Response to letter to the Editor on ‘Utility of the Hospital Frailty Risk Score in patients undergoing endovascular treatment for ruptured aneurysms’

We greatly appreciate the letter by Estes et al sent in regard to our published work “Higher Hospital Frailty Risk Score (HFRS) is associated with increased complications and healthcare resource utilization after endovascular treatment of ruptured intracranial aneurysms.” In that study, we sought to assess the potential use of HFRS in identifying the ‘frailty risk’ of patients with ruptured aneurysms undergoing endovascular treatment, leveraging the use of ICD10CM coding that is available within the National Inpatient Sample. The authors have raised a few concerns, including the lack of temporality in the National Inpatient Sample to distinguish between preoperative and postoperative conditions, the limited external validity of patients outside of age and neurosurgical context, and the poor discriminatory accuracy of postoperative outcomes. We concur with the authors’ concerns about the use of the National Inpatient Sample, as the dataset is limited to single inpatient hospital admissions. We also agree that single coding discrepancies may exist, such as the initial billing coding, timeline of neurological deficits and medical complications, baseline comorbidities that were newly diagnosed during the indexed admission (but might have also been chronic), and postoperative complications. We agree that more delineated coding methodologies are necessary to calculate an accurate frailty coding system limiting potential misclassification bias. As eluded to by the authors, this is one of the significant limitations using large administered billing datasets compared with other frailty tools such as the Risk Analysis Index. However, a metric based entirely on ICD10 coding also has the key strength of being able to be calculated quickly on, or even prior to, admission. Implementation of an automated frailty score within the hospital electronic medical record, such as the HFRS, could help inform timely clinician decisionmaking and warrants further investigation. Furthermore, we agree with the authors concerns as the original Lancet authors acknowledged poor discrimination of their adjusted model on predicting postoperative outcomes. However, by way of comparison, more recent studies have attempted to explore the validation of the HFRS in unique disease pathologies, such as acute heart failure, myocardial infarction, and stroke, and have seemed to achieve good discrimination (ie, Cstatistics 0.70 or above). In addition, other studies have recently incorporated HFRS evaluation for younger patient populations and shown similar postoperative outcomes. This underscores the importance of investigating and validating frailty indices for unique pathologies in their specific disease context, as patients might have different baseline characteristics, comorbidities, and clinical presentation. A better way in which to investigate the utility of these frailty indices would be through multiinstitutional, prospective studies, which could reduce the limitations and in which, coding might be more refined. In conclusion, we again thank the authors for their comments, which have given us the opportunity to clarify our work. Ultimately, our goal was to evaluate the potential use of a new frailty tool and to create awareness on this topic, while understanding the limitations above. We agree with the authors that until these results are validated, the utility of this frailty index in riskstratifying endovascular management of ruptured intracranial aneurysms may be limited. In the future, further welldesigned, prospective institutional studies that have both the granularity and temporality to accurately define frailty will be necessary to help care for, and riskstratify, our patients.