Evidence-based policy and practice on community benefit: information support for developing a bibliography.

“Community benefit” (CB) is described in a revenue ruling from the Internal Revenue Service (IRS) that justifies the tax-exempt status of nonprofit hospitals and requires these hospitals to provide community services. According to the instructions for IRS form 990, schedule H, CB spans a wide array of potential activities, varying from clinical services, to research, to community outreach and workforce education [1]. As the nation continues to make changes to the health care delivery system, the contributions of nonprofit health care organizations to the community will likely remain a charged topic, one warranting empirical evidence to enhance understanding. Rigorous studies about CB can provide a more complete picture of its structure, management, and effectiveness, as well as its impact on the social, economic, and health status of the community. Frontline hospital administrators are striving to make the implementation of CB programs driven more by evidence of effectiveness than by dollar requirements or historical trends. Health care executives, policy analysts, and researchers all express an increased interest in data pertaining to CB. 
 
Librarians will be asked to assist researchers and policy analysts in finding existing research studies and compiling future data. However, because the majority of the existing literature on CB pertains to practice or regulations, the definitions are complex, and the published literature is widely dispersed, this task can be immense and daunting. This article describes one approach to searching for research evidence about CB and recommends methods that will make such searches easier in the future.


INTRODUCTION
''Community benefit'' (CB) is described in a revenue ruling from the Internal Revenue Service (IRS) that justifies the tax-exempt status of nonprofit hospitals and requires these hospitals to provide community services. According to the instructions for IRS form 990, schedule H, CB spans a wide array of potential activities, varying from clinical services, to research, to community outreach and workforce education [1]. As the nation continues to make changes to the health care delivery system, the contributions of nonprofit health care organizations to the community will likely remain a charged topic, one warranting empirical evidence to enhance understanding. Rigorous studies about CB can provide a more complete picture of its structure, management, and effectiveness, as well as its impact on the social, economic, and health status of the community. Frontline hospital administrators are striving to make the implementation of CB programs driven more by evidence of effectiveness than by dollar requirements or historical trends. Health care executives, policy analysts, and researchers all express an increased interest in data pertaining to CB.
Librarians will be asked to assist researchers and policy analysts in finding existing research studies and compiling future data. However, because the majority of the existing literature on CB pertains to practice or regulations, the definitions are complex, and the published literature is widely dispersed, this task can be immense and daunting. This article describes one approach to searching for research evidence about CB and recommends methods that will make such searches easier in the future.

BACKGROUND
Since 2007, the Saint Louis University (SLU) School of Public Health (SPH) has been promoting rigorous policy and health services research on CB. In an effort to move the field more toward evidence-based policy and programming, SLU SPH planned a conference, ''Community Benefit: Moving Forward with Evidence-Based Policy and Practice,'' held on June 27, 2009, in Chicago [2]. The purpose of this invitational meeting was to develop the initial agenda for CB research and policy analysis for the next three years.
Given this orientation, the planning committee invited the SLU SPH liaison librarian to compile a bibliography with CB-related research literature published between 1990 and the present for the conference. The librarian then worked with an SPH research assistant between October 2008 and February 2009 to develop a comprehensive search on research literature pertaining to ten aspects of CB. These aspects included: (1) community leadership and health status; (2) community needs and population segments; (3) economics, finance, charity care, and bad debt; (4) ethics; (5) evaluation and program effectiveness; (6) governance; (7) quality and performance measurement; (8) health professions education; (9) state and federal policy; and (10) structure and staffing. These ten aspects are derived from IRS form 990, schedule H, and are used by the practitioners in nonprofit hospitals to organize their CB portfolios.
The immediate goal of the literature search was to develop a representative bibliography on CB research and policy studies that could be used by conference participants to prepare for the meeting. The longerterm goal was to make widely available a bibliography on CB that would expedite the work of researchers and policy analysts, as well as of health administrators, health educators, and health sciences librarians.

Database identification
The conference sought to develop a research agenda on CB, and thus research articles were the major component of the bibliography. Online databases covering the field of health care administration were the initial resources for the search. The databases available at the SLU Medical Center Library during the period when the search was conducted included MEDLINE (OvidSP platform), Health Management (ProQuest), Global Health (EBSCO platform), ABI/ Inform (ProQuest), Academic Search Premier, Sociological Abstracts, and Web of Science. To identify the optimal databases, a pilot search was conducted on the economics, finance, charity care, and bad debt aspect of CB across all the initial databases. OvidSP MEDLINE and ProQuest Health Management covered more relevant publications than other databases, and most citations retrieved from other databases also appeared in the results lists of these two databases. Therefore, OvidSP MEDLINE and ProQuest Health Management were used to retrieve academic research publications on CB for the conference.

Identification of search terms
A search with the keyword ''community benefit'' from OvidSP MEDLINE and ProQuest Health Management yielded about 200 citations, most of which were relevant to CB. However, as CB is such a broad concept and the phrase ''community benefit'' may not always appear in relevant articles or citations, a concept search, using subject headings, was advisable. To locate research literature on each of the 10 aspects of CB, the overall search was broken down into 10 individual searches. In each individual search, ''community benefit'' was the main concept and 1 aspect of CB represented the other concept. The challenge was that no existing subject headings in either Medical Subject Headings (MeSH) or the ProQuest subject list directly represented the concept of ''community benefit''; this was the case as well for many of the 10 aspects. Therefore, multiple subject headings were used to represent each aspect. The subject terms identified from the 2 databases and their relationships are presented in Table 1.

Formulation of search strategy
When the actual searches were performed in OvidSP MEDLINE, an ''Exploded'' search was applied to all the MeSH terms. In searching for each aspect of CB, the subject headings chosen to retrieve articles on the concept of ''community benefit'' was searched with a Boolean ''OR'' with the keyword ''community benefit'' and then searched with a Boolean ''AND'' with subject headings representing one aspect of CB. As CB is a provision required by the IRS in the United States, the results were combined with ''United States.'' As the bibliography focused on research literature published between 1990 and the present, English language, journal articles, and 1990 to 2009 limits on language (lg), publication type (pt), and publication year (yr) were applied at the end of the search.
A similar strategy was used with the advanced search feature in the ProQuest Health Management database. ''Subject'' was selected from the drop-down index list when subject terms were entered. ''Document language'' and ''Document type'' were selected to narrow down each search with the input of ''English'' and ''Journal article.'' A limit on publication year was applied in the ''Date range'' box.
Using ''community needs and population segments'' as an example, the conceptual search strategy in OvidSP MEDLINE was as follows: ((EXP community health services OR EXP communityinstitutional relations OR EXP tax exemption OR EXP uncompensated care OR community benefit$.mp.) AND (EXP hospitals)) AND (EXP health services needs and demand OR EXP needs assessment) AND EXP United States AND ''English''[lg] AND ''Journal Article''[pt] AND ''1990-2009''[yr]

Management of search results
About 500 articles on the 10 aspects of CB were retrieved from the 2 databases. After carefully reading the abstracts of the retrieved articles, 135 relevant articles were selected and sorted into each of the 10 aspects. The relevance of the retrieved articles was determined by the agreement between the librarian and the research assistant, with the consensus of the conference planning committee and the faculty member heading the conference. Based on the retrieved literature, 2 additional topics were created during the classification process: general community benefit and examples of community benefit programs.

BIBLIOGRAPHIC COMPILATION
After organizing the retrieved research articles, the planning committee was asked to review the list of articles to suggest additions or changes that would enhance the conference discussion. White papers and reports that were not part of the published academic literature retrieved from the databases being examined were added to the list.
To make it easy for attendees to prepare for the conference and, subsequently, to facilitate research, a website was developed to post the bibliography [3]. Articles on the website were organized first according to aspect, then according to access. The access classification separated the articles into those that were freely available online and those that had access restrictions due to licensing provisions.
The bibliography provided background information prior to the conference and facilitated conference attendees learning about different aspects of and perspectives on CB. Therefore, it served the intended purposes of provoking thoughtful discussion at the conference and received compliments from the planning committee and the conference attendees.

DISCUSSION
The current search strategy yielded research publications on CB sufficient for the purpose of the conference. Although searching with the keyword ''community benefit'' retrieved quite a few relevant articles, searching with subject headings produced more relevant articles. A comparison was made of the search on the economics, finance, charity care, and bad debt aspect of CB in OvidSP MEDLINE between searching with a Boolean ''AND'' with the keyword ''community benefit'' and searching with a Boolean ''AND'' with the subject headings representing the concept of community benefit. The results showed that searching with the selected subject headings yielded about forty relevant articles that were not retrieved by only using the phrase ''community benefit'' to search. This mini experiment indicated the importance of applying a concept search to retrieve CB research literature. It also indicated the need to add the term ''Community Benefit'' to MeSH so that only one subject heading, rather than several subject headings, would be required for a comprehensive search. Search filters [4] might also be developed that interface with the PubMed search engine to ease the search process for each specific aspect of CB.
As time allows, further refinement to the search should be done to produce a more comprehensive bibliography. Only online databases were searched, and these were limited to journal articles, resulting in a somewhat restricted view of the CB literature. White papers, reports, gray literature, and government documents should also be included in the bibliography, as they provide solid background information and are useful for designing research. Although some white papers and reports were added to the bibliog-

CONCLUSIONS
This project provides an example of a successful collaborative relationship that health sciences librarians can create with their liaison schools. In addition to receiving regular information services from the liaison librarian, the faculty and students of the liaison school also recognize a liaison librarian's expertise in organizing, managing, and retrieving information on a complex subject. Recognizing the need to have solid literature as the basis for research and policy studies, and now more than ever realizing the complexity of the subject of CB, the SLU SPH faculty will continue to involve the liaison librarian as an integral part of a CB research team. CB is emerging as a new field of academic and policy research. The project described here demonstrates the invaluable role a liaison librarian can play in shaping a new field of study and represents an initial foundation about the study of CB literature that should be expanded by librarians to facilitate future study of CB.