Aitken, Elizabeth M. MLS; Powelson, Susan E. MLS; Reaume, Renée D. MLS; Ghali, William A. MD, MPH
The concept of clinical librarianship, whereby librarians are integrated into the health care setting and care process, has been practiced since the 1970s.1,2 There is a growing body of literature on clinical librarianship, including a systematic review1 and a 2010 article discussing new paradigms for clinical informationists.3 Brettle and colleagues4 provide an excellent systematic review on the evaluation of clinical librarianship studies to date, suggesting the need for reduced bias in such research and examination of specific outcomes of including clinical librarians on the health care team. Studies by King,5 Marshall,6 and Klein and colleagues7 all found a direct effect of clinical librarian involvement on patient care decisions and length of stay. Health professionals have been found to be more willing to seek information when collaborating with a clinical librarian, and clinical research teams changed their information-seeking behaviors with guidance from clinical librarians.3,8 A review of the literature, however, clearly indicates that there is a need for more evidence-based studies of the effect of clinical librarian interventions.3 Our study adds to the literature by providing such a controlled assessment. We demonstrate positive effects of a clinical librarian program on trainee attitudes, self-reported search tendencies, and, most important, clinical decision making.
Our objective was to place a clinical librarian on an internal medicine teaching unit whose members were providing direct patient care and to perform a prospective, controlled assessment of the effect of the clinical librarian's involvement on care providers and the decisions that they made. Grefsheim and colleagues3 point out that a controlled comparison would add to our understanding of the effect of clinical librarians on care. Our study aims to fill this gap. Using pre and post surveys in both intervention and control groups of providers, we measured self-reported information retrieval practices, general satisfaction, and attitudes toward having a librarian on the team, as well as the associated effect on clinical decision making.
Setting and study design
The intervention took place on the general internal medicine inpatient medical teaching unit of a large, tertiary care teaching hospital in Calgary, Alberta, Canada in 2009. Our study involved a direct comparison between an intervention group, whose members were provided access to a clinical librarian, and a concurrent control group, who did not receive the intervention.
All medical residents and clinical clerks (i.e., final-year medical students) on the inpatient general internal medicine service were eligible to participate in this study. Program administrators preassigned the residents and clinical clerks as usual to their four-week rotations through the inpatient general internal medicine service. This was blind to the librarian, and the administrators assigned teams using their usual criteria independent of this project. Attending physicians on the service changed every two weeks. The intervention and control teams each typically contained one attending physician, four residents, two or three clerks, and a pharmacist. We presumed the prior, evidence-based library training of the two cohorts to be similar.
The protocol for this study was reviewed by the conjoint health research ethics board, Faculty of Medicine, University of Calgary.
The control group was a team of residents, clerks, a pharmacist, and attending physicians with identical structure to the intervention group, working on the same unit with the same pool of incoming patients assigned to either the control or the intervention group. There was no crossover between the control and intervention groups.
We assigned an experienced health science librarian (E.A.) with a masters in library sciences from an American Library Association–accredited program and more than 20 years' experience in health libraries to the intervention team for a six-month period from January 26 through July 26, 2009. The librarian had previously rounded at a different hospital and had significant experience providing instruction in evidence-based information retrieval in lecture and small-group settings. The librarian spent an average of 10 to 12 hours per week with the team, typically attending morning report as well as accompanying the physician team to the bedside for intake rounds relating to new patient admissions, and to a team conference typically held in the afternoons when the team's patient list was reviewed and discussed. There were also some occasions on which the librarian undertook formal classroom teaching. The same individual filled the role of clinical librarian throughout the project, though other librarians weighed in on the basis of their experience and for feedback on process.
The clinical librarian's role on the intervention team, as described above, had three key components. The first was an educational role. The clinical librarian had multiple opportunities for teaching interventions, ranging from formal teaching of evidence-based medicine retrieval concepts and access techniques, to group discussion of information search techniques, quick demonstrations, and brief, one-on-one tutorials, depending on demand and the team's schedule. The second component of the clinical librarian role was service. The clinical librarian was called on to perform mediated searching and document retrieval on request. The third component was an anticipatory role. The librarian would provide clinical information to the resident or clerk in charge of a specific patient with specific medical issues and corresponding health information needs. To fulfill this role, the clinical librarian attended case presentations and bedside history then searched for relevant materials. She then typically provided the information to the appropriate resident or clerk the same day whenever possible.
The clinical librarian performed all searches at the same bank of computers where teams congregate to enter order data and update charts. By ensuring close proximity with the team, the librarian was typically available to answer further questions. Although most literature search questions related to direct patient care, some team members requested assistance for research projects and information alert services. Among all search-related questions, there was a strong emphasis on point-of-care tools and clinical queries. The librarian relied on online resources including Medline (Ovid), PubMed, Embase, Cochrane Database of Systematic Reviews, American College of Physicians' Information and Education Resource, Dynamed, First Consult, Natural Standard, Clinical Evidence, and UptoDate.
Measure of interest
To assess outcomes of the intervention, we measured participants' responses to a paper survey for the intervention and control groups both just before and just after their rotation. We gave all participants a letter outlining the purpose of the project, assurances of anonymity, and contact information to which participants could direct any questions.
The items on the pre and post survey for the intervention group were identical, but the post survey included additional questions about the effect and perceived value of the librarian role. Questions were multiple-choice, closed-ended, Likert scales, or open-ended (Supplemental Digital Appendix http://links.lww.com/ACADMED/A64). The intervention group members were asked to report how they used and viewed the different types of information support. The survey placed special emphasis on the librarian role. We asked further questions to determine team member attitudes toward having the clinical librarian as part of their team. Finally, participants were asked whether they changed a diagnosis or treatment plan based on new information provided by the librarian or on new skills acquired through contact with the clinical librarian. In addition, all survey responders were given opportunities to provide additional, open-ended comments.
The control team completed pre and post surveys that were identical to the intervention group's, and respondents could opt not to answer questions that spoke to their experience working with the clinical librarian.
We calculated simple descriptive statistics based on the survey data, as well as proportions for various responses to survey questions.
Description of participants
There were 34 responses from the intervention team. These included 20 residents, 12 clinical clerks, and 2 other team members who did not specify their level of training. There were 16 responses from the nonintervention team, including 9 residents, 5 clinical clerks, 1 pharmacist, and 1 unknown. During the six-month study period, 42 medical trainees rotated through each team, giving an overall response rate of 50 out of 84, or 59%.
Receptivity to clinical librarian concept
The intervention and control respondents strongly agreed (22 of 34 intervention, 65%; 9 of 19 control, 47%) or agreed (12 of 34 intervention, 35%; 6 of 19 control, 32%) on the post survey with the statement, “I think that the ability to find current, high-quality information is very important to patient care.” These responses indicated a generally receptive and enthusiastic environment amongst all the residents and clerks on both intervention and control teams.
The intervention survey measured participants' attitudes about the librarian intervention by asking participants to indicate their agreement with a statement about the usefulness of working directly with a librarian on the patient care team. Only 12% (3/26) disagreed before the intervention that working directly with the librarian would be useful, and that number dropped to 6% (2/34) after the intervention. The number of respondents indicating that they strongly agreed or agreed after the intervention slightly increased from 24 of 26 (92%) before the intervention to 32 of 34 (94%) after the intervention, indicating that participants' receptivity to the addition of a librarian to the team was generally positive. Thirty-two of 34 (94%) intervention respondents indicated that they would recommend a clinical librarian service to colleagues, further confirming the participants' approval of the clinical librarian role on the care team.
Types of librarian support used
Our assessment showed a balance between using the librarian's skills as support and obtaining search strategy advice. Seventy-nine percent (27/34) of respondents asked for search strategy advice, 68% (23/34) asked the librarian to perform a mediated search, and 74% (25/34) learned from the librarian as part of a group discussion. Only 6% (2/34) did not use the librarian's assistance.
Changes in evidence retrieval tendencies
The intervention surveys included questions designed to describe changes in participants' evidence retrieval tendencies (Table 1). Identical questions were posed on pre and post surveys about participants' likelihood of using specific categories of resources to answer a patient care question. There was an increase in the number of subjects likely to use a journal article or PubMed and a corresponding decrease in those unlikely to use those same resources. Similarly, the number of respondents indicating “maybe” about their willingness to use other databases such as the Cochrane Database of Systematic Reviews or Embase increased from 11 of 29 (38%) before the intervention to 15 of 34 (48%) after the intervention. Responses from the control team, meanwhile, did not reflect the same increases in likelihood of using a medical journal article or PubMed to answer a patient care question. The intervention team was markedly more likely to consider using a librarian's assistance than were their control counterparts.
Effect on clinical decision making
Although responses to individual questions relating to resource use showed only subtle changes in self-reported use of specific resources or confidence in identified search tasks, we posed additional questions to determine the effect of the librarian intervention on diagnostic and therapeutic decisions. We asked two key questions (Table 2): (1) Did literature you obtained for yourself using skills learned by having a clinical librarian on your team ever help you change a diagnosis or a treatment plan? and (2) Did literature that was obtained for you through the support provided by the clinical librarian ever help you change a diagnosis or a treatment plan?
The results of these two important questions indicated a strong, positive effect of having a clinical librarian on the team. Notably, 30 of 34 (88%) respondents on the intervention team reported changing a treatment based on new information skills taught, and 27 of 34 (79%) changed a treatment plan based on a search done by the librarian. The corresponding numbers of responses related to diagnosis change are smaller but still notable (Table 2).
Participants were able to provide qualitative comments in response to two questions. The first was “Would you recommend a clinical librarian service to colleagues? Why or why not?” The responses cite librarian efficiency, lack of practitioners' time to search, and focus on evidence as prime reasons to recommend the intervention. The second question was “What was the most important thing you learned from working with the clinical librarian?” The responses ranged from increased awareness of available resources and integration of searching into daily practice to improved strategy.
Our intervention took a twofold approach. We wished to determine the perceived value and effect of having a librarian on an internal medicine teaching team for mediated search support and for teaching retrieval of evidence-based resources at the point of care. The respondents were receptive to the addition of a clinical librarian to the team, and a large majority of the intervention group indicated that they would recommend a clinical librarian service to colleagues, an indication of a high perceived value of having a librarian on the team. Quantitative and qualitative responses indicate that the team members found the services useful. We also assessed changes to diagnosis and treatment plans, and the notable findings in that area are a strong indication of success for the intervention because it seems to have affected patient care, a significant positive finding.
Whereas our study is a single-center assessment of a clinical librarian intervention, our study setting is that of a rather typical academic tertiary care center that we consider to be representative of many medical teaching units throughout the developed world. We add to the evidence base because we were able to take advantage of an established parallel team structure in the center to conduct a prospective study with a concurrent control group. As mentioned earlier, the librarian spent a significant amount of time with the intervention team at the bedside hearing the case presentation firsthand during morning walk-rounds and was, therefore, more able to respond to or anticipate information needs. The new physicians in our study group were open to and interested in evidence-based information and receptive to having librarians work with them. The self-reported change to diagnosis and treatment plans is a significant and compelling affirmation that clinical librarians can be a valuable addition to the patient care team and that education and support are key pillars of any clinical librarian program.
Our findings raise the question of what constitutes a good clinical librarian program. Esparza's9 review of this topic described bedside rounds, interdisciplinary conferences, morning report, and roving reference (where the librarian informally circulates on a medical unit to offer ad hoc searching and teaching assistance) as four typically successful aspects of a clinical librarianship program. Sargeant and Harrison,10 meanwhile, add clinical meetings and guideline development to the list. Flynn and McGuinness11 suggest that information skills and informatics training, research collaboration, and current awareness are additional valuable opportunities. Sackett and Straus12 propose an “evidence cart” for providing access to information to be used quickly and efficiently during the care process.
On the basis of our experience, a clinical librarian is valuable as part of the care team. The study by Vaughn13 evaluated clinicians' search strategies and asked clinicians whether they would have asked questions had the librarian not been present at rounds. Among respondents to this question, 74% said no. Thus, there seems to be value in this educational and supportive role. Searching for clinical information in real time or “near real time”14 is critical, and participants in our study found the formal education in this subject provided by the librarian at the start of rotations to be useful. The residents and clerks reported being pleased to discover where to find resources, decision support tools, and clinical queries.
Librarians who understand the clinical environment can make better resource choices and help vendors shape their products to suit the needs of physicians. The type of inquiry-based learning that is common to residency programs lends itself to integrating evidence-based medicine and information-based support. The morning intake of new patients was very time consuming for the whole team, but the process allowed for the presentation of the whole story for a clinical question and was an incredible learning opportunity for the librarian to become more familiar with the language and process of acute care, leading to increased understanding, development of relationships, and a sense of belonging on the care team.
These educational and supportive programs like the one we tested need adequate staffing and resources to adapt to changing clinical paradigms and to fully incorporate clinical librarians at the point of care. Our survey results indicate providers' receptivity to the presence of a clinical librarian on the care team, desire to use evidence well, and willingness to alter treatment plans and diagnosis based on information facilitated by a clinical librarian.
A caveat to our study finding is that we focused narrowly on the role of the clinical librarian. Others have focused on informationists.3 There are recognized distinctions between these roles. The broader and more specialized informationist role has even greater potential to inform providers and enhance care, such that our study may actually underestimate the potential positive effect of having an “information specialist” in the care setting.
Although our results are encouraging, our study has limitations. As mentioned above, this is a single-institution evaluation, which may or may not be applicable outside of an academic situation. The survey instrument relied on self-reporting to gauge the effect of the intervention, but respondents' reported treatment plan changes and high willingness to recommend the service to colleagues is not ambiguous. Also, we surveyed a relatively small number of residents and clerks. Though quite typical for this type of evaluation, the modest number of respondents is an acknowledged limitation. It is possible that members of the intervention team could have shared information obtained through the clinical librarian with the control team, thus contaminating our results. Although possible, we do not think that this would affect the validity of our conclusions because any information sharing related to the project would have caused the groups' survey responses to be more similar. The changes in patient care practices are self-reported, and we have assumed that these changes are positive ones, based on better informed clinical decision making. The Donabedian15 model of quality of care predicts that an improvement in process will lead to an improvement in care and, consequently, improved patient outcomes. We would submit that an addition of a clinical librarian to the patient care team is an improvement in process. The measurement of a clinical librarian's effect on other, more tangible outcomes of patient care would be an interesting basis for further study. A final limitation and caveat to our findings is that our study did not explore economic considerations or business models for implementing clinical librarian initiatives on a broad scale. This needs to be addressed in future research and managerial processes.
In conclusion, our study has added to the knowledge base on clinical librarian initiatives by providing evidence from a prospective controlled study. We demonstrate that the clinical librarian intervention that we tested led to positive effects on provider attitudes, self-reported provider information-retrieval tendencies, and, notably, clinical decision making. The three key elements of intervention—librarian teaching, on-demand searching, and anticipatory provision of information to clinicians—seem to positively affect clinical care and may ultimately provide benefits to the patients cared for by the intervention team. Future research should focus on the business model and economic considerations surrounding widespread implementation of such a clinical librarian intervention.
Ethical approval was applied for and a comfort letter was received from Dr. Glenys Godlovitch, chair of the conjoint health research ethics board, Faculty of Medicine, University of Calgary.
Preliminary results were presented at the Canadian Health Libraries Association Conference, June 7–11, 2010, Kingston, Ontario, Canada, and at the fifth International Clinical Librarian Conference, June 13, 2011, Birmingham, United Kingdom.