CDF Methodologies

The National Institute of Clinical Excellence (NICE) reviews all new and promising cancer treatments which have received a marketing authorisation and makes recommendations on their use within the National Health Service (NHS) if the clinical and economic evidence, provided by the company, shows that the treatment is not only beneficial to patients but that the drug is cost effective. If there is any clinical uncertainty, and as such, the treatment is not recommended for routine commissioning, but there is enough evidence to say the treatment could be of benefit to patients, then the NICE appraisal committee can recommend a treatment for use in the Cancer Drugs Fund (CDF).

CDF funding allows for a period of manged access whilst additional data is collected to address areas of committee uncertainty. Data may be collected through ongoing clinical trials and/or through real world use of the treatment in the NHS. The Systemic Anti-Cancer Therapy (SACT) data set is used to collect data on real world treatment use.

This is a working document that details the steps taken to clean and extract data from SACT and applications captured on the NHS England’s Blueteq system. It details the methodologies applied to extract these data and analyse outcomes. SACT data are collected for all systemic anti-cancer therapies, regardless of the funding route for the drug, including treatments that are provided through baseline commissioning, the CDF and free of charge as part of an Early Access to Medicines Scheme (EAMS)¹.

NHS England’s Blueteq database captures the Cancer Drugs Fund population. The National Disease Registration Service (NDRS) has access to the Blueteq database for the Cancer Drugs Fund evaluation purposes.

The lawfulness of this processing is covered under Article 6(1)(e) of the United Kingdom (UK) General Data Protection Regulations (GDPR) (processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller). NHS England, through the National Disease Registration Service (NDRS), does have statutory authority to process confidential patient information (without prior patient consent) afforded through the National Disease Registries (NDRS) Directions 2021 issued to it by the Secretary of State for Health and Social Care, and has issued the NDRS Data Provision Notice under section 259 of the Health and Social Care Act 2012 regarding collection of the Blueteq data from NHS England.

This document will be updated as new methodologies are developed or current methodologies refined.

CDF applications - identification of the cohorts of interest

NHS England collects applications for CDF treatments through their online prior approval system (Blueteq®). The Blueteq application form captures essential baseline demographic and clinical characteristics of patients needed for CDF evaluation purposes. Where appropriate, Blueteq data are included in this report.

Consultants must complete a Blueteq application form for every patient receiving a CDF funded treatment. As part of the application form, consultants must confirm that a patient satisfies all clinical eligibility criteria to commence treatment.

NDRS has access to the Blueteq database and key data items such as NHS number, primary diagnosis, and drug information of all patients with an approved CDF application which met the treatment eligibility criteria.

CDF applications - de-duplication criteria

Before conducting any analysis on CDF treatment data, the CDF database is examined to identify duplicate applications. The following de-duplication rules are applied.

1. If two trusts submit an application for a CDF treatment, for the same patient (identified using the patient’s NHS number), and both applications have the same approval date, then the record where the CDF trust (the trust applying for CDF treatment) matches the SACT treating trust is selected.
2. If two trusts submit an application for a CDF treatment, for the same patient, and the application dates are different, then the record where the approval date in the CDF is closest to the regimen start date in SACT is selected, even if the CDF trust (the trust applying for CDF treatment) didn’t match the SACT treating trust.
3. If two applications are submitted for a CDF treatment and the patient has no regimen start date in SACT capturing when the specific drug was delivered, then the earliest application in the CDF is selected.

Analysis is limited to CDF applications made from the date when the drug entered the CDF. Any treatments delivered before the CDF entry date are excluded as they are likely to be patients receiving treatment through an Early Access to Medicines (EAMS) programme or a compassionate access scheme run by the pharmaceutical company.

Figure 1. Process to identify the CDF cohort of interest

The below image shows the process to identify the CDF cohort of interest, from the initial CDF applications made for a specific CDF drug, from the CDF entry date to the data collection end date. Please refer to the text description for a full summary of the content of this image. 

SACT is the primary data source when there is no clinical trial, in this situation SACT will be used to answer areas of clinical uncertainty that have been raised at the NICE committee. Where SACT is the primary data source a 4-month reporting period is allowed from the end of the data collection for treatment information to be collected and reported to the SACT data set. This reporting window ensures the data is as correct and complete as possible. Trusts submit records to the SACT data set 12 weeks after treatment activity. During the intervening time, trusts upload their initial SACT data for the month to the submission portal, correct all local and critical errors (errors arise if a submission fails to pass validation on the upload portal), map all regimens to national standards and finally submit the data. An additional month is allowed for the SACT submission to pass through the data collection system at NDRS before data is available for analysis.

SACT is the secondary source when there is an ongoing clinical trial that will act as the primary data source and answer the clinical uncertainty raised at the NICE committee. Where SACT is the secondary data source SACT follow-up stops 4 months before the end of data collection period to allow more rapid delivery of the SACT report after the formal end of data collection.

The SACT snapshot is taken on the first Saturday of every month and made available to analysts, normally by the middle of the month. At the time of reporting, all patients are sent for tracing to obtain their vital status using the Personal Demographics Service (PDS)².

NHS numbers are used to link SACT records to data in NHS England’s Blueteq system for specific drugs in the CDF identified in Figure 1. Information on treatments in SACT are examined to ensure the correct SACT treatment records are matched to the CDF application, this includes information on treatment dates (regimen, cycle, and administration dates) and primary diagnosis codes in SACT.

Data completeness of the SACT data set has improved since its implementation in 2012 with all trusts now submitting their systemic anti-cancer therapies. This improvement is a result of the wider use of e-prescribing systems and the recruitment of several SACT data liaison officers at NDRS. NDRS also has a dedicated SACT Helpdesk that supports trusts in their SACT data submissions.

NDRS have produced several routine outputs to demonstrate data completeness of SACT data for the submitting trusts. These reports help trusts to review their performance and highlight data gaps. These reports can be found on the Cancerstats2 website³. This website can only be accessed behind the N3 (NHS) network

For CDF indications, NDRS have set up an additional process in which the SACT data liaison officers follow-up with trusts if any of their CDF treatment activity is missing. As such, the quality of SACT data for CDF applications is higher than general SACT submissions. An overview of the process for patient follow up is provided below (Figure 2 and Table 1). 

Figure 2. CDF follow-up review

The below image shows the process for CDF follow-up review. Please refer to the text description for a full summary of the content of this image. 

Analysts at NHS England generate an updated Blueteq reference list on a monthly basis, for the duration of data collection. This list contains patient NHS numbers, date of birth, diagnosis, CDF treatment, requesting consultant, and the Blueteq approval date. The Blueteq reference list of patients is uploaded as a data table into the Cancer Analytical System (CAS) which is used to store and analyse all SACT data.

SACT data uploaded by trusts becomes available to analysts through CAS on the second week in the month. At this stage patient identifiers from the SACT data set are cross-referenced to the Blueteq reference list and CDF patients flagged as ‘present’, ‘missing’ or ‘incomplete’ based on the following criteria:

Table 1: Criteria to evaluate status of CDF patients in SACT

Patients identified as incomplete or missing are flagged for follow up and allocated to the appropriate member of the data liaison team, based on geographic coverage.

Treatment duration

Treatment duration is one of the main analyses provided in CDF annual and final reports.

Start date is defined as the date the patient started their CDF treatment. This date is identified in the SACT data set as the patient’s earliest treatment date for the treatment of interest. Data items used to determine a patient’s earliest treatment date are:

1. Start date of regimen – SACT data item #22.
2. Start date of cycle – SACT data item #27.
3. Administration date – SACT data item #34.

The earliest of these dates is used as the treatment start date.

The same SACT data items (#22, #27, #34) are used to identify a patient’s final treatment date. The latest of these 3 dates is used as the patient’s final treatment date.

Additional explanation of these dates is provided below:

Start date of regimen

A regimen defines the drugs used, their dosage and frequency of treatment. A regimen may contain many cycles. This date is generally only used if cycle or administration dates are missing.

Start date of cycle

A cycle is a period of time over which treatment is delivered. A cycle may contain several administrations of treatment, after each treatment administration, there will be a delay before the next treatment administration. For example, a patient may be on a 3-weekly cycle with treatment being administered on the 1st and 8th day, but nothing on days 2 to 7 and days 9 to 20. The patient’s next cycle would start on the 21st day.

Administration date

An administration is the date a patient receives treatment, using the above as an example the administrations for a 3 weekly cycle would be on the first and eighth day and then again on the 21st day, which would be the first day of their next cycle.

The interval between treatment start date and final treatment date is the patient’s time on treatment.

All patients are then allocated a ‘prescription length’, which is a set number of days added to the final treatment date to allow for the fact that they are effectively still ‘on treatment’ between administrations. The prescription length should correspond to the typical interval between treatment administrations.

If a patient dies between administrations, then their censor date is their date of death, and these patients are deemed to have died on treatment unless an outcome summary is submitted to the SACT database confirming that the patient ended treatment due to disease progression or toxicity before death.

Adding a number of days to a patient’s final treatment date ensures we do not underestimate treatment duration or the number of deaths on treatment. See Appendix A and B for the number of days added for each drug. The number of days detailed in appendix A and B are based on the administration schedule recommended by the drug manufacturer and cross referenced with the observed cycles or administrations captured in SACT. The appendices will be updated as new drugs enter the CDF.

Treatment duration calculator

Treatment duration is calculated for each patient as:

Treatment duration (days) = (Final treatment date – Treatment start date) + prescription length (days)

This date would be the patient’s censored date, unless a patient dies in between their last treatment and the prescription length added, in this case, the censored date would be the patients date of death.

Once a patient’s treatment duration has been calculated, the patient’s treatment status is identified as one of the following:

No longer receiving treatment (event), if:

• the patient has died.
• the outcome summary, detailing the reason for stopping treatment has been completed:
  • SACT v2.0 data item #41
  • SACT v3.0 data item #58 - #60.
• IV - there is no further SACT records for the patient following a three-month period.
• Oral - there is no further SACT records for the patient following the period of the treatment end indicator (Appendix B)

If none of the above apply, the patient is assumed to still be on treatment and is censored.

Overall survival

Overall survival (OS) is calculated from the CDF treatment start date, not the date of a patient’s cancer diagnosis. Survival from the treatment start date is calculated using the patient’s earliest treatment date, as described above, and the patient’s date of death or the date the patient was traced for their vital status.

All patients in the cohort of interest are submitted to the PDS to check their vital status (dead or alive). Patients are traced before any analysis takes place. The date of tracing is used as the date of follow-up (censoring) for patients who have not died.

OS is calculated for each patient as the interval between the earliest treatment date where a specific drug was given to the date of death or date of follow-up (censoring):

OS (days) = Date of death (or follow up) - treatment start date

The patient is flagged as either:

This document represents standard methodologies used by the NDRS analytical team for the analysis of CDF treatments. Bespoke methodologies will be developed as required to support treatment evaluation and will be documented as appropriate. This document will be reviewed and updated as methodologies are developed.

References

1. Early access to medicines scheme. [Internet]. GOV.UK: 2020 [cited 2023 December]. 
2. The Personal Demographics Service (PDS). [Internet]. NHS Digital: 2023 [cited 2023 December].
3. CancerStats2 data platform* (opens in a new window). NHSE: 2023 [cited 2023 December].

*Please note that this platform requires an N3/HSCN secure network connection.