Article type
Year
Abstract
Background: Minimizing risk of bias in animal research is essential for preserving scientific integrity and ensuring that clinical trials in humans are based on valid evidence. Numerous guidelines have been developed to improve the design and execution of preclinical drug and animal toxicology research, yet little has been done to systematically gather and evaluate the assessment instruments.
Objectives: We performed a systematic review to identify and evaluate instruments for assessing risk of bias and/or reporting research in preclinical and animal toxicology research.
Methods: We searched Medline from January 1966 to November 2011 to identify all articles meeting our inclusion criteria: (1) published report focusing on the development of a quality assessment instrument for animal studies and (2) written in the English language. We extracted data on risk of bias criteria (e.g. randomization, blinding, allocation concealment) and other study design features. We recorded the number of criteria assessed by each instrument, whether and how the instrument calculated an overall quality score, and the method used to develop the criteria.
Results: Thirty distinct instruments were identified. Core methodological criteria included randomization (25 instruments), blinding (23 instruments), and sample size calculation (18 instruments). In general, authors failed to empirically justify why these criteria were included, and defended their recommendations qualitatively, through consensus, or by citing evidence of risk of bias in human studies. Furthermore, 7 instruments calculated a score for assessing methodological quality, 29 instruments have not been tested for validity or reliability, and the number of criteria assessed by these instruments ranged from 2 to 26.
Conclusions: This review identified core quality assessment criteria for animal research and provides the rationale for empirically testing additional criteria. Furthermore, we recommend testing these instruments for validity and reliability so that risk of bias in animal research can be accurately assessed.
Objectives: We performed a systematic review to identify and evaluate instruments for assessing risk of bias and/or reporting research in preclinical and animal toxicology research.
Methods: We searched Medline from January 1966 to November 2011 to identify all articles meeting our inclusion criteria: (1) published report focusing on the development of a quality assessment instrument for animal studies and (2) written in the English language. We extracted data on risk of bias criteria (e.g. randomization, blinding, allocation concealment) and other study design features. We recorded the number of criteria assessed by each instrument, whether and how the instrument calculated an overall quality score, and the method used to develop the criteria.
Results: Thirty distinct instruments were identified. Core methodological criteria included randomization (25 instruments), blinding (23 instruments), and sample size calculation (18 instruments). In general, authors failed to empirically justify why these criteria were included, and defended their recommendations qualitatively, through consensus, or by citing evidence of risk of bias in human studies. Furthermore, 7 instruments calculated a score for assessing methodological quality, 29 instruments have not been tested for validity or reliability, and the number of criteria assessed by these instruments ranged from 2 to 26.
Conclusions: This review identified core quality assessment criteria for animal research and provides the rationale for empirically testing additional criteria. Furthermore, we recommend testing these instruments for validity and reliability so that risk of bias in animal research can be accurately assessed.