Back

Project DELTA - integrateD diagnostic solution forEarLy deTection of oesophageal cAncer: a study protocol for data mining and risk prediction algorithm development

Hippisley-Cox, J.; Coupland, C.; Mei, W. X.; Tan, P. S.; Fitzgerald, R.; Sasieni, P.; Ghimire, B.; Offman, J.

2021-02-18 primary care research
10.1101/2021.02.15.21251618 medRxiv
Show abstract

IntroductionOesophageal cancer is the sixth most common cause for cancer related deaths with over 450,000 new cases and 400,000 resulting deaths per year globally. Most cases in the UK are adenocarcinoma with some of the poorest outcomes from this cancer type in Europe -- mainly due to late diagnosis. The main risk factor for oesophageal adenocarcinoma is chronic reflux disease and due to the high prevalence and non-specific nature of these symptoms most patients are often managed with acid-reflux medications (e.g. Proton Pump Inhibitors (PPIs)) without referral for endoscopy. For those patients that are referred the endoscopy is normal in over 70% of cases, and there is not enough capacity within the NHS for endoscopy especially considering colon cancer screening. The primary aim of this project is to improve early identification of individuals at risk of oesophageal cancer and reduce over-use of prescription antacids. Methods and analysisWe will conduct a longitudinal cohort study consisted of adults 40 years and over who are free of oesophageal cancer at study entry, using the QResearch database for data gathered between 2000 and 2020. The main exposure is the use of prescription antacids which includes PPI, H2RA, and other aluminium and magnesium containing antacids. The exposure will be categorised based on active ingredients, dose, and duration of use and will be modelled as a time-varying covariate. Ethics and disseminationEthical approval for this project was obtained from the QResearch Scientific Committee [Ref: OX39, project title "DELTA - integrated Diagnostic solution for Early detection of Oesophageal cAncer"]. This project has been supported by patient and public involvement panels. We intend to submit the findings for peer-reviewed publication in an academic journal and disseminate them to the public. Strength and limitations of this studyO_LIThis is an open cohort study comprising a nationally representative sample of English population. C_LIO_LIThe cohort consists of GP clinic data linked to hospital records, the English national cancer registry and English national death registry. C_LIO_LIThis study has access to detailed information on acid-suppressant prescriptions, allowing analysis with consideration of the specific compound, dose, and duration of exposure. C_LIO_LIThis study is limited by high rates of missing data for cancer grade and stage, although completeness has improved in recent years, this will be accounted for using appropriate multiple imputation techniques. C_LI

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.