PROBAST (Prediction model Risk Of Bias ASsessment Tool) was developed for developing, validating, or updating/extending diagnostic and prognostic (multivariable) prediction model studies through a consensus process involving a group of experts in the field.

A multivariable prediction model describes any combination or equation of two or more predictors (age, biomarkers, disease stage, sex, signs, or symptoms) for estimating for an individual the probability or risk of having (diagnosis) or developing (prognosis) a particular outcome [1].

PROBAST comprises four (4) domains (analysis, outcome, participants, and predictors) covering 20 signaling questions to facilitate risk of bias (ROB) assessment. This explanation and elaboration document describes the rationale for including each domain and signaling question and guides researchers, reviewers, readers, and guideline developers in using them to assess the risk of bias and applicability concerns. All concepts are illustrated with published examples across different topics [1].

Applying PROBAST consists of 4 steps:

1. Specify your systematic review question(s).
2. Classify the type of prediction model evaluation.
3. Assess risk of bias and applicability (per domain).
4. Overall judgment of risk of bias and applicability.

PROBAST was designed for use in systematic reviews of prediction model studies and to address the lack of suitable tools designed specifically to assess ROB and applicability of primary prediction model studies. However, it can also be used as a general tool for critical appraisal of (primary) prediction model studies

• To assess varied forms of systematic review questions [1].
• To assess any diagnostic or prognostic prediction models geared at individualized predictions, regardless of the methods used to develop, validate, or update the model, outcomes being predicted, or predictors used [1].
• PROBAST addresses studies on multivariable models intended to make diagnostic and prognostic individualized predictions, including studies on [1]
  • Developing new prediction models
  • Developing and validating the same prediction models
  • Developing new prediction models versus validating existing models
  • Extending (e.g., adding new predictors to) or updating (e.g., adjusting model coefficients) existing prediction models
  • Validating existing prediction models, and
  • Combinations of these purposes.

• PROBAST is not designed to assess predictor-finding studies, particularly those where multivariable modeling aims to identify predictors associated with the outcome rather than to develop a model for individualized predictions [1].

• PROBAST is unsuitable for assessing comparative studies [1].

• PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration

[1] Moons KG, Wolff RF, Riley RD, Whiting PF, Westwood M, Collins GS, Reitsma JB, Kleijnen J, Mallett S. PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration. Annals of internal medicine. 2019 Jan 1;170(1):W1-33.