The amount data generated in medical research is overwhelming and evaluating and synthesizing information to make sense of it all can be a daunting task. Searches of literature databases often return published papers addressing the same issue, but with contradictory or equivocal results. To this end, systematic reviews that employ meta-analyses to synthesize results across multiple studies have proven to be useful evidence-based approaches and arrive at conclusions that can help guide clinical decision and policy making.
Traditionally, meta-analyses are based on aggregated data acquired from the study publications that are synthesized to generate summary estimates of effects. However, with the increased emphasis of data sharing, use of individual participant data (IPD) in meta-analyses (sometimes referred to as mega-analyses) have gained in popularity and are considered the “gold-standard” of meta-analyses. IPD meta-analyses are based on synthesis of raw, participant-level data across relevant studies that are pooled and reanalysed together as one dataset (one-stage approach) or a two-stage approach in which IPD from each study is reanalyzed separately to obtain relevant summary data which are then combined into the appropriate meta-analyses model. IPD meta-analyses have many statistical and clinical advantages over meta-analyses based on aggregate data, and can overcome selective reporting and publications biases. Furthermore, by accessing participant-level data, IPD meta-analyses allow researchers to explore additional analyses or outcomes than those reported in the original publication.
There are challenges, however, as IPD meta-analyses can be more time consuming and resource intensive, requiring development of collaborations and data-sharing agreements, as well data cleaning and harmonization across multiple heterogeneous datasets. Informatics’ platforms and data repositories are also needed to support data curation and sharing, and because participant-level data is exposed, consideration must be given to establishing a data governance framework that supports compliance to data privacy regulations and data security.
This Research Topic aims to explore the approaches and challenges of IPD analyses, as well as original research using this approach to advance evidence-based medicine. Potential subtopics include (but are not limited to):
• One stage versus two stage approaches
• Statistical issues and methods
• Data management and assessment of data integrity
• Data pooling and harmonization
• Platforms/repositories that support data sharing and management
• Data governance issues, including data protection and privacy
• Data de-identification methods
• Data sharing and collaboration
• Approaches to minimize potential biases
• Study selection processes and strategy
• Regulatory and ethical considerations
• Reporting guidelines (i.e., PRISMA-IPD)
• Historical perspective and reviews
• Original studies using IPD meta-analyses
We welcome all types of manuscripts, including original research papers, reviews and opinions.
The amount data generated in medical research is overwhelming and evaluating and synthesizing information to make sense of it all can be a daunting task. Searches of literature databases often return published papers addressing the same issue, but with contradictory or equivocal results. To this end, systematic reviews that employ meta-analyses to synthesize results across multiple studies have proven to be useful evidence-based approaches and arrive at conclusions that can help guide clinical decision and policy making.
Traditionally, meta-analyses are based on aggregated data acquired from the study publications that are synthesized to generate summary estimates of effects. However, with the increased emphasis of data sharing, use of individual participant data (IPD) in meta-analyses (sometimes referred to as mega-analyses) have gained in popularity and are considered the “gold-standard” of meta-analyses. IPD meta-analyses are based on synthesis of raw, participant-level data across relevant studies that are pooled and reanalysed together as one dataset (one-stage approach) or a two-stage approach in which IPD from each study is reanalyzed separately to obtain relevant summary data which are then combined into the appropriate meta-analyses model. IPD meta-analyses have many statistical and clinical advantages over meta-analyses based on aggregate data, and can overcome selective reporting and publications biases. Furthermore, by accessing participant-level data, IPD meta-analyses allow researchers to explore additional analyses or outcomes than those reported in the original publication.
There are challenges, however, as IPD meta-analyses can be more time consuming and resource intensive, requiring development of collaborations and data-sharing agreements, as well data cleaning and harmonization across multiple heterogeneous datasets. Informatics’ platforms and data repositories are also needed to support data curation and sharing, and because participant-level data is exposed, consideration must be given to establishing a data governance framework that supports compliance to data privacy regulations and data security.
This Research Topic aims to explore the approaches and challenges of IPD analyses, as well as original research using this approach to advance evidence-based medicine. Potential subtopics include (but are not limited to):
• One stage versus two stage approaches
• Statistical issues and methods
• Data management and assessment of data integrity
• Data pooling and harmonization
• Platforms/repositories that support data sharing and management
• Data governance issues, including data protection and privacy
• Data de-identification methods
• Data sharing and collaboration
• Approaches to minimize potential biases
• Study selection processes and strategy
• Regulatory and ethical considerations
• Reporting guidelines (i.e., PRISMA-IPD)
• Historical perspective and reviews
• Original studies using IPD meta-analyses
We welcome all types of manuscripts, including original research papers, reviews and opinions.
