Article type
Year
Abstract
Background: The National Health and Medical Research Council (NHMRC) 'levels of evidence' hierarchy has been routinely utilised by Australian evidence-based medicine practitioners, clinical practice guideline developers and health technology assessment agencies since 1999. This hierarchy ranked interventional studies according to the likelihood that bias has been minimised by the design of the study but was not ideal to assess studies that addressed diagnostic, prognostic, aetiologic or screening research questions.
Objective: To revise and update the NHMRC hierarchy of evidence with the aim of retaining the existing levels I-IV but increasing its relevance for diagnostic, prognostic, aetiologic and screening research questions.
Methods: Methodological experts contracted by the NHMRC utilised empirical evidence, along with a commissioned literature review of existing schema for assessing bias in individual studies, to support decision-making regarding a revised hierarchy of evidence. This work was partnered by the development of a grading system to encompass the whole body of evidence.
Results: A suitable framework upon which to model the revised NHMRC hierarchy was identified, and reflected an individual study design's susceptibility to bias within the context of a particular research question. Consistency was maintained in the hierarchy of 'levels' across all research questions; empirical evidence was utilised to support the relationship between study design and ranking wherever possible; and systematic reviews of lower level studies were themselves ascribed a ranking. The impact of ethics on the hierarchy of study designs was acknowledged in the framework, along with a consideration of how harms (safety) should be assessed. Feedback received during the piloting and wide public consultation of this revised hierarchy will be presented, along with the final integration process.
Discussion: The revised NHMRC hierarchy of evidence is an interim measure that may go on to inform a formal NHMRC review within the next two years. It will ensure that systematic reviews, evidence-based clinical practice guidelines and health technology assessments that address interventional, diagnostic, prognostic, aetiologic and screening research questions have a common standard against which to judge the likelihood of bias in the design of the study.
Objective: To revise and update the NHMRC hierarchy of evidence with the aim of retaining the existing levels I-IV but increasing its relevance for diagnostic, prognostic, aetiologic and screening research questions.
Methods: Methodological experts contracted by the NHMRC utilised empirical evidence, along with a commissioned literature review of existing schema for assessing bias in individual studies, to support decision-making regarding a revised hierarchy of evidence. This work was partnered by the development of a grading system to encompass the whole body of evidence.
Results: A suitable framework upon which to model the revised NHMRC hierarchy was identified, and reflected an individual study design's susceptibility to bias within the context of a particular research question. Consistency was maintained in the hierarchy of 'levels' across all research questions; empirical evidence was utilised to support the relationship between study design and ranking wherever possible; and systematic reviews of lower level studies were themselves ascribed a ranking. The impact of ethics on the hierarchy of study designs was acknowledged in the framework, along with a consideration of how harms (safety) should be assessed. Feedback received during the piloting and wide public consultation of this revised hierarchy will be presented, along with the final integration process.
Discussion: The revised NHMRC hierarchy of evidence is an interim measure that may go on to inform a formal NHMRC review within the next two years. It will ensure that systematic reviews, evidence-based clinical practice guidelines and health technology assessments that address interventional, diagnostic, prognostic, aetiologic and screening research questions have a common standard against which to judge the likelihood of bias in the design of the study.
PDF