Advancing Education, Research, and Quality of Care for the Head and Neck oncology patient.
Introduction: With the rising proportion of elderly patients in the United States, more elderly patients are undergoing surgical intervention for head and neck cancer. Numerous studies have evaluated the impact of frailty scores on surgical outcomes. However, the use of natural language processing (NLP) to assess frailty in head and neck cancer patients remains understudied. NLP offers a promising approach to help clinicians easily identify patients with functional statuses that may be higher risk for surgical intervention.
Methods: We retrospectively reviewed all patients who underwent head and neck surgery with postoperative admission at UCSD between January 2023 and August 2023. An NLP model was developed and trained to extract data from clinical notes based on the Vulnerable Elders Survey-13 (VES-13). A frailty score was considered positive if the VES-13 score was ≥3. Logistic regression was applied to assess the relationship between postoperative complications and a positive frailty score.
Results: A total of 128 patients were included in the study, of whom 89 (69.5%) had a positive frailty score based on the NLP analysis. Nine patients (7.0%) experienced acute postoperative complications, including new-onset atrial fibrillation, postoperative infections, aspiration pneumonia, uncontrolled pain, asystole requiring CPR, sepsis, and death. Notably, all acute postoperative complications occurred in patients with a positive frailty score. A positive frailty score was also associated with a longer hospital stay (3.14 days vs. 1.38 days, p=0.04). No significant differences were observed in 30-day readmission rates or 30-day return to the operating room based on the NLP-derived frailty score.
Conclusion: NLP-derived frailty scores can provide valuable insights to clinicians, enabling better counseling of patients regarding their risk of acute postoperative complications