Article type
Year
Abstract
Introduction: The development of the specialised EPOC register raises a number of methodological issues. In particular, systematic identification of relevant studies, by retrospective and prospective electronic searching or by journal handsearching, is particularly difficult due to poor indexing in major databases of relevant interventions and the scatter of journals in which they are published.
Objective: To develop and maintain a comprehensive register of studies within the scope of EPOC including professional, organisational, financial and regulatory interventions to improve quality of care.
Methods: Electronic searching of bibliographic databases and hand searching key journals.
Results: Handsearching: The yield of relevant studies from handsearching tends to be low. 'Medical Care', a journal identified as likely to give the highest yield, produced only 168 relevant studies between the years 1969-95; an average of 6 per year. Screen searching Ovid's Core Biomedieal Collection for all records indexed as original or miscellaneous and published in the BMJ, Lancet, JAMA or Annals of Internal Medicine in 1995-6 (over 7000 articles) resulted in 57 studies being identified as relevant. Electronic searching: A "gold standard" of studies known to be within the scope of EPOC has been created and is being used to assist in the development of a Medline search strategy. Many MeSH terms have been identified which could be of potential relevance to EPOC studies but preliminary analysis suggest most have very low precision when measured against the gold standard. Text word searching has produced more promising results. The trade off between sensitivity and precision is being explored to determine the optimal strategy. Electronic storage: Register records are maintained in an Idealist database and comprise fields downloaded from Medline, including MeSH terms and abstracts. In addition fields giving details of study design, clinical problem, type of intervention and targeted behaviour are also included. Initially, the majority of entries had been located in a non-systematic fashion, derived from citations contained in two large reviews. Progress to date: The register currently has 680 studies, although we believe that eventually the register may contain 2-3,000 studies. The register can be downloaded in a read only format from the EPOC web site.
Discussion: Systematic identification of relevant studies is problematic but electronic searching appears the more promising. The register will provide a valuable resource for EPOC reviewers, researchers and policy makers
Objective: To develop and maintain a comprehensive register of studies within the scope of EPOC including professional, organisational, financial and regulatory interventions to improve quality of care.
Methods: Electronic searching of bibliographic databases and hand searching key journals.
Results: Handsearching: The yield of relevant studies from handsearching tends to be low. 'Medical Care', a journal identified as likely to give the highest yield, produced only 168 relevant studies between the years 1969-95; an average of 6 per year. Screen searching Ovid's Core Biomedieal Collection for all records indexed as original or miscellaneous and published in the BMJ, Lancet, JAMA or Annals of Internal Medicine in 1995-6 (over 7000 articles) resulted in 57 studies being identified as relevant. Electronic searching: A "gold standard" of studies known to be within the scope of EPOC has been created and is being used to assist in the development of a Medline search strategy. Many MeSH terms have been identified which could be of potential relevance to EPOC studies but preliminary analysis suggest most have very low precision when measured against the gold standard. Text word searching has produced more promising results. The trade off between sensitivity and precision is being explored to determine the optimal strategy. Electronic storage: Register records are maintained in an Idealist database and comprise fields downloaded from Medline, including MeSH terms and abstracts. In addition fields giving details of study design, clinical problem, type of intervention and targeted behaviour are also included. Initially, the majority of entries had been located in a non-systematic fashion, derived from citations contained in two large reviews. Progress to date: The register currently has 680 studies, although we believe that eventually the register may contain 2-3,000 studies. The register can be downloaded in a read only format from the EPOC web site.
Discussion: Systematic identification of relevant studies is problematic but electronic searching appears the more promising. The register will provide a valuable resource for EPOC reviewers, researchers and policy makers