship quality. Trainees generally rated the quality of their mentorship highly, with a mean score of 3.07 ± 0.14 on a five-point scale (0 = Very Poor, 4 = Excellent; N = 58). Participants often rated their mentorship as “Good” [n = 15, 25.9% (15/58)] or “Excellent” [n = 27, 46.5% (27/58)]. Only a small number of participants reported “Very Poor” [n = 2, 3.4% (2/58)] or “Poor” [n = 3, 5.2% (3/58)] mentorship quality. Approximately 16% (n = 11) of respondents marked this question as not applicable to them. There were no significant differences between the mean quality of mentorship scores when tested by Wilcoxon rank sum tests as reported by low-SES, URiM, First-Generation, or international students compared to their respective counterparts. Fisher’s exact test indicated a significant association (p = 0.044) between international student status and reporting career development as a helpful mentorship aspect. All international students (n = 8, 100%) reported this need, compared to 59.0% (n = 36/61) of non-international students. There was no association between helpful mentorship aspects and country of study or program.

Reported mentorship and training needs of trainees (n = 69 responses). Mentorship needs and skills/training needs are denoted by green and blue, respectively.

DomainItemn% (n/69)Mentorship needsResearch guidance6289.9%Career development4463.8%Networking4362.3%Personal support4565.2%ChallengesLack of time5173.9%Lack of mentorship1927.5%Quality of mentorship1014.5%Limited access to resources2840.6%Insufficient funding3246.4%Limited opportunities for collaboration1623.2%None of these45.8%Access to resourcesStatistical software4058.0%Library access6188.4%Funding opportunities2739.1%Research assistants2231.9%Collaboration opportunities3652.2%Grant writing offices1115.9%Consistent lab/space1623.2%No resources22.9%Skills/training needsStudy design3144.9%Quantitative analysis4463.8%Qualitative analysis3449.3%Manuscript writing2942.0%Grant writing4565.2%Research project management3449.3%No additional training needed68.7%

Full summary table of the self-identified needs, challenges, and available resources of trainees by domain and item selected (n = 69).

Research skills

Trainees’ confidence in research-related skills varied across domains on a five-point scale (0 = Not Confident, 4 = Very Confident), presented fully in Table 3. Participants reported the highest levels of confidence in formulating a research question and presenting findings. High levels of confidence were also reported for conducting literature reviews, qualitative analysis, and writing for publication. Confidence was notably lower in grant writing, with nearly one-third (31.9%) of respondents reporting no confidence in this area. For the research skill of quantitative analysis, approximately 42% of participants indicated low confidence (ratings 0 or 1); the mean score was 2.93 (±0.15).

Research skillMean
score ± SEConfidence rating (n, %)0 = not
confident1234 = very
confidentFormulating a research question3.75 ± 0.110, 0%9, 13%14, 20.3%31, 44.9%15, 21.7%Conducting literature reviews3.67 ± 0.121, 1.4%10, 14.5%14, 20.3%30, 43.5%14, 20.3%Quantitative analysis2.93 ± 0.158, 11.6%21, 30.4%16, 23.2%16, 23.2%8, 11.6%Qualitative analysis3.17 ± 0.157, 10.1%16, 23.2%14, 20.3%22, 31.9%10, 14.5%Writing for publication3.20 ± 0.157, 10.1%15, 21.7%15, 21.7%21, 30.4%11, 15.9%Grant writing2.14 ± 0.1322, 31.9%26, 37.7%10, 14.5%11, 15.9%0, 0%Presenting findings3.75 ± 0.132, 2.9%8, 11.6%16, 23.2%22, 31.9%21, 30.4%

Likert-scale confidence ratings in research skills among trainees to the question: “rate your confidence in the following research skills.”

We used Wilcoxon rank sum test (Supplementary Appendix Table B) to compare the mean confidence ratings in research skills of low-SES, URiM, First-Generation students, and international students, and their counterparts. International students had significantly higher confidence in quantitative analysis (median = 4, IQR = 1), writing for publication (median = 4.50, IQR = 1.2), and grant writing (median = 3.0, IQR = 0.5) than non-international students (respectively: median = 3.0, IQR = 2; median = 3.0, IQR = 2.00; median = 2.0, IQR = 1.0). International students had a slightly higher confidence score in conducting literature reviews (median = 5.0, IQR = 1.25) compared to non-international students (median = 4.0, IQR = 1.00), but this difference was not significant (W = 146, p = 0.054). Finally, those with URiM status (median = 5.0, IQR = 1.25) had significantly higher confidence in presenting research findings than non-URiM trainees (median = 4.0, IQR = 1.00).

Needs for additional research skills training were also examined (Figure 1 and Supplementary Appendix Table B). In a series of Chi-square and Fisher’s exact tests, we found no statistically significant associations between any skill need and low-SES, URiM, international student status, or first-generation status. There were, however, significant associations between countries of study and self-reported needs for quantitative analysis (p < 0.001), qualitative analysis (p = 0.016), and manuscript writing skills (p = 0.002). Participants from the US and Canada were more likely to report needing quantitative analysis training (n = 34, 75.6%; n = 8, 72.7%, respectively) compared to those from the Netherlands (n = 1, 8.3%). Respondents from the US reported the highest need for qualitative analysis (n = 27, 60.0%) and manuscript writing (n = 25, 55.6%), followed by Canada (n = 4, 36.4% and n = 2, 18.2%) and Netherlands (n = 2, 16.7% and n = 1, 8.3%). The sole respondent from Uganda reported all three needs. There was also a significant association between program and need for more training in manuscript writing (p = 0.0005). This need for manuscript writing training was highest among MD students (n = 14, 78%), Bachelors (n = 1, 100%), DNP (n = 1, 100%), and post-graduate trainees (n = 1, 100%). In contrast, the need was much lower among PhD students (n = 2, 15%), those in more than one program (0%), and MD/Ph.D students (0%).

Access to resources

When asked what resources they have access to, most trainees (n = 61, 88.4%) indicated they had library access, while only 11 (15.9%) participants indicated they had access to a grant writing office. Two participants indicated that they had access to no research resources. The most frequent response pattern was selecting only Library Access (n = 12, 17.4%). A full frequency table of the resources trainees indicated that they have access to can be found in Table 2.

A Wilcoxon rank sum test indicated a statistically significant difference in the number of resources selected between those who reported being of low-SES and those who did not (W = 681.5, p = 0.005). Respondents who were not of low-SES selected more resources (median = 3.00, IQR = 2.75) compared to those who reported low-SES (median = 2.00, IQR = 2.00). This indicates a significant disparity in access to resources, with trainees from low-SES backgrounds having access to fewer resources.

A series of chi-square and Fisher’s exact tests on resources available to trainees indicated statistically significant associations between country of study and access to both statistical software (p = 0.002) and research assistants (p = 0.001). Participants from Netherlands were most likely to report access to statistical software (n = 12, 100%) and research assistants (n = 8, 66.7%), whereas access to these resources was lowest among US respondents (n = 22, 48.9%; n = 8, 17.8%, respectively). Fisher’s exact test indicated a statistically significant association between program and access to funding opportunities (p = 0.026). Access to funding opportunities was most common among Ph.D. students (n = 7, 54%), students in “more than one program” (n = 3, 100%), and MD/Ph.D. students (n = 3, 60%). In contrast, access to funding was less common among MD students (n = 4, 22%), medical residents (n = 3, 18%), and not observed among Bachelors students (0%).

Challenges faced in primary care

Regarding challenges faced by trainees in primary care, lack of time was cited most frequently, while quality of mentorship was least (Table 2). A series of Fisher’s exact tests indicated several statistically significant associations between participant demographics and challenges faced. Low-SES respondents were more likely to report “quality of mentorship” as a challenge (n = 6, 31.6%) compared to those without low-SES (n = 4, 8.0%; p = 0.022). International students were less likely to report “lack of time” (n = 3, 37.5%) than non-international students (n = 48, 78.7%; p = 0.024) but more likely to report “insufficient funding” (n = 7, 87.5% vs. n = 25, 41.0%; p = 0.021). Program was statistically significantly associated with the likelihood of reporting several primary care challenges: specifically, lack of time (p = 0.0009), lack of mentorship (p = 0.049), and access to resources (p = 0.049). Lack of time was reported most frequently by medical residents (n = 17, 100%), MD students (n = 15, 83%), and students in more than one program (n = 3, 100%). Lower proportions were observed among Masters students (n = 1, 25%), Bachelors (0%), and post-graduate trainees (0%). Lack of mentorship was most commonly reported by Bachelors (n = 1, 100%), DNP (n = 1, 100%), MD students (n = 8, 44%), and Masters students (n = 2, 50%). The challenge was less common among Ph.D. students (n = 1, 8%), and absent in all other trainees. Access to resources was reported most frequently by MD students (n = 13, 72%) and Ph.D. students (n = 5, 38%). Lower proportions were seen among students in more than one program (n = 1, 33%) and absent in all other trainees. Finally, there was a significant association between gender and reporting “lack of mentorship” (p = 0.046). Women were more likely to report this challenge (n = 19, 33.9%) compared to men, other, or not disclosed (all 0%). There was no association between challenges faced by trainees and country of study.

Qualitative

We had a total of 66 optional open-text responses from 29 participants [42.0% of total included sample size (29/69)]. We identified a total of nine overarching themes from an inductive thematic approach, as described in the Methods: (1) Guidance and Mentorship, (2) Networking, (3) Training, (4) Time, (5) Funding, (6) Resources, (7) Support, (8) Sustainability, and (9) Institutional Limitations. The most common theme response was Guidance and Mentorship, appearing across all four questions. All themes categorized by question posed can be found in Table 4.

What additional support do you need from your mentor or research team? What aspects of mentorship have been the most helpful to you?n/18 (%)Guidance and Mentorship (i.e., planning for the future, next steps of career, research guidance)13 (72.2%)Time (i.e., limited availability, dedicated research time, and the push-pull between clinical and research responsibilities)4 (22.2%)Support (i.e., needs support from institution, project management)4 (22.2%)Networking (i.e., building supportive systems beyond the institution, fostering connections, and collaborating with peers)2 (11.1%)Training (i.e., more training needed, research methodology training)2 (11.1%)Funding (i.e., grant writing, low-salary, limited grants)1 (5.6%)Institutional Limitations (i.e., limited primary care prioritization by institution, lack of investment by field)1 (5.6%)What resources or opportunities do you feel are lacking in your current research environment?n/16 (%)Funding (i.e., grant writing, low-salary, limited grants)7 (43.8%)Resources (i.e., no physical space, limited equipment, no libraries)4 (25.0%)Time (i.e., limited availability, dedicated research time, and the push-pull between clinical and research responsibilities)4 (25.0%)Training (i.e., more training needed, research methodology training)3 (18.8%)Support (i.e., needs support from institution, project management)3 (18.8%)Institutional Limitations (i.e., limited primary care prioritization by institution, lack of investment by field)3 (18.8%)Guidance and Mentorship (i.e., planning for the future, next steps of career, research guidance)2 (12.5%)Networking (i.e., building supportive systems beyond the institution, fostering connections, and collaborating with peers)2 (12.5%)What support would help you overcome these challenges?n/21 (%)Funding (i.e., grant writing, low-salary, limited grants)8 (38.1%)Guidance and Mentorship (i.e., planning for the future, next steps of career, research guidance)6 (28.6%)Time (i.e., limited availability, dedicated research time, and the push-pull between clinical and research responsibilities)5 (23.8%)Institutional Limitations (i.e., limited primary care prioritization by institution, lack of investment by field)5 (23.8%)Networking (i.e., building supportive systems beyond the institution, fostering connections, and collaborating with peers)4 (19.1%)Support (i.e., needs support from institution, project management)3 (14.3%)Training (i.e., more training needed, research methodology training)2 (9.5%)Sustainability (i.e., burnout, wellness, stress)1 (4.8%)Is there anything else you would like to share about your needs as a primary care research trainee?n/11 (%)Funding (i.e., grant writing, low-salary, limited grants)6 (54.6%)Support (i.e., needs support from institution, project management)4 (36.4%)Guidance and Mentorship (i.e., planning for the future, next steps of career, research guidance)3 (27.3%)Time (i.e., limited availability, dedicated research time, and the push-pull between clinical and research responsibilities)3 (27.3%)Sustainability (i.e., burnout, wellness, stress)3 (27.3%)Institutional Limitations (i.e., limited primary care prioritization by institution, lack of investment by field)3 (27.3%)Resources (i.e., no physical space, limited equipment, no libraries)2 (18.2%)Networking (i.e., building supportive systems beyond the institution, fostering connections, and collaborating with peers)1 (9.1%)

Qualitative responses organized by question and corresponding themes, presented in order from most to least represented.

Some codes overlap and are co-coded across themes, and certain themes are not represented within specific questions. Text items in bold indicate the open-ended and free-text question posed to participants in the online survey.

Theme 1: Guidance and Mentorship

Guidance and Mentorship emerged as a main theme in every open-text question posted to participants. Additionally, we identified several subthemes, based on implications from participant responses such as: planning for the future, next steps of career, and research guidance. A sample of quotes that exemplify the theme of Guidance and Mentorship are provided below:

“Research is especially important for pursuing competitive specialties in medicine. So having a mentor who understands this and is determined to help get published work, not just the experience.” [ID#89, Medical (MD) Student in the US]

“Career development and research guidance since the world is heading toward research and project management.” (ID#91, Bachelor’s Student in Uganda)

“At first, research and personal guidance was most important to me. As the years went by, career development played a bigger role.” (ID#85, Medical Resident and Ph.D. Student in Netherlands)

Theme 2: Networking

Respondents frequently emphasized the importance of building supportive systems beyond their institution, fostering connections, and collaborating with peers, which were coded under the theme of Networking. Networking was often coded in addition to the theme Guidance and Mentorship.

“Real networking-not simply a cocktail hour. Hearing how reviewers feel about grants they receive, not simply being told what elements to create. Pilot funding that would allow me to become competitive.” (ID#31, Post-doctoral Student in the US, in regards to overcoming challenges)

“Proper primary care connections through my medical school.” [ID#75, Medical (MD) Student in the US]