In Reply We thank Dr Palatini for his interest in our study and agree that observational studies such as ours might be affected by unmeasured confounding factors, as we acknowledged in the article. Indeed, we made extensive efforts to exclude patients treated with β-blockers owing to cardiac or liver disease and further controlled statistically for a number of other potential confounding factors, using well-accepted techniques to minimize confounding. Despite all of these measures, some bias may persist due to imbalances in unmeasured factors, so the results should be interpreted accordingly.