Predictive Modeling and Risk Stratification of Patients With Enlarged Vestibular Aqueduct. (18th November 2021)

• Record Type: Journal Article
• Authors: Farrokhian, Nathan; Kocharyan, Armine; Ruthberg, Jeremy; Piper, Robin; Rivas, Alejandro; Semaan, Maroun; Otteson, Todd; Manzoor, Nauman F.
• Is Part Of: Laryngoscope. Volume 132:Number 7(2022)
• Page Start: 1439
• Page End: 1445
• Publication Date: 2021-11-18
• Subjects: Enlarged vestibular aqueduct -- pediatric hearing loss -- predictive modeling; Otolaryngology -- Periodicals; 617.51005
• DOI: 10.1002/lary.29936
• Languages: English
• ISSNs: 0023-852X

Abstract

Abstract : Objectives/Hypothesis: To investigate patient‐specific characteristics that independently predict for progressive hearing loss in patients with enlarged vestibular aqueduct (EVA). Utilize multivariable predictive models to identify subgroups of patients with significantly different progression risks. Study Design: Retrospective analysis of patients evaluated at an academic tertiary care center. Cohort included 74 ears of patients with a diagnosis of EVA as defined by the Cincinnati criteria. Methods: Hearing trajectories were characterized, and a Kaplan–Meier estimator was utilized to determine progressive phenotype probabilities across the first 10 years after diagnosis. Cox proportional hazard regression was used to identify patient characteristics that independently altered this probability. Stratified risk groups were delineated from generated nomogram scores. Results: Male gender was associated with a 4.53 hazard ratio for progressive hearing loss (95% confidence interval [CI], 2.53 to 12.59). Each millimeter increase in operculum size was independently associated with an 80.40% increase in expected hazard (95% CI, 40.18 to 120.62). Each dB increase in air pure tone average at time of diagnosis decreased expected hazard by 1.59% (95% CI, −3.02 to −0.17). The presence of incomplete partition type II was associated with a 2.44 hazard ratio (95% CI, 1.04 to 5.72). Risk groups stratified by median nomogram score evidenced the discriminative ability of our model with the progression probability in the high‐risk group being six times higher at 1 year, nearly five times greater at 3 years, and three times greater at 9 years. Conclusions: EVA patient characteristics can be used to predict hearing loss probability with a high degree of accuracy (C‐index of 0.79). This can help clinicians make more proactive management decisions by identifying patients at high risk for hearing loss. Level of Evidence: 4 Laryngoscope, 132:1439–1445, 2022