A predictive model for swallowing dysfunction after oral cancer resection

• Published in: British journal of oral & maxillofacial surgery Vol. 59; no. 9; pp. 1043 - 1049
• Main Authors: Ferro, A., Basyuni, S., Bosley, R., Santhanam, V.
• Format: Journal Article
• Language: English
• Published: Scotland Elsevier Ltd 01.11.2021
• Subjects: Deglutition; Enteral Nutrition; Feeding Issues; head and neck oncology; Humans; Length of Stay; Mouth Neoplasms - surgery; reconstruction; Retrospective Studies; Tracheostomy; Feeding Issues; reconstruction; Enteral Nutrition; head and neck oncology
• ISSN: 0266-4356; 1532-1940
• DOI: 10.1016/j.bjoms.2021.01.007

Aggressive therapy of oral cancers is associated with significant postoperative morbidity. Patients with feeding issues may require nutritional support. In our unit, patients identified as developing feeding issues are reactively referred for specialist input through a feeding issues multidisciplinary team meeting (FiMDT). Reactive feeding increases length of patient stay (LOS) and may contribute to patient morbidity. We aimed to develop a model to pre-emptively identify patients likely to develop feeding issues postoperatively, to facilitate the establishment of a preoperative referral pathway to increase patient flow. All referrals to a Head and Neck multidisciplinary team meeting over a five-year period were identified and preoperative factors were extracted. Linear regression was used to confirm that FiMDT was an independent predictor of LOS. Logistic regression was used to determine if referral to FiMDT could be predicted based on preoperative factors only. A total of 203 patients met inclusion criteria for analysis. Inpatient referral to FiMDT was an independent predictor of LOS. Significant predictors of inpatient FiMDT referral included tracheostomy, patient age, and alcohol intake. The resulting model was 90% sensitive and 93.8% specific with a threshold of 0.2. We have shown that inpatient FiMDT referral is an independent predictor of patient length of stay, and that the odds of referral can be robustly predicted. We aim to use this model in redirecting emphasis to a preoperative referral pathway for improved patient flow.