Table 2Strength of evidence rating domains

DomainLevelCriteria
Risk of biasGeneralDegree to which studies have high likelihood of protection against bias; derived from assessment of the risk of bias in individual studies; incorporates both study design and conduct; grading this domain requires assessment of aggregate quality of studies within each major study design and integration into overall risk of bias score; limitations of design for reducing bias in addressing a Key Question should be taken into account. If studies differ substantially in risk of bias, may give greater weight to those studies with low risk of bias.
LowAt least 1 good quality RCT.
MediumAt least: 1 fair quality RCT;
or 1 good quality observational study and 1 additional study of good or fair quality.
HighDoes not meet minimum requirements for low or medium.
ConsistencyGeneralDegree to which studies are similar in effect sizes; degree to which studies have same direction of effect (even in presence of statistical heterogeneity).
ConsistentEffect sizes have same direction.
InconsistentEffect sizes are in different directions.
UnknownSingle study evidence base.
DirectnessGeneralA single direct link between intervention and health outcome; intervention and comparator(s) compared head-to-head within a study.
DirectDirect head-to-head comparison of interventions within a study or assesses a health outcome.
IndirectNot a direct head-to-head comparison of interventions within a study or assesses an intermediate outcome.
PrecisionGeneralDegree of certainty surrounding an effect estimate.
PreciseUncertainty around an effect compatible with only one of these: clinically important superiority, inferiority or noninferiority. In absence of meta-analysis, individual studies consistently report precise and/or statistically significant results.
ImpreciseUncertainty around an effect compatible with both clinically important superiority and inferiority. In absence of meta-analysis, individual studies do not consistently report precise and/or statistically significant results.

RCT = randomized controlled trial

From: Methods

Cover of Screening for Methicillin-Resistant Staphylococcus Aureus (MRSA)
Screening for Methicillin-Resistant Staphylococcus Aureus (MRSA) [Internet].
Comparative Effectiveness Reviews, No. 102.
Glick SB, Samson DJ, Huang E, et al.

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.