Analysis of the workforce and workplace for rheumatology and the research activities of rheumatologists early in their careers


  • The contents of this article are solely attributable to the authors, and do not reflect the views of the American College of Rheumatology or any of its officers or staff. The American College of Rheumatology exercised no role in the design or analysis of this study or in the decision to publish the findings.



To assess the workforce and workplace in rheumatology, and the research work of early-career rheumatologists.


Early-career rheumatologists were defined as practicing physicians who joined the American College of Rheumatology (ACR) in 1991–2005, were 49 years of age or younger when they joined, and reside in North America. This cohort participated in a Web-based survey distributed by the ACR. A total of 247 surveys (21.2% response) were used for this analysis. Survey questions were designed to obtain core insights about the workforce, workplace, research activities, funding, and the demographic profile of respondents.


Respondents from all workplaces—clinical, academic, federal, and industry—engaged in clinical care, teaching, administration, and research. The time devoted to these tasks was employer dependent, and workplaces shaped the scale and scope of research. Patient-oriented research was predominant across all workplaces. Disease, population, and translational research were intermediate, and few respondents pursued basic or prevention-oriented research in any type of workplace. Rheumatologists obtained extramural funds (21.3%) and intramural funds (78.7%) to pay portions of their salaries for time spent on research. Receiving a National Institutes of Health K08/K23 award was associated with receipt of a federal research project grant (P < 0.001). Respondents associated investigative work with reduced earnings, a perception validated by an estimated drop in pre-tax annual earnings of 2.3% for each half-day/week dedicated to research (P < 0.01).


The results of this study justify interventions for closing gaps embedded in investigational rheumatology. These include improved funding for clinical research, increasing the number of K08/K23 awards, and recruiting rheumatologists from underrepresented demographic groups.

Clinical research has a pivotal position in the overall structure of medicine, for several reasons. First, and foremost, clinical investigation is the proving ground for all innovations and discoveries that advance the practice of medicine. Second, clinical research serves as the training ground for producing the next generation of investigators needed to sustain medical progress. Third, a reward structure has evolved among clinical investigators, leading to peer recognition and the opportunity to obtain funds required for the production of a public good. In the case of rheumatologists, clinical investigation serves as the mechanism to advance clinical practice through the development of improved diagnostic methods, new treatments, and preventative measures that enhance the lives of the millions of patients with disorders characterized by persistent pain and accumulated disability (1).

For any medical or surgical specialty, the scope and scale of clinical research are unknown beyond snapshots of the broad aims and expenditures of research programs sponsored by federal agencies or the pharmaceutical industry (2–4). As a consequence, the workforce and workplace for clinical investigation are enigmatic and unexamined, even after explicit warnings that this essential arm for advancing clinical practice is in jeopardy (5–8). The present study was designed to assess the workforce and workplace for rheumatology and the extent and prevailing type of research among rheumatologists early in their careers. Our findings provide fresh insights about the workforce and the workplace for rheumatology and justify interventions to address gaps in both the scope and the scale of clinical research in the field.


Survey participants.

The American College of Rheumatology (ACR) is the primary professional organization dedicated to advancing the practice of rheumatology in North America. For the present study the ACR agreed to distribute a Web-based questionnaire, produced by the authors, to members' e-mail addresses. A letter of invitation from the ACR preceded respondent access to the survey instrument.

Rheumatologists early in their careers were identified based on 5 criteria: joined the ACR between January 1, 1991 and December 31, 2005, earned an MD or equivalent degree, held an active license to practice medicine, reside in Canada or the US, and were 49 years of age or younger at the time of the survey. These criteria were chosen since internists/pediatricians in the US complete fellowship training at ∼34 years of age (9), gain support for their first research project grant at ∼44 years of age if they compete for federal grants/contracts (10), and function as a principal investigator (PI) or co-PI for ∼5 years.

Prospective respondents received an original e-mail, and 2 followup e-mail reminders if they failed to return a survey questionnaire between December 2007 and February 2008. A total of 265 rheumatologists responded. Incomplete surveys were discarded, yielding 247 questionnaires available for analysis (final response rate 21.2%). The survey protocol adopted for this study was consistent with that used by the ACR to survey members about services or issues affecting the practice of rheumatology. The mean response rate for 3 ACR surveys performed in 2007–2008 was 25.7% (Batts L, American College of Rheumatology: personal communication), a value approximating the response rate achieved here.

Participant instructions noted that the Board of Directors of the ACR had approved the survey, and that institutional review board approval for an exempt protocol was obtained from the Office for the Protection of Research Subjects, University of Illinois at Chicago. Respondents were advised that participation was voluntary, confidentiality would be maintained, and none of the research conducted or published would divulge the responses of individual physicians. Questionnaires were submitted anonymously over the Internet. Survey security was maintained by using secure servers to direct participant responses over the Internet and store data.

Design of survey instrument.

A preliminary version of the survey questionnaire was tested for ambiguity and errors. A random sample of 50 rheumatologists, satisfying all “early-career” criteria, was asked to respond to a prototype questionnaire. Pilot survey participants were instructed to provide written comments on any question that was ambiguous, awkward, or irrelevant. A total of 18 participants returned questionnaires and provided comments used to reformat the survey. The responses of pilot survey participants were excluded from this report.

In the final version of the survey questionnaire, participants were asked to identify their employment sector, designate the number of half-days/week spent on administration, clinical care, teaching, and research, specify the source(s) of salary support for time spent on research, and indicate their annual pre-tax compensation within ordered ranges. Respondents were asked to specify 1 or more types of research that typified their investigative activities over the last 12 months. The definition of clinical research used was developed to be inclusive, conform to the recommendations of a consensus development conference convened to codify clinical investigation, and respect definitions adopted by others (6, 11, 12).

A series of questions was used to examine incentives and disincentives for pursuing research careers. These questions were answered via a 4-point Likert scale (13) that was collapsed to consider 2 outcomes: agree or disagree. A second set of questions addressed institutional support for research during residency and fellowship training, extramural funding for post-fellowship training in research, and the receipt of extra- and/or intramural funds for research projects as a PI or co-PI. The last question set addressed respondent age, citizenship (Canada or US), sex, and ethnic background. The ethnic background of individuals graduating from allopathic schools of medicine in the US was estimated by calculating the mean distribution of ethnic groups graduating in 1991–2005 (9).

Statistical analysis.

A unique, computer-generated case number was assigned to questionnaires to protect respondent confidentiality throughout the study. Data were analyzed with Stata software, version 6.0. Pearson's chi-square goodness-of-fit-test was used to determine whether the cohort of survey respondents differed from the population of prospective participants. Results involving multiple comparisons among groups relied on testing whether distributions differed across groups or, in the case of means, whether the means differed between or among groups. Categorical variables were compared using Pearson's chi-square goodness-of-fit-test, and continuous variables were compared by analysis of variance or paired t-test.

A multiple regression model was used to test the association between the receipt of grants by PIs and co-PIs and other explanatory variables of research performance. Probability estimates of regression coefficients assumed that tested variables were distributed randomly. Assumptions underlying the unmodified ordinary least squares model included linearity, full rank, exogeneity of independent variables, homoscedasticity of error terms, and exogenously generated data. Two-sided P values are reported.

A Tobit regression model was used to determine the association between pre-tax annual earnings and various outcome measures to accommodate the censored earnings that respondents reported within specified ranges. Dependent variables with binary responses were tested using the logit regression model with the assumption that the natural log of the probability ratio is approximated by a linear function. Certain respondents failed to report the number of half-days spent on research or to provide information related to the submission of proposals for research project grants. The absence of a response, in a few cases, was assumed to represent zero or to be identical to the “no-response” entered by most respondents. This transformation was based on the assumption that rheumatologists who are not involved in research would likely overlook the need to document a zero response.


Respondent sample.

The cohort of prospective participants was compared with the group who actually responded, to ascertain whether the 2 populations differed with respect to sex or age distribution, country of residence (US or Canada), and state of residence within the US. These parameters did not differ significantly between the respondents and the overall cohort who had received the questionnaire (P > 0.20), which indicated that the 247 survey participants were representative of ACR members early in their careers.

Respondent profile.

A respective 10.5% and 19.9% of the respondents had earned their baccalaureate and MD (or equivalent) degrees outside of North America. Residency and fellowship training were accomplished in North America by 98.6% of respondents.

Respondents completed medical school at 27.1 ± 0.2 years of age (mean ± SEM) and finished fellowship training at 33.7 ± 0.2 years of age, an elapsed interval of 6.6 ± 0.2 years. Women and men completed pre- and postclinical training at similar ages (P > 0.25). More than 99% of the respondents reported that they were board-certified or board-eligible in either adult or pediatric rheumatology, an indication that respondent training was consistent across these 2 related subspecialties.

Most respondents resided in the US (95%) and the remainder in Canada (5%). The proportions of citizens, permanent residents, and noncitizens among the US respondents were 85.8%, 8.5%, and 5.7%, respectively. The proportions of citizens, permanent residents, and noncitizens among the Canadian respondents were similar to those among rheumatologists from the US (P > 0.50). The sex distribution of early-career participants from the US was 44.8% women and 55.2% men. Respondent data from both countries were merged into a single data set since no significant differences were evident (P > 0.50) in any demographic measure (age, sex, ethnic background).

The self-identified ethnic backgrounds of rheumatologists from the US were compared with those of graduates of allopathic schools of medicine (Table 1). The percentages of respondents from Asian and Hispanic/Latino backgrounds approached the percentages among US graduates overall (Table 1). The number of black or African American rheumatologists, in contrast, was underrepresented in this survey sample by 6.1% while the number of Caucasians was overrepresented by a similar amount when compared with the cohort of individuals earning an MD degree in 1991–2005 (Table 1).

Table 1. Self-identified ethnic backgrounds of the early-career rheumatologists in the present study and of graduates of allopathic schools of medicine in the US
Ethnic backgroundSelf-identified background, %Graduates of US medical schools, %*
Women (n = 107)Men (n = 132)Women and men combined (n = 239)
  • *

    The ethnic backgrounds of US allopathic medical school graduates were determined by estimating the mean number of individuals within each specified ethnic group who graduated each year, from 1991 to 2005 (9). An average of 15,713 physicians graduated per year between 1991 and 2005 in the US, the same 15-year sample window adopted for the present survey of early-career rheumatologists.

  • The number of American Indians and Alaskan Natives was not determined in this study, but is listed here to coincide with the demographic profiles of graduates from allopathic schools of medicine in the US (9).

  • The proportion of individuals who self-identified as “other” may include rheumatologists of ethnicities (e.g., American Indian, Alaskan Native, Hawaiian Native, Pacific Islander) that were not listed in the survey instrument for this study, and/or individuals whose ethnicity is unknown.

American Indian or Alaskan Native0.7
Asian American16.814.415.516.9
Black or African American0.
Hispanic or Latino American5.
Other or unknown1.

Workplace assessment.

The time devoted to clinical care, teaching, research, and administration was estimated by asking respondents to specify the number of half-days/week committed to each of these activities (Table 2). Clinical care took place at all types of workplaces. Respondents employed in a solo or group practice or health system (workplaces referred to below as “clinical care providers” or simply “providers”) spent a mean of 7.9 half-days/week on clinical service, a time commitment exceeding that in other workplaces (P < 0.05). Respondents employed by academic medical centers/teaching hospitals, federal government, and the pharmaceutical/biotechnology industry worked a mean of 3.5, 2.3, and 1.5 half-days/week, respectively, on clinical service (Table 2).

Table 2. Time (half-days/week) spent by early-career rheumatologists in clinical service, teaching, research, and administration, by workplace type*
WorkplaceClinical serviceTeachingResearchAdministrationTotal effort
  • *

    Values are the mean ± SEM half-days/week (mean percent effort). Results are based on 226 of 247 eligible respondents. The number of respondents employed in clinical care, academic medical centers (AMCs), federal government, and industry was 113, 96, 10, and 7, respectively. Certain respondents were not affiliated with the specified workplaces and were excluded from this analysis. The total number of half-days/week deviates from the expected value of 10.0 to reflect the actual time reported by respondents for each workplace.

  • Total time devoted to teaching medical students, residents, and fellows.

  • Total time devoted to administrative work and committee assignments.

  • §

    Solo or group practice, health system, or hospital.

  • Academic medical center or teaching hospital accredited to sponsor a residency program in internal medicine or pediatrics, or a fellowship program in adult or pediatric rheumatology or both, as approved by the Accreditation Council for Graduate Medical Education.

  • #

    HHS = Department of Health and Human Services; DOD = Department of Defense; VA = Department of Veterans Affairs.

Clinical care provider§7.9 ± 0.2 (74.7)0.4 ± 0.1 (3.6)1.4 ± 0.1 (15.3)0.8 ± 0.1 (6.4)10.5 ± 0.1
AMC or teaching hospital3.5 ± 0.2 (34.9)1.3 ± 0.1 (11.9)4.1 ± 0.3 (38.7)1.5 ± 0.1 (14.5)10.4 ± 0.2
Federal (HHS/DOD/VA)#2.3 ± 0.4 (21.6)0.6 ± 0.2 (6.8)4.8 ± 0.9 (52.8)1.5 ± 0.3 (18.8)9.2 ± 0.3
Pharmaceutical or biotechnology industry1.5 ± 1.3 (13.6)0.5 ± 0.4 (4.5)5.1 ± 0.9 (44.1)3.8 ± 1.5 (37.8)10.9 ± 0.7

Rheumatologists from all types of workplaces participated in teaching medical students, residents, or fellows (Table 2). Respondents at academic medical centers devoted an average of 1.3 half-days/week to teaching, whereas those who worked primarily for clinical care providers, the federal government, and industry spent 0.4, 0.6, and 0.5 half days/week, respectively, in teaching (P < 0.01 versus those at academic medical centers) (Table 2). Respondents from clinical care providers spent a mean of 1.4 half-days/week conducting research. Those from academic medical centers reported dedicating 4.0 half days/week to research, while those in the federal and pharmaceutical/biotechnology sectors spent a respective 4.8 and 5.1 half days/week on research (Table 2). Administrative work, including committee responsibilities, involved 0.8 half days/week among those who were employed by providers, 1.5 half days/week in academic and federal workplaces, and 3.8 half days/week in industry (Table 2).

Scope and scale of research.

The extent and type of research pursued by respondents was workplace dependent (Table 3). No between-sex disparities were evident, at any workplace type, in the percentages of respondents involved versus not involved in research or in the distribution of respondents pursuing basic versus clinical research (P > 0.25).

Table 3. The effect of workplace on the type of research work performed by early-careeer rheumatologists*
WorkplaceType of research work
NoneBasicTranslationalDisease- orientedPatient- orientedPopulation- orientedPrevention- oriented
  • *

    Values are the mean percent of investigative effort. Results are based on 244 of 247 eligible respondents. Respondents were allowed to specify 1 or more types of research activity to reflect the type of investigative work accomplished in the 12 months preceding this survey. Investigative activities were defined using terminology adopted by others (6, 11, 12) and appeared as follows in the survey instrument: Basic research—laboratory-based research involving the development of new drugs, technologies, or devices; Translational research—bench-to-bedside or bidirectional research involving human subjects known to the investigator, excluding the use of human specimens (cells/tissues) for laboratory studies; Disease-oriented research—requires use of human subjects to investigate the mechanisms or natural history of disease, or improve the detection or diagnosis of disease; Patient-oriented research—clinical trials, including phase I, II, III, and IV trials of drugs, biologics, devices, and the evaluation of therapeutic interventions; Population-oriented research—outcomes studies of populations, health services and cost-effectiveness research, studies of health quality including best practices and medical errors, epidemiology and genetic studies, and community-based clinical trials; Prevention-oriented research—research on primary and secondary prevention of disease in patients, and health promotion via behavioral modification.

  • See Table 2 for description of each workplace type and for definitions.

Clinical care provider30.
AMC or teaching hospital5.514.412.220.422.720.44.4
Federal (HHS/DOD/VA)
Pharmaceutical or biotechnology industry6.312.518.818.837.56.30

In the clinical care workplace, 69.1% of providers pursued research, although this proportion was below that at workplaces providing dedicated research time (Table 3). Patient-oriented, disease-oriented, and population-oriented research was pursued by 42.4%, 10.8%, and 7.9% of providers, respectively. Employees of clinical care providers pursued some basic, translational, and prevention-oriented research, but the proportion of respondents engaged in these endeavors was limited to 2.2–3.6% (Table 3).

Findings regarding extent and type of research performed were similar for employees of the federal government and of the pharmaceutical/biotechnology industry (P > 0.15). Respondents at academic medical centers devoted analogous amounts of time to patient-, disease-, and population-oriented research (P > 0.25) as indicated by respective commitments of 22.7%, 20.4%, and 20.4%. Basic and translational research involved respective efforts of 14.4% and 12.2%, while prevention research was limited to 4.4% (Table 3).

Research funding.

Analysis of the extra- and intramural funds used to defray a portion of respondents' salaries for time spent on research indicated no disparities between women and men in extra- and intramural support available (P > 0.25). Grants/contracts from federal and nonfederal sources supported the salaries of 21.3% of respondents. For the remaining 78.7% of respondents, research time was paid for by intramural funding. Intramural funds were derived from clinical earnings (39.7%), medical schools and/or hospitals (14.5%), endowments/other sources (8.3%), and the salaried commitments of federal employers (8.3%) or pharmaceutical employers (7.9%).

The proposals submitted and grants/contracts received by early-career rheumatologists from the US indicated that 21.3% of respondents applied for post-fellowship research training from federal and nonfederal sources, and 14.2% submitted applications for National Institutes of Health (NIH) K08/23 awards. An average of 1.8 proposals was required per K08/23 award (Table 4), yielding a success rate of 41.2% per respondent or 22.2% per application. Fewer training proposals were submitted to the Department of Veterans Affairs (VA) than to the NIH, but the success rates did not differ significantly (P > 0.20) (Table 4). The receipt of an NIH K08/23 award was a consequential marker of research project grants awarded to PIs and co-PIs (P < 0.001). Specifically, rheumatologists receiving a K award were estimated to receive an average of 2.8 and 1.2 federal research project grants as PIs or co-PIs, respectively. The association between the receipt of a K award and the subsequent receipt of a research project grant was independent of sex (P > 0.25); female and male recipients of K awards were just as likely to receive NIH R01 awards.

Table 4. Analysis of proposals submitted and awards received by early-career rheumatologists for post-fellowship research training and research project grants*
Type of proposal or grant awardSources of funding available to rheumatologists
  • *

    Data are from a total of 239 early-career rheumatologists in the US. Except where indicated otherwise, values are the mean ± SEM. HHS = Department of Health and Human Services; NIH = National Institutes of Health; VA = Department of Veterans Affairs; DOD = Department of Defense; NA = data not available.

  • Estimates of post-fellowship training proposals are based on a cohort of 51 applicants who submitted 149 proposals. Proposals for NIH K08/23 awards were submitted by 34 rheumatologists. Estimates for research project grants submitted by principal investigators (PIs) are based on a cohort of 38 rheumatologists who submitted 233 proposals. Estimates for research project grants submitted by co-PIs are based on a cohort of 17 rheumatologists who submitted 85 proposals.

  • Agency for Healthcare Research and Quality, Centers for Disease Control and Prevention, Food and Drug Administration, Health Resources and Services Administration, and Substance Abuse and Mental Health Services Administration.

Training (post-fellowship)      
 Proposals/applicant1.8 ± 0.301.4 ± 0.201.8 ± 0.22.6 ± 0.81.6 ± 0.2
 Success/applicant, %41.220.0NANANA
 Total no. submitted6371352122
Research as PI      
 Proposals/applicant2.8 ± 0.40.9 ± 0.51.1 ± 0.61.5 ± 0.52.7 ± 0.32.9 ± 0.51.8 ± 0.3
 Grants/applicant2.0 ± 0.31.3 ± 0.31.0 ± 0.102.4 ± 0.32.3 ± 0.42.0 ± 0.7
 Success/applicant, %51.180.010042.243.850.0
 Total no. submitted82993644620
Research as co-PI      
 Proposals/applicant2.2 ± 0.41.1 ± 0.9002.1 ± 0.62.3 ± 0.74.3 ± 2.3
 Grants/applicant1.9 ± 0.40.6 ± 0.5001.4 ± 0.72.0 ± 0.71.7 ± 0.9
 Success/applicant, %45.875.050.041.640.0
 Total no. submitted2914151413

Respondents from the US produced 318 proposals as PIs or co-PIs (Table 4). Prospective PIs submitted an average of 2.8 research project proposals to the NIH and received an average of 2.0 awards, yielding a success rate of 51.1% per respondent (Table 4) or 31.1% per proposal. Fewer proposals were submitted to non-NIH agencies of the Department of Health and Human Services and to the VA, but success for applications submitted by respondents as PIs or co-PIs exceeded that for NIH awards (P < 0.05) (Table 4). The success of investigator-initiated proposals considered by philanthropic, industry, and other funding sources was 42%, 44%, and 50%, respectively (Table 4). Proposals submitted by co-PIs were funded on the same order (P > 0.15) as those of PIs (Table 4).

Individual and institutional commitments to research.

Respondents were queried about their interest in pursuing investigative careers as medical students, residents, and fellows. The proportion of positive responses increased from 28.7% at the time they were medical students, 38.6% when they were residents, and 60.3% when they were fellows. The same cohort was asked to indicate whether, as medical students, residents, and fellows, they were aware of opportunities to participate in faculty-mentored research projects. Only 22.7% indicated that they were aware of such opportunities as medical students and 25.5% were aware as residents, whereas 67.6% acknowledged that as fellows they were aware of the possibility of participating in faculty-sponsored research.

Potential deterrents to pursuing a research career were examined to identify interventions that could enhance the workforce for investigative rheumatology. This analysis indicated that the time and energy required for a meaningful research career was not considered to be more demanding than that required for a similar career involving clinical service (P > 0.25). Most rheumatologists (72.4%) viewed job uncertainty as a significant concern regarding a career involving research relative to one focused primarily on clinical service (P < 0.01). Additionally, the majority of respondents (75.6%) perceived that the earnings of investigators were less than those of providers (P < 0.01). This perception was tested by using a Tobit regression model specifying the natural logarithm of pre-tax annual earnings as a function of 2 key explanatory variables—years post-fellowship and half-days/week devoted to research—as well as variables to control for sex, employment sector, and total work hours per week. The results indicated that pre-tax annual earnings increased linearly at 1.8% per year of post-fellowship experience. However, the earnings of rheumatologists devoting ≥2 half-days/week to investigative work fell by 2.3% per year for each half-day/week spent on research (P < 0.01) relative to individuals spending ≤1 half-day/week on research work (Figure 1).

Figure 1.

Top, Relationship between the natural log of pre-tax annual earnings and years post-fellowship for early-career rheumatologists. The equation/linear function demonstrates that mean annual pre-tax earnings increased by 1.8 % per year for each year of post-fellowship experience. Middle, Relationship between the natural log of pre-tax annual earnings and half-days/week devoted to research work among early-career rheumatologists. The equation/linear function indicates that mean annual pre-tax earnings decreased by 2.3% for each half-day/week spent on research work among early-career rheumatologists who spend ≥2 half-days/week on research. The equations shown in the top and middle panels are based on a Tobit regression model that controlled for sex, and total work time/week, and for bias occasioned by the nonreporting of some variables as noted in Methods. The model used for the equations shown in the top panel also controlled for half-days/week spent on research. Bottom, Institutional support for research work of early-career academic rheumatologists. Values are the mean and SEM constant dollars invested in junior faculty as a function of the number of half-days/week respondents spent on research. Institutional investments were based on the mean value of start-up packages (total institutional dollars [corrected to constant dollars] for personnel, equipment, supplies, and other research expenses) provided to early-career rheumatologists employed at academic medical centers.

Institutional support for early-career academic rheumatologists was estimated by using the total dollar value of start-up packages provided for personnel, equipment, supplies, and other research expenses. Investment in the research programs of early-career academic rheumatologists increased linearly with the number of half-days/week devoted to research (Figure 1).


The present study provides important new insights about the workforce and workplace in rheumatology and extends knowledge about the amount and type of research accomplished by rheumatologists early in their careers. The findings rely on self-reported responses to a Web-based questionnaire to assess the administrative, clinical, didactic, and investigative work of rheumatologists affiliated with academic, clinical, federal, and industry employers. The results fill a gap created by the lack of a national protocol for appraising clinical investigation and its practitioners. Our report establishes a template for estimating the scope and scale of clinical research within medical subspecialties and assessing explanatory variables that could be part of a framework for interventions aimed at enhancing opportunities for innovation and discovery in rheumatology.

The analysis of administrative, clinical, teaching, and research work established that the time devoted to these tasks is workplace dependent (Table 2). A novel finding of this assessment derived from the analysis of workplaces according to the extent and type of research pursued by early-career rheumatologists. Research, of all types—basic, translational, disease, patient, population, and prevention—was accomplished at academic, federal, industrial, and provider workplaces. However, the scope and scale of research are a function of the workplace (Table 3). The provider workplace, for example, emphasized patient-oriented studies, a focus exceeding all other research endeavors by 4–5-fold. Providers may direct or participate in various clinical protocols, including registries, cohort studies, and multicenter clinical trials of new drugs. While the survey did not distinguish among these possibilities, it seems likely that most provider research involved patient enrollment in protocols developed and funded by industry or federal sponsors since research time among rheumatologists in the provider setting was restricted to 15.3% of total professional effort (Table 2). Certain rheumatologists in private practice, however, also participate in registries and other prospective cohort studies.

Rheumatologists employed in federal and pharmaceutical workplaces devoted more time to research than any other activity, and the distribution of investigative approaches was similar in both types of workplace (Table 3). The minor distinction between the 2 workplaces pivoted on translational research in the pharmaceutical industry, an investigative emphasis exceeding that of all other workplaces. Research in the academic workplace was dispersed among multiple investigative activities. The diversity of investigative pursuits within the academic workplace contrasts with the focal specificity of research in other workplaces.

Prevention-oriented research emerged as an orphan enterprise across all workplaces considered here. Stakeholders should consider this void as a singular opportunity to plan interventions aimed at promoting prevention research among early-career rheumatologists. Such research would provide a platform to establish individuals' risk for developing disease, provide protocols to prevent disease onset, and make it possible to intervene at the earliest possible time when disease occurs (14–16). Adoption of such a paradigm would begin the transformation of rheumatology from having a curative to a preventative focus (14).

Analysis of the proposals submitted by respondents for research grants/contracts provided new information about rheumatologists receiving K08/23 awards. We observed a compelling association (P < 0.001) between receipt of K08/23 awards and subsequent award of an estimated average of 2.8 NIH research project grants. While the survey questionnaire did not control for all possible variables needed to predict the success of K awardees as future recipients of federal research grants, the unequivocal success of K awardees observed here provides logic for proposing that the number of K08/23 awards be increased from an estimated 5% of the 240 individuals who complete fellowship training in adult or pediatric rheumatology each year (17) to ∼8%, or 19–20 per year.

The present assessment of the academic workforce indicates that investigative rheumatology is hobbled at multiple levels. First, research work is highly leveraged on intramural funds as opposed to extramural grants/contracts. This assertion stems from the finding that almost 80% of academic rheumatologists relied on intramural funds to cover a portion of their salary for time spent on research. Clinical earnings were used to generate ∼40% of the intramural funds used by early-career academics—almost twice the support derived from extramural grants/contracts. Next, the time available for research was limited by the need to accomplish service-related tasks (patient care, teaching, administration) among early-career academics. Finally, only an estimated 5% of academic rheumatologists received an NIH K08/23 award for post-fellowship research training (Table 4).

Beyond this set of workforce constraints resides a host of limitations related to an ailing and outmoded national infrastructure for sustaining clinical research (18). For instance, there is a disparity approaching 2:1 for funding of basic versus clinical research by federal agencies (19, 20), and a recent report documents the difficulty of obtaining funding for clinical research and population-based studies from individual institutes within NIH (21). The summed constraints in both the workforce and federal–institutional partnership prompt the suggestion that stakeholders in rheumatology—academic, federal, industry, philanthropic, professional—pursue interventions for the gradual systematic growth of clinical research in rheumatology. The intent of this proposal is to enlarge the proportion of rheumatologists with research training (K awardees) so they can subsequently gain extramural support for research project grants and have the time and funding to advance the clinical practice in rheumatology.

Consideration of the deterrents for pursuing investigative careers indicated that 75.6% of respondents perceived that providers held an earnings advantage over researchers. This perception was verified by the finding that pre-tax annual earnings fell by 2.3% for each half-day/week devoted to investigative work among rheumatologists spending 2 or more half-days/week on research (Figure 1). The observed disparity in annual earnings does not reflect the presumed remuneration to mid-career and established researchers; this was not addressed since the present study focused on early-career investigators. Clinical investigation, moreover, remains the source of nonmonetary rewards, including peer recognition for innovation and discovery, the opportunity to obtain extramural funds for the production of a public good, and the self-satisfaction of championing advances in clinical practice. The significance of a “researcher cohort” to rheumatology far outweighs their number in the clinical workforce since a single innovation offers the potential to benefit thousands of patients by transforming the clinical management of a particular condition (22).

The assessment of self-identified ethnic backgrounds identified a gap in the proportion of African American respondents within the rheumatology workforce (Table 1). The underrepresentation of rheumatologists from any demographic background is of intrinsic importance to investigative rheumatology because the ethnic profile of volunteers agreeing to participate as subjects in a clinical trial reflects the ethnic background of physicians directing the study (23). Reliable estimates of the safety/efficacy of new drugs or treatments are unachievable unless the participants in a clinical study approximate the demographic profile of the US population (24, 25). Disparities in the ethnic backgrounds of practicing physicians indicate the need to pursue interventions designed to recruit future generations of rheumatologists that mirror the demographics of graduates from allopathic schools of medicine in the US (13, 26–28).

The present study was based on a limited number of survey participants. The proportion of responders to this survey, however, was consistent with that of other surveys conducted by the ACR. Further, statistical comparisons of responders and nonresponders established that both cohorts were similar (P > 0.20) in terms of sex and age distribution, state of residency within the US, and country. Finally, the demographic profile of survey respondents conforms to data reported from the annual survey of graduate medical education for rheumatologists (17) and data on the workforce for rheumatology (29). Despite this evidence, the results should be viewed as an index rather than an absolute measure of the workforce, workplace, and research accomplished by early-career rheumatologists.

In conclusion, this study calls attention to multiple issues that limit advances in investigative rheumatology. The findings provide stakeholders—academic, federal, industry, philanthropic, professional—with an evidence-based rationale to pursue multiple interventions for remodeling clinical research in rheumatology. The present results are not intended to detract from recent initiatives, championed by the NIH, to improve the climate for clinical scholarship (30–32). Instead, the findings complement and extend the efforts made by NIH and provide a rational framework for initiating a national dialogue aimed at establishing a public and private infrastructure to prudently and amply support investigative rheumatology (33). Only then will a bright future be assured for innovation and discovery in rheumatology, ultimately benefiting patients with musculoskeletal diseases.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Desjardins had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Desjardins, St.Clair.

Acquisition of data. Desjardins.

Analysis and interpretation of data. Desjardins, Ehrenberg.


The authors thank each of the rheumatologists who took the time to respond and return a survey questionnaire. La Tanya Batts and Steve Blevins at the American College of Rheumatology assisted with all phases of this project including the selection of rheumatologists for the preliminary and full surveys and the distribution of survey materials. Thanks are also due to the staff of the Survey Research Institute at Cornell University for programming and administering the survey, and to Kenneth T. Whelan, Cornell Higher Education Research Institute, who managed the statistical analysis of the entire survey project. Dr. Desjardins acknowledges the encouragement provided by Dr. Gerald S. Moss, Dean, College of Medicine, University of Illinois at Chicago for his early endorsement of the Clinical Scholars Project.