Article
Grey Literature Searching for Health Sciences
Systematic Reviews: A Prospective Study of Time Spent and Resources Utilized
Ahlam A. Saleh
Assistant Librarian Arizona
Health Sciences Library
University of Arizona
Tucson, Arizona, United
States of America
Email: asaleh@ahsl.arizona.edu
Melissa A. Ratajeski
Reference Librarian
Health Sciences Library System
University of Pittsburgh
Pittsburgh, Pennsylvania, United States of America
Email: mar@pitt.edu
Marnie Bertolet
Assistant Professor
Graduate School of Public Health
University of Pittsburgh
Pittsburgh, Pennsylvania, United States of America
Email: mhb12@pitt.edu
Received: 4 June 2014 Accepted: 7
Aug. 2014
2014 Saleh, Ratajeski, and Bertolet.
This is an Open Access article distributed under the terms of the Creative
Commons‐Attribution‐Noncommercial‐Share Alike License 4.0
International (http://creativecommons.org/licenses/by-nc-sa/4.0/),
which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly attributed, not used for commercial
purposes, and, if transformed, the resulting work is redistributed under the
same or similar license to this one.
Abstract
Objective – To identify estimates of time taken to search grey literature in
support of health sciences systematic reviews and to identify searcher or
systematic review characteristics that may impact resource selection or time
spent searching.
Methods – A survey was electronically
distributed to searchers embarking on a new systematic review. Characteristics
of the searcher and systematic review were collected along with time spent
searching and what resources were searched. Time and resources were tabulated
and resources were categorized as grey or non-grey. Data was analyzed using
Kruskal-Wallis tests.
Results – Out of 81
original respondents, 21% followed through with completion of the surveys in
their entirety. The median time spent searching all resources was 471 minutes,
and of those a median of 85 minutes were spent searching grey literature. The
median number of resources used in a systematic review search was four and the
median number of grey literature sources searched was two. The amount of time
spent searching was influenced by whether the systematic review was grant
funded. Additionally, the number of resources searched was impacted by
institution type and whether systematic review training was received.
Conclusions – This study characterized the amount of time for conducting systematic
review searches including searching the grey literature, in addition to the
number and types of resources used. This may aid searchers in planning their
time, along with providing benchmark information for future studies. This paper
contributes by quantifying current grey literature search patterns and
associating them with searcher and review characteristics. Further discussion
and research into the search approach for grey literature in support of
systematic reviews is encouraged.
Introduction
A properly conducted systematic review summarizes the
evidence from all relevant studies on a topic concisely and transparently
(Cook, Mulrow, & Haynes, 1997). The searches to support these reviews need
to be extensive often including extended searches of the grey literature.
Grey literature can be described in a number of ways,
but commonly has been defined by the 1997 Luxembourg Convention on Grey
Literature definition as literature: “which is produced on all levels of
government, academics, business and industry in print and electronic formats,
but which is not controlled by commercial publishers” (Farace, 1998, p.iii). In
2004, at the Sixth International Conference on Grey Literature, a postscript
was added to further expand on the “commercial publishers” aspects of this
definition. In recent years, the definition has stimulated new discussion due
to changes in the environment such as the evolving landscape of information
dissemination and the introduction of new avenues of scientific communication.
According to two-thirds of respondents in the Grey Literature Survey (Boekhorst, Farace, & Frantzen, 2005),
“Grey Literature is best described by the type of document it embodies” (p.6).
Some examples of grey literature include: reports, conference abstracts,
dissertations, and white papers (GreyNet International, 2013).
Systematic review support seems to be of growing
interest to health sciences information professionals. A 2013 survey of
librarians and directors about emerging roles of biomedical librarians found
that support of systematic reviews was one of the top six most common reported
new roles (Crum & Cooper, 2013). Furthermore, in a 2013 systematic review
of the literature from 1990-2012 on changing roles for health sciences
librarians, systematic review librarian was identified as one of the newer
roles in the field (Cooper & Crum, 2013).
In recent years, education opportunities relating to
systematic review searching have transpired. This is no surprise, given that
information professionals planning to get involved in systematic reviews would
need to familiarize themselves and learn more about the process and
specifically about the search process, as it is distinct from routine
literature searches. As an example of
these emerging education opportunities, a search of the Medical Library
Association’s Educational Clearinghouse (http://cech.mlanet.org/), found eight
Continuing Education (CE) courses with the words “systematic review” in the title,
and there were at least three CE courses found listed on the Canadian Health
Libraries Association website (http://www.chla-absc.ca/node/119). Systematic
review as a topic is also emerging in library science curricula as can be found
in select content covered in the Certificate of Advanced Study in Health Sciences
Librarianship (HealthCAS), Reference Services and Instruction in Healthcare
Environments course previously offered through the University of Pittsburgh.
The University of Alberta, School of Library Information Studies has a course
entitled “Systematic Review Searching”, and the Texas Woman’s University
medical library curriculum planned a
course for systematic reviews that would
be launched in Spring 2014 (C. Perryman, personal communication, July 7,
2013).
In 2011, the Institute of Medicine (IOM) released
standards for systematic reviews which indicate that researchers should “take
action to address potentially biased reporting of research results” (p.84). To
address this bias, Standards 3.2.1 and 3.2.4 call for the inclusion of grey
literature searches in all systematic reviews and handsearching of selected
journals and conference abstracts (Institute of Medicine [IOM], 2011). These
standards may lead to wider acceptance of including grey literature searching
in systematic review methodology. As librarians
become increasingly integrated into systematic reviews, as called for by the
IOM Standard 3.1.1 which specifically states “Work with a librarian or other
information specialist trained in performing systematic reviews to plan the
search strategy”, they must be prepared to search the grey literature or at
least provide guidance on resources and search strategies (IOM, 2011,
p.84). Locating grey literature can
often be challenging, requiring librarians to utilize a number of databases
from various host providers or various websites, some of which they may not be
familiar with or aware of. Additionally,
searchers may need to spend time learning various search interfaces and the
nuances of each resource, such as how to download references or if the search
query boxes have term limits (Wright,
Cottrell, & Mir, 2014).
Investigation into the grey literature search process
for systematic reviews may reveal useful information that can be applied by
information specialists planning and preparing for systematic review searches.
There is limited information on the time it takes to search grey literature in
support of systematic reviews and we are not aware of any studies which relate
searcher or systematic review characteristics to either time spent searching
the grey literature or which grey literature resources are selected for the
search. Therefore, we sought to explore these aspects of the process of grey
literature searching in support of systematic reviews.
Literature Review
Acceptance of the inclusion of grey literature in
systematic reviews has varied over time. A 2006 survey showed approximately 90%
of systematic reviewers and approximately 70% of editors felt grey literature
probably or definitely should be eligible for inclusion in systematic reviews,
while a prior 1993 survey showed that 78% of meta-analyst and methodologist
respondents felt that unpublished material should definitely or probably be
included in systematic reviews, and that only 47% of journal editors felt this
way (Cook et al., 1993; Tetzlaff, Moher, Pham, & Altman, 2006). The biggest
concerns were about the lack of peer review and quality of the studies found in
the grey literature. However, other studies detail the benefit of including
grey literature in systematic reviews (Crumley, Wiebe, Cramer, Klassen, &
Hartling, 2005; Savoie, Helmer, Green, & Kazanjian, 2003). Two Cochrane
reviews further support why it is important to search for grey literature. A
2007 Cochrane systematic review on the use of grey literature in meta-analyses
of randomized trials found that non-grey literature trials tended to be larger
and showed an overall larger treatment effect when compared to grey literature
trials (Hopewell, McDonald, Clarke, & Egger, 2007). Another Cochrane
systematic review on time to publication for results of clinical trials found
that positive result trials tended to be published earlier than negative or
null result trials and positive result trials were more likely to be published
than negative or null result trials (Hopewell, Clarke, Stewart, & Tierney,
2007). Therefore, when conducting searches for systematic reviews searching for
grey literature may be used as a means to minimize the introduction of bias
such as publication and time-lag bias.
There is a time cost for including a grey literature
search in a review. How much time is difficult to estimate however and, because
librarians in an academic or hospital role are often juggling other
responsibilities, time management is crucial.
There seems to be limited research specifically reporting on the time
taken to conduct literature searching for systematic reviews. If time is
reported, it is often grouped with other tasks such as article retrieval and
screening, or the search time is listed as one number, not denoting the time
differences for various resources such as PubMed vs. a grey literature resource
such as ClinicalTrials.gov. In an examination of 37 meta-analyses, Allen and
Olkin (1999) found that the average systematic review took 1139 hours to
complete (with a range of 216 to 2518 hours). Of this time, 588 hours accounted
for protocol development, searches, retrieval, abstract management, paper
screening, blinding, data extraction and quality scoring and data entry. In a
single meta-analysis conducted by Steinberg et al. (1997), a description of the
time to complete various systematic review tasks, including screening,
extracting data, and writing the manuscript was reported. They estimated the
total hours for conducting the review to be 1046 hours (26 weeks) of which 24
hours was used to conduct the literature search. Guise and Viswanathan (2011)
estimate that it would take 1-4 weeks to run comparative effectiveness review
searches.
Greenhalgh and Peackock (2005) reported the time taken
for electronic database searches for their systematic review (including
developing the search, refining, and adapting to other databases) as
approximately two weeks of a librarian specialist’s time. This article also
looked at how productively the time was spent. The two weeks of a librarian’s
time “yielded only about a quarter of the sources - an average of one useful
paper every 40 minutes of searching” (Greenhalgh & Peacock, 2005, p. 1065).
Greenhalgh and Peackock (2005) go into further detail and compares electronic
searching with handsearching. A handsearch of 271 journals took approximately a
month of time resulting in 24 papers that were included in the final report –
“an average of one paper per nine hours of searching” (Greenhalgh &
Peacock, 2005, p. 1065).
Using traditional electronic databases to search the
literature does not always identify all relevant studies. This can be due to a
number of reasons including lack of appropriate indexing terms, lack of
indexing all sections of a journal, or research methods not being fully
described in the abstract (Hopewell, Clarke, Lefebvre, & Scherer, 2007). Therefore
handsearching of journals or conference proceedings would be particularly
relevant to include in the systematic review search methodology. The Cochrane
Handbook for Systematic Reviews of Interventions defines handsearching as
follows: “Handsearching involves a manual page-by-page examination of the
entire contents of a journal issue or conference proceedings to identify all
eligible reports of trials” (Cochrane Collaboration, 2011, Section 6.2.2.1).
Handsearching of conference proceedings could be considered as a type of grey
literature searching.
Studies other than Greenhalgh and Peackock have also
looked at the time required for handsearching. Additional reported time ranged
from 30 minutes per journal issue to 45 minutes - 3 hours per year of a title
and this may vary based on subject matter (Adams, Power, Frederick, &
Lefebvre, 1994; Armstrong, Jackson, Doyle, Waters, & Howes, 2005; Croft,
Vassallo, & Rowe, 1999; Jadad, Carroll, Moore, & McQuay, 1996).
There are a variety of studies that have examined how
characteristics such as search experience, relate to search quality, search
speed, search effectiveness and the search process (Al-Maskari & Sanderson,
2011; Debowski, 2001; Fenichel, 1981; Hsieh-Yee, 1993; Kuhlthau, 1999). Many of
these studies seem to focus on varying experience levels within end-user
groups. Though, Tabatabai and Shore (2005) explored how experts (highly
experienced librarian professionals), intermediates (final year master of
library information studies students) and novices (undergraduate teachers)
searched the Web. Significant differences in patterns of search among the
different groups were found in cognition, metacognition, and prior knowledge
strategies (Tabatabai & Shore, 2005). Previous literature also demonstrates
interest in exploring different experience levels in relation to performance of
a systematic review. A study (Riaz, Sulayman, Salleh, & Mendes, 2010)
reporting on systematic reviewers noted that new systematic reviewers
experienced problems with time taken to conduct the review, defining the
research question and inclusion/exclusion criteria, and data management that
were not faced by experienced systematic reviewers. Given this insight from the
literature, it seems useful to consider subject characteristics as variables
that may potentially impact outcome searches.
Aims
The purpose of this study was to explore grey
literature searching for health sciences systematic reviews. More specifically:
Methods
Study Recruitment
University of Pittsburgh Institutional Review Board
(IRB) approval was obtained to conduct the study. Participants were recruited
through listservs, social media and email contacts. A total of 19 listservs,
including information professional types and those whose subscribers were
thought to have an interest in systematic reviews, were used for recruitment.
Social media sources included posting on Facebook, posting on popular librarian
blogs, and having fellow librarians post tweets. The prerequisite for study
enrolment was: the participant must currently be embarking on a literature
search to support a systematic review or plan to in the near future. However,
searching should not have taken place prior to enrolment. The searcher did not
have to be a librarian but had to be the person responsible for building the
searches and running them in each database used.
Data Collection
A survey instrument was developed and pilot tested on
a small sample which included two systematic review course instructors (a
clinician and a biostatistician) and five information specialists. Revisions to
the survey forms were made based on feedback. The survey forms were distributed
in two parts (see Appendix for surveys). Part one of the survey collected
demographic information about the searcher and their systematic review
experience and was required to be submitted directly after study enrolment.
Part two of the survey was provided to participants after they submitted part
one. Return of the part two form was expected upon completion of the systematic
review searches. This part of the survey collected information about the
systematic review including topic, population, and if the systematic review was
grant funded. Participants were also asked to document the name of the
resources searched, platform or vendor of the resource, resource URL when
applicable, and the time taken to search the resource (electronic database,
website searching or handsearching). Reported time was to account for choosing
terminology, developing the strategy, refining and running the search.
In order to minimize the potential of bias in the
searching, it was not disclosed to subject participants that this was a survey
specifically focused on grey literature searching in systematic reviews.
Rather, participants were asked to include all resources searched for the
systematic review including handsearching. Other means of searching for
studies, such as citation tracking (i.e. snowballing) or contacting key authors
or experts were not specifically requested in our data collection form.
Email reminders were sent twice to participants who
did not submit their completed data forms.
Data Analysis
From the completed surveys, a list of the resources
participants used in their systematic review searches was compiled. For some
resources various interfaces were used to complete the search i.e. OVID MEDLINE
vs. PubMed. In these cases the resources were classified as the same despite
the interface used with the exception of Cochrane Library and the Centre for
Review and Dissemination (CRD) resources. Some users searched both Cochrane
Library and the CRD version of the databases: Database of Abstracts of Reviews
of Effects (DARE), NHS Economic Evaluation Database (NHSEED) and Health
Technology Assessment (HTA), for the same review. Cochrane Library individual
resources were all grouped under “Cochrane Library,” and if a CRD search
platform was also used, that resource was categorized separately as “CRD
database.”
Once the list of resources was compiled, each resource
was labeled as grey literature or non-grey literature. To categorize the
resources, the librarian authors used the 1997 Luxembourg definition of grey
literature, reviewed the database content, contacted the database content
producers as necessary, and used their own expert opinion. Resources that
predominately included literature from journals were categorized as non-grey.
Resources that included citations mostly from book chapters, theses, reports,
conference materials or other type of grey literature were categorized as grey.
Because many resources may cover both grey and non-grey literature for these
resources, assignment of grey vs. non-grey was based on the authors’ estimation
of which type of content was the majority. Handsearching was denoted as a grey
literature resource because conference proceedings were primarily handsearched
for participant systematic reviews which included handsearching. Once
categorization was completed, the number of grey literature vs. non-grey
literature resources and the time taken to search each type of resource was
tallied.
The six outcomes analyzed were the number of grey,
non-grey and total literature resources along with the amount of time searching
each of those resources. These outcomes were compared across the searcher and
the systematic review characteristics. Continuous variables were compared
across groups using a Kruskal-Wallis test instead of a t-test because
non-normality in some of the variables violated t-test assumptions. Boxplots
comparing the amount of time searching the literature across groups were
graphed on the log scale due to a few extremely large values. Kruskal-Wallis
tests were performed on the log scales for the amount of time searching the
literature across groups to be consistent with the boxplots. Sensitivity
analysis without the log transformations yielded similar results.
Results
Out of 81 initial respondents, 17 (21%) completed both
parts of the study and were included in the data analysis. Nineteen respondents
were excluded because they did not meet the study prerequisites. The remaining
respondents withdrew from the study or did not complete both parts of the
survey. Of the 17 final participants, 15 reported that their primary
professional role was a librarian/information professional.
Searcher Characteristics
Most study participants were from an academic
environment (Figure 1). Hospital was reported by 24% and Other, which included
one non-government agency and one independent research company, made up 11% (n=
2) of participants’ institution. The participant country representation was
mostly comprised of United States (US), Canada, and the United Kingdom (UK)
with a few additional countries represented by Other (Figure 2).
Figure 1
Type of institution where searcher was employed (n= 17).
Figure 2
Country where searcher was employed (n= 17).
Figures 3 and 4 show that most study participants had
greater than 10 years of experience in their profession and at least 5 years of
experience in assisting in systematic reviews. The number of systematic review
searches that respondents had contributed to ranged from 0 to greater than 50
(Figure 5).
Figure 3
Searchers’ years of experience in profession
(n= 17).
Systematic Review Characteristics
Half of the searchers responded that they would be a
co-author on the systematic review that they completed the searches for, while
the others responded they would not (33%) or were not sure (17%).
With regards to study population, the survey asked about
age of the target population. Thirty-nine percent of the systematic reviews
focused on adults, 17% pediatric, and 44% focused on both. The breakdown for
the methodological focus of the systematic reviews included Therapy (33%),
Diagnosis (11%), Prognosis (11%), Other (39%), and Unsure (6%).
An attempt was made to explore whether any relationships existed between systematic review topic
and time spent searching and resources used. However, because of the wide range
of topics for a small sample size, we were not able to analyze the data in a
meaningful way.
Most of the reviews were completed under the guidance
of a systematic review producing entity (56%; n= 18). Also the majority of
systematic reviews were not grant funded (67%) (Figure 6).
Figure 4
Searchers’ years of experience contributing to systematic reviews (n= 17).
Figure 5
Number of systematic reviews
searcher has contributed to in the past (n= 17).
Figure 6
Grant funded systematic
reviews (n= 18).
Time Spent Searching and Number of Resources Searched
Tables 1 and 2 show the time survey participants
reported searching resources for their systematic review and the mean number of
resources used per systematic review.
Using the Kruskal-Wallis test we examined whether the
time spent searching resources for a systematic review or the number of
resources used for a systematic review search varied by the characteristics of
a systematic review (grant funded, under guidance of a systematic review
producing entity, etc.) or of an individual searcher (institution, country,
systematic review training, etc.).
Figures 7-10 use boxplots to visualize the
statistically significant (p ≤ 0.05) findings. In the boxplots the thick dark horizontal
line represents the median value. The bottom and top of the box are the first
(Q1) and third (Q3) quartiles where 25% of the data is below Q1 and 75% is
below Q3. The “whiskers” extending from the boxes show the spread of the data
with outliners represented as dots (as in Figures 9 and 10).
The time spent
searching can be obtained by exponentiating the log (minutes) in Figures 7 and
8. Specifically, whether or not the systematic review was grant funded was
associated with the amount of time spent searching for both grey literature
(median [Q1, Q3] for grant funded: 544.5 [211.4, 1339.4] minutes and not grant
funded: 66.7 [36.6, 109.9] minutes, p=0.03)
and non-grey literature (median [Q1, Q3] for grant
funded: 1480.3 [365.0, 3294.5] minutes and not grant funded: 270.4 [164.0,
492.7] minutes, p=0.05) (Figures 7 and 8). The number of resources searched for
the systematic reviews varied by the searchers’ institution (median [Q1-Q3] for
academic: 2.5 [2.0, 4.5]; hospital: 1.0
[1.0, 1.3]; and Other: 8.5 [7.8, 9.3], p=0.02) and whether the searcher
received systematic review training (median [Q1-Q3] for trained: 3.0 [3.0,
5.0]; and untrained: 6.0 [4.0, 7.0], p=0.045) (Figures 9 and 10).
Resources Searched
Figures 11 and 12 show the most common grey and
non-grey literature resources reported as being used by the 17 study
participants.
Discussion
The results of this study found that the average total
time spent searching electronic databases and handsearching the literature for
a systematic review was 24 hours with a range of 2 to 113 hours with 50% of the
participants reporting spending less than 8 hours. Our study also more
specifically identified time taken to search the grey literature for a
systematic review. All systematic reviews reported (n= 18) included some form
of grey literature searching. The average time taken to conduct the grey
literature search was approximately 7 hours, with range of 20 minutes to 58
hours, with 50% of the participants reporting spending less than 1.5 hours. The
grey literature search represents 27% of the total time of the literature
search on average, with 50% of the participants spending 20% or less of their
time searching grey literature.
Table 1
Time Spent Searching Resources for Systematic Review
Searches
|
Time Range (minutes) |
Time Mean (minutes) |
Time Median (minutes) |
Quartiles (Q1,Q3) |
Grey Literature Resources |
20-3480 |
395 |
85 |
(45,240) |
All Resources |
96-6780 |
1457 |
471 |
(255,2104) |
Table 2
Number of Resources Utilized for Systematic Review
Searches
|
Resources Range (number) |
Resources Mean (number) |
Resources Median (number) |
Quartiles (Q1,Q3) |
Total Resources Searched |
3-27 |
9 |
8 |
(5,10) |
Grey Literature Resources Searched |
1-14 |
4 |
2 |
(1.25,5.5) |
Non-Grey Literature Resources Searched |
2-13 |
5 |
4 |
(3,5.75) |
Figure 7
Time spent searching grey literature for a systematic
review (n= 17).
Figure 8
Time spent searching non-grey literature for a
systematic review (n=17).
Figure 9
Number of grey literature resources used for a
systematic review (n= 18).
Figure 10
Number of non-grey literature resources used for a
systematic review (n= 18).
Figure 11
Top grey literature resources searched for a
systematic review (n= 18).
Figure 12
Top non-grey literature
resources searched for a systematic review (n= 18).
When reporting time spent searching, survey
respondents were asked to account for choosing terminology, developing the
strategy, refining, and running the search. Depending on how the instruction
“running the search” was interpreted response times could be skewed low.
Although not specifically stated, we aimed to have the time requested on the
survey to capture the time required for searchers to navigate and learn the
nuances of the databases used for grey literature searching as they prepare the
search approach. Grey literature electronic resources are often limited and
crude in search capability. Truncation of terms may not be possible, search
boxes may be limited in the number of characters they accept, little or no search
help documentation may be provided and no export feature may be available.
Because of these limitations the searcher may spend more time searching the
database than anticipated. Grey literature resources are also not routinely
searched by the average health sciences librarian for everyday work. Because of
this, database unfamiliarity may also require searchers to spend more time
searching the grey literature.
As previously mentioned, all reviews reported included
some form of grey literature searching. An average of four grey literature
resources were searched per review, with a range of one to 14 resources and
with 50% of the participants reporting using one or two grey literature
resources. The number of resources selected for the grey literature search may
be restricted due to time constraints if there is pressure to complete the
systematic review within a short time period. Also resource selection may be impacted
by resource access which may be limited depending on institutional
subscriptions. We did not ask searchers what, if anything, impacted or limited
the number of resources selected, especially for grey literature. We did ask if
a resource was purchased as a one-time paid subscription for purposes of a
search for the systematic review, however no respondents reported this to be
the case. Locating resources to search for grey literature is important and
worth noting as one of the challenges with searching the grey literature. It is
possible that problems locating grey literature resources impacted the number
of resources used by participants in the study.
As shown in Figures 11 and 12, MEDLINE and Cochrane
Library were used in all the reviews reported in this study. Following the
Cochrane Library, the next most commonly used grey literature resources were
the World Health Organization International Clinical Trials Registry Platform
(WHO ICTRP) (n= 6) and the metaRegister of Controlled Trials (mRCT) (n= 5).
It is not surprising that the Cochrane Library,
produced by the Cochrane Collaboration, led the ranking of the grey literature
resource searched. The Cochrane Collaboration, which recently celebrated their
20th anniversary is widely known for generating systematic reviews
and is respected by both researchers and librarians alike (Friedrich, 2013).
The Cochrane Library provides access to the CENTRAL database, which includes
randomized and controlled clinical trials obtained not only from electronic database
searches but also from the journal and conference proceeding handsearching
efforts of the Cochrane Collaboration. Furthermore, the Cochrane Library
includes access to the HTA database and the NHSEED which contain indexed
reports. . For this reason, the Cochrane Library was classified as a source of
grey literature for the purpose of this study. It should be noted that 56% of
the systematic reviews reported in this survey were conducted under the
guidance of a systematic review entity. This may include the Cochrane
Collaboration, for which documentation in the editorial policies requires the
search of CENTRAL for all Cochrane Systematic Reviews (Chandler, Churchill,
Higgins, Lasserson, & Tovey, 2013). It is possible that this could have
influenced why the Cochrane Library was found to be the top grey literature
resource used. The two grey literature resources that were the next most
commonly used were the WHO ICTRP and mRCT. Both of these resources allow
federated searching across multiple trial registries including
ClinicalTrials.gov. Searching such resources would ideally reduce the need to
search several trial registries, a possible explanation for their appearance in
our top grey literature resources searched list and perhaps why they may have
appeared higher on the list compared to ClinicalTrials.gov.
Also evaluated was whether time spent searching or the
number of resources selected varied by searcher or review characteristics. As
shown in Figures 7 and 8 the amount of time spent searching for both grey and
non-grey literature was impacted by whether or not the systematic review was
grant funded. More time was spent searching both types of resources if the
systematic review was grant funded. The explanation for this finding is
unclear. It may be that there are differences as to how funded systematic
reviews are conducted compared to non-funded systematic reviews. A 2007 study
(Reed et al., 2007) on the association of funding and quality of published
medical education research found differences in the quality of funded studies.
Perhaps a funded systematic review has more resources available in terms of
manpower and technology because of funding.
Another finding was that the number of resources
searched was impacted by the searcher characteristics. The number of literature
resources varied by institution (academic, hospital or other). When performing
all two-way comparisons, we found that those who work in academic settings used
fewer grey literature resources than those who work in other settings (p =
0.03) and that those who work in hospitals use fewer numbers of grey literature
resources than those who work in other settings (p = 0.049). Academic- and
hospital-affiliated information professionals are often juggling a multitude of
responsibilities and therefore may not have as much time to devote to each
systematic review search and therefore it is possible that this leads to fewer
grey literature resources being used in the search. The number of non-grey
literature resources searched also varied by whether the individual received
systematic review training. If systematic review training was received, the
number of non-grey literature resources was decreased. It is unclear why such a
relationship is evident. Perhaps those with training can more readily select
the pertinent databases for the systematic review topic or feel more confident
with being selective. Our survey cannot conclude this however. Searchers with
less training may take precaution and therefore select a large number of
resources to search and spend a long time searching. Again, there is no certain
explanation for our results. Many variables may contribute, including whether
there is a fair amount of time allotted to locating studies.
Limitations
There are several limitations in this study that are
worth noting. We utilized convenience sampling, recruiting mainly through known
listservs. Of the respondents who started the survey, only 21% completed the
survey. The reasons for this non-completion are not known, although it is
possible that the second part of the survey was viewed as too onerous or that
the planned systematic review never progressed to the searching phase. There
was limited power in the analysis due to the small sample size, so results
should be viewed as suggestive rather than predictive. Also, because the small
sample size consisted of mainly library professionals, results may not be
generalizable to all those undertaking a systematic review search. Part two of
this survey study utilized prospective methodology, asking participants to
record information about their systematic review searching as they worked.
Perhaps if a retrospective methodology was used more potential participants
would have met our inclusion criteria, resulting in a larger sample size.
However, it was the feeling of the authors that a prospective survey would
allow the least biased capture of time spent searching. A further limitation
regarding methodology of the study was the subjective categorization of
resources into grey literature or non-grey literature by the authors.
The time for searching that we obtained through this
study may be underreported due to the following: five survey participants
reported that a portion of the searching was completed by another individual
but only reported the name of the resource, not time spent searching. Two
survey participants did not report the time spent searching each resource
individually; they only reported the total time as whole which was spent
searching all resources for the review.
Future Directions
This study draws attention to the need for further
research on search methodology in systematic reviews and grey literature.
Furthermore, there is a need for guidelines on conducting systematic review
searches including grey literature searching. Some earlier literature
demonstrates searchers’ recognition of issues with searching grey literature,
and with the approach to searching for comprehensive literature reviews. In a Canadian report (Dobbins, Robeson, Jentha &
DesMeules, 2008), a methodology for the grey literature search to support
evidence syntheses in public health was explored. Tyndall (2008) opens up
discussion of the grey literature search process by suggesting a hierarchy for
acceptable grey literature. Bidwell‘s COSI model (Bidwell & Jensen, 2003),
a model proposed for searches to support technology assessment reports is an
example of an approach that can be revisited to address this matter. The COSI
model relates to the whole search process and uses a framework or protocol to
categorize resources into three levels of priority based on expectation of
yield. The acronym COSI, and levels of priority, include CO for core search, S
for standard search, and I for ideal search (Bidwell & Jensen, 2003). If
there is sufficient time to undertake the search, the Ideal search of resources
is proposed. Information professionals conducting searches for systematic
reviews have little guidance to assist them with the approach and how to
conduct the searches. For information specialists embarking on systematic
review searches, questions may arise as to which resources to search, how many,
how far to go in breadth of resources. Perhaps, a similar concept to the COSI
model can be used to develop a framework for guidance in grey literature
searching in support of systematic reviews.
Hopewell, McDonald, et al. (2007) found that studies
included in systematic reviews, located through grey literature searching, were
more commonly conference abstracts or unpublished data (trial registers, file
drawer data, data from individual trialists). Additional types of grey
literature identified in the review include: book chapters, unpublished
reports, pharmaceutical company data, in press publications, and letters and
theses (Hopewell, McDonald, et al., 2007). The study findings could potentially
be used to support development of a framework that would prioritize resources
by the document types that they include, with the thought that certain types of
documents are more often found useful for inclusion into systematic reviews.
The aforementioned are just several examples for illustration purposes. A more
national/international collaborative effort by an authoritative body would be
best to propose systematically developed guidance.
It should be noted that a number of guides exist which
list grey literature resources to search and explain the importance of
searching grey literature (Canadian Agency for Drugs and Technologies in
Health, 2013; Centre for Reviews and Dissemination, 2009; Cochrane
Collaboration, 2011). Many of these guides can be helpful to identify possible
resources but some are merely lists, and while some explain the importance of
select types of items they do not highlight an approach or framework for
following through with the entire grey literature search process. Furthermore,
only a select few are designed specifically for information professionals.
Exploration of the systematic review search process
regarding grey literature including time taken to search, resources used, and
investigation of any relating characteristics which may impact these factors is
one step forward toward engaging the health sciences library community in
further discussion about grey literature. Many information professionals are
multi-tasking, such as is the case with academic health science librarians and
hospital librarians, and therefore time management is of great interest in
order to efficiently integrate systematic review searching into one’s routine
responsibility.
Conclusion
We sought to prospectively explore the time taken to
conduct grey literature searches, via electronic database searching and
handsearching, for systematic reviews and to evaluate whether any relationship
exists between searcher and systematic review characteristics. The mean time
taken to conduct grey literature searches was approximately 7 hours, with 50%
of the searchers reporting less than 1.5 hours spent. This mean time represents
27% of the total time taken to complete the systematic review literature search.
Time spent searching both the grey and non-grey literature was influenced by
whether or not the systematic review was grant funded. The time estimates given
in this study are for searching-related activities only and do not include
other potential librarian efforts involved in participating in the synthesis of
a review such as meetings with the review requestor and the systematic review
team and managing citations. However, the time estimate provided for searching
both the grey literature and non-grey literature resources can provide
direction for librarians when meeting with researchers, writing a grant for a
completing a systematic review, or simply managing their own time. The top
resources used by the participants in this study might provide a reference
point for librarians working on a systematic review.
In light of recently established systematic review
standards, we expect some changes in the landscape of systematic review
searching. Additionally, we hope that in the near future grey literature
searching standards for systematic reviews are developed by the librarian
community for information professionals.
Acknowledgements
This project is supported in part by the National
Institutes of Health through Grant Numbers UL1 RR024153 and UL1TR000005.
We would like to thank Dave Piper for review of the
manuscript.
References
Adams, C. E., Power, A., Frederick, K., & Lefebvre, C. (1994). An
investigation of the adequacy of MEDLINE searches for randomized controlled
trials (RCTs) of the effects of mental health care. Psychological Medicine, 24(3), 741-748. doi: http://dx.doi.org/10.1017/S0033291700027896
Al-Maskari, A., & Sanderson, M. (2011). The effect of user
characteristics on search effectiveness in information retrieval. Information Processing & Management, 47(5),
719-729. doi: http://dx.doi.org/10.1016/j.ipm.2011.03.002
Armstrong, R., Jackson, N., Doyle, J., Waters, E., & Howes, F.
(2005). It's in your hands: the value of handsearching in conducting systematic
reviews of public health interventions.
Journal of Public Health, 27(4), 388-391. doi: http://dx.doi.org/10.1093/pubmed/fdi056
Bidwell, S., & Jensen, M. F. (2003). Using a search protocol to
identify sources of information: the COSI model. Etext on Health Technology
Assessment (HTA) Information Resources. In National
Information Center on Health Services Research and Health Care Technology
(NICHSR). Retrieved from http://www.nlm.nih.gov/archive/20060905/nichsr/ehta/chapter3.html
Canadian Agency for Drugs and Technologies in Health. (2013). Grey
matters: a practical deep-web search tool for evidence-based medicine, Retrieved from http://www.cadth.ca/media/pdf/Grey-Matters_A-Practical-Search-Tool-for-Evidence-Based-Medicine.doc
Centre for Reviews and Dissemination. (2009). Systematic reviews: CRD's
guidance for undertaking reviews in healthcare. Retrieved from http://www.york.ac.uk/inst/crd/pdf/Systematic_Reviews.pdf
Chandler, J., Churchill, R., Higgins, J., Lasserson, T., & Tovey, D.
(2013). Methodological standards for the conduct of new Cochrane Intervention
Reviews. In Cochrane Editorial Unit
(Ed.) Methodological Expectations of
Cochrane Intervention Reviews (MECIR) |
Retrieved from http://www.editorial-unit.cochrane.org/sites/editorial-unit.cochrane.org/files/uploads/MECIR_conduct_standards%202.3%2002122013.pdf
Cochrane Collaboration. (2011). In J. P. T. Higgins
& S. Green (Eds.) Cochrane Handbook
for Systematic Reviews of Interventions,
Retrieved from http://www.cochrane-handbook.org
Cook, D. J., Guyatt, G. H., Ryan, G., Clifton, J.,
Buckingham, L., Willan, A., McIlroy, W., Oxman, A. D. (1993). Should
unpublished data be included in meta-analyses? Current convictions and
controversies. Journal of the American
Medical Association, 269(21), 2749-2753. doi: http://dx.doi.org/10.1001/jama.1993.03500210049030
Cook, D. J., Mulrow, C. D., & Haynes, R. B.
(1997). Systematic reviews: synthesis of best evidence for clinical decisions. Annals of Internal Medicine, 126(5),
376-380. doi: http://dx.doi.org/10.7326/0003-4819-126-5-199703010-00006
Cooper, I. D., & Crum, J. A. (2013). New
activities and changing roles of health sciences librarians: a systematic
review, 1990-2012. Journal of the Medical
Library Association, 101(4), 268-277. doi: http://dx.doi.org/10.3163/1536-5050.101.4.008
Croft, A. M., Vassallo, D. J., & Rowe, M. (1999).
Handsearching the Journal of the Royal Army Medical Corps for trials. Journal of the Royal Army Medical Corps, 145(2),
86-88. doi: dx.doi.org/10.1136/jramc-145-02-09
Crum, J. A., & Cooper, I. D. (2013). Emerging
roles for biomedical librarians: a survey of current practice, challenges, and
changes. Journal of the Medical Library
Association, 101(4), 278-286. doi: http://dx.doi.org/10.3163/1536-5050.101.4.009
Crumley, E. T., Wiebe, N., Cramer, K., Klassen, T. P.,
& Hartling, L. (2005). Which resources should be used to identify RCT/CCTs
for systematic reviews: a systematic review. BMC Medical Research Methodology, 5, 24. doi: http://dx.doi.org/10.1186/1471-2288-5-24
Dobbins, M.,Robeson, P,Jentha, N., & DesMeules , M (2008).
Grey Literature: A methodology for searching the grey literature for
effectiveness evidence syntheses related to public health: A report from
Canada.Health Inform, 17(1), 9-12.
Farace, D. J. (1998). Forward. In D.J. Farace
(ed.). GL'97 Conference Proceedings: Perspectives
on the Design and Transfer of Scientific and Technical Information, Amsterdam: TransAtlantic.
Fenichel, C. H. (1981). Online searching: Measures that
discriminate among users with different types of experiences. Journal of the American Society for
Information Science, 32(1), 23-32. doi: http://dx.doi.org/10.1002/asi.4630320104
Friedrich, M. J. (2013). The Cochrane Collaboration
turns 20: Assessing the evidence to inform clinical care. Journal of the American Medical Association, 309(18), 1881-1882.
doi: http://dx.doi.org/10.1001/jama.2013.1827
Greenhalgh, T., & Peacock, R. (2005).
Effectiveness and efficiency of search methods in systematic reviews of complex
evidence: audit of primary sources. British
Medical Journal, 331(7524), 1064-1065. doi: http://dx.doi.org/10.1136/bmj.38636.593461.68
GreyNet International. (2013). Document types in grey
literature. Retrieved 07/09/2013, from http://www.greynet.org/greysourceindex/documenttypes.html
Guise, J.-M., & Viswanathan, M. (2011). Overview
of best practices in conducting comparative-effectiveness reviews. Clinical Pharmacology &Therapeutics, 90(6),
876-882. doi: http://dx.doi.org/10.1038/clpt.2011.239
Hopewell, S., Clarke, M.J., Lefebvre, C., &
Scherer, R.W. (2007). Handsearching versus electronic searching to identify
reports of randomized trials. In Cochrane
Database of Systematic Reviews, MR000001. doi: http://dx.doi.org/10.1002/14651858.MR000001.pub2
Hopewell, S., Clarke, M.J., Stewart, L., &
Tierney, J. (2007). Time to publication for results of clinical trials. In Cochrane Database of Systematic Reviews,
MR000011. doi: http://dx.doi.org/10.1002/14651858.MR000011.pub2
Hopewell, S., McDonald, S., Clarke, M.J., & Egger,
M. (2007). Grey literature in meta-analyses of randomized trials of health care
interventions. In Cochrane Database of
Systematic Reviews, MR000010. doi: http://dx.doi.org/10.1002/14651858.MR000010.pub3
Hsieh-Yee, I. (1993). Effects of search experience and
subject knowledge on the search tactics of novice and experienced searchers. Journal of the American Society for
Information Science, 44(3), 161-174. doi: http://dx.doi.org/10.1002/(SICI)1097-4571(199304)44:3<161::AID-ASI5>3.0.CO;2-8
Institute of Medicine. (2011). Finding what works in health care : Standards for systematic reviews.
Washington DC: National Academies PressRetrieved from www.nap.edu/openbook.php?record_id=13059
Kuhlthau, C. C. (1999). The role of experience in the
information search process of an early career information worker: Perceptions
of uncertainty, complexity, construction, and sources. Journal of the American Society for Information Science, 50(5),
399-412. doi: http://dx.doi.org/10.1002/(Sici)1097-4571(1999)50:5<399::Aid-Asi3>3.0.Co;2-L
Reed, D. A., Cook, D. A., Beckman, T. J., Levine, R.
B., Kern, D. E., & Wright, S. M. (2007). Association between funding and
quality of published medical education research. Journal of the American Medical Association, 298(9), 1002-1009.
doi: http://dx.doi.org/10.1001/jama.298.9.1002
Savoie, I., Helmer, D., Green, C. J., & Kazanjian,
A. (2003). Beyond Medline: Reducing bias through extended systematic review
search. International Journal of
Technology Assessment in Health Care, 19(1), 168-178. doi: http://dx.doi.org/10.1017/S0266462303000163
Tabatabai, D., & Shore, B. M. (2005). How experts
and novices search the Web. Library &
Information Science Research, 27(2), 222-248. doi: http://dx.doi.org/10.1016/j.lisr.2005.01.005
Wright, J. M., Cottrell, D. J., & Mir, G. (2014). Searching for
religion and mental health studies required health, social science, and grey
literature databases. Journal of Clinical
Epidemiology, 67(7), 800-810. doi: http://dx.doi.org/10.1016/j.jclinepi.2014.02.017
Appendix
Survey -Part One
CONTACT INFORMATION - solely used for communication
with you regarding the study
First Name
Last Name
Institution/Organization
E-mail
Phone
INFORMATION ABOUT THE SEARCHER
1.
Country of Employment
2.
List Academic Credentials (e.g., PhD, MLIS, MPH)
3.
Type of Institution
·
Academic
·
Hospital
·
Industry/Corporation
·
Government Agency
·
Organization/Agency (non-Government)
·
Other (If Other, specify)
4.
Position Title
5.
Select an option which represents your primary
professional role
6.
Years of experience in your profession (that you
selected in Question 5)?
7. Have you had any formal training on how to
conduct literature searching
for Systematic Reviews? YES/NO
8.
Approximately how many years of experience do you have
contributing to Systematic Reviews?
9.
Approximately how many Systematic Reviews have you
contributed to in the past?
Survey -Part Two
INFORMATION ABOUT THE SYSTEMATIC REVIEW (SR)
1.
Please list the title or the topic of the SR.
Note- This
information will only be used to classify the SRs into subject categories.
2.
Please select the population age group included in the
SR.
3.
Which category best describes the type of Systematic
Review?
4.
Is this SR grant funded? YES/NO
5.
Is this SR being produced under the guidance of a
systematic review funding or producing agency/organization? YES/NO/NOT SURE
6.
Should the SR be published, are there plans for you to
be a coauthor? YES/NO/NOT SURE
7.
If you are NOT a librarian/information professional,
is there one involved in the SR?
If you are a
librarian/information professional select N/A
YES/NO/ N/A
If YES, in
what primary role?
INFORMATION ABOUT SEARCHING FOR THE SYSTEMATIC REVIEW
(SR)
8.
Is cited reference searching (checking who cited a
paper) or checking the reference lists of papers of interest, planned as part
of the search effort for this SR?
9.
Will a methodology filter be used in the SR search? (A
methodology filter is a pre-designed search strategy with terms related to
research methodology. Examples include the Cochrane Highly Sensitive Search
Strategy and Clinical Queries.) YES/NO
If YES, for
which resources?
10.
Were any resources purchased
as a one-time paid subscription for purposes of a search for the SR? YES/NO
11.
May we contact you to ask
for the final number of studies included and where the citations were
originally found? YES/NO
12.
Use the following table to
document all resources searched for the SR.
If any journals or conference proceedings were
handsearched, label with Handsearch followed
by a dash and the Title of the Resource in the RESOURCE NAME column. Indicate
in RESOURCE Platform column whether the print or electronic version was
handsearched
If a resource was purchased as a one-time paid subscription
for purposes of a search for this SR, please mark with an asterisk (*) preceding the resource name in the
RESOURCE NAME column.
Use n/a if
an item does not apply, to indicate that you are unable to identify the
information, or if the information is not available
Three examples are provided, although the time is not
documented
ANOTHER1 |
DATE SEARCHED |
RESOURCE NAME |
RESOURCE Platform/ Interface/ Vendor |
RESOURCE URL |
TIME in minutes2 |
|
1/22/2010 |
* BIOSIS Previews |
Dialog |
n/a |
Insert Time Here |
|
2/10/2010 |
clinicaltrials.gov |
n/a |
clinicaltrials.gov |
Insert Time Here |
x |
1/4/2010- 3/1/2010 |
Handsearch- Proceedings of the Nutritional Society (2007-2009) |
print |
n/a |
Insert Time Here |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 Mark with an
X resources you did not search yourself but that were searched for the SR
2 Time should
account for choosing terminology, developing the strategy, and running the
search