Development and Validation of A Predictive Scoring Model for Complications following Endoscopic Endonasal Skull Base Surgery

Objective: The endoscopic endonasal approach (EEA) has evolved from an unconventional procedure to an established technique in skull base surgery. A prior study from our institution of 1002 patients who underwent EEA in the modern era (2010–2018) reported the safety and risk factors associated with perioperative and postoperative complications. We now have developed a scoring model based on risk factors in this prior cohort to predict complications following EEA. The initial model was then validated on a cohort of patients who underwent EEA at our center since the original study, and the model was subsequently refined to improve predictive ability.

Methods: After obtaining institutional review board approval, the authors analyzed data from our initial 1,002 patient cohort and developed a scoring model that most accurately predicted postoperative complications as evidenced by the area under the receiver operating characteristic (ROC) curve. A retrospective review of intraoperative and postoperative complications was then conducted on patients from years 2018 to 2020 immediately following the initial group of 1,002 to serve as a validation cohort.

Results: The authors identified 295 validation cohort patients that met the inclusion criteria. The most common pathology was pituitary adenoma (n = 151 [51.2%]), followed by meningioma (n = 25 [8.5%]). There were no intraoperative deaths. Higher grade complications (those requiring reoperation, ICU level care, or death) were identified after 33 (11.1%) EEA cases. The initial scoring model contained four variables (age, pathology type, case complexity, and approach type) and yielded area under ROC of 0.723 and 0.717 for high-grade complications in the initial and validation cohorts, respectively. A slightly refined scoring model using the new patient cohort identified four variables (age, pathology type, case complexity, and operative time) yielding an ROC curve area of 0.744 for high-grade complications in this validation cohort. This final scoring model demonstrated good ability to differentiate between low-risk (score = 0, n = 126, complication rate = 2.4%), medium-risk (score 1 or 2, n = 105, complication rate = 12.4%), and high-risk (score 3+, n = 64, complication rate = 26.5%) procedures when assessing higher grade complications.

Conclusion: This EEA scoring model is simple and can be utilized in preoperative counseling with patients and in discussions with other providers. A high score in this complication model does not suggest that a particular patient is not a candidate for surgery but rather, it highlights the importance of operating with caution and counseling the patient and their family appropriately. Patients with complex skull base pathology have inherently high-complication rates with open approaches and EEA. Additionally, this model can be used to define how complex a patient cohort is in any EEA publication, provided they include patient age, pathology, surgical approach, and operative times. This may help reconcile disparate complication rates in smaller, single-center series. Multicenter external validation of this model is the next step in our current research.

Fig. 1

Fig. 2

Publication History

Article published online:
15 February 2022