Abstract
BACKGROUND: Current guidelines recommend endoscopic surveillance for Barrett oesophagus (BE), but the value of surveillance is still debated. Using a combination of primary care, secondary care and cancer registry datasets, we examined the impact of a prior BE diagnosis, clinical and risk factors on survival from oesophageal cancer and adenocarcinoma. METHODS: Retrospective cohort study of patients aged 50 and above diagnosed with malignant oesophageal cancer between 1993 and 2014 using Clinical Practice Research Datalink (CPRD). All prior BE diagnoses and endoscopies were identified from CPRD and Hospital Episode Statistics. Histology information was obtained from linked cancer registry data. We used flexible parametric models to estimate excess hazard ratios (EHRs) for relative survival. We simulated the potential impact of lead-time by adding random lead-times from a variety of distributions to all those with prior BE. RESULTS: Among our oesophageal cancer (n = 7503) and adenocarcinoma (n = 1476) cohorts only small percentages, 3.4% and 5.3%, respectively, had a prior BE diagnosis. Two-year relative survival was better among patients with BE: 48.0% (95% CI 41.9-54.9) compared to 25.2% (24.3-26.2) without. Patients with BE had a better prognosis (EHR = 0.53, 0.41-0.68). Survival was higher even if patients with BE had fewer than two endoscopies (50.0%; 43.6-57.3). A survival benefit was still observed after lead-time adjustment, with a 20% absolute difference in 2-year survival using a 5 year mean sojourn time. CONCLUSIONS: Patients with a prior BE diagnosis had a survival advantage. This was not fully explained by surveillance endoscopies.