Long-term follow-up and further analyses of the National Chronic Kidney Disease Audit

Chronic kidney disease (CKD) is estimated to affect 1 in 10 people of the adult population in the UK. Although it is known that the bulk of patients with CKD are managed by primary care, it is not known who has progressive kidney disease, whether better management could improve outcomes other than developing kidney failure and how care (including the interphase between primary and secondary specialist care) could be improved further.

London School of Hygiene and Tropical Medicine (LSHTM) and University College London (UCL) seeks to link data from the National CKD Audit to Hospital Episodes Statistics and Mortality data from the Civil Registrations dataset to create the National CKD Audit Research database. For both organisations, the legal basis for processing the NHS England (previously known as NHS Digital) data under UK GDPR is Article 6 (1)(e) - "processing is necessary for the performance of a task in the public interest or in the exercise of official authority vested in the controller" and Article 9 (2)(j) - "processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes".

History of the National CKD Audit giving rise to the current research data application to NHS England:

The National Chronic Kidney Disease Audit (NCKDA) took place in England and Wales in 2014-2016. It was tendered by the Health Quality Improvement Partnership (HQIP) to a commercial provider (Informatica systems). Informatica systems extracted the primary care data from consenting GP practices and provided pseudonymised data to LSHTM for analysis.

The NCKDA aimed to provide a comprehensive picture of management and outcomes for people with chronic kidney disease (CKD) stages 3-5 managed in primary care in the region.

The following key questions were addressed based on the NICE CKD guidance:

1. Are patients with risk factors tested for CKD?

2. Are people who have repeated abnormal kidney function values appropriately coded?

3. Measure the number of patients assessed for cardiovascular risk and the numbers receiving cholesterol lowering treatment

4. What percentage of people with CKD received other NICE-recommended key aspects of CKD management applicable in primary care?

5. For people with CKD, what are the rates of acute complications (hospitalisation with acute kidney injury, hospitalisation with cardiovascular events/interventions, and death)?

Audit data were collected from primary care up to June 2016, and subsequently linked to HES/PEDW/ONS data in September 2016 for audit reporting purposes. Data analyses using the linked Audit data were carried out at LSHTM. The most recent Audit report (published in Dec 2017) reported on short term outcomes. This showed that people who have CKD based on their biochemical data but who have no corresponding diagnosis in their GP record die more often than those who are coded for their disease, and very high burden of hospitalisations in this this population, again more often if they were not formally identified by a diagnostic GP code in the health record.

However, there is a substantive proportion of patients who were only diagnosed with CKD in 2016 for whom there are no meaningful data on outcomes (i.e. in whom audit question 5 in particular was not addressed, this includes hospitalisation with acute kidney injury, [short term outcomes] hospitalisation with cardiovascular events/interventions, and death [long term outcomes] ). Therefore, there are a number of patients for whom there is not sufficient follow-up time to assess their outcomes, and some outcomes, such as cardiovascular outcomes, kidney failure and death require longer periods of follow-up to assess impact of care. In addition, understanding interfaces between primary and secondary care, i.e. supply/care pathways requires further data validation by linkage to the data held at the UK Renal Registry.

Such validation is necessary for understanding features of renal service delivery; originally this was planned for years 4 and 5 of the audit, but NHS England had decided in 2016 to not continue funding the Audit. The linked Audit outcome (hospitalisation and death) data for which HQIP was not the data controller were deleted at the contract end in December 2017.

After securing HRA approval for the establishment of this NCKDA research database, and section 251 approval for linking NCKDA data for research purposes without consent, prior to the end of the Audit in December 2017, HQIP signed a data-sharing agreement with UCL and LSHTM, for UCL and LSHTM to retain a copy of the historical Audit data for future research, and for set up of the National CKD Audit research database.

The Data Sharing Agreement with NHS England is led by LSHTM, who was the data processor for the Audit database, in collaboration with UCL, and wishes to address the short falls of the terminated Audit that were listed above by establishing the National CKD Audit Research database. The team at UCL agreed to house the identifiers required for future data linkage, whilst LSHTM holds the pseudonymised clinical data.

The audit website contains information on how to opt out from the audit, and there has been no individual contact to the study team asking for individual opt out from the study. After the data sharing agreement with HQIP was signed, a copy of the identifiable data held at Informatica systems (NHS number, study pseudo-identifier) was moved to the secure server at UCL and the original data at Informatica systems were deleted.

In brief, the plan is to establish a research database by retaining the existing NCKDA data (the entire database derived from primary care records), and to augment these data by carrying out linkages with other databases (HES, Civil Registrations (deaths) data, PEDW, UK Renal Registry, and datasets held at NICOR) under the research approvals that were obtained.

Research questions that the research data base will address are as follows:

1. What are the long-term outcomes of patients at risk of and with CKD and which aspects of primary care management influences outcomes?

This study question relates to the original question posed by the audit data collection which HQIP could not answer with the previous Audit-linkage as at the time of the first linkage there was too short follow-up for a considerable subset of study participants. Outcomes include hospitalisations, cause-specific hospitalisations (e.g. angina, stroke, heart failure, infection, acute kidney injury etc), progressive renal disease, requirement of acute and chronic dialysis (both planned and unplanned), and deaths. All of these outcomes (apart from acute kidney injury) are long-term outcomes of care.

As outlined above, the preliminary audit analyses are suggestive that death rates may be higher in those people with kidney disease (based on their kidney function test results) who are not formally recognised to have kidney disease based on the coded disease list in the GP health record. Similar observations were made for hospitalisation rates, though there may be competing risk by mortality.

In order to better understand how GP care of people with kidney disease impacts on their health outcomes, the research team therefore require information on mortality to deal with competing risk of death, and also cause-specific mortality to capture renal and cardiovascular deaths. It is here that data from cardiac registries held at NICOR will be very useful as some of these outcomes are only partially captured by HES.

2. How and where are patients with CKD managed? - Investigating the patient journey from primary care to dialysis.

This question is crucial to address as currently there is little knowledge on where the best (i.e. most complete) clinical data on patients with more advanced kidney disease are held as these patients are often seen by specialists in secondary care, thus primary care data may be incomplete. To understand which out-patient specialist sees the patient, HES outpatient records are required for those with kidney disease. The ethics permissions for this particular study question do not cover those at risk of kidney disease and hence outpatient data are only requested for the subset of people with known kidney problems.

3. What is the burden of progressive kidney disease and what are its consequences?

The audit has shown that testing of kidney disease varies by underlying risk profile, therefore to date there is no official estimate on the numbers affected by progressive kidney disease. The audit database can be used to derive an estimate of the burden of progressive kidney disease, and if linked to long-term outcome data (cardiovascular hospitalisations, death, dialysis) would inform on cardiac and renal outcomes of patients with progressive kidney disease, and also mortality (most patients with CKD die before they ever reach dialysis)

4. To which extent does acute kidney injury contribute to progressive kidney disease and incidence of acute and chronic dialysis?

This question cannot be simply addressed with linking hospital admission data to the UK Renal Registry as there is significant concern that the awareness for acute kidney injury (AKI) amongst hospital doctors may depend on the patient's underlying CKD status, thus introducing differential misclassification. The data from the AKI Think Kidneys programme (held at the UK Renal Registry) may address this question with gold standard data. In addition, mortality data are needed as patients with severe AKI may have died before having been identified as a chronic dialysis patient.

5. What are the disability adjusted life years, years of life lost, healthy life expectancy and cost associated with CKD, AKI and progressive renal disease?

There are no precise UK estimates for any of these important figures. The data from the above research studies can contribute to a more precise estimation of the impact of CKD on health, and when combined with estimates from existing relevant randomised trials simulations could be run to investigate which known evidence based interventions would be most cost-effective to reduce the impact on patients' lives due to CKD, as well as costs to the community. Mortality data are needed to calculate life expectancies.

6. Data validation to enable analyses of single existing datasets (e.g. primary care, or HES data) so that these do not require future linkages

Questions include: How good are primary care codes of transplantation and dialysis?, How well is dialysis/transplantation captured in HES compared to the UK renal registry?, Can data from HES inform on acute dialysis?, Is there differential misclassification of AKI coding in hospital records by CKD status (see above)? The referral date held in the UK Renal Registry database could be used to validate primary care referral entries. How much extra information on renal disease progression is held in renal clinics for those with CKD stages 4 & 5? For this type of analysis HES outpatient data are needed, as well as data on admitted care and mortality data (as other data sources may only collect reliable information on survivors and this bias is important to detect)

In summary, hospital data from NHS England are key to addressing the questions listed above, informing on specialist and critical care and outcomes of people at risk of and with CKD in primary care. In addition there is the need of accurate death data to understand what people with CKD die of, whether there is a difference due to coding of kidney disease, and the role of competing mortality, duration of follow-up and causes of death.

Data will be accessed by students under the supervision of the London School of Hygiene and Tropical Medicine.

The data will not be used for commercial purposes and not used for direct marketing. Any data access for people wanting to use the research database is reviewed by the steering committee.

Any research outside of the currently 6 specified research questions listed above will have to undergo new ethics and new data-sharing approvals by all data controllers who provided original data.

Funding source:

The previous iterations of the DSA were funded by HQIP/National CKD Audit. Additional funding has been secured for pilot analyses of these linked data when these become available from Kidney Research UK. However, none of these organisations will have any influence on the research findings, and they play no role in defining the means or purpose for which the data will be used. This extension DSA is funded by the Health Foundation.

This is an extension only application to retain previously disseminated data to continue to hold and process this data to complete analysis. No further data will flow under this version of the agreement.

An explanation of the role of each organisation involved in the set-up of the NCKDA is as follows:

1. UCL previously provided list of NHS numbers with NCKDA pseudo-id (study ID) to NHS England in the previous iteration of this agreement. These were the only identifiers that flowed to NHS England.

2. NHS England linked data for HES & Civil Registration Mortality data, and returned clinical and mortality data to LSHTM with NCKDA pseudo-id (study ID), with the NHS numbers removed.

3. UCL provided a list of NHS numbers with NCKDA pseudo-id to NWIS

4. NWIS linked data for PEDW and returned clinical data to LSHTM with NCKDA pseudo-id, with the NHS numbers removed

5. UCL provided a list of NHS numbers with NCKDA pseudo-id to NICOR

6. UK Renal Registry provided a list of NHS numbers with Renal registry pseudo-id to NICOR

7. NICOR generated Master-patient index of NHS number to match with NICOR, the UK Renal Registry, and the NCKDA.

8. NICOR returned Renal registry pseudo-id and NCKDA pseudo-id to the UK renal registry

9. UK Renal Registry sent the clinical data with NCKDA pseudo-id to LSHTM.

10. NICOR sent the clinical data with NCKDA pseudo-id to LSHTM

- University College London (UCL) - UCL requested in a previous iteration of the agreement, to link data from the National Chronic Kidney Disease (CKD) Audit to Hospital Episodes Statistics and Mortality data to create the National CKD Audit Research database. UCL then worked with the London School of Hygiene and Tropical Medicine to determine the nature and processing of the data and to process the data.

UCL provided NHS England with a list of NHS numbers to enable linkage with third parties (The National Institute for Cardiovascular Outcomes Research/NICOR and the UK Renal Registry). They are, therefore, a data controller who processed the data.

In relation to the identifiers, (NHS numbers, NCKDA pseudo-id) have already flowed from Informatica systems to UCL secure data haven.

- London School of Hygiene and Tropical Medicine (LSHTM) - the LSHTM is the other organisation working with UCL to link data from the National Chronic Kidney Disease (CKD) Audit to Hospital Episodes Statistics and Mortality data to create the National CKD Audit Research database. They are working with UCL to determining the nature and processing of the NHS England data, and will process the data. They are, therefore, a data controller who will process data.

- Healthcare Quality Improvement Partnership (HQIP): HQIP were the data controller for the original primary care data extraction of the NCKDA, and tendered the Audit on behalf of NHS England and Wales. They have given permission for the data to be used by UCL and LSHTM (see SD11, SD11.1), but play no role in defining the means or purpose for which the data will be used, and therefore play no role in this application.

- NICOR - the NCKDA research database will also contain data from NICOR, allowing the individual validation of key data items in HES for the population in the Audit (which will require linkage to the HES data). This is one of the stated aims of the establishment of the data base. However, NICOR play no role in determining the nature of the research or processing, and will not have access to the NHS England data.

- UK Renal Registry (UKRR)- the NCKDA research database will eventually contain data from the UKRR, allowing the individual validation of key data items in HES for the population in the Audit (which will require linkage to the HES data). This is one of the stated aims of the establishment of the data base. However, NICOR play no role in determining the nature of the research or processing, and will not have access to the NHS England data.

- Informatica Systems: they were original Audit provider for the NCKDA under the tender agreement with HQIP. Prior to the termination of the Audit, they transferred a copy of the identifiable data to UCL and subsequently deleted all Audit data on their systems. Since then they have had no further involvement, and have no role in defining the means or purpose for which the data will be used for this application.

- Queen Mary, University of London (QMUL) - researchers at QMUL previously contributed to the writing of two national Audit reports from the NCKDA, but have had no further involvement, and have no role in defining the means or purpose for which the data will be used in this application.

Data flows for this current agreement for linkages carried out by NHS England:

All data flows have been reviewed by ethics and CAG and have section 251 approval for linkage without individual consent.

1. Any identifiable data flows only involve transfer of the NHS number and a pseudo-identifier (study ID) to allow data linkage.

2. Data flows to LSHTM involve pseudonymised linked data.

1. The Patient Identifiable data:

Data flows to NHS England include the identifiable data held at UCL (NHS number) with the pseudo-id for the clinical data held at LSHTM. UCL will supply data from the entire Audit dataset irrespective of GP location in order to track patients who have moved across national boundaries. Cohort details will be transferred as a password-protected and encrypted CSV file to NHS England. Details of patients who have chosen to opt out during 2014-16 will not be sent to NHS England as these were never extracted. Data extraction stopped in 2016, and no new data were extracted since then. There has been no contact asking for individual opt outs from 2014 up to now.

NHS England will check whether there are any national opt-outs that UCL were not made aware of and quantify the total number for the study. LSHTM anticipate that this number will be very small, as all research opt-outs were removed from the outset prior to Audit data extraction.

Data requested from NHS England include hospitalisation, outpatient and critical care and death data. In order to supplement the data from primary care in terms of comorbidities (allowing the control for confounding for the objectives outlined above) LSHTM require data on past hospital admissions and care received in the patient population. LSHTM require data from Audit participants who are located either in England or Wales, as some patients from Wales may have had acute care in England (and vice versa).

2. The Pseudonymised data:

NHS England will supply HES in and outpatient records, critical care records (for England) and mortality data from the Civil Registration dataset (for England and Wales) for the cohort provided for the agreed period to LSHTM with the pseudo-id to be incorporated in the existing database held at LSHTM which only uses pseudonymised data. NHS England will not supply records for those who have opted out from research.

Further linkages:

A similar process will take place to obtain hospitalisation and outpatient data for people seen by health services in Wales, with identifiable data flowing from UCL to NWIS, and NWIS supplying pseudonymised hospital outpatient and admission data to LSHTM. Therefore, the NCKDA research database will have hospital data and death data for England and Wales.

Data will also be linked to data held at the UK Renal Registry and data held at the National Institute for Cardiovascular Outcomes Research (NICOR) to gather further information on kidney and cardiovascular outcomes. The data for NICOR and the UK Renal Registry will be using NICOR as trusted third party for linkage, and therefore are not shown on this simplified diagram. The full data flow diagram can be found in the data management protocol. In brief, audit identifiable data (NHS number) and pseudo-identifier will be sent to NICOR. NICOR will then send linked pseudonymised data to LSHTM for analysis. A similar process applies to UK renal registry data where NICOR will act as trusted third party.

Legal/ethical basis:

The entire set-up of the NCKDA research database has been ethically approved for research purposes as outlined above, and with approval by CAG for linking data without individual consent. The NCKDA was originally set up and funded by NHS England and Wales to audit care and to determine how care impacts on long-term health outcomes, and so there is in addition a need to capture these data for historical research/statistical purposes.

Identifiable primary care data will only be handled by UCL as outlined above for linkage purposes. Data will only be accessed by individuals within UCL who have authorisation as per the data-sharing agreement with HQIP and ethics/CAG to process the data for the purposes described above, all of whom are substantive employees of UCL.

The core clinical pseudonymised data will only be accessed by 2 individuals within LSHTM who have authorisation as per the data-sharing agreements with HQIP and ethics/CAG to process the data to produce subsets of the data that will be accessed by researchers using the data. These subsets of data will only be extracted after the project has been reviewed and approved by the NCKDA steering group. The NCKDA steering group consists of former representatives of the Audit steering group when the Audit was live (before 2017), representatives from UCL, LSHTM, NICOR, Wales, and the UK Renal Registry.

Research project-specific data would be analysed on a project by project basis. Analysts who used the project specific data are trained in analysing routine electronic health record data under appropriate supervision by LSHTM and UCL at LSHTM (other locations are not suitable based on the existing data sharing agreements). Results from analyses of the pseudonymised clinical data will be published in peer reviewed clinical and epidemiological journals. All outputs will be aggregated with small numbers suppressed in line with the HES Analysis Guide.

There will be no data linkage undertaken with NHS England data provided under this agreement other than that which is already stated.

Data will only be accessed and processed by substantive employees of UCL, or LSHTM and will not be accessed or processed by any other third parties not mentioned in this agreement.

What will be produced as a result of the data processing?

1) How does accurate CKD diagnosis/coding in the primary care record impact long-term outcomes (up to 10 years

follow-up)? Specifically, progression to end-stage renal disease (ESRD)/dialysis, transplantation. (These data will come from the UK renal registry and not NHSD)

(2) How and where are patients with CKD managed? - investigating the patient journey from primary care to end-stage renal disease (including dialysis, transplantation or conservative care). As patients with CKD are multimorbid, there is great fragmentation in care delivery and the audit can look at the impact of this fragmentation on outcomes using already linked data (HES in- and outpatient data). Specifically, many patients at risk of or with CKD suffer from

cardiac conditions and therefore the audit allows to better understand how variation in care for cardiac and/or renal patients impacts on outcomes.

Specifically, we have created outpatient variables to capture specialist care provision and used the inpatient data to create variables which capture cardiac care, and are currently working on looking at cardiac care in particular to inform analyses plans for national linkages. A paper will be written by end of this year describing outcomes of CKD patients requiring acute cardiac care.

(3) To which extent does acute kidney injury contribute to progressive kidney disease and incidence of acute and chronic dialysis? By linking the data from the AKI Think Kidneys programme (held at the UK Renal Registry) to primary care data (this dataset) to establish baseline CKD status prior to AKI we would be able to address this important national question with gold standard data.(These data will come from the UK renal registry and not NHSD)

(4) Data validation to enable analyses of single existing datasets (eg. primary care, NICOR or HES data) so that these

do not require linkages. Ongoing - using derived variables.

How good are primary care codes of transplantation and dialysis? The referral date held in the UK Renal Registry database could be used to validate primary care referral entries. How much extra information on renal disease progression is held in renal clinics for those with CKD stages 4 & 5? - ongoing. Renal outpatient visits are captured using HES OP derived variables.

(5) Validation of existing risk prediction algorithms with regards to long-term renal outcomes.

As each result becomes available, there will be submissions to peer reviewed journals (BMJ, JASN, AJKD, PLOS, BMC, JAMA, Heart, etc), presentations and conference contributions (UK Kidney week, European Renal Association, American Society of Nephrology, British cardiovascular Society, British Society for Heart Failure). Ethics approvals are only until 2027, but we anticipate that we will not need the raw HES data going forward once the derived variables have been approved.

What level of data will be contained in the output?

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide

Dissemination and communication approach

The findings of the research will be shared within the renal and cardiovascular community. The planned analyses will be discussed with patients attending the Patient Council at the UK Renal Registry. There is an Audit website which informs about the audit and its findings to date. The Audit steering group anticipate engaging with the renal, cardiovascular, diabetes and hypertension societies as well as the RCGP to publicise the findings of research and to inform future quality improvement projects.

Exploitation of results

Findings from the research will analyses plans using data from future national linkages in cardiovascular and renal research. This information will be shared in research papers and algorithms used within the HDR UK for linked primary and secondary care data. This information will be made publicly available.

The early stages of CKD are usually asymptomatic. Hence it is important that those who are at risk are tested at appropriate intervals so that CKD is identified early, with the opportunity to institute appropriate management to prevent kidney disease progression and cardiovascular disease (CVD) complications. Most patients with CKD will be identified and managed by their GP and there a number of Read codes used by practice computer systems which identify these patients and enable a practice register to support regular monitoring and treatment decisions.

The specific benefit that relates to Question 1 of the Research Questions; "What are the long-term outcomes of patients at risk of and with CKD and which aspects of primary care management influences outcomes?" is as follows: Currently it is unknown how primary care activities affect long-term outcomes of people at risk of or living with kidney disease.

For example, the December 2017 Audit report of the NCKDA data showed that those who have chronic kidney disease (CKD) stages 3-5 based on biochemical test results but who do not have a corresponding read code in their primary care record have worse outcomes than those who do (i.e. they die faster, have more hospital admissions). This raises the question what the outcomes are of patients at risk of CKD or with CKD in practices who are less good in identifying those with CKD have worse outcomes than those who do.

The answer to this question has considerable implications for continued incentivisation of testing for and identification of CKD, including whether earlier stages of CKD (1-2) should be identified to improve cardiovascular outcomes and prevent premature deaths. The mortality data will be important to better understand the distribution of cause of deaths. Note that most patients with CKD die before they require dialysis.

The specific benefit that relates to Question 3 of the Research Questions; "What is the burden of progressive kidney disease and what are its consequences?" is as follows: The statistician who carried out the NCKDA analyses has obtained a MRC PhD fellowship to investigate within the next 3 years the burden of progressive kidney disease and its consequences (Question 3). The answer to this question is currently unknown and of clear importance to the planning of health care delivery.

The specific benefit that relates to Question 2 of the Research Questions; "How and where are patients with CKD managed? - Investigating the patient journey from primary care to dialysis" is as follows: HES and Mortality data will be used to examine the association between variation in coding/ management of CKD with outcomes to better understand if and how these are related. Linked data from outpatient records will identify those CKD patients who were not just managed in primary care and provide a more holistic picture of the burden of health needs of the CKD population. This will also contribute to answering Question 1.

In summary: Patient information will be used to improve patient care by improving the diagnosis of chronic kidney disease (CKD) in primary care and to improve diagnosis and care using electronic patient systems.

Patient information will be used to serve the wider public interest by identifying gaps or shortfalls in commissioning services and providing a more comprehensive picture of care and outcomes of people with CKD within primary care in England and Wales. The research data base will for the first time give a comprehensive picture of renal and cardiovascular care in those at risk of and with CKD. This is also contributing to answering Question 2 of the research questions.

The linked data will for the first time be able to address the question whether short-term outcomes of CKD management such as acute kidney injury (AKI) contribute to later outcomes such as progressive kidney disease, and dialysis start (Question 4: To which extent does acute kidney injury contribute to progressive kidney disease and incidence of acute and chronic dialysis?)

Overall, these linked data will be incredibly useful for health economic analyses which to date have only focused on subsets of patients with CKD (Question 5. What are the disability adjusted life years, years of life lost, healthy life expectancy and cost associated with CKD, AKI and progressive renal disease?)

The multiple linkages will be extremely useful to understand key data items that are available in other datasets for kidney patients (Question 6. Data validation to enable analyses of single existing datasets (e.g. primary care, or HES data) so that these do not require future linkages.

The National Institute for Clinical Excellence (NICE) issued guidance on the early identification and management of CKD in adults in primary and secondary care in 2008 (National Institute for Health and Care Excellence, 2008) with a recent update in 2014 (National Institute for Health and Care Excellence, 2014). These guidelines are currently reviewed.

Findings from the research database will reflect how adherence to these guidelines at the time was associated with outcomes and address open research questions as listed in the respective guidelines. Engagement with all stakeholders and dissemination of findings (see section 5c) will enable further quality improvement efforts and more joint up care, and updates to any relevant further guidelines.

Currently LSHTM are working on obtaining additional funding from the Wellcome trust, BHF, NIHR and Kidney Research UK for analyses of the database, and LSHTM would anticipate that within the next 3-5 years, LSHTM have analysed and published the data with regards to the 6 broad research questions covered by the audit research ethics approval.

Two NCKDA reports have been published, which detail the work that was done using these audit data, and further 2 manuscripts on data quality, and a paper that demonstrated the association between accurate coding of CKD status in primary care and reduced risk of hospitalisation.

The team has written up the outcomes of the coding for CKD in primary care (as a proxy of GP recognition that a patient has kidney disease) and this is published in BMJ Open: Association between practice coding of chronic kidney disease (CKD) in primary care and subsequent hospitalisations and death: a cohort analysis using national audit data - PubMed (nih.gov).

Furthermore, one research paper has been published BMJ Open which describes the completeness and accuracy of identifying acute myocardial infarction (AMI) hospitalisations in two different secondary care datasets (HES and the Myocardial Ischaemia National Audit Project (MINAP, part of NICOR)). A second paper details processes of care and outcomes (all-cause and cardiovascular mortality during AMI hospitalisation and post-discharge, readmissions) associated with CKD stage (under review).