We appreciate the comments by Dr. Agarwal. Pitfalls of administrative databases are well recognized, and rightful concerns have been published in the past. The underestimation of postoperative stroke rates due to possible coding errors is a previously described limitation of the Nationwide Inpatient Sample database and was noted in our report as well. Nonetheless, strengths of Nationwide Inpatient Sample and its validation have also been described in previous publications and highlighted in detail in our Methods section. In our study, postprocedural complications were calculated using patient safety indicators, which were established by the Agency for Healthcare Research and Quality. We also identified other complications using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes in any secondary diagnosis field to prevent classification of pre-existing condition as a complication. Hertzer et al also agree that in-hospital death is unlikely to be miscoded along with cost of care or length of stay.

Symptomatic patients were identified if they had carotid artery stenosis with cerebral infarction or stroke (ICD 9 code: 433.11) as presenting diagnosis. Hence, the patients undergoing carotid artery stenting with only a principal diagnosis of stroke were classified as symptomatic. After excluding these symptomatic patients, any subsequent strokes were coded using ICD-9-CM code 997.X or 438.XX. Moreover, it has been previously noted that strict neurologic evaluation requirements for Centers for Medicare & Medicaid Services reimbursement are likely to result in more accurate identification of preprocedural symptomatic status. Supplementary Table 1 highlighted in the letter by Dr. Agarwal solely refers to the ICD codes used in calculating the Charlson co-morbidity index and not those used for classifying clinical status of patients with carotid artery stenosis in our study. Hence, the point of duplication of stroke in the preprocedural and postprocedural codes is incorrect.

Finally, comments of Dr. Agarwal, regarding duplication of ICD 9 code, does not affect the objective of our study which shows decrease in mortality, length of stay, cost of care and complications with increase in annual hospital volume.