Chapter  99:  Community-based Participatory Research: Assessing the Evidence

Community-based Participatory Research: Defining the Approach

CBPR has been proposed as an approach that combines research methods and community capacity-building strategies to bridge the gap between knowledge produced through research and translation of this research into interventions and policies.^{2, 7, 9–12, 17–20}

CBPR's distinction from other community-based research approaches, which view “community” as a setting or location, is the recognition of community as a social entity with a sense of identity and shared fate. Working with rather than in communities, CBPR attempts to strengthen a community's problem-solving capacity through collective engagement in the research process. The seminal review of community-based research literature by Israel and colleagues¹¹ defines CBPR as “[a] collaborative approach to research that equitably involves, for example, community members, organizational representatives, and researchers in all aspects of the research process. The partners contribute unique strengths and shared responsibilities to enhance understanding of a given phenomenon and the social and cultural dynamics of the community, and integrate the knowledge gained with action to improve the health and well-being of community members” (p. 177)

In their review of participatory research studies,^21–23 Green and colleagues offer the following definition:²³ “Participatory research is systematic inquiry, with the collaboration of those affected by the issue being studied, for purposes of education and taking action or effecting social change” (p. 194). Using their own findings, this Canadian group developed a set of criteria for evaluating research proposals²³ that we have adapted and propose to refine further to apply to articles in our evidence tables. Green and colleagues defined community²³ as “any group of individuals sharing a given interest; this definition includes cultural, social, political, health, and economic issues that may link together individuals who may or may not share a particular geographic association. This definition also includes the traditional concept of community as a geographic entity” (p. 186). Although many researchers and practitioners offer definitions and descriptions of community and CBPR, no clear consensus has emerged to move the field forward during a time when interest is growing rapidly.^24–30

Nevertheless, common themes are that the CBPR approach (a) recognizes the importance of social, political, cultural, and economic systems to health behaviors and outcomes; (b) engages community members in choosing research topics, developing projects, collecting data, and interpreting results; (c) emphasizes both qualitative and quantitative research methods; and (d) puts high priority on translation of the findings of basic, intervention, and applied research into changes in practice and policy. More difficult to prescribe, however, is the degree to which each of these criteria must be fulfilled to satisfy the elements of CBPR.

Community-based Participatory Research: Clarifying the Benefits

Done properly, CBPR should benefit community participants, practitioners, and researchers alike. CBPR creates bridges between scientists and communities, allowing both to gain in knowledge and experience.^31–35 This collaboration assists in developing culturally appropriate measurement instruments, thus making projects more effective and efficient.^{36, 37} Finally, CBPR establishes a level of trust that enhances both the quantity and the quality of data collected.^{31, 38–40} The ultimate benefit is the prospect of examining the community's own unique circumstances to test and adapt best practices to its own needs.^{2, 31, 33, 36, 41–47}

Assessing the Need

In November 2001, the Agency for Healthcare Research and Quality (AHRQ), in collaboration with several Federal agencies and the W.K. Kellogg Foundation, convened a 2-day conference “to promote and support the use of CBPR, to develop strategies to advance CBPR, and to explore the use of CBPR as a resource for policymakers to help guide their program development.”⁴⁸ AHRQ organized the meeting specifically to address three key barriers to CBPR: (1) poor community incentives and capacity to be partners in CBPR projects; (2) poor academic incentives and capacity for researchers to act as partners in CBPR projects; and (3) inadequate funding and insensitive funding mechanisms.

Conference participants, through working groups and extensive discussion, produced three sets of recommendations aimed at funders, community members, and academics. The information generated is to be used to “describe the current context or environment for CBPR, to develop strategies to promote CBPR, and to provide funding organizations with input from communities as they work together to improve the health and well-being of those in communities.”⁴⁸ Among the recommendations was a request that an AHRQ Evidence-based Practice Center (EPC) synthesize evidence on the conduct and evaluation of CBPR. A national group could then use such a review as the basis for CBPR guidelines with the following anticipated benefits: enhanced stature for CBPR; guidance to potential partners entering into CBPR projects; and improved assessment criteria and mechanism for funders to review CBPR proposals.

AHRQ awarded this evidence report to the RTI International-University of North Carolina Evidence-based Practice Center (RTI-UNC EPC). Our systematic review consolidates and analyzes the body of literature that has been produced to date on CBPR in several areas relating to the following key questions:

• What defines community-based participatory research?

• How has CBPR been implemented to date with regard to the quality of research methodology and community involvement?

• What is the evidence that CBPR efforts have resulted in intended outcomes?

• What criteria and processes should be used for review of CBPR in grant proposals?

Expected Audiences

The RTI-UNC EPC team anticipates that its report and subsequent publications will assist several audiences. Community leaders interested in initiating research projects will find guidance on expectations of what a true collaboration might look like, including their obligations as research partners. Public health and health services researchers and practitioners new to CBPR will gain insights into their obligations as partners with communities in research. Funders in both Federal and foundation arenas will find criteria that they can use to evaluate CBPR proposals.

Organization of the Report

Chapter 2 describes our methods, including key questions and analytic framework, our search strategies and inclusion/exclusion criteria, and our approach to grading the quality of articles and rating the strength of evidence. In Chapter 3, we present the results of our literature search and synthesis of retained articles. Chapter 4 further discusses the findings and offers our recommendations for future research. Our references and included studies and a listing of excluded studies follow Chapter 4. Appendixes include a detailed description of our search strings (Appendix A), an example of our quality assessment form (Appendix B), detailed evidence tables (Appendix C), peer reviewers (Appendix D), and suggested guidelines for funders and applicants (Appendix E).

Inclusion and Exclusion Criteria

Based on the final key questions specified following the expert meeting and further discussions with our TEAG, we generated a list of inclusion and exclusion criteria for each key question. Generally, we included human studies; all ages and both sexes, English language only; and studies done in the United States and Canada (English-speaking North America). We included a broader set of international studies for purposes of describing the history and definition of CBPR, but systematically reviewing empirical studies conducted in vastly different sociocultural and political climates would have far exceeded the scope of this effort.

Exclusion criteria (apart from the obverse of the above) included editorials, letters, and commentaries; articles that did not report information related to the key questions; and studies that did not provide sufficient information to be abstractable. We identified several manuscripts that were limited to descriptions of CBPR processes and partnership development that did not include sufficient information on projects or outcomes; we also excluded these studies from our review.

On the advice of our TEAG and based on our cumulative definition of CBPR, we elected to limit our review to studies that defined community at the level of study participants; thus, we excluded studies that used participatory techniques to involve health professionals in the research process. For example, an extensive body of research in the literature addresses participatory action research as a method to include and empower nurse professionals in continuing education and career development.^{49, 50} Likewise, many studies involve physicians and other health care professionals in the process of identifying barriers to health care delivery and testing intervention approaches to address these barriers.^{51, 52} Although these types of investigations represent an important approach to involving those who can both improve the research process and enhance the potential for implementing findings, we elected to narrow our review to participatory research involving primarily community members, worksite employees, and other individuals not involved with the health care delivery process.

We did not restrict the search by date of publication. The last of our systematic searches was conducted on March 3, 2003. After that date, we continued to search for citations that were necessary to provide a complete overview of studies that we had already identified through our systematic searches and TEAG suggestions. We performed these latter searches on individual author names or study names (or both), mainly during the process of data abstraction. We were also able to obtain advance copies of articles to be published in a special issue of the Journal of General Internal Medicine focusing on CBPR, which appeared in July 2003.

Relevant Data Sources

For KQ 1, 2, and 3, we used three strategies to include all the current valid research related to the key questions: systematic searches based on search terms and author names, consultation with the TEAG, and hand searches of reference lists. First, we searched standard electronic databases such as MEDLINE®, Cochrane Collaboration resources, PsycInfo, and Sociofile using specified search terms. Based on the inclusion/exclusion criteria above and the additional key terms identified by our expert meeting attendees, we generated a list of Medical Subject Heading (MeSH) search terms (Table 3). The TEAG reviewed these terms to ensure that we were not missing any critical areas and suggested additional searches on specific authors and studies. We included these names in our systematic search strategy below. This list represents our collective decisions on the MeSH terms to use for all searches.

Second, we consulted with the TEAG about any studies that were under way but not yet published. Key among the sources of information identified through the TEAG was the special CBPR issue of the Journal of General Internal Medicine (July 2003). This publication date was relatively late in our abstraction process, so we were concerned that we would miss this important source of literature. Fortunately, we were able to obtain and abstract data from these journal articles before they were published.

Third, we conducted hand searches of the reference lists of relevant articles to ensure that we did not miss any relevant studies that we had not identified through our MeSH terms. In conducting systematic reviews, we often find it necessary to pull additional articles to gain full information about a particular study. The CBPR literature represents an extreme case of this situation.

Because CBPR work requires long-term and deliberate collaborations before, during, and perhaps after a research project, this process often results in numerous articles through which the investigators describe their methods and results. This phenomenon is exacerbated by journal limitations on length of submissions, which tends to promote fragmentation of the work into multiple articles. Our original search terms often did not capture these additional citations because the authors do not specifically use CBPR or related terminology in describing their efforts. Moreover, in some cases, we determined that we missed relevant (sets of) articles because they simply had never been categorized or indexed as relating to CBPR at all, evidently because the investigators did not refer to their CBPR methodology. We were able to identify them only from review articles relating to CBPR. The review articles were especially important because they often included extensive, completed, often well-funded projects that covered a wide array of CBPR elements of the type we needed to examine in this evidence report (e.g., those of the Urban Research Centers).

For KQ 4, we compiled any peer-reviewed publications that could contribute to the research questions. Very few articles directly addressed CBPR funding issues per se;⁵³ rather, the materials we found tended to describe funding mechanisms for CBPR, such as Urban Research Centers funded by the Centers for Disease Control and Prevention (CDC)^54–58 and the Environmental Justice funding mechanism of the National Institute of Environmental Health Sciences (NIEHS).⁵⁹ We also reviewed the Web sites for several funding agencies supporting CBPR, talked with Federal staff involved with the Interagency Working Group for Community-based Participatory Research,⁶⁰ and interviewed individuals at the CDC and National Institutes of Health (NIH) who were involved with developing CBPR Requests for Applications (RFAs) and the grant review process more generally.

Literature Search Results

Across the four key questions, we identified a total of 650 abstracts for review through our systematic searches. We identified an additional 599 abstracts by using names and search phrases suggested by our expert meeting attendees and TEAG. While reviewing these abstracts, we identified 159 additional citations through hand searches that we considered necessary to decide whether the study qualified for inclusion in our review. Finally, we retained and pulled 298 articles for complete review and excluded 113 studies.

A common reason for exclusion was that the study was a review article listing several CBPR studies, with insufficient information on any individual study to be included in an evidence table. Another frequent reason for exclusion was that, on review, the study did not have sufficient elements of community involvement and/or research to be considered CBPR. Other reasons for exclusion included lack of relevance to the topic (for instance, not health related), or unabstractable information (as with process evaluations that focus on participatory processes with no details on research collaborations) [see: list of Excluded Articles, page 107].

Ultimately, we retained 55 articles for KQ 1; we were unable to obtain three identified articles through interlibrary loan requests or Web searches. For KQ 2 and 3, we identified 123 articles that constituted 60 studies. For KQ 4, we used 7 articles to inform the results and discussion.

Of the 123 articles identified for KQ 2 and 3, a sizable proportion (55 articles or 45%) were identified through hand searches. A key limitation of employing secondary and tertiary sources to identify CBPR studies is that these studies are often not self-identified as CBPR. Although a separate review article may have mentioned elements of their participatory approach, the authors may not have intended to conduct a full-fledged CBPR study. For these studies, evaluation against elements of a CBPR scale is perhaps unfair and creates unnecessary inconsistencies among the pool of included studies.

Therefore, we chose to limit our reliance on hand searches by considering citations relevant only to the intervention mentioned in the article originally obtained through our systematic searches. For instance, in the case of the Health is Gold! study, several other interventions had been conducted as well, but we chose to limit review of these citations to the intervention identified in the July 2003 issue of the Journal of General Internal Medicine. Using this strategy prevented an exponential expansion in our scope of work while still allowing us to capture a larger pool of studies. In addition, it brought some degree of consistency to the studies included in the final analysis, in that all the studies were identified by CBPR or related key words. As a consequence of this strategy, however, we cannot claim this review to be exhaustive.

An additional limitation of this review is that it necessarily depends on results having been reported in peer-reviewed publications. Articles that focus on process evaluation may not provide any details on study design and methodology. Conversely, articles focusing on study outcomes may choose either not to report the CBPR process or to report it only partially, depending on the focus of the journal article and limitations on length. Furthermore, no clearly established standards for reporting CBPR elements exist. Given the great variability of reporting, we are able to provide only information on whether these elements were reported; their absence cannot be taken as proof that the study did not incorporate these elements. By the same token, the relative absence of negative findings in this report is likely to be attributable to a form of publication bias, in which unsuccessful collaborations are rarely reported.

Designing Abstraction Procedures

We employed our analytic framework and feedback from the expert meeting and TEAG to guide development of our abstraction tables (see Appendix B ^*), which we designed to approximate the final evidence tables as closely as possible. We also used the framework and feedback to guide the quality rating system (described below). We divided both the abstraction tables and quality ratings into primary research and primary community-based participatory elements. In this way, we were able to describe the studies more fully and evaluate the research and community participation elements separately rather than forcing community participation elements into research methodology categories.

For KQ 2 and 3, because of the multiplicity of articles from a single study, the first step in data collection required grouping articles by study. The Study Director reviewed all articles marked for inclusion and grouped them by study and then sent all articles relating to a single study to our abstractors. Abstractors sometimes identified additional articles necessary to complete the evidence table, and they also recommended articles for exclusion. The abstracts also determined whether the group of articles related to multiple interventions (listed under the same study name) and, if so, forwarded queries to the senior reviewer to select the relevant intervention for abstraction. Once we had compiled a complete set of articles pertaining to a single study, the abstractors keyed the data into an evidence table. The senior reviewer paired with the abstractor performed quality control assessments by reviewing each of the evidence tables against the original articles and making revisions where needed.

Training Abstractors

All abstractors attended two training sessions. At the first session, we explained the process and goals of data abstraction; we then sent the abstractors home with an article to review. We reconvened the group and, through a review of the test article, ensured that the abstractors understood what was expected of them. At that time, we determined that the abstractors were able to abstract the data as required and began the data abstraction process. The Research Coordinator monitored progress and routed the data abstractors' questions or issues to the Study or Co-Scientific Directors.

Developing Data Abstraction Forms

For KQ 1, one of the Scientific Directors (EE) took sole responsibility for generating a data abstraction form, and it formed the basis for the respective evidence table. For KQ 2 and 3, the Study Director (MV) and the Co-Scientific Directors (EE, AA) together created a single form that served as a data abstraction form as well as the template for the respective evidence tables. We revised and refined the form through multiple rounds of pretesting on different articles spanning the entire range of interventions to ensure that it would adequately capture all relevant issues. We solicited feedback from the data abstractors during training to refine further these various forms.

Developing Evidence Tables and Preparing the Draft Evidence Report

The two final evidence tables are found in their entirety in Appendix C.^* The first covers evaluated interventions and the second interventions either not completed or not evaluated. Entries are sorted by study design and then listed alphabetically by their study names. When articles gave no “official” study names, we used the key focus of the study. Entries in the evidence table may combine information from multiple articles to provide more complete information on a given study. A list of abbreviations used in the tables appears at the beginning of the appendix.

Grading the Quality of Individual Articles and Rating the Strength of the Evidence

We also developed forms to guide our evaluations of the quality of individual articles in this literature and the degree to which investigators had implemented CBPR principles in their research. Specifically, we developed two quality rating forms: one related to research quality that drew on previous work of the RTI-UNC EPC^{61, 62} and the other rated the quality of collaboration with a community.

CBPR reflects significant diversity in outcomes, research methodology, and measures.Thus, we elected to grade the quality of only two types of studies (often represented by a set of published articles): (1) those that represented a completed intervention study and (2) those that represented an observational study that was not limited to a baseline needs assessment, but rather was designed to allow extrapolation to a broader population. While this limits the scope of the research graded for quality, it allows application of a consistent set of research criteria.

We tested several drafts of our quality grading instruments and revised them numerous times to assure that they captured the desired information. The final grading forms can be found in Appendix B. Research elements of intervention studies were grouped into the following nine categories: (1) the research question, (2) study population and external validity, (3) control/comparison group, (4) intervention, (5) internal validity and intervention fidelity, (6) primary outcome measures, (7) statistical analysis, (8) blinding, and (9) funding source. CBPR elements rated included the following 10 dimensions: (1) selection of research question, (2) proposal development, (3) financial responsibility for grant funds, (4) study design, (5) recruitment and retention, (6) measurement instruments and data collection, (7) intervention development, implementation, (8) interpretation of findings, (9) dissemination of findings, and (10) application of findings to health concern identified.

One key element of quality grading involves whether the articles or investigators at least disclosed their funding sources, because of the potential for bias associated with the funding source, whether private or public.⁶² (An example might be funding from the Dairy Council for a CBPR study promoting milk consumption.) We did not directly include information about funding source in our quality grading scheme, because of the dissimilarity between this element (on the one hand) and items drawn from epidemiology or validated methods research (on the other). In the final evidence report, evidence tables record either the actual funding source or the fact that the investigators did not supply the information in their published articles.

Two senior investigators completed study quality assessments by rating the studies separately, comparing the scores, discussing any discrepancies until these were resolved, and assigning a single score. We assigned a score of “1i” for insufficient information, “1p” for poor, “2” for fair, and “3” for good.

KQ 1: Definition of Community-based Participatory Research

In exploring this topic, we sought to answer three important questions:

• What are the essential elements of CBPR?

• What are the “best practices” of CBPR, including the characteristics of successful investigator-community partnerships?

• What are the major expected outcomes from both the research and community perspectives?

We identified 58 peer-reviewed articles that were conceptual in orientation; that is, they synthesized the evolution of, values for, or lessons learned from collaborative research. All articles used CBPR or similar terms, such as action research, collaborative community action research, community-centered praxis, participatory action research, participatory evaluation, and participatory research. Of these 58, we were able to retrieve and review 55 articles; three were not retrievable through interlibrary loan requests or Web site searches by the time we prepared this report. Our review of the abstracts of these three articles suggests that their acquisition would not materially change our results. The articles came from the fields of anthropology, community development, community psychology, disability research, environmental health, health education, health sociology, injury research, mental health, nursing, organization development, patient care, and reproductive health.

We used three reviewers to abstract content from these 55 articles, using a matrix of 28 cells, representing specific CBPR domains in which to enter abstracted verbatim text. The matrix appears in Appendix B. The 28 domains were named as essential elements of participation; essential elements of research; best practices; and expected outcomes for seven components of research (identification of issues and concerns; study design and funding; participant recruitment and retention; measures and data collection; intervention design and implementation; data analysis, interpretation, and dissemination; and partnership structure). One of the Scientific Co-Directors (EE) reread the 55 articles to verify the verbatim text entered onto each cell of the matrix, read through the text entered for each domain, and then summarized the meaning of abstracted text as themes.

KQs 2 and 3: Intervention Studies and Outcomes

As expected, we found a striking degree of variability in the study designs, substantive concerns, and scope of community involvement of CBPR studies. The extent to which these elements were reported in the published literature varied appreciably as well. We looked to the key questions to help us organize this assortment of studies and to decide whether the CBPR studies had achieved their intended outcomes. Specifically, we considered (a) whether the study had an explicitly intended outcome resulting from a planned intervention and (b) whether the outcome was evaluated in sufficient detail in the published literature available to us.

We defined an intervention as an organized and planned effort to change behavior among individuals, communities' norms or practices, organizational structure or policies, or environmental conditions. Our overriding principle was consistency; we used a definition of interventions that would have a similar meaning across different studies. As an example, although some studies using a participatory action research approach viewed participation in the study as the intervention or the means to achieve their goal of empowerment, we did not classify these studies as having an intervention. We did not restrict interventions to those involving the research community; we included evaluations of studies in which the intervention occurred before researchers became extensively involved in the process. In addressing the evaluation of the intervention, we considered whether the intervention was reported as completed and whether it had been evaluated in a manner that allowed us to make conclusions about whether the intended outcomes had been achieved.

Of the 60 studies relevant to KQs 2 and 3, 30 studies listed interventions and 30 were noninterventional studies (see Table 4 for a list of study names, abbreviations and citations, Table 5 for a summary of characteristics). Evidence Table 1 (Appendix C) comprises 12 of the 30 interventional studies that reported the intervention as complete and evaluated it in a manner that allowed us to assess whether intended outcomes had been achieved. In judging an intervention to be complete (as opposed to ongoing), we considered only whether the intervention had been evaluated; we did not consider whether the intervention was implemented to a lesser degree or in a manner that was different than the intention. Evidence Table 2 (Appendix C) consists of the remaining 18 interventional studies that reported an ongoing intervention (for which we could not find any later citations through our additional searches) and studies with completed interventions that were not fully evaluated (Table 6 presents summary results).

We did not attempt to create an evidence table for the 30 studies that had no interventions. CBPR studies may often focus on basic research questions, initially, without an intervention but with a commitment to disseminating and translating results into interventions and policy. While there is much to be learned about the CBPR approach from these studies, the 30 studies without interventions varied in the extent to which information was abstractable; we present summary information in Table 7.

KQ 4: Funding Criteria for Community-based Participatory Research

Based on our discussions with the TEAG and AHRQ, we understood our task for KQ 4 to be primarily one of synthesizing our findings from the evidence review for the purpose of guiding future funding applicants (proposal writers), reviewers, and agencies toward submitting and funding the best possible CBPR. To this end, we used the findings for KQs 2 and 3 to identify the strengths and weaknesses of currently funded CBPR and highlight some of the challenges that CBPR researchers face. As noted earlier, we also reviewed articles identified from the literature that addressed existing funding mechanisms specifically focusing on CBPR.

Some articles described broader challenges faced by CBPR researchers and the benefits that may accrue from such research to both communities and investigators.^{2, 11, 53} Other articles addressed future research and funding priorities that included CBPR, such as those for the National Institute on Disability and Rehabilitation Research (NIDRR),⁶³ or the challenges of securing funding to sustain CBPR efforts.⁶⁴

We also reviewed Web sites and talked with individuals in Federal agencies about issues of generating requests for applications (RFAs) for grants and of reviewing and funding CBPR proposals. We focused the Web search and discussions primarily on agencies and their study (review) sections associated with translational research, which we thought to be the most likely recipients of CBPR submissions. These include translational grants sections of the National Institute for Diabetes, Digestive, and Kidney Diseases (NIDDK), National Institute of Environmental Health Sciences (NIEHS) and the Demonstration and Education section (R18) for the National Heart, Lung and Blood Institute. With the Centers for Disease Control and Prevention (CDC) we reviewed and discussed the recently funded RFA “Community-based Participatory Prevention Research,” in 2002 and 2003, 26 grants were funded under this mechanism so the number of CBPR manuscripts submitted and published should rise markedly by the end of this decade.

Finally, we learned more about the Interagency Working Group for Community-based Participatory Research initiated by NIEHS and established in February 2002.⁶⁰ This group was set up Dr. Olden, Director of NIEHS, inviting other agencies to join in the formation of the Interagency Working Group. The purpose of this group is “to strengthen communication among Federal agencies with an interest in supporting CBPR processes in the conduct of biomedical research, education, health care delivery, or policy.” As this group is still in a formative stage, its members expressed considerable interest in the results of this evidence review as a guide to their future efforts.

Overview

Through our synthesis of verbatim abstractions from 55 articles entered onto the matrix of CBPR domains, we derived a summative definition of CBPR. This deliberately short, workable definition guided our work; we believe that it can serve the purposes of AHRQ, sponsor of this evidence report, other Federal agencies that extensively support CBPR, and other interested parties and agencies.

CBPR is a collaborative research approach that is designed to ensure and establish structures for participation by communities affected by the issue being studied, representatives of organizations, and researchers in all aspects of the research process to improve health and well-being through taking action, including social change. To expand this definition, we conclude that CBPR emphasizes (1) co-learning about issues of concern and, within those, the issues that can be studied with CBPR methods and reciprocal transfer of expertise; (2) sharing of decisionmaking power; and (3) mutual ownership of the products and processes of research. The end result is incorporating the knowledge gained with taking action or effecting social change to improve the health and well-being of community members.

The following sections present the results from our systematic review of the literature in this area, which formed the basis for the definition. We emphasize the essential elements of community participation, the essential elements of research, and best practices for these types of investigations. Other key issues concern the outcomes expected from the perspectives of both the community and the investigators.

Essential Elements of Community Participation

According to all 55 articles we reviewed for this key question, participation in the products and process of research by people who experience the issue being studied is considered fundamental to CBPR. Their participation has been justified on the basis of enhanced knowledge production and as a human right. Community members have a right to participate in research because they

• are uniquely qualified and capable to investigate their lived experiences;^65–73

• should have the opportunity, as co-learners, to generate relevant knowledge and create critical awareness of collective self-reliance that are of immediate and direct benefit;^{11, 66, 74–81} and

• are entitled to own the means of knowledge production and to hold the status and roles of the researcher in relation to the participants.^{20, 53, 78, 82–88}

Moreover, participation by community members who experience the issue being studied can enhance the quality of the process and products of research by

• providing descriptions, rich in detail, of the local social context and real-world constraints (i.e., replicability), which will improve conceptual robustness and explanatory utility of a study's findings;^{78, 89, 90}

• Establishing congruence between the study and local reality (i.e., increasing face validity), particularly for defining the problem, adapting methodology to specific ecologies and contexts, and determining the nature of acceptable solutions;^{75, 78, 87, 89, 90} and

• Improving adequate response rates and minimizing attrition because the research question and data collection methods are likely to be context sensitive and culturally relevant (i.e., dependability).^{78, 88, 91}

Community members' participation in research is viewed as a necessary condition for the researcher and the researched to (a) redefine their relationship, (b) discover new understanding of the situation and their options, (c) make choices, (d) reduce frustration with past failed attempts, and thereby, (e) build their collective capacities to improve health and well-being of community members.^{11, 66, 72, 75–79, 85} As a necessary condition, participation in CBPR has been characterized as a concept with multiple dimensions, a process with several modes, and a core value of democracy.

Democratic systems of decisionmaking give a central place to participation in open discussion by guaranteeing public reasoning and deliberative interactions.^{67, 74} The values placed on participation are tolerance of different points of view, including agreeing to disagree, and the importance of learning from one another.⁸³ Knowledge development, therefore, is not value-free but rather is political in nature.^{67, 68, 70, 71, 74, 83, 86, 87, 92} That is, power accrues to those who are able to create knowledge and access systems of knowledge that name the problem, organize people and resources around the problem, and mobilize solutions.^{67, 83, 87} Hence, decentralization of power in research decisionmaking is necessary to ensure participation of people who have a stake in the process and products of research, regardless of their status or prior experience with conducting research.

Participation in research of community members affected by the issue being studied has also been defined as a planned and directed process, which can be a social process or a means for empowerment. As a social process, participation is based on theories of group formation and functioning to facilitate open dialogue on divergent views, accommodate conflict, and agree on structures for collaborative decisionmaking.^{11, 70, 83, 84, 93, 94} As a means for empowerment, the purpose of participation is to engage the research group in actively examining the reasons for and consequences from either formal or informal activities of investigation through discussion, whereby needs are identified, decisions are made, and mechanisms are established to improve community life, services, and/or resources.^{84, 95}

This group process has been described as gradually moving the group through different modes of participation.^{73, 79} Although not reflective of all CBPR approaches, the four modes of participation, originally conceptualized by Biggs,⁹⁶ are as follows:

• Contractual: Researchers contract for services (e.g., interviewing) or resources (e.g., time or property) from local people who agree to take part in the research, inquiry, or experiment.

• Consultative: Local people are asked for their opinions and advice before the intervention is designed.

• Collaborative: Researchers and local people work together on a study that is designed, initiated, and managed by researchers.

• Collegiate: Researchers and local people work together as colleagues, each with different skills to offer for mutual learning, to develop a system for independent research among local people.

This notion of gradually shifting control from researchers to local people is also reflected in the literature on participation in research as a process of empowerment. Townsend and colleagues⁸⁶ defined empowerment, in a participatory research context, as a process of learning to critique and transform individual feelings, thoughts, and actions, as well as those of the organizations of society, so that the power and resources of research can be shared equitably. Drawing from theories of adult learning (e.g., Freire)⁹⁷ and action theory (e.g., Habermas),⁹⁸ empowerment is understood as changing not only a participant's personal experience with the power of research but also the power exerted through policies and other forms of institutional control over research.^{68, 73, 78, 80, 81, 83, 85–87, 92, 99}

Nonetheless, local people's participation in research does not guarantee that power and resources will be shifted to them because research partnerships cannot be entirely horizontal.⁸⁴ That is, complete equity is constrained by community norms, institutional inertia, and internalized expectations that allow the more powerful participants, however well intentioned, to determine what level of participation at which stage of research is most valuable for whom.^{84, 99} When participants are conscious of how power is organized by the policies and institutions that govern research, the researchers and the researched are more likely to redefine the power relationship between them.^{11, 20, 70, 75, 76, 78, 83, 84, 93} Whereas, failure to reflect on and openly discuss how power dynamics vary at each stage of research can inhibit meaningful participation and result in a sense of powerlessness and cynicism, when the many tasks involved with research become burdensome or unfeasible, and when the results do not meet expectations.^{73, 84}

Suggestions to researchers for potential collaborators include

• professional staff at a workplace (such as medical practitioners, health and human service workers, and therapists);^{65–68, 75, 80, 84, 86, 100}

• representatives of local organizations or agencies (such as managers, supervisors, nonprofessional workers, and clients);^{11, 74, 81, 86, 94, 100–103} and

• members of a local community (such as citizens, residents of a neighborhood or hamlet, and members of community-based organizations).^{11, 56, 69, 73, 74, 79, 81, 87, 88, 99–101, 104}

Participants from one or all of these three categories can serve as researchers and research collaborators. The rationale is that research needs such collaborators for two additional reasons: (1) to gain entry into the world of the people who experience the issue being studied, and (2) to instill accountability and responsibility for what researchers learn to see.^{66, 67, 87} Researchers can maximize reciprocity for the construction and validation of instruments, findings, and conclusions by examining the multiple world views on the issue that collaborators provide.⁸⁷

Participatory research that is community based, such as CBPR, emphasizes enlarging the role and representation of communities as collaborators.^{11, 56, 59, 76, 83, 101} Community, as a collaborator, has been defined as a unit of identity, which is a social and cultural entity that can actively engage and influence its members in all aspects of the research process. Within any local area, people associate through multiple and overlapping networks with diverse linkages based on different interests.^{11, 20, 59, 90} This emphasis on community comes from the view that, for lay (nontechnical) people, their community holds the strongest potential for collective power to negotiate the production and use of knowledge with the institutions and systems that govern the research enterprise.^{11, 56, 59, 76, 83, 90, 100}

Hence, for our evidence report on CBPR, we reviewed studies that included among their collaborators any of the following types of groups: community-based organizations and their executive directors, community as a unit of identity, community residents, clients served by an organization, or nonprofessional workers at a worksite. Many of these studies also included professional and management staff of professional organizations as collaborators. Therefore, we excluded from our review studies that collaborated solely with professional and management staff of professional organizations.

Essential Elements and Best Practices for CBPR Research

The field of public health generally agrees that CBPR is a collaborative process and approach to research for learning about health and illness while contributing to the good health of a community with whom the research is being conducted.^{11, 56, 67–70, 72, 77, 78, 81, 86, 90, 92, 94, 99} However, disagreement arises about whether the stages of research and methods of inquiry of a collaborative approach are the same as those of conventional research^{69, 83} or distinctively different.^{84, 87, 92}

Nonetheless, consensus does exist on the distinguishing characteristics of a collaborative approach to research. The two core ideas are (1) the reciprocal co-learner relationship between the researcher and the researched^{20, 67–70, 76, 78, 85, 87} and (2) the immediate and direct benefit of using new knowledge for taking collective action and effecting social change.^{11, 65, 66, 71, 74, 81–83, 89, 90, 93, 94}

Establishing a reciprocal co-learner relationship is viewed as a systematically planned encounter between researchers and their community collaborators during each stage of research. In this, they (a) meet face-to-face to define their relationship, (b) enter into dialogue on the requirements for equalization of power in the processes and products of research, and (c) set, alongside each other, their respective legitimate knowledge and expertise for examining and addressing a particular issue.^{68, 76} The criteria for determining the quality of a reciprocal co-learner relationship, put forth by Badger,⁷⁵ are

• Reflexive validity: Recognizing and exploiting how researchers and a community's respective experiences, values, and actions have affected the research situation and interpretation of findings.

• Dialectical validity: Constant analysis and report of movement between theory, research, and practice by examining tensions, contradictions, and complexities of the research situation.

• Critical validity: Analyzing the process of change, intentions, actions, ethical implications, and consequences.

• Face validity: Subjective judgment of researchers and community that findings appear to fit reality.

Moreover, the three potential uses of research that may be of immediate and direct benefit to a community collaborator have been defined as conceptual, instrumental, and persuasive.⁸⁴ Conceptual uses of research aim to change the way people think about problems and their solutions. The experience of collaborating in research can help communities better understand change-related processes, such as the politics of information utilization in change efforts, or the social context in which definitions of the problem are created and revised.^{70, 71, 73, 83, 87, 89} Instrumental uses occur when the results dictate direct changes in existing programs or services.^{11, 59, 69, 72, 78, 81, 83, 89, 90, 92, 94, 99, 105} Persuasive uses of research gather sufficient evidence to support a particular position or to influence policy.^{53, 74, 80, 82, 83, 85–89, 93, 99}

Therefore, the essential research elements of a collaborative approach have been categorized below under its two distinctive characteristics: (1) the reciprocal co-learner relationship between researchers and communities, and (2) the immediate and direct use of new knowledge for taking collective action and effecting social change. With regard to “best practices” for each research element, which are derived through empirical testing, we report on recommended guidelines for operationalizing each element from our review of 55 articles that are conceptual rather than empirical.

Reciprocal Co-Learner Relationship. The first important element in this category holds that a structure or mechanism is created for shared decisionmaking between researchers and community. Examples from the literature include a community advisory board, technical advisor group, task force, planning committee, evaluation committee, coordinating committee, or steering committee.^{56, 57, 70, 78, 80, 81, 84, 89, 102, 106} Such decisionmaking bodies must develop and then operate under guiding principles for collaboration.^{56, 57, 78, 106} The rationale is that in any collaborative relationship, conflict and contradictions are not only inevitable, but in fact are necessary for moving forward with trust building, power dynamics, and accommodating conflict at every stage of the research.^{83, 93}

Another important element is that the study be designed to remove previous barriers to community participation in research. Some public health scholars and practitioners assert that minimal direct benefit accrues to communities that have given their time, resources, and good will to a study that has “pathologized” them.^{68, 73, 87} That is, when research pathologizes social problems, the common outcomes are individually focused solutions (as opposed to community-focused) controlled by noncommunity entities, thereby once again disenfranchising communities.^{73, 87}

To remove barriers to community participation in research, the following guidelines have been recommended:

• Offer educational experiences, such as “vision workshops,” for both researchers and communities to understand resources and strengths of local people; generate awareness of shared concern with the problems inhibiting social progress of a community; transfer new skills during the research process; and discuss the details of research methods and tensions of matching experimental designs with community action.^{70, 74, 76, 93, 107}

• Hold group meetings and structured interviews to ascertain concerns about research and discuss methodological options, given a community's resources.⁸²

• Hire local coordinators.¹⁰⁷

• Make written plans detailing types of expertise required at each stage of research.¹⁰²

• Create issue-specific operational mechanisms, such as ad hoc groups, for internal review of operations and measures of accountability.^{70, 102, 106}

• Appoint researchers as guardians of the data during the study, and assure guardianship to the community at the end of the study. However, the shared decisionmaking body is obligated to offer original researchers the opportunity to continue analysis before it offers data to new investigators, and the latter must agree to follow guiding principles of collaboration established by the research partners.⁷⁸

• Evaluate the collaborative processes involved throughout the cycle of problem analysis, intervention design, implementation, and institutionalization.⁹⁰

Immediate and Direct Use of New Knowledge. Several considerations arise in thinking about how new knowledge from CBPR work should be applied. First, socioeconomic determinants of health are assessed, addressed, or both. The purpose of assessing and addressing such determinants is to engage researchers and communities in examining how people's personal experiences with health disparities are linked to policies, social structures, and other forms of institutional control.⁸⁶ To assess socioeconomic determinants of health, experts have suggested two research strategies as best practices. One is for the study to take an ecological perspective on health so that it generates a holistic understanding of the power that systems exert on everyday life.^{11, 79} Another strategy is to conduct a power analysis that examines where there is systematic disadvantage, failure to advocate, or merit that is not being recognized or acknowledged.⁸³ The new knowledge can then be incorporated into the study's problem definition and development of a conceptual framework. The eventual design of a multilevel intervention would address, for example, training families to monitor and protect their homes from air pollutants. The intervention might also include organizing affected communities to present their findings to legislative bodies and advocate for changing policy that is biased toward locating polluting industry near rural communities that are often poor and home to people of color.⁸⁸

Second, the research team should be cognizant and respectful of community needs and priorities during the study's implementation. A high degree of cooperation and flexibility between researchers and communities can be achieved through the best practice of building regular “feedback loops” into the stages of research, one step at a time, and directly reflecting evidence from the previous step.^{66, 84, 94} To be flexible to community needs and priorities, movement through the stages of research is cyclical, repetitive, and iterative.^{11, 66, 94}

Feedback loops create forums for meaningful discussion between researchers and communities on significant community issues, which can also help overcome distrust.^{56, 75} A reflexive discussion is one in which researchers and their community collaborators acknowledge that their respective experiences, actions, and values have affected the situation and its interpretation.⁷⁵ For example, a community may see different uses for the data than what was originally planned. This issue could be addressed at the next scheduled feedback session with a committee specifically formed for this purpose to enhance the research team's flexibility in addressing unforeseen needs and priorities.⁸⁴ Moreover, the research team's flexibility will enable them to adjust to the pace at which a collaborative research approach can proceed with success.⁷⁴

Third, the study's duration and purpose contribute to capacity building among individual researchers and their institutions as well as among individual participants or their larger community. Researchers taking a collaborative research approach have an obligation to maintain a long-term relationship of trust in their dual role of researcher-educator, with the purpose of capacity building.^{11, 78} Four stages of building collective capacity have been recognized: (1) identifying common ground; (2) establishing self as a community player with an issue-based agenda; (3) working on a common project; and (4) working on a multiagency, multisector project.⁹² Through a collaborative research approach, capacities that can result include those related to formation of critical consciousness of their unrealized capabilities and potential, improvement of the lives of those involved in the study, and reformation of underlying political structures.⁸²

Fourth, formation of critical consciousness of their situation to find answers to unrealized capabilities and potential is another important element of use of new knowledge. Participants' sense of isolation or alienation is reduced by being engaged in systematic discussion and reflection during the study. By focusing on their community, the residents' awareness of their shared strengths and concerns is increased.^{82, 93}

Fifth, improvement of lives of those involved in the study means that residents' unique knowledge of what will work in their community is integrated into information sharing and problem solving during the study. Increasing participants' power to claim a larger share of decisionmaking for their community makes it more likely that findings can be applied to address the health and social issues raised as a result of the research. Community participants can increase control over their lives by nurturing community strengths and problem-solving abilities.^{11, 69, 78, 82, 93, 100}

Sixth, reforming underlying political structures is another key action. The ultimate goal of a collaborative research approach is to change social structures, dealing with institutional control and conflict.⁶⁶ The acts of creating knowledge and using it to communicate a community's perspective to policymakers are fundamentally about the right to speak.⁸² Although these steps may not guarantee shifting power to communities to decide on policy, a community's capacity to interact directly with policymakers is a necessary first step toward understanding and changing oppressive situations.^{73, 82, 88}

Finally, findings should be (1) used to address the original health concern, (2) disseminated and interpreted to participants, (3) applied to a health-related intervention or policy change, and (4) used to sustain research-related interventions by the community. When new knowledge is constructed from multiple perspectives and meanings, differences in interpretation of findings are inevitable and intellectual growth can occur.⁸⁴ Hence, community collaborators must remain fully involved with decisions on what, where, when, to whom, and how to disseminate findings, apply them toward an action, and sustain them.⁸² Products for dissemination include advocacy documents for relevant agencies and authorities, mass media reports, training manuals, and scientific papers and manuscripts.⁸⁴ To ensure full collaboration in co-authoring communications about findings, experts recommend developing dissemination guidelines.⁷⁸ Before submitting manuscripts or presenting at conferences, co-authors discuss findings with the study's shared decisionmaking body. Any collaborator who disagrees with the interpretation or method of dissemination is invited to submit an alternative interpretation as an addition to the main communication, albeit written or oral, to be submitted at the same time. No single collaborator has the power of veto.

Macleod offers the following recommendations for disseminating findings:⁸⁴

• Frame results to limit potential for blaming people for their problems.

• Communicate results openly, even when some stakeholders will not benefit.

• Establish and maintain credibility of persons who conducted the research.

• View feedback and dissemination as an on-going process of dialogue with stakeholders.

• Be aware of political considerations behind feedback from stakeholders.

• Stay as jargon-free as possible, even with well-trained audiences.

• Use oral presentations as a means for assessing the validity of findings.

• Develop a task force of community members to study any recommendations.

With regard to application of findings, we examined the three potential uses of research described earlier (i.e., conceptual, instrumental, and persuasive).⁸⁴ Conceptual application of the findings involves developing theory that is sensitive to a community's context and culturally relevant. Through understanding the social contexts in which findings are applied, the public health field can move toward developing better theories of the problem.^{70, 71, 73, 83, 87, 89} Instrumental application of the findings includes documenting the process by which the findings are used in designing interventions or effecting social changes that attempt to solve public health problems.^{2, 11, 59, 69, 72, 78, 81, 83, 89, 90, 92, 94, 99} When the application of findings begins and ends with the behaviors of individuals, however, it is not considered social change (i.e., persuasive).⁸⁷ Persuasive application of findings alters the structure, policies, and other forms of institutional control over a community or individual's health and well-being.^{53, 74, 80, 82, 83, 85–89, 93, 99}

We detected disagreement on how a collaborative research approach contributes to the sustainability of research-related interventions. Some conclude that a long-term commitment by all collaborators is necessary.^{11, 56, 69, 74, 78, 90} For others, however, achieving community autonomy or self-reliance is necessary for sustaining interventions that emerged from the study.^{79, 80}

Expected Outcomes from the Community and Research Perspectives

A few scholars note that outcomes from a collaborative research approach include those of a capacity-building intervention.^{78, 82} Inclusiveness of community residents in learning to integrate questioning with reflection—which is the power of research—enables them to challenge and increase the power of conceptualizing the problem, selecting methodology, defining goals and objectives, securing funding, training trainers, sampling and recruiting participants, constructing measures, conducting analysis, interpreting results, disseminating findings, and advocating for policy change.^{53, 56, 59, 69, 73, 77, 80, 81, 86, 88, 103, 108} Hence, because the data are “grounded” in the experiences of people living along the margins of health and well-being, the findings are more likely to lead to collective action for structural and personal change.^{71, 86, 94, 109–111}

At the same time, this grounding in a community's local context can increase the face validity of findings on disparities in health status and practices. Arguably, problem definition, measures, and acceptable solutions need to be connected to social determinants of health; in other words, broad political and economic processes that have drawn capital, people, services, and other resources from low-income, rural, and inner-city communities.^{75, 78, 87, 89, 90, 112} Moreover, by enabling the decisionmaking power of a community to determine with researchers the most context-sensitive and culturally relevant methodology, CBPR approaches can raise the dependability of findings for identifying priorities and possible solutions. The reason is that eligibility criteria, recruitment strategy, data collection methods, and analysis procedures will reflect indigenous mechanisms and structures for communicating information and opinions and exerting influence.^{78, 88, 91} Finally, including community collaborators can increase the replicability of findings on health improvements. Replication by others is more likely to follow from documenting the details of how behavioral and social change processes, which are conceptually robust and have explanatory utility, combine resources beyond a community with the competencies, influence, and other assets embedded in a community.^{78, 89, 90, 112}

Overview of CBPR Studies

To answer KQ 2, we drew from the 60 studies identified as CBPR. To be included, articles were required to use basic community participation methods and to include some element of data collection and analysis, be it quantitative or qualitative. This is not an exhaustive list of all CBPR studies ever published; we suspect several other studies may exist that we could not identify because of the limitations of MEDLINE indexing terms, the nature of this literature and the work it represents, and our systematic review methodology.

Many of the studies reviewed in this report comprise multiple citations; to allow for both readability and easy access to the complete list of citations, we provide the full study name, the abbreviated name by which we refer to the study in this review, and the complete list of citations in Table 4. When we cite the study for the first time in the text, we cite all references; thereafter, we use the abbreviated study name. Table 5 summarizes the numbers of these 60 studies with certain characteristics related to populations, clinical or social topics, and similar matters. These points are discussed in more detail in the following sections.

Number and Time Between Publications. We found an average of two publications per study: 35 studies published only one article, but the remaining 25 studies produced, on average, 3.5 articles. This suggests a skewed distribution, with some studies generating multiple publications over a period of several years. By design, some CBPR studies include both a focus on an intervention and an evaluation of the intervention. The complexity of CBPR collaborations combined with journal restrictions on the length of the article are likely to contribute to the multiplicity of articles in these instances.

Also, CBPR collaborations may take longer, in general terms, than some other types of research and, thus, more time to publish results. Not counting the East Baltimore Health Promotion Study, which spanned 17 years between the first publication and the last, the 24 studies with more than one publication took about 2.5 years from the first publication to the last. Because several of these studies were not completed as of late 2003, we believe that our findings likely understate both the average number of articles generated by a study and the average length of time taken to publish the results.

Period of Research and Publication. The number of CBPR studies has increased sharply in recent years, especially since 2000, and the trend is likely to continue. This phenomenon may be attributable to several critical incidents. With the launching in 1998 of the Federal Department of Health and Human Services' Initiative to Eliminate Racial and Ethnic Disparities in Health by the Year 2010, national attention has generated an environment for innovation in public health research and practice for achieving the Healthy People 2010 objectives. Public and private funding institutions have been sponsoring special funding mechanisms, which explicitly require proposed studies to take a CBPR approach. A Federal Interagency Committee has been formed to advance the use of CBPR; it involves the National Institutes of Health (NIH), CDC, AHRQ, Department of Agriculture, Housing and Urban Development, Federal Highway Administration, Agency for Toxic Substances and Disease Registry, National Science Foundation, and Environmental Protection Agency.⁶⁰ Most recently, the 2003 Institute of Medicine Report, Who Will Keep the Public Healthy? Educating the Public Health Professionals for the 21 ^st Century, identifies the use of CBPR as one of eight areas of critical importance in which all public health professionals need to be trained.

Substantive Health Concerns. Several studies took a broad approach to defining health, and these studies constituted the largest group in this literature base. Among studies that took a narrower focus, environmental health was the leading concern because of NIEHS' long-standing interest in CBPR.

Communities of Interest. The definition of community typically included elements of both sociodemographic characteristics and location. Of these 60 studies, the highest proportion of studies (24 studies, or 40 percent) defined their community primarily along racial and ethnic lines, followed by health concerns (18, or 30 percent), location (12, or 20 percent) and occupation (5, or 8 percent).

Among the 24 studies that defined community primarily through race and ethnicity, eight focused on Native Americans, five each on African-Americans, Latino and Asian populations, and one on multiple ethnic groups. Of note, studies conducted with communities of color concentrated on those of low socioeconomic status, using a combination of indicators, such as level of education completed, median family income, health insurance coverage, enrollment in entitlement programs, or English language skills.

Funding. We were able to determine funding sources for 55 of the 60 studies. The majority (53 percent) of these studies reported a single funding source, but a significant minority (33 percent) mentioned at least two funding sources. Several studies were funded by a few key CBPR funding mechanisms. They include the Urban Research Centers, previously funded by the CDC and Environmental Justice and Community-based Participatory Research in Environmental Health of the NIEHS.

A total of 75 funding sources could be classified as Federal or national funding, state funding, foundation or private funding, or university funding. Government agencies at the national level were the predominant source of support; of these, NIEHS and CDC were the two most commonly named funders. Foundations or private sources of funding such as the Kellogg Foundation and the Robert Wood Johnson Foundation also played a significant (albeit smaller) role in supporting CBPR, followed by state agencies such as local departments of health and universities.

Implementation of CBPR: Research Methodology

We were best able to evaluate research methodology by distinguishing among three categories of studies. Of 60 studies, 30 were completed interventions or ongoing interventions; of these, 12 evaluated the intervention and 18 had either not completed the intervention or not evaluated it fully. The remaining 30 studies either did not have an intervention or did not report one. To assess fairly the actual study design, measurement, and data collection and analysis across studies, we considered it necessary to separate studies that implemented and evaluated planned interventions from those that were nonintervention. Noninterventional studies inevitably have different study aims and reporting standards than interventional studies. Similarly, we thought it necessary to distinguish those studies that had completed and fully reported the results of their interventions from those that had not. The following three subsections describe these separate bodies of literature. Tables 6, 7, and 8 present study design and data collection methods for the studies in the completed intervention, not completed or fully evaluated, and noninterventional groups, respectively.

Studies That Implemented and Evaluated Interventions. Table 6 lists the 12 studies that completed evaluated interventions. They are listed by study design and then alphabetically. Although these research teams used several study designs to evaluate interventions, experimental and quasi-experimental designs were used more frequently than nonexperimental methods. Table 6 provides citations, study design, intervention and key results. In addition, it gives two quality grades, one for research design and one for elements of community-based participation. Quality grades could range from 1 to 3, with higher scores reflecting better studies.

Of the 12 studies in this category, 4 were randomized controlled trials (RCTs); they include Communities Mobilizing for Change on Alcohol or CMCA,^113–118 East Baltimore Health Promotion;^119–122 Health is Gold,¹²³ and the Sierra Stanford Partnership.^{124, 125} Five of the 12 were quasi-experimental studies; these include HIV Testing and Counseling for Latina Women;^126–130 Internet Access and Empowerment;¹³¹ the Korean Study Breast and Cervical Cancer Screening Intervention;^{132, 133, 133, 134} the Okanagan Diabetes Project,¹³⁵ and the Wai'anae Cancer Research Project.^136–138 Studies with nonexperimental designs include the New York Immunization Project¹³⁹ and the Stress and Wellness Project,^{32, 140–143} and Women Dedicated to Demolishing Denial: HIV Risk Reduction for Lesbians and Bisexual Women.^{144, 145} One of three nonexperimental studies was a one-group pretest and posttest study (NY Immunization); another was a nonexperimental design with data collection throughout the period of the intervention, (Women and HIV Denial); and the third was initiated with a nonexperimental design (Stress and Wellness), but because of changes in operations at the study site, it eventually became a natural experiment comparing two sites, with pretest and posttest data.

The predominant data collection method was quantitative. Five studies used a combination of qualitative and quantitative data collection methods (HIV Latina, Internet Access, Okanagan, Wai'anae, and Stress and Wellness); and one used only qualitative methods (Women and HIV Denial). Two studies mentioned blinded data collection (Sierra Stanford and Stress and Wellness).

Two studies reported that they changed their measures, based on input from community members, to be more culturally relevant (Wai'anae and Korean Study). Three other studies mentioned that they applied instruments that had been previously used in the literature (Internet Access, Stress and Wellness, and Sierra Stanford), but it is unclear whether these were previously validated instruments.

All the studies in this category reported multiple primary variables and outcomes. All but one (Internet Access) assessed socioeconomic determinants of health.

All studies using experimental, quasi-experimental, and one-group pretest and posttest designs reported the statistical significance of their findings. Of the five studies that used qualitative data either alone or in combination with quantitative methods (HIV Latina, Internet Access, Okanagan, Stress and Wellness, and Wai'anae), two (Stress and Wellness and Wai'anae) mentioned that community members checked results as a way of verifying the findings with participants. Four studies used a triangulation of data sources (such as medical records, surveys of multiple interest groups and media records) to validate their conclusions (CMCA, East Baltimore, Stress and Wellness, and Okanagan).

Interventions Either Not Completed or Not Fully Evaluated. In the absence of clear information on implemented study design, we classified these studies based on the intended study design. This group of studies (see Table 7) illustrates the long-term nature of much CBPR work and the fact that many studies require several publications issued over several years to report the full findings of the project. Of the 18 ongoing interventions, 4 were part of ongoing experimental designs (Community Action Against Asthma,^146–148 PRAISE!,^{149, 150} Seattle King County Healthy Homes Project¹⁵¹ and Seattle King County Vaccines¹⁵²); 1 was intended to be a quasi-experimental design (TEAL¹⁵³); and 13 were nonexperimental designs (Elderly in Need,^{92, 154} East Side Village Health Worker Partnership,^{106, 112, 155–163} Haida Gwaii Diabetes Project,⁷⁷ Healthy Homes, Healthy Child,^{100, 164, 165} Kahnawake,^{78, 166–169} La Vida,¹⁷⁰ Mom Empowerment, Too!,¹⁷¹ the Nuclear Risk Management for Native Communities Project,⁷⁰ Preventing Agricultural, Chemical Exposure in North Carolina Farmworkers (PACE),^{172, 173} The Partners for Improved Nutrition and Health Project (PINAH),¹⁷⁴ Preventing Halloween Arson,¹⁷⁵ Survival Guide,^{176, 177} and Women and Heart Disease.¹⁷⁸ Table 7 provides a list of citations, study designs and the intended intervention for these studies. Two of the 13 studies with nonexperimental designs discussed plans for later RCTs to test the effectiveness of the interventions (Survival Guide and PACE).

These 18 investigations published findings from baseline data, formative work, and process data. Among this group of studies, information was generally not sufficient to determine whether they had implemented the intervention as intended, which is an issue of research fidelity. These data are more commonly reported when final outcomes data are presented, so this information gap may be expected to be addressed for some of these studies in the future.

Compared to the fully evaluated interventions, a similar portion of these studies used a combination of qualitative and quantitative methods (39 percent for incomplete interventions, compared to 42 percent for fully evaluated interventions). Many of these projects are ongoing studies and have not yet reported their final outcomes data. On average, the first publication from these studies appeared in the peer-reviewed literature 4.5 years ago, compared to 9 years ago for completed interventions.

Four studies reported that the community reviewed and revised their instruments and concepts (ESVHWP, PRAISE, Seattle Homes, and Survival Guide). Although several studies reported using previously developed instruments, the information was insufficient in most cases to determine whether the instruments had been previously validated. Eight studies reported their intent to use multiple sources of information, including archival records, surveys and focus groups of multiple interest groups, environmental assessments, and clinical data from blood sample and pulmonary function tests (CAAA, PRAISE, TEAL, ESVHWP, Healthy Home, Kahnawake, Preventing Arson, and Survival Guide).

Although no study presented sufficient data to qualify as fully evaluated interventions, 11 studies provided information on findings from analysis of psychosocial data, process evaluation, the research process, or more descriptive aspects of the intervention (CAAA, PRAISE, Seattle Vaccines, Elderly in Need, La Vida, Kahnawake, ME2, PINAH, Preventing Arson, Survival Guide, and Women and Heart Disease).

The Halloween Arson study represents an unusual case in that the intervention was conducted (in response to ongoing violence in Detroit around the Halloween period) by a coalition of community members and organizations without any input from researchers or an evaluation plan. Later, researchers in the Urban Research Center at the University of Michigan retrospectively evaluated the intervention in collaboration with community members. In many research efforts using traditional non-CBPR methods, the community is not likely to be involved in designing the intervention. In this case, however, the researchers were not involved in intervention design but were later called in to use a retrospective research method and analysis strategy.

Noninterventional Studies. Table 8 provides key information on the 30 studies we reviewed that had no clear intervention either implemented or planned. The table provides citations, study design, and objective for these studies. Of these 30 noninterventional studies, 27 were nonexperimental and primarily exploratory in nature. The other three were observational studies that were designed to permit extrapolation to individuals beyond the study population (African Americans Building a Legacy of Health,¹⁷⁹ Hospice Access and Use by African-Americans,¹⁸⁰ and Oregon Migrant Farm Workers^{181, 182}). Although these studies are classified as noninterventional for the purposes of this review, these studies may have resulted in the implementation of an intervention as a result of the findings. Several of the studies in this category resulted in significant policy change in either civic or private institutions. For the purposes of this report, these studies are considered to be noninterventional because they were not designed with an explicit intervention, nor did they undertake the evaluation of any intervention that might have resulted from their findings. Because the 30 studies without interventions were varied in the extent to which information was abstractable, we do not present detailed evidence tables; summary information is provided in Table 8.

The purpose of these studies varied and several had multiple objectives. We classified studies according to what appeared to be their primary objective in the literature available to us. More than half the studies (16 of 30) were predominantly concerned with understanding the problem at hand. Of these 16 studies, 2 focused on identifying health problems (Poultry Slaughterhouse Study¹⁸³ and¹⁸⁴ HERE¹⁸⁵); 8 were explorations of health-related knowledge, attitudes and practices (James Bay Cree Diabetes,⁷⁶ TAS Together for Agricultural Safety Project,¹⁸⁶ Perspectives of Pregnant and Postpartum Latino Women on Diabetes, Physical Activity, and Health,¹⁸⁷ The Native Hawaiian Smokers Survey,¹⁸⁸ Controlling Pesticide Exposure to Children of Farmworkers,¹⁸⁹ Hospice Access and Use by African-Americans,¹⁸⁰ Diabetes in East Harlem,¹⁹⁰ and Disability community¹⁹¹); and 6 were intended to serve as a needs assessment involving community members in identifying health issues, concerns, and determinants that might ultimately be used to develop an intervention study or to inform community action (Aboriginal grandmothers,^{192, 193} Positively Fit,¹⁹⁴ Bingham,¹⁹⁵ Housing Options,¹⁹⁶ Madison County,¹⁹⁷ Participatory Action Research for Community Health Promotion¹⁹⁸).

Ten studies moved beyond problem identification. Of these, six assessed factors influencing risk (Oregon Migrant Farm Workers;¹⁸¹ Chinese American Elderly with Osteoporosis;¹⁹⁹ Community Health and Environment Program;^200–202 Ethnocultural Communities Facing AIDS;^203–208 The Harlem Birth Right Project,²⁰⁹ Welcome Home Ministries^{210, 211}), two examined prevalence (The Glades Health Survey,²¹² West Harlem Environmental Action [WE ACT]^{213, 214}), and two examined the impact of environmental or policy change (EJS;^{215, 216} Evaluation of the Blended Funding Project²¹⁷). Although most CBPR studies are designed to increase community capacity or engender empowerment as a byproduct of the collaboration, four projects described this as the major objective of the study (African Americans Building a Legacy of Health,¹⁷⁹ Healthy Neighborhoods,^{69, 218} Participatory Action Research for Hmong Women,²¹⁹ South Asian women²²⁰).

Of the 29 studies in this category that provided information on data collection methods, the majority (62 percent) used qualitative methods either alone, or in combination with quantitative methods. In 12 projects, this was the sole data collection approach (Oregon Migrants, Aboriginal, Bingham, Controlling pesticides, Disability community, James Bay, Madison County, Perspectives of Latinas, Positively Fit, South Asian, Welcome Home, and Housing Options). In another six studies, the investigators combined qualitative and quantitative methods (CHEP, ECFA, HERE, Hospice Access, TAS, and Harlem Birth Right). Eleven studies (38 percent) reported using only quantitative methods (AABLH, Chinese Elderly, Diabetes in East Harlem, EJS, EBFP, Healthy Neighborhoods, PAR CHP, Poultry Slaughterhouse, Glades, Native Hawaiian, and WE ACT).

Over half of the studies (17 of 30) documented the involvement of the community in making measurement instruments more culturally relevant or mentioned field testing their instruments to improve their reliability (Oregon Migrants, Aboriginal, Chinese Elderly, Diabetes in East Harlem, Disability Community, ECFA, EJS, Healthy Neighborhoods, Housing Options, Hospice Access, James Bay Madison County, Native Hawaiian, PAR CHP, Poultry Slaughterhouse, TAS, and Harlem Birth Right).

Half the studies (15 of 30) presented baseline data, general findings or process evaluation results (Aboriginal, Bingham, CHEP, ECFA, Healthy Neighborhoods, HERE, Hospice Access, Housing Options, La Vida, PAR CHP, Perspectives of Latinas, Poultry Slaughterhouse, South Asian, Harlem Birth Right, and Welcome Home). The rest were primarily descriptions of either the research process or building the community-research collaboration.

Over a third of the studies (11 of 30) reported the use of multiple sources of evidence to validate their findings (Aboriginal, Bingham, CHEP, Controlling pesticides, ECFA, HERE, Hospice Access, Housing Options, Harlem Birth Right, TAS, and WE ACT).

Finally, many of these studies provided rich qualitative and quantitative data regarding the lengthy process of partnership development between universities and communities. Additionally, the studies described how the collaborative process benefited study design, data collection, and participant recruitment or retention, even if they did not include a formal evaluation of this process.

Level of Community Involvement in the Research Process

We reviewed all 60 studies to record evidence of the level of community involvement in the research process (Table 9). As with other sections of this review, our findings are limited by the information available in the published literature. Therefore, our report of the extent of community involvement is necessarily based on the perspectives of the authors of the published articles, which may not always have included the community partners.

The subsections below discuss specific elements of community involvement. The following analyses will typically begin by presenting the number of studies reporting any community involvement for each of these elements, with a comprehensive list of citations. However, in further analysis that lists the specifics of each element of community involvement, we provide illustrative rather than comprehensive citations. We have employed this approach because we found that in several instances, authors stated the nature of community involvement without providing additional detail. In other instances, we may have detected sufficient ambiguity about the extent of community collaboration to limit our abstraction of the data. Limitations of resources and time prevented us from seeking clarification from the authors in these instances.

Priority Setting and Hypothesis Generation. Twenty-eight studies involved the community in setting priorities and generating hypotheses. Often, community-based organizations were already concerned with an issue before researchers approached the community (e.g., Kahnawake). Sometimes residents needed to be recruited to form a Community Advisory Committee. The extent of community involvement varied greatly. Some studies changed or expanded priorities based on community input (James Bay, Survival Guide, CHEP, East Baltimore, HERE, La Vida, PAR CHP, and PAR Hmong); others mainly used community involvement to confirm priorities (Disability Community, NRMNC, Diabetes in East Harlem, and Health is Gold). One article reported a community organization that took the lead role, approaching the researchers about its community's priorities and desired research (WE ACT).

Of the 12 projects that assessed the effectiveness of an intervention, 8 reported community involvement (Sierra Stanford, Wai'anae, Health is Gold, HIV Latina, East Baltimore, Women and HIV Denial, Stress and Wellness, and Korean Study). Despite a priori notions that RCTs are less flexible than other study designs and that they tend to be dominated by researchers' concerns, we found that three of the four RCTs that evaluated interventions involved the community in setting priorities (Sierra Stanford, Health is Gold, and East Baltimore). In the case of the East Baltimore, the interests of community leaders were taken into account following a needs assessment to select hypertension and smoking as specific health issues.

Methods Selection. In all, 50 studies reported involving the community in selecting methods, but such participation occurred on different levels. Most studies reported using an advisory committee that cooperated with the researchers. Some committees reviewed proposed methods and suggested changes in wording or terminology to increase cultural appropriateness (Aboriginal, Madison County, ECFA, EJS, James Bay, and Housing Options).

Several communities were actively involved in designing surveys to emphasize particular issues of interest for the community. In one instance, the Haida Gwaii diabetes project, community involvement resulted in the exclusion of alcoholism, a major topic, because of controversy about the issue within the community.

Another frequently used method of involvement was to pretest surveys in the community. Evaluation of these pretest results led to changes in survey questions and improved clarity and validity (Chinese Elderly, TAS, Oregon Migrants, and ESVHWP). Some studies reported using qualitative results of focus groups or interviews to design an appropriate survey instrument (HERE and Hospice Access).

One group stated that it increased its sample size to address community concerns (Harlem Birth Right). Only one article described a complete change in data collection methods pursuant to community input. Residents of Madison County, for the Madison County study, stated a strong aversion toward surveys because of earlier experiences. Subsequently, the project adopted group interviews as a more acceptable method of data collection.

Proposal Development and Funding. Researchers usually took the lead role in proposal development, using their greater experience in the task of obtaining financial support, and they often applied for grants before the actual community involvement started. Fourteen studies mentioned community involvement in proposal development. Community involvement took place mainly in the form of advisory committees, but there were also examples of partnership steering committees in which community partners were involved as equal partners. In one instance (WE ACT), the community approached the researchers and initiated the proposal.

Nineteen studies reported shared funding. Communities mainly used funds to pay for staffing. In one study (Stress and Wellness), the community contributed some of the direct funding (taken from union funds) to maintain the research.

Study Design and Implementation; Data Collection Tools, Recruitment, and Retention. Twenty-eight studies described the active participation of the community in study design and study implementation. Some communities served in the form of advisory boards or steering committees to discuss possible challenges to study implementation (PRAISE, Okanagan, Internet Access, ESVHWP, CAAA, and Stress and Wellness). Another community took on a more active role proposing appropriate study designs to researchers (PAR CHP) or steering them away from potentially unsuccessful designs (ECFA). In several cases, community involvement tried to ease recruitment and study implementation by using local staff to administer surveys or interviews (Wai'anae, Seattle Homes, PACE, Disability Community, Okanagan, ESVHWP, Women and HIV Denial, and TEAL) or to act as survey helpers who were fluent in the languages of the target group (HERE).

Fifty studies reported community involvement with respect to recruiting and retaining subjects. Contact with community members generally raised the participation rate (Stress and Wellness, CHEP, EJS, ESVHWP, Oregon Migrants, and Positively Fit).

Community advisory boards or community-based organizations were often actively involved in recruiting participants. A commonly used strategy of recruitment was to seek participants within the social networks of community members who were involved in the research project (Health is Gold, PRAISE, Okanagan, PINAH, ESVHWP, Native Hawaiian, Disability Community, Seattle Homes, and Internet Access). Sierra Stanford emphasized personal contacts before the enrollment of the participants. One study (PRAISE) added an interim intervention for the delayed intervention control group, following advice of community members who were involved in the study. Another study (Chinese Elderly) changed from door-to-door recruitment to community meetings because team leaders thought that the latter would be more culturally appropriate for this particular community. In the HERE study, a union launched a mini-campaign to raise participation. Recruitment within social networks or the participation of volunteers led to high participation rates but also introduced the risk of selection bias; however, such bias was not measured directly,.

Intervention Design and Implementation. Of 30 studies with a planned or implemented intervention, more than 90 percent (28 studies) reported community involvement in intervention design,and implementation. Even among the 30 studies without a planned intervention (fully evaluated or otherwise), one-third of the studies (10 of 30) reported that communities were engaged in designing interventions for the community based on the results.

The magnitude of community involvement varied across these studies. Some researchers used findings of earlier community-based descriptive or exploratory studies as a base for intervention development (Healthy Home, Stress and Wellness, and East Baltimore). Others relied on advisory committees that co-designed the intervention and guaranteed its cultural appropriateness (ME2, PRAISE, Okanagan, PINAH, TEAL, and Health is Gold). Still others involved community organizations with active and creative leadership roles in shaping and implementing interventions (Sierra Stanford, South Asian, Survival Guide, East Baltimore, NY Immunization, ESVHWP, Stress and Wellness, Women and Heart Disease, ESVHWP, Stress and Wellness, and CMCA).

Two studies (Health is Gold and PRAISE) reported that, as a response to concerns of the community either during proposal writing or after funding, they implemented a delayed intervention for the control group. Another study stated that researchers agreed to implement the intervention sooner than intended after negotiations with its community steering committee (ESVHWP).

Feedback from communities also resulted in changed and adapted interventions to deal with the needs and priorities of the target groups (PACE, PINAH, and Health is Gold). Some studies undertook additional efforts to be flexible in addressing community needs and removing barriers specific to the intervention community that could otherwise have compromised participation or intervention. These steps included providing native speakers, child care, transportation, or small stipends (ME2, South Asian, Survival Guide, Healthy Home, Health is Gold, and Korean Study). One study related a negative impact of community involvement; the Korean Study Breast and Cervical Cancer Intervention could not be fully implemented because of a lack of community staff.

Translation of Research Findings. We reviewed the studies to identify those in which communities were involved in translating research findings into demonstrable policy change, either in civic bodies or at private institutions and local levels. Three of the 60 studies reported demonstrable policy change in civic bodies as a result of the intervention (EJS, CMCA, and PAR CHP) through the efforts of the community collaborators. EJS led to a presentation of findings to the House Agricultural Committee of the North Carolina General Assembly, followed by subsequent changes in policy. As a result of the CMCA study, policies were altered to reduce youth access to alcohol through changes in procedures and practices in the communities via alcohol merchants, law enforcement and criminal justice, community events, hotels, media, treatment agencies, and religious venues. PAR CHP, partly through supporting data from its survey, prompted the city council in the community to pass an ordinance to create nonsmoking areas. Five studies resulted in changes at private institutions or local levels through the efforts of community collaborators (Bingham, Healthy Neighborhoods, HERE, Stress and Wellness, and Poultry Slaughterhouse).

Five studies had the potential for change in policy through the generation of plans addressing the specific health concern (AABLH, ECFA, James Bay, TEAL, and NRMNC). They did not report the impact of these plans, however.

Integration and Sustainability. Thirteen studies reported on the sustainability of programs or interventions. An additional 28 studies detailed the integration or application of findings to achieve changes that affect health or other aspects of daily life.

Some projects achieved temporary sustainability of programs by acquiring additional grants for further research (CHEP, Oregon Migrants, and Kahnawake) or through local funding (Healthy Neighborhoods, Wai'anae, East Baltimore, and Glades) initiated by community organizations. One screening program reported sustainability as a result of the community's closer contact to health clinics during the research (Korean Study).

Multiple studies reported sustainable changes in policies or other aspects of daily life through the presentation and application of findings (Healthy Neighborhoods, CHEP, CMCA, HERE, Stress and Wellness, NRMNC, Bingham, Poultry Slaughterhouse, Madison County, PAR CHP, and EJS). For example, Healthy Neighborhoods was able to re-establish evening and night bus services and to have tobacco billboards removed. The HERE project managed to reduce the workload of hotel room cleaning staff. Communities also frequently used the CBPR project findings to develop action plans for other programs and to apply for grants (Native Hawaiian, Glades, Survival Guide, Diabetes in East Harlem, and Perspectives of Latinas).

Community Involvement in All Aspects of Research. Of the 60 studies relevant to KQs 2 and 3, three studies reported community involvement in all aspects of the research (Wai'anae, Kahnawake, CHEP, and HERE). Of these studies, one was an evaluated intervention with a quasi-experimental design (Wai'anae); another was an incompletely evaluated intervention (Kahnawake); and two were were nonexperimental studies that did not include any interventions (CHEP and HERE).

Improved Research Quality Outcomes

As discussed in Chapter 2, we scored the 12 studies with completed interventions in terms of two outcome evaluations: average scores for research quality and for adherence to the principles of community participation (recorded in Table 6). Higher scores reflect better quality. The average scores could range from 1 to 3, based on the quality grading form provided in Appendix B. Although the scores on these two dimensions are not directly comparable, the average research quality scores ranged from 1.5 to 2.8 with a mean of 2.3, while the community participation quality scores ranged from 1.6 to 3.0 with an average of 2.2

As would be expected, research quality scores reflected research design rigor. Experimental studies averaged 2.7; quasi-experimental, 2.2; one-group pretest and posttest design, 1.9; and the one nonexperimental intervention study, 1.5. Community participation scores appeared less closely associated with study design, with the experimental studies averaging 2.3; quasi-experimental, 2.2; one-group and posttest design, 2.3; and the nonexperimental study, 1.95.

We also conducted quality ratings on the three observational studies that we deemed were of sufficiently strong design to permit generalizability to a population beyond that of the study sample. Many observational studies reviewed served primarily as baseline data for a community assessment or an intervention study with no attempt at representative sampling techniques, thus were not included in the quality ratings. We used slightly different criteria for research quality ratings with the observational studies, primarily related to the lack of an intervention. Research quality rating scores for the three observational studies were 1.4, 2.6 and 2.1, with community participation scores of 1.6, 2.6, and 2.0, respectively.

Quality rating scores for research elements primarily reflect internal and external validity. Recognizing that RCTs are not always feasible or ethically appropriate in CBPR where one group would be denied an intervention, we rated the intervention studies based on specific criteria reflecting reliability and validity rather than requiring a randomized controlled trial for the highest quality rating. While the four experimental completed intervention studies were all RCTs, a study using group assignment with careful matching of intervention and comparison groups would also have been included. Studies were downgraded, for example, if the study population differed significantly from the population to which findings were generalized, if there was significant loss to followup, or if the intervention and comparison groups were not comparable demographically. For observational studies, we downgraded those that failed to adequately justify their sampling procedure or the control of confounders.

In abstracting data from these studies, we documented evidence of either enhanced or diminished research quality attributable to the CBPR method; we focused on the categories of methodology, measures, recruitment, intervention, analysis, dissemination, and outcomes. Of the 12 completed intervention studies, 11 reported enhanced intervention quality related to community involvement. Only two studies reported improved outcomes related to CBPR. Eight noted enhanced recruitment, four reported improved research methods and dissemination, and three described improved measures. Very little evidence of diminished research quality resulting from CBPR was reported. One study suggested possible recruitment bias (NY Immunization) and another reported that the CBPR approach pulled staff away from intervention delivery, thus reducing the exposure to the intervention (Korean Study).

Community Capacity Outcomes

Improved community capacity is rarely discussed as the objective of the study or the intervention. However, in describing their CBPR methods, authors clearly considered improved community capacity to be an essential component of the process. Of the 60 studies in this review, 47 reported improved community capacity as an outcome associated with the study. Generally, authors focused on the greater capacity of the participant community rather than that of the research community, possibly reflecting the biases of the authors who were primarily academic researchers. Only nine studies documented the improved capacity of the researchers and research organization from collaboration with the community (James Bay, CAAA, Health is Gold, Kahnawake, Poultry Slaughterhouse, Disability Community, NRMNC, ESVHWP, and Korean Study). In our review of the definitional literature, however, development of individual investigator and research institution capacity to interact better with the community on research issues is a significant expectation of CBPR.

Seven studies mentioned the communities' enhanced capacity to create change (Poultry Slaughterhouse, HERE, Madison County, Native Hawaiian, TAS, Oregon Migrants, and Stress and Wellness). Increases in community capacity happen either directly through the research results or indirectly through the process of participating in the research.

Studies demonstrated enhanced community capacity in numerous ways. Additional grant funding obtained by the community was one such outcome (Haida Gwaii, CHEP, Welcome Home, Stress and Wellness, Healthy Neighborhoods, NRMNC, and ESVHWP). Another positive result was the jobs created by the collaboration (ESVHWP, NRMNC, Wai'anae, and Project TEAL). Skills building (CMCA and East Baltimore) and partnership and coalition development (ESVHWP, Okanagan, and Wai'anae) were other beneficial outcomes of the CBPR activities. Finally, numerous studies mentioned the communities' enhanced capacity to conduct research, either in combination with other outcomes of community capacity or as the sole evidence of enhanced community capacity (James Bay, Disability community, Korean Study, PRAISE, Sierra Stanford, Healthy Home, WE ACT, Internet Access, NY Immunization, AABLH, Women and HIV Denial, Controlling pesticides, EJS, La Vida, PAR CHP, PACE, and Wai'anae).

Health Outcomes

Among the 12 studies evaluating completed interventions addressing health outcomes, 2 dealt with physiologic health outcomes (East Baltimore and Okanagan). Three studies assessed cancer screening behavior (Health is Gold, Korean Study, and Wai'anae) and four others addressed other types of behavior change, such as alcohol consumption, immunization rates, and safer sex behavior (CMCA, HIV Latina, NY Immunization, and Women and HIV Denial). Finally, three studies measured the impact of the intervention on psychosocial outcomes such as emotional support, empowerment, and employee well-being (Sierra Stanford, Internet Access, and Stress and Wellness).

The four RCTs reviewed all resulted in at least some modest positive effects; eight non-RCTs showed more mixed results. Given the highly varied health outcomes, measurement strategies, and intervention approaches used, comparing studies to assess relative impact on health outcomes is not possible. Cost-effectiveness data would have allowed us to compare similar outcomes from CBPR studies and more traditional research studies, but no study provided such data.

From our review of the published data on these studies, we were unable to determine whether the modest positive findings reported could be attributed to CBPR methods. Several authors mentioned positive effects of their CBPR approaches on research quality and participation rates, but we could not ascertain whether these benefits directly improved study outcomes relative to nonparticipatory research approaches.

Current Approaches by Funders to Solicit and Review CBPR Proposals

The CDC and NIEHS have been at the forefront of Federal funding for CBPR to date. Specific initiatives by these agencies include many of the studies we reviewed. For example, the CDC funded three Urban Research Centers in 1995, and NIEHS sponsored two CBPR funding vehicles—Environmental Justice and Community Based Participatory Research in Environmental Health—since 1993. In 2002–2003, the CDC funded 26 new projects under the “Community-based Participatory Prevention Research” grant mechanism.

Private foundations also support CBPR; the W. K. Kellogg Foundation and Annie E. Casey Foundation are among the leaders in the private sector. The Kellogg Foundation funded a Community-based Public Health Initiative (CBPHI) in 1991 that included several sites that emphasized community-university-agency partnerships to address health disparities. This program prompted the creation of the Community Health Scholars Program, designed to fund postdoctoral applicants seeking training in CBPR (http://sph.umich.edu/chsp/index.shtml).

The considerable interest at the Federal level in funding CBPR is further evidenced by the creation of an Interagency Working Group for Community-based Participatory Research, which has begun to assemble information about existing funding mechanisms for CBPR.⁶⁰ Given the rising interest and monetary support for this work, AHRQ sponsored a national meeting in 2001 to explore the current role of CBPR and how best to foster good proposals and successful initiatives in this arena. Participants at that meeting strongly recommended that AHRQ commission this systematic review of issues relating to CBPR, with a view to clarifying this entire research enterprise for current and potential supporters.

Depending on the agency, CBPR proposals may be reviewed through existing study sections or through a special emphasis panel. Because CBPR is an excellent approach to translational research, study sections designated for this purpose are particularly appropriate. Many parts of the National Institutes of Health (NIH) refer to these as R18 proposals. These would include, for example, Demonstration and Education Research within the National Heart, Lung and Blood Institute and Translational Research within the National Institute of Diabetes and Disgestive Kidney Diseases (both of which use special emphasis or ad hoc panels for review).

A new study section within the National Cancer Institute is Community Level Health Promotion. Standing study sections generally require a multiyear tenure by committee members, and they review all grants deemed relevant to their focus. A special emphasis panel or ad hoc committee is assembled specifically for the purpose of reviewing responses to a Request for Application (RFA) or more narrowly defined research area. The advantage of a special emphasis panel is that specific instructions, pertinent to the proposals being reviewed, are sent to reviewers for each meeting. Reviewers selected are also more likely to be content experts with respect to the focus of the RFA.

Reviewers for all proposals generally receive review criteria to guide their efforts. These criteria often follow the framework of the standard proposal format and commonly include such broad sections as Significance, Innovation, Approach (methods), Investigators, Research Environment, Budget, and Human Subjects.

Discussions with individuals from the NIH and CDC who are involved with generating RFAs and refining the review process highlighted the need for brief guidance materials about CBPR for reviewers less familiar with this approach. They recommended fact sheets that could be distributed between sessions to standing panels (with the assumption that guidance arriving with a large box of grants will be less likely to be read) or with other orientation materials for special emphasis panels. Also recommended were guidelines for those writing RFAs designed to encourage CBPR submissions and offer guidance for researchers submitting CBPR proposals.

Criteria for High-Quality Grant Applications

As described above, a few special funding mechanisms to date have focused specifically on promoting CBPR. Perhaps the bigger challenge is to obtain funding for CBPR through more conventional review mechanisms in which reviewers may be less familiar with and perhaps even skeptical about CBPR. Not only will a broader range of funding options for CBPR expand the options for funding CBPR efforts; it can serve to educate other scientists about the potential rigor and “added value” of CBPR.

Conventional Research Criteria

Researchers who are applying for funds to support CBPR often fail to address all the criteria for high-quality conventional research, and this may be the biggest mistake in seeking CBPR funding. We identified relatively few high-quality completed interventions or observational studies relative to what appears to be many excellent collaborations based on CBPR principles. This mismatch raises the question of whether researchers assume that effectively combining high-quality conventional research with CBPR collaborations is not possible. If so, they may simply choose not to embark on such ventures.

CBPR Criteria

In addition to meeting criteria for conventional research proposal review, a proposal based on CBPR should clearly describe the added value that this approach brings. This is particularly important when reviewers can be assumed to be unfamiliar with CBPR, which is still probably a safe assumption. The proposal should not simply describe CBPR criteria; it should also discuss the potential benefits for both research quality and the community. Table 1 (p. 15) provides a detailed framework of CBPR principles and their benefits. This information is also available on the RTI Web site (http//www.rti.org) in the document “CBPR Reviewer and Applicant Guidelines” (CBPR Exhibit 1).

Guidance for Funding Organizations and Applicants

Based on the results of our literature review, discussion with Federal funders, a review of funding agency Web sites, and the criteria for funding outlined above, we have created three concise documents that provide suggested guidance to funding organizations, reviewers, and applicants: “CBPR Reviewer and Applicant Guidelines,” “CBPR Reviewer Checklist,” and “CBPR Requests for Applications and Peer Review.” These materials are included online (athttp://www.rti.org/) as CBPR Exhibits 1, 2, and 3, respectively. For a more detailed checklist, we refer the reader to work by Green and colleagues, “Guidelines and Categories for Classifying Participatory Research Projects in Health Promotion,”²²¹ which appraises the extent to which proposals or projects align with principles of participatory research.

Because the grant proposal and review process is somewhat standardized across the U.S. Department of Health and Human Services agencies (using the PHS-398 package, for instance), and because these agencies are likely to involve the most rigorous review process, we elected to use the review criteria generally used by these agencies in developing a prototype guideline document. The “CBPR Reviewer and Applicant Guidelines” document (Exhibit 1) is adapted from NIDDK review criteria for translational research, with components for CBPR that we have added for this particular purpose. However, these guidelines are our recommendations and have not been formally adopted by AHRQ or other components of HHS.

The “CBPR Reviewer Checklist” (Exhibit 2) goes one step further, adding to these suggested guidelines more detail regarding what should be expected in a high-quality proposal involving CBPR. Because this example is modeled on what we might expect or advise for Federal research agencies, it may not translate directly to grant review mechanisms that foundations and other funding sources might use. These are highly variable across such funding organizations, but we believe that their review procedures will often include the primary components covered in Exhibits 1 and 2; thus, such organizations could adapt this checklist to their own purposes in a fairly straightforward manner.

Finally, as outlined in “CBPR Requests for Applications and Peer Review” (Exhibit 3), our discussions with funders and review of the literature led us to recommend that review panels include academic experts in the content area and in CBPR methods, and that the panels also involve individuals who have expertise in both arenas. Our discussions did not lead to a clear recommendation regarding how community members should be involved in the peer review process for CBPR. Some precedent exists for “citizen involvement” on academic and industry advisory committees and review panels for activities such as Institutional Review Boards. Federal staff, with whom we discussed this issue, reported limited experience with community members on review panels, and they had mixed feelings about the best way to include community representatives in the process.

An underlying concern is the potential discomfort for community members who are put into a situation in which the language and subject matter are quite foreign. One NIH contact described a situation in which community members participated in a review for which no prior orientation had been held to enable them to discuss their respective perspectives. This resulted in a very tense and unproductive session. Thus, on the one hand, without a thorough understanding of research principles, lay persons may find it difficult to understand and contribute to much of the discussion. On the other hand, a community member is uniquely qualified to help reviewers critique the proposed approach to community participation.

In short, more careful and creative thought is needed concerning how to solicit input from community members. Some possible solutions to consider include the following:

• Provide extensive orientation for individual community members serving on review panels.

• Oriente the academic panel members to the role of community members.

• Convene an orientation meeting before the formal review to discuss review expectations, ground rules, questions, and concerns.

• Invite community representatives who have been involved in CBPR and hence are more knowledgeable about research.

• Ask community representatives to read abstracts and participate in the discussion but not to serve as a primary or secondary reviewer.

• Ask community representatives to read abstracts and relevant CBPR components of proposals and be asked to assess those components.

• Ask principal investigators to submit two versions of the proposal abstract: one for a lay audience and one for academics.

• Hold primary reviewers for each proposal responsible for engaging community representatives in the discussion in a positive and nonthreatening manner.

• Require the resulting summary statement to include a section reflecting comments from community representatives, which may increase the likelihood that the primary reviewers will involve community representatives in a meaningful way.

Quality of Research Methodology

An inherent challenge faced by anyone trying to evaluate the quality and impact of CBPR methodology is the fact that being true to the methods makes it nearly impossible to compare CBPR rigorously to research carried out with more traditional research methods. The problem begins early in the process in that the purest form of CBPR requires that the community identify the health problem to be addressed. One could not readily compare the process and outcome of a study for which the community chooses diabetes as a research focus and the researchers choose HIV/AIDS.

Although in theory one could preselect a study outcome and measure and then conduct a two-arm trial randomizing half the participants to a CBPR approach and half to traditional research methodologies, the two approaches would almost certainly yield different sets of measures, interventions, and recruitment and retentions strategies, leaving very little for comparison other than the final outcome measure. One would be left wondering whether the outcomes achieved were potentially biased by different factors in each study that could be the result of the research method used, such as interviews conducted by individuals hired from the community in the CBPR arm as compared to interviews done by graduate students in the traditional arm.

In the absence of randomized trials comparing CBPR with non-CBPR approaches, we are left with trying to draw conclusions from what investigators report in published journal articles. We have found that publication of intervention research (conducted by either CBPR or traditional methods) is associated with significant challenges related to page limitations of journals. Authors of such studies must often publish their findings and study methodology in separate pieces. This problem is further compounded for CBPR work; researchers must report years of partnership development and collaboration in very few words and in a small number of journals willing to accept this more descriptive science. As a result, we found that articles lacked information about the implementation of CBPR, from both the community participation and the research perspectives.

In our review, we were careful to assess research quality based on factors such as internal and external validity rather than a strict adherence to traditional study designs. For example, rather than specifying that a randomized controlled trial (RCT) is the highest quality study design, we assessed the degree to which the study sample was representative of the larger population to which generalizations would be made, whether intervention and comparison groups were comparable, the quality of the measures, and loss to followup. Study designs that included a delayed intervention control group intended to provide benefit to those randomized to the control condition were rated as very high-quality studies. Similarly, studies that gave thoughtful attention to the identification of a nonrandomized comparison that preserved internal validity while responding to community concerns were also given high marks.

To date, a limited number of CBPR studies have been published that represent a complete and fully evaluated intervention or an observational/epidemiologic study that can be generalized beyond the participants involved in an intervention study (baseline data). Recent special issues for journals focusing on CBPR have reported on studies with high-quality research methods, as with the July 2003 issue of the Journal of General Internal Medicine. Other journals (including the American Journal of Public Health and the Journal of Interprofessional Care) have issued similar calls for CBPR articles, but these occurred after our evidence review period. Much of the research reported in these special issues was generated as a result of studies funded through the Centers for Disease Control and Prevention (CDC), the National Institute for Environmental Health Sciences (NIEHS), and several foundations. As funding agencies and high-quality peer-reviewed journals begin to recognize the legitimacy and potential value of CBPR, these steps offer further encouragement to researchers combining both excellent research methods and adherence to the principles of CBPR.

Although the potential for trade-offs between addressing community concerns about research and maintaining high-quality study designs has been cited as a possible challenge to high-quality research,^{2, 58} our review does not suggest a strong trend in the direction of solid community-based participatory methods combined with weak research design or measurement (Table 6). Similarly, the strongest research methods do not appear to be combined with weaker community participation elements. Again, owing at least in part to page limitations in those journals publishing rigorous experimental research, researchers tend not to describe fully their research methodology, adherence to CBPR principles, and the degree to which the collaboration may have benefited or threatened the research quality. Future CBPR researchers should consider identifying creative approaches to condensing this information in tabular format or making it available on the Web.

Most of the studies we reviewed were nonexperimental in design; only a limited number included any sort of intervention. When multiple papers were published about a single study, we combined the information in a single table row of our evidence tables and treated the data as a single unit. We did not inflate the relative number of nonexperimental studies by the spread of content across several articles. Many papers described the partnership development process and reported on formative data related to their processes and assessments of community concerns. In our view, many of the nonexperimental studies had been funded with small grants to develop partnerships around an identified health issue that did not provide sufficient resources to conduct an intervention or rigorous evaluation.

We also speculate that few larger intervention and/or experimental trials were funded in the past because review panels were not receptive to a CBPR approach. To the traditional researcher, asking study “subjects” to identify the focus of research, help design the intervention, and provide feedback on measurement instruments and data analysis might be viewed as scientific heresy. At the same time, researchers skilled in community collaboration may or may not be equally skilled in using rigorous research methodology and thus able to convince reviewers of the strength of the complete CBPR approach.

Additional possible explanations for the relative lack of completed evaluations of CBPR interventions is the “lack of fit” between the dynamics of true community collaborations and the peer-review funding approach to setting research priorities, maintaining timelines, and exercising budgetary control. Partnership development between communities and researchers takes time; if such work is to be truly community-guided, then it requires a different way of thinking about choosing research topics and allocating funding.

CDC and the National Institutes of Health (NIH) are major sources of health-related funding. Both agencies are divided into institutes and centers primarily related to specific disease entities, such as diabetes, cardiovascular disease, and HIV/AIDS. For the most part, these agency divisions generate funding opportunities and review proposals. This results in what is sometimes referred to as “categorical funding,” which ultimately leads to putting researchers in the position of choosing a health issue and then looking around for a community where this topic can be studied. With the exception of some foundations, such as the W.K. Kellogg Foundation with the Community-based Public Health Initiative, and Federal agencies, such as the CDC with the Urban Research Center Initiative and more recently the CBPR initiative, few funding opportunities allow the flexibility of research partners selecting the focus of their research based on concerns identified within the community.

Length of funding is also an issue. In true CBPR, by the time the partnership has formed and the health outcome is identified, time in the funding cycle (usually a maximum of 5 years) is generally inadequate to implement and complete a well-developed intervention and rigorous evaluation. Several solutions have been proposed. Israel and colleagues recommend the use of planning grants to facilitate partnership development and identification of the research focus.²²⁶ The planning grant could be a “stand-alone” funding option or linked to a larger followup funding opportunity. The CDC REACH (Racial and Ethnic Approaches to Community Health) Initiative, for example, makes followup funding for longer-term work contingent on successful partnership development and issue identification. The CDC's Community-based Participatory Prevention Research effort requires the community-university research collaborative to demonstrate an existing track record before applying. This approach rewards researchers who choose to become involved in community collaborations before the potential for funding becomes an incentive.

Finally, budgetary restrictions may inhibit the generation of high-quality CBPR. Perhaps more important than the total amount is flexibility in budget management and expenditures. As communities receive an increasing number of requests to participate in research projects, often receiving little direct benefit in return (such as an epidemiologic study where risks are identified but no intervention is delivered in return), they are understandably demanding more involvement regarding the decisions about expenditures. For example, funds could be used to hire graduate students to conduct telephone surveys or to hire and train community members who are currently unemployed, thus infusing funds directly into the community while building capacity among community members. Budgetary restrictions (such as no overhead dollars to be spent on food) that may be an irritation to academics can have more serious consequences for research in the community, where food is considered an essential component of social interaction and serves as an incentive or an acknowledgment for research participation. Indirect expenses, in general, represent a disparity between universities and the communities, where the academic institution receives substantial overhead, but few indirect costs of the community organization are covered.

Level of Community Involvement

In our review, community involvement extended through all areas of research, although the extent of involvement varied by the stage of the research. The strongest involvement was in recruitment of study participants, design and implementation of the intervention, and interpretation of findings. Many authors argued that community involvement, especially in these areas, led to greater participation rates, increased external validity, decreased loss to followup, and increased individual and community capacity.

Disadvantages to such methods were not frequently reported. They may include some loss of internal validity, often through introduction of selection bias (recruitment), and lack or sometimes even loss of randomization if contamination occurs as community members become more knowledgeable and share intervention strategies with control or comparison groups. Disadvantages may also inclue highly motivated intervention groups not representative of the broader population and possible biased interpretations of findings.

In many cases, distinguishing between advantages or disadvantages associated with CBPR can be difficult. For example, on the whole, community mobilization can yield high and sustained attendance rates at intervention group sessions but also produce some “spillover effect” in the control group. Whether this is, on balance, a good or bad thing for the research process is open to debate.

Our review suggests that hypothesis generation and proposal development remained mainly in the hands of researchers. Most studies involved some form of community advisory boards that worked closely with the researchers in setting priorities, developing interventions, and assuring a culturally appropriate approach. Only a few, however, involved a steering committee or decisionmaking board that actually took an active lead role.

If this leadership pattern could be attributed to the community's lack of decisionmaking power and experience or lack of ownership of the research, the publications we reviewed did not make it clear. Some articles addressed the persistent challenge for researchers to maintain scientific validity and to share ownership with community groups and address participant interests. In one diet and cancer study (PRAISE!), researchers scrambled to create a non-nutrition intervention for the delayed intervention control group when it appeared that this group was so enthused about the project that they intended to create and implement their own nutrition intervention early in the project. Other researchers reported mid-course adjustments in the intervention or measurement approach based on input from the community.

Some studies reported that application of findings influenced policy changes that led to a sustainable improvement for the community. Others received further funding that was obtained by the community. Apart from these obvious successes, some studies suggested that empowerment of the community was a positive result of participation in the research.

Improving Research Quality

To achieve the highest research quality, researchers must select the strongest possible study design, measurement approach, data collection plan, and analysis strategy to address their specific research question or specific aims. If community input suggests that an RCT to test a diabetes intervention would be unacceptable because the control group would receive no benefits, it is incumbent on the researcher to work with the community to select and justify the strongest possible alternative design, such as a delayed intervention control. The research partner must present arguments in the proposal that identify the potential costs and benefits of a variety of different approaches from both the research and community perspectives.

In addition, researchers must give community members credit for the ability to understand complex research challenges if they present the issues clearly and thoughtfully. One of the many benefits of involving community members as research partners is that they begin to see the long- term gains associated with research — for example, improved intervention approaches, increased potential for funding and dissemination, “ammunition” to advocate for effective policy changes — even as they come to understand the relatively short-term bother of the data collection activities themselves (e.g., blood draws, long surveys). This can have a positive effect on the immediate study and on the potential for study participants to become involved with future research efforts.

Given the substantial number of good-quality but incomplete CBPR intervention studies we identified, an increasing number of initiatives to fund CBPR work, and journal editors giving special attention to this research, the number of high-quality CBPR publications is likely to rise significantly in the next few years.

Improving Community Capacity

Authors of the studies we reviewed here rarely brought up enhanced community capacity as an explicit goal of a CBPR project. Rather, they mentioned it in descriptions of the collaborative process and clearly considered it to be a critical component. Studies were much more likely to report capacity building on the part of the community rather than on the part of the researchers or their institutions.

In our review of the definitional literature, however, development of the capacity of individual investigators and research institutions to interact more collaboratively with the community on research issues is a significant expectation of CBPR. Researchers, who are the traditionally designated “experts” in conventional academic-community partnerships, may find it hard to view themselves as learning from their community partners. When published studies results discuss capacity building on the part of the researchers, we may rightly conclude that such learning has taken place.

Improving Health Outcomes

Among the limited number of fully evaluated complete interventions that we located for our review, the stronger or more consistent positive health outcomes were generally found in the higher-quality research designs. This could serve as an incentive to CBPR research partnerships to pay adequate attention to the “R” component of CBPR.

Given the long-term nature of true CBPR efforts, one could argue that the potential scope of related health outcomes cannot be realized from one 5-year study focused on a specific chronic illness. If a CBPR effort successfully builds individual and community capacity, future benefits may include improved lifestyle habits, increased institutional responsiveness to workers' health concerns, or changes in policy that facilitate a healthier environment. Associated positive health outcomes might have nothing to do with those initially targeted by the study. None of the studies we reviewed could have captured such long-term and indirect potential benefits of CBPR.

Criteria and Processes for Reviewing CBPR Proposals

Although our review focused on published CBPR papers rather than grant proposals, it provided some insight into the quality of research that has been funded. We were somewhat surprised by the limited number of high-quality completed intervention and observational studies identified in our review. Because we included only completed interventions and epidemiologic studies in our quality rating system, we may have missed some high-quality research projects focused on formative data collection or cross-sectional survey findings that did not meet these criteria.

We have discussed some potential reasons for the limited number of high-quality published studies describing completed interventions. They included unfamiliarity with CBPR principles or skepticism about involving research participants in the research enterprise, challenges of developing a research partnership and completing a study within the traditional funding frameworks, and a focus of many reports on the development of research partnerships rather than outcomes. As described in the next section, we have reason to believe that this number of completed projects will grow in the near future because of several initiatives promoting the funding and publication of CBPR.

With the abundance of interest in funding CBPR efforts, understanding what we have learned to this point and how this can be applied to improving this field of research in the future is critical. Guidelines for applicants and reviewers are also essential, as are recommendations for funding agencies interested in supporting this type of work. Indeed, our review suggests that the stronger studies were somewhat more likely to be funded by Federal agencies with more stringent review processes than, for example, State or community-level organizations.

If we are to continue in our efforts to understand the quality and impact of CBPR, funders must structure their Requests for Applications (RFAs) to elicit responsive applications adhering to CBPR principles, and reviewers must be adequately familiar with the nuances and potential added value of CBPR to identify proposals with the greatest potential to move this field forward. Exhibits 1 through 3 (in Appendix E^*) offer guidelines to support this effort.

Challenges of the Literature Review

As with many systematic efforts to review the literature, this one was hampered by our inability to initially narrow the scope of the literature using existing Medical Subject Headings (MeSH) terms or key words. MEDLINE^® searches for CBPR articles are particularly challenging because the literature is newly emerging and the MeSH indexing is not yet adequate for the task. We considered many terms while constructing our searches (Table 10). Terms with asterisks occur frequently in the relevant citations and CBPR literature, and terms in quotes are key words, not MeSH terms.

Searching MEDLINE^® and combining these three concepts yielded more than 1,300 citations. These multiple searches yielded numerous articles of varying relevance; moreover, formal MEDLINE^® searches did not always identify highly relevant articles. When we probed, we could find no consistent coding. Thus, we supplemented these searches with citation searches in previously identified articles and with recommendations from experts in the field.

As CBPR becomes better recognized and understood, the MeSH indexing should become more sensitive. We recommend building a uniform set of MeSH headings to describe CBPR and encouraging journal editors to suggest the use of these terms as appropriate.

Future Growth of CBPR

Based on several developments in CBPR research uncovered in our review, we believe that the number of high-quality CBPR studies published is likely to increase substantially in the near term. First, NIEHS continues to fund proposals emphasizing CBPR and environmental justice. Second, NIEHS hosted a conference in 2000 on successful models of CBPR to “expand the acceptance, use, and applicability of CBPR as a valuable tool in improving the public health of the nation” (p. 1), followed by a report titled Successful Models of Community-Based Participatory Research."²²⁷ Third, AHRQ convened a CBPR planning conference in 2001; AHRQ also initiated the EXCEDE program-90 national leaders interested in advancing CBPR. Fourth, the Journal of General Internal Medicine published a special issue on CBPR in 2003 (funded by AHRQ), as did the American Journal of Public Health, also in 2003. Finally, the Journal of Interprofessional Care will sponsor a CBPR theme issue in 2004. Fifth, the CDC, through the Urban Research Centers and the Prevention Research Centers, continues to fund this type of research. Sixth, the W.K. Kellogg Foundation has increased support to train Community Health Scholars, with an emphasis on CBPR methods. Seventh, a CDC initiative (totalling $13 million) seeks to support “multi-disciplinary, multi-level, participatory research that will enhance the capacity of communities and population groups to address health promotion and the prevention of disease, disability, and injury”; 26 proposals for 3-year projects have been funded. Eighth, formation of a Federal interagency workgroup for CBPR⁶⁰ will strengthen communication among Federal agencies with an interest in supporting CBPR processes.²²⁸ Ninth, an Environmental Health Perspectives' Supplement, “Advancing Environmental Justice Through Community-Based Participatory Research.²²⁹ Finally, a report by the Community-Campus Partnership for Health, “Developing and Sustaining Community-University Partnerships for Health Research: Infrastructure Requirements.”²³⁰

Given the predicted increase in high-quality CBPR publications in the near future, we recommend that AHRQ or another agency committee sponsor an updated evidence review of CBPR within a few years to assess the development of this field and to refine, insofar as necessary, our proposed guidelines for proposal development and review.

Environmental and Policy Change

In many areas of health promotion and disease prevention, researchers and community activists alike are beginning to focus their efforts further “upstream” on the socioecologic model, which means placing a greater emphasis on policy and environmental change that facilitate health-promoting choices at the individual level. The belief is that individuals currently facing a “toxic environment” related to air quality, availability of healthy foods, opportunities for physical activity, and ease of access to alcohol and cigarettes may be better served by community-level change than by intensive efforts aimed at individual behavior change.

CBPR fits well with this trend toward “upstream” approaches to health promotion through its ability to mobilize community action. Although some approaches to environmental and policy changes require State or national legislative decisions, many other environmental enhancements can occur through micro-level policy change within the community or workplace. For example, some CBPR efforts were able to identify workplace health and safety issues of great concern to the workers, form working groups, and begin to address some of the issues (the Stress and Wellness and Poultry Slaughterhouse projects illustrate these steps; see Table 8 in Chapter 3 for the full set of references). Better funding for this research effort might have allowed for a stronger study design able to demonstrate effectiveness.

Conventional and CBPR researchers alike face many challenges in the area of study design and measurement as we move our research upstream. However, CBPR approaches to community collaborations are well positioned to engage communities and achieve the desired changes. Seeking the best possible balance between research methodology and community collaboration is critical to move the field forward.

Improving the Quality of CBPR Reports

New guidelines from international groups provide clear instructions on how randomized controlled trials (CONSORT) and observational studies (MOOSE) should be reported.^{231, 232} Systematic reviews such as this one are frequently hampered by the lack of standardization in the peer-reviewed literature, leading to many studies being left out or an inability to draw useful conclusions about a particular field of research. If studies are incompletely or inaccurately documented, their quality rating is likely to be downgraded (fairly or not).^{233, 234}

Just as we have proposed guidelines for the CBPR proposal writing and peer review (study section) process, perhaps recommendations are needed for improving the quality of reports for CBPR studies. O'Toole, in the Journal of General Internal Medicine special issue on CBPR, suggested the need for a “common language” regarding CBPR and describes a potential process for CBPR findings in the health sciences literature; he articulates this approach as “research-plus” that is methodologically rigorous while maintaining important contributions to the relevance and translation of research.²³⁵

Publication guidelines, like those for proposal review should reflect the increasing rigor required of authors in the evidence-based practice field while recognizing the unique situation facing researchers who are balancing research rigor with commitment to community collaboration. For CBPR to gain more credibility and receive more research dollars, researchers and community members must hold themselves to the highest possible standards on both sides of this issue.

Support for CBPR from the Community of Scholars

If CBPR is to achieve its full potential as a research process or methodology uniquely designed to address some of the most challenging health care issues of our time, full support is required from the “community of scholars,” located in neighborhoods as well as universities. Funding agencies must understand the full benefits and complexities of CBPR to generate RFAs that elicit high-quality proposals incorporating the essential research and participatory elements of this approach. Communities must take the risk to become full partners in the research enterprise, contributing their unique knowledge and experience while safeguarding their interests. Reseachers must combine excellent science with compassionate and respectful community partnerships; journals must create opportunities to highlight and disseminate CBPR research products; and health care providers and policymakers must be guided by the evidence that results from the collaborative efforts. Enhancing any one component of this cycle is likely to have a positive effect on the others, ultimately strengthening and sustaining community-based participatory research.

Technical Expert Advisory Group

We also extend our appreciation to the members of our Technical Expert Advisory Group (TEAG), who provided advice and input during our research process. The RTI-UNC EPC team solicited the views of TEAG members from the beginning of the project. TEAG members also provided insights into and reactions to work in progress and advice on substantive issues or possibly overlooked areas of research. TEAG members participated in refining the analytic framework and key questions and discussing the preliminary assessment of the literature, including inclusion/exclusion criteria, and also provided input on the information and categories, including evidence tables. The TEAG was both a substantive resource and a “sounding board” throughout the study. It was also the body from which expertise was formally sought at several junctions. TEAG members are listed below:

Alex Allen, MSA

AJA Consulting

Princeton, NJ

Gwen Bampfield-Wright, JD,MSW

The Access Network

Ridgeland, SC

Jack Geiger, MD

City University of New Work

School of Medicine

New York, NY

Russell Glasgow, PhD

AMC Cancer Center

Canyon City. CO

Barbara Israel, DrPH

University of Michigan

School of Public Health

Ann Arbor, MI

Deborah Jones-Saumty, PhD

CEO, American Indian Association

Talihina, OK

Meredith Minkler, DrPH

University of California at Berkeley

School of Public Health

Berkeley, CA

Jesus Ramirez-Valles, PhD, MPH

University of Illinois at Chicago

School of Public Health

Chicago, IL

Monika Suchowierska, MA/BCBA

University of Kansas

Lawrence, KS

Glenn White, PhD

University of Kansas

Lawrence, KS

Peer Reviewers

We gratefully acknowledge the following individuals who reviewed the initial draft of this report and provided us with constructive feedback. External reviewers comprised clinicians, researchers, representatives of professional societies, and potential users of the report. We would also like to extend our appreciation to David Atkins, MD, and Kaytura Felix Aaron, MD, from AHRQ for contributing peer review comments. Our peer review panel also includes six members of the TEAG: Alex Allen, Gwen Bampfield-Wright, Barbara Israel, Deborah Jones-Saumty, Mereidith Minkler, and Jesus Ramirez-Valles. Peer review was a separate duty for these individuals and not part of their commitment as TEAG members. All are active professionals in the field. The peer reviewers were asked to provide comments on the content, structure, and format of the evidence report and to complete a checklist. The peer reviewers' comments and suggestions formed the basis of our revisions to the evidence report. Acknowledgments are made with the explicit statement that this does not constitute endorsement of the report.

Alex Allen, MSA

AJA Consulting

Princeton, NJ

Ann Beal, MD

Commonwealth Fund

New York, NY

Paul Estabrooks, PhD

Kaiser Permanente

Denver, CO

Deborah Jones-Saumty, PhD

CEO, American Indian Association

Talihina, OK

Jesus Ramirez-Valles, PhD, MPH

University of Illinois at Chicago

School of Public Health

Chicago, IL

Sarena Seifer, MD

Community Campus Partnerships for Health

Seattle, WA

JoAnn Umilani Tsark, MPH

Research Director

Papa Ola Lokahi

Honolulu, HI

Gwen Bampfield-Wright, JD,MSW

The Access Network

Ridgeland, SC

Tom Bruce, MD

University of Arkansas

College of Public Health

Little Rock, AR

Barbara Israel, DrPH

University of Michigan

School of Public Health

Ann Arbor, MI

Meredith Minkler, DrPH

University of California at Berkeley

School of Public Health

Berkeley, CA

Victor Rubin, PhD

Policy Link

Oakland, CA

Shobha Srinivasan, Ph.D.

National Institute of Environmental Health Sciences

Research Triangle Park, NC

Tony Whitehead, PhD

University of Maryland

College Park, MD