Epidemiology of traumatic brain injury in England

The aim of processing NHS England data is to determine at a population level, the incidence of head injury (the event of a person receiving a blow to the head), traumatic brain injury (a dysfunction in the normal working of the brain due to a head injury), intracranial haemorrhage, neurosurgical intervention, and death within 28 days of injury.

Traumatic brain injuries can result in significant morbidity and mortality, and leads to a significant healthcare need for patients, the healthcare system and society. To understand the need, it is essential to understand the size of the problem. Patients may experience a head injury. A head injury is distinct from a brain injury, because it is possible to sustain a head injury without a brain injury. Patients or their associates may choose to attend an Emergency Department because they sustained a head injury. Within the Emergency Department they may be diagnosed with a brain injury. They may be admitted into the hospital or they may be discharged home. Epidemiological studies report all these rates, i.e. Emergency Department attendance rate, hospital admission rate, head injury rate and brain injury rate. Between 1974 and 2013, 30 studies of the epidemiology of head injuries were published internationally. Of those, 22 (73%) reported total ED attendances per year; 20 (67%) reported ED attendance rate; 12 (40%) reported hospital admission rate; 28 (93%) reported head injury rate and 25 (83%) reported brain injury rate. Of note only one of those studies purported to be a national analysis of head and brain injury epidemiology, but this was extrapolated from survey data and so was not truly national. This research aims to be the first truly national study of head and brain injury epidemiology, and to be comparable to any previous research. In order to achieve this, accurate ED attendance data, in addition to admission data, is required for analysis. For this reason, specific fields within ECDS as defined in the DARS product specification, minimised as far as possible, have been requested for all ED attendances.

The processing will be carried out under Articles 6(1)e and 9(2)j of the General Data Protection Regulations (GDPR). Article 6(1)e states that processing of data is lawful if ‘processing is necessary for the performance of a task carried out in the public interest’. Consequently, the lawful basis for processing is under ‘Public task’. Article 9(2)j states that processing is permitted for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes for the special category health data. The research is in the public interest, will involve processing using sophisticated statistical techniques, and is directly related to health.

The data requested will be processed using the population of England as a denominator for national population incidence calculations, and the number of patients attending Emergency Departments in a year for Emergency Department attendance rate calculations. The data will be stratified by age category, sex, ethnicity, geography and deprivation index in order to identify subgroups of the overall population that are at increased risk of significant outcomes. This will be invaluable information for healthcare planners. Those independent variables will be used in logistic regression models to identify risk factors for five dependent variables (outcomes) including head injury, traumatic brain injury, intracranial haemorrhage, neurosurgery, and death.

There are five previously published articles on the epidemiology of traumatic brain injury in the UK, which are based on data from the 1970s onwards. None are national, four are based on in-patient cohorts, and so bias to the severe end of the TBI spectrum, and one is based on data from a single Emergency Department. The existing data are therefore both out of date and incomplete and are of limited utility to health care planners.

The data subjects are all patients who attended an Emergency Department in England in 2019. The population of interest are those with a head injury, based on the criteria defined below.

Patients with head injury will be defined as follows:

1. A patient having one or more of the chief complaints:

a. Injury of head (disorder)

b. Traumatic injury (disorder) AND linked Diagnostic Imaging Data Set (DIDS) record of having had a Computerized Axial Tomography (CT) head, or CT head and cervical spine, but NOT CT head, cervical spine and other e.g. chest, abdomen & pelvis. This is because National Institute for Health and Care Excellence (NICE) guidance on imaging after head injury advises CT cervical spine if a head CT is being done and there is a suspicion of cervical spine injury.

2. Or, a patient having any chief complaint and one or more of the Emergency Care Data Set (ECDS) diagnoses of

a. Closed fracture of skull (disorder)

b. Open fracture of skull (disorder)

c. Crushing injury of skull and intracranial contents (disorder)

d. Traumatic brain injury with no loss of consciousness (disorder)

e. Traumatic brain injury with brief loss of consciousness (disorder)

f. Traumatic brain injury with moderate loss of consciousness (disorder)

g. Traumatic intracranial subdural hematoma (disorder)

h. Traumatic intracranial extradural hematoma (disorder)

i. Traumatic intracranial subarachnoid haemorrhage (disorder)

j. Cerebral haemorrhage following injury (disorder)

k. Moderate head injury (disorder)

l. Major head injury (disorder)

m. Diffuse brain injury (disorder)

n. Contusion of brain (disorder)

Patients with intracranial traumatic brain injury or haemorrhage are subsets based on the ECDS diagnosis. Patients who had neurosurgery are defined based on the Hospital Episode Statistics (HES) Admitted Patient Care (APC) procedure codes. Mortality within 28 days of index event will be taken from Office for National Statistics (ONS) data (Civil Registrations Deaths).

Data from the Emergency Care Data Set (ECDS), the Diagnostic Imaging Data Set, Hospital Episode Statistics Admitted Patient Care and Civil Registrations Deaths data sets are being requested. These are necessary to identify emergency department (ED) attendances, reason for attendance being head injury, diagnosis of attendance being brain injury, in the overall ED population. They are also necessary to determine whether any independent characteristics such as ethnicity, language, accommodation status or index of deprivation is associated with head or brain injury. DIDS is necessary to determine ED patients who had a CT scan of the head or head and cervical spine. HES APC is necessary to determine those patients who had neurosurgery. Civil Registration Deaths is necessary to identify patients who died and whether it was within 28 days of the index event.

All ECDS records are requested for several reasons. 1) In order to describe the population incidence of attendance to the ED with head injury amongst the population of England. 2) To describe the population incidence of TBI amongst the population of England. 3) To describe the incidence of head injury amongst the population of people that attend EDs in England. 4) To identify the association (if any) with TBI of independent variables including age, sex, ethnicity, language, accommodation and index of multiple deprivation. 5) Because an understanding of the overall population attending, specifically with respect to characteristics of patients that could be associated with head or brain injury and used in some of the mathematical analysis (namely age, sex, ethnicity, language, accommodation, geographical region, and deprivation measure) will be required to enable readers of the results to make comparisons with characteristics of patients in their area, or that attend their hospitals. 6) Because part of the mathematical analysis (also called statistical modelling), will employ a technique called hurdle modelling. This uses characteristics from the overall population (including those patients that do not have head injury) to create a final mathematical model that predicts brain injury in patients with head injury.

No identifiable data is being requested from NHS England. 16 months of data has been requested, which includes one year of analysis (2019) and two months on either side. This will account for the prevalent pool effect, which is a phenomenon that occurs when an episode could be either a first or a follow up episode. For instance, a patient attending an ED with a head injury in January could be attending because they had just suffered a head injury, or because they were returning with ongoing symptoms following a head injury in the preceding December. Without including ‘buffer months’ on either side of the study period, the returned patients in January, whose index event is outside the study period, and those that have a head injury in the December within the study period and return in the following January (outside the study period), could not be identified. These follow up attendances are important for an accurate understanding of the population incidence and for health care planning. Data for all of England has been requested. There are no alternative less intrusive ways of performing this research.

Data has been minimised as far as possible. Only data directly relevant to the study is requested. No personal identifiers are requested, the minimum number of datasets (four) are included and for a single year period of analysis (plus two months each side). A pseudonymised identifier is required to link the four requested databases. A single year of data (plus two months each side) is necessary to accurately report the population incidence of the defined outcomes. The data for all of England has been requested because this study is relevant to all of England and because geographical variation in head and traumatic brain injury rates is critical for health care planning. Since people of all ages sustain head injuries, it is not possible to filter by age. Consequently, age but not date of birth is requested. The clinical procedure is filtered by a specific Operating Procedure Codes (OPCS). All patients’ emergency department episodes for the study period will be requested because patients reattend with head injury, either because of a repeated head injury or ongoing symptoms related to the initial injury (index event). Because within the study there is a predefined two-month period after which a repeat attendance is not likely to be due to the index event, to avoid the prevalent pool effect, two months prior to the analysis period and two months after the analysis period are included in the data request. The number of fields within ECDS, HES APC, DIDS and ONS mortality has been minimised to include only those relevant to answering the study question.

The Royal London Hospital Emergency Department Research Team Patient and Public Involvement (PPI) group consists of six members. The study team shared the study protocol with the PPI group and asked for their views and feedback. The study protocol was then updated based on the PPI group feedback. The PPI group act in an advisory role only and Barts Health NHS Trust make the final decision regarding any changes/updates to the study (as the data controller). In addition to the local PPI group, the charity Headway has been invited to be represented on the study committee.

The sole data controller who also process the data is Barts Health NHS Trust. This work is funded from the Principle Investigator account of the Chief Investigator, held by the Barts Health NHS Trust Joint Research Methods Office. There are no other funders/commissioners.

The study committee members' organisations are Barts Health NHS Trust (Royal London Hospital), Queen Mary University London, Hospital Sulpetrie (Paris, France), Lancashire Teaching Hospital NHS Foundation Trust, Addenbrookes Hospital NHS Trust and London School of Hygiene and Tropical Medicine. Apart from Barts Health NHS Trust (the data controller who determines the purpose for processing), members of other organisations are only acting in an advisory capacity and will not have access to the NHS England data. The Information Commissioner’s Office “Are we a joint controller?” checklist was used (available at https://ico.org.uk/for-organisations/guide-to-data-protection/guide-to-the-general-data-protection-regulation-gdpr/key-definitions/controllers-and-processors/) to determine that none of the members on the committee are joint data controllers.

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 i.e. employees, agents and contractors of the data recipient who may have access to that data).

There will be no data linkage undertaken with NHS England data provided under this agreement that is not already noted in this agreement.

Data from the Emergency Care Data Set (ECDS), the Diagnostic Imaging Data Set (DIDS), Hospital Episode Statistics Admitted Patient Care (HES APC), and Civil Registrations Deaths is being requested. Each data set will be at patient level with each patient having a pseudonymised identifier. The datasets will be aggregated at record level based on the pseudonymised identifier as part of processing within the Barts Health NHS Trust environment.

Data flows:

- NHS England will filter the requested datasets according to criteria explained in section 5a (Objective for Processing) of the agreement.

- NHS England will then upload the data to the Secure Electronic File Transfer (SEFT) account.

- Barts Health NHS Trust will then download the data to the Trust's secure network drive.

- Data will then be analysed to produce the required outcomes of the study.

No linkage to any other data other than those requested in this application is planned.

No matching of data to publicly available data sets is planned.

No attempt to re-identify individuals will be made.

Data processing will only be carried out by substantive employees of Barts Health NHS Trust who will have been trained in data protection and confidentiality (this includes the PhD students who are members of the study committee as they are also substantively employed by the data controller). All individuals processing data will have completed the Office for National Statistics Safe Researcher Course. The data will be stored in a password protected drive within a networked drive of Barts Health NHS Trust. System access will be granted by the Barts Health NHS Trust IT team. A monthly log of data processing episodes, and a final total log of all episodes will be generated.

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

The final results will be submitted to a high impact factor peer reviewed journal such as The Lancet, and be presented at national and international conferences including but not limited to The International Conference on Emergency Medicine and The Royal College of Emergency Medicine Scientific Conference. The output is expected to be cited frequently.

The study expert committee includes members of influential national bodies including the committee that manages the National Institute for Health and Care Excellence guidelines for the management of Head Injury. The results of this study are likely to influence future iterations of this guideline, which is the basis for head injury management in all Emergency Departments in England and Wales.

Publicity following from the study results, and dissemination of the results outside of the scientific community, will be managed by the Barts Health NHS Trust communications department, and streams will include social media and news reports. The social media presence is yet to be created.

The outcomes of the study (containing aggregated data with small numbers suppressed in line HES Analysis guidance) will be shared with The Royal London Hospital Emergency Department Research Team Patient and Public Involvement (PPI) group and their feedback obtained.

Specialist charity groups e.g. Headway will be contacted and given copies of the study results.

Once the data has been made available by NHS England, the study team anticipate a requirement of six months to process and analyse, and a further six months to write up and submit for publication.

Estimates of the population incidence of head injury are based on data from fifty years ago that there are 1.4 million patients sustaining a head injury. Data on population incidence of mild traumatic brain injury (TBI) are even less reliable. However, it is thought to be common, and the consequences of even mild TBI are thought to be devastating with as many as 50% of patients with mild TBI having ongoing symptoms a year after injury and 10% who were in work prior to the injury still unable to return to work at a year.

Commissioners and acute care trusts may, for the first time, have information on local head and traumatic brain injury patterns, which may inform TBI services and the creation of head injury clinics. Independent factors such as age, sex, ethnicity, housing status and deprivation will be tested for association with head and traumatic brain injury. Understanding the association between characteristic properties of the population and a disease process is key for designing public health measures to reduce the disease burden.

In addition, this research hopes to address and support several recommendations identified in a recent call for a concerted effort to tackle the health problem in England posed by traumatic brain injury (Maas, Menon et al., The Lancet Neurology, 2017). Several recommendations were set including performing rigorous epidemiological studies to capture the changing patterns of epidemiology and to identify high-risk groups and key targets for improved prevention and management of TBI. This research is designed to address this recommendation, by defining a priori potential high-risk groups, based on age, sex, ethnicity, language, accommodation status (homelessness), economic deprivation, and geographical area. The policy makers and clinical community hope to then use the limited resources that are allocated to TBI education, prevention and treatment to focus any impact on groups or regions that are found to be high risk.

A further recommendation was to standardise epidemiological monitoring of TBI to allow accurate measurement of incidence, prevalence, and mortality, and comparison of rates of access to community, hospital, and residential care. The UK healthcare system is near unique in that it is a single payer system with multiple centrally held records. The methodology employed in this research, namely defining five head injury outcomes, and utilising nationally held pseudonymised patient level records, may be beneficial for future research due to its simple and replicable methodology. The study protocol will be made available for free in an open access environment for replication.

TBI in specific populations was explicitly named as a research priority. Defined populations include children, adolescents, and the elderly. Because this research will stratify by age, the rates of the five outcomes in this research will be specifically measured within these populations. Describing TBI by mechanisms and cause of injury is also a research priority and is part of the descriptive part of the analysis plan in this research. This may be beneficial to determine incidence and outcomes for the defined population and may inform policy changes, clinical care improvements and further research.

Effective strategies for TBI prevention are urgently needed and hope to deliver cost savings that help to fund research and improved access to health care for TBI. By identifying high risk groups, this research will, in addition to answering the core research question, be hypothesis generating and identify further areas of epidemiological and public health research that may be best placed to reduce the impact of head injuries and TBI.

Furthermore, outcomes that quantify the overall burden of disability from TBI need to be developed. Understanding the proportion of patients who develop high severity outcomes such as surgery or death following TBI is a critical contributor to quantifying disability. This research is uniquely placed to answer that question.

No yielded benefits have been achieved so far as there was missing data fields that was not requested in the previous iteration which was required in order to complete the analysis.