A web-based prediction score for head and neck cancer referrals

Area of Science:

• Oncology
• Medical Informatics
• Public Health

Background:

• The UK's National Health Service (NHS) Cancer Plan mandates specialist review within 2 weeks for suspected cancer referrals.
• Studies indicate low diagnostic yield (6.3%-14.6%) for head and neck cancers via the 2-week referral pathway, questioning its cost-effectiveness.
• Most head and neck cancer diagnoses occur through alternative referral routes, suggesting inefficiencies in the current system.

Purpose of the Study:

• To develop a predictive scoring system to identify patients at high risk for head and neck cancer.
• To aid general practitioners (GPs) in making more accurate and efficient 2-week wait referrals.
• To enhance the cost-effectiveness of cancer diagnostic pathways.

Main Methods:

• Retrospective analysis of data from 1075 patients with 2-week head and neck cancer referrals.
• Inclusion of patient demographics, risk factors, and investigation results.
• Application of logistic regression, machine learning, statistical inference, and model validation, with prospective testing on 235 patients.

Main Results:

• A logistic regression model was developed to predict the probability of head and neck cancer.
• A web-based prototype tool was created for GPs to calculate cancer probability and support referral decisions.
• The prototype demonstrated a sensitivity of 31% and a specificity of 92%.

Conclusions:

• A preliminary online scoring system prototype has been developed to assist GPs with head and neck cancer referrals.
• The system shows promising specificity (92%) for identifying patients unlikely to have cancer.
• Further data collection and model refinement are recommended to improve sensitivity and long-term utility for GPs.