In Reply Sun and colleagues raise important points regarding primary hemorrhage, surgical technique, and nonopioid analgesics. Our decision to not examine primary hemorrhage in our study 1 was driven by 2 considerations. First, primary hemorrhage occurs within 24 hours of surgery (eg, on the day of surgery or the day after). Therefore, any putative effect of perioperative opioid prescribing on the risk of primary hemorrhage (eg, by reducing the use of nonsteroidal anti-inflammatory drugs [NSAIDs]) would have to be strong and immediate. However, prior studies do not suggest that the association between NSAID use and hemorrhage is large, if it exists at all. [2][3][4][5] Given the lack of a reasonable a priori mechanism between perioperative opioid prescribing and primary hemorrhage, any detected association would most likely be incidental or due to confounding. Second, our definition of perioperative opioid prescription included opioid prescriptions filled from a week before surgery to 1 day after surgery. To establish a clear temporal relationship between perioperative opioid prescriptions and primary hemorrhage, we would have needed to exclude prescriptions filled the day after surgery while only analyzing primary hemorrhages that occurred on the day after surgery.The relationship between surgical technique and primary hemorrhage warrants further study. Although data on surgical technique was not available in our claims database, recently launched data registries such as RegENT may be useful for investigating this relationship in the future. 6 Clinical trials may also be useful, but these studies would be costly and would need to enroll a large number of patients to achieve sufficient power to detect differences in primary hemorrhage between techniques given the relative rareness of this complication.We agree with Sun and colleagues that information on nonopioid medications would have added valuable information to this study. However, we were limited by our insurance claims database. Our data would have allowed us to estimate NSAID use by analyzing perioperative NSAID prescriptions filled through insurance, but this analysis would have far underestimated the true rate of NSAID use. This is because NSAIDs are available over-the-counter at prices that are typically lower than the amount privately insured patients have to pay to fill NSAID prescriptions through insurance in the United States.Finally, we agree that an important objective for future studies is collecting data on opioid consumption, pain, and complications after tonsillectomy. We are currently collecting such data for adolescents and young adults undergoing tonsillectomy. We are currently planning such a study and look forward to disseminating our results to the clinical and research community in the near future.