Portal de Periódicos da CAPES

Fahrenheit 451?: a ‘burning question’ on the evidence for book withdrawal

2009; Wiley; Volume: 26; Issue: 2 Linguagem: Inglês

10.1111/j.1471-1842.2009.00844.x

1471-1842

Andrew Booth,

Pharmaceutical Practices and Patient Outcomes

Andrew Booth With space at a premium in most libraries, particularly those located in a health service context, the health librarian frequently encounters pressures to withdraw outdated items from their library book stock. Should this be left to the experience and expertise of the ‘librarian on the spot’? Or is there, perhaps, a body of evidence to inform this process—making it more systematic and rigorous? This question came to my mind recently. Indeed, it reawakened distant memories of a Regional Stock Withdrawal Initiative which I coordinated in South East Thames.1 As 20 years might be construed a decent enough interval between uses of a terrible punning title (although some readers will disagree!) I have allowed this book-burning analogy to rise phoenix-like within the collective health librarian consciousness. Before formally considering the evidence base, we should at least acknowledge that ‘books are different’. First, given the variety of topics covered by your average health library, it is not desirable to implement some standard cut-off date regardless of discipline. Books do not have a ‘sell-by date’ to be implemented across every shelf. Nevertheless, particularly within a health context, information within books is consistently volatile. As Antman and colleagues demonstrated,2 textbooks are particularly inefficient in capturing current knowledge. It is not exaggerating to say that outdated health information can kill, although fortunately the frequency with which such instances are reported (note, I do not say occur!) is low enough not to require our taking out indemnity insurance. We should also appreciate that updating of editions of popular textbooks is almost exclusively determined by commercial considerations and not by a philanthropic desire to refresh prevalent medical knowledge at appropriate intervals. A second factor relates to recognition, to misquote John Donne, that ‘no book is an island entire of itself’. Each item in a book collection must be considered in relation to the collection as a whole. A library with many alternatives to a particular book considered for withdrawal may be less ‘forgiving’ than one where no such alternatives exist. Collection development policies also imbue an individual book with greater collective significance when it is designated to an area of the collection labelled as ‘core’. The existence of such policies provide an important antidote to unwelcome connotations of ‘weeding’ that evoke vacuuming up Cochrane's classic Effectiveness and Efficiency3 and replacing it with a pocket guide to evidence-based medicine! Finally, we must acknowledge factors in favour of book withdrawal. Some suggest that weeding of a book stock has a positive effect on the use of the collection that remains, although this phenomenon remains contested. Furthermore a ‘do nothing’ option is not cost neutral—librarians rarely consider that an item lying unused on a shelf is costing space, light, heat and cleaning, not to mention opportunity costs.4 Of course, this must be weighed against the time and effort required to remove it from the shelves and, sometimes, the equal time and effort it takes to remove it from the catalogue. Deferring a decision to a later date may betray a lack of professional courage rather than reflecting prudent management. In my first professional post, I relegated old volumes to the plentiful shelves of our postgraduate centre's breakout rooms, only to discover later that I had removed any possibility that these items would be used, although they still figured on the catalogue. I had also perpetuated an ongoing waste of space and added a supernumerary interim stage to the process of throwing the volumes out for good! Recently, I re-encountered a posting to the lis-medical Jiscmail discussion list dating back about a year and a half: ‘Hello all, I am looking at formalizing my collection management policy in the light of both a monster weed we are undertaking and work to shift various aspects of our collecting into a more electronic world’.5 Putting aside the B-movie connotations of the ‘monster weed’, I was reminded that one of the exemplar ‘burning questions’ identified by Eldredge6 was: ‘When weeding a hospital library collection, how does one determine the publication year before which materials can be weeded from the collection?’. I therefore formulated the following hypothetical scenario. A health librarian working in a multidisciplinary education centre library built 15 years ago is concerned at the lack of bookshelf space remaining for new acquisitions. The librarian decides to examine the evidence base for book withdrawal, specifically within a health library context. To initiate the search, he or she formulates the following SPICE question:7 A search of Library and Information Science Abstracts (lisa) using ‘withdrawals’, of cinahl using ‘weeding’ and ‘withdrawals’ and the Library, Information Science and Technology Abstracts (lista) database using ‘Weeding’ and ‘DISCARDING of books, periodicals, etc.’ complemented by a simple targeted Google Scholar search (e.g. ‘weeding and health and libraries’) reveals only six relevant items published in the decade between 2000 and 2009 (Table 1). Such a disappointing yield attests that book withdrawal in health libraries, arguably more critical given the volatility of health information, has received less attention than in the more ‘politicized’ domains of public and school libraries. This Using Evidence in Practice feature is not intended to offer mini-reviews of the evidence base, so the interested reader is referred to individual items in Table 1 and the reference list. Instead, we will illustrate more generalizable observations about the quantity and quality of the evidence base. Including the specific setting of a health library, considered an important feature of a focused question,7 was particularly detrimental for the number of items retrieved. Clearly, there is a significant trade-off between relevance and availability of the evidence. If we broaden our question to other contexts, which do we include or exclude on the basis of relevance? While excluding public and school libraries may be justified, how do we allow for possibly confounding disciplinary differences if we admit small academic libraries? More significant is the mismatch between the type of material that I was seeking to address my question and that retrieved from my ‘quick and dirty’ search. I was looking for a methodology by which I might implement a weeding policy. Overall items I retrieved were rich on ‘know-how’ but poor on evidence. They also tended to focus on macro-level considerations for the process as a whole rather than on the micro-level operational detail of conducting the weeding. Interestingly, one item draws parallels with a physician's diagnosis and treatment, making it easier to argue that weeding is legitimately viewed as a facet of evidence-based practice.8 Several items were retrieved through multiple database routes (lisa/lista/cinahl). However, two of the most useful items, a report in a regional newsletter of an MSc Project11 and possibly the profession's most unobtrusive randomized controlled trial in a research section newsletter,9 were more difficult to identify. The former was located through a Google search and not covered by any of the formal databases. The latter was located not through library-specific databases but through cinahl. This account from a poster has not subsequently been published in the peer-reviewed library literature. As this is a randomized controlled trial, I subsequently conducted a known item search to see if it was included in the Cochrane Library—with no success. A search on ‘weeding’ on the Cochrane Library yields only a single record—on seasonal allergic rhinitis! Regarding the contexts of the included articles, two were multiple accounts of a single dental library initiative,8,10 one related to a nursing library12 and two were academic health sciences libraries.9,13 The single most relevant item, the MSc project account,11 was conducted in the UK National Health Service so scored best on immediate relevance. However, how do we handle the signal-to-noise balance, given that the newsletter is unlikely to have been peer reviewed? On the positive side, four of the six articles (from Interim,11 Hypothesis9 and two from Journal of the Medical Library Association12,13) were easily accessed in full-text format without charge. However, the item most likely to hold the potential to resolve my scenario, the Masters dissertation itself,14 would likely prove more challenging to obtain. Despite the unsatisfactory nature of my evidence quest, unexpected benefits were derived during the course of the search. For example, I am reminded that book withdrawal should not be tackled in isolation, but should be fully integrated within the wider context of collection management.15 This includes more strategic issues such as identifying core collection areas in the first place, as well as ‘preventive’ strategies, such as including a policy for gifts and donations. Such features make the subsequent process of weeding much easier in the long run. More importantly, a key theme I had not expected, emerged relating to the importance of communication with readers about what you are doing and why you are doing it.16 Arguably, this aspect of ‘public relations’17 is far more critical to the success of the weeding process than technicalities regarding cut-off dates and areas for exemption. Indeed, we might go further to suggest that involving users in the decision-making process, including eliciting suggestions for new acquisitions, could turn around perceptions of the weeding event from a negative ‘purge’ to a much more positive ‘collection refreshment’. Certainly, we would not expect users simply to anticipate plumes of black, or white, smoke at the library windows to signal the professional staff's collegial activity!18 This brief case study of a burning question from a realistic health library setting vividly illustrates limitations of the existing evidence base. Not that the evidence base is unhelpful. Arguably, if we were to consider a wider range of generic articles,19,20 and even the classic textbook on the topic (assuming it has not been thrown away!),21 we might obtain the necessary ingredients for a successful stock withdrawal policy, together with a significant fund of practical wisdom and know-how. In fact, the most interesting item of research, as opposed to more discursive ‘evidence’, demonstrates that adding multiple entry points to a catalogue record did not translate into an increase in circulation.22 No doubt, too, a wider search would yield more systematic methodologies for stock withdrawal, such as the CREW method used in public libraries.23 Such methodologies may well translate to a health library context. It is clear too that, if our decision making incorporates consideration and review of even a small body of the available evidence, we will achieve a more holistic, considered and justifiable outcome than reliance alone upon librarian expertise, experience and intuition. This is illustrated in our recognition of the user perspective, even where this is only acknowledged as requiring better communication concerning what is going on. What is apparently missing is what is most desired—a rigorous, context-sensitive methodology that is easy to implement and that can simply be taken ... off the shelf!