Association between hospital interval and survival in patients with oral cancer: A waiting time paradox

BACKGROUND: In early diagnosis studies on symptomatic cancer, survival was the most recommended outcome. The magnitude and impact of the patient interval and primary care interval is well-known in oral cancer; however, the hospital interval and its influence on surviving this neoplasia are not well known. AIMS: To quantify the interval between the first contact with the specialist and the start of treatment for patients with oral cancer and to evaluate whether there was a link between this interval and disease survival. METHODS: We designed a hospital-based study that included 228 patients diagnosed with oral/oropharyngeal squamous cell carcinoma between 1998 and 2008 at A Coruña University Hospital (Spain) who were followed up until 2016. The data were extracted retrospectively from hospital medical charts. The study interval was defined in the context of the "pathways to treatment" model as the interval from the first specialist visit (start point) to the start of treatment (end point). We calculated the total interval (from first symptom to treatment) to evaluate the relative length of the hospital interval, and we considered the variables age, sex, location, comorbidity and tumour classification stage. Survival time was defined as the interval from the first treatment to death or censoring. RESULTS: The median hospital interval was 20 days, with an interquartile range of 15–29.1 days. The most relevant prognostic variable was the tumour stage (III-IV: Exp. ß = 2.8, p = 0.001). The hospital interval was part of the multivariate model, and its association with mortality showed a V-shaped association, where patients with short hospital intervals (3–18 days) and those with long hospital intervals (26–55 days) had significantly higher mortality than those with medium hospital intervals (19–25 days). CONCLUSION: The hospital interval represents a relevant interval for the patient’s path towards treatment, has prognostic implications and is subject to a severity bias (waiting time paradox) that should be avoided.