Risk Adjustment In Neurocritical care (RAIN)

Risk Adjustment In Neurocritical care (RAIN) Understanding recovery from head injury to inform care NIHR HTA grant: 07/37/29 NIHR CRN Portfolio Study ID: 7349

Scope RAIN Study Idea to research grant

Risk Adjustment In Neurocritical Care RAIN First idea (2001) Adopted by NCCNet (March 2007) Health Technology Assessment (HTA) themed call in emergency medicine Expression of interest (May 2007) Outline proposal (October 2007) Full proposal (February 2008) Response to reviewers (May 2008) Final full proposal (November 2008) Contract signed (December 2008) National Institute for Health Research (NIHR) HTA grant number: 07/37/29

Scope RAIN Study Idea to research grant Aim/objectives/phases

Aim To provide an accurate risk prediction model for TBI to aid clinicians to evaluate clinical outcome(s) effectiveness of care whether, and best time, to transfer cost-effectiveness of care

Objectives of RAIN To validate risk prediction models for acute TBI To evaluate the optimum location and comparative costs of care Why? TBI leading cause of death and disability worldwide Major cause of death/disabilitydisability in those aged under 40

RAIN phases Study preparation March 2009 to August 2009 (six months) Collect/validate data September 2009 to July 2011 (23 months) Validate models March 2011 to October 2011 (eight months) Cost-effectiveness analysis September 2010 to December 2011 (16 months) Report writing and dissemination September 2010 to February 2012 (18 months)

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure

Investigators Role: design and conduct of RAIN Chief Investigator Dr David Harrison Senior statistician, ICNARC Co-investigators Prof David Menon Dr Martin Smith Neurocritical care Dr Paddy Yeoman Mr Peter Hutchinson Neurosurgery Dr Fiona Lecky Emergency medicine Dr Richard Grieve Health economics Prof Kathy Rowan Director, ICNARC

Study Steering Group Role: overall supervision of RAIN on behalf of the funder (HTA) and sponsor (ICNARC) Investigators plus independent members Prof Monty Mythen (Chair) Critical care Mr Jonathan Hyam Neurosurgery Dr Ian Tweedie Neuroanaesthesia Ms Julie Bridgewater Lay representative (Headway) And Dr Gita Prabhu Study Coordinator Research fellow to be appointed

Study Management Group Role: day-to-day running of RAIN Emma Walmsley (Research Administrator) David Harrison (Chief Investigator) Gita Prabhu (Study Coordinator) Phil Restarick(Research Coordinator) Kathy Rowan (Co-investigator) All based at ICNARC

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation

RAIN (NIHR HTA: 07/37/29) Appointments: Study Coordinator (Gita Prabhu) December 2008 Study Administrator (Andrew Craven) February 2009 Research Coordinator (Phil Restarick) April 2009

RAIN (NIHR HTA: 07/37/29) Ethics approvals (March 2009) England, Wales and Northern Ireland (09/MRE09/10) Scotland (09/MRE00/15) National Information Governance Board for Health and Social Care (NIGB) (August 2009) (ECC 2-06(d)/2009) Main R&D approved (August 2009) Site specific R&D approvals (ongoing) Portfolio adoption (August 2009) Final recruitment of units (ongoing)

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation Local funding CLRN

Comprehensive Local Research Network (CLRN) funding England has been divided into 25 regions, each of these is a CLRN (see next slide) Each CLRN receives a budget to manage and deliver research for its region Many CLRNs have critical care leads (see next slide) RAIN is a NIHR CRN Portfolio study (Study ID: 7349) so units taking part in the study in England are eligible to apply for funding for NHS support costs through their CLRN

Comprehensive Local Research Network (CLRN) funding Even if there is no critical care lead in your CLRN, you can still apply for funding RAIN fully adopted onto NIHR CRN Portfolio following approval of main R&D For further information on CLRN funding: www.ukcrn.org.uk

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation Local funding CLRN Literature review to inform dataset

Systematic literature review Update to Cochrane Injuries Group systematic review (Perel et al. 2006) Over 1800 abstracts reviewed for eligibility Three candidate risk prediction models selected as developed on large multicentre databases predicting six-month unfavourable outcome have some degree of external validation

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation Local funding CLRN Literature review to inform dataset Risk prediction models

J Neurotrauma 2005; 22:1025-39. BMJ 2008; 336:425-9. PLoS Med 2008; 5:e165.

Hukkelhoven et al Data two multicentre RCTs (n=2269) Variables in model Age, GCS motor score, pupil reactivity, hypoxia, hypotension, Marshall CT class, tsah Validation European Brain Injury Consortium (n=796)

CRASH Trial Collaborators Data one multicentre RCTs (n=2482 high income) Variables in model Basic: Age, GCS, pupil reactivity, major extracranial injury CT: + five specified CT findings Validation Slightly modified models validated in IMPACT dataset (n=8509)

Steyerberg et al (IMPACT) Data Eight RCTs and three observational studies (n=8509) Variables in model Core: Age, GCS motor score, pupil reactivity Extended: + hypoxia, hypotension, Marshall CT class, tsah, extradural haematoma Lab: + glucose, haemoglobin Validation Core model and simplified extended model validated in CRASH dataset (n=6681)

Limitations of the models Rely heavily on data from RCTs may not be representative of typical care Never validated in a critical care setting not all models validated at all Hukkelhoven and IMPACT models developed on older data latest 1997

Conclusion further prospective validation in independent cohorts is needed to strengthen the generalisability of the models MRC CRASH Trial Collaborators

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation Local funding CLRN Literature review to inform dataset Risk prediction models Dataset/data collection burden

RAIN dataset Dataset developed to include all data required for the three models plus variables to describe population with TBI To be collected in three types of units neurocritical care unit in neuro. centre general critical care unit in neuro. centre general critical care unit outside neuro. centre

Data collection burden adult TBI Neurocritical care unit in neuro. centre average 7 per unit per month range 4 to 10 per unit per month General critical care unit in neuro. centre average 5.5 per unit per month range 2.5 to 10 per unit per month General critical care unit outside neuro. centre average 0.5 per unit per month range 0 to 3 per unit per month

RAIN dataset Dataset defined in seven sections: Patient TBI pre-hospital Source TBI at hospital First CT Outcome GP

RAIN dataset - Patient Full name Full postal address Telephone number Sex NHS number Hospital number Date of birth Residence/Dependency prior to hospital

RAIN dataset TBI pre-hospital Date/Time of TBI Cause/Type of TBI Intoxication at time of TBI Major extracranial injuries Blood pressure Oxygen saturation Pupil reactivity/size Glasgow Coma Score AVPU

RAIN dataset - Source Date/Time of admission to hospital Date/Time of admission to unit Location prior to admission to the unit

RAIN dataset TBI at hospital Temperature Blood pressure Heart rate Oxygen saturation Arterial blood gas Serum glucose Haemoglobin Platelets Coagulation (PT/APTT) Pupil reactivity/size Glasgow Coma Score

RAIN dataset First CT Traumatic subarachnoid haemorrhage present Brainstem pathology present Basal cisterns Third ventricle Midline shift present Lesion(s) present Petechial haemorrhages High/mixed density lesion Evacuation of haematoma Volume of largest high/mixed density lesion

RAIN dataset - Outcome CCMDS organ support Confirmed diagnosis of TBI Spinal cord injury present Date/Time/Status on discharge from unit Level of care received at discharge from unit Expected outcome post hospital discharge Location to which the patient was discharged Date/Time of death/brainstem death Date/Status on discharge from hospital

RAIN dataset - GP GP Practice name GP name/address or Practice code

Data entry/submission Secure, web-based system For Case Mix Programme (CMP), few CMP fields duplicated to link datasets RAIN and CMP data will be linked and validated at ICNARC

Scope RAIN Study Idea to research grant Aim/objectives/phases Infrastructure Preparation Local funding CLRN Literature review to inform dataset Risk prediction models Dataset/data collection burden Why take part?

Why take part? Contribute to improving the care of TBI patients Gain insight into TBI care and have national comparative data for your unit Receive six month follow-up of your patients by the RAIN Study team

How to take part Identify a Local Collaborator (LC) who will be responsible for the Study locally Fill out the site-specific information for your Trust Attend a RAIN Dataset Familiarisation Course LC responsible for completion of: assessment of First CT assessment of expected outcome of patient Sign Research Agreement to indicate acceptance of LC responsibilities

Responsibilities of the LC The LC will be responsible for setting up a screening process to ensure that all eligible admissions (suspected/actual acute TBI) are recruited into the RAIN Study completing the First CT variables, according to the definitions, for all eligible admissions recruited to the RAIN Study completing the variable Expected outcome for all eligible admissions recruited to the RAIN Study ensuring access is available to the secure, web-based data entry system ensuring accurate and timely data collection and entry onto the secure, web-based system

Risk Adjustment In Neurocritical care (RAIN) Understanding recovery from head injury to inform care For further information about RAIN email: rain@icnarc.org