Aim: To determine which features of history and demographics predict a diagnosis of malignancy or peptic stricture in patients presenting with dysphagia.
Methods: A prospective case-control study of 2000 consecutive referrals (1031 female, age range: 17-103 years) to a rapid access service for dysphagia, based in a teaching hospital within the United Kingdom, over 7 years. The service consists of a nurse-led telephone triage followed by investigation (barium swallow or gastroscopy), if appropriate, within 2 wk. Logistic regression analysis of demographic and clinical variables was performed. This includes age, sex, duration of dysphagia, whether to liquids or solids, and whether there are associated features (reflux, odynophagia, weight loss, regurgitation). We determined odds ratio (OR) for these variables for the diagnoses of malignancy and peptic stricture. We determined the value of the Edinburgh Dysphagia Score (EDS) in predicting cancer in our cohort. Multivariate logistic regression was performed and P < 0.05 considered significant. The local ethics committee confirmed ethics approval was not required (audit).
Results: The commonest diagnosis is gastro-esophageal reflux disease (41.3%). Malignancy (11.0%) and peptic stricture (10.0%) were also relatively common. Malignancies were diagnosed by histology (97%) or on radiological criteria, either sequential barium swallows showing progression of disease or unequivocal evidence of malignancy on computed tomography. The majority of malignancies were esophago-gastric in origin but ear, nose and throat tumors, pancreatic cancer and extrinsic compression from lung or mediastinal metastatic cancer were also found. Malignancy was statistically more frequent in older patients (aged >73 years, OR 1.1-3.3, age < 60 years 6.5%, 60-73 years 11.2%, > 73 years 11.8%, P < 0.05), males (OR 2.2-4.8, males 14.5%, females 5.6%, P < 0.0005), short duration of dysphagia (≤ 8 wk, OR 4.5-20.7, 16.6%, 8-26 wk 14.5%, > 26 wk 2.5%, P < 0.0005), progressive symptoms (OR 1.3-2.6: progressive 14.8%, intermittent 9.3%, P < 0.001), with weight loss of ≥ 2 kg (OR 2.5-5.1, weight loss 22.1%, without weight loss 6.4%, P < 0.0005) and without reflux (OR 1.2-2.5, reflux 7.2%, no reflux 15.5%, P < 0.0005). The likelihood of malignancy was greater in those who described true dysphagia (food or drink sticking within 5 s of swallowing than those who did not (15.1% vs 5.2% respectively, P < 0.001). The sensitivity, specificity, positive predictive value and negative predictive value of the EDS were 98.4%, 9.3%, 11.8% and 98.0% respectively. Three patients with an EDS of 3 (high risk EDS ≥ 3.5) had malignancy. Unlike the original validation cohort, there was no difference in likelihood of malignancy based on level of dysphagia (pharyngeal level dysphagia 11.9% vs mid sternal or lower sternal dysphagia 12.4%). Peptic stricture was statistically more frequent in those with longer duration of symptoms (> 6 mo, OR 1.2-2.9, ≤ 8 wk 9.8%, 8-26 wk 10.6%, > 26 wk 15.7%, P < 0.05) and over 60 s (OR 1.2-3.0, age < 60 years 6.2%, 60-73 years 10.2%, > 73 years 10.6%, P < 0.05).
Conclusion: Malignancy and peptic stricture are frequent findings in those referred with dysphagia. The predictive value for associated features could help determine need for fast track investigation whilst reducing service pressures.
Keywords: Barium swallow; Deglutition disorders; Dysphagia; Esophageal neoplasms; Esophageal stenosis; Gastroscopy; Predictive value of tests.