In January 2017, the National Lung Cancer Audit (NLCA) will publish its first annual report based on data sourced from the National Cancer Registration and Analysis Service (NCRAS). Transition to this new methodology has proved to be difficult for some trusts, and anxieties about the accuracy of the data has prompted enquiries to the NLCA project team that are worthy of further explanation and discussion.
All trusts now submit the Cancer Outcomes and Services Dataset (COSD) to NCRAS, and these data are used to provide contemporaneous feedback on data completeness (as well as performance data) on the CancerStats portal. For NHS trusts in England, NCRAS receives data from multiple other sources as well as COSD, allowing registration staff to build the most accurate and complete cancer registration, and it is these fully processed data that are analysed by the NLCA team to produce an annual report. As a result, trust-level data in the annual report may be slightly different to the data that trusts expect based on CancerStats.
Lung cancer care is increasingly complex, and it is now not unusual for patients to move across different organisations for their diagnosis and treatment.
Furthermore, NCRAS is now able to identify a substantial number of ‘extra cases’ (approximately 6,000 in England for 2015) compared with historical Lung Cancer Audit Dataset (LUCADA) records. Naturally, trusts have questioned why their data include more cases than they were expecting, and local analysis has confirmed that these are real cases that have been appropriately allocated, but that they were not picked up by the multidisciplinary team (MDT) for a variety of reasons. The NLCA will continue to include these cases because the audit is commissioned to report on all cases of lung cancer, and not just those known to the MDTs.
Lung cancer care is increasingly complex, and it is now not unusual for patients to move across different organisations for their diagnosis and treatment. Historically, we have always tried to allocate patients to the ‘trust first seen’ because this is usually the site of the decision-making MDT, and the same approach is appropriate for the cancer registration data. Unfortunately, in around 18% of patients these data are missing, and some cases have conflicting entries. In order that these cases are not lost to the audit analysis, we have developed an algorithm that uses other data supplied to NCRAS to determine the most likely ‘trust first seen’. Testing of the algorithm shows that it is very accurate, although inevitably some errors do occur. We urge trusts to ensure that the ‘trust first seen’ field is completed accurately in their COSD submissions.
The NLCA team welcome feedback, both good and bad, and we always try to respond constructively. For those organisations that are still struggling with discrepancies between data they submit and data in the annual report, local NCRAS data improvement leads can supply patient-level data to help them to understand it.
Paul Beckett, NLCA clinical lead