LAUSR

factors not captured in any of the 3 risk scores are needed in a new scoring system to more accurately predict rebleeding, death, or surgery. This is consistent with the poor score performances that we reported. However, we are not clear what Dr Fan meant by his description of our statistical approach in assessing 3 prognosis scores as “arbitrary.” We reported the commonly utilized statistics for determining accuracy (eg, sensitivity, specificity, overall accuracy, ROC area) for the 3 scores alone, showing that their accuracy was poor. This is not arbitrary but we agree it does not use any other information. There is certainly a need for improvement in scoring systems for predicting risk and outcomes of severe LGI bleeding and we look forward to seeing and using them clinically or in future research.