Survival outcomes and HES-based Charlson Comorbidity Scores for women recruited to the Bridging the Age Gap in Breast Cancer study

Background

This work is part of an on-going study - The Bridging the Age Gap in Breast cancer study. This is a multicentre cohort study collecting direct prospective data on older women with breast cancer from multiple UK sites.

The Age Gap study, which started in January 2013, is using statistical and modelling techniques to determine the age, comorbidity, frailty and disease characteristics of women over 70 with early breast cancer to provide guidance on 2 primary questions:

1. What are the personal and cancer characteristics of women who can be safely advised that surgery is unlikely to confer any advantage for them?

2. What are the personal and cancer characteristics of women who should be advised to have adjuvant chemotherapy after surgery?

A preliminary disease and outcome statistical model has been derived using pre-existing data from the UK primary breast cancer registry held by the West Midlands Cancer Intelligence Unit (WMCIU) and the NHS Hospital Episode Statistics data (available to Northern and Yorkshire Cancer Intelligence Service, not within this agreement). The statistical model is being used to develop a web-based algorithm to support clinicians in decision making related to older women. However, the cancer registry data have a number of limitations relating to the completeness and quality of comorbidity data. In addition, staging and co-morbidity data may be less accurate in women treated non-surgically as there will be no post-operative pathology data returns.

To overcome these limitations, a UK-wide data collection exercise to gather detailed data on older women, their primary disease, health status and treatment details and medium term outcomes was performed. These data are more complete and contain additional information on co-morbidities which is particularly important for informing decisions in this older age group. As part of the project the intention is to supplement the initial 2 year direct follow up (via direct data collection for the study) by longer term follow-up via cancer registry returns, for up to 10 years of follow up.

The study has full ethics approval and was awarded NIGB approval at inception. All patients have consented to take part in the study & were also asked to consent specifically to permit the study group to seek cancer registry return data from them in the longer term. Consent was approved by both the Ethics Committee and the NIGB.

The recruitment period for the 3,400 women was over 5 years, starting in 2013 and completed in Spring 2018. Data on cancer type, health & fitness, treatment & outcomes are collected for 2 years after diagnosis. The earliest tranche of recruits are therefore not under direct follow up any more.

The aims of this study are as follows:

Objective 1 : Updated survival outcomes

To update the outcomes of women with invasive breast cancer (c50) recruited to the Bridging the Age Gap cohort study at three time points.

The researchers wish to update the outcomes of these women at the following three time points:

(a) study recruitment closure (Spring 2018),

(b) median 5 year follow up (Spring 2020) and

(c) median 10 year follow up (Spring 2025)

These time points reflect the 5/10 year follow up from the median time of recruitment for the cohort - participants have been recruited from 2013 to 2018.

These longer term follow up data will be used to revise and validate the preliminary statistical model. The updated statistical model will be used within the web-based algorithm to support clinicians in decision making related to older women. The updated statistical models will also be used to develop a health economic model to estimate long-term health outcomes and costs for different intervention strategies for older women.

At all three time points, the University would like to be able to derive Overall Survival (OS) and Breast Cancer Specific Survival (BCSS) and therefore want to receive vital status, date and cause of death .

The researchers are also interested in obtaining data on recurrence, but are aware that, currently, the quality of such data is poor and will therefore not be requesting this for the first timepoint. However, at the two later time points, assuming the quality of the data has improved, researchers may also be interested in obtaining breast cancer recurrence data. This would be subject to a further application to NHS Digital.

Objective 2 : To obtain the HES-derived proxy Charlson comorbidity Index (CCI) on patients recruited to the study

As part of the Cohort Study the University have also collected detailed comorbidity data using a direct data collection of the Charlson comorbidity Index (CCI) (the Charlson Comorbidity Index predicts the ten-year mortality for a patient who may have a range of comorbid conditions).

The study team would like to obtain the HES-derived CCI on all patients in the cohort to validate the accuracy of the HES –derived CCI against the CCI calculated directly from the cohort data. This is a uniquely valuable opportunity to undertake such a validation. Please note, HES data was disseminated under version 0.18, no HES data will be disseminated under this version of the application.

The GDPR lawful basis for the University of Sheffield to process this data is Article 6(1)(e) 'task in the public interest' and Article 9(2)(j) 'scientific research' as the project aims to improve treatment decision making for older women with breast cancer.

The University of Sheffield have previously supplied the details of patients in the Bridging the Age gap cohort study, who have consented to sharing their data with the study, to NHS Digital. This included all patients who have consented to cancer registry access (around 92% of recruited women) and who have not voluntarily discontinued the study and requested their data to be deleted or withdrawn their consent (approx. 3 or 4 patients).

Data already sent to NHS Digital:

For women who have consented to registry access the University provided NHS Digital with information consisting of NHS number, postcode, date of birth and sex in order for NHS Digital to identify patients from the Bridging the Age Cohort.

Data previously requested from NHS Digital :

(1) Survival follow-up cohort

For the survival follow-up subgroup (a subgroup of the whole cohort (those diagnosed prior to June 2015) who have already completed their two year follow up within the Bridging the Age Gap study) the University would like to request the following information from NHS Digital :

(i) Vital status,

(ii) Date of death

(iii) Cause of death (all fields)

(2) Whole cohort:

The University will calculate the Charlson comorbidity Index (CCI) from the linked HES and mortality data at the time of patient baseline assessment.

Note: The Charlson Index (Charlson et al 1987) includes the following co-morbidities: myocardial infarction, congestive cardiac insufficiency, peripheral vascular disease, dementia, cerebrovascular disease, chronic pulmonary disease, diabetes, liver disease, kidney disease, and other cancers.

The CCI Score (excluding the cancer component) is calculated for each patient using the relevant diagnostic codes recorded for any inpatient or day case hospital admission in the 18 months before diagnosis of their breast cancer.

The cancer component of the Charlson Comorbidity Index is derived from the cancer registry data, in a method consistent with other similar registry data analyses (Lavelle et al 2012, Morris et al 2011).

This agreement will follow Step 1 only, as no further HES data has been requested.

The data will not be made available to third parties. All organisations party to this agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by “Personnel” (as defined within the Data Sharing Framework Contract ie: employees, agents and contractors of the Data Recipient who may have access to that data).

Use of the NHS Digital-supplied data

(1) Study of long term outcomes of women in cohort study receiving current standard care for early breast cancer

As part of the project a preliminary disease and outcome statistical model has been derived using pre-existing data from the UK primary breast cancer registry held by the West Midlands Cancer Intelligence Unit (WMCIU) and the NHS Hospital Episode Statistics data.

The requested data will supplement the initial 2 year direct follow up via direct data collection for the study by providing longer term follow-up of outcomes for patients in the cohort study

The initial statistical model (using merged retrospective data from the WMCIU and HES) will be revised and refined using data from the cohort study and the updated survival outcomes. These data will be used to test and revise the initial statistical model.

Initially the study team will update the statistical model using the (up to three years’ worth of) longitudinal data from the cohort study alone as the data source. This will enable researchers to examine whether evidence is emerging of differences in model coefficients. Researchers will undertake comparison using the original statistical models to predict probabilities of early outcomes for the cohort and examine them against observed outcomes. Finally, researchers will seek to develop a new set of statistical models for time to breast cancer and non-breast cancer death linking the longer-term data from the cancer registries with the short to medium term outcomes data from the prospective cohort study.

Two further follow-up data (at median follow-up of 5 years and 10 years) will allow the models to be revisited and improved over time, based on the long term data for the cohort patients.

(2) Comparison of HES-derived proxy CCI compared with CCI derived from cohort study data

The Cohort Study has collected detailed co-morbidity data using a direct data collection of the Charlson co-morbidity Index (CCI). By requesting the HES derived CCI on all patients in the cohort researchers will be able to validate the accuracy of the registry CCI.

Outputs will contain only data that is aggregated (with small numbers suppressed in line with the HES Analysis Guide).

The study has an open access website which gives updates on the status of the study on a regular basis.

The study has worked with a team of patient representatives, and with Age UK and Breakthrough Breast Cancer, during the creation of the study (in line with NIHR recommendation), and will continue to do so in order to disseminate the outputs. The outputs will be shared widely, including with the West Midlands Cancer Intelligence Unit, the Northern and Yorkshire Cancer Intelligence Service, and through members of the MRC Cognitive Function and Aging (CFAS) project.

Specific outputs will include:

(a) a web-based decision support tool (to be made available nationally, to UK breast units).

(b) a research report for NIHR

(c) publications in medical journals

It is anticipated that publications will follow the research team’s track record of publishing in medical journals (BMJ, Lancet), specialist breast cancer journals (e.g. British Journal of Cancer, European Journal of Cancer) , and health economics journals (e.g.Value in Health ). Initial publications to be submitted Summer 2020.

The end product will be a revised statistical model incorporating retrospective and prospective collected data. Record level information will not be included within the model. The revised model will be completed Summer 2020. The model will be made available to support clinicians in decision making related to older women. This will be made freely available to clinicians in the NHS. The statistical model will be further updated when 5 and 10 year median follow- up data is made available.

In addition to the development of a statistical model that can be used to predict individual outcomes, the University are proposing to construct an economic model to inform National Guidelines on treatment choice for older women with breast cancer. The economic model is likely to take an individual level modelling approach, incorporating outputs from the updated statistical model (using both retrospective and prospective data) that allows the impact of different strategies for treatment selection to be compared. The model will be completed by end 2020.

The study itself has already led to the publication of a number of papers including:

~ Quality of life versus length of life considerations in cancer patients: A systematic literature review.

~ Adjuvant Chemotherapy for Breast Cancer in Older Women: An Analysis of Retrospective English Cancer Registration Data.

~ Bridging the age gap in breast cancer: evaluation of decision support interventions for older women with operable breast cancer: protocol for a cluster randomised controlled trial.

~ What influences healthcare professionals' treatment preferences for older women with operable breast cancer? An application of the discrete choice experiment.

The broad aim of the project is to improve treatment decision making for older women with breast cancer.

Objective 1

An update of the survival data will be required in February/March 2020. The updated survival status will provide updated evidence for the statistical model used within the decision support tool. By this timepoint, there will be 5 years average follow up and all patients will have a least 2 years followup.

This will be used to provide improved evidence on optimum treatment for older women with breast cancer, because to date much of the evidence has focused on younger women. Within this older age group it is particularly important to predict cancer outcomes and how these interact with frailty, age and co-morbidity.

At present older women with breast cancer are treated less aggressively than younger women. Some of this variance to normal practice is fully justified. It is known that a very frail elderly lady, with a short predicted life expectancy, will struggle to come through surgery without significant side effects and that her cancer may well be controlled for the rest of her natural life by anti-oestrogen tablets. However, at present there are no guidelines for just how elderly and how unfit a woman needs to be before surgery is of no benefit. At present, rates of non-surgical therapy are highly variable across the UK with a 4 fold variation in rates of non-surgical treatment between health regions.

This means that in some areas fitter, older women are denied surgery and placed at risk of later local disease progression, whereas in other areas, where surgery is the norm, many frail older women are subjected to surgery when it will have no benefit for them and may cause them significant harm.

Similarly, rates of chemotherapy use in older women are very low relative to those in younger women despite the fact that there is evidence that fitter younger women in the over 70 age group may benefit. The breast clinicians have little or no evidence to help them to decide what the cut offs for these treatments may be.

This work will benefit the health and social care systems by providing a freely-available web based tool to support clinician and patient decision-making for older women with breast cancer. The tool will help ensure that patients receive the most appropriate treatment and aims to improve outcomes for older women - in terms of survival and/or quality of life.

Objective 2 HES data

This data was requested previously. An update is NOT required.

The validation of the HES-derived CCI will provide valuable information to the health services community on the strengths and limitations of using the HES-derived CCI. This results of this validation will be published in a peer-reviewed medical journal.

Analysis of the disseminated data is now underway and key papers and results will be made available later in 2020.