The project aims to estimate the minimum threshold number of patients required for a reliable cancer survival estimate. The threshold number will assist analysts to optimise the level of granularity, in terms of geography or patient-subgroups, in official statistics. Greater granularity in official statistics drives more effective change in cancer services. Cancer patient net survival statistics become unreliable when the number of patients included in estimation is low. The 95% confidence interval about the estimate is too wide (lacking precision) and may not capture the true survival at the significance level stated (lacking accuracy). Reliable stand-alone (e.g. for rare cancers) or comparative (e.g. deprivation sub-groups) survival estimates are required to measure progress in cancer services over time and in different patient sub-groups. The PhD project will 1) explore the definition of ‘reliable’, 2) the various conditions upon which the minimum threshold may depend, 3) the potential of different statistical approaches (with their assumptions) to improve the reliability (precision and accuracy) of small-number estimates. The ultimate aim is to provide guidance on the minimum threshold number, a ‘rule-of-thumb’, required of a reliable survival estimate, but underpin this with a theoretical rationale that is supported by simulation experiments
Based on your current searches we recommend the following search filters.
Based on your current search criteria we thought you might be interested in these.