Using RCRD and staging data

1. Measure the proportion of early-stage cancers as the number of persons diagnosed at stage 1 or 2 over all staged cancers.
2. The early-stage proportion measured within the Rapid Cancer Registration Data has a stable relationship to that in the National Cancer Registration Data, although there is an offset between them in the early-stage percentage measured.
3. Data should be aggregated over a long enough period to smooth out random noise, a 12-month period is recommended.
4. Comparisons within data sets (either within the NCRD or within the RCRD) are preferred to direct comparisons between the two.

NDRS collects, curates and quality-assures the National Cancer Registration Data (NCRD).  This data is the ‘gold standard’ for measuring the incidence and properties (including stage at diagnosis) of cancer in England.  In response to the COVID-19 pandemic and the desire for more rapid reporting, NDRS developed an algorithmically generated Rapid Cancer Registration Data set (RCRD) using the standard administrative data sets which flow most rapidly into NDRS. This represents the best available database of 'proxy cancer registrations' (including proxy staging information) for more recent periods (typically available from three to four months post-diagnosis) than are available from finalised NCRD data (which can take 21-24 months).

The NHS Long Term Plan (LTP) was published in January 2019.  One of the key ambitions is:

“By 2028, the NHS will diagnose 75% of cancers at stage 1 or 2.”

Progress against this target should be measured as the percentage of stageable cancers that are diagnosed at stage 1 or 2 (or equivalent), i.e.

This calculation can be completed with data from NCRD and RCRD, though there are differences in cancer types included. Stageable cancers within the NCRD are defined by NDRS in the publication Case-mix adjusted percentage of cancers diagnosed at stages 1 and 2 in England (this also covers which speciality staging systems are equivalent to stage 1 and 2). The RCRD aims to provide stage information for 13 cancer types. Information on these 13 cancer types combined should be used to measure progress against the 75% ambition.

Overall, there is a stable relationship between the early-stage percentage measured in RCRD and the early-stage percentage measured in NCRD, although NDRS continues to monitor this. The early-stage percentage measured in RCRD is typically offset by 1-2% above that in the NCRD but, on the whole, the RCRD early-stage percentage closely tracks the NCRD early-stage percentage over time.  The stability of the agreement between the two is better or worse for different cancers, and also varies by patient age-group, deprivation, ethnicity and for sub-national geographies. For a more detailed exploration of these points consult the technical documentation below. In general we recommend comparisons within datasets (either within the NCRD or within the RCRD) rather than direct comparisons between the two.

There is a choice to be made about the length of period over which to aggregate data when measuring change over time. Shorter periods introduce more random noise (due to averaging over smaller numbers of patients) and are vulnerable to biases including seasonality.  In particular, we know the most recent month’s data in the RCRD has a slight bias due to a lack of time for stage data to accumulate following clinical activity.  However, longer periods can be less timely, harder to associate with activity and average out trends over time.  Overall, we recommend using a 12-month aggregation (which can include the most recent available month of data) in the RCRD for this early-stage percentage data, rather than comparing data for individual months. The 12-month aggregated data can be considered on a rolling basis to mitigate timeliness concerns.

Across both sources, there is observed variation between Integrated Care Boards (ICB) / Cancer Alliances in the early-stage percentage. This may be due either to case-mix or to genuine variation, and to some extent is driven by variation in particular cancer types (prostate cancer in particular). Caution in interpreting the data is needed due to small number variation caused by lower counts available for local geographies compared to England as a whole.  This is especially the case where a geography’s early-stage percentage is being segmented by another factor, such as deprivation. NDRS continues to explore the use of RCRD data to measure early stage percentage at local area levels.