Why does it take so long to publish data?

Our most recent annual report, published at the BTOG conference in Dublin in January 2018, included data on patients diagnosed in 2016. Why does it take so long to publish data? To answer this question, it is useful to look in detail at the process outlined in the graphic available to download below.

In England, all hospitals are mandated to submit data on patients diagnosed with cancer to the National Cancer Registration and Analysis Service (NCRAS). This dataset is called the Cancer Outcomes and Services Dataset (COSD) and includes general cancer data fields as well as some lung-specific ones (such as lung function). This data is usually collected through the local multidisciplinary team (MDT) meetings, recorded on local MDT software systems, and then uploaded on a monthly basis running about a month behind the date of diagnosis.

Once NCRAS receive notification of a cancer diagnosis through a COSD submission, an initial cancer registration record is created. This is done relatively quickly and allows the NLCA to report back to trusts on their data completeness though the CancerStats web portal. However, this information has not been checked for accuracy and will not contain details of the treatment that the patient receives over the subsequent weeks and months.

As new information about the patient's treatment and outcomes becomes available, this can be added to local software and uploaded to COSD in subsequent months. At the same time, NCRAS receive information from other data sources, including pathology and death data, culminating in a final registration record some 6 months after the initial registration. Thus, patients diagnosed towards the end of 2016 will not have their final registration completed until August 2017. At this point, NCRAS are able to link the final registration record with other sources of data such as Hospital Episode Statistics (HES), Systemic Anti-Cancer Therapy Dataset (SACT) and National Radiotherapy Dataset (RTDS), as well as determining whether the patient is still alive. This data has to be quality assured within NCRAS before it can be released to our analytical team at the University of Nottingham.

Following data cleaning to prepare the data for analysis, we conduct a large number of complex analyses on the data. We go through a process of reformatting this data in a way that can then be shared with local organisations so that they can "sense-check" it before publication, and it is not uncommon for important issues with data quality and accuracy to become apparent only at this late stage. The Annual report cannot be written until the data has been analysed and agreed - however, we try to "pre-write"" the report so that in we have a draft version ready very soon after sharing data with organisations.

Publishing a national annual report is not straightforward either, and we have a lengthy process to gain approval from HQIP, NHS England and the Welsh Government. Indeed we only gained agreement to publish our last report 5 days before the actual release date!

Although other national audits are able to analyse and report data back to organisations within a few weeks or months, the nature of the lung cancer diagnostic/treatment pathway, and the data we collect and report on means that there is inevitably quite a delay between diagnosis and reporting. Whilst reporting on accurate and complete data has to take priority, we will continue to do our best to report back as quickly as possible, and our new "mid-year" reports reflect our commitment to this.

Dr Paul Beckett is the senior clinical lead at the NLCA