Article type
Year
Abstract
Background:
Most systematic reviews are planned after all or most eligible trials have completed and are based on aggregate data extracted from publications. Prior knowledge of trial results may introduce bias, and reliance on published data can limit the range of possible analyses and lead to reporting biases. Our framework for collaborative, prospective, adaptive meta-analysis of aggregate data, “FAME” (Tierney 2021), aims to tackle these issues. Planning prospectively reduces the potential for bias in review and meta-analysis methods. Collaborating with trialists provides information on when trials are due to complete so that the earliest timing for reliable meta-analysis can be assessed. Collecting more complete and detailed trial results allows for more thorough analyses and may further reduce possible sources of bias.
Objectives:
We aim to demonstrate how the use of all or even some FAME principles (“FAME-lite”) may improve the quality of Cochrane reviews.
Methods:
The FAME principles that might be adopted in Cochrane reviews are:
• Start the review early while most trials are ongoing or yet to report
• Collaborate with trialists to clarify trial status, eligibility and plans and to obtain information to inform risk of bias assessments
• Assess the earliest possible timing for reliable meta-analysis based on the accumulating data
• Develop protocol (before trial results are known) and collect harmonised and detailed trial results from trialists to allow more thorough and less biased analyses
• Interpret results, taking account of both available and unavailable data, and assess the value of updating
Results:
We will demonstrate the advantages of applying FAME in prostate cancer (Vale 2016; Rydzewska 2017; Burdett 2019; Vale 2020) and a similar prospective approach in COVID-19 (Sterne, 2020; Shankar-Hari, 2021). Recognising that it might not be possible or desirable to apply all elements of FAME, we will highlight those principles that might be used most readily in Cochrane (or other) aggregate data systematic reviews. We will also consider some of the potential challenges for Cochrane and suggest potential solutions.
Conclusions:
Adopting all or some FAME principles could help to reduce bias and improve the timeliness, trustworthiness and utility of Cochrane reviews.
Most systematic reviews are planned after all or most eligible trials have completed and are based on aggregate data extracted from publications. Prior knowledge of trial results may introduce bias, and reliance on published data can limit the range of possible analyses and lead to reporting biases. Our framework for collaborative, prospective, adaptive meta-analysis of aggregate data, “FAME” (Tierney 2021), aims to tackle these issues. Planning prospectively reduces the potential for bias in review and meta-analysis methods. Collaborating with trialists provides information on when trials are due to complete so that the earliest timing for reliable meta-analysis can be assessed. Collecting more complete and detailed trial results allows for more thorough analyses and may further reduce possible sources of bias.
Objectives:
We aim to demonstrate how the use of all or even some FAME principles (“FAME-lite”) may improve the quality of Cochrane reviews.
Methods:
The FAME principles that might be adopted in Cochrane reviews are:
• Start the review early while most trials are ongoing or yet to report
• Collaborate with trialists to clarify trial status, eligibility and plans and to obtain information to inform risk of bias assessments
• Assess the earliest possible timing for reliable meta-analysis based on the accumulating data
• Develop protocol (before trial results are known) and collect harmonised and detailed trial results from trialists to allow more thorough and less biased analyses
• Interpret results, taking account of both available and unavailable data, and assess the value of updating
Results:
We will demonstrate the advantages of applying FAME in prostate cancer (Vale 2016; Rydzewska 2017; Burdett 2019; Vale 2020) and a similar prospective approach in COVID-19 (Sterne, 2020; Shankar-Hari, 2021). Recognising that it might not be possible or desirable to apply all elements of FAME, we will highlight those principles that might be used most readily in Cochrane (or other) aggregate data systematic reviews. We will also consider some of the potential challenges for Cochrane and suggest potential solutions.
Conclusions:
Adopting all or some FAME principles could help to reduce bias and improve the timeliness, trustworthiness and utility of Cochrane reviews.