MDADI as a Screening Tool for VFSS-Graded Swallowing Impairment in Head and Neck Cancer Patients.

Abstract

To determine clinically useful MD Anderson Dysphagia Inventory (MDADI) thresholds associated with Videofluoroscopic swallow study (VFSS)-graded swallowing impairment in head and neck cancer (HNC) patients and to explore whether a reduced-item MDADI could preserve discrimination while minimizing patient burden. This retrospective cross-sectional study analyzed HNC patients from a prospectively maintained dysphagia surveillance program who completed the MDADI within 120 days of VFSS. Associations between MDADI and VFSS metrics were assessed using Spearman and point-biserial correlations. Swallowing impairment was defined using Dynamic Imaging Grade of Swallowing Toxicity (DIGEST) and Penetration-Aspiration Scale (PAS) with clinically relevant dichotomization. Optimal MDADI composite score thresholds were derived using Youden's J statistic, and exploratory item reduction investigated the ability of a concise subset to preserve discrimination of clinically significant impairment. A total of 208 patients contributed 470 MDADI-VFSS observations for PAS and 443 for DIGEST (mean follow-up 3.7 years and MDADI-VFSS interval 56 ± 35.9 days). MDADI composite scores showed moderately strong and moderate inverse correlations with DIGEST (r = -0.63, p < 0.001) and PAS (r = -0.50, p < 0.001), respectively, with good discrimination across both outcome definitions (AUC 0.78-0.88). Optimal MDADI thresholds ranged from 74 to 78 with ~74 best identifying DIGEST ≥ 2. A four-item MDADI short form (F3, E3, E4, and P6) showed strong agreement with Composite score (r = 0.96), with good discrimination of DIGEST-defined impairment (AUC = 0.85). MDADI scores correlate with VFSS-graded impairment in HNC patients. A composite threshold near 74 may serve as a clinically useful screening parameter for identifying patients requiring swallowing evaluation.