Article type
Year
Abstract
Background:
Systematic reviews (SRs) and meta-analyses (MAs) of prediction model studies range in breadth and can aim to identify, appraise and summarise the evidence about existing models and their predictive performance. They differ in many important ways from SRs of intervention studies. Emphasis is on predictive performance of models obtained from internal or external validation, rather than effect sizes. Although TRIPOD provides guidance for reporting primary prediction model studies and PRISMA 2020 is available for SRs of interventions, there is currently no tailored guideline for reporting SRs of prediction model studies.
Objectives:
To develop a tailored checklist for the transparent reporting of SRs and MAs of diagnostic and prognostic prediction model studies.
Methods:
We formed an executive committee responsible for developing TRIPOD-SRMA. Reporting items from existing guidelines such as TRIPOD, PRISMA 2020 and others were used to produce the initial checklist. A modified Delphi approach was then used to elicit views from a wider group of experts with experience of SRs and primary prediction model studies. In the first round, 86 individuals were invited to participate, of whom 43 participants responded and formed the Delphi panel. Online surveys were used to inform the consensus process on which items to include and gather feedback to refine the checklist. After two Delphi rounds and consideration of all feedback, TRIPOD-SRMA was finalised and approved by all members of the executive committee.
Results:
The new TRIPOD-SRMA checklist will be presented. It consists of 26 reporting items within six sections. Some reporting items remain unchanged or minimally modified from PRISMA 2020 as there are common elements to all SRs. However, 11 items have been tailored or included specifically to address SRs of prediction model studies. A checklist for abstracts is also included.
Conclusions:
TRIPOD-SRMA and the accompanying checklist for abstracts provide the first reporting guideline for SRs and MAs of prediction model studies. Use of TRIPOD-SRMA by journals and authors will help to improve transparency of SRs of prediction model studies and ensure readers can more easily evaluate the evidence for existing prediction models within a clinical field.
Patient, public and/or healthcare consumer involvement: None.
Systematic reviews (SRs) and meta-analyses (MAs) of prediction model studies range in breadth and can aim to identify, appraise and summarise the evidence about existing models and their predictive performance. They differ in many important ways from SRs of intervention studies. Emphasis is on predictive performance of models obtained from internal or external validation, rather than effect sizes. Although TRIPOD provides guidance for reporting primary prediction model studies and PRISMA 2020 is available for SRs of interventions, there is currently no tailored guideline for reporting SRs of prediction model studies.
Objectives:
To develop a tailored checklist for the transparent reporting of SRs and MAs of diagnostic and prognostic prediction model studies.
Methods:
We formed an executive committee responsible for developing TRIPOD-SRMA. Reporting items from existing guidelines such as TRIPOD, PRISMA 2020 and others were used to produce the initial checklist. A modified Delphi approach was then used to elicit views from a wider group of experts with experience of SRs and primary prediction model studies. In the first round, 86 individuals were invited to participate, of whom 43 participants responded and formed the Delphi panel. Online surveys were used to inform the consensus process on which items to include and gather feedback to refine the checklist. After two Delphi rounds and consideration of all feedback, TRIPOD-SRMA was finalised and approved by all members of the executive committee.
Results:
The new TRIPOD-SRMA checklist will be presented. It consists of 26 reporting items within six sections. Some reporting items remain unchanged or minimally modified from PRISMA 2020 as there are common elements to all SRs. However, 11 items have been tailored or included specifically to address SRs of prediction model studies. A checklist for abstracts is also included.
Conclusions:
TRIPOD-SRMA and the accompanying checklist for abstracts provide the first reporting guideline for SRs and MAs of prediction model studies. Use of TRIPOD-SRMA by journals and authors will help to improve transparency of SRs of prediction model studies and ensure readers can more easily evaluate the evidence for existing prediction models within a clinical field.
Patient, public and/or healthcare consumer involvement: None.