search strategy critique of cochrane systematic review i

3
LIBR534 – March 2010, Assignment #3 ( prepared by Sandra Barron) Activity #1 – Critique of search strategy Acupuncture for tension-type headache” Cochrane Systematic Review (1) Pros of search strategy: good choice of mainstream databases for thorough overview of the literature for acupuncture & headache Medline is the premier international biomedical database, and essential in locating primary studies/ research/ narrative reviews; Embase is international in scope and comparable to Medline; required for coverage; complementary view of topic keywords are well chosen and cover main terms adequately; controlled terms match thesauri of databases (e.g. EMTREE headings used for Embase; MeSH used for Medline); truncation is used liberally throughout the searches to improve recall liked the ‘streamlined simplicity’ of the search strategy; root of search terms appears, while ends are replaced by a symbol such as * or $, so all forms of word are retrieved search strategy seems reasonable in terms of the topic except for some suggestions noted in next section I liked the search strategy for its balance of overview and attention to details Cons of search strategy: To improve recall, other databases should be searched such as CINAHL (55% unique content, not in Medline or Embase), PsychINFO, PsycArticles and PubMedCentral. In the latter two, good results are obtained using keywords [acupuncture AND tension headache]. PsychINFO was searched using controlled terms [Acupuncture AND Muscle Contraction Headache] recall can be improved further with additional controlled (MeSH) terms such as [Acupuncture Points] and [Treatment Outcome]; alternatively, precision could be improved by using more specific controlled terms: replace [Headache Disorders] and [Headache] with narrower terms such as [Tension-Type Headache] = MeSH term; [Tension Headache] = EMTREE term; quality of search results is questionable because ‘migraine’ was not eliminated due to broad controlled terms [Headache Disorders; Headache] rather than more specific [Tension Headache] as noted; wild cards are not used for unknown or extra characters in search, but may have been helpful for terms such as “randomi?ed” Surprises in the search strategy: strategy is less thorough than expected; exploding/ focusing of terms not used (exception: [exp Headache Disorders]) Why is this surprise? Because consistency between databases helps to determine specific recall and precision needs subheadings were not used at all; no limiters either such as ‘Randomized Controlled Trial’ and ‘English Language’ and ‘Humans’; keywords are kept to a minimum controlled terms are broad (more specific terms will help to eliminate ‘migraine headaches’ from search [Tension-Type Headache] = MeSH term; [Tension Headache] = EMTREE term) filters (i.e. hedges) are used to search for randomized controlled trials (RCTs) rather than limiters such as OvidSP Medline’s Publication Type “Randomized Controlled Trials”; instructions in Cochrane Handbook are not followed explicitly according to Yoshii et al. (2), search strategy descriptions should include: databases searched, name of host, date search was run, years covered by search, complete search strategy, one or two sentence summary of strategy, language restrictions; search strategy did not include (i) date search was run and (ii) language restrictions possible reasons for omissions: (i) word limits for reviews may tempt authors to minimize descriptions of search strategies; (ii) earlier version of Cochrane Handbook likely used for this study was confusing (2) detailed search strategy reporting is important because it allows for reproducibility, evaluation of the quality of a search and the credibility of the review. It aids in replication when updating a review; and can be used as a template for search strategies for reviews on similar topics (2)

Upload: dean-giustini

Post on 18-Jun-2015

3.529 views

Category:

Health & Medicine


5 download

DESCRIPTION

Sandra Barron, MLIS Student, does a great critique of the search strategy in “Acupuncture for tension-type headache” Cochrane Systematic Review

TRANSCRIPT

Page 1: Search strategy critique of Cochrane Systematic Review I

LIBR534 – March 2010, Assignment #3 ( prepared by Sandra Barron)

Activity #1 – Critique of search strategy

“ Acupuncture for tension-type headache” Cochrane Systematic Review (1)

Pros of search strategy:

• good choice of mainstream databases for thorough overview of the literature for acupuncture & headache• Medline is the premier international biomedical database, and essential in locating primary studies/ research/ narrative

reviews; Embase is international in scope and comparable to Medline; required for coverage; complementary view of topic• keywords are well chosen and cover main terms adequately; controlled terms match thesauri of databases (e.g. EMTREE

headings used for Embase; MeSH used for Medline); truncation is used liberally throughout the searches to improve recall

• liked the ‘streamlined simplicity’ of the search strategy; root of search terms appears, while ends are replaced by a symbol such as * or $, so all forms of word are retrieved

• search strategy seems reasonable in terms of the topic except for some suggestions noted in next section

• I liked the search strategy for its balance of overview and attention to details

Cons of search strategy:

• To improve recall, other databases should be searched such as CINAHL (55% unique content, not in Medline or Embase),• PsychINFO, PsycArticles and PubMedCentral. In the latter two, good results are obtained using keywords [acupuncture AND

tension headache]. PsychINFO was searched using controlled terms [Acupuncture AND Muscle Contraction Headache]• recall can be improved further with additional controlled (MeSH) terms such as [Acupuncture Points] and [Treatment

Outcome]; alternatively, precision could be improved by using more specific controlled terms: replace [Headache Disorders] and [Headache] with narrower terms such as [Tension-Type Headache] = MeSH term; [Tension Headache] = EMTREE term; quality of search results is questionable because ‘migraine’ was not eliminated due to broad controlled terms [Headache Disorders; Headache] rather than more specific [Tension Headache] as noted; wild cards are not used for unknown or extra characters in search, but may have been helpful for terms such as “randomi?ed”

Surprises in the search strategy:

• strategy is less thorough than expected; exploding/ focusing of terms not used (exception: [exp Headache Disorders])• Why is this surprise? Because consistency between databases helps to determine specific recall and precision needs• subheadings were not used at all; no limiters either such as ‘Randomized Controlled Trial’ and ‘English Language’ and

‘Humans’; keywords are kept to a minimum• controlled terms are broad (more specific terms will help to eliminate ‘migraine headaches’ from search [Tension-Type

Headache] = MeSH term; [Tension Headache] = EMTREE term)• filters (i.e. hedges) are used to search for randomized controlled trials (RCTs) rather than limiters such as OvidSP Medline’s

Publication Type “Randomized Controlled Trials”; instructions in Cochrane Handbook are not followed explicitly

• according to Yoshii et al. (2), search strategy descriptions should include: databases searched, name of host, date search was run, years covered by search, complete search strategy, one or two sentence summary of strategy, language restrictions; search strategy did not include (i) date search was run and (ii) language restrictions

• possible reasons for omissions: (i) word limits for reviews may tempt authors to minimize descriptions of search strategies; (ii) earlier version of Cochrane Handbook likely used for this study was confusing (2)

• detailed search strategy reporting is important because it allows for reproducibility, evaluation of the quality of a search and the credibility of the review. It aids in replication when updating a review; and can be used as a template for search strategies for reviews on similar topics (2)

Page 2: Search strategy critique of Cochrane Systematic Review I

Extended search techniques used in SR:

• one reviewer regularly checked PubMed for new entries up to April 15, 2008 using search terms [acupuncture AND headache], checked conference abstracts, and inquired about new studies from researchers (digital browsing) (1)

• grey literature (ongoing or unpublished studies) was accessed by searching three separate online clinical trial registries (1)

Activity #2 – Current and future search skills

Three important search skills that I have learned in LIBR 534 centre around the Five Steps of Evidence-based

Medicine (EBM); in particular, the first two steps of EBM are most important for the health librarian because they

require (i) framing a patient scenario into an answerable clinical question and (ii) retrieving the best evidence available

using a good search strategy (D. Giustini, personal communication, January 28, 2010). The Medical Library Association

(MLA) Policy Statement (3) says that an expert searcher has the “ability to accurately identify an information need

through effective personal interaction and to clarify and refine the need and retrieval requirements.” The first search

skill I acquired is to use PICO to structure a clinical question during the reference interview. Consequently, the patient

scenario is broken into facets or categories: patient/population/problem, intervention, comparison, and outcomes. In

the first step, a clinician decides what patient population will be studied, what tests, treatments, and alternatives will be

considered, and what outcomes are to be measured. The health librarian uses PICO to combine these elements in an

orderly way and create a focused, answerable clinical question. (4)

The second learned search skill I learned is applying a search strategy that coincides with evidence-based

practice. The MLA (3) refers to this as “knowledge of database subject content, indexing or metadata conventions, and

online record format to determine relevance to the information need and the method of retrieval access.” This search

strategy may be conducted by the health librarian or it may be taught to a health professional or researcher (D. Giustini,

personal communication, February 11, 2010). An EBM search strategy involves understanding hierarchies of evidence to

search filtered sources according to the evidence-based pyramid. More rigorous studies, such as systematic reviews and

randomized controlled (double/triple blind) trials (RCTs) are at the top of the pyramid, while ideas, editorials and

opinions are lower. In addition, it is important to understand where to find these high quality (pre)filtered sources. For

instance, the Cochrane Library is a better source for systematic reviews than Medline although you can search SRs in

clinical queries (D. Giustini, personal communication, February 4, 2010). Effective searching of medical literature

requires an understanding of overall methods, strengths and weaknesses of search strategies and tools. In general

terms, health librarians use the most relevant search terms, keywords and concepts from the PICO model to find the

best evidence available (4). LIBR 534 class lectures and assignments have provided the means to search various

databases (Medline, CINAHL, Embase etc.), pre-appraised sources (Cochrane and related tools), and grey literature.

The MLA (3) defines one aspect of expert searching as the “ability to identify and search resources beyond the

electronically available published literature, including the older published literature, gray literature, unpublished

information, and Web documents.” The third search skill I learned is searching for grey literature outside of the

Page 3: Search strategy critique of Cochrane Systematic Review I

conventional print and electronic publishing and indexing methods. Grey literature is defined as original and recent

information produced in electronic and print formats not controlled by commercial publishing and not found easily

through conventional channels (D. Giustini, personal communication, March 4, 2010). Grey literature may be found in

many places including certain databases (e.g. TRIP Database), conference proceedings, research reports, theses and

dissertations, and open search tools such as Google Scholar, Scirus, PubMedCentral, NLM Gateway, etc. (D. Giustini,

personal communication, March 4, 2010); (5). Trends in digitization are creating new sources of grey literature such as

emails, blogposts, and even tweets. One of the best sources of grey literature is the library catalogue. The impact of

grey literature in medicine is significant – it allows for a more comprehensive view of the topic and fills in evidence gaps.

Plus, it reduces the risk of introducing bias into a review (D. Giustini, personal communication, March 4, 2010).

There are several search skills noted in the MLA policy statement (3) that I want to learn more about in the

future. These revolve around the Five Steps of EBM noted earlier. For instance, the ability:

(i) To evaluate retrieved evidence

(ii) To expertly process retrieved results for presentation by editing and applying data mining techniques

(iii) To document the search process for end-user information or legal purposes.

These skills would lessen the subsequent work for end-users. According to the MLA (3), health professionals and

researchers generally do not have the combined skills listed in the policy statement. Acquiring a complete set of skills is

necessary for health librarians to fulfill their roles as online instructors and consultants. The emphasis on evidence-

based practice makes these search skills even more relevant and indispensible in today’s health care settings.

Reference List

(1) Linde K, Allais G, Brinkhaus B, Manheimer E, Vickers A, White AR. Acupuncture for tension-type headache (review). The Cochrane Library [Online] 2009 [cited 2010 Mar 13]; (4):[54 screens]. Available from: http://www2.cochrane.org/reviews/en/ab007587.html

(2) Yoshii A, Plaut DA, McGraw KA, Anderson MJ, Wellik KE. Analysis of the reporting of search strategies in Cochrane systematic reviews. J Med Libr Assoc [serial online]. 2009 Jan [cited 2010 Mar 13]; 97(1): 21-9. Available from: http://tiny.cc/mOzTZ

(3) Medical Library Association. Role of expert searching in health sciences libraries. J Med Libr Assoc [serial online]. 2003 Jan [cited 2010 Mar 14]; 93(1): 42-4. Available from: http://tiny.cc/Gzvx2

(4) HLWiki Canada. Evidence-based health care. [Online]. 2010 [cited 2010 Mar 15]; Available from: http://tiny.cc/UnJwT

(5) HLWiki Canada. Open search. [Online]. 2010 [cited 2010 Mar 15]; Available from: http://tiny.cc/Pv4iX