Article type
Abstract
Background
We undertook a systematic review comparing irradiation to no irradiation of platelets for transfusion. Forty-four studies were included, of which 43 were in-vitro studies of paired/split samples.
A suitable tool for risk of bias (ROB) assessment of these studies could not be identified, as ROB tools currently available focus on clinical or animal research, and some questions, prompts, and domains are irrelevant to paired or split samples, and lack assessment of other potential sources of bias.
Objective
To develop a ROB tool for laboratory studies using paired (split) samples.
Methods
We examined widely accepted and well-validated ROB tools for clinical and animal studies, and ultimately adapted a tool used in animal studies (SYRCLE, Hooijmans 2014 (BMJ)). We tested and implemented our changes in an iterative process across the 43 in-vitro studies included in our review.
Results
We used the domain headings and assessments (low, high, unclear) of the original Cochrane ROB tool and have compared the two tools in Table 1. Our description of each domain and associated prompts focuses the user on considerations along the pathway of the components being assessed.
Notable differences include:
Selection bias: including only paired samples in in-vitro studies removes issues around baseline imbalance that randomisation is designed to overcome.
Performance bias: blinding of participants is not applicable to in-vitro samples, but storage of samples becomes more important.
Detection bias: many outcomes in in-vitro studies are objective, but quality assurance and/or calibration of equipment becomes more important.
Conclusions
We have developed and used a ROB tool suitable for use in paired sample in-vitro studies that is simple to use, and includes assessment of all domains laid out by Cochrane.
This new ROB tool should help reviewers assess the risk of bias of in-vitro studies, and highlights the importance of clear reporting for the scientists conducting the trials, resulting in robust evidence production and reporting.
Further development and validation of the tool is required.
We undertook a systematic review comparing irradiation to no irradiation of platelets for transfusion. Forty-four studies were included, of which 43 were in-vitro studies of paired/split samples.
A suitable tool for risk of bias (ROB) assessment of these studies could not be identified, as ROB tools currently available focus on clinical or animal research, and some questions, prompts, and domains are irrelevant to paired or split samples, and lack assessment of other potential sources of bias.
Objective
To develop a ROB tool for laboratory studies using paired (split) samples.
Methods
We examined widely accepted and well-validated ROB tools for clinical and animal studies, and ultimately adapted a tool used in animal studies (SYRCLE, Hooijmans 2014 (BMJ)). We tested and implemented our changes in an iterative process across the 43 in-vitro studies included in our review.
Results
We used the domain headings and assessments (low, high, unclear) of the original Cochrane ROB tool and have compared the two tools in Table 1. Our description of each domain and associated prompts focuses the user on considerations along the pathway of the components being assessed.
Notable differences include:
Selection bias: including only paired samples in in-vitro studies removes issues around baseline imbalance that randomisation is designed to overcome.
Performance bias: blinding of participants is not applicable to in-vitro samples, but storage of samples becomes more important.
Detection bias: many outcomes in in-vitro studies are objective, but quality assurance and/or calibration of equipment becomes more important.
Conclusions
We have developed and used a ROB tool suitable for use in paired sample in-vitro studies that is simple to use, and includes assessment of all domains laid out by Cochrane.
This new ROB tool should help reviewers assess the risk of bias of in-vitro studies, and highlights the importance of clear reporting for the scientists conducting the trials, resulting in robust evidence production and reporting.
Further development and validation of the tool is required.