Update on the Stock and Supply of Health Services Researchers in the United States

Abstract

Objective

This study examines the stock and supply including educational pipeline of health services researchers (HSRers) in the United States.

Data Sources

National Institutes of Health Research Portfolio Online Reporting Tools, Agency for Healthcare Research and Quality Grants On‐Line Database, Health Services Research Projects in Progress, PubMed, Integrated Postsecondary Education Data System, AcademyHealth membership database, and social networks.

Study Design

Exploratory descriptive analysis of individuals involved in health services research (HSR) in the United States as of 2015/16.

Principal Findings

The HSRer stock grew by 25 percent to 45 percent between 2007 and 2015/16, which was slower than the previous measurement period. The growth in the number of doctoral and master's degrees conferred in core HSR fields has been slowing in recent years. Minorities are underrepresented among HSRers, but this trend may improve over time given the diversity of the educational pipeline. Hispanics, however, were generally underrepresented in the educational pipeline.

Conclusions

The average annual growth rate of HSRers appears to be within range of national recommendations. Additional work is needed to assess whether supply of HSRers is meeting demand and to ensure a competent and diverse educational pipeline that meets the needs of an evolving health system.

Investment in health services research (HSR) is an explicit strategy stated in the Department of Health and Human Services’ Fiscal Year 2014 to 2018 Strategic Plan to achieve the goal of strengthening health care and meet the objective of improving health care quality and patient safety (ASPE 2017). Despite this stated strategy, funding to support HSR and training to grow the next generation of health services researchers (HSRers), especially through the Agency for Healthcare Research and Quality (AHRQ), the only federal agency that solely focuses on HSR, has been under threat for decades including recent budget proposals that entirely defund AHRQ (OMB 2017, 2018). As AHRQ fights for sustained funding, they face the challenge of finding up‐to‐date data on the stock and supply of HSRers. Without this data, AHRQ is unable to assess return on investment and determine funding priorities with regards to training. When AHRQ (then known as the Agency for Health Care Policy and Research) experienced declining federal support for HSR education and training, the Institute of Medicine produced a report in 1995 on the state of the health services research (HSR) workforce. Since then, only one other study was published in 2009 to update the stock of the HSR workforce. This study uses methods from prior studies, leverages existing data sources not previously used, and explores new data sources to update the literature on the stock of HSRers from the last count nearly a decade ago. The results from this study may be informative to funders and other stakeholders in HSR including employers and educational institutions as they make decisions on investments in the field.

AcademyHealth defines “health services research” as the “multidisciplinary field of scientific investigation that studies how social factors, financing systems, organizational structures and processes, health technologies, and personal behaviors affect access to health care, the quality and cost of health care, and ultimately our health and well‐being. Its research domains are individuals, families, organizations, institutions, communities, and populations.” The multidisciplinary nature of the training received by the investigators, work conducted by the investigators, and the composition of the team on which they work makes the boundaries of the HSR workforce challenging to define. McGinnis and Moore (2009) identified 10 key disciplines (i.e., health economics, public health, medicine, psychology, political science, health care management, nursing, health policy, public policy, and sociology) to define the HSR workforce (McGinnis and Moore 2009). As the field of HSR expands, these disciplines remain core but expansion of this definition warrants examination as new fields such as translational science, implementation science, medical informatics, and data science emerge. Using these 10 disciplines to define the boundaries of HSR and replicating methods established by the 1995 IOM report, McGinnis and Moore (2009) estimated a stock of 11,596 HSRers in 2007, which they considered to be a dramatic increase over IOM's estimate of 5,000 HSRers in 1995. A National Research Council (NRC) Committee deemed McGinnis and Moore's estimate to be a conservative count (NRC 2011). NRC Committee then recommended that HSR training by AHRQ and NIH be expanded, and the number of HSRers trained each year should be the same as the growth rate of national health expenditures (NRC 2011).

Assessing whether the growth of HSRers has kept pace with NRC's recommendation is not known given that data on the HSR workforce, including researchers working in the field and the educational pipeline, are not systematically collected. Monitoring the growth of HSRers is important for many stakeholders in addition to AHRQ. There are several other major funders of HSR such as Robert Wood Johnson Foundation, Patient‐Centered Outcomes Research Institute, National Institutes of Health, and Veterans Affairs that need this information to make funding determinations. Policy makers are debating how major policies such as the Affordable Care Act of 2010 (ACA) and Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) are impacting access and affordability of health insurance. Health systems want to monitor the impact of new delivery models such as patient‐centered medical homes and Accountable Care Organizations on quality of care. HSRers are specifically trained, and can be employed, to answer these questions. Universities offering programs in HSR need to know whether there is continuing interesting to support the existing programs, and if they need to expand (or reduce) their programs. Advocates and professional organizations supporting HSR such as AcademyHealth need to understand the size and characteristics of their population to ensure they are representing their base.

This study fills the gap in knowledge by replicating methods used in previous studies to update the literature describing the stock and supply of HSRers in the United States, which provides an additional data point for monitoring the growth of HSRers. This study also leverages data not previously used to monitor the educational pipeline of students studying in HSR‐related fields and explores the value of new data sources from social networking sites. This study ends with a discussion of the challenges of defining the boundary of the evolving HSR workforce, the implications to stakeholders of HSR, and recommendations on data collection moving forward.

Data and Methods

This study replicates to the extent possible the past descriptive approaches used by the 1995 IOM Committee on Health Services Research and a follow‐up study published by McGinnis and Moore to create an unduplicated count of HSRers (Field, Tranquada, and Feasley 1995; McGinnis and Moore 2009). Detailed information on data and methods is available in Appendix SA1.

AcademyHealth Membership and Activities

As one of the largest professional societies for HSR, we mined the online AcademyHealth membership database to identify characteristics of HSRers. The database included searchable fields of industry, type of work (primary and secondary), expertise, degree, and location. We extracted the total number of members in the United States who identified “health service research” as the primary type of work and secondary type of work. We also obtained a list of presenters from one of AcademyHealth's largest HSR events, the Annual Research Meeting, which is a national HSR forum attracting both HSRers and non‐HSRers (AcademyHealth 2015).

Social Network Membership

Since the last publication of the supply of HSRers, several professional social networks have emerged. We explored ResearchGate.net, which launched in 2008 and has been a rapidly growing social network of authors through their co‐authorships and citations (ResearchGate.net n.d.). Although ResearchGate did not specifically use the term “health services,” they used a closely related category of “healthcare policy and economics.” We also explored LinkedIn.com, which launched in 2003, to identify individuals who listed the skill “health services research” (LinkedIn.com n.d.). While we could not extract specific names, we focused on obtaining information regarding education, years of experience, employment location, and job function. We compared this count of individuals with those that self‐identified into a LinkedIn group called “health services research” as well as those in AcademyHealth's LinkedIn group.

Journals

We identified individuals actively conducting HSR in two ways—publications and grants. We used PubMed to extract the first three authors (i.e., last name, first initial, and middle initial) in Medical Care and Health Services Research (NLM n.d.3). Allowing for a growing number of HSRers, we used a 5‐year publication window (January 1, 2010 to December 31, 2015) rather than the 1‐year publication window used by McGinnis and Moore (2009), allowing for the possibility that active HSRers may not publish each year and that authors with recent publications continue to work in the HSR field. We restricted our search to those published in the English language, studies conducted on humans, and studies published in MEDLINE. We used Medical Subject Heading (MeSH) terms to assess the range of journals in which HSRers published and the number of publications by HSRers that appeared using those terms (see Appendices SA2 and SA3 for MeSH terms) (NLM n.d.2).

Investigators

We used three sources to identify individuals leading HSR studies—National Institutes of Health (NIH) Research Portfolio Online Reporting Tools (RePORTER), Agency for Healthcare Research and Quality (AHRQ) Grants On‐Line Database (GOLD), and HSRProj. We extracted a list of PIs (i.e., last name, first name, and middle initial), performing organization, and administering institute or center from NIH RePORTER for federally funded grants conducted in fiscal years 2010 to 2015 (NIH n.d.). Again, the 5‐year window allows for the possibility that active HSRers may not obtain a grant every year and that investigators with recent grants continue to work in the HSR field. We selected NIH grants that were categorized under the NIH spending category called “health services” awarded in the United States. We captured PIs receiving grants from Department of Veterans Affairs (VA) through NIH RePORTER if the project term “health services” was identified. We identified AHRQ grant recipients using a combination of NIH RePORTER and AHRQ GOLD (AHRQ n.d.). We included investigators listed in HSRProj, a voluntary database of research‐in‐progress in HSR housed at the National Library of Medicine that was used in prior studies on the HSR workforce (NLM n.d.1). HSRProj allowed us to identify HSR‐related grants with support from foundations, Centers for Disease Control and Prevention, Health Resource and Services Administration, Centers for Medicare and Medicaid Services, and other grant‐making organizations. AcademyHealth staff provided an unduplicated list of investigators names (i.e., last name, first name, and middle initial), performing organization, and sponsoring agency for projects with an initial start year between 2010 and 2015.

Pipeline

To examine the educational pipeline of doctoral and master's level students training in HSR‐related programs, and identify a pool of students who could potentially enter the field of HSR, we used the Integrated Postsecondary Education Data System (IPEDS). IPEDS is a mandatory reporting system of data on institutional characteristics and resources (e.g., degrees and certificates conferred) collected by Department of Education National Center for Education Statistics from universities that receive federal financial assistance (e.g., Title IV of the Higher Education Act of 1965 and Title VI of the Civil Rights Act of 1964) (NCES n.d.2). We extracted the number of degrees conferred between 2010 to 2015 to students in master's degree programs and doctoral degree programs in research/scholarship (vs. professional practice or “other”) from public and not‐for‐profit universities in the United States. We used the 2010 Classification of Instructional Programs (CIP) codes to identify a core set of programs and an expanded set of programs related to HSR (see CIP list in Appendix SA4) (NCES n.d.1). We extracted information on the gender, citizenship, and racial/ethnic profile of these students from IPEDS.

Comparison of Approaches to Create Unduplicated Count of HSRers

The 1995 IOM report used three sources to create an unduplicated count of self‐identified HSRers: (1) membership files from AcademyHealth (then known as the Association for Health Services Research), (2) a list of PIs of projects reported in HSRProj, and (3) names obtained from brochures of health research centers, AcademyHealth annual meetings, and honorary societies (Field, Tranquada, and Feasley 1995). Similar to IOM, McGinnis and Moore (2009) extracted names from member lists obtained from AcademyHealth, PIs in HSRproj, and participants in AcademyHealth activities (e.g., conference presenters and interest group members). They added names of authors from major HSR journals and a combined list of attendees and presenters at select conferences (McGinnis and Moore 2009). Our study used the list of PIs from HSRProj and added names from NIH RePORTER and AHRQ GOLD; participants in AcademyHealth annual research meetings, excluding IG presenters; and authors from the same HSR journals identified by McGinnis and Moore (2009), but using a 5‐year versus 1‐year time horizon. We did not use names from the AcademyHealth membership directory.

Results

Count of Health Services Researcher Workforce

Following McGinnis and Moore's approach of counting unique authors in two key HSR journals, investigators listed in HSRProj, and presenters at the 2015 Annual Research Meeting of AcademyHealth, we found 14,526 uniquely identified HSRers as of 2015 (Table 1) compared to the 11,596 HSRers in 2007 (McGinnis and Moore 2009). Including investigators from NIH RePORTER and AHRQ GOLD, we identified an additional 2,217 HSRers, for a total of 16,743 (Table 1) as of 2015.

Table 1.

Number of Health Services Researchers Identified from Sources

Sources	Count
AcademyHealth Annual Research Meeting Presenters (2015)	5,792
Select Journals (First three authors published from 2010 to 2015)
Health Services Research (714 articles)	1,500
Medical Care (1,209 articles)	2,580
Sponsored Projects (Investigators listed between 2010 to 2015)
NIH RePORTER and AHRQ GOLD (16,041 grants)a	5,031
HSRProjb	6,630
NIH, AHRQ, VA sponsoring agencies	4,740
Other sponsoring agencies	1,890
Duplicated count	21,533
Unduplicated count to compare with 11,596 health services researchers found in 2011 by McGinnis & Moore (ARM, Select Journals, HSRProj)	14,526
Unduplicated count expanded (including NIH RePORTER, AHRQ GOLD)	16,743
AcademyHealth Membership Directory (2016)
All members in the United States	4,343
Type of work identified as “Health Services Research”	1,395
Secondary type of work identified as “Health Services Research”	163
LinkedIn.com (2016)
Self‐identified skill of “Health Services Research” in the United States	6,126
Members of “Health Services Research Group”	6,331
AcademyHealth Group	3,206
ResearchGate.net (2016)
Researchers in “Healthcare Policy and Economics” in the United States	20,300

^aPIs in FY 2010 to 2015 in United States; NIH Spending Category called “Health Services” for NIH grants; project term search includes “Health Services” for VA grant source.

^bInvestigators in project initial year 2010 to 2015.

AHRQ GOLD, Agency for Healthcare Research and Quality Grants On‐Line Directory; HSRProj, Health Services Research Projects in Progress; NIH RePORTER, National Institutes of Health Research Portfolio Online Reporting Tool.

Source: Author extraction and calculation from AcademyHealth (2015, n.d.); AHRQ (n.d.); LinkedIn.com (n.d.); NIH (n.d.); [Link], [Link]; ResearchGate.net (n.d.).

Compared to other sources of data obtained in 2015/16, AcademyHealth's membership data included 1,395 members that primarily identify as HSRers out of the 4,343 total U.S. members. On LinkedIn, 6,126 members self‐identified HSR as a skill and 6,331 members were in the LinkedIn “Health Services Research Group,” two groups which likely included duplicates. ResearchGate had the highest count at 20,300 researchers in health care policy and economics, a count based on authorship.

MeSH Terms and Publications

Using the MeSH term of “health services research” alone as a major topic heading (i.e., main topic of the paper as determined by NLM MEDLINE indexers), we found 12,184 articles between 2010 and 2015 (Appendix SA5). The most common journals in which articles were identified using HSR as the major topic are listed in Appendix SA6; Medical Care and Health Services Research were not among the top five most common journals. None of the top five journals listed were among the expanded list of journals in the study by McGinnis and Moore (2009).

On the upper end of the estimated number of potential contributors to HSR, the broader MeSH term of “health care” as the major MeSH topic heading resulted in 750,000 articles between 2010 and 2015 (see Appendix SA5 for the number of articles that were identified within a 5‐year window using each of related health services MeSH subheadings). Undoubtedly, these articles were not all related to HSR and included multiple authors as well as duplicate authors.

Characteristics of Health Services Researchers

Although we did not have a source of information with age, we gained clues about the age distribution of HSRers based on information on seniority and years of experience. About 20 percent of LinkedIn HSRers were working at an entry level in 2016 (Table 2). The majority of LinkedIn HSRers was working at a senior level, and/or held manager or director titles in their organization. Generally aligning with the distribution of job titles, two‐thirds of LinkedIn HSRers had over 10 years of experience with another 27 percent with 6 to 8 years of experience (Appendix SA7). While bias could be present due to the voluntary nature of data entry into these social networks, the experience distribution suggests that social networks are not only for a younger generation.

Table 2.

Characteristics of Health Services Researchers from LinkedIn.com, 2016

Characteristics	Health Services Researchers (N = 6,126)
	N	%
Job function (Five Most Common)
Research	2,270	37.1
Education	2,294	37.4
Health care services	2,353	38.4
Program and project management	420	6.9
Consulting	410	6.7
Degrees
PhD	2,445	39.9
MD	1,936	31.6
ScD	142	2.3
DrPH	276	4.5
MPH	1,865	30.4
MS	1,011	16.5
Seniority
Unpaid	24	0.4
Training	110	1.8
Entry	1,140	18.6
Senior	3,525	57.5
Manager	1,514	24.7
Director	1,248	20.4
VP	298	4.9
CXO	366	6.0
Partner	161	2.6
Owner	147	2.4

Multiple responses to categories are allowed so columns do not add to 100%.

Source: Author extraction from (LinkedIn.com n.d.).

According to ResearchGate, the trend among publishing authors in health care policy and economics was the opposite of LinkedIn with about two‐thirds of ResearchGate authors having 3 to 6 years of experience and another 17 percent with less than 3 years of experience in 2016. While the methods behind the years of experience calculation were not transparent, and given that ResearchGate relies on publications, years of experience may be reflective of the number of years over which an individual published. The experience trends reported by ResearchGate were aligned with the experience level reported by respondents to AcademyHealth's 2013 salary survey (AcademyHealth 2013).

Only 40 percent of LinkedIn HSRers identified as having a PhD. About a third of LinkedIn HSRers had a medical degree (note that individuals could identify multiple degrees). This trend closely aligned with the highest degree reported by respondents in AcademyHealth's 2013 salary survey (AcademyHealth 2013). The share of individuals with a medical degree was consistent with the 38.4 percent of LinkedIn HSRers who stated that their job function included health care services (individuals could identify multiple job functions).

Where Health Services Researchers Work

According to the AcademyHealth membership database in 2016, half of those members who identified HSR as their primary type of work were in universities (Figure 1). The next most common industry of work was a research or policy organization that was not a university (10.7 percent). These findings were relatively consistent with the finding that about three‐quarters of LinkedIn HSRers identified research and/or education among their job functions (individuals could identify multiple job functions) (Table 2). Despite only 3.9 percent of AcademyHealth members identifying the Federal Government as their place of work, the VA and CDC had the highest share of HSRers according to LinkedIn (Table 3). Universities, however, were likely to be the dominant employer if one were to aggregate the counts across all educational institutions.

Figure 1.

Industry in Which AcademyHealth Members Identified as Health Services Researchers Work, 2016 (N = 1,395). Source: Author extraction from (AcademyHealth n.d.) [Color figure can be viewed at http://www.wileyonlinelibrary.com/]

Table 3.

Most Commonly Listed Performing Organization by Data Source

Rank	LinkedIn, 2016 (N = 6,126 people)		NIH RePORTER, 2010–2015 (N = 16,041 grants)		NIH, AHRQ, VA Funded Studies in HSRProj, 2010–2015 (N = 4,740 studies)		Other Funded Studies in HSRProj, 2010–2015 (N = 1,890 studies)
	Performing Organization	N	Performing Organization	N	Performing Organization	N	Performing Organization	N
1	Department of Veterans Affairs	159	Johns Hopkins University	514	University of Michigan	126	University of North Carolina, Chapel Hill	76
2	Centers for Disease Control and Prevention	146	University of Michigan	447	Johns Hopkins University	123	University of Pennsylvania	46
3	University of California, San Francisco	113	University of Washington	444	University of Washington	100	University of Washington	43
4	Harvard Medical School	94	University of California, San Francisco	437	University of California, San Francisco	96	Harvard University	35
							Johns Hopkins University	35
							University of Michigan	35
5	University of Washington	92	University of Pennsylvania	429	University of Pennsylvania	88	University of California, San Francisco	32
6	University of California, Los Angeles	91	University of California, Los Angeles	391	University of North Carolina, Chapel Hill	85	University of California, Los Angeles	28
7	University of North Carolina, Chapel Hill	82	University of North Carolina, Chapel Hill	342	Oregon Health & Sciences University	80	Columbia University	26
8	Johns Hopkins Bloomberg School of Public Health	70	University of Pittsburgh at Pittsburgh	293	Columbia University	78	Duke University	22
9	RAND Corporation	63	Yale University	278	University of California, Los Angeles	72	University of Minnesota	21
10	Brigham and Women's Hospital	60	Duke University	276	Duke University	71	University of Illinois at Chicago	19
							Mathematica Policy Research	19
Total		970		3,851		919		167
Share of N		15.8%		24.0%		20.6%		8.8%

Source: Author extraction and calculation from AcademyHealth (2015); LinkedIn.com (n.d.); AHRQ (n.d.); NIH (n.d.), and NLM (n.d.1).

Another indicator of where HSRers worked is the performing organizations of a sponsored project (Table 3). While each sponsored project did not necessarily translate to a single unique HSRer, a handful of organizations consistently appeared across all data sources such as Johns Hopkins University, Harvard, University of Washington, University of North Carolina at Chapel Hill, and the University of California system. Interestingly, the top 10 universities with the highest share of funded projects represented about one‐fifth to a quarter of all funded projects when looking across both NIH RePORTER and HSRProj.

Educational Pipeline

We found that a total of 2,667 individuals were conferred a research/scholarship doctorate in one of five “core” HSR fields (see Appendix SA4 for codes used to define “core” HSR fields). Using our expanded definition, there were an additional 7,000 doctoral students that could be counted among the pipeline of HSRers (see Appendix SA4 for codes used to define “expanded” HSR fields). The doctoral pipeline for the core fields slightly fluctuated in size over time and was generally slowing in recent years. The expanded fields saw a decline over time, particularly in nursing sciences, which was one of the largest contributors to the count of the HSR pipeline under our expanded definition of the HSR.

The pipeline of master's degree students was about 30 times as large as the pipeline of doctoral students using the core HSR definition. There was a similar split in students between public health and the more specifically defined health services/health care programs. Using the expanded definition of HSR, there were about 10 times as many master's students as doctoral students. The number of master's degrees conferred dipped in 2012/2013, returning to 2011/2012 levels by 2014/2015. In the expanded definition of HSR, the number of students with relevant master's degrees slightly increased over time, particularly in medical informatics and biostatistics.

The most common institutions that conferred doctoral or master's degrees in HSR‐related fields had some overlap, with the most common performing organizations receiving funded projects such as Johns Hopkins University, University of North Carolina at Chapel Hill, and University of California campuses (Appendix SA9). Online universities such as Webster University were among the institutions training a large volume of master's‐level HSR students.

Diversity

While we were not able to obtain data on the diversity of the stock of HSRers, we gained clues about the future HSR workforce based on the 5‐year demographic trends of the educational pipeline of students in core fields related to HSR. The HSR workforce appeared to remain female dominated as of 2015/16 with an increasing share of women over time (Table 4) compared to 56–60 percent women in HSR workforce identified in McGinnis and Moore's study as well as AcademyHealth's 2013 salary survey (McGinnis and Moore 2009; AcademyHealth 2013).

Table 4.

Diversity of Pipeline in Core Fields, 2010 to 2015

Year	2010–2011		2011–2012		2012–2013		2013–2014		2014–2015
Completed master's degree
Females	10,087	69.4%	11,505	70.4%	10,905	70.0%	11,255	70.5%	11,495	70.4%
Hispanic, any race	840	5.8%	890	5.4%	949	6.1%	1,094	6.9%	1,097	6.7%
Non‐Hispanic
White	7,195	49.5%	8,395	51.4%	7,932	50.9%	7,918	49.6%	8,130	49.8%
Black/African American	2,053	14.1%	2,400	14.7%	2,282	14.7%	2,352	14.7%	2,287	14.0%
Asian	1,448	10.0%	1,775	10.9%	1,678	10.8%	1,765	11.1%	1,817	11.1%
American Indian/Alaskan Native	79	0.5%	92	0.6%	80	0.5%	81	0.5%	72	0.4%
Hawaiian Island/Pacific Islander	24	0.2%	41	0.3%	36	0.2%	31	0.2%	33	0.2%
Multi‐race	141	1.0%	224	1.4%	257	1.7%	315	2.0%	379	2.3%
Unknown race/ethnicity	1,293	8.9%	1,170	7.2%	1,027	6.6%	1,145	7.2%	1,182	7.2%
Non‐citizen	1,471	10.1%	1,350	8.3%	1,332	8.6%	1,268	7.9%	1,342	8.2%
Total	14,544		16,337		15,573		15,969		16,339
Completed doctoral degree—Research/scholarship
Females	259	62.4%	371	68.2%	376	63.5%	373	69.3%	394	68.2%
Hispanic, any race	19	4.6%	24	4.4%	23	3.9%	29	5.4%	37	6.4%
Non‐Hispanic
White	171	41.2%	228	41.9%	266	44.9%	272	50.6%	300	51.9%
Black/African American	55	13.3%	81	14.9%	64	10.8%	90	16.7%	102	17.6%
Asian	42	10.1%	64	11.8%	42	7.1%	31	5.8%	46	8.0%
American Indian/Alaskan Native	2	0.5%	3	0.6%	1	0.2%	2	0.4%	4	0.7%
Hawaiian Island/Pacific Islander	0	0.0%	1	0.2%	1	0.2%	2	0.4%	2	0.3%
Multi‐race	1	0.2%	2	0.4%	6	1.0%	8	1.5%	6	1.0%
Unknown race/ethnicity	67	16.1%	69	12.7%	101	17.1%	42	7.8%	29	5.0%
Non‐citizen	58	14.0%	72	13.2%	88	14.9%	62	11.5%	52	9.0%
Total	415		544		592		538		578

Hispanic and Non‐Hispanic rows add to 100%. Core fields include Public Health, General; Health/Health Care Administration/Management; Health Policy Analysis; Health Services Administration; Health Services/Allied Health/Health Sciences, General. Fields are defined using the Classification of Instructional Programs.

Source: NCES (n.d.2).

According to AcademyHealth's Membership Survey, the percentage of minorities increased over time (18.0 percent in 2013 v. 22.1 percent in 2016 of members were non‐white or Hispanic in 2013), although minorities were generally underrepresented among AcademyHealth's membership relative to the general population (AcademyHealth 2013; Edmunds, Johnson, and Wolfe 2018). The pipeline of HSRers appears to be more diverse than the AcademyHealth membership with approximately half of those who completed a master's degree or doctoral degree in a core field were non‐white or Hispanic (Table 4). Black/African Americans were the dominant minority group among those completing a master's (14.0 percent) and doctoral degree (17.6 percent) in a core HSR field. The percentage of Hispanics identified in the core educational fields of HSR, however, was underrepresented when comparing to the percentage of Hispanics reported in the general population.

The percentage of minorities completing a master's degree had been relatively stable between 2010 and 2015, with a slight gain among Hispanics. The percentage of Hispanics also grew among those completing a doctoral degree. The increase in Hispanics came alongside a decline in the reporting of “unknown race/ethnicity” among those receiving either a master's or doctoral degree, and a decline in the share of non‐citizens and Asians completing a doctoral degree in a HSR‐related field. Notably, white, non‐Hispanics gained 10.7 percentage points, and black/African Americans gained 4.3 percentage points among those who completed a doctoral degree in a HSR‐related field.

Discussion

This study found that the stock of HSRers increased by about 3,000 HSRers, or 25 percent, between 2007 and 2015 when replicating to the extent possible the methodologies of previous studies. Under our expanded approach that added PIs found in NIH RePORTER and AHRQ GOLD, the number of HSRers increased by 5,000, or approximately 45 percent, compared to the count in 2007. This rate is less than the doubling of the HSRer workforce between 1995 and 2007. The growth of the HSRer workforce is fast relative to the overall health care workforce, which grew by about 20 percent over the last 10 years, which has made health care the fastest growing industry in the economy (Frogner 2018). The annual growth rate of HSRers was 7.2 percent between 1995 and 2007, and 2.9 to 4.7 percent between 2007 and 2015. Using our lower end estimate of HSRers, the growth rate is lower than the National Research Council (US) Committee to Study the National Needs for Biomedical, Behavioral, and Clinical Research Personnel (NRC) (2011) recommendation that the number of HSRers should grow at the rate of national health expenditures, which has been 4.2 percent over this time frame (CMS 2018); our higher end HSRer estimate is on par with the recommendation. In either scenario, these trends may be useful for AHRQ and other federal agencies to argue that the current funding rates for HSR should be sustained, if not increased, to maintain these growth rates. Caution is needed in the interpretation of these growth rates, the descriptive nature of the study, and lack of statistical tests for significance.

Despite the apparent growth in overall HSRers and a threefold to fourfold increase in the number of doctoral and master's students compared to what was estimated in 2007 (Ricketts 2009), the educational pipeline appears to be slowing in recent years. The number of doctoral degrees conferred in the core HSR fields has remained relatively flat in growth over the last few years and has been declining in the expanded HSR fields. In core HSR fields, the number of master's degrees conferred dropped in 2012/2013, resulting in an overall slowdown in master's degrees conferred over the last 5 years. Master's degrees conferred grew, however, when considering degrees in medical informatics and biostatistics. This slowdown in the educational pipeline may be concerning to educational institutions that rely on tuition revenue and research assistance to support faculty productivity. It is not clear whether this slowdown is due to lack of interest by students as a reflection on of their perception of market demand for HSR skills, and/or lack of available federal funding to attract and support students.

This study faces several challenges that make assessing the adequacy of the stock and supply of HSRers difficult. NRC stated that the approach used by McGinnis and Moore (2009) resulted in a conservative estimate of HSRers. This suggests that our count of 14,526 to 16,743 may be on the lower end of the HSRer count. Finding the “true” number is a challenge without a consensus on the boundaries of HSR and consistency in the use of data. We are not able to identify an error bar around our estimates given that our approach is not based on a statistical sampling. While we attempted to replicate their approach, there are subtle yet important differences in our approach and that of McGinnis and Moore, including the range of years used to count authors and PIs as well as the source of PI information. Also, McGinnis and Moore reported a higher estimate of 19,203 HSRers in 2007 when they considered names from an expanded set of conferences and journals. We do not have an equivalent comparison group because it was not clear which among the burgeoning number of conferences and journals that are relevant to HSR to include. Across the various sources we explored, ResearchGate resulted in the highest count at 20,300 individuals related to HSR, which is only slightly higher than McGinnis and Moore's highest count.

This study is not able to assess whether the stock or supply of HSRers is adequate to meet the demand for HSR. It is assumed that with the passage of policies such as ACA and MACRA, and societal concerns about the affordability and accessibility of high quality of care, that policy makers, employers, and other health care leaders are demanding an evidence base to evaluate these changes and to create new policies. The number of HSRers needed to produce this evidence base is unknown. More work is needed to connect how HSR is disseminated, consumed and used to influence policy to understand the value of recent HSR investments and to help make decisions on how to prioritize funding to support HSR activities. This gap could be filled through surveys of HSR consumers as well as better data collection and monitoring of how HSR is cited and referenced in policy debates.

A challenge in conducting this study was the lack of available data to clearly identify HSRers. We relied on a combination of sources to identify HSRers, many of which overlap and are not easily combined to remove duplicates. Even though many journals have appeared over time that publish HSR‐relevant papers, identifying those HSR‐relevant papers and authors that identify as an HSRer is not clear. Authors who only publish within the gray literature are not easily identifiable or extractable from public databases if captured at all. Individuals who do not publish or seek funding may only be identified through sources such as the AcademyHealth membership database or LinkedIn. Social networking sites like LinkedIn, however, are a relatively untested data source for monitoring workforce, not universally used, and may not be kept up‐to‐date.

An additional challenge was defining the boundaries of HSR and the career pathway by which one becomes a HSRer. This study focused only on the HSR workforce in the United States, future studies should consider the number of HSRers abroad, and the engagement of U.S. HSRers with their international counterparts through the use of, for example, social network analysis. Also, many individuals may have entered HSR without formal training through team‐based research or through applications of methods such as economics and data science to health care topics. Individuals may be conducting HSR without realizing they were doing so because they either did not know of the field of HSR or identified primarily under a different field such as health economics, clinical practice, data science, or translational science. As such, this study may have missed a broad swath of people who do not enter the realm of AcademyHealth, and rather participated in other professional societies such as American Society of Health Economics, International Society of Pharmacoeconomics and Outcomes Research, Society of General Internal Medicine, Pediatrics Academic Societies, and so on. On the other hand, we may be overestimating the number of HSRers by assuming that every person identified as an author in a HSR journal or conducting a study related to HSR considers themselves an HSRer. Our count is high when one considers that only 1,395 declared HSR as their primary field of work with an additional 163 declaring it as their secondary field among the over 4,000 AcademyHealth members. Similarly, only 6,126 identified HSR as a skill in their LinkedIn profile.

Conclusions

We found that the number of HSRers appears to have increased compared to prior studies, but the growth may be slowing, especially when considering the educational pipeline trends. The future HSR workforce may be increasingly diverse, although Hispanics are underrepresented in the pipeline. Findings from AcademyHealth's Membership survey suggest that minorities are underrepresented among HSRers, but this trend may improve over time given the diversity of the pipeline. More work, however, could be done to recruit minorities into the HSR field, and to cultivate mentorship of minorities to ensure that diversity is reflected across leadership ranks. Future monitoring of whether the diverse pipeline eventually translates to a diverse leadership in senior ranks in HSR is recommended.

A potential threat to the future HSRer stock is the available funding for HSR predoctoral and postdoctoral training (Mor and Wallace 2018). Some schools have recognized the financial challenges that students face in financing their education, and thus have restricted the pipeline of HSR students to a number that they were able to ensure support throughout their doctoral education, which may be reflected in the slowdown in doctoral degrees conferred over the last few years. Better monitoring of the long‐term outcomes of individuals who received training funds at both the predoctoral and postdoctoral levels is needed to understand the impact of these training funds on the supply of HSRers. Also, further investigation is warranted on whether underrepresented minority HSRers receive a proportionate share of funding, and to identify any potential barriers that may impede this goal.

Future efforts are needed to streamline data collection to monitor the HSR workforce. Given the growing availability of data online, infrastructure could be designed in such a way to add sources of data such as conference program or other social networking sites to provide a better and richer profile of the HSRers. Also, with the growing availability of web crawling and data mining tools, a relatively minimal investment could result in high dividends to set up an infrastructure that could more accurately and regularly monitor the supply (and perhaps also demand) of the HSRers.

Supporting information

Appendix SA1: Additional Details on Data and Methods.

Appendix SA2: Medical Subject Heading (MeSH) Terms: Health Care Category.

Appendix SA3: Definition of MeSH Term “Health Services Research.”

Appendix SA4: Classification of Instruction Programs.

Appendix SA5: Article Count by Key Medical Subject Headings (MeSH) as Major Topic, 2010–2015.

Appendix SA6: Most Frequent Journals with Appearance of Medical Subject Heading (MeSH) Term as Major Topic: “Health Services Research,” 2010–2015.

Appendix SA7: Comparison of Years of Experience from LinkedIn.com and ResearchGate.net, 2016.

Appendix SA8: Completed Degrees by Selected Fields, 2010 to 2015.

Appendix SA9: Institutions with Highest Number of Completed Doctoral Degrees in Research/Scholarship among Core Fields, 2011–2015.