Trauma Audit and Research Network HES and Civil Registrations Mortality application

The Trauma Audit and Research Network (TARN) is a non-commercial organisation affiliated with the University of Manchester and funded by membership fees from participant Trusts that submit data. TARN is the mandated organisation for the audit of trauma care in England, as set out in and endorsed by the Information Standards Board (ISB) 1606. Data submission is mandated by NHS England for Hospitals to be considered as trauma units / major trauma centres. TARN holds Europe's largest database of traumatic injury, with Hospitals within participating Trusts entering details of relevant cases into TARNs online system. Any Trusts receiving trauma patients are eligible to submit data.

The aim of TARN is to support service improvement by providing analytical feedback to Trusts, in reports such as the major trauma dashboards. TARN supports the funding mechanism Best Practice Tariff. The TARN site hosts the ‘Best Practice Tariff’ report, which major trauma centres use to report to commissioners and receive payment. To ensure that analyses are as accurate as possible, TARN needs to ensure that its dataset is as complete as possible.

TARN requires pseudonymised and identifiable HES data to measure completeness of submission to the TARN database for individual Trusts and Hospitals. Trusts are fed back results for their individual Trust, via the secure NHS email account. Identifiable HES data will help Trusts (via TARN) ensure all cases that are eligible for payment are submitted and can be reported on.

3 years of HES pseudonymised and identifiable HES data would greatly assist in the execution of the outlier policy and will ensure greater accuracy in the outcomes. When monitoring outliers two complete years of TARN data are examined to monitor performance over time. Any positive or negative outlier Hospitals are contacted, and a data quality review is performed. Underpinning this process is a review of the number of cases a Trust has submitted vs. the expected number based on HES data. Whether a Hospital has submitted all deaths based on HES for those years is also reviewed. Having the data for 3 years will increase the accuracy of the outcomes and allow this process to be executed effectively and efficiently as currently data is obtained from the Hospital IT departments and analysed for TARN eligible cases.

Case ascertainment is also used in published TARN reports, the Severe Injury in Children’s report and the Major Trauma in older persons report. Both available to view https://www.tarn.ac.uk/Content.aspx?ca=4&c=3572.

Healthcare Quality Improvement Partnership (HQIP) National Clinical Audit Benchmarking (NCAB) is an online publication and is a key example of how case ascertainment is used to inform performance (https://ncab.hqip.org.uk/reports/card/audits/TARN/), an example of case ascertainment based on information from TARN has been provided to NHS Digital. Case ascertainment is also used to inform the national Major Trauma Peer Review process to ensure this information is as accurate as possible TARN needs to ensure its dataset is complete as possible.

Recent projects completed by TARN have also used Case ascertainment. There are 4 recent projects (2 published, 1 accepted and 1 ongoing) where case ascertainment has been used 1 - Time to definitive care within major trauma networks in England

N. R. Haslam , O. Bouamra, T. Lawrence, C. G. Moran and D. J. Lockey . BJS open 2020. DOI: 10.1002/bjs5.50316 2 - Changing the System - Major Trauma Patients and Their Outcomes in the NHS (England) 2008–17. Christopher G. Moran , Fiona Lecky , Omar Bouamra , Tom Lawrence , Antoinette Edwards , Maralyn Woodford , Keith Willett , Timothy J. Coats. EClinicalMedicine 2–3 (2018) 13–21. https://doi.org/10.1016/j.eclinm.2018.07.001 3 - National Temporal Variation in Major Trauma. Will Kieffer, Daniel Michalik, Jason Bernard, Omar Bouamra, Benedict Rogers. Trauma (Sage publication). In press. 4 – Ongoing Older person trauma. Will Eardley. To ensure all analyses are as accurate as possible, TARN needs to ensure that its dataset is complete as possible.

TARN requires identifiable Civil Registrations Mortality data to supplement data already collected (concerning the acute phase of care for patients that have suffered a traumatic injury). TARN need to determine the outcome of patients post discharge. Civil Registrations Mortality data will be used as part of a statistical model to determine rates of survival at 30 days which is then fed back to Hospitals. TARN can then compare the results of this model to another which considers outcome at discharge and possibly determine the efficacy of post discharge support. Furthermore, measuring mortality in this fashion is in line with other mortality assessments such as the Summary Hospital-level Mortality Indicator (SHMI). The information may be fed back to the treating Hospital at patient level to provide individualised feedback in the context of their injuries and treatment, any feedback will be done using the secure NHS email address.

TARN require identifiable HES and Civil Registration Mortality data. Identifiable HES and Civil Registration Mortality data may possibly be shared, at patient level, with the Hospital who originally treated the patient to allow individualised feedback in the context of the patient’s injuries and treatment. Any feedback at patient level will be done using the secure NHS email address.

TARN has developed an outcome prediction model using a 30-day cut-off for mortality/survival of patients. Patients discharged before the 30-day cut-off are considered Alive even if they died after their discharge. To avoid an overestimation of survival TARN seeks a “true” outcome within 30 days, this requires a data linkage using the NHS number of TARN patients, with Civil Registration Mortality data to obtain the outcome. Primarily, NHS Digital data will be used as part of a statistical model to determine rates of survival at 30 days which is then fed back to Hospitals.

The combined (identifiable and pseudonymised) HES and identifiable Civil Registration Mortality data will provide all information needed to complete this project.

The University of Manchester is the Data Controller who will also process data. There are no commissioners involved, data will only be accessed and processed by substantive employees of The University of Manchester.

This project relies on Article 6 (1) (e) and Article 9 (2) (j) as the GDPR legal basis for processing purposes. ’Processing of personal data “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”. 9 (j) for research purpose is used as 9(i) cannot be used as TARN is not a health professional. Special category 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. TARN is the mandated organisation for the audit of trauma care in England, as set out in and endorsed by Standard ISB 1606.

TARN requires HES data to measure completeness of submission to the TARN database for individual Trusts and Hospitals. The HES data will be stored and processed entirely in the Data Safe Haven. The Extract for the HES V TARN comparison will be password protected and emailed out to the individual hospitals via the secure NHS email account. In this way the identifiable HES extract will not be available on the TARN website. HES data helps TARN to help Trusts ensure that all cases that are eligible for payment are submitted and can be reported on.

TARN requires HES data from NHS Digital for patients with at least one ICD-10 code relating to traumatic injury. Any of these patients are potentially eligible for the TARN audit, which is concerned only with traumatically injured patients.

TARN does not audit all traumatically injured patients: TARN further processes the received data to create a reduced cohort that correlates with TARN inclusion criteria. Records would be linked to Hospitals and Trusts using Organisation Data Service (ODS) codes. The purpose of this is to derive a number of TARN-eligible cases for each participating Hospital and Trust, which is then used to assist Hospitals with case identification and to measure completeness of data.

TARN requires 2 different copies of HES. Pseudonymised HES is required as this extract will not have patient objections applied and as such will be a complete record of all the cases which will enable to audit to calculate true case ascertainment figures for each of the eligible trusts.

Identifiable HES is required as the extract will be used to notify Trusts of potentially individual eligible cases that were not reported to the Audit, via the secure NHS email.

PSEUDONYMISED EXTRACT

• Containing no identifiable information from which no data will be removed because of national patient opt-out. The two pseudonymised and identifiable extracts will not contain common identifiers that could be used to link them.

• The dataset will be filtered by ICD-10 codes to identify cases that appear to be eligible for the TARN audit

• The pseudonymised HES dataset containing all potentially eligible episodes will be used to determine the total number of eligible cases used in data completeness (case ascertainment) calculations

IDENTIFIABLE EXTRACT

• Containing NHS Number for all eligible patients and excluding data for all patients who have opted-out of having their data shared. Data filters applied to limit data to specific primary or secondary diagnoses relating to traumatic injury 2 extracts requested. Datasets will be filtered by ICD-10 codes to identify cases that appear to be eligible for the TARN audit.

• The two pseudonymised and identifiable extracts will not contain common identifiers that could be used to link them

• The HES data required is the case (submission) ID, TARNs own data will be filtered to this. TARN will use the identifiable NHS number for matching, but the NHS number will then be deleted immediately.

• The identifiable HES dataset containing NHS number will be used to notify Trusts of potentially eligible cases that were not reported by linking to the TARN database using NHS number. This is achieved through the provision of lists of patients (containing NHS number, arrival date, age, length of stay in Hospital, discharge destination and the first 5 ICD-10 codes) sent securely and confidentially to the Hospital that provided the treatment. Only data relevant to the specific Hospital is sent to each Hospital via secure NHS email address and no provider will receive data about individuals for whom they did not provide treatment. Data will be only sent via NHS email address to confirmed users of the TARN EDCR, with access to that hospital, to another NHS email address. In this way no NHS Trust will receive data that did not originate from them.

The information may be fed back to the treating Hospital at patient level to provide individualised feedback in the context of their injuries and treatment.

TARN also requires Civil Registration deaths secondary care cut data with NHS number as the sole identifier. TARN has developed an outcome prediction model using a 30-day cut-off for mortality/survival of patients. Patients discharged before the 30-day cut-off are considered Alive even if they died after their discharge. To avoid an overestimation of survival TARN seeks a “true” outcome within 30 days, this requires a data linkage using the NHS number of TARN patients, with Civil Registration Mortality data to obtain the outcome. Primarily, NHS Digital data will be used as part of a statistical model to determine rates of survival at 30 days which is then fed back to Hospitals.

TARN require a flag at 30 days of patient outcome, whether the patient is dead or alive. TARN would like to ascertain the outcome of patients 30 days post discharge. Outcome (alive or dead) at 30 days from injury has historically been used in the calculation for comparative survival rate (Ws). This is calculated using observed and expected survivors and the total number of patients in the Hospital’s rate of survival dataset.

Ws Excess deaths or survivors (W) standardised according to Hospital case mix using the TARN fraction. A Hospital with the same case mix as the overall TARN population will have identical W and Ws values. A Hospital whose case mix differs from the overall TARN population will have different W and Ws values. However, many patients are discharged before this 30-day point. To include these patients, TARN need to know whether patients died at or before the 30-day point after leaving Hospital. To do this, TARN needs information about post discharge deaths from NHS Digital and use this information in the calculations of Ws for hospitals submitting to TARN.

Data will only be processed as described in this agreement and access will be restricted to substantive employees of The University of Manchester and Hospitals that provided the treatment with which data may be shared. No data will be sent to other organisations other than to Hospitals about the patients they have treated.

TARN requests one data drop data for 2016/17, 2017/18, 2018/19 and then an annual drop for the following year 2019/2020. Each year TARN will request the latest available year of data and delete the oldest year of data that TARN hold, if older than 3 years TARN will delete agreed data within 3 months of receipt of new data on a rolling basis. TARN will provide a data destruction certificate.

TARNs outputs relevant to this application are:

Case ascertainment measures at site level which are published in the Major Trauma Dashboard. Circulated securely and confidentially to Major Trauma Centres in February, May, August, and November. The Dashboard measures were drawn up and agreed by clinical experts on the Clinical Reference Group (CRG) to allow effective benchmarking in relation to specific measures between Major Trauma Centres.

Case ascertainment measures at site and Trust level which are published in the TARN clinical reports. Circulated securely and confidentially to TARN member Hospitals in March, July, and November. Case ascertainment measures contain no identifiable or record level personal data. All published data is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Case ascertainment at site level which are published in the TARN clinical reports. Shows comparison to the target of 80% case ascertainment and compares them to last year. These are circulated securely and confidentially to TARN member Hospital in March, August, and November

The Civil Registration data will be used as part of the TARN clinical reports to provide a true 30-day outcome of patients. Circulated securely and confidentially to TARN member Hospitals in March, July, and November. The Civil Registration data will also be used as part of the TARN Older Persons and Children's reports.

Outcome (alive or dead) at 30 days from injury has historically been used in the calculation for Ws. However, many patients are discharged before this 30-day point. To include these patients, TARN need to know whether patients died at or before the 30-day point after leaving Hospital. To do this, TARN need to receive information about post discharge deaths from the Civil Registry and use this information in one of the calculations of Ws for Hospitals.

The aim of TARN is to improve patient care. TARN carries out several planned and ad hoc analyses for hospitals and trusts to highlight good performance and areas where performance could be improved. A key element of TARNs work is analysing rates of survival to identify hospitals with excess deaths, based on injury and demographic profiling. This work is dependent on the quality/accuracy of information emanating from the trusts so the use of HES data to monitor and improve completion rates contributes towards this overall aim. Using HES data to calculate data completeness assists in determining whether apparent poor performance is related to poor data collection, or whether other issues exist that need to be examined further.

Using HES data allows TARN to derive a denominator of expected cases, which in turn allows TARN to identify which sites may be missing patients. TARN can then work with those sites on case identification. An additional benefit of improving case identification is Best Practice Tariff, where Major Trauma Centres receive payment for meeting national standards on a patient by patient basis. The notification to trusts of potentially eligible patients that were not reported by the provision of patient lists and linked to the TARN dataset using NHS number has resulted in significant improvements in rates of case ascertainment. Improvements ranged from 5 to 50% with an average improvement of 20%.

Information recorded on TARN is regularly fed back to trusts in the form of reports and analyses, these may then be used to identify areas of concern and/or best practice in the process of care. The reliability of the reports and analysis is directly related to the proportion of the relevant population they represent. In instances where a low proportion of the population is represented within analysis; bias may be introduced or important results missed.

The Case ascertainment and Survivor/Death ratio derived from the HES data therefore enable TARN to validate the information TARN receive from Hospitals, by effectively putting a Data Quality stamp on each Hospital’s reports.

3 years of HES Data would greatly assist TARN in the execution of TARNs outlier policy. When monitoring outliers TARN look at 2 complete years of TARN data and monitor performance over time. TARN will then contact any positive or negative outlier hospitals and perform a data quality review. Underpinning this process is a review of the number of cases a Trust has submitted vs the expected number based on the HES data. TARN will also review whether a hospital has submitted all deaths based on HES for those years. Having the data for 3 years will allow this process to be executed more effectively and efficiently as currently TARN must get the data from the hospital IT departments and then analyse it for TARN eligible patients.

Taken from NIC-326033:

Following the provision of patient lists linked to the TARN dataset there have been significant improvements in case ascertainment where hospitals have engaged with the information, ranging from 5 to 50% with an average of 20%. Significant improvements in case ascertainment continue to be seen at trusts that have engaged with the TARN/HES data. Those that have done so having an average case ascertainment rate of over 90%, compared to 70% for those that have not. One example is Birmingham Children’s hospital, a children’s major trauma centre – they used the TARN/HES data to identify an extra 42 cases, improving their data completeness from 81 to 94%. As they’re a major trauma centre these additional cases will also have attracted best practice tariff payment. Another example is Sunderland Royal Infirmary, prior to using the TARN/HES data they felt that they were identifying all of their eligible cases, however their case ascertainment rate was consistently around 80%, making use the data they found a tranche of patients admitted to a medical ward that they were inadvertently missing . After submitting these cases their data completeness rose considerably.

The combined TARN/HES data has also proven to be a useful tool for responding to outlier surveillance. It can be used to determine if a site is failing to submit a particular cohort of patients (e.g. over 70s, died etc.) and case ascertainment figures are used to inform the reliability of a given sites outlier status.