Evidence-appraisal glossary
C-statistic
The c-statistic (concordance statistic) measures how well a prediction model separates people who have the outcome from those who do not. It is the probability that a randomly chosen person with the event has a higher predicted risk than one without. 0.5 means chance; 1.0 means perfect ranking.
Also called: Concordance statistic, Concordance index, Harrell's C.
What it is
The c-statistic quantifies a model's discrimination: its ability to rank higher-risk people above lower-risk people. For a yes/no outcome it equals the area under the ROC curve (AUC). For time-to-event data it generalizes to Harrell's concordance index, which handles censoring. Conventionally, 0.7 to 0.8 is acceptable, 0.8 to 0.9 is strong, and above 0.9 is outstanding; 0.5 is no better than a coin flip.
How to use it when reading a study
- Read it as ranking quality, not accuracy: a high c-statistic does not mean predicted risks are numerically correct. Check calibration separately.
- Look for a confidence interval and whether it came from an external validation sample; c-statistics inflate on the data used to build the model.
- A model can have a respectable c-statistic yet add little over existing predictors, so ask what it improves upon.
- Very high values can signal overfitting or leakage rather than real skill.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.