Reference management software for systematic reviews and meta-analyses: an exploration of usage and usability

Various software programs have been adopted by, and specifically developed for, authors of systematic reviews. These tools range from statistical software to comprehensive systematic reviews programs, such as the Cochrane Collaboration’s Review Manager (RevMan) software. Reference management software programs, including EndNote, RefWorks and Zotero, are one such tool.

Reference management software, also known as bibliographic software, citation management software, or personal bibliographic file managers is “any software product used for storage and retrieval of bibliographic records” [1]. First developed in the 1980s, these programs were initially marketed to researchers as a means of creating online indexes of personal print-article collections [2–4]. As electronic databases, such as MEDLINE, became generally accessible and more easily searchable, researchers began to use this software to maintain databases of all research relevant to their fields of interest [3].

Numerous reference management programs are currently available. Although all programs facilitate the capture, organization, and elimination of duplicate records from electronic database searching, they vary with respect to cost, overall functionality, and networking capabilities. Products such as EndNote, Papers, and RefWorks are licensed or sold outright, while others (e.g.: Mendeley and Zotero), are available at little or no cost to the user. While some, such as EndNote and Reference Manager, run on single-station computers, many others, including RefWorks, Mendeley and Zotero, are web-based. Single-station software usage is generally not affected by website time-lags, down times, or record limits, all of which may impinge on the usability of web-based products. That said, the benefits of these web-based programs include the ability to store reference databases on secure servers, and access databases from multiple computers or other electronic devices. Web-based programs also provide users with enhanced networking functions that readily support the sharing of records among researchers [3, 5].

The identification, collection, and organization of relevant studies are instrumental to the successful completion of systematic reviews. To this end, a “rigorous data management plan” is essential [6]. Many guides to undertaking systematic reviews, including the Cochrane Handbook for Systematic Reviews of Interventions, recommend reference management software as a means of assisting in the organization and selection of component studies for inclusion in these reviews [6, 7].

In the context of systematic reviews, reference management programs facilitate the capture and organization of studies identified through electronic database searching, the identification and elimination of duplicate records from multiple database searches, the transfer of references to Cochrane RevMan and other systematic reviews software, and the accurate citing of references within manuscripts [8, 9]. Thus, an author’s decision to use, or not use, this software may impact on the accurate reporting of the number of studies reviewed for inclusion and exclusion in a systematic review.

The purpose of this study was to determine the extent to which authors are using reference management software to produce systematic reviews; identify which programs are used most frequently and rate their ease of use; and assess the degree to which software usage is documented in published studies.

We reviewed the full text of systematic reviews published in core clinical journals. We surveyed the corresponding authors of included studies to verify and supplement information in published reports, and gather frequency and ease-of-use data on specific reference management programs. This study received ethics approval from the University of Calgary’s Conjoint Health Ethics Board.

From a total of 163 systematic reviews identified in ACP Journal Club, 111 authors, representing 122 papers, were included in this study. Based on eligibility criteria, 41 papers were excluded from this study. Reasons for exclusion were: an inability to determine or verify the email address of the corresponding author (n = 30); study authors only searched 1 electronic database to identify studies to include in their systematic review (n = 8), or no electronic databases were searched (n = 3). Of the 111 surveys that were emailed to authors, 78 surveys were returned, for a response rate of 70.27%.

Of the 62 authors who reported using reference management software, 4 (2 EndNote and 2 Reference Manager) stated that they would be switching to a different software package in future reviews. Stated reasons for switching software packages included: incompatibility issues between Reference Manager and Mac computers, difficulty importing references into EndNote, and, in one instance, a cost savings that would result from replacing EndNote with Zotero.

Forty-seven authors (60.3%) responded to our final open-ended survey question asking for comments on the usability of reference management programs. Thirteen authors remarked on the ease of use and overall importance of their chosen tool. As one respondent stated:

“I don’t think you can write manuscripts for prestigious journals without one of these bibliography management programs…..there will always be recommended revisions, or a rejection… in which case you’ll have to reconfigure your references. To do [this] by hand is too cumbersome and time-consuming”.

A number of authors also identified specific challenges associated with these programs. Among these were: record errors that occurred when downloading references from electronic databases, such as MEDLINE (n = 4), difficulties in identifying and deleting duplicate records from reference management databases (n = 2); PC/Mac-incompatibilities for users of EndNote and Reference Manager (n = 3); errors in journal output styles (n = 3); difficulties in transferring reference databases from one software package to another (n = 2), and delays in accessing RefWorks databases (n = 2). Five respondents indicated that they were not only using this software to collate studies identified through database searching, but also to record reviewers’ decisions with respect to the inclusion or exclusion of studies.

Our study reveals that most (but not all) authors of systematic reviews are using reference management software tools. Despite this, only a small minority are reporting this usage in their published studies. Further, there appears to be no apparent relationship between choice of reference management software and perceived functionality or usability. Nor was software choice associated with the degree to which usage was reported or not reported in published studies.

Currently, there are more than 28,000 scholarly journals in active publication [10]. Collectively, these journals publish approximately 1.8 million articles per year [10]. Given the trend, across disciplines, towards integrating evidence into daily practice, it is unsurprising that an increasing number of systematic reviews are being funded and produced. Previous research has demonstrated that authors typically search multiple electronic databases to identify studies relevant for systematic reviews [11–14]. Estimates of overlap between these databases have ranged between 8% and 60% [11–14]. Software programs that can facilitate the collection, organization and removal of duplicate studies from database searches are invaluable in the production of large-scale systematic reviews.

The most significant finding from this study was the disparity between stated and reported software use. Few (4.8%) authors who had used reference management software reported this information in their published manuscripts. As a point of contrast, 76.9% of the authors in our sample reported on the statistical software packages used to analyze study results. A recent study on the use of statistical software in health services research also found that a significant number of authors were including statistical software information in their published manuscripts [15]. This discrepancy in reporting may be the result of an absence of guidelines which specifically address this aspect of reporting. Whereas The International Committee of Medical Journal Editors’ Uniform Requirements for Manuscripts Submitted to Biomedical Journals recommend that authors “specify the computer software” when describing the statistical methods they employed to analyze study data, no mention is made, here or elsewhere, of the importance of reporting on the use of reference management software [16].

Although the PRISMA checklist for the quality reporting of systematic reviews does not specifically suggest that authors report on reference management software usage, its use can impact the number of unique studies identified and reported in a systematic review [17]. A failure to identify duplicate records captured in database searches may result in an over-reporting of irrelevant studies. Similarly, an over-reliance on reference management software to identify and remove duplicate records from a reference database may cause relevant studies to be overlooked. As such, reference management software can have a real impact on the quality of a completed review.

Whereas numerous studies have been published citing the benefits of reference management and other sofware programs in the production of systematic reviews, the degree to which these programs have been adopted by the research community is largely unknown [2, 3, 6, 9, 18–22]. In 2007, Senarath published a research letter on the reference management software experiences of 22 researchers in Sri Lanka [8]. Although informative in terms of highlighting the functions and perceived advantages of this software, to our knowledge, ours is the first study to explore reference management software usage and reporting among authors of systematic reviews.

Our study has caveats and limitations. Although we were able to achieve a strong response rate of 70% from our email survey, a telephone survey could have yielded a higher response rate. Furthermore, researchers who declined to participate in our study may have different perspectives on reference management software usage than those who participated. Secondly, as mentioned previously, we limited our study sample to clinical reviews published in ACP Journal Club. This sampling frame may have resulted in the identification of published systematic reviews that have a better-than-average quality of reporting. Thirdly, an analysis of systematic reviews produced in languages other than English, and other disciplines, such as education or social work, could yield different findings. Finally, with the exception of one open-ended question, our survey did not seek to identify perceived benefits and deficits with respect to individual software programs.

An overriding caveat to our study, and its implications in terms of the importance of reporting reference management software usage, is that there is, as yet, no firm evidence to suggest that reference management software can have a positive impact on the quality of completed reviews. That said, the “value of a systematic review depends on what was done, what was found, and the clarity of reporting” [17]. As stated by Moher and colleagues “the reporting and conduct of systematic reviews are, by nature, closely intertwined” [17]. As such, a detailed transparent plan for the identification and selection of component studies, including the use or lack of use of reference management software, is an important element of any review protocol. It can reflect the methodological expertise of the authors, and the transparency, reproducibility and, ultimately, the quality of the study that has been produced.

Despite being underreported, reference management software is frequently used by authors of systematic reviews. The transparency, reproducibility and quality of systematic reviews may be enhanced through increased reporting of reference management software usage.