Case-mix adjusted percentage of cancers diagnosed at stages 1 and 2 in England, 2019

The case-mix adjusted indicator is currently in a testing phase and is not yet fully developed. It has not been submitted for assessment as a National Statistic to the UK Statistics Authority. We welcome any feedback from customers and stakeholders in the development of this indicator to ensure quality at an early phase in the development process.

ICD-10 revision 5 was introduced in 2013. This bulletin and accompanying data tables codes all cancers according to ICD-10 revision 5 for registrations and deaths between 2013 to 2019. All previous years are coded according to the original ICD-10 (see Impact paper on ICD-10 revision 5 for more detail).

In October 2018, the government set out a plan for 75% of all stageable cancers to be diagnosed as an early stage by 2028. The improvement in recording of stage continues to be part of the work programme for the National Disease Registration Service.

Stage is a measure of how much a cancer has grown and spread, with more advanced stages meaning the cancer is bigger or has spread to other parts of the body (lymphatic spread or metastasis). There are usually fewer treatment options at an advanced stage. National public health interventions, such as screening programmes and information and educational campaigns (for example, Be Clear on Cancer and Help Us, Help You), aim to increase the percentage of early diagnoses.

A cancer is considered stageable if a staging system exists for where the cancer originates in the body (the site) and the type of cells in the cancer. For example, a staging system doesn’t exist for the types of cancer found in the heart.

The staging system used here is typically TNM. The TNM system puts cancers in a group from 1 to 4 depending on the cancer, or tumour, size (T); which, if any, lymph nodes have cancer cells (N); and if the cancer has spread (metastasised) to other parts of the body (M). Different versions of TNM are used (versions 5 (for colorectal), 7 and 8), in line with international guidance. For some cancers, a site/group-specific staging system is used instead of TNM:

• International Federation of Gynecology and Obstetrics (FIGO) staging for gynaecological (ovary, cervical and uterus) cancers
• Ann Arbor staging for non-Hodgkin lymphomas
• International Staging System (ISS) for myelomas
• Binet staging for chronic lymphocytic leukaemia (CLL)
• Chang staging for Medulloblastoma
• International Neuroblastoma Risk Group staging system (INRGSS) for neuroblastoma
• National Wilms Tumour Study staging for Wilms tumours.

For these cancer sites/groups, TNM stage has been used where the site/group-specific stage was unknown. Cervical cancer is the exception, whereby a cancer is only considered staged if a FIGO staging value is available. These site/group-specific staging systems can be mapped to TNM as follows:

•     Stage 1 = TNM stage 1, FIGO stage 1, Ann Arbor stage 1, NWTS 1
•     Stage 2 = TNM stage 2, FIGO stage 2, Ann Arbor stage 2, NWTS 2
•     Stage 3 = TNM stage 3, FIGO stage 3, Ann Arbor stage 3, NWTS 3
•     Stage 4 = TNM stage 4, FIGO stage 4, Ann Arbor stage 4, NWTS 4-5

For this publication, cancers that have been staged in an alternate system (Binet; ISS; Chang; INRGSS) and cannot be mapped to TNM, have been included as two separate categories

• Staged – other early = Binet A-B; ISS 1-2; Chang M0-M2; INRGSS L1-L2
• Staged – other advanced = Binet C; ISS 3-4; Chang M3-M4; INRGSS M&MS

Both TNM staging system and site/group-specific staging systems can be mapped to early (referred to as stages 1 & 2) or advanced stage (referred to as stages 3 & 4) as follows:

• Stages 1 & 2 = Stage 1, Stage 2, Staged – other early
• Stages 3 & 4 = Stage 3, Stage 4, Staged – other advanced

In this publication the rare subset of breast cancers known as "Paget's disease of the nipple", which are classified as "stage group 0", have been included with non-malignant tumours where appropriate and do not appear in the malignant cancer statistics presented. There are less than 200 of these tumours diagnosed per year. This rule has also been applied to the small number of occurrences of other cancers with "stage group 0". This may cause a small discrepancy in counts and totals compared to other publications.

TNM was updated to version 8 and introduced for all cancers registered since January 2018, except head and neck where it was introduced for 2019 registrations onwards. This has an impact on the time series of certain cancer sites/groups, in that a noticeable change in the unadjusted percentage of cancers diagnosed at stages 1 and 2 can be seen from 2017 to 2018 and 2018 to 2019. The greatest change in the unadjusted percentage of cancers diagnosed at stages 1 and 2 between 2017 and 2018 was seen for thyroid cancer and between 2018 and 2019 for oropharynx cancer, with both experiencing a sharp increase.

A sensitivity analysis was performed to assess the impact of the introduction of the new TNM version on the case-mix adjusted estimates for the 3-year rolling case-mix adjusted average. The sensitivity analysis removed oropharynx as a cancer site/group and compared the estimates. Oropharynx was chosen because it was the cancer site/group with the greatest increase in the unadjusted percentage of cancers diagnosed at stages 1 and 2 from 2018 to 2019, the time when TNM version 8 was introduced for head and neck cancers. For England, the results were as follows:

• 2013 to 2015: 54.8% without oropharynx, 55.3% with oropharynx
• 2014 to 2016: 54.9% without oropharynx, 55.4% with oropharynx
• 2015 to 2017: 54.8% without oropharynx, 55.3% with oropharynx
• 2016 to 2018: 54.8% without oropharynx, 55.3% with oropharynx
• 2017 to 2019: 54.9% without oropharynx, 55.2% with oropharynx

The standard deviation for the differences between the 2 values for each CCG was 0.1% for all 5 time periods. Therefore, it was concluded that no methodological changes were required to account for the TNM version update. A similar sensitivity analysis was performed previously investigating the change from TNM7 to TNM8 in 2018. As stated in the Technical Documentation, this will be reviewed for the next release, when the inclusion of TNM version as a case-mix variable will be considered.

Important points for consideration:

• the case-mix adjusted ‘percentage of cancers diagnosed at stages 1 and 2’ indicator adjusts for the following patient-level characteristics: age, sex, cancer site/group and area-level deprivation
• the case-mix adjusted values of the ‘percentage of cancers diagnosed at stages 1 and 2’ are applied to the characteristics of the baseline population – this is to enable comparisons of changes to the indicator which are unrelated to changes to the underlying patient characteristics over time
• the smaller populations at a CCG level create wider confidence intervals. The reliability measure accompanying the output provides an indication of the robustness of the value – reliability above 0.7 is considered sufficient, with 0.9 considered the threshold for pay-for-performance measures

The cancer sites/groups identified for this indicator are stageable sites/groups that meet 2 criteria: at least 1,500 cancers diagnosed annually in England and an average of at least 70% of cancers having a complete stage over the past 3 years. Therefore, some cancer sites/groups are excluded from the indicator due to higher levels of missing stage at diagnosis. They were excluded on the basis that their inclusion would compromise the statistical robustness of the measure.

Due to the impact of the COVID-19 pandemic on the NDRS workforce and working arrangements, cancer registrations were delayed for the 2019 registrations, and there was a slight decrease in staging completeness. For 2019, completeness was below 65% for cervix, thyroid and stomach cancers, this must be taken into consideration when interpreting 1-year estimates. The sites contributing to the CMA indicator, taking into account staging completeness, will be reviewed for the next release.

The ‘percentage of cancers diagnosed at stages 1 and 2’ indicator was developed to inform policy and to monitor the quality and effectiveness of interventions aiming to increase diagnosis at an earlier stage. It also provides contextual information for the NDRS cancer survival rates, which are also published by stage.

Without case-mix adjustment, the unadjusted ‘percentage of cancers diagnosed at stages 1 and 2’ indicator for a CCG will be impacted if:

• cancers less likely to be diagnosed at stages 1 and 2 occur more frequently in the CCG than the national average, which will make the CCG’s measure look worse than it is
• cancers more likely to be diagnosed at stages 1 and 2 occur more frequently in the CCG than the national average, which will make the CCG’s measure look better than it is

For example, breast cancer is more frequently diagnosed at stages 1 and 2 than lung cancer. Therefore, without case-mix adjustment, CCGs with a higher than average occurrence of breast cancer will tend to perform better on the ‘percentage of cancers diagnosed at stages 1 and 2’ indicator, compared to CCGs with a higher than average occurrence of lung cancer.

Reliability gives an indication of how robust the performance measure is. The possible range lies between 0 and 1, with values closer to 1 indicating higher reliability. Low reliability indicates that variation by chance, due to small numbers, is having an unduly high influence on the performance measure. Lower reliability is likely to reflect smaller sample sizes, due to smaller populations within a CCG or use of annual rather than 3-year summary statistics.

The 75% ambition indicator measures the 1-year rolling percentage of all staged malignant cancers diagnosed, excluding non-melanoma skin cancer, that are recorded as presenting at stage 1 or 2, in England in 2013 to 2019. The indicator is not case-mix adjusted and is presented only for Cancer Alliances.

The two-sided proportions Z-test was used to test whether proportions were different from each other. A p-value of 0.05 was used to determine if the test was statistically significant.