Article type
Year
Abstract
Background: Comprehensive searching for relevant controlled trials forms the essential foundation for systematic reviews. In executing searches, beyond an understanding of the rules that must be applied to each search platform, an awareness and understanding of how and why record retrieval can still vary between databases searched is helpful.
Objective: The objective of this exercise was to demonstrate some anomalies which may affect record retrieval from electronic searches and to reinforce the importance of searching a range of databases in the interests of meticulous and systematic searching to identify reports of trials.
Method: Search strategies were developed for two reviews: Single dose dextropropoxyphene, alone and with paracetamol (acetaminophen) for postoperative pain (Cochrane Pain, Palliative Care & Supportive Care Review Group) and Surgical techniques for removal of mandibular third molar teeth (Cochrane Oral Health Group) and run on CENTRAL; MEDLINE; EMBASE; and the respective review groups trials registers. For each review the search strategies were revised appropriately for each database using relevant controlled vocabulary but applying the same freetext terms throughout. The records were downloaded and checked for duplications and where the search strategy had not retrieved a record from a particular database the reasons for exclusion were identified.
Results: Failure of some MEDLINE records to be retrieved in the search was shown to be due to fewer indexing terms in the record when compared to the EMBASE record. Other reasons were due to EMBASE records being entered into the database and indexed earlier than MEDLINE records. Retrieval of records from CENTRAL was also affected by whether the record in CENTRAL was sourced from MEDLINE or EMBASE. There is a delay of three months in publication in CENTRAL of records from MEDLINE and EMBASE. This gap increases to six months prior to publication of the next quarterly issue of The Cochrane Library.
Lack of a proximity searching facility in the review groups trials register, (ProCite® bibliographic management program) was the cause of failure to retrieve some records from the register that came up in other databases. Another potential problem was identified where additional spaces found in the abstracts of some records negated phrase searching where this occurred.
Conclusions: The Cochrane Collaboration continues its commitment to helping reviewers by the development of the Cochrane Central Register of Controlled Trials in The Cochrane Library and through working with the National Library of Medicine to ensure records are correctly tagged by publication type. Searching a range of databases and search platforms remains a prerequisite to address the diversity of variables within search facilities and is fundamental to achieving the meticulous and comprehensive searching that is the foundation for systematic reviewing. The reviewers subject knowledge and the expertise of the information specialist provide the complementary and essential elements for effective electronic searching.
Acknowledgements: The co-operation of the reviewers/authors of the reviews identified in this exercise is gratefully acknowledged.
Objective: The objective of this exercise was to demonstrate some anomalies which may affect record retrieval from electronic searches and to reinforce the importance of searching a range of databases in the interests of meticulous and systematic searching to identify reports of trials.
Method: Search strategies were developed for two reviews: Single dose dextropropoxyphene, alone and with paracetamol (acetaminophen) for postoperative pain (Cochrane Pain, Palliative Care & Supportive Care Review Group) and Surgical techniques for removal of mandibular third molar teeth (Cochrane Oral Health Group) and run on CENTRAL; MEDLINE; EMBASE; and the respective review groups trials registers. For each review the search strategies were revised appropriately for each database using relevant controlled vocabulary but applying the same freetext terms throughout. The records were downloaded and checked for duplications and where the search strategy had not retrieved a record from a particular database the reasons for exclusion were identified.
Results: Failure of some MEDLINE records to be retrieved in the search was shown to be due to fewer indexing terms in the record when compared to the EMBASE record. Other reasons were due to EMBASE records being entered into the database and indexed earlier than MEDLINE records. Retrieval of records from CENTRAL was also affected by whether the record in CENTRAL was sourced from MEDLINE or EMBASE. There is a delay of three months in publication in CENTRAL of records from MEDLINE and EMBASE. This gap increases to six months prior to publication of the next quarterly issue of The Cochrane Library.
Lack of a proximity searching facility in the review groups trials register, (ProCite® bibliographic management program) was the cause of failure to retrieve some records from the register that came up in other databases. Another potential problem was identified where additional spaces found in the abstracts of some records negated phrase searching where this occurred.
Conclusions: The Cochrane Collaboration continues its commitment to helping reviewers by the development of the Cochrane Central Register of Controlled Trials in The Cochrane Library and through working with the National Library of Medicine to ensure records are correctly tagged by publication type. Searching a range of databases and search platforms remains a prerequisite to address the diversity of variables within search facilities and is fundamental to achieving the meticulous and comprehensive searching that is the foundation for systematic reviewing. The reviewers subject knowledge and the expertise of the information specialist provide the complementary and essential elements for effective electronic searching.
Acknowledgements: The co-operation of the reviewers/authors of the reviews identified in this exercise is gratefully acknowledged.