Assessing the role of optimal information size in systematic reviews
<p><b>Background and objectives</b></p> <p>The concept of optimum information size (OIS) was first proposed by Pogue and Yusuf in a 1997 article published in Controlled clinical trials, as “the minimum amount of information required in the collective literature for reli...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Published: |
2018
|
| Subjects: |
| _version_ | 1797110552938414080 |
|---|---|
| author | Garcia-Alamino, J |
| author2 | Perera, R |
| author_facet | Perera, R Garcia-Alamino, J |
| author_sort | Garcia-Alamino, J |
| collection | OXFORD |
| description | <p><b>Background and objectives</b></p> <p>The concept of optimum information size (OIS) was first proposed by Pogue and Yusuf in a 1997 article published in Controlled clinical trials, as “the minimum amount of information required in the collective literature for reliable conclusions about an intervention to be reached”. OIS estimates, which are mainly based on standard sample size calculations, are influenced by several variables, including the control event rate (CER), effect size (ES) and heterogeneity.</p> <p>The overall aims of this thesis are to provide an accurate overview of OIS as depicted in the literature, establish if and how OIS is being used, and assess awareness and perceptions regarding OIS estimation from benchmark bodies and authors of systematic reviews with meta-analyses.</p> <p><b>Methods</b></p> <p>The literature was systematically reviewed in order to investigate what information has been published to date on OIS and sequential analyses applied to systematic reviews with meta-analyses.</p> <p>Published meta-analyses that included OIS estimates were reviewed in order to establish historical use and evolution of OIS in meta-analyses. On a more technical level, key statistical parameters — specifically, CER, ES and heterogeneity — were explored in terms of their impact on OIS estimates. </p> <p>Finally, surveys were conducted, firstly of Cochrane Review Groups (CRGs), in order to determine opinions and policies regarding the OIS methodology, and secondly, of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement authors and systematic review authors, in order to assess level of awareness and perceptions of OIS.</p> <p><b>Results</b></p> <p>Although OIS estimates are increasingly included in meta-analyses, most especially since 2010, only a small fraction of all meta-analyses published annually estimate OIS. The use of certain statistical parameters, particularly CER, ES and heterogeneity, is arbitrary and there are no guidelines or consensus in regard to which values to use. While the number of CRGs that had heard of and were considering using OIS estimates was substantially higher in 2016 than in 2010, ultimately only 50% of CRGs stated in 2016 that they were considering using OIS. It was found that few recently published meta-analyses achieved OIS, which would suggest that statistical power may have been inadequate to allow firm conclusions to be drawn. In terms of the range of values used for ES and heterogeneity, the results point to wide variability, although this variability is partially explained by outcome type. No clear and unequivocal relationship could ultimately be established for CER in relation to follow-up duration. While the concept of OIS generates substantial interest among systematic review authors, PRISMA Statement authors tend to be more sceptical, possibly due to the lack of methodological development of OIS estimation. </p> <p><b>Conclusion</b></p> <p>OIS estimates are increasingly being used by authors of systematic reviews, yet there are no clear guidelines or consensus regarding the principal statistical assumptions that affect OIS estimates. Most PRISMA Statement authors and systematic review authors expressed an interest in knowing more about OIS, despite a lack of familiarity with the methodology.</p> |
| first_indexed | 2024-03-07T07:56:24Z |
| format | Thesis |
| id | oxford-uuid:a5ec4338-5bc8-421b-8403-76e5aa27e770 |
| institution | University of Oxford |
| last_indexed | 2024-03-07T07:56:24Z |
| publishDate | 2018 |
| record_format | dspace |
| spelling | oxford-uuid:a5ec4338-5bc8-421b-8403-76e5aa27e7702023-08-16T16:11:28ZAssessing the role of optimal information size in systematic reviewsThesishttp://purl.org/coar/resource_type/c_db06uuid:a5ec4338-5bc8-421b-8403-76e5aa27e770Meta-analysisEvidence based medicineSystematic reviewsORA Deposit2018Garcia-Alamino, JPerera, RHeneghan, C<p><b>Background and objectives</b></p> <p>The concept of optimum information size (OIS) was first proposed by Pogue and Yusuf in a 1997 article published in Controlled clinical trials, as “the minimum amount of information required in the collective literature for reliable conclusions about an intervention to be reached”. OIS estimates, which are mainly based on standard sample size calculations, are influenced by several variables, including the control event rate (CER), effect size (ES) and heterogeneity.</p> <p>The overall aims of this thesis are to provide an accurate overview of OIS as depicted in the literature, establish if and how OIS is being used, and assess awareness and perceptions regarding OIS estimation from benchmark bodies and authors of systematic reviews with meta-analyses.</p> <p><b>Methods</b></p> <p>The literature was systematically reviewed in order to investigate what information has been published to date on OIS and sequential analyses applied to systematic reviews with meta-analyses.</p> <p>Published meta-analyses that included OIS estimates were reviewed in order to establish historical use and evolution of OIS in meta-analyses. On a more technical level, key statistical parameters — specifically, CER, ES and heterogeneity — were explored in terms of their impact on OIS estimates. </p> <p>Finally, surveys were conducted, firstly of Cochrane Review Groups (CRGs), in order to determine opinions and policies regarding the OIS methodology, and secondly, of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement authors and systematic review authors, in order to assess level of awareness and perceptions of OIS.</p> <p><b>Results</b></p> <p>Although OIS estimates are increasingly included in meta-analyses, most especially since 2010, only a small fraction of all meta-analyses published annually estimate OIS. The use of certain statistical parameters, particularly CER, ES and heterogeneity, is arbitrary and there are no guidelines or consensus in regard to which values to use. While the number of CRGs that had heard of and were considering using OIS estimates was substantially higher in 2016 than in 2010, ultimately only 50% of CRGs stated in 2016 that they were considering using OIS. It was found that few recently published meta-analyses achieved OIS, which would suggest that statistical power may have been inadequate to allow firm conclusions to be drawn. In terms of the range of values used for ES and heterogeneity, the results point to wide variability, although this variability is partially explained by outcome type. No clear and unequivocal relationship could ultimately be established for CER in relation to follow-up duration. While the concept of OIS generates substantial interest among systematic review authors, PRISMA Statement authors tend to be more sceptical, possibly due to the lack of methodological development of OIS estimation. </p> <p><b>Conclusion</b></p> <p>OIS estimates are increasingly being used by authors of systematic reviews, yet there are no clear guidelines or consensus regarding the principal statistical assumptions that affect OIS estimates. Most PRISMA Statement authors and systematic review authors expressed an interest in knowing more about OIS, despite a lack of familiarity with the methodology.</p> |
| spellingShingle | Meta-analysis Evidence based medicine Systematic reviews Garcia-Alamino, J Assessing the role of optimal information size in systematic reviews |
| title | Assessing the role of optimal information size in systematic reviews |
| title_full | Assessing the role of optimal information size in systematic reviews |
| title_fullStr | Assessing the role of optimal information size in systematic reviews |
| title_full_unstemmed | Assessing the role of optimal information size in systematic reviews |
| title_short | Assessing the role of optimal information size in systematic reviews |
| title_sort | assessing the role of optimal information size in systematic reviews |
| topic | Meta-analysis Evidence based medicine Systematic reviews |
| work_keys_str_mv | AT garciaalaminoj assessingtheroleofoptimalinformationsizeinsystematicreviews |