Magnetic resonance imaging has a limited role in investigating odynophagia if examination is normal: A binary logistic regression analysis of 484 patients presenting to a tertiary head and neck clinic

Introduction Investigation of head and neck cancers places a significant burden on the National Health Service and effective resource allocation is of perpetual importance. Existing risk calculators are designed to stratify the likelihood of underlying malignancy according to symptoms, but are less relevant in secondary care as they do not integrate clinical examination findings (e.g. naso‐endoscopy). We looked to evaluate the utility of magnetic resonance imaging (MRI) in investigating patients with unilateral odynophagia and a normal clinical examination. Methods A retrospective analysis was conducted over a 54‐month period; 484 consecutively‐recruited adult patients who underwent MRI of the neck for suspected malignancy were included. Imaging reports, case notes, and histopathology results were reviewed. Patients with previously diagnosed/treated malignancy, undergoing surveillance, or those with pathology of the salivary glands, oral cavity or thyroid gland, were excluded. A multivariate binary logistical regression model was performed to calculate the odds ratios and probabilities of malignancy for each presenting symptom, with and without negative nasoendoscopy findings. Results The overall incidence of malignancy within the cohort was 173/484 patients (35.7%; 95% CI 31.5%–40.2%) with no cases of malignancy in patients presenting with odynophagia and a normal nasoendoscopy (0/39). The presence of a neck lump was significantly associated with malignancy, (odds ratio [OR] 2.03, p = .001; 95% CI 1.59–2.58), as was dysphagia (OR 1.52, p = .009; 95% CI 1.11–2.11). Conversely, globus was found to have an inverse association (OR 0.41, p