Framework for assessing the internal validity of studies of prognosisa

Study FeatureQualities Sought
PatientsRecruited at a common point (usually early) in the course of the disease
Ideally complete (all eligible patients included)
Source and selection of patients is explained
Diagnostic and inclusion criteria well-described and defined
Clinical and demographic characteristics fully described
Follow-upSufficiently long to be meaningful
OutcomeAppropriate
Unbiased
Assessed blinded to prognostic information
Fully defined and known for all patients
Candidate prognostic variablesFully defined, including method of measurement
Available for all patients
AnalysisContinuous predictors analyzed appropriately
Statistical adjustment for all important factors
TreatmentsRandomized or at least standardized
Fully described and explained
  • a Modified from Altman.10