PROBAST
PROBAST is a tool that helps researchers and reviewers assess the quality and relevance of studies that develop, validate, or update prediction models in healthcare. It checks for bias and whether the results can be applied to the intended population and setting.
At a glance
Use when
Conducting systematic reviews of prediction model studies, developing clinical guidelines, or evaluating the credibility of published prediction models.
Avoid when
Assessing non-prediction model study designs (e.g., RCTs, observational studies for causal inference) or when no structured quality appraisal is needed.
Inputs
A study or set of studies on diagnostic or prognostic prediction models (e.g., model development, validation, or updating studies).
Outputs
Assessment of risk of bias and applicability across four domains, leading to an overall judgment of study quality and relevance.
How it works
PROBAST (Prediction model Risk Of Bias ASsessment Tool) is a validated checklist developed through expert consensus to evaluate risk of bias and applicability in studies involving diagnostic and prognostic prediction models. It comprises 20 signaling questions across four domains: participants, predictors, outcome, and analysis. It supports systematic reviewers in critically appraising prediction model studies for inclusion in reviews and guideline development.
- HTA domains
- Clinical Effectiveness, Patient and Social Aspects, Aspects Beyond HTA
- Assumptions
- The tool assumes that prediction model studies can be systematically evaluated using standardized criteria and that bias and applicability concerns can be identified through structured signaling questions.
- Strengths
- Developed through expert consensus and peer-reviewed publication.,Applicable to both diagnostic and prognostic prediction models.,Provides clear signaling questions with guidance and examples.,Freely available with supporting materials at www.probast.org.
- Limitations
- Requires trained users to interpret signaling questions correctly.,May not cover all context-specific biases in highly specialized areas.,Focused on study-level assessment, not model performance metrics.
- Also known as
- Prediction model Risk Of Bias ASsessment Tool
Questions this answers
- › Is the study sample representative of the target population?
- › Were predictors measured in a reliable and consistent way?
- › Was the outcome defined and assessed appropriately?
- › Was the statistical analysis methodologically sound?
- › Are the results applicable to the intended clinical or policy context?
- › What is the overall risk of bias in the prediction model study?
References & sources
Related methods
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.