I had trepidations going into this panel session: panels have a reputation of being either really good or really bad, no inbetween. I am extremely happy to report that this panel fell into the ‘really good’ category. My colleagues who attended agreed and all of us believe that this kind of panel should be held every year at the MLA conference. It was that worthwhile.
The panelists were Lisa Kruesi, Carol Lefebvre, Terry Ann Jankowski and Rebecca Jerome. The preparation for the panel required each panelist to address two questions that have been submitted to them over the year and provide a presentation on their answers. (I am not clear whether this was because of a call, questions they receive due to their recognized expert status or some other submission process.) Moderated by Sarah Sutton with wit and let’s get to it-ness kept everything on track leaving room for discussion at the end.
Carol Lefebvre presented first on her two questions: (a) “What guidelines should I use to support systematic reviews?”; and (b) “When to use NOT in a systematic review strategy?”. For question (a), Lefebvre began by reviewing the issues around participating in a systematic review: managing expectations, the trapdoor of time and dedication when the researcher is a student; who pays (for librarian time in a systematic review); librarian expertise. She then proceeded to give an extensive list of resources to assist librarians in creating a process, based on standards, that works for them. A (very) brief list of these resources includes the Cochrane Handbook (http://handbook.cochrane.org/), Cochrane Standards, AHRQ Research Resources, and IOM systematic review guidelines/standards. The Peer Review of Electronic Search Strategies (PRESS) checklist was noted as an essential tool to ensure search quality. Duke University has a very nice guide that lists out the systematic review process and contains a lot of the aforementioned resources. Finally, participants were invited to consult and join the Cochrane Information Retrieval Interest Group (http://irmg.cochrane.org/) for collegial support for their systematic review process and questions.
For question (b), Lefebvre’s key points were to exercise extreme caution when using NOT, and to remind the audience that the use of limiters (e.g. human in PubMed) will exclude pre-indexed records. Cochrane advocates the use of the double negative to get around the inadvertent loss of those relevant records that are both human and animal studies. Lefebvre presented an example of her preferred style to catch those records that would otherwise be lost by a limit – consult her powerpoint when it becomes available.
Lisa Kruesi’s first question, submitted by a Canadian, was: (c) “How do I go about finding health statistics related to rehabilitation?” She gave an overview of grey literature resources pertaining to rehabilitation statistics e.g. NSW project: Rehabilitation Design, AIHW (Australian), and Qmentum Project (Rehabilitation Services) as sources to obtain Canadian and other national statistics to provide as benchmarks. For a general, where-to-get-started-with-health-statistics, she mentioned the NLM Health Statistics resources (http://www.nlm.nih.gov/hsrinfo/datasites.html; https://www.nlm.nih.gov/nichsr/usestats/). Kruesi’s second question (d)“What is best practice in documenting/package results?” was a nice follow up to Lefebvre’s question (a). Her recommendation – set up a template first. This template should have the following elements: the question; summary of the evidence, result count etc.; tailoring of the evidence (ordering the evidence by best evidence levels); search statement/strategy; format of search results (i.e. Endnote, Excel file); library branding (which may/should include a disclaimer); and then provide a method for the research to provide feedback on the service (link to evaluation form etc.). She further elaborated on the tailoring element that this requires critically appraising the literature (and that you only get better through practice which is quite labour-intensive for novice searchers). Ability to do this is often question-dependent and how much time you have. In her experience, for more skilled searchers it takes 2.5 -3 hours to synthesize and appraise the evidence for a question.
Terry Ann Jankowski gave a very nice review of indexing quirks in her answer to the question (e) “Why do I get different results for the same search strategy/term?”. There are several elements that impact result count, including year coverage, term selection, update schedule, and the default fields the database searches if you are not command line searching. She then used the example of frozen shoulder v. frozen shoulder* v. “frozen shoulder” to illustrate the differences in interpretation between Ovid Medline and PubMed. Moral of the story: use both keyword/free text and controlled vocabulary and know the database you are searching! AMENDED: There was a question from the audience to have her clarify what was a “pre-coordinated term”: as defined, pre-coordinated headings occur when a single heading combines 2 (or more) concepts. In the case of the concept for “pregnant teens”, the pre-coordinated heading is “pregnancy in adolescence”.
The final panelist Rebecca Jerome segued nicely from Jankowski and noted while she had two questions, one flowed from the other: (f) “How do you construct an effective command line search/how do you effectively use floating subheadings?”. She proceeded to expound on Kruesi’s question (d) by listing the systematic review process: conducting a thorough reference interview to establish a clear, focused question; doing background reading on the topic to get a understanding of the terminology used; performing preliminary searches to test how the various terms perform; framing a research draft which you then review with the researcher; document all the terms used including the discarded terms. This process explains why you chose one or more subject headings, why the choice to isolate a particular field (e.g. publication type) for a term, etc. within a concept set. I highly recommend the term documentation template Jerome showed in her presentation – I plan to consult it when her presentation becomes available.
Jerome then proceeded to clarify the use of floating sub-headings using a current search she is conducting. If, for example, the question contains the concept of treatment (in a broad sense) it is important to express that with the synonymous term therapy. The controlled term therapy is an exploding subheading, which includes all forms of therapy such as drug therapy. The term then is expressed as part of a command line search like this: therapy[sh]. The danger to using floating subheadings is they can be imprecise.
There was a question from the audience of whether it was possible to copyright a search strategy as the requester had experienced a researcher taking “ownership” of a search the requester had performed. Sarah Sutton asked the audience how many of us had experienced a similar situation (there was several of us in the audience that had, including myself). I had volunteered that my colleague Angela Osterreicher had devised a citation shell at the top of the cover letter template we use. In the cover letter we invite researchers to please use the citation so as to provide an easy way to acknowledge our work in a paper or presentation. Sarah had asked me to provide it here for your information; I cannot say it prevents people from denying our contribution but I hope it gives them pause before doing so.
Literature search: insert question title here. Librarian/Informationist’s name. (Date of search or day sent). Location: Institution Name.
Literature search: Models of referral/collaborative practice between primary and specialist physicians. Szwajcer, Andrea. (September 25, 2012). Winnipeg, MB: University of Manitoba Health Sciences Libraries.