What criteria do consumer health librarians use to develop library collections? a phenomenological study *

Objectives: The criteria for determining whether resources are included in consumer health library collections are summarized in institutional collection development policies (CDPs). Evidence suggests that CDPs do not adequately capture all of these criteria. The aim of this study was to describe the resource review experience of librarians and compare it to what is described in CDPs. Methods: A phenomenological approach was used to explore and describe the process. Four consumer health librarians independently evaluated cancer-related consumer health resources and described their review process during a semi-structured telephone interview. Afterward, these librarians completed online questionnaires about their approaches to collection development. CDPs from participating libraries, interview transcripts, and questionnaire data were analyzed. Researchers summarized the findings, and participating librarians reviewed results for validation. Results: Librarians all utilized similar criteria, as documented in their CDPs; however, of thirteen criteria described in the study, only four were documented in CDPs. Conclusions: CDPs for consumer health libraries may be missing important criteria that are considered integral parts of the collection development process. Implications: A better understanding of the criteria and contextual factors involved in the collection development process can assist with establishing high-quality consumer health library collections.

Providing consumers with quality medical information is important because receiving unclear information or misinformation can have serious implications for one's health management and decision making [6,12]. Consumer health libraries can intervene by providing educational and informational support to consumers [13]. This support can be tailored to meet the needs of individuals who experience barriers when seeking information, including those with lower educational attainment and non-English speaking groups [7,[13][14][15][16][17][18][19]. Librarian expertise is important to managing highquality consumer health libraries. Librarians review print and online resources to develop collections that include only those resources that are of high quality and suitable for consumers [4,20].
Most librarians utilize collection development policies (CDPs) to assist with their review processes [2]. However, anecdotal evidence suggests that the assessment of consumer health collections is not always straightforward. Prior to conceptualizing this study, the authors had an informal discussion with consumer health librarians about the suitability of specific resources for their respective library collections, and several unexpected review criteria were unearthed. In particular, when speaking about the suitability of a resource about sexuality and cancer, the authors were interested to learn that it might not be considered appropriate for library collections in every setting because of librarians' regard for the religious sensibilities of the dominant population of patrons. We were further interested to learn that a resource about death might not be permitted in the general library collection because of beliefs in the inauspicious nature of the written word ''death.'' Further discussion ensued as to whether these ''rules'' were included in the respective CDPs or whether they were just common knowledge. As the discussion continued, the authors discovered that many ''rules'' were not included in CDPs and that they represented, instead, the librarians' tacit knowledge. This prompted our interest in a formal study to describe the collection development process and to compare policies with processes.
The consumer health libraries at the University Health Network (UHN) in Toronto, Canada, and the Centro di Riferimento Oncologico (CRO) in Aviano, Italy, have CDPs that librarians use to inform collection development. The purpose of this study was to describe the resource review and selection experience of librarians and compare it to what was described in CDPs. The UHN and CRO have a memorandum of understanding for collaborative program development and research and sought to do this work together. The authors focused on cancer-specific resources for this study because cancer was the primary health topic covered by the participating libraries, and it was the topic area of expertise among the librarians at these sites.

Study design
Phenomenological research aims to capture and understand the lived experience of a group or an individual in the context in which the experience occurs [21]. Understanding these lived experiences can lead to the development of practices or policies that can include a more comprehensive description of the phenomenon [21]. We employed a phenomenological approach for this study so that we could derive a rich description of the collection development process directly from those who develop consumer health library collections in their everyday life roles. We focused on the description of the librarians' experiences of collection development rather than referring to processes described in library training text books [22].

Study participants
UHN comprises four hospitals that are located in the city centre of Toronto. Librarians from two of the UHN hospitals, the Princess Margaret Cancer Centre and the Toronto Western Hospital, participated in this study. CRO is one of eight comprehensive cancer centers in Italy and is located in Aviano. Similar to the UHN hospitals, CRO has its own library for patients and families.
Four librarians were recruited to participate in the study, two librarians from the UHN Patient & Family Libraries and two librarians from the CRO Patient & Family Library. The sample was a typical case sample, a sample that is expected to behave as most of their counterparts would. Though small, it was an acceptable size for phenomenological research [21]. The four librarians volunteered to participate in the study after a meeting between a librarian from one of the UHN libraries and a librarian from the CRO library, which brought to light differences in collection development practices between UHN and CRO.

Data collection
After receiving approval from the UHN Research Ethics Board, informed consent was obtained from all study participants.
To document the collection development process, each of the four librarians participated in a sixtyminute interview (Appendix A, online only), completed an online questionnaire (Appendix B, online only) about their respective library collections, and submitted their CDPs for review.
Interview. Study participants were given four consumer-oriented pamphlets related to cancer in either English or Italian, as appropriate. The librarians were asked to review each pamphlet to determine its acceptability for inclusion in their respective library collections. A study team member who did not participate in data analysis selected the pamphlets. Topics addressed in the pamphlets included diseasespecific information (multiple myeloma), medical treatment options (clinical trials), side effects management (sexuality and intimacy), and bereavement (sympathy).
Each librarian was given one week to conduct and document the review process for the pamphlets in the study package before participating in a telephone interview. Telephone interviews followed a semistructured script. The script began with a question about whether each pamphlet would be acceptable for inclusion in their libraries' collections. This was followed by questions designed to walk the interviewer (research assistant) through the review process that lead to the librarian's decision. To limit bias or misinterpretation, the CRO librarians completed the interviews in Italian with a certified Italian medical interpreter, and the UHN librarians completed the interviews in English with the research assistant. The interviews were recorded and transcribed. A certified Italian medical interpreter completed Italian transcription.
Questionnaire and document submission. Following the interview, librarians completed an online questionnaire about their library collections and how they were developed. The questionnaire took between fifteen and thirty minutes to complete. Finally, librarians submitted the CDPs from their respective libraries, one for the CRO library and two for the UHN libraries.

Data analysis
Three sets of data were analyzed: the interview transcripts, responses to the online questionnaire, and CDPs.
The goal of phenomenological research is to answer a specific research question in a way that lends itself to a comprehensive description of the phenomenon of interest. This is in contrast to more common qualitative methods, such as grounded theory where the objective is to generate a research question or theory. The study reported here had a specific research question-''What criteria do consumer health librarians use to inform collection development?''-and the authors wished to answer it in a way that would Consumer health librarians' library collection criteria provide a comprehensive description of the collection development process.
Data analysis followed a general phenomenological approach, as described by Moustakas and others [22][23][24]. The researchers began by identifying significant statements in the transcripts that depicted the phenomenon of collection development. These statements were grouped into themes or clusters. These clusters were used to construct (1) descriptions of the processes used for resource vetting and selection and (2) descriptions of the context in which these processes occurred. From these descriptions, a summary was written that described the resource review and selection experience of all four librarians.
The interview recordings were transcribed in their respective languages. The Italian transcripts underwent English translation by a certified medical interpreter for data analysis in English. Descriptions were extracted from the transcripts independently by each of the three researchers, and only those agreed upon by all three were used to formulate the final summary.
The review process documented in the interviews was compared to that outlined in the CDPs, as were the online questionnaire responses. Lastly, the CDPs were compared across sites. All study results were summarized and sent back to the participant librarians for review and validation.

The libraries, librarians, and collection development policies
The CRO librarians utilized a ''Quality Evaluation Grid'' to assist them in reviewing and selecting resources. The UHN librarians used a set of resource review criteria outlined in their CDPs to guide their resource assessments. The CDPs for all library sites were based on guidelines issued by various library associations. The UHN libraries based their policies on the National Library of Medicine Collection Development Guidelines, the Medical Library Association guidelines, and the HONcode from the Health On the Net (HON) Foundation, while collection development at the CRO was influenced by the Consumer and Patient Health Information Section (CAPHIS) of the Medical Library Association. A variety of specialists contributed to the construction of the CDPs, including patient education leads, information specialists, and librarians. However, clinicians (e.g., physicians, psychologists) were involved in actual collection development at only the UHN libraries. All library sites also had a process in place for reviewing and updating older resources through regular weeding practices and recommendations.
All library collections comprised books, pamphlets, videos, and online resources, with the addition of audio and electronic books in the UHN libraries. The librarians acquired approximately 80% of the collections by searching for new material, often identifying resources and references from recommendations issued by cancer-related organizations. Recommended resources from clinicians, patients, and families through book donations to the library made up 20% of both the UHN and CRO collections.
Although the processes were similar and the collections contained similar formats, UHN libraries were larger, serving approximately 100 patrons in a day and housing more than 2,550 pamphlets and 2,197 circulation materials in 1 location, and 800 pamphlets and 754 circulation materials in the other. In contrast, the CRO library served approximately 400 patrons in a year and housed about 120 resources.
The UHN and CRO librarians had different library education and training. UHN librarians reported receiving formal library or information-related training (i.e., the master of library science, master of information studies), while the CRO librarians gained their library knowledge and patient education expertise over time through hands-on experience and had undergraduate training from nonlibrary fields of study.

Resource review criteria and process
Examination of the collection development policies showed that UHN and CRO librarians utilized similar criteria in selecting appropriate resources for their library collections ( Table 1). The four major criteria in the CDPs of both UHN library sites and the CRO library are: (1) relevance, (2) credibility, (3) currency, and (4) accessibility. One extra criterion was contained in the CRO CDP that was not stated in the UHN policy, (5) interaction. Each of these criteria is described below. 1. Resources that were deemed ''relevant'' or ''appropriate'' had content that met the information needs of the library patrons. Because the patrons at the UHN and CRO libraries were mainly patients, family members, and health care staff, relevant topics were those that addressed patient support and health care services, organizations, education, and research. 2. ''Credible'' or ''reliable'' resources contained accurate and clearly sourced information, and were free of bias and conflicts of interest. Credible resources were those that were transparent about their missions or purposes. The content came from appropriately trained and qualified professionals, with authors' names and credentials or organizations clearly stated and easily identifiable. Contact information was also provided. 3. ''Currency'' or ''date'' referred to the timeliness of the information. Current resources were up-to-date and labeled with the date of creation or latest revision. 4. ''Accessibility'' or ''usability'' referred to how available the resource was to consumers in terms of language level and format. Resources written in plain language were favored above those that were not. Resources in video and audio formats were favored above dense books and text-heavy pamphlets. This criterion also included consideration of the accessibility of resources to consumers with disabilities (i.e., large print books). 5. A resource was considered ''interactive'' if contact information was included with the purpose of encouraging feedback and interaction between the resource's (pamphlet's) creators and users.

The librarians' collection development experience
After reviewing the four pamphlets included in the study package, both of the UHN librarians determined that only two of the four pamphlets would qualify for inclusion in their respective library collections. The two CRO librarians both determined that all four pamphlets would qualify for inclusion in their library collections. Analysis of the interview transcripts showed that all four of the UHN and CRO librarians made their selections using the review criteria articulated in their formal CDPs. However, transcripts revealed an additional eight criteria that were not included in any of the CDPs. Five of these were process related, and three were related to context. The processrelated criteria were (6) practicality, (7) duplication, (8) content review, (9) disclaimer, and (10) tone. The context-related criteria that influenced the librarians' decisions were (11) limitations and leniency, (12) proximity, and (13) certification label.
Practice-based criteria: process related 6. ''Practicality'' referred to the inclusion of practical information or tips that could assist patients and family members in how to apply the information, access services, or programs mentioned in the pamphlet or to find more information about a certain topic. All of the study participants agreed that practicality was not essential for all resources, but the addition of practical information, such as directions or side effect-management tips, are helpful. 8. During the vetting process, the librarians performed an informal ''content review'' on each of the four pamphlets. They scanned the content to assess how comprehensive a resource was and if any information should be flagged for a more in-depth review by the most appropriate content expert (for example, a physician, therapist, or social worker). 9. A ''disclaimer'' statement cautions readers that the information provided is not intended to substitute professional medical advice and recommends patients discuss their health routines with their health care teams before implementing any changes.
It does say that your doctor will advise you, so that's a good thing to put in…And it also says ''consult your doctor'' again, which is good. ''Don't act on this before consulting your doctor.'' So that's a bit of a disclaimer there.
Disclaimer was included in the criteria ''credibility'' by the CRO librarians. However, UHN librarians treated ''disclaimer'' as a separate vetting criteria and a crucial addition to a resource.
10. The ''tone'' of a resource referred to the wording or language style used in a resource. During their reviews, the librarians considered whether a pamphlet's tone complemented the type of information it was delivering.
[The Intimacy & Sexuality pamphlet] is well done, it is clear and simple. It [takes] an informal approach to the issue, it remains close to the reader while, for example, the text on clinical trials was more medical, and it had a more scientific cut, the text in this pamphlet has a more human approach.
Librarians also avoided inappropriate use of words or phrasing that might cause a negative response from the readers.
They…used the word degenerate. So usually when we're creating our own pamphlets that have been plain-language edited, we don't use such highbrow words. We try to tone it down. And this hasn't been.
Practice-based criteria: context related 11. ''Limitations and leniency'' were external factors that the librarians indicated had an influence on Consumer health librarians' library collection criteria collection development. Limitations could refer to the limited availability or the limited access a librarian might have with respect to resources on a certain topic. For example, the librarians mentioned that consumer resources on rare medical conditions were often lacking. Leniency referred to the compromises librarians made in vetting and selecting pamphlets given these resource or access limitations. Librarians alluded to compromising on certain vetting criteria in order to provide information that consumers need on uncommon topics. For example, one librarian decided to include the clinical trials pamphlet in the collection despite some unmet vetting criteria. 12. ''Proximity'' refers to the influence of nearby resource centers on the development of the library's own collection. The librarians might decide to complement their collections with the other services by sharing resources with other facilities and to refer patrons to them rather than adding material to their own collections. Alternatively, they might strive to create a comprehensive collection by including essential resources that are available elsewhere.
Publications on [bereavement] are not many. But for structural reason[s]-the next-door building is a hospiceand they have a library with their own material. So when patients are transferred to the hospice, few return here for informational support. 13. ''Certification labels'' are awarded to certain resources if they meet criteria outlined by an organization, such as the HON Foundation or ONCO label. The certification label distinguishes high-quality and transparent information, and reflects positively on the organizations that produce those resources. Certification assisted the librarians in their vetting processes and in resource selection because it indicated that a resource had already been vetted by others.
The absence of a certification label did not exclude a resource from the library collection but did prompt the librarians to be more thorough and cautious in their vetting, particularly if the resource was published by a new, unknown organization.
[The HON-certified logo] verifies the [information] is credible…health information. So this is one of the indications when we can trust or not trust a website if they're not sure. For this one, it's probably [a] credible resource based on all the information I found on the website. So there's not this stamp…That doesn't mean that that's not credible. It's just one of those things that I've noticed.

DISCUSSION
Study results showed that CDPs for the CRO and UHN libraries contained similar review criteria, but that librarians from both organizations also used unwritten criteria. CDPs from all organizations emphasized relevance, credibility, currency, and accessibility as core requirements of appropriate and high-quality resources. Similar findings are reflected in other publications that find commonly agreed upon criteria to include currency, indication of information source (credibility), reliability, relevance, and format (accessibility) [20,25]. However, despite similarities in the CDPs between the library sites, our study revealed that there are tacit criteria, not in the CDP, that also guide resource selection. Extracting textural and structural descriptions from the interview transcripts captured several tacit review criteria and external factors that might have influenced the librarians' decisions to include or exclude a particular pamphlet. As such, the CDP does not provide a complete description of the collection development strategies that these librarians used.
This observation is important because librarians may be unaware of their own tacit review criteria, or their value to others, and may therefore leave these criteria out of training manuals and documents like CDPs. Tacit knowledge is only transferred through extensive personal contact and regular interaction [26]; as such, unless these tacit criteria can be recorded in the CDPs, the actual collection development strategy of the library may not be followed.
There has been significant debate about the value of CDPs, and indeed, their value may be questioned if CDPs contain only the most basic or obvious review criteria [27]. However, if a protocol could be developed to augment CDPs with tacit knowledge gained through the experience of librarians, the worth of CDPs to novice or untrained librarians could be significant. The authors propose that this study can provide a framework for developing such a protocol.
Based on discussion with librarians prior to the start of the study, the authors anticipated extracting some interesting culturally nuanced tacit criteria from the interviews. None were revealed through the study. The authors suspect that this was because the study questions were framed to ask for the specific opinions of the participating librarians rather than to ask the librarians to reflect on the suitability of the resources in their particular contexts.

CONCLUSION
The purpose of a phenomenological study is to generate descriptive data that answer a specific research question. These data can then be used to further prospective investigation of the phenomenon [28]. Findings from this study present preliminary insight into (1) what the collection development process is for consumer health libraries and (2) how collection development policies are implemented in that process. Further investigation is needed to replicate and verify study findings. Further work could be undertaken to explore the implications of ''incomplete'' CDPs on the quality of collections or to audit collections to determine whether CDPs are followed and to explore whether other criteria surface from this type of review. There is also potential for further analysis into each practice-based criterion for a deeper understanding.