health information

ABOUT PRACTICE-BASED RESEARCH NETWORKS

Supporting Better Science in Primary Care: A Description of Practice-based Research Networks (PBRNs) in 2011 Kevin A. Peterson, MD, MPH, Paula Darby Lipman, PhD, Carol J. Lange, MPH, Rachel A. Cohen, MPH, and Steve Durako, BA

Background: Bound by a shared commitment to improving medical care through systematic inquiry, practice-based research networks (PBRNs) provide a basic laboratory for primary care research and dissemination.

Methods: Data from US primary care PBRNs were collected as part of the 2011 Agency for Healthcare Research and Quality PBRN registration process. Data addressed PBRN characteristics, research activi- ties, and perceived strengths and weaknesses.

Results: One hundred forty-three primary care PBRNs were registered with the resource center in 2011, including 131 that were identified as either eligible for Agency for Healthcare Research and Qual- ity recognition (n � 121) or as developing (n � 10). These PBRNs included 12,981 practices with more than 63,000 individual members providing care to approximately 47.5 million people. PBRNs had an average of 482 individual members (median, 170) from 101 practices (median, 32).

Conclusions: PBRNs are growing in experience and research capacity. With member practices serving approximately 15% of the US population, PBRNs are adopting more advanced study designs, disseminat- ing and implementing practice change, and participating in clinical trials. PBRNs provide valuable ca- pacity for investigating questions of importance to clinical practice, disseminating results, and imple- menting evidence-based strategies. PBRNs are well positioned to support the emerging public health role of primary care providers and provide an essential component of a learning health care system. ( J Am Board Fam Med 2012;25:565–571.)

Keywords: Family Medicine Research, Practice-based Research, Practice-based Research Networks

Primary care practice-based research networks (PBRNs) enhance the performance of clinical re- search in community settings and speed the dissem- ination of new knowledge into practice.1,2 Bound by a shared commitment to improving medical care

through systematic inquiry, PBRNs provide a basic laboratory for primary care research and dissemi- nation involving every state and territory in the United States.3,4 The ability of PBRNs to involve “real-world” practices in clinical research provides new opportunities to engage understudied popula- tions, to study a range of health problems, and to accelerate community adoption of new knowledge and best practices.5,6

The Agency for Healthcare Research and Qual- ity (AHRQ) has a long history of supporting pri- mary care research networks. In 2002, the AHRQ created the National PBRN Resource Center to identify existing networks and promote growth in their capacity for clinical research. Led initially by the University of Indiana and National Opinion Research Center at the University of Chicago, in late 2007 the PBRN Resource Center was awarded

This article was externally peer reviewed. Submitted 11 April 2012; revised 16 June 2012; accepted

19 June 2012. From the Department of Family Medicine and Commu-

nity Health, University of Minnesota Medical School, Min- neapolis (KAP, CJL); and Westat, Rockville, MD (PDL, RAC, SD).

Funding: This study was supported by the Agency for Healthcare Research and Quality contract no. HHSA 290200710037.

Conflict of interest: Authors are affiliated with the AHRQ funded PBRN Resource Center.

Corresponding author: Kevin A. Peterson, MD, MPH, De- partment of Family Medicine and Community Health, University of Minnesota Medical School, 717 Delaware Street SE, Ste. 425, Minneapolis, MN 55414 (E-mail: [email protected]).

doi: 10.3122/jabfm.2012.05.120100 PBRNs in 2011 565

to the University of Minnesota and Westat. The Resource Center has supported a wide variety of learning opportunities to foster new PBRNs and promote collaboration, including an annual re- search conference, peer learning groups, technical assistance, and development of a range of electronic tools and collaborative resources.

An important role of the Resource Center is the maintenance of a PBRN registry. To maintain AHRQ recognition, PBRNs provide annual up- dates of administrative information and research activities. These data are summarized on the public PBRN website (pbrn.ahrq.gov) and inform learn- ing activities planned by the Resource Center. Data on characteristics and productivity of primary care PBRNs from 2003 to 2004 were last published in 2007.7 This article updates the status of PBRNs from the 2011 registration data, explores the rela- tionship between key characteristics of PBRNs and general indicators of research capacity, and pro- vides a perspective on changes over time.

Methods PBRN Criteria To be eligible for AHRQ certification, a network must have a minimum of 5 practices (15 providers) devoted principally to the care of patients, and be united by a shared commitment to expand the sci- ence base of clinical care and better understand the health care events that unfold daily in their practice settings.8 PBRNs must be located in the United States or a US territory and have at least 50% of their membership in family medicine, pediatrics, general internal medicine, nursing, physician assist- ing, osteopathy, or chiropractic. In addition to PBRNs, the AHRQ registry records two other types of networks: “Developing networks” are in the early stages of PBRN formation or have not conducted at least one study, and “affiliate net- works” do not meet requirements for membership as a primary care PBRN. Affiliate membership is generally provided to non–primary care PBRNs (eg, pharmacy and dentistry) and PBRNs based outside of the United States. Although registration is voluntary, annual registration is required to re- ceive support from the Resource Center. The AHRQ intermittently re-examines and refines the registration form to identify needs and plan support activities.

2011 Registration The 2011 registration form contained 39 items addressing PBRN characteristics, productivity, and research interests. Self-reported strengths and challenges were used to identify peer learning group topics, consultation service needs, and im- portant topics for the annual PBRN research con- ference.

Registry data were collected electronically using 2010 Checkbox software (Checkbox Survey Solu- tions, Inc, Watertown, MA). Personalized E-mails were sent to all previously registered PBRN direc- tors and coordinators with an embedded link to a prepopulated form from the previous year. Tech- nical assistance was provided to PBRNs to facilitate completion. Reminder E-mails were sent monthly from December 2010 through March 2011, with follow-up telephone calls beginning in mid-Febru- ary 2011. Telephone follow-up included determin- ing the status of nonregistered PBRNs. New net- works interested in registration completed a screening questionnaire. If eligible, new networks were provided with a secure Internet account and an invitation to register using a blank registration form.

Registry data addressed the number and scope of PBRN research studies during the previous year (2010). Respondents also were asked to report on the strengths and challenges that impact gover- nance, membership, operations, conduct of re- search, or use of information technology. These variables were chosen for their relevance to PBRN operations/functioning and to allow comparison with previous registry data.

Statistical Methods Self-reported data from each PBRN regarding the number of practices, health care providers, and patients served were aggregated to generate overall estimates. Analyses were conducted to identify po- tential relationships between specific PBRN char- acteristics and research productivity. Three cate- gorical PBRN characteristics were identified: (1) whether the specialty area of network members was primarily family medicine; (2) whether the network was affiliated with a funded Clinical Translational Science Award (CTSA); and (3) geographic cover- age (national vs not national). These were com- pared using 3 measures of productivity/capacity: (1) number of studies conducted in 2010, (2) type of study designs used, and (3) use of electronic health

566 JABFM September–October 2012 Vol. 25 No. 5 http://www.jabfm.org

record (EHR) data for research. Analyses of differ- ences in the continuous variable (mean number of studies in 2010) by PBRN characteristics were con- ducted using analysis of variance. Associations be- tween categorical characteristics (type of study de- sign, use of EHR for research) were made using Fischer exact test. P � .05 was considered statisti- cally significant. Analyses were conducted using SAS 1 software, version 9.1 (SAS Institute, Cary, NC).

Results Description of PBRNs As of May 2011, 143 PBRNs were registered with the PBRN Resource Center, an increase in 30 from the prior registration year. These included 121 primary care PBRNs, 10 developing networks, and 12 affiliate networks. The 131 established and de- veloping PBRNs included 12,981 practices with more than 63,000 individual members providing care to approximately 47.5 million people. The following PBRN descriptive data are based on these 131 US primary care PBRNs.

PBRN Organizational and Member Characteristics In 2011, PBRNs had been functioning for an aver- age of 8.6 years (median, 8 years; range, �1–39 years). Sixty-two PBRNs have been registered con- tinuously since 2008. A total of 133 different reg- istered PBRNs were recorded by the AHRQ be- tween 2008 and 2011. Three of these had previously reported disbanding and one disbanded during the 2011 registration process. Two PBRNs had merged with other PBRNs, one no longer met primary care eligibility requirements, and 5 re- ported being active but did not complete 2011 registration.

All PBRNs have a director, 89% (116 of 131) have a coordinator or associate director, and 73% (95 of 131) have a public website. PBRNs reported an average of 482 individual members (median, 170 members; range, 0 –14,952 members) and a mean of 101 member practices (median, 32 practices; range, 0 –1600 practices). PBRN member practices serve a mean of more than 415,000 patients (me- dian, 169,000 patients; range, 0 –7,000,000 pa- tients) receiving care from a mean of 219 physicians per PBRN (median, 118 physicians; range, 0 –1848 physicians).

As shown in Table 1, 41.5% of PBRNs identify their membership as “mixed,” with some combina- tion of family medicine, general internal medicine, pediatrics, nursing, or other specialty. Most PBRNs report that at least 75% of their clinicians belong to a single specialty, with approximately one third of networks indicating they were predomi- nately family medicine. Approximately half (49%) of current PBRNs report that the majority of their member practices have an EHR, with 19% (25 of 131) reporting that less than one fourth of member practices have an EHR.

More than 80% of PBRNs report that they are local, state-based, or regional. The 25 PBRNs that identify as national PBRNs have members or prac- tices in at least 10 states. The national PBRNs fall into 2 major categories: (1) networks sponsored by national professional organizations or EHR com- panies, and (2) networks focused on a specific sub- ject such as a patient population (ie, homeless) or innovation (ie, collaborative care).

Most PBRNs (65%) reported primary affiliation with a university, and most of the remainder (30%) reported affiliation primarily with a non-profit or professional organization. The most common pri- mary funding source is federal grants (63%); how- ever, a variety of additional funding sources were identified. Half of all PBRNs report affiliation with a funded CTSA.

Approximately one third of PBRNs (n � 39) reported membership in an organized “network of networks,” and 36% participated in a multinetwork project in 2010. Half of the PBRNs that did not participate in such a project have plans to do so.

PBRN Research Scope The research focus for more than 60% of PBRNs is underserved, low-income, and minority popula- tions. Smaller percentages targeted inner-city or rural populations. Only one third (32%) report that the network does not focus on a specific population group. Diabetes was the most commonly studied health condition (50%), whereas obesity was a cur- rent focus for more than one third of PBRNs (42%). The specific study designs used in the past 5 years are reported in Table 2 in descending order of frequency. The most common research designs were observational epidemiology, health systems/ outcomes research, and best practice modeling. More than one fourth (28%) have conducted a clinical trial.

doi: 10.3122/jabfm.2012.05.120100 PBRNs in 2011 567

The number of studies conducted by the 121 registered PBRNs ranged from 1 to more than 40. Most PBRNs (56%) had conducted �8 studies, whereas 36% had conducted more than twice that number. On average, PBRNs conducted 4 studies per year (range, 2– 6 studies). In the majority of networks (94%), members and practices decide in- dependently whether to participate in PBRN stud- ies. Almost 70% of the PBRNs have used an EHR for research.

PBRN Strengths and Challenges Table 3 lists areas that many (�40%) of the 131 registered PBRNs report as either a strength or a challenge. Generally, PBRNs identify areas related to the conduct of practice-based research (eg, agenda setting, study development, study manage- ment) as strengths. Most PBRNs continue to be challenged by member compensation and provider training, community involvement, and infrastruc- ture support.

Comparison by PBRN Characteristics No significant differences in research productivity/ capacity were identified by the categorical PBRN characteristics of specialty, CTSA affiliation, or geographic coverage. Although 63 registered PBRNs (52%) reported a formal affiliation with a CTSA, in 2011 these PBRNs conducted the same number of studies as PBRNs not affiliated with a CTSA (mean, 4 studies; P � .32). In addition, no significant differences in the type of study designs used or in the use of EHRs were found between PBRN specialty designations or geographic cover- age.

Discussion A description of the current state of AHRQ-regis- tered PBRNs provides a framework for recognizing

Table 1. Practice-based Research Network (PBRN) Characteristics (N � 131)

Primary specialty of PBRN Mixed 53 (40.5) Family medicine 42 (32.1) Pediatrics 16 (12.2) General internal medicine 8 (6.1) Nursing 4 (3.1) Other 8 (6.1)

Presence of members by clinical discipline Physicians 122 (93.1) Nurse practitioners 93 (71.0) Physician assistants 76 (58.0) Other clinician types 67 (51.2)

Members with an EHR (%) 51–100 64 (48.9) 26–50 35 (26.7) 1–25 22 (16.8) None or unknown 10 (7.6)

Geographic coverage Regional 40 (30.5) State 36 (27.5) Local or citywide 30 (22.9) National 25 (19.1)

Primary affiliation University 85 (64.9) Not-for-profit 38 (29.0) Professional organization 5 (3.8) For-profit 1 (0.8) None 0 (0) Other 2 (1.5)

Funding sources (during past 5 years)* US government 83 (63.4) Academia 63 (48.1) Nonprofit 51 (38.9) State/local government 32 (24.4) Professional organization 31 (23.7) For-profit/commercial 23 (17.6) Government outside United States 1 (0.8) Other sources 14 (10.7)

Target populations* Underserved 79 (60.3) Low income 71 (54.2) Minority 71 (54.2) Inner city 51 (38.9) Rural 49 (37.4) No target population 42 (32.1) Other (unspecified) 16 (12.2)

Relationship to CTSA None 53 (40.5) Funded with CTSA 66 (50.4) Planning with CTSA 12 (9.2)

Continued

Table 1. Continued

Multinetwork project/study (during past year) None 36 (27.5) None yet, but planning 38 (29.0) No, but have in the past 10 (7.6) Yes 47 (35.9)

Data provided as n (%). *Multiple responses permitted. EHR, electronic health record; CTSA, Clinical Translational Science Award.

568 JABFM September–October 2012 Vol. 25 No. 5 http://www.jabfm.org

changes in PBRNs over time. Comparing the re- sults of the 2011 registry data with data from a 2003 to 2004 cohort of PBRNs demonstrates consider- able growth in PBRN research capacity during the past decade. This includes sustained and continued growth in the number of PBRNs, from 28 networks in 1994 and 86 in 2003 to 131 in 2011. As the number of networks has increased, the number of member practices and the number of patients that potentially could be involved in or impacted by practice-based research also has increased. In 2003, the 86 networks that met AHRQ criteria had a total

of 1871 practices. The 131 PBRNs in 2011 in- cluded nearly 13,000 practices, a growth of nearly 6-fold. The current estimate suggests that approx- imately 15% of the US population currently re- ceives health care from a PBRN-member practice. Of the 45 PBRNs reporting one ongoing or com- pleted research project in 2004, only 20 had con- ducted �7 studies. In contrast, in 2011, 68 PBRNs (56%) reported having completed �8 studies, and 22 report having completed �40 studies. With in- creasing experience, PBRNs also are using more advanced study designs. Half of the registered PBRNs now report using designs for modeling best practices, and 30% are performing implementation research or clinical trials.

From 2008 to 2011, a total of 141 unique pri- mary care PBRNs were registered. Although not all PBRNs re-register every year, only 4 are known to have disbanded. The growth in number of PBRNs seems to be largely because of new PBRNs forming in nontraditional primary care disciplines, new PBRNs associated with CTSA community engage-

Table 2. Scope of Practice-based Research Network (PBRN) Research Studies (N � 121)

Studies ever conducted, n 1–3 29 (24.0) 4–7 24 (19.8) 8–16 25 (20.7) 17–39 21 (17.4) �40 22 (18.2)

Used EHR for research 83 (68.6) AHRQ priority health conditions studied (during

past 5 years)* Diabetes mellitus 60 (49.6) Obesity 51 (42.1) Pulmonary disease/asthma 44 (36.4) Cardiovascular disease 42 (34.7) Mental health disorders 34 (28.1) Cancer 32 (26.4) Substance abuse 18 (14.9) Development delays 14 (11.6) Infectious disease (including HIV/AIDs,

sexually transmitted diseases) 13 (10.7)

Dementia 11 (9.1) Arthritis and joint disease 10 (8.3) Pregnancy and childbirth 10 (8.3)

Study designs used (during past 5 years)* Observational epidemiology 66 (54.5) Health systems/outcome research 63 (52.1) Best practice research/modeling 61 (50.4) Implementation research 39 (32.2) Clinical trials 34 (28.1) Comparative effectiveness research 31 (25.6) Methodological research 22 (18.2) Nonpractice-based community health

intervention 18 (14.9)

Pharmaceutical clinical trials 12 (9.9)

Data provided as n (%). Ten developing PBRNs that had not completed a research project were excluded from the description of PBRN research. *Multiple responses permitted. EHR, electronic health record; AHRQ, Agency for Health care Research and Quality.

Table 3. Most Commonly Reported Practice-based Research Network (PBRN) Strengths and Challenges*

Strengths Leadership 86 (65.6) Study development 79 (60.3) Study management 76 (58.0) Data management 71 (54.2) Network staff 70 (53.4) Diversity of patient population 68 (51.9) Capacity to conduct research 60 (45.8) Geographic distribution 59 (45.0) Access to patient data 58 (44.3) IRB/HIPAA 58 (44.3) Recruitment 54 (41.2) Computer access at practices 53 (40.5) Research agenda setting 52 (39.7)

Challenges Infrastructure funding 89 (67.9) Infrastructure support 69 (52.7) Compensation strategies 64 (48.9) EMR availability/interface 57 (43.5) Community participation 55 (42.0) Provider training 52 (39.7) Member involvement 52 (39.7)

Data provided as n (%). *Reported by �40% of the practice-based research networks (N � 131). IRB, institutional review board; HIPAA, Health Insurance Pri- vacy and Accountability Act; EMR, electronic medical record.

doi: 10.3122/jabfm.2012.05.120100 PBRNs in 2011 569

ment activities, and more comprehensive identifi- cation of PBRNs across all states and territories. Although some PBRNs have merged, no evidence of PBRN fragmentation was seen.

The current data demonstrated no significant association between selected PBRN characteristics and research productivity over the prior year as measured by number of studies, study designs, or use of EHR data. This confirms findings from other investigators, and the association between effective PBRN infrastructure characteristics and research productivity remains unclear.8

PBRNs frequently participate in multinetwork research projects. This is consistent with observa- tions from other investigators that PBRNs are in- creasingly engaging in formal relationships with other research organizations, including CTSAs.9

The current analysis suggests, however, that formal relationships with CTSAs are still in preliminary stages and have not yet resulted in more research projects per year, nor has CTSA affiliation resulted in a significant difference in self-reported research capacity, as explored in this article.

Many of the self-reported challenges of PBRNs remain the same as those reported 8 years ago. The ability to coordinate a study in the community, geographic distribution of members’ practices, and diversity all remain identified as PBRN strengths. Infrastructure funding, support, and compensation strategies remain the biggest challenges. Although in 2004, 40% (of 83 networks) reported that com- munity involvement was a strength; in 2011 only 28% found community participation a strength, and 42% found it a challenge. PBRNs are continu- ing to address ways to improve community partic- ipation, and this may reflect both increased aware- ness and increased pressure to promote meaningful participation of communities in research. The abil- ity to secure funding continues to be a challenge (for 65% in 2004 and 68% in 2011). Although computer access at the practice was a strength, EHR interfaces were identified as a challenge.

Study Limitations The current analysis of AHRQ PBRN registration data has a number of limitations. Given the AHRQ’s focus on primary care, PBRNs with non–primary care specialties (affiliate members) were excluded. The data set also relies on the interest of PBRNs to join the community of AHRQ-registered networks and may exclude other functional primary care re-

search organizations. These data are self-reported and are not validated. In addition, practices or pro- viders belonging to multiple PBRNs would be du- plicated in aggregate estimates. Another important limitation is that the purpose of the data collection was for supporting the administrative and technical assistance roles of the Resource Center in meeting the needs of the PBRN community, rather than explicitly addressing research questions pertaining to the strategic value or impact of PBRNs in ad- vancing research in primary care settings. Finally, the registry does not address the number of mem- ber practices within the PBRN that are active par- ticipants in the research process, so estimates based on the overall number of practices may overempha- size actual research capacity.

Assessing the Future of PBRNs and Practice-based Research Despite the limitations of the findings, the oppor- tunity to describe PBRNs in 2011 and consider factors associated with their sustainability and re- search capacity can provide insights that are valu- able for policymakers, PBRN participants, and the broader primary care community. In the current health care environment, PBRNs are positioned to address the emerging public health role of primary care providers and provide an essential component of a learning health care system.10 PBRNs can draw on the experience and insight of practicing clini- cians to identify and frame research questions so that new findings can be applied directly to clinical practice. The 2012 PBRN registration form in- cludes information about how networks engage member practices in research and how they dissem- inate evidence-based approaches to care and best practices to their members in the community.

The role of PBRNs continues to evolve in the direction of a stronger focus on health improve- ment, primary care transitions, and providing con- tinuing education and maintenance of certifica- tion.11,12 PBRNs are continuing to increase their capacity to investigate questions of importance to clinical practice, to disseminate results, and to im- plement evidence-based strategies. By blending re- search design and community practice experience, PBRNs provide research findings that are recog- nized as relevant by primary care clinicians. A bet- ter understanding of how challenges such as mem- ber compensation, provider training, and community involvement affect the capacity of practices to par-

570 JABFM September–October 2012 Vol. 25 No. 5 http://www.jabfm.org

ticipate would advance the ability of PBRNs to fulfill the promise of supporting better science in primary care.4

References 1. Westfall JM, Mold J, Fagnan L. Practice-based re-

search–“Blue Highways” on the NIH roadmap. JAMA 2007;297:403– 6.

2. Nutting PA, Beasley JW, Werner JJ. Practice-based research networks answer primary care questions. JAMA 1999;281:686 – 8.

3. Durako S, Peterson K. (2011). The AHRQ PBRN Resource Center Update. AHRQ PBRN 2011 Na- tional Conference Proceedings. AHRQ, DHHS, Wash- ington, DC. June 23, 2011. Retrieved from https:// portal.pbrn.ahrq.gov/portal/server.pt/community/ pbrn_portal_annual_pbrn_conference/1293/2011_ archives/27943.

4. Green LA, Hickner J. A short history of primary care practice-based research networks: from concept to essential research laboratories. J Am Board Fam Med 2006;19:1–10.

5. Mold JW, Peterson KA. Primary care practice-based research networks: working at the interface between research and quality improvement. Ann Fam Med 2005;3(Suppl 1):S12–20.

6. Lindbloom EJ, Ewigman BG, Hickner JM. Practice- based research networks: the laboratories of primary care research. Med Care 2004;42:III45–9.

7. Tierney WM, Oppenheimer CC, Hudson BL, et al. A national survey of primary care practice-based re- search networks. Ann Fam Med 2007;5:242–50.

8. Green LA, White LL, Barry HC, Nease DE Jr, Hudson BL. Infrastructure requirements for prac- tice-based research networks. Ann Fam Med 2005; 3(Suppl 1)S5–11.

9. Fagnan LJ, Davis M, Deyo RA, Werner JJ, Stange KC. Linking practice-based research networks and Clinical and Translational Science Awards: new op- portunities for community engagement by academic health centers. Acad Med 2010;85:476 – 83.

10. Delaney BC, Peterson KA, Speedie S, Taweel A, Arvanitis TN, Hobbs FD. Envisioning a learning health care system: the electronic primary care re- search network, a case study. Ann Fam Med 2012; 10:54 –9.

11. Institute of Medicine. Primary care and the public health: exploring integration to improve population health. Washington, D.C.: The National Academies Press; 2012.

12. Williams RL, Rhyne RL. No longer simply a prac- tice-based research network (PBRN) health im- provement networks. J Am Board Fam Med 2011; 24:485– 8.

doi: 10.3122/jabfm.2012.05.120100 PBRNs in 2011 571