ePrints

The utility of a symptom model to predict the risk of oesophageal cancer

Lookup NU author(s): Dr Michael Mather, Emerita Professor Janet WilsonORCiD, Mary Doona, Dr Ben Talks, Professor Michael Griffin, Jason PowellORCiD, Dr Michael DrinnanORCiD

Downloads

• Published version [.pdf]

Licence

This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).

Abstract

© 2022 Royal College of Surgeons of Edinburgh (Scottish charity number SC005317) and Royal College of Surgeons in Ireland. Objective: To assess whether extra-oesophageal symptoms are predictive of oesophageal malignancy. Methods: A prospective, single-centre cross-sectional questionnaire study at a tertiary referral unit for oesophageal cancer using the Comprehensive Reflux Symptoms Scale (CReSS) questionnaire tool. Respondents with oesophageal malignancy were compared with historical cohorts undergoing airway examination or upper gastrointestinal endoscopy and found to have benign diagnoses. We developed a model for predicting oesophageal cancer using linear discriminant analysis and logistic regression, assessed by Monte Carlo cross validation. Results: Respondents with oesophageal malignancy (n = 146; mean age 70.5; male: female, 71:29) were compared with those undergoing airway examination (n = 177) and upper gastrointestinal endoscopy (n = 351), found to have benign diagnoses. No single questionnaire item, or group of co-varying items (factors), reliably discriminated oesophageal cancer from other diagnoses. Individual items which suggested higher risk of oesophageal malignancy included dysphagia (area under the curve (AUC) 0.68), low appetite (AUC 0.66), and early satiety (AUC 0.58). Conversely, throat pain (AUC 0.38), bloating (AUC 0.38) and heartburn (AUC 0.37) were inversely related to cancer risk. A forward stepwise regression analysis including a subset of 12 CReSS questionnaire items together with age and sex derived a model predictive of oesophageal malignancy in this cohort (AUC 0.89). Conclusion: We demonstrate a model comprised of 12 questionnaire items and 2 demographic parameters as a potential predictive tool for oesophageal malignancy diagnosis in this study population. Translating this model for predicting oesophageal malignancy in the general population is a valuable topic for future research.

Publication metadata

Author(s): Mather MW, Wilson JA, Doona M, Talks BJ, Fullard M, Griffin M, Powell J, Drinnan M

Publication type: Article

Publication status: Published

Journal: Surgeon

Year: 2023

Volume: 21

Issue: 2

Pages: 119-127

Print publication date: 01/04/2023

Online publication date: 15/04/2022

Acceptance date: 11/03/2022

Date deposited: 24/06/2022

ISSN (print): 1479-666X

ISSN (electronic): 2405-5840

Publisher: Elsevier Ltd

URL: https://doi.org/10.1016/j.surge.2022.03.006

DOI: 10.1016/j.surge.2022.03.006

PubMed id: 35431110

Altmetrics

Share