OBJECTIVES: The goals were (1) to describe and to characterize pediatric clinical research networks (PCRNs) in the United States and Canada, (2) to identify PCRN strengths and weaknesses, (3) to evaluate the potential for collaboration among PCRNs, and (4) to assess untapped potential interest in PCRN participation.
METHODS: Data collection included (1) initial identification of PCRNs through an Internet search and word of mouth, (2) follow-up surveys of PCRN leaders, (3) telephone interviews with 21 PCRN leaders, and (4) a survey of 43 American Academy of Pediatrics specialty leaders regarding untapped interest in network research.
RESULTS: Seventy exclusively pediatric networks were identified. Of those, specialty care networks constituted the largest proportion (50%), followed by primary care (28.6%) and disease-specific (21.4%) networks. A network profile survey (response rate: 74.3%) revealed that ∼90% held infrastructure funding. Nearly 75% of respondents viewed cross-network collaborations positively. In-depth telephone interviews corroborated the survey data, with cross-network collaboration mentioned consistently as a theme. American Academy of Pediatrics specialty leaders indicated that up to 30% of current nonparticipants might be interested in research involvement.
CONCLUSIONS: Pediatric networks exist across the care continuum. Significant numbers of uninvolved practitioners may be interested in joining PCRNs. A strong majority of network leaders cited potential benefits from network collaboration.
WHAT'S KNOWN ON THIS SUBJECT:
Although a very large amount of scientific literature has been generated by PCRNs and several articles described particular networks, no published articles have examined and compared PCRNs as a group or attempted to gauge the interest of networks in collaboration.
WHAT THIS STUDY ADDS:
Through surveys and interviews with network leaders, we learned that pediatric networks exist across the care continuum. Significant numbers of uninvolved practitioners may be interested in joining PCRNs. A strong majority of network leaders cited potential benefits from network collaboration.
Pediatric clinical research networks (PCRNs) have a 50-year record of advancing pediatric care. Although network models and missions vary, PCRNs can generally be defined as multiinstitutional groups “conducting research on human subjects or using data from human subjects that is relevant to improving the quality of human health.”1 PCRNs grew out of the particular challenges to discovery in pediatric populations. Childhood diseases are relatively rare. Adverse clinical study outcomes for pediatric diseases are comparatively infrequent and remote to disease onset, and the same situation prevails for outcomes of pediatric preventive measures. Gathering adequate sample sizes and monitoring subjects longitudinally is logistically difficult for researchers at individual institutions. PCRNs arose to meet these challenges and have been productive in their individual endeavors.
The first PCRNs arose in the pediatric oncology community in the 1950s.2,3 The rationale for these networks was straightforward, that is, individual institutions had insufficient numbers of patients with particular cancers to conduct adequately powered trials of new therapies over reasonable time periods. Ultimately, several of the pediatric oncology networks coalesced into the Children's Oncology Group. In the 1970s, pediatric rheumatologists established a research network.4 A regional and a national pediatric primary care research network were launched in the 1980s.5,6 At the same time, neonatologists established 2 networks, one based at large academic medical centers7 and another that included neonatologists at large community hospitals.8 Many other networks (disease-specific, specialist, and primary care) have been formed since then.9 These networks have a rich tradition of achieving findings that have been translated into improved care and outcomes. The networks, however, have been working largely in isolation. A sense of community and collaboration has been created across primary care practice-based research networks through the innovative Primary Care Research Networks program of the Agency for Healthcare Research and Quality.10 To date, however, no formal mechanisms exist to facilitate collaboration among PCRNs, and it is unclear to what extent network leaders and participants would be amenable to such collaboration.
Although a large amount of scientific literature has been generated by PCRNs and several articles have described particular networks, we are aware of no publications that examine and compare PCRNs as a group. Therefore, we undertook an investigation with the goals of (1) describing and characterizing PCRNs in the United States and Canada, (2) identifying relative strengths and weaknesses common to PCRNs, (3) evaluating the potential for collaboration among PCRNs, to improve efficiency and effectiveness, and (4) assessing untapped potential interest in PCRN participation among pediatricians.
This investigation was conducted by researchers with Pediatric Research in Office Settings, the national, primary care PCRN of the American Academy of Pediatrics (AAP). Institutional review boards of the AAP and the University of Vermont granted approval for study activities. The researchers used multiple methods to gather information on pediatric network research. These approaches included (1) a 2-stage survey of PCRNs identified through an Internet search and through interviews, (2) telephone interviews with 21 leaders in pediatric network research, and (3) a survey of 43 AAP leaders regarding untapped interest in network research.
Descriptive Data on PCRNs
Researchers gathered data on primary care, specialty care, and disease-specific PCRNs. Networks were identified through searches of Web sites, including the National Institutes of Health Roadmap Initiative Inventory and Evaluation of Clinical Research Networks, the Agency for Healthcare Research and Quality Practice-Based Research Network Resource Center, the American Academy of Family Physicians Federation of Practice-Based Research Networks, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, as well as through word of mouth. The investigation focused exclusively on pediatric networks (those not conducting any adult studies). Seventy such networks were identified.
PCRN leaders were identified through information posted on the Web sites noted above and on network-specific Web sites. When leadership had changed, the current leader was identified. Contact information was verified through an initial e-mail. After the verification, leaders were sent an e-mail with an embedded hyperlink to an online survey asking about network characteristics such as longevity, productivity, funding, and study topic areas.
Telephone Interviews With PCRN Leaders and Stakeholders
Twenty-one leaders and stakeholders in PCRN research (eg, clinicians active in PCRNs) participated in semi-structured 1-hour interviews between February and July 2007. The informant list was developed initially to ensure a diversity of PCRN types and experiences. Additional informants were approached to explore themes that emerged in the initial interviews (eg, funding issues). A physician researcher led each telephone interview, using a standard script with additional questions to address emerging concepts; health researchers took notes and recorded calls. Health researchers produced detailed summaries, which were confirmed by the physician researcher and informants. Qualitative analysis, by using a grounded-theory approach, consisted of 2 main activities, that is, postinterview debriefing and coding/theme compilation, with the use of Atlas.ti (Scientific Software Development GmbH, Berlin, Germany) to organize data. Health researchers independently coded each summary for themes and subthemes. Coding variations were resolved through physician researcher and health researcher discussion. Consensus was achieved for all themes and subthemes.
Assessment of Untapped Pediatrician Interest in Network Research
Fifty-five AAP section and council chairpersons were surveyed, by using an online survey program, in January and February 2008. The AAP sections and councils are groups within the AAP that address subspecialty and clinical interests, including age-based groupings, organ system–based groupings, other clinical groupings such as child abuse and neglect, and pharmacology. The survey's purpose was to assess the section and council leaders' impressions of their members' current participation in clinical research, as well as potential interest in increasing participation in research (including research networks).
Descriptive Data on PCRNs
Of the 70 exclusively pediatric networks identified, specialty care networks (n = 35) represented the largest proportion (50%), followed by primary care networks (n = 20 [28.6%]) and disease-specific networks (n = 15 [21.4%]). Of the 70 networks approached, 52 (74.3%) completed the profile survey. The 18 nonresponding networks included 7 primary care (38.8%), 7 specialty care (38.8%), and 4 disease-specific (22.2%) networks. Table 1 reveals that regional networks were in the minority, with the majority of networks being national or international. Although the median age of networks at the time of data collection (ie, 2009) was 8.0 years, there was a wide range in the distribution, with networks having been established as recently as 2006 and as early as 1955. On average, specialty care networks had been in existence the longest (mean: 14.0 years; SD: 12.5 years), followed by their primary care (mean: 11.6 years: SD: 6.9 years), and disease-specific (mean: 7.3 years; SD: 2.5 years) counterparts. The median numbers of data collection efforts and articles published (6 efforts and 10 articles) do not fully describe the variation in network experience. One network, the Children's Oncology Group, was particularly prolific, having been in existence since 1955, having participated in >1000 data collection efforts, and having published >5000 articles. Given the longevity and performance of the Children's Oncology Group, we chose to regard it as an outlier and removed it from analyses related to productivity. Even when the Children's Oncology Group was removed from the distribution, there was a wide range of experience, with numbers of completed collection efforts ranging from 0 to 80 (mean: 11.45 efforts; SD: 15.8 efforts) and numbers of published articles ranging from 0 to 143 (mean: 23.81 articles; SD: 36.1 articles). Supplemental Table 4 provides examples of the breadth of network study types and topics. PCRN investigations include the full range of clinical research, from qualitative analyses, diagnostic test evaluations, and registries to various types of outcomes research, pharmacokinetic studies, and randomized controlled trials.
Network leaders also were surveyed about funding. Sixteen networks (30.7%) reported multiple sources of current infrastructure support. In contrast, 6 networks (11.5%) indicated no current source of infrastructure funding. As Table 2 indicates, the most-frequently cited sources of current infrastructure funding were the federal government (61.5%), charitable foundations (26.9%), universities (15.4%), and professional organizations (11.5%). With respect to project-specific funding, 40.4% networks reported multiple sources, whereas 28.8% of networks indicated no current source of project funding. Once again, the federal government and charitable foundations were the most common funding sources, but industry, which was the source of little infrastructure funding, funded specific projects for 25.0% of networks.
We examined the associations between the predominant types of infrastructure funding (federal) and 2 specific variables, that is, network type and longevity. No specific association was found between federal funding and network type. For analytic purposes, we made longevity a dichotomous variable, with young networks being defined as those formed at the 8-year longevity median described above or later (n = 24) and older networks being defined as those in existence for >8 years (n = 25; data for 3 networks were missing). Again, no significant association was found. With respect to project-specific federal funding, we used the same analytic scheme to determine whether there were associations between such funding and network type or longevity. There was no association between project-specific funding and network type. With respect to longevity, larger proportions of older networks reported the receipt of project funding; this difference approached but did not reach statistical significance (χ12 = 2.761;P = .097).
Analysis of average numbers of data collection efforts suggested that primary care networks launched more efforts than their counterparts, with a mean of 20.9 efforts for primary care networks (SD: 24.6 efforts), compared with means of 9.1 efforts (SD: 9.9 efforts) for specialty care networks and 4.6 efforts (SD: 2.1 efforts) for disease-specific networks. The difference among groups approached but did not achieve statistical significance (P = .077). Slightly different findings were noted when the average numbers of peer-reviewed publications were examined according to network type. In this respect, specialty care networks seemed more productive than their counterparts, with a mean of 33.7 articles (SD: 46.3 articles), compared with means of 17.8 articles (SD: 28.0 articles) for primary care networks and 11.8 articles (SD: 10.3 articles) for disease-specific networks. The overall difference among network types, however, was not significant (P = .227).
Another set of descriptive analyses addressed the correlates of peer-reviewed publication numbers. One analysis examined network longevity and noted that networks with longer histories (ie, formed before the 8-year longevity median) were significantly more prolific than their younger counterparts, with an average of 39.67 versus 6.58 publications (t35 = 3.199; P = .003). A second analysis examined whether total numbers of data collection efforts were associated with numbers of peer-reviewed publications; there was no overall association (χ21 = 2.29; P = .13). Finally, we examined the link between longevity and number of data collection efforts completed. Older networks outpaced their younger counterparts, with an average of 17.4 such efforts (n = 16), compared with 8.2 efforts for the younger groups (n = 18). With a total sample size of 34 for this analysis, however, no statistical significance emerged (t32 = 1.682; P = .102).
Evaluation of Potential for Collaboration
As part of the network profile survey, network leaders were asked to respond to a list of potential benefits of participation in a coalition of networks. Overall, 73% cited ≥2 tangible benefits of such a coalition. Table 3 details specific benefit areas. A majority of respondents suggested that benefits would accrue in the areas of collaboration on study proposals (60%), advocacy for networks (58%), and connections with national pediatric organizations (54%). Most other areas (ie, strategies for recruitment, addressing institutional review board issues, and ideas for policy-making) received support from approximately one-half of the respondents. Of note, 10% of respondents wrote in “sharing best practices” in response to this question, although this was not listed among the response categories.
There was no association between longevity and numbers of benefits endorsed (3.80 benefits were endorsed for older networks, compared with 2.68 efforts for their younger counterparts; t40 = 1.55; P = .129), and there was no difference between older and newer networks with respect to the perceived specific benefit of collaborating on study proposals (72.7% vs 51.9%; χ21 = 2.225; P = .136). A similar analysis was conducted to examine the potential association between numbers of benefits acknowledged and network type. Primary care networks endorsed a mean of 4.23 such benefits, whereas specialty care networks (3.18 benefits) and disease-specific networks (2.09 benefits) endorsed fewer. An overall analysis of variance suggested a trend toward differences among the groups, but that difference did not reach significance (F = 2.82; P = .069). In a comparison of network types and specific benefits endorsed, there was a difference among network types with respect to collaboration on study proposals (76.9% for primary care networks, 64.3% for specialty care groups, and 27.3% for disease-specific groups; χ22 = 6.651; P = .036). Posthoc analyses revealed the overall difference to have been driven by the difference between the primary care and disease-specific networks (χ21 = 5.916; P = .015) and that between the specialty care and disease-specific networks (χ21 = 4.353; P = .037).
Qualitative Data From Telephone Interviews
Sixteen of the 21 informants were PCRN leaders, with the remainder representing stakeholders or participants in the PCRN process. Dominant barrier-related themes included funding challenges (subthemes were funding of core activities, funding of individual studies, and cross-subsidization of PCRN expenses with local funds), data management (subthemes were data flow, informatics, and data standardization), governance (subthemes were standardization of protocol review, protocol implementation, management of expectations, review of external proposals), inter-network relationships (subtheme was competition between PCRNs), and ethical issues (subtheme was lack of local pediatric expertise). Solution-related themes included collaboration between networks to address challenges (subtheme was forum for sharing best practices), and human subjects collaborations (subtheme was central institutional review boards). Supplemental Table 5 contains quotations illustrating each theme and subtheme.
Assessment of Untapped Interest in Network Research
Ultimately, 78% of AAP section and council chairpersons completed this survey (N = 43). Questions addressed pediatric research generally and interest according to their members' participation in research and PCRNs specifically. Results suggest that a large proportion of respondents (44.2%) considered pediatric research to be inadequate. Respondents provided answers to several categories of questions, with gross estimates in 10-point categories as possible answer (eg, 0–10%, 11–20%, 21–30%, etc). Respondents provided median estimate of 21% to 30% for members who were participants in some type of research and median estimates of 21% to 30% for nonparticipants who might be interested in research. They suggested that a minority of their members currently were network members (median estimates: 0%–11%) but a significant proportion of the rest would be interested (median estimates: 21%–30%). An overwhelming 93% of respondents thought that the AAP could play an important role in enhancing pediatric clinical research capacity; among the possibilities for doing so, 69.8% thought that the AAP should make members aware of PCRNs in their clinical areas.
The large numbers of PCRNs span the gamut of care. Consistent with their history, specialty care networks have been in existence the longest and represent a plurality of all such networks. Disease-specific networks are relative newcomers. Each network type seems to have relative strengths and weaknesses. Primary care networks seem to launch more data collection activities but are weaker in developing articles, which may be attributable to their nonacademic tradition or possibly to a relative lack of protected time or resources. Irrespective of network type, there is some evidence that longevity may be associated with productivity. Longevity may be not only a measure of time but also a proxy measure for funding success.
Clearly, networks have varying degrees of funding, both infrastructure and project-specific support. Of particular concern was the fact that almost 30% of networks had no current project funding. The ability to secure continuous funding streams is clearly a key to networks' ability to endure. The sample interview quotations from network leaders support this concept (Supplemental Table 5). With respect to infrastructure funding, which was cited by several of the interviewees, the prevailing sentiment was that the essence of a network is its stability over time, and this stability becomes tenuous in the absence of project funding unless infrastructure funding exists. Equally important is maintenance of the infrastructure to demonstrate a network's vitality during the process of obtaining project funding.
Networks strongly supported the notion of a coalition. Respondents were strongly supportive of a cross-network coalition in their responses to the question, “Which of the following ways could your network benefit from being part of a coalition?” Collaboration on study proposals (in an increasingly competitive funding environment) and advocacy for networks (PCRNs find themselves to be the proverbial “square peg in the round hole”) were the 2 potential benefits cited most often, with collaboration being a particular focus of primary care groups. One “write-in” benefit that received multiple endorsements was sharing of best practices across networks. Although many other items on the list were intended to capture this implicitly, we think that the specific free-text identification of this issue by multiple respondents underscores its importance to networks. Of note, the separately conducted network leader interviews clearly supported the notion of coalitions as one potential solution to the challenges of PCRN research. Interestingly, although it is possible that a collaboration or coalition might have disadvantages, network leaders and stakeholders never mentioned the topic during many hours of interviews.
The survey of AAP section and council chairpersons provided some insight into the perceptions of AAP leaders regarding the topic of research capacity. Two messages seemed clear, namely, that there is untapped potential for adding clinicians interested in research to existing PCRNs and that there is resounding support for the role of the AAP in enhancing pediatric research capacity, including a greater role in promoting research networks.
This work has a number of limitations. First, the survey instrument analyses were constrained by small numbers, which limited the power to detect differences. We recognize that some adult-focused research networks also may conduct pediatric studies, and those networks were not included in this sample. Second, the samples we obtained, particularly the network profile sample, might have been biased toward more-successful networks, because such networks might have been more willing to share their success stories than less-successful networks. Third, the survey of sections and councils provided admittedly very approximate estimates regarding pediatric capacity. It is possible that the AAP section and council leaders were overestimating the interest in research on the part of their members. A more-direct survey of AAP members would yield better information, but costs of such an effort precluded that work. Also, only approximately one-half of the nation's board-certified pediatric subspecialists are members of the AAP. Last, the nature of the surveys precluded deeper investigation of some of the findings. For example, although networks seemed interested in the benefits of collaboration on study proposals, it is unclear in what areas or to what extent a cross-specialty proposal would be acceptable.
Nevertheless, these data provide the first detailed examination of PCRNs. Three conclusions seem warranted. First, networks exist across the continuum of care and, as indicated by survey and interview results, have different strengths and challenges. Second, significant numbers of nonparticipating practitioners across the care continuum may have an interest in joining networks. Last, a clear majority of networks cite multiple potential benefits that they think would follow from an affiliation or coalition of networks.
Results of this investigation suggest substantial potential for collaboration across the full range of PCRNs. To that end, we have begun to bring network leaders together to explore the possibilities for collaboration, at Pediatric Academic Societies special interest group meetings in 2009 and 2010 and through a series of National Institutes of Health-sponsored conferences, the first of which was held in November 2009. These meetings have initiated the sharing of best practices and the exploration of evolving issues, such as the impact of the National Institutes of Health Clinical and Translational Science Award program on PCRNs. Through these and other ongoing efforts (eg, the Agency for Healthcare Research and Quality practice-based network program), we expect to see PCRN researchers leverage their collective experience into another 50 years of substantive contributions to the knowledge base on child health.
Support for the study was provided by the AAP and the National Institutes of Health (grant 1R13EY019972-01).
We thank the individuals who agreed to be interviewed and/or surveyed and to share their thoughts for the study. We also thank Harry Pellman, MD, for serving as a subject in the pilot phase of interview development and the Pediatric Research in Office Settings steering committee for review.
- Accepted July 2, 2010.
- Address correspondence to Eric J. Slora, PhD, American Academy of Pediatrics, Division of Primary Care Research, 141 Northwest Point Blvd, Elk Grove Village, IL 60007. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
Funded by the National Institutes of Health (NIH).
- PCRN =
- pediatric clinical research network •
- AAP =
- American Academy of Pediatrics
- 1.↵National Institutes of Health. Inventory and evaluation of clinical research networks. Available at: http://www.clinicalresearchnetworks.org. Accessed April 2, 2008
- Schilsky RL,
- McIntyre OR,
- Holland JF,
- Frei E III.
- Christoffel KK,
- Binns H,
- Stockman JA III.,
- et al
- Wasserman RC,
- Slora EJ,
- Bocian AB,
- et al
- Hack M,
- Horbar JD,
- Malloy MH,
- Wright L,
- Tyson JE,
- Wright E
- 8.↵Investigators of the Vermont-Oxford Trials Network Database Project. The Vermont-Oxford Trials Network: very low birth weight outcomes for 1990. Pediatrics. 1993;91(3):540–545
- Harris DL
- 10.↵Agency for Healthcare Research and Quality. AHRQ support for primary care practice-based research networks (PBRNs). Available at: http://www.ahrq.gov/research/pbrn/pbrnfact.htm. Accessed December 2, 2009
- Copyright © 2010 by the American Academy of Pediatrics