The Distribution of Highly Sensitive Troponin in the Critically Unwell and Associated Mortality

A blood test called troponin forms a key part of the diagnosis of a heart attack (myocardial infarction). Troponin refers to a group of proteins that help regulate the contractions of the heart and skeletal muscles. High troponin levels can indicate a problem with the heart. The heart releases troponin into the blood following an injury, such as a heart attack. The upper limit of normal for this test is determined by the manufacturer of the test, based on the troponin levels in healthy individuals, generally aged between 18-40. Recently assays have been developed that provide a higher degree of sensitivity and as such allow rapid exclusion of a heart attack within a few hours. This benefit does however raise another issue; elevated levels of high sensitivity troponin are now frequently seen in patients who have not suffered a heart attack, particularly in association with severe illness states, such as in patients in intensive care. In the absence of clinical features of a heart attack an elevated high-sensitivity troponin leaves clinicians uncertain as to how to interpret this result and whether it should result in changes to patient care.

The first part of the study, which is in the advanced stages of analysis, collected high-sensitivity troponin results from 1,612 consecutive patients (over the age of 18 years) who had blood samples requested by clinicians at University Hospital Southampton (UHS) during their stay in one of three intensive care units (Cardiac Intensive care, General Intensive Care and Neurosciences Intensive Care). The aim was to provide a description of the distribution of high-sensitivity troponin and to describe factors associated with it's release. Further aims were to assess whether there was an association between high-sensitivity troponin concentration and inpatient mortality. A secondary aim of the study was to apply to NHS Digital (for which this agreement is in place) for longer term mortality data (one year mortality was agreed as the most appropriate endpoint after discussion with the Confidentiality Advisory Group).

This study has highlighted that elevated concentrations of high-sensitivity troponin are frequently seen in patients in intensive care and therefore this test should not be used as a binary arbiter of whether someone has had a type 1 myocardial infarction (a traditional myocardial infarction associated with coronary artery obstruction). The

cause of the elevated high-sensitivity troponin is likely to be due to myocardial injury, rather than a type 1 myocardial infarction in most cases. The study has also demonstrated that the high-sensitivity troponin concentration is closely associated with markers of illness severity. Furthermore, outside cardiac intensive care, high-sensitivity troponin concentrations have been shown to be closely associated with inpatient mortality.

Having established these important observations in the original study, UHS now wish to move onto the pre planned secondary aim to assess whether high-sensitivity troponin taken in critical illness is a predictor of longer term mortality. Does this bio marker just act as a predictor of short term risk, as it is associated with illness severity, or does it predict future events because it acts as a marker of previously unknown cardiovascular disease?

This study will assess the one year outcomes of the cohort (mortality and cause of death). This study is of significant public interest for three reasons. Firstly, as demonstrated by the original study high-sensitivity troponin elevation is frequently seen in patients in critical care; secondly - clinicians are often unsure whether these results are of clinical significance. Finally, if this study does demonstrate an association between high-sensitivity troponin levels and outcomes then further studies will result to assess whether any medical interventions, (particularly proven cardiovascular therapies), could alter the prognosis in this group. This study therefore has significant potential implications for the healthcare of a large population of patients both in the UK and internationally.

As such the study meets the standards set out in section Article 6 (1)(E) of the General Data Protection Regulation for processing these data: "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". Furthermore for these reasons it also meets the expectations set out for processing data set out in Article 9(2)(J) of the General Data Protection Regulation: "processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of 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 key ethical question raised by this study is whether it is appropriate to process these data without consent. As already discussed the potential improvements that this study could offer to patient care both in the UK and internationally mean that this study is undoubtedly in the public interest. Furthermore this study will have minimal risk to the patients included in the study and their personal data will be kept pseudonymised. It is therefore clear that the study has an ethical mandate. This opinion is also held by both the Research Ethics Committee and the Confidentiality Advisory Group, as demonstrated by their support for this study.

The data requested will help the research team to assess whether there is an association between high-sensitivity troponin levels and clinical outcomes. Specifically mortality status will allow the team to assess whether there is a link between high-sensitivity troponin levels and mortality. The cause of death is important because it would seem more likely that high-sensitivity troponin levels will predict cardiovascular mortality rather than all cause mortality. Without the cause of death the research team will be unable to evaluate this.

The data returned from NHS Digital will be record level pseudonymised (just with the unique study identifier that the CRG at UHS will send) to allow the CRG at UHS to link these data with the high-sensitivity troponin levels. If data were provided on an aggregate level it would not be possible to fully assess the association between high-sensitivity troponin levels and clinical/mortality outcomes. One year outcomes should provide the research teams with a clear picture of the relationship between high-sensitivity troponin levels and outcomes. This is the minimum time frame required to answer the primary objectives of this study. It is likely that some patients will have moved or have been admitted to hospital whilst away from their permanent residence. Therefore the data will not be restricted to one particular geographical area, such as Southampton. It is not possible to achieve the aims of the study without processing these data. Furthermore the research team have specifically asked for the minimum data required to complete the objectives of this study, any less and it would not be possible to robustly evaluate the association between high-sensitivity troponin levels and clinical outcomes.

The results from these analyses will result in improvements in the way clinicians interpret high-sensitivity troponin levels but may also precipitate further studies to assess whether any medical interventions could alter the outcomes in at risk groups identified by this study.

This study is a pre specified follow up of the distribution of highly sensitive troponin in the critically unwell and associated mortality study which recruited 1,612 consecutive patients in one of the three adult intensive care units at UHS in whom a biochemistry sample was performed at UHS between the 30th January 2019 and 30th July 2019. There are no additional phases. This particular study will be referred to as the distribution of highly sensitive troponin in the critically unwell and associated mortality - one year follow up.

UHS will are the sole Data Controller. UHS and the Keele Cardiovascular Research Group (KCRG -part of the University of Keele (UoK) will are the data processors. The KCRG was set up as a specialist centre for the analysis of large cardiovascular data sets. As such the KCRG were involved in the analysis of the original the distribution of highly sensitive troponin in the critically unwell and associated mortality study and it would be remiss not to continue to use their expertise to ensure that the data supplied by NHS Digital provide robust answers to the objectives of this study. The spreadsheet that will be sent to KCRG will contain the age (not date of birth) and gender of the patients in the cohort, alongside date and cause of death from NHS Digital, but no other additional identifiable data will be shared. KCRG will act under instructions from UHS - therefore reinforcing that UHS are sole data controller for this project.

The original study was funded by an unrestricted research grant from Beckman Coulter (BC). BC had no input in the study specifics. This sub-study requires no input from industry and specifically BC have no ongoing role. There are no other organisations involved. There are no funders/commissioners involved.

Only substantive employees of University Hospital Southampton NHS Foundation Trust and the Keele Cardiovascular Research Group (who have been trained in data protection and confidentiality) will have access to the data and only for the purposes described in this document. The data will be stored on secured computers within each of the listed data processing organisations.

University Hospital Southampton NHS Foundation Trust (UHS) provided NHS Digital with the following identifiers of the cohort for the linkage to the data requested

- NHS number,

- date of birth,

- gender

- Study ID.

NHS digital returned mortality data (date of death and cause of death) with each unique study ID. The CRG at UHS will then use the Study ID to link the data provided by NHS digital to the high-sensitivity troponin levels. Once this is complete the data will be sent to the KCRG to allow both the CRG and KCRG to undertake further statistical analysis. Re-identification of these patients will not be possible because the database will only contain the unique study identifier and gender of the patient - no other patient identifiers will be included. The patient identifiers are kept in a separate database. KCRG have no access to these other patient identifiers.

UHS ensures that the identifying data they hold are segregated with the pseudonymised data and linkage by them is completed using the unique study ID.

No contact will be made with the patients selected. The data will only be used for this study.

The NHS Digital data is crucial to provide the most accurate information on the outcome of patients recruited to this study and will form the basis of a robust research study that will be submitted to peer reviewed journals.

The data from this study will not be used for commercial purposes, not provided in record level form to any other third party not mentioned in this agreement, and not used for direct marketing or commercial purposes.

All outputs will be aggregated with small numbers suppressed in line with the HES analysis guide.

NHS Digital reminds all organisations party to this agreement of the need to 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).

There will be a multi-faceted approach to the dissemination of the results of this study. Firstly the results will be presented at respected medical conferences. Secondly the data will be published in respected peer review journals. These two activities will ensure that the key messages are available to the scientific community. The results will also be published (in a non-scientific format) on the UHS web pages. Finally the promotion of the published article through the use of social media will also widen the audience.

The results of this study are likely to be of immediate clinical relevance to clinicians by providing them with a better

understanding of the clinical significance of the high-sensitivity troponin levels in intensive care. Furthermore, if these data demonstrate that the high-sensitivity troponin levels are associated with cardiovascular mortality then this is likely to stimulate further research to evaluate whether medical interventions can alter the outcomes seen in this group of patients.

Whilst the aggregate results will be published to allow clinicians, scientists and the public to understand the implications of raised high-sensitivity troponin levels, the data themselves will not be released and will be stored and then destroyed following the standards set out by NHS Digital.

Once the study has completed UHS plan to ensure that the results are available widely both to the medical community and the public. The outputs to the medical community will take the form of peer reviewed publications and presentations at medical conferences. The main output to the public will be through the updates and explanation of the results in the trusts patient facing research website.

All outputs will contain only aggregate level data with small numbers suppressed in line with the HES analysis guide.

An extension Agreement is required due to delays that resulted from the COVID-19 pandemic. As such the manuscripts are now being sent to journals for publication but in order to address any issues raised by the reviewers University Hospital Southampton Foundation Trust need to maintain access to the data.

The expected measurable benefit of the data access and the study will be an important contribution to the optimal use and interpretation of high sensitivity troponin assays in intensive care. Use of high sensitivity troponin assays in intensive care is flawed in two ways. Firstly, there is widespread misunderstanding about the meaning of a raised high-sensitivity troponin; such a result is often incorrectly labelled as implying type 1 myocardial infarction, due to low awareness of myocardial injury. Secondly the upper limit of normal supplied by the manufacturer of the blood test is based upon a predominantly healthy population which is not reflective of the population of patients seen within intensive care. Thus, whilst the current application of high-sensitivity troponin is largely flawed, data are accumulating that the assay (determining the quality) may have a clinically valuable alternative role as a bio marker of cardiovascular risk.

The proposed study represents a unique opportunity to assess whether the high-sensitivity troponin levels in intensive care patients are indeed associated with longer term mortality. If so, the "never means nothing" hypothesis of high-sensitivity troponin should lead to changes in the way that intensive care clinicians interpret the assay in practice. It would no longer be an assay mistakenly used to "rule in" type infarction, but rather a bio marker for patient risk and prognosis.

The dissemination of this data is in the public interests because it has the potential to change the way that large groups of patients are managed. The data are likely to result in changes in the way that high-sensitivity troponins are requested and also interpreted by clinicians in intensive care. Furthermore as already discussed it is likely that these data will precipitate further study to assess whether cardiovascular interventions could change the outcomes in these patients. This study will have wide implications, because, as demonstrated by the original study elevated concentrations of high-sensitivity troponin are frequently seen in patients within the intensive care environment.

The expected benefit will be achieved by potential changes in clinical practice. It is expected that the dissemination of these data will occur within six months of NHS Digital releasing that data and then changes in clinical care will follow afterwards. Due to delays resulting from the COVID-19 pandemic there were delays receiving the data and progress with analysis was also delayed.

The manuscripts have now been written and are being submitted to clinical journals currently so that the results can be disseminated. Furthermore the data have resulted in the acceptance of two abstracts for presentation at the European Society of Cardiology Congress in one week. Again this will help to ensure that the results are widely distributed