Theme 3: Generating and Implementing Innovation

Randomised trials

Project 1, 2 and 3 are interventional trials. The applicants from the University of Cambridge and several LMICs have significant experience in undertaking randomised trials. We recently finished recruitment for the NIHR-funded RESCUE-ASDH trial which assessed two surgical techniques for patients with acute subdural haematomas. RESCUE-ASDH had significant participation from many LMICs (the highest recruiting site was NIMHANS, Bangalore) and we learnt lessons on how to successfully complete large randomised studies and ensure good follow up in LMICs. Theme 3 will also leverage close links with the NIHR Brain Injury MedTech Co-operative (​​).

PROJECT 1 ​- ​What is more effective at reducing intracranial pressure in traumatic brain injury – mannitol or hypertonic saline?

Project Lead:

Background: ​Post-traumatic intracranial hypertension is very common and if not treated promptly, can cause brain ischemia, stroke and ultimately death. ​Hyperventilation, use of hyperosmolar agents and the evacuation of intracranial mass lesions are common approaches in reducing the intracranial pressure (ICP). Currently, mannitol (20%) is considered the gold standard hyperosmolar agent and is more widely used but is known to cause lethal adverse effects such systemic hypovolemia and hypotension and severe electrolyte derangements. Hypertonic saline (3%) is another agent that is being increasingly used in emergency and intensive care settings in patients with TBI. It is considered as effective as mannitol at reducing ICP and does not have significant side effects ​(22)​. However, there is a lack of high-quality randomised trials examining its efficacy.

Aim​: To determine whether hyperosmolar therapy regimes involving hypertonic saline (HTS) or mannitol improve neurological function at 3, 6 and 12 months (as assessed by the GOS-E questionnaire) following severe TBI in LMICs. Secondary aims are to assess the effects of hyperosmolar therapy with either HTS or mannitol on clinical, patient-centred outcomes in the emergency room and ICU, in hospital and up to 12 months follow-up post randomisation.

Study design: ​The trial is an allocation concealed, open label, randomised controlled clinical trial with an internal pilot and blinded assessment of primary outcome at 3, 6 and 12 months. The main trial will be preceded by a six-month internal pilot study to test recruitment feasibility and compliance with assigned treatment.

Inclusion criteria​: >18 years old, have been admitted to an emergency room (ER) with moderate/severe TBI. ​Exclusion criteria​: devastating brain injury with withdrawal of life sustaining treatment anticipated in the next 24 hours, pregnancy, severe hypernatremia (serum Na > 155 mmol/L) in ER.

Study setting​: The main trial will take place in approximately 10 hospitals in Colombia, Brazil and India with ER and ICUs that routinely manage TBI patients. The participation of Tanzania in the pilot is currently being explored. Sites will need to establish experience with receiving and acting on protocolised advice in relation to the administration of hyperosmolar therapy. This will be addressed with a ‘run-in’ period, while the pilot phase is ongoing with all sites having access to an educational package.  Trial interventions​: Patients will be randomly assigned to boluses of either: (1) 2ml/kg 20% mannitol intravenous bolus (osmolarity 1026 mOsm/L) – or equivalent osmolar dose using the concentration used locally by participating study centres, (2) 2 ml/kg 3% HTS intravenous bolus (osmolarity 1110 mOsm/L) – or equivalent osmolar dose using the concentration used locally by participating study centres. One or two doses will be administered prior to surgery in the emergency room and continued 1 to 4 doses per day for the first 4-5 days in the ICU

Primary outcome measure​: Extended Glasgow Outcome Scale (GOS-E) measured at 6 and 12 months after randomisation. ​Secondary outcome measures​: These will focus on quality of life, short-term outcomes, complications, use of resources, and economic analysis.

Sample size​: In order to detect a treatment effect of 13% (63.5% to 50.5%) in the proportion of patients having an unfavourable neurological outcome (GOS-E: dead, vegetative state, lower severe disability, upper severe disability) compared to favourable outcome (GOS-E: lower moderate disability, upper moderate disability, lower good recovery, upper good recovery) between mannitol and HTS, 250 patients will be required in each arm (total planned size is 500 patients). This is based on a superiority hypothesis of a difference between the two interventions, using 80% power, two-sided significance level of 5% with a dropout / withdrawal rate of up to 10%.

Impact​: This will be an important trial that will be delivered in LMICs and its results will be complementary to the results of the NIHR-funded SoS UK-wide multicentre randomised trial of osmotherapy (​​). The two trials together are expected to define the exact role of osmotherapy for TBI.

PROJECT 2​ – ​When should blood be transfused to patients with traumatic brain injury?

Project Lead:

Background: Anaemia is frequent among patients with TBI and it is associated with an increased risk of poor outcomes. The optimal haemoglobin (Hb) level to trigger red blood cell (RBC) transfusions in TBI patients is not clearly defined. This is the substantive phase of a previously published external pilot trial (44 patients randomised at the ​University of São Paulo Faculty of Medicine ​ Clinics Hospital​ ). The pilot trial reached feasibility criteria and confirmed that a substantive trial is necessary and feasible ​(23)​.

Aim​: To compare the effects of two Hb transfusion thresholds (7 and 9 g/dL) on neurological recovery after traumatic brain injury.

Study design​: Multicentre, randomised, open-label clinical trial, with blinded outcomes assessment. Inclusion criteria: ​Adult patients with moderate or severe TBI (hospital admission Glasgow coma score ≤ 12) admitted to a participating intensive care unit (ICU) in Brazil and with a Hb concentration less than 9 g/dL within 7 days from trauma.

Exclusion criteria​: Patients with a GCS of 3; bilateral dilated pupils at admission; previous known neurological sequelae; pregnant; Jehovah’s witnesses; haemorrhagic shock at randomisation; or deemed moribund due to other causes.

Interventions: ​The enrolled patients will be randomised to a restrictive Hb transfusion threshold group (7g/dL) or to a liberal group (9g/dL). Haemoglobin concentrations will be assessed at least daily, and the patients will be randomised when Hb reaches 9 g/dL. The patients will be given single units of cross-matched, red blood cells (RBC) packs (recommended 1 pack). Hb concentrations will be reassessed within 3 hours after termination of the transfusion and/or before the initiation of another transfusion. The allocated transfusion strategy will be maintained for up to 14 days or to ICU discharge, whatever comes first. All other interventions will be at discretion of the patient assistant team, and will not be influenced by the trial investigators, including transfusion before ICU admission or during surgery.

Primary outcome measure​: Based on the extended version of the Glasgow outcome scale (GOS-E) at 6 months after randomisation, there will be two co-primary outcomes analyses: ordinal (based on the proportional-odds model) and dichotomous (favourable or not, defined as lower moderate disability or better). ​

Secondary outcome measures​: GOS-E at hospital discharge, transfusion requirements, intracranial pressure (ICP) occurrence and management resources use, transcranial doppler (TCD) parameters, head CT evolution (new or worsening ischemia and progressive haemorrhagic injury), hospital and ICU length of stay, duration of mechanical ventilation, adverse events, and mortality.

Sample size​: The sample size calculation was based on the results of the TRAHT feasibility trial​3​, in which the restrictive group exhibited 44% favourable outcomes (versus 62% for the liberal group). A planned sample size of 260 patients would provide 80% power with an alpha level of 0.05 for the estimated risk reduction (incorporating up to 10% loss to follow-up).

Impact​: Packed red blood cells are a precious resource and this trial will define the optimal strategy of managing anaemia following TBI. The pilot phase has been completed and published and collaborating centres in Brazil have been identified.

PROJECT 3 ​- ​For patients who have suffered a traumatic brain injury or stroke and need an operation, there are two techniques commonly used. Which one is more effective?

Project Lead:

Background: ​A significant proportion of patients with TBI or stroke require an operation to control brain swelling and remove a haematoma if present. There are 2 operations that can be performed to achieve this: a Decompressive cran​io​tomy (DCO) or a Decompressive cra​nie​ctomy (DC). A DCO involves opening the skin, removing a piece of skull, removing a clot if present, replacing the bone back but not rigidly fixing it to the surrounding skull and finally closing the skin. A DC is similar but the piece of skull is left out prior to closing the skin. Patients having a DC will require a 2nd operation at a later stage to rebuild their skull. A scoping review published by the existing Group provides evidence that the outcomes after DC and DCO are comparable ​(24)​. We have also undertaken a global survey of 208 surgeons worldwide which showed that DCO is an operation performed often, especially in LMICs. However, there are no randomised trials in existence comparing DC and DCO.

Aim: This randomised trial aims to compare 2 different surgical techniques for patients with TBI or stroke that are candidates for decompression due to brain swelling and/or intracranial haematoma.

Trial design: ​RESCUE CD is a multi-centre, pragmatic, parallel group, superiority randomised trial. It will commence with a pilot (6 months) and then subsequently progress to the substantive phase, subject to meeting specific progression criteria. Patients will be randomly allocated to DCO or DC.

Inclusion criteria​: Adult patients (aged >16 years) with a TBI or stroke and the admitting neurosurgeon feels that the patient requires a surgical decompression for the management of TBI or stroke either by a decompressive ​craniotomy ​ (DCO) or a decompressive ​craniectomy (DC) (recommended bone flap ≥ 11 cm anteroposterior length).

Exclusion criteria​: Bilateral fixed and dilated pupils and/or brainstem injuries on CT, uncorrected coagulopathy, severe pre-existing disability.

Recruitment and setting​: We aim to recruit 570 patients (285 in each arm; 10% loss to follow-up rate) in order to detect an 14% absolute difference in the rate of favourable outcome (defined as lower moderate disability, upper moderate disability, lower good recovery or upper good recovery on the GOSE) at 6 months. The duration of the trial is 60 months and will be undertaken in approximately 25 sites based in Brazil, Cameroon, Colombia, Ethiopia, India, and Pakistan.

Primary outcome measure​: GOSE (extended Glasgow Outcome Scale) at 6 months after randomisation.

Secondary outcome measures​: These will focus on quality of life, short-term outcomes, complications, use of resources, and economic analysis.

Sample size: ​A 14% increase in the rate of favourable outcome (defined as lower moderate disability, upper moderate disability, lower good recovery or upper good recovery on the GOSE) at 6 months is considered a clinically important treatment effect on the basis of our survey. Assuming a favourable outcome rate of 35% in the control group and using a 2-sided test at the 5% significance level, a sample size of 570 patients (allowing for a 10% loss to follow-up) will enable us to detect an absolute difference of 14% in the proportion of participants with a favourable outcome with a power of 90% [35% vs 49%; 90% power and 2-sided significance 0.05].

Impact​: A randomised trial showing that DCO is superior or even comparable to DC would have huge economic advantages and benefits in terms of optimising healthcare resource usage in the countries where there is the greatest need to do so.

PROJECT 4 – ​Can we create a rapidly deliverable source of guidance to centres with limited resources treating patients with traumatic brain injury?

Project Lead:

Background​: Recently investigators based in Cali, Colombia created a stratified protocol for TBI treatment that provides a range of options that (a) accommodate varying available resources, and (b) fill the gaps in evidence-based guidelines using a consensus of clinical expertise. The protocol, B​eyond One O​ption for ​T​reatment of Traumatic Brain Injury: A ​Stra​tified ​Protocol [​BOOTStraP​] provides step-by-step treatment options spanning low, intermediate, and high resource settings and across the four acute treatment phases of field transport, emergency department (ED), neurosurgery, and intensive care unit (ICU). If effective, BOOTStraP could provide an inexpensive and rapidly-deliverable resource of information throughout LMICs and underserved areas of HICs where expertise and resources are limited. We have initiated a pilot project to test the feasibility of collecting data about TBI patients and implementing the BOOTStraP protocol across the treatment phases of prehospital, ED, neurosurgery, and ICU, in resource-limited environments. The project is currently operating in two centres in Colombia. The pilot, which is part of the extension of the current Group, intends to lay the groundwork ​for this proposed larger study that can validate feasibility findings and will have the power to detect a statistically significant influence of protocol adherence on patient outcomes.

Aim​: (1) Finalise the pilot database, consistent with the NINDS Common Data Elements (CDEs), which includes a set of indicators of available resources, patient and injury characteristics, adherence to BOOTStraP-indicated treatments, and patient outcomes. (2) Conduct a prospective, observational study of resources, treatments, mortality and morbidity for patients with severe TBI from field transport to 6 months post-injury, in 4 centres in Colombia. Aim 2(a): Describe resources and treatments, mortality and morbidity. Aim 2(b): Estimate level of adherence to BOOTStraP overall and examine consistency in adherence to BOOTStraP across field transport, ED, neurosurgery, and ICU phases. Aim 2(c): Evaluate the relationship between resource availability and adherence to BOOTStraP. (3) Compare the effectiveness of treatments, and adherence to BOOTStrap, on outcomes of mortality and neurological morbidity at 6 months post-injury.

Design​: This will be a prospective, observational study in which we will collect data over 27 months about patients with severe TBI received in one of 4 study hospitals in Colombia. The analytic design is comparative effectiveness.

Setting​: Four hospitals in Colombia; one high resource, one low resource, and two medium resource centres.

Inclusion criteria​: Indications of TBI, Glasgow Coma Scale Score (GCS) <= 8 or GCS motor<= 5 if verbal or eyes cannot be assessed), on admission or within the first 48 hours after injury, admission to study hospital within 24 hours of injury, all ages included.

Sample size​: Average of 6 severe TBI patients per month per center for 27 months = 648. Loss-to-follow-up of 15%,= 550 patients.

Treatments​: Treatments are provided as algorithms for options according to the availability or lack of specific resources (accessed via ​​).

Analysis​: The association of each treatment indicator and outcomes will be examined using multiple analytic approaches to correct for potential confounding, including: 1) multivariable models; 2) propensity score methods, and 3) instrumental variables. All analyses will directly adjust for the centre or account for clustering of patients within centres. All statistical tests will be 2-sided with significance evaluated at the 5% level.

PROJECT 5 ​- ​Can antibiotic impregnated ventriculoperitoneal shunts reduce infection rates?

Project Lead:

Background: ​Tikur Anbessa University Hospital in Ethiopia and its affiliated hospitals at Addis Ababa perform about 300 ventriculoperitoneal shunt insertions each year. Published data have shown that a shunt infection affects 24% of patients ​(25)​. Shunt infections are associated with prolonged hospital treatment, multiple operations, and reduced cognition and quality of life for patients. The economic implications of shunt infections for the patients, their families, and the society are also substantial. Local clinicians are concerned about the high shunt infection rate and are interested in the implementation of antibiotic-impregnated shunts as a possible solution, given that there is now ​class I evidence about their use from the NIHR-funded BASICS multi-centre randomised trial ​ (26)​ . This trial showed that the use of antibiotic-impregnated shunts reduced the risk of infection by 60% (relative risk reduction) and can save £135,753 per shunt infection avoided.

Aim​: We aim to perform a before–after intervention cohort study at the Tikur Anbessa University Hospital and affiliated hospitals in Addis Ababa, Ethiopia in order to examine whether (i) the use of antibiotic-impregnated shunts is feasible and (ii) reduces the risk of shunt infection. We hypothesise that the use of antibiotic-impregnated shunts will lead to a reduction in the risk of shunt infection. We also wish to assess the sustainability of using antibiotic-impregnated shunts in the long-term.

Methods​: This will be a before–after intervention cohort study at the Tikur Anbessa University Hospital and affiliated hospitals in Addis Ababa, Ethiopia with prospective assessment of outcomes. The study will comprise 4 periods: baseline (12 months), including the start of formal shunt infection surveillance with establishment of a dedicated study database; intervention (9 months), consisting of the rollout of the use of antibiotic-impregnated shunts in all participating hospitals; follow-up (6 months), representing the first evaluation period of the effect of the intervention; and sustainability (15 months; including 6 months follow-up), representing the long-term follow-up when the antibiotic-impregnated shunts will likely have become part of the regular process of care.

Setting​: The study will include patients of any age who have hydrocephalus of any aetiology requiring a first ventriculoperitoneal shunt at the Tikur Anbessa University Hospital and its affiliated hospitals. Patients will be excluded if they have evidence of active and ongoing CSF or peritoneal infection; a previous indwelling ventriculoperitoneal shunt; multiloculated hydrocephalus requiring multiple shunts or neuroendoscopy; or a known allergy to rifampicin, clindamycin.

Data collection​: Data will be collected at baseline (around the time of first shunt operation), any routine postoperative assessments, and, where applicable, at unscheduled visits and admissions, and at shunt revision and removal (if required).

Data analysis/sample size​: With an estimated baseline risk of shunt infection at 24%, based on published data, we estimate that infection will be reduced by 10% (absolute risk reduction) between the before and after intervention periods. This is more conservative compared to the relative risk reduction observed in the BASICS randomised trial. Assuming an 80% statistical power at a 5% level of significance, an expectation of a 1:2 size ratio between time periods, and a 10% drop-out, we aim to include 645 operations in total (215 in baseline phase, 215 in intervention phase, and 215 in sustainability phase). We will use a logistic regression model with mixed-effects to assess the effect of using antibiotic-impregnated shunt on outcomes.

Primary outcome measure​: The primary outcome will be shunt failure due to infection at 6 months after shunt insertion. The secondary outcomes will include adverse events and complications; quality of life using the EQ-5D-3L, EQ-5D-3L (proxy), or EQ-5D-3L-Y versions of the EQ-5D health-related quality-of-life questionnaire; and the proportion of patients receiving an antibiotic-impregnated shunt during each study period. We will also explore the possibility of undertaking an economic evaluation (e.g. incremental cost per first shunt infection averted for impregnated and standard shunts).

Impact​: Throughout the study, we will engage with key stakeholders in order to establish a sustainable model for procurement of antibiotic-impregnated shunts, at least for the very high-risk patient groups (e.g. children). To that end, we have already engaged in discussions with manufacturers, who appear keen to collaborate on this project with a view to achieving sustainable procurement.

PROJECT 6 – ​Can ophthalmoscopy using a simple mobile device and an app be used to automatically predict patients who may have raised intracranial pressure?

Project Lead: Brandon G. Smith (bgs30(at)

Background​: Direct ophthalmoscopy is an indispensable diagnostic tool to evaluate adults and children with suspected intracranial hypertension. Optic nerve oedema has been well defined as an objective and important sign suggesting central nervous system pathology requiring urgent evaluation with brain imaging methods ​(27)​. Nevertheless, direct ophthalmoscopy is not always available in rural settings or in places where healthcare screening may be done by non-medical personnel. Furthermore, many individuals within resource limited settings do not have free access to brain imaging leading to significant financial burden on families with individuals who are suspected of harbouring intracranial pathologies. Aim​: To develop a simple mobile device to allow direct ophthalmoscopy in non-hospital settings with automated, artificial intelligence based, detection of papilloedema.

Methods​: We will prospectively include patients from the Ophthalmology and Paediatric Departments at the participating centres in Zambia and Ethiopia who are being evaluated for optic disc pathology for any reason. We aim to prospectively collect fundus imaging using various commercially available mobile devices as well as commercially available fundus cameras. Imaging will be reviewed by an ophthalmologist and labelled as papilloedema or no papilloedema. The first part of the study will be used to collect a training dataset of images to develop a deep learning algorithm for automated detection of papilloedema. This will be followed by a validation dataset which will be used for confirming the accuracy of the deep learning model.

Impact​: We aim to use the algorithm alongside mobile phone technology and mobile appliances which can work with optical cameras both in the peripheral as well as hospital/clinic settings. Automated diagnosis of papilloedema could be used to guide management and stratify the need for brain imaging in adult and paediatric patients suspected of intracranial hypertension. An automated papilloedema detection technology could also be used by non-medical personnel conducting health visits in the community allowing for wider screening potential for the local populations without direct access to ophthalmology clinics and/or brain imaging.

PROJECT 7​ – ​Can patient follow up be improved by using SMS-based health questionnaires?

Project Lead: Brandon G. Smith (bgs30(at)

Background​: To fully understand the full burden of TBI, the disability and quality of life of survivors must be fully evaluated. As a whole, long-term outcome data (beyond discharge from acute care) is generally not reliably recorded in  LMICs. There is a need to develop more systematic, diligent processes to follow up TBI patients across the spectrum of injury. This project encompasses the development of a follow-up technology based upon short-message service (SMS). According to the Global System for Mobile Communications Association (GSMA), SMS technology has seen rapid and worldwide growth, with just over 5.2 billion unique mobile subscribers as of June 2020 ​(28)​, representing 68% of the world’s population with respect to the United Nations 2019 estimates ​(29)​. According to 2018 data provided by the World Bank, LICs have a value of 66 mobile cellular subscriptions per 100 people, while MICs 106 subscriptions per 100 people ​(30)​. In respect of this coverage, SMS technology was deemed most appropriate for an early prototype to adopt. The Patient Response Interface through SMS (PRISMS) platform will function to deliver a text-message survey based on the existing gold-standard measure for traumatic brain injury, the Glasgow Outcome Scale Extended (GOS-E), henceforth referred to as (GOS-E-SMS). Each survey is designed to be delivered through a timed sequence of text messages to the personal devices of patients, identifying the need for any additional in-person follow-up assessments or to expedite assessments the timing of assessments found in the more traditional follow-up pathway of outpatient clinics at 3, 6, 12 months and beyond.

Aim: ​To examine if clinicians and service users perceive that SMS can be used to follow up TBI patients (utilising the telehealth usability questionnaire (TUQ) ​(31)​. We will also aim to validate GOS-E-SMS against GOS-E.

Study design: ​Service evaluation of a SMS-based follow-up technology over 12 months, including a preliminary pilot study for validation of the SMS GOS-E.

Recruitment and setting​: We aim to achieve a convenience sample of 50 patients. Patients will be enrolled on a rolling basis upon their first outpatient clinic visit, typically at 1-month post-injury.

Primary outcome measure: ​GOS-E-SMS validation and telehealth usability questionnaire results at 3, 6 and 12 months from the patients enrolled in the service evaluation, and the clinical team using the PRISMS platform. ​

Secondary outcome measures​: fortnightly GOS-E scores collected by SMS, coinciding with the in-person GOS-E delivered at 1, 3, 6 and 12 months delivered in clinic.