The short and long-term cardiovascular consequences of critical illness: The C3 Study

The C3 Study: The short and long-term cardiovascular consequences of critical illness is designed to find out which patients are at risk of heart attacks/strokes up to several years after discharge from an Intensive Care Unit (ICU).

This study will also investigate whether treatments and events occurring in an Intensive Care Unit ICU contribute to this risk.

The mortality of patients who survive a period of treatment on an ICU is considerably higher than an age and sex matched general population. There is evidence of a high rate of major adverse cardiovascular events (MACE) amongst ICU survivors. MACE events include nonfatal stroke, nonfatal myocardial infarction, heart failure and cardiovascular death.

There are no data resources in existence that combine data collected during the episode of critical illness (occurring in ICU and during admission to hospital), with longer-term health data such as repeat hospital admissions, cardiovascular events, treatment and other related medical conditions.

The aim of this study will be to find out which patients are at risk of MACE after discharge from an ICU. This study will also investigate whether treatments and events occurring in ICU contribute to this risk.

In order to perform this study, a data resource needs to be generated that is sufficiently large and detailed. In an ICU, patients’ vital signs, treatments and blood tests are often electronically recorded as part of normal care. Following successful treatment on ICU patients remain admitted to the hospital before being discharged. During this time, they continue to accumulate electronic test results such as laboratory reports and diagnostic data. During this period on the wards they may also experience a heart attack or stroke which it is vital that is captured for purposes of the study.

By linking these highly detailed electronic health care records with NHS long-term follow-up data, the study team can unpick what factors increase patients’ long-term risks and identify patients at highest risk of having heart attacks/strokes years after ICU care.

This study will provide new knowledge about the associations between baseline cardiovascular risk, the disease resulting in ICU admission and therapies / events on ICU with subsequent MACE events, to allow the ongoing risk of these events to be determined. This may identify modifiable risk factors and allow for preventative treatments, improving the health outcomes of this vulnerable group of patients.

University of Oxford’s justification for processing is GDPR Article 6 (1) (e): The processing necessary to perform this task is in the public interest and the task has a clear basis in law. This is an issue of patients suffering critical illness and the clinicians treating it, what happens to patients after they leave ICU and whether the study team can identify those at risk and potentially prevent it.

The dissemination of the aggregated results of this study pose no risk to the public.

GDPR Article 9 (2)(j): processing is necessary for scientific research purposes and shall be proportionate to the aim pursued, respect the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

The data requested will allow the study team to build a novel database linking intensive care therapies and individual patient responses with longer term time to event data (both fatal and non-fatal) medical events requiring hospitalisation. Using this data, the study team can study in unprecedented detail both the event rate and methods of predicating who is at greatest risk or suffering an adverse event.

This study is staffed by two NIHR funded doctoral research fellows who are substantive employees of University of Oxford and work for the Critical Care Research Group in the University of Oxford. The themes of the two research programmes are focused on cardiovascular health after intensive care from both the myocardial event rate and the cardiac rhythm perspectives, respectively. This request relates to The C3 (short- and long-term consequences of critical illness) study (c3study.org). This is an NIHR funded observational study of patients admitted to certain intensive care units with the aim to find out which patients are at risk of heart attacks/strokes up to several years after discharge from an ICU. The team will also study how much the treatment the patients received on ICU contributes to this risk.

• Conduct a detailed review of what has been written in this area

• Find out how many patients suffer strokes/heart attacks after ICU care in England

• Work out which diseases and ICU treatments make heart attacks/strokes more likely

• Determine which patients are at greatest risk of heart attacks/strokes up to several years after leaving ICU

By linking these highly detailed ICU records with NHS long-term follow-up data, the study team can unpick what factors increase patients’ long-term risks and identify patients at highest risk of having heart attacks/strokes years after ICU care.

This study is a standalone project, and no wider collaboration is planned.

Retrospective observational cohort study of patients >16 years of age admitted to an Intensive Care Unit (ICU) in one of the study sites. This is a non-interventional cohort study comparing those patients who experience cardiovascular events and patients in the post ICU population who do not.

This study will provide new knowledge about the associations between baseline cardiovascular risk, the disease resulting in ICU admission and therapies / events on ICU with subsequent MACE events, to allow the ongoing risk of these events to be determined. By understanding who is at risk, it may be possible to modify known risk factors. Where the risk factors cannot be modified, the study team may be able to add preventative treatments.

The study team will create a new database to containing patients’ vital signs, treatments and blood tests, data routinely collected in ICU. By linking these highly detailed ICU records with NHS long-term follow-up data (HES & Civil Registration Data data), the study team can unpick what factors increase patients’ long-term risks and identify patients at highest risk of having heart attacks/strokes years after ICU care.

To understand how events in ICU effect cardiovascular risk after ICU discharge it is important to obtain data about subsequent events. Linking ICU data to key data items in the HES and Civil Registration data datasets are key in this regard.

Only pseudonymised data will be retained and used for analysis by the study team. HES data is an essential feature of this study providing the follow-up and coding data required to define the outcome of interest. HES data has been received on a per patient basis as it must be linked to the study dataset. However, only pseudonymised data was received by the coordinating site (Critical Care Research Group Oxford University) from NHS England (all postcodes have been translated into LSOA codes by NHS England). No data from NHS England will be transferred to the participating sites. At the co-ordinating site the dataset to which it is being linked will not contain any direct identifiers

To adequately answer the study’s questions, many years’ worth of data is needed. This will provide the scale required to provide robust results and predictive model development. Relatively few people are admitted to ICU every year at each hospital (around 1000 admissions for a teaching hospital ICU). The study will include admissions dating back to 2006 (around the time the digital systems needed to conduct the study started to be used on ICU) to gain about 14k admissions to ICU per hospital. Some of these NHS Trusts have several ICUs but not all the ICUs have been using the digital systems for the entire 14 years. The study team therefore estimates that the 4 NHS Trusts currently in the study will provide around 84k admissions. The study needs a minimum of 5 years of HES data prior to the incident admission to ICU to assess co-morbidities and risk factors. The request for this number of years of data is therefore based on a recruitment period starting in 2006 to the current day and 5 years prior to the first admission.

Obtaining data from different types of hospital (district general and teaching hospitals) from different geographical locations, will make the study results more generalisable to the UK as a whole. The published data will only ever be published in aggregated form (this includes all mediums both published and websites etc.) and will apply small number suppression to prevent the results of small sub-groups of patients being published. The study results may will then be heeded by more hospitals, potentially benefiting more patients.

In studies of this size, individual patient consent is not feasible. No alternative study method was available that could provide the necessary data. The study team have therefore taken steps to ensure data processing is minimised and as unobtrusive as can be. The study team obtained full support from the Confidentiality Advisory Group.

The study will extract only data that will aid with the answering of the questions of the study. The study team require up to 120K ICU encounters (defined as a single visit to ICU) in order to overcome many of the challenges studying this outcome represents. The size of the cohort required has been determined by an a priori sample size calculation to provide adequate power to detect risk factors associated with MACE events after ICU admission. The number of years of historical data required were determined by this sample size. All patient episodes within the specified timeframe are required so ensure readmissions and repeated events are properly recorded. Cohort minimisation will occur at each participating site. Only those patients with valid admissions to participating ICUs will be included

HES – Used for timings and coding of co-morbidities and events that occur in the 5 years prior to ICU admission and the 5 years post.

Lower Layer Super Output Areas (LSOA) codes - Map different areas of the country to published deprivation indices which can be used to adjust for variations in deprivation across the England.

Civil Registration of Deaths – Limited to timing and cause of death with LSOA and occupation added for the purposes of correction for deprivation.

The study team have gone to some lengths to ensure that personal data collected will be relevant and limited to what is necessary in relation to the purposes of the study.

The organisations involved are:

a. Co-ordinating site (data controller, processor and guardian) – Critical Care Research Group, University of Oxford.

b. The participating sites (which will submit identifiers directly to NHS England and pseudonymised data to the participating site):

i. Oxford University Hospitals NHS Foundation Trust

ii. Royal Berkshire NHS Foundation Trust

iii. Imperial College NHS Foundation Trust

iv. King College NHS Foundation Trust

By having each site submit their identifiers directly to NHS England using their encrypted file transfer service, this improves data security as the identifiers do not need to be aggregated or stored for any period of time by the co-ordinating centre.

c. Internal linkage is performed by NHS England to Civil Registration of Deaths, HES APC and HES A&E.

The protocol and CAG application forms discuss third party linkage to NICOR via NHS England. This is currently on hold from a study management perspective and will not proceed without future amendment of both the CAG and this NHS England Data Sharing Agreement.

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)”

The participating sites have submitted identifiers to NHS England. These individuals identifiers have been linked to the corresponding Hospital Episode Statistics (HES), HES Admitted Patient Care, HES A&E (Accident and Emergency) Data, and, Civil Registration Data by NHS England.

The individual sites have each contributed to the study cohort by uploading the local identifiers via the NHS England secure upload system. Members of the study team attended the individual sites and assisted with this process, but these identifiers will never be transferred to the co-ordinating site. All the participating / individual sites have been combined to form one study cohort by NHS England to form one single cohort.

Pseudonymised patient level data from HES and Civil Registration data was requested from NHS England by the co-ordinating site, University of Oxford.

Data flow consisted of:

- Each participating site allocating participants a unique study ID.

- Identifiers (specified below) submitted from each site to NHS England using their secure file transfer system

- NHS England merging these identifiers into a single cohort

- NHS England linkage to the requested datasets.

- NHS England sending the co-ordinating site (Critical Care Research Group – University of Oxford) historical and yearly updates from the above-mentioned data products for the duration of the agreement as specified.

The co-ordinating site will hold and process the pseudonymised dataset.

Data from NHS England (HES/ Civil Registration data), and the participating sites will be linked and then subject to a process of data cleaning and data quality assessment. The resulting dataset will then be used for statistical analysis in keeping with the statistical analysis plan detailed in the study protocol.

All linkage will be performed via/by NHS England. The study database (held by the co-ordinating centre will not hold any direct identifiers. The identifiers used for linkage to NHS England (listed also in the s251 approval form are)

- NHS number

- Data of birth

- Sex

- Postcode

There is no matching or linkage with any other public data.

There will be no requirement/attempt to re-identify individuals.

Data processing is only carried out by substantive employees of the data processor(s) and or data controller(s) who have been appropriately trained in data protection and confidentiality. All data processing takes please within a secure system which is designed in keeping with the principles of a data safe haven. The data does not leave the system at any time. The system is compliant with the NHS England DSPT Toolkit.

All data is held within the environment which is owned and run solely by the data controller/data processor and data guardian / co-ordinating site – Critical Care Research Group, University of Oxford.

This study aims to publish at least two journal articles within the first 3 years of the study tackling the areas of cardiovascular risk and atrial fibrillation following critical illness respectively. These will be published in high-impact open access peer-reviewed journals (e.g European Society of Intensive Care Medicine Journal and Critical Care).

All published output will be accompanied by a corresponding press releases including lay summaries of the findings and its applicability to patients. Findings will be presented at national and international conferences to experts in the field.

Outreach will occur to specialist interest groups such as ex-patients and their families via:

- ICU charity “ICUsteps” (https://www.icusteps.org/)

- Intensive Care Society’s patient group

- Oxford patient forum

Outputs could include commentary, tweets and statements of support

When major findings are published, the departmental (Critical Care Research Group at University of Oxford) press office will assist with press releases, social media messages and interviews. The study has a significant social media presence using a Twitter account comprising of patients and health care professionals. All findings and comments will be manged through this Twitter account.

The study website will be updated with all the details of the above.

The study team will engage with policy makers, such as Intensive Care Society should the results suggest that areas of clinical practice might be improved.

Through addressing questions about the impacts of critical illness / ICU on subsequent cardiovascular disease amongst survivors, the outputs of this work will inform:

- Clinicians regarding the ongoing/future cardiovascular risks to their patients (journals and conference presentations)

- Patients now and in the future of their short, medium and long-term cardiovascular risks (twitter and specialist interest groups)

- The wider health care community and health services that are tailored to the treatment of patients during and after ICU (combination of journals, conference presentations, twitter and specialist interest groups)

Likely action/change/decision from this work:

Derive a method of identifying which patients are at risk of heart attacks, strokes and arrythmias and understand common risk factors. This work will identify potentially modifiable risk factors that may lead to identification and treatment of vulnerable patients.

Magnitude of impact:

The research will be able to estimate both the size of the affected population and therefore the number of potentially preventable cardiac events following ICU in England.

Actions leading to benefit:

Through publication of several scientific papers and presentations at scientific meetings on the topic the study team will raise the understanding of the scale of this problem faced by patients following ICU. This work should lead to further funded work that result in prevention and treatment options.

Measuring benefit:

Ultimately the benefit will depend on which risk factors are identified and the degree to which they can be modified in clinical practice.

Achievement:

The aim is to achieve these goals within the 3-year study period.

This has been an impactful project so far. The University of Oxford collected all the data as planned from 2 different sites and created a research resource with un paralleled detail specifically to research cardiovascular health after ICU. It has supported to PhD projects, two predictive models have been created and tested and the University of Oxford have plans to utilise this data further in the future, hopefully putting at least one model into clinical practice.