BACKGROUND

The field of primary care remains an unattractive career choice for many medical students. Medical schools have made efforts to increase student exposure in the area of primary care, but the proportion of graduates choosing a primary care career has not increased. Indeed, in the United States, the proportion of graduates choosing a primary care specialty fell from 60.7 % in 1997 to 42.1 % in 2006.1 Similarly, between 2005 and 2009, only 28 % of recent graduates in the United Kingdom planned to go into general practice.2 The primary care physician is defined as a specifically trained physician providing first-contact care, taking continuing responsibility for the patient’s care, and dealing with all health problems.3, 4 Depending on the country, this includes general practitioners or family physicians, and may also include general pediatricians and general internists.

Research has highlighted that health systems with a strong primary care base are associated with improved health outcomes and a more equitable health distribution among populations.5 However, generalists make up only about 30 % of all physicians in OECD (Organisation for Economic Co-operation and Development) countries,6 although proportions vary, from 50 % in Australia and Canada, to around 30 % in the United Kingdom, to 12 % in the United States (or approximately one-third if general internists and general pediatricians are counted as well7). General pediatricians make up a small proportion, ranging from under 2 % (Australia) to almost 10 % (United States).6 In order to address the shortage in primary care physicians, medical schools must provide motivation for students to choose a career in primary care.

Specialty choice is complex, influenced by gender, career motivation and life goals,8 as well as attitudes about social responsibility.1 In a review of the North American literature published through 1993, Bland, Meurer and Maldonado identified student- and curriculum-related determinants of primary care specialty choice as described in their paper published in 1995 (Appendix 1).9 In a separate paper in 1996, they published a detailed description of the quality of studies on this topic,10 concluding that “research in this area predominantly uses the weaker study designs and uses few instruments with known reliability or validity”. Based on these findings, the authors suggested strategies to increase the percentage of primary care physicians and to improve research in this field (Appendix 2).

Three other literature reviews have examined primary care specialty choice and have reached similar conclusions (Appendix 1),1113 but were also limited to North American studies. Since Bland and colleagues’ comprehensive review, medical schools around the world have developed new initiatives in recognition of the need to increase students’ exposure to primary care. Nevertheless, the overall proportion of students choosing a primary care career has not risen. Therefore, we conducted a systematic review of the literature published within the last two decades, with the aim of providing an update of our knowledge about the effectiveness of curricular interventions on primary care specialty choice. We included studies from across the world in order to explore the impact of interventions in a global context.

We sought to answer the following questions:

  1. (1)

    Which interventions in undergraduate medical education can increase the proportion of students choosing a primary care specialty?

  2. (2)

    What are the characteristics of successful interventions, and which factors can explain their impact?

  3. (3)

    What is the state of the quality of the published literature in this area, and how has it evolved since the recommendations published in 1996?

METHODS

We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement to guide the reporting of our review,14 and we registered our protocol in the PROSPERO (international prospective register of systematic reviews) database (record number CRD42014009422).15

Definitions

According to the most frequently used definitions of primary care, we included the specialties of family medicine, general practice, general internal medicine, and general pediatrics.

Search Strategy

We searched MEDLINE (PubMed), EMBASE, ERIC (EBSCO interface), CINAHL (EBSCO interface), PsycINFO (EBSCO interface), The Cochrane Library, and Dissertations & Theses A&I on February 20, 2015 for papers published since January 1, 1993. We developed a search strategy for each database by combining the key terms “undergraduate medical student”, “medical education”, “specialty choice”, and “primary care”, as well as numerous synonyms and subject headings. Our full search strategy for MEDLINE is presented in Appendix 3. Adjusted search strategies for the other databases are available on request. We also scanned the reference lists of included studies for relevant papers.

Eligibility Criteria

We included original research on interventions focusing on primary care and targeted at medical students, including curricular components, longitudinal interventions, special curriculum tracks, and extracurricular interventions. We only included studies reporting outcomes related to career choice (career intentions during medical school, specialty choice at graduation, and final practice choice). We included studies published in any language, provided they had an abstract in English. We excluded studies of osteopathic medical students, since osteopathy as practiced in the United States does not exist in other countries, where osteopathy is considered an allied health profession.16

Study Selection Process

One reviewer (EP) performed the search, imported the potentially relevant citations into a reference management program (EndNote version X5.0.1; Thomson Reuters, New York, NY, USA) and discarded duplicates. We used a two-stage process to select studies. All titles and abstracts were screened for exclusion criteria by two reviewers (EP and RT, see "Acknowledgements"), and were included for full-text review if chosen by at least one researcher. Two authors (EP and CC) then independently reviewed all available full-text articles for inclusion criteria. Discrepancies were resolved through discussion with another reviewer (DH).

Data Extraction

Two reviewers (EP and DH) developed a data extraction sheet and pilot-tested it on three articles. EP then extracted data from each article, including study, sample and intervention characteristics, and outcomes relevant for the review. DH independently extracted data from a random sample of six articles to confirm the reliability of the data extraction sheet.

Quality Assessment

All studies were assessed for quality using two scores (rated from 1 to 5) according to published recommendations,17 reflecting the methodological quality of the study and the quality of the information provided in the article. We also graded each article for strength of findings on a scale from 1 to 5 (see Table 1 for details of scores).18

Table 1 Quality Grading of Studies Included in a Systematic Review on Medical Students’ Primary Care Career Choices

Five reviewers (EP, DH, HM, MN, and NJP) pilot-tested the quality scales on six articles. EP then graded all studies, and two authors (DH, HM) each graded half of the studies to ensure inter-coder reliability. Differences of opinion arose for six studies and were resolved by discussion. We did not exclude any articles from the review based on the quality grading, but we recorded the most common threats to quality.

Analysis and Synthesis of Included Studies

We could not identify sufficiently large and homogeneous groups of studies to permit quantitative synthesis given the variability in interventions and outcome measures. We therefore synthesized the results narratively, categorizing them into similar types of interventions.

RESULTS

Study Selection

Our search strategy yielded 2333 unique citations (Fig. 1), of which 68 articles met our inclusion criteria. We identified four additional articles from reference lists of included papers.

Figure 1
figure 1

Systematic review flowchart of the literature search and selection process of studies on medical students’ choice of a career in primary care, published between January 1, 1993 and February 20, 2015

Characteristics of Included Studies

The 72 articles described 66 different initiatives.1990 A summary of the interventions and their impact is provided in Appendix 4. A more extensive summary of outcomes is available in Appendix 5.

Table 2 provides a summary of the characteristics and methods of the included studies. All articles were written in English, with the exception of one article published in German.39 Only 12 % of studies published from 1993 through 1999 took place outside the United States, but this proportion increased to 52 % of studies published from 2000 through 2014. Most studies (n = 65) used quantitative methods. Surveys were the most common data collection method. None of the papers mentioned a conceptual framework or underlying theory for the study.

Table 2 Characteristics of 72 Studies Included in a Systematic Review on Primary Care Specialty Choice

Quality Ratings of Included Studies

Twenty-four studies achieved a high quality score (defined in Table 1). The most common threats to quality were the risk of confounding due to the absence of adjusted statistical analyses, risk of selection bias (through selective admission of students to the intervention), risk of recall bias (in surveys including retrospective items), small sample size, and the lack of a control group (see Appendix 5). The distribution of study quality by regions, intervention type and impact is presented in Table 3.

Table 3 Distribution of 72 Studies Included in a Systematic Review According to Region, Study Quality, Intervention Type, and Impact on Primary Care Career Choice

Impact of Interventions on Primary Care Specialty Choice

Compulsory Primary Care Clerkships

The duration of the clerkships varied from a few days to 12 weeks. The majority (n = 23) took place in ambulatory settings; one clerkship took place in a hospital and four in mixed settings. The most frequently described outcomes were students’ career intentions immediately after the clerkship, with 17 studies reporting a positive effect (Table 3).19,23,28,3033,3639,41,45 One study surveying students after three different clerkships (general practice, internal medicine and surgery) suggested that the positive effect of clerkships may not be specific to primary care.23 The impact on students’ career choices at graduation varied across studies. Most interventions with a positive impact were clerkships in family medicine22,26,30,31,39 or general practice,19,32,34 as opposed to other types of clerkships (such as community placements). Only one study included a long-term outcome—8 years after a general practice placement, physicians’ attitudes were still positively influenced—but this was unrelated to final career choices.25 Outcomes did not differ according to the timing of the clerkship in the curriculum (preclinical versus clinical years, final versus penultimate year).

Some studies suggested that the impact of clerkships depended on their intrinsic qualities, such as student perception of educational value,22 student satisfaction with attending physicians and the quality of teaching,38 or the specialty of teaching physicians.28 In a small qualitative study, graduates stated that contact with family physician role models and the opportunity to see the diverse nature of work in family practice had been important experiences influencing their specialty choice.44

Longitudinal Programs

Of 21 different programs (described in 26 papers), 15 were primary care courses added to the conventional curriculum.4953,5559,6366,6872 The remaining six programs were special curricula or medical schools focusing on primary care and/or rural health.47,48,54,6062, Contrary to the studies on clerkships, the most frequently studied outcomes of longitudinal programs were career choices after graduation (n = 15). Of these, seven programs had a higher proportion of graduates choosing primary care compared to traditional graduates.

All programs but one67 included clinical placements. With one exception,72 all high quality studies were from the United States, and most of them had a positive impact (Table 3). Components of these programs included longitudinal preclinical preceptorships,47,49,50 family medicine faculty advisors,52 clinical rotations at rural or regional sites,47,48,50,52,54 workshops or seminars on primary care,4850 and community-based projects.50 Most of them also recruited students according to predictors of career choice previously described in the literature, including the students’ childhood community and interest in primary care.47,50,52,54

Most programs (n = 19) were offered to students on a voluntary basis, or used specific recruitment and admission procedures to select students interested in primary care, thus introducing a potential selection bias. The majority of studies comparing career choices of program participants to non-participants found that participants were more likely to choose primary care specialties. However, the effect of the selection process could not be separated from the impact of the program itself. Only two studies reduced selection bias by randomly choosing program participants among interested students. One found a significant association between program participation and residency choice,49 whereas the other suggested that the effect of preexisting interest was a stronger predictor of specialty choice.56 One study found an additive effect of a preclinical curriculum and a clinical longitudinal preceptorship on students’ choice of primary care specialty.47

Two qualitative studies explored student attitudes in community-based curricula. In the first, students commented on persistent negative preconceptions about general practice, but stated that good-quality attachments to general practitioners were able to reverse these negative attitudes.72 The quality of clerkships was rated highly when they offered opportunities to see a diverse mix of patients, to have interactive experiences, and to be in a practice team that made them feel welcome. The other study took place in the context of a state-mandated initiative.71 Students felt that primary care was imposed on the school by outside agencies, and that medical school faculty were censored and not allowed to promote subspecialty careers. They also felt unable to obtain unbiased career counseling.

Electives

The duration of electives varied from 3 days to 8 weeks, and most (n = 9) included clinical primary care experiences. Five studies compared participants to non-participants. Of these, three studies found participants more likely to choose a primary care specialty than non-participants.73,74,81 Only one study included a measure of pre-existing interest in primary care, for which the authors found no difference between the two groups.76 Five studies used before-and-after questionnaires to measure the impact of the elective, without using control groups. Three of these studies found that the elective had a positive influence on students’ attitudes towards a career in primary care.75,80,82

Interest Groups

Three papers described student-led interest groups, offering a range of activities for students interested in family medicine or primary care. One study found that students participating in an interprofessional primary care interest group were more likely than non-participants to choose a primary care residency.83 The other two studies suggested a positive influence of a family medicine interest group on student interest in family medicine, but could not ascertain that the group actively influenced career choice.84,85

Student-Run Free Clinics

Participation in a student-run free clinic was found to increase students’ interest in primary care,87 but did not influence specialty choice.86 , 88

Integrated Residency Program

One paper described an integrated residency in general internal medicine, combining the requirements of the last year of medical school and the first year of residency.89 Although a large proportion of participants chose to pursue general internal medicine, the study was too small to allow firm conclusions.

Participation in Primary Care Research

Jones and colleagues interviewed students undertaking a year-long intercalated research degree in primary care.90 The course did not change students’ career intentions, as most of them were already committed to general practice, but students perceived that the course extended their skills and reinforced their career choice.

DISCUSSION

Twenty years after the last comprehensive literature review, as reported by Bland and colleagues, our main conclusions are similar to those in their publication: longitudinal programs are the only strategy that significantly increases the proportion of medical school graduates choosing a primary care specialty, as illustrated by several studies in our review that reported on long-standing longitudinal programs. Most of these programs have been focused within a rural context in the United States and Australia, where primary care physician shortages are an important issue. We found no reports on similar interventions in Europe or other parts of the world (with the exception of one medical school for rural practice in a Japanese study21). Still, primary care physician shortages have developed in several European countries,91 and remain unresolved in emerging countries such as China, India and South Africa.6 Medical schools around the world, therefore, could be inspired by some of these enduring American and Australian programs.

We found conflicting evidence on the impact of compulsory clerkships. They often positively influenced students’ attitudes about primary care, but without affecting career choice. Interestingly, most clerkships with a positive impact were in family medicine or general practice, as opposed to those, for example, in general internal medicine or rural practice. This resonates with Bland and colleagues’ finding that family practice is the only primary care discipline in which a clerkship is correlated with specialty choice. The variety of clerkships in our review explains the fact that our findings are less clear-cut, although they suggest that clerkships can help students in clarifying the role and content of primary care practice, and may act to counterbalance negative preconceptions towards primary care. Electives may further strengthen the interests of individual students and help clarify future career pathways.92

The State of Quality of the Studies in Our Review

We found that the quality of the specialty choice literature had not improved during the last two decades. Our findings reaffirm some of the major threats to quality that had previously been noted, such as the frequent use of instruments without known reliability or validity and the use of study designs at greater risk of bias.10 Observational studies are rarely useful for evaluating initiatives, as the impact of other influences is difficult to estimate.93 Conceptual frameworks are helpful for formulating research questions and discussing generalizability,94 yet none of these were mentioned in any of the studies in our review. Only a minority of studies included several independent variables in their analyses, despite the fact that many influencing factors have been described in the literature, such as gender, background and parental income. The recommendations published in 1996 are therefore still relevant for medical education researchers today. Based on our findings, we suggest several additions to these recommendations (Table 4).

Table 4 Recommendations for Medical Education Researchers for Improving the Quality of Research in the Area of Primary Care Specialty Choice, Based on the Present Review and on Previous Recommendations

Strengths and Weaknesses of the Review

Our review adds a global perspective to the previous specialty choice reviews. Although a majority of studies are still produced in the United States, international publications have increased since 2000. These studies detail initiatives in diverse contexts but also introduce greater heterogeneity. The clerkships, for example, are varied in their duration, setting, and content. This limits their comparability, and may explain some of the contradictory findings across studies. Another explanation is the insufficient use of optimal study designs, threatening the internal and external validity of many of the included studies. The global perspective adds further difficulties, including the various definitions of primary care and variations in the duration of medical education. These issues highlight the difficulties that medical schools may encounter in implementing evidence-based interventions within their own specific context.

We preferred a general score over a strict point-based scoring system for evaluating the quality of the studies. Although this score could be viewed as more subjective, we limited this risk by pilot-testing the score within the review group and by having all articles graded by at least two authors, achieving good reliability. Contrary to previous reviews, we did not exclude studies based on their quality score, but preferred to critically present the full scope of the literature.

Our review has further limitations. It was limited to the pre-graduate medical curriculum, and therefore we cannot draw conclusions regarding the impact of political and other contextual influences. Although we searched several databases, including an educational resource, we did not search further for grey literature, and our search was limited to published papers. We excluded ten papers because we could not retrieve the full-text version. However, we read their abstracts and could not identify findings that would have added evidence.

Implications of the Review

Our findings suggest that longitudinal programs are the most effective for promoting primary care, yet clerkships remain the most frequently described intervention. Building on Bland and colleagues’ recommendations, and based on our findings, we provide updated recommendations for medical educators wishing to make changes to strengthen primary care in their curriculum (Table 5). The impact of any program depends on its quality, which is closely related to preceptors’ teaching behaviors. Good precepting includes actively involving students in patient assessment, giving feedback, and spending time teaching.95 We also encourage medical schools to discuss selection strategies and career support, which must remain unbiased despite our will to promote primary care.96 Although our review is limited to pre-graduate medical education, the broader context of cultural aspects, political influences and financial incentives must be considered. Educational efforts should not be expected to increase the number of primary care physicians if primary care practice is not encouraged within the medical system, as suggested by some of the studies in our review.21,35,41,43,46

Table 5 Recommendations for Curricular Interventions to Increase the Proportion of Medical Students Choosing Primary Care, Based on the Findings of the Present Literature Review and on Previous Recommendations Published in 1995

Conclusions and Recommendations for Future Research

Our findings suggest a need to develop multifaceted strategies organized in coherent longitudinal programs in order to increase the proportion of students who wish to become primary care physicians. We confirm the main findings of a review published two decades ago, whilst adding a global perspective. We also note that the quality of the research has not improved since the last comprehensive review. However, some literature demonstrates that it is possible to conduct controlled experimental studies to evaluate initiatives. It is essential to conduct high-quality research to evaluate new programs and to share the outcomes so that others can replicate successful initiatives. We encourage medical educators to be inspired by Bland, Meurer and Maldonado’s recommendations, which remain highly relevant two decades after their publication, and which we have updated. Last but not least, we must not forget that even a well-designed primary care curriculum will only be successful if the broader context is favorable to primary care.