Drs. Evans and Reilly raise several points regarding possible sources of bias in our study. We do not concur with their statement that "many study participants with missing or uninterpretable test results were dropped." Of the 1098 patients admitted with chest pain during the study time frame, 459 met entry criteria and were enrolled. As stated in our paper, most patients who were excluded did not have active chest pain while under medical supervision or could not quantify their chest pain. These patients did not receive therapy and thus were not candidates for the study.

We indeed focused on patients with active chest pain in the emergency department. While this did eliminate patients who had pain only before presentation, this approach was necessary to minimize recall bias in patients who had taken their own nitroglycerin before coming to the hospital.

Work-up bias was a potential confounder in our study, an issue that we addressed in the Discussion section. We limited the study to patients admitted to the hospital in order to enroll patients with an adequate burden of disease. Patients discharged from the emergency department commonly have a very low likelihood of active coronary artery disease. We focused our study on the higher-risk patients admitted to the hospital, in whom active disease was more likely. We do not have any formal data on pain relief by nitroglycerin in patients discharged from the emergency department and concur that this is worthy of future investigation.

In summary, we appreciate and have previously acknowledged some of the concerns raised by Drs. Evans and Reilly. Expanding the study sample to include any and all patients with chest pain dilutes and negates the very specific question our study posed: In patients admitted with suspected myocardial ischemia, does nitroglycerin-induced pain relief suggest ischemia?