Status of global health fellowship training in the United States and Canada.

Background Increasing numbers of residency graduates desire global health (GH) fellowship training. However, the full extent of training options is not clear. Objective To identify clinical GH fellowships in all specialties in the U.S. and Canada and to describe their demographics, innovative features, and challenges. Methods The authors surveyed program directors or designees from GH fellowships with a web-based tool in 2017. Results The authors identified 85 programs. Fifty-four programs (63.5%) responded confirming 50 fellowships. One- third of fellowships accepted graduates from more than one specialty, and the most common single-specialty programs were Emergency Medicine and Family Medicine. Fellowships most commonly were 24 months in duration with a median size of one fellow per year. Funding and lack of qualified applicants were significant challenges. Most programs were funded through fellow billing for patient care or other self-support. Conclusion The number of U.S. and Canadian GH fellowship programs has nearly doubled since 2010. Challenges include lack of funding and qualified applicants. Further work is needed to understand how best to identify and disseminate fellowship best practices to meet the diverse needs of international partners, fellows, and the patients they serve and to determine if consensus regarding training requirements would be beneficial.


Introduction
Over the past four decades, interest in global health (GH) among physicians-in-training has increased dramatically. 1-3 GH fellowships -which provide advanced training in GH beyond the clinical requirements of residency -have existed since at least 1997. 4 As GH medical school electives, residency tracks, and fellowships become more common, it is important that trainees, program directors, international partners, and future employers understand the scope and value of these experiences. The first survey of U.S. GH fellowships documented the growing number and variety of GH fellowship opportunities available in 2010 and described program characteristics such as size, duration, specialty, and educational activities. 5 Subsequently, profiles of individual GH fellowships [6][7][8][9][10] and reviews of GH opportunities within subspecialty fellowships [11][12][13][14][15][16] have been published.
However, no subsequent studies have examined trends across all specialties.
Our objectives with this study were to identify all active U.S. and Canadian GH fellowships in all specialties and to describe their features including innovations, challenges, and graduate activities.

Methods
A GH fellowship was defined as formal medical training beyond the usual requirements and length of residency. Fellowships that followed the completion of an accredited residency program or were integrated within a residency program (but extended its length) were included. Fellowships that were solely research-based were excluded to improve comparability amongst programs.
We identified GH fellowship programs from multiple sources, including 1) the Global Health Fellowship Database (globalhealthfellowships.org); 5 2) peerreviewed and gray literatures; 3) epidemiologic e82 snowball sampling, in which participants identified programs not currently listed in the Global Health Fellowship Database; and 4) web searches. Inclusion criteria were programs which: 1) required an additional training period beyond residency requirements, 2) self-identified as 'global health' or were identified as such by others through snowball recruitment, and 3) included a clinical training component.
We contacted fellowship directors or their programs' listed point of contact using publicly-available information. Study participants completed a webbased survey (Survey Monkey, San Mateo, CA). We reminded non-respondents to complete the survey with email and, if needed, telephone reminders. We collected data from March to July 2017.

Results
We identified 85 potential fellowship programs. Fiftyfour programs responded (63.5%), of which 50 (92.6%) offered a GH fellowship (Supplementary Materials, Figure s1). Of the four remaining respondents, two had closed their fellowships, one never had a fellowship, and one is intending to start a fellowship. Thirty-one programs did not respond but were considered probable active fellowships based on careful review of their websites. We requested and received permission to use each program's information such as location and contacts in the Global Health Fellowship Database (globalhealthfellowships.org). Our data reflect survey responses from the 50 confirmed fellowships unless otherwise indicated. Table 1 lists program characteristics such as duration, location, and size. The majority of programs were located on the East Coast of the U.S. (Supplementary Materials, Figure s2).

Fellowship program characteristics
Coursework was primarily completed in resource-rich areas of North America (n=39, 86.7%). Research and policy/advocacy work were primarily done in resource-limited settings in low-and middle-income countries (LMICs) (research: n=42, 91.3%; policy: n=33, 82.5%). Clinical work was commonly performed in resource-rich settings in North America (n=34, 73.9%) and resource-limited settings in both North America (n=24, 52.2%) and in LMICs (n=35, 76.1%) (Supplementary Materials, Table s2). a more than one option could be chosen b self-support includes fellow covering own expenses and/or generating revenue domestically through patient care in clinic, urgent care, hospital, or community health center Abbreviation: LMICs = low-and middle-income countries

Fellowship program challenges and innovations
Program representatives ranked six challenges (6 = most and 1 = least significant). Mean ranking is presented here. Lack of funding (4.5) and qualified applicants (4.1) were ranked most challenging. Lack of political/institutional support (3.7), experienced GH faculty (3.6), fellowship accreditation (2.6), and international placement sites (2.5) were ranked less challenging. e84 Respondents could provide free-text responses for other perceived challenges and innovative or important aspects of their programs (Table 2).

Fellowship graduate characteristics
Respondents estimated that from 2012-2016 their programs each graduated a cumulative total of 0-19 graduates (median 2). Thirteen programs (26.0%) had yet to graduate a fellow so were excluded from postfellowship analyses.
Twenty-six programs tracked their graduates' activities through surveys, interviews, or informal contact. Graduates commonly participated in direct patient care (n=24, 92.3%), education (n=22, 84.6%), and research (n=14, 53.8%). Fewer than half of graduates participated in advocacy, policy development, or administration. Sixteen respondents provided an estimate of the proportion of their graduates working three or more months per year in LMICs (range 0-100%, mean 49.6%).

Comparison of 2010 and 2017 fellowships
In the 2010 survey by Nelson et al., 80 programs in the U.S. self-identified as GH fellowships. 5 However, residency track-only programs were not specifically excluded from that study. Because of the substantial differences in depth of training and oversight between a residency track and a fellowship program, 17

Discussion
We identified 81 total U.S. and Canadian GH fellowships, and 50 programs across various medical specialties responded to our survey. We found that lack of funding and qualified applicants were the greatest challenges for fellowship programs.
The majority of respondents in our survey (95.7%) report some type of fellow self-support as a means of funding the training program. Although complex, current fellowship billing rules provide an opportunity for sustainable global health education programs that serve domestic or (indirectly) international underserved populations. In the U.S., Accreditation Council for Graduate Medical Education (ACGME)accredited fellowship programs (e.g., sports medicine, hospice and palliative medicine, and many others) bill for fellow services at a designated fraction of the fee charged for the same service by an attending. These programs also typically receive some funding through the U.S. government and the hospital in which the fellow is based. However, if a residency graduate joins a non-accredited fellowship (e.g., global health), the fee charged for the fellow's service is the same as the attending physician's fee.
The fellow's income is typically lower than the attending because fellowship programs use some of the receipts to cover expenses related to education and administration of the fellowship. This selfsupport funding model may make training programs more attractive to leaders, decision-makers, and communities. 19 Detailed tracking of GH fellowship graduates is needed to understand the long-term outcomes of training and create a compelling argument for a positive return-on-investment for government funding. [20][21][22][23] We estimate the total number of U.S. fellowship programs (according to our definition) grew from 39 in 2010 to 74 in 2017 (increase of 89.7%). This exceeds growth seen in GH training opportunities for medical students and residents. 2,3 Out of 1,063 U.S. family medicine (FM) residents surveyed who were planning fellowship training, only 2.1% intended to apply for FM GH fellowships. 24 Further study is warranted to determine how well fellowship opportunities match the demand for post-residency GH training. This could include subgroup analysis by specialty, region, or format/content of programs so programs struggling with vacancies could learn from subgroups that excel at recruitment.
e86 Despite challenges, respondents described a multitude of fellowship innovations. Programs reported innovative teaching opportunities, advanced training courses, and varied settings for patient care that were consistent with best practices for international partnerships. [25][26][27][28][29][30] Our study identified many opportunities for growth in the field of GH fellowship training such as improving interprofessional training, building partnerships with tribal councils, honing advocacy skills, and pairing fellows from high-resource and low-resource institutions. In the face of the rapid increase in GH fellowship programs and the common problems of funding and lack of qualified applicants, it is critical to continually reassess and prioritize needs of the international partners to ensure mutual benefit for all participants.
Next steps in the field of GH fellowship training should include discussion amongst U.S. and Canadian program leaders, current and potential international partners, and GH fellows to optimize fellowship structure, funding, and competencies. Preliminary work to define GH competencies at the residency and fellowship level has been published already. [31][32][33][34][35][36] A demographic survey of fellows and potential fellows is needed to inform this work. Understanding factors such as ethnicity, gender, sexuality, and socioeconomic background may help educators and partners prioritize competencies and overcome unintended biases that may be influencing their programs.
While our response rate was higher than typical webbased surveys, 37-39 the actual number of fellowships could be larger if our search failed to identify programs, or smaller, if selection bias led to a greater proportion of closed programs among our 31 nonresponders.
In addition to the fellow demographic study described above, future studies could characterize non-clinical, research-based programs, alternatives for physicians preparing for a career in GH (e.g., diploma or certificate programs in tropical medicine, public health, or health administration) and why some GH fellowship programs have closed. Further study of funding models and matching of high-quality fellowships sites and fellow candidates would be beneficial. Such global fellowships may want to establish a type of voluntary registry so that the data can be updated regularly and changes monitored more easily.

Conclusion
The number of U.S. and Canadian GH fellowship programs has nearly doubled since 2010. Major challenges include lack of funding and qualified applicants. Further study is needed to assess 1) whether the quickly growing number of GH fellowships may have exceeded applicant demand, 2) how training programs can meet the needs of both international partners and a diverse group of fellows, and 3) how to incorporate and align innovations and best practices in education, research, and advocacy to ensure improved patient outcomes. Although our study did not identify any GH program accredited by the ACGME, fellowship program leaders should consider whether consensus on core competencies and minimum training requirements would be beneficial for fellows, their employers, and patients.  27. We would like to ensure that we identify all global health fellowships in the U.S. and Canada.
Please list any global health fellowships of which you are aware (existing or in development) that are not already listed on our database. If possible, please include fellowship program title, location, academic institution, and/or any contact information you may have. 28. Thank you again for your willingness to complete this survey. If you have any additional comments or questions, please feel free to include them below.
29. If you would like us to email you a summary of the survey results, please provide your email address in the space provided.