Community health centers (CHCs) have great potential to participate in the development of evidence-based primary care but face obstacles to engagement in clinical translational research.
Community health centers (CHCs) have great potential to participate in the development of evidence-based primary care but face obstacles to engagement in clinical translational research.
To understand factors associated with CHC interest in building research infrastructure, Harvard Catalyst and the Massachusetts League of Community Health Centers conducted an online survey of medical directors in all 50 Massachusetts CHC networks.
Thirty-two (64%) medical directors completed the survey representing 126 clinical sites. Over 80% reported that their primary care providers (PCPs) were slightly to very interested in future clinical research and that they were interested in building research infrastructure at their CHC. Frequently cited barriers to participation in research included financial issues, lack of research skills, and lack of research infrastructure. In bivariate analyses, PCP interest in future clinical research and a belief that involvement in research contributed to PCP retention were significantly associated with interest in building research infrastructure.
CHCs critical role in caring for vulnerable populations ideally positions them to raise relevant research questions and translate evidence into practice. Our findings suggest a high interest in engagement in research among CHC leadership. CTSAs have a unique opportunity to support local CHCs in this endeavor.
With the passage of Affordable Care Act (ACA) and the expansion of health insurance coverage, more people will have access to primary care. Community health centers (CHCs) will play a significant role in meeting this demand particularly for vulnerable disadvantaged populations. A strong primary care system that provides evidence-based, patient-centered care is crucial. As critical access providers, CHCs have great potential to help build the foundation for evidence-based primary care and participate in examining outcomes of emerging models. Today, there is already a long tradition of practice-based research[3, 4] and multiple practice-based research networks (PBRNs) across the nation have provided important findings for primary care.[2, 5] However, while some community health centers have been part of PBRNs, others are largely disconnected from the research enterprise. Recognizing this, the National Association of Community Health Centers (NACHC) called for the development of a research agenda for community health centers.
Given the populations they serve and their close connections to their geographic communities, CHCs have an important role to play in the evolving primary care research agenda. Their contribution to the clinical research enterprise could include developing relevant research questions, engaging their patients in appropriate studies, expanding partnerships with academics to build the research base, and utilizing the resulting evidence to improve patient care. The engagement of CHCs in clinical research may ultimately improve evidence translation and uptake, help address health care disparities, enhance clinical systems improvement and increase provider career satisfaction and retention.
To date, however, CHCs engagement in research has been hampered by numerous organizational culture and infrastructure obstacles, including conflicting priorities, provider's lack of research skills and limited access to data, research partners and related resources.[9-14] In order to address this gap, the Harvard Clinical and Translational Science Center (Harvard Catalyst) Safety Net Infrastructure Initiative embarked on an initiative to build infrastructure for primary care research in Massachusetts community health centers. As a first step, Catalyst Safety Net staff developed a partnership with the Massachusetts League of Community Health Centers. Staff then collaboratively developed and implemented a survey of all Massachusetts community health centers to: (1) understand the current level of engagement and interest in quality improvement and clinical research among Massachusetts CHCs and (2) determine predictors of interest in building infrastructure for clinical research. The intent was to subsequently use the results to inform programmatic efforts to build research capacity among CHCs.
Massachusetts has 50 community health center organizations with over 180 practice sites across the state. These include federally qualified health centers as well as look-alikes and hospital owned CHCs. Most are largely in fee-for-service arrangements with insurers. There is a wide range of health center size with some serving less than 10,000 patients annually (n = 6) and others serving over 50,000 patients annually. These may be the only or largest providers of care to their community's publically insured residents. In 2011, members of the Catalyst Safety Net Initiative approached the Massachusetts League of Community Health Centers (Mass League) to assist in the development and distribution of a survey designed to ascertain current research infrastructure and level of interest. The Massachusetts League of Community Health Centers was established in 1972, as a nonprofit, statewide association representing and serving the needs of the state's community health center organizations.
Given that the Safetynet Initiative is part of the Community Engagement core and the Community Innovation and Research Program (CHIRP) of Harvard Catalyst, the project attempted to engage community health center providers in many aspects of the research process. In particular, the development of the survey and its distribution was an inclusive endeavor. Community providers were instrumental in the development of the survey questions, the pilot phase and distribution. This approach enhanced the relevancy of the questions and set the groundwork for future research endeavors by building trust between academics and community health center leadership and providers.
The survey was initially developed in collaboration with three primary care providers practicing in CHCs from the Cambridge Health Alliance (CHA), a safety net public hospital system that includes a network of community health centers. These individuals were critical in helping to identify the main areas of focus for the survey. The selection of survey items and constructs was also informed by Cooke and Shaw's work with practice based research structures (PBRNs) in the United Kingdoms’ National Health Service.[15-17] This work identified key elements for successful primary care research infrastructure including; strong and equitable partnerships with scientists, top level support to ensure that research was valued, effective management of partnerships, and use of existing infrastructure and expertise in the primary care organization[15-17] In addition, AHRQ identified a variety of elements necessary to support PBRN efforts: coordination, fiscal support, and leadership. Based on this work and our discourse with primary care providers at CHA we hypothesized that four factors would contribute to a CHC's interest in building research infrastructure; existing infrastructure, provider experience with research, provider interest in Quality Improvement (QI) and research, and organizational priorities (Figure 1).
With these factors in mind, we developed a survey consisting of 50 open and closed-ended questions. The survey took about 20 minutes to complete. In order to examine key infrastructure constructs, questions were asked about site level demographic variables, prior experience with clinical research, skills of current staff in research, access to tools such as the electronic medical record and its functionality and analytic support, and relationship of research to provider retention. Additionally, questions about the top barriers and facilitators to clinical research, and finally, a question about interest in building clinical research infrastructure were asked. Prior to fielding the CHC survey, the Mass League piloted the survey with several medical directors and feedback was incorporated.
During the development of the survey there was ample discussion about the meaning of research and quality improvement. The providers felt strongly that QI and research represented two ends of a spectrum, and data gathered through these strategies could be utilized to answer important questions. Thus both were included in the survey and a clear definition for each was provided. To distinguish clinical research from clinical quality improvement the following definition was utilized: “Most QI initiatives are designed to improve quality within an organization (e.g., reduce patient wait time in your clinic). In contrast, clinical research is often designed to develop knowledge that can be generalized to other populations and settings (e.g., to understand risk factors for patient lack of adherence to hypertension medication).”
The CHC survey was distributed via Survey Monkey in August 2011 to 50 CHC medical directors by the Mass League with a letter encouraging participation. These respondents were chosen given their familiarity with the skill sets of their primary care providers and organizational level research and QI infrastructure. Each medical director was offered a $25 dollar gift certificate upon completion. To ensure a high response rate and completeness of data, extensive individual follow-up communication in the form of phone calls and emails was conducted with those that did not complete the survey and to those who partially completed the survey. In addition, the Mass League was helpful in providing demographic data where it was missing.
Upon completion of the survey, preliminary results were presented to CHC medical directors at a Mass League annual meeting in May 2012 to discuss and interpret the data. Comments from the meeting were documented and contributed to the final analysis plan.
The primary outcome was “interest in developing research infrastructure,” measured by an affirmative response to the question “Are you interested in developing research infrastructure at your CHC?” Independent variables included site level demographic characteristics collected as part of the HHS Uniform Data System (UDS) requirements; size of the health center as measured by mean number of patients served annually, mean number of sites operated by the CHC; percent white/non-white patients; and percent Medicaid; and mean number of primary care providers. In addition, other variables of interest include a series of questions on electronic health record (EHR) functionality for which a composite score was developed based on the percent of 9 possible features of EHR system that a CHC has and uses. The resulting EHR usage index was used as a continuous variable for analysis. Additional questions included “level of successful research collaboration”; “provider interest in future clinical research and quality improvement”; and “involvement of clinical research contribution to provider retention.” These questions utilized 5 point Likert scale responses which were examined and collapsed into two categories (not at all, slightly, moderately, very, extremely) based on the distribution of responses.
First, frequencies for all variables were generated. Subsequently, bivariate analyses were conducted using Fisher's exact test (due to small sample size) to assess the correlation between the dependent variable of interest (interest in building research infrastructure-Y/N) and the independent variables. Independent variables such as number of sites, number of patients, % ethnicity, % race, annual visits by type and percent insurance were all treated as continuous variables while the remaining variables were treated categorically (yes/no or Likert scales). Due to the small sample size, it was not possible to run predictive multivariate models.
Sixty-four percent (n = 32) of medical directors completed the survey. They represented 32 health center networks comprised of 126 clinical sites. Another 8 medical directors commenced the survey but did not proceed beyond the set of demographic questions. These medical directors were not included in the analysis. The majority of respondents (81%) reported being interested in building clinical research infrastructure at their sites. Demographics of the health centers are presented in Table 1.
|CHC* Classification [n = 32]|
|FQHC† Section 330 grantee||22 (69)|
|FQHC “Look Alike”‡||2 (6)|
|Hospital community based center||8 (25)|
|Academic affiliations [n = 29]|
|Mean # of clinic sites [n = 32]||3.94 (7.03)|
|Mean # of patients served in calendar year 2010 [n = 32]||30,768 (48,973)|
|Mean # of primary care providers [n = 29]||26 (25)|
|% Ethnicity mean|
|Hispanic/Latino [n = 32]||33 (26)|
|Non-Hispanic/Latino [n = 30]||62 (25)|
|Unreported/refused to report [n = 26]||7 (9)|
|% Race mean|
|Asian [n = 30]||8 (18)|
|Black/African American [n = 32]||21 (21)|
|American Indian/Alaska native [n = 29]||1 (2)|
|White [n = 31]||1 (27)|
|More than one race [n = 27]||3 (8)|
|Unreported/refused to report [n = 29]||16 (21)|
|Mean # of annual visits by type|
|Primary care visits [n = 30]||68,171(56,655)|
|Dental visits [n = 30]||9,754 (10,542)|
|Mental health [n = 29]||12,043 (27,062)|
|Ob/gyn visits [n = 30]||3,984 (6,170)|
|Vision visits [n = 28]||3,224 (5,659)|
|Mean% patient's insurance|
|Self-pay/uninsured [n = 30]||18 (11)|
|Medicaid (includes MCO) [n = 30]||42 (15)|
|Private/commercial [n = 30]||16 (12)|
|Medicare [n = 30]||16 (13)|
|Other [n = 29]||9 (6)|
|Mean EMR usage index [n = 32]||0.74 (0.29)|
With regard to characteristics of the CHCs, the majority reported that both QI (90%) and clinical research (80%) were beneficial to patients and more than one-half (60%) reported having staff engaged in research with external collaborators while all reported having providers involved in QI efforts. In addition, all medical directors reported that their primary care providers (PCPs) had some interest in future QI while 94% reported some PCP interest in future clinical research (Table 2). The primary barriers to conducting research (Figure 2) included lack of financial support, lack of research skills and knowledge and lack of research infrastructure.
|Provider experiences with research|
|Do any PCPs* at your CHC† have training in clinical research/QI‡? [n = 32]|
|Have PCPs been involved in QI efforts? [n = 32]|
|Have PCPs been involved in clinical research? [n = 32]|
|Do any PCPs at your CHC have external collaborators/research partners? [n = 20]|
|To what extent would increased involvement in QI at your CHC contribute to retention of PCPs? [n = 32]|
|Not at all||2 (6)|
|To what extent would increased involvement in clinical research at your CHC contribute to retention of PCPs? [n = 32]|
|Not at all||4 (13)|
|How interested are PCPs in future QI efforts? [n = 32]|
|Not at all||0 (0)|
|How interested are PCPs in future clinical research? [n = 32]|
|Not at all||2 (6)|
|Interest in building research infrastructure (n = 32)|
|Transformation to patient-centered medical home (n = 32)|
|Systems change (n = 32)|
In bivariate analyses, PCP interest in future clinical research or quality improvement and any belief that research or quality improvement would contribute to retention of PCPs was significantly related to interest in building research capacity. The results of bivariate analyses for hypothesized factors are shown in Table 3. The mean number of primary care providers was significant (p = .0015) and the mean number of patients served and viewing systems change as a priority (all p = 0.06) both approached significance.
|Mean number of patients served in 2010 [n = 32] (SD*)||13,043 (8623)||34,858 (53,538)||0.06|
|Mean EHR Usage Index [n = 32] (SD)||0.63 (0.33)||0.76 (0.28)||0.33|
|Number of sites mean (SD) [n = 32]||1.67 (1.03)||4.46 (7.72)||0.09|
|Mean number of primary care providers (SD) [n = 29]||11.3 (5.0)||29.3 (26.9)||0.0015|
|Academic affiliation [n = 29]||1.00|
|No||0 (0)||4 (17%)|
|Yes||5 (100%)||20 (83%)|
|Provider experiences with research|
|Do any PCPs† at your CHC‡ have training in clinical research? [n = 32]||0.10|
|No||1 (17%)||5 (19%)|
|Yes||3 (50%)||20 (77%)|
|Unsure||2 (33%)||1 (4%)|
|PCPs have past experience with QI§ [n = 32]||–|
|0 (0%)||0 (0%)|
|6 (100%)||26 (100%)|
|PCPs have past experience with clinical research [n = 32]||0.17|
|No||4 (67%)||8 (31%)|
|Yes||2 (33%)||18 (69%)|
|Do any PCPs at your CHC have external collaborators/research partners? [n = 20]||0.37|
|No||1 (50%)||3 (17%)|
|Yes||1 (50%)||15 (83%)|
|Extent to which involvement in QI would contribute to retention of PCPs [n = 32]||0.05|
|Not at all||2 (33%)||0 (0%)|
|Slightly||3 (50%)||8 (31%)|
|Moderately||1 (17%)||11 (42%)|
|Very||0 (0)||5 (19%)|
|Extremely||0 (0)||2 (8%)|
|Extent to which involvement in research would contribute to retention of PCPs [n = 32]||0.04|
|Not at all||3 (50%)||1 (4%)|
|Slightly||2 (33%)||13 (50%)|
|Moderately||1 (17%)||10 (38%)|
|Very||0 (0)||2 (8%)|
|Extremely||0 (0)||0 (0)|
|PCPs have interest in future QI [n = 32]||0.04|
|Not at all||0 (0%)||0 (0%)|
|Slightly||2 (33%)||0 (0%)|
|Moderately||2 (33%)||10 (38%)|
|Very||1 (17%)||13 (50%)|
|Extremely||1 (17%)||3 (12%)|
|PCPs have interest in future clinical research [n = 32]||0.04|
|Not at all||2 (33%)||0 (0)|
|Slightly||2 (33%)||10 (38%)|
|Moderately||1 (17%)||13 (50%)|
|Very||1 (17%)||3 (12%)|
|Extremely||0 (0%)||0 (0%)|
|PCMH care model is a priority [n = 32]||0.64|
|No||2 (33%)||1 (4%)|
|Yes||4 (67%)||25 (96%)|
|Systems Change (team approaches, payment structures, etc.) is a priority [n = 32]||0.06|
|No||3 (50%)||3 (12%)|
|Yes||3 (50%)||23 (88%)|
Over 30 CHC medical directors attended a meeting of medical directors at the Mass League annual conference. Frequency data from the survey was presented. Participants commented that for providers to be involved in research there needed to be a balance between protected time and revenue loss that included an understanding of both long- and short-term investments. While providers involved in research could generate extra revenue from grants, buy-in from health center administration was critical to justify revenue loss. There was agreement that participation in research could be a good strategy for retention with the caveat that this was not appropriate for all providers. Lastly, there was consensus that CHCs needed to identify their own unique contribution to research that would yield information to improve the health of the populations that they served.
In this statewide survey of community health centers, we found that the majority of health center networks were interested in building future research infrastructure. Factors significantly correlated with this interest included number of PCPs, PCP interest in either quality improvement or clinical research and a belief that participation in research would enhance provider retention to even a small extent.
For CHCs, the factors associated with interest in building clinical research revolved around their PCPs including the number of PCPs and PCP interest in both quality improvement and clinical research. Medical directors seemed to perceive this engagement as a strategy to improve career satisfaction and thereby retain PCPs. This is not surprising given that CHCs often face high provider turnover rates and may be particularly vulnerable to provider loss given the increasing demand for primary care providers in Massachusetts post health care reform. Studies have illustrated that physicians become less satisfied over time and that a greater worker load is associated with lower perceived career satisfaction and rising job demands that result in high turnover rates. Factors related to physician retention in CHCs include facility characteristics, compensation, satisfaction with mission, preparation for practice and opportunities for professional experiences such as engagement in research.[18, 22] In a prior survey of Massachusetts CHC primary care physicians, retention was significantly related to having opportunities for research. Safety net PCPs have high demands with large patient volumes and work burden, and participation in research may lead to more job autonomy and in turn increase career satisfaction. While many of the medical directors agreed that involvement in research could enhance physician retention, they also cautioned that this strategy would only be appropriate for some PCPs-those with existing research skills or experience. To more fully examine which PCPs would benefit from research engagement, more research is needed to understand the relationship between research engagement and career satisfaction from the PCP perspective.
While the majority of medical directors reported that their PCPs were interested in clinical research engagement, they reported numerous barriers to this engagement. These range from financial barriers (lost patient revenue, cost of data extraction, cost to cover PCP time) to limited PCP skills and knowledge about research, to inaccessible or inappropriate research partners. Harvard Catalyst has had some success overcoming these barriers through its Safety Net Initiative which has included activities such as training PCPs in research skills, providing mentoring, linking PCPs to academics, and supporting small amounts of provider time to participate.
One important and unexpected finding of the survey also merits discussion. While the survey response rate was high, many medical directors had difficulty responding to the demographics questions despite the fact that it is the same data which is reported annually for FQHC health centers. In several follow-up conversations, respondents indicated that they did not have access to this information because it was in someone else's purview (e.g., QI manager or executive director) or that it was not easily available to them via current data systems. As safety net providers seek to manage their populations, it will become increasingly important to understand the total number of patients and their characteristics served by a given CHC. This information is not only important for clinical management but is a first step needed to engage in research. CHC providers need to be able to review, manipulate, and report on patient population data in order to determine adequate sample sizes and devise appropriate research designs. Additional research is needed to better understand the barriers that medical directors and PCPs experience in trying to obtain this type of information.
This exploratory study has a number of limitations. First it was conducted in only one state and results may not be generalizable to other health centers in other states. While the response rate was high (>60%) the total number of sites was low. Thus we were unable to test a predictive model for building research infrastructure. Second, as respondents were only the CHC medical directors, their views may not represent of all the doctors or nonphysician staff at their centers. With that said, this is one of the first studies to examine correlates of interest in building research infrastructure in CHCs. For translational purposes, this further elucidates our understanding of how and when CHCs may want to contribute to the research enterprise. This is particularly important for the clinical translational science awardees of the National Institutes of Health who are attempting to build this infrastructure and translate research into practice in primary care.
CHCs play a critical role in caring for vulnerable populations, ideally positioning them to raise research questions relevant to these groups and encourage translation of evidence into practice. Our findings suggest that there is high interest in engagement in research among CHC leadership and engagement may promote provider retention. However, there is a need to contribute appropriate resources to support infrastructure needs and engagement of PCPs. CTSAs have a unique opportunity to work collaboratively with local CHCs in this endeavor to address these barriers and enhance the clinical translational continuum.
The authors would like to acknowledge Ellen Hafer and Keith Maxwell of the Massachusetts League of Community Health Centers for their assistance with the survey and support with distribution. We would also like to thank Shimon Shaykevich and Blessing Dube for help with statistical analysis.
This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (NIH Award #UL1 RR 025758 and financial contributions from Harvard University and its affiliated academic health care centers). The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic health care centers, the National Center for Research Resources, or the National Institutes of Health.