A STARSHIP analysis
Peter is a 10-year-old boy in Paediatric Intensive Care (PICU).
He was admitted 2 days ago following a serious road traffic accident in which he was knocked off his bicycle by a passing car and sustained a left-sided fronto-temporal fracture with underlying subdural haematoma. He was GCS 6 at the scene.
The neurosurgical team inserted an intracranial pressure bolt, and he has had ongoing monitoring whilst remaining muscle relaxed with rocuronium. He is well sedated and supported on noradrenaline to titrate his blood pressure.
On discussion with your consultant you have been targeting a Cerebral Perfusion Pressure of 45, as a midpoint in the Brain Trauma Foundation guidelines given his older age, but you’ve been called to his bedside frequently through the night because the nurses have found that both his intracranial pressure and his blood pressure have been extremely labile and maintaining a steady cerebral perfusion pressure has been a challenge.
The nurse asks if there is anything else she should monitor or target, and what this means for Peter’s likely outcome.
Every year in the UK, around 800 children (5.6 per 100,000) are admitted to PICU following severe traumatic brain injuries or TBI.
The most common long-term impact of severe traumatic brain injury is persistent neurocognitive, physical, behavioural and social disabilities, which can vary from mild to significantly life-changing [2,3,4].
At 5 years post-injury, less than half of children with severe TBI demonstrate good recovery, and more than half have ongoing disability affecting their independence and participation. Approximately 3-14% of children do not survive severe traumatic brain injury, with the lower end reflecting the mild to moderate injuries and the higher end reflecting the strictly severe TBI cases [4,5,6,7].
During their admission, these children may need periods of invasive monitoring of their intracranial pressure and careful physiological support to maintain adequate cerebral perfusion. For a primer on TBI management considerations, take a look at these DFTB posts.
Targeted Temperature Management in Paediatric Traumatic Brain Injury – Don’t Forget the Bubbles
Managing raised intracranial pressure in severe traumatic brain injury – the basics – Don’t Forget the Bubbles
Reproduced from ‘Cerebral perfusion pressure in neurotrauma: a review’, White and Venkatesh 2008 https://pubmed.ncbi.nlm.nih.gov/18713917/
One of the brain’s compensatory mechanisms is cerebrovascular autoregulation, which acts through a number of mechanisms, including altering arteriolar resistance to maintain cerebral blood flow across variations of arterial blood pressure (and thus variations of cerebral perfusion pressure).
These changes to the cerebrovascular resistance can mitigate the effect of changes in cerebral perfusion pressure on cerebral blood flow across the zone of normal autoregulation, though they will fail at the point where the arterioles are maximally dilated or constricted. In TBI, however, this mechanism is often impaired and becomes progressively more ineffective as the evolving injury and pathophysiological mechanisms following the primary injury lead to disruption of cerebral autoregulation. This then leads to secondary injury induced by CPP changes, causing either hypo- or hyperperfusion.
The Brain Trauma Foundation offers guidance on many aspects of post-TBI care, including a recommendation for a Cerebral Perfusion Pressure (CPP) fixed minimum threshold of 40mmHg (and a recommended threshold range of 40-50 in older children) to prevent hypoperfusion-induced secondary injury. However, they caution that this CPP target in children is currently only weakly evidenced and primarily extrapolated from adult data.
There is growing evidence in adult literature of the use of dynamic, patient-specific indices, such as the Pressure Reactivity Index (PRx), to assess cerebrovascular autoregulation and guide treatment.
Smith, C.A., Bögli, S.Y., Placek, M.M., Cabeleira, M., White, D., Daubney, E., Young, A., Beqiri, E., Kayani, R., O’Donnell, R. and Pathan, N., 2025. Dynamic versus fixed cerebral perfusion pressure targets in paediatric traumatic brain injury: a STARSHIP analysis. EClinicalMedicine, 86.
The study
The STARSHIP traumatic brain injury trial, led by S Agrawal, was a prospective, multicentre observational study that assessed the relationship between cerebrovascular autoregulation (CA), as measured by the pressure reactivity index (PRx), and outcomes in children with severe traumatic brain injury (TBI) requiring invasive monitoring.
The information was available at the bedside to the treating clinicians, but no intervention was undertaken as a result of this data, and interventions for CPP and ICP were at the treating team’s discretion in line with the current BTF guidelines. The primary STARSHIP study revealed that impaired cerebrovascular autoregulation, indicated by an elevated PRx, is independently associated with higher mortality and unfavourable neurological outcomes at 12 months. Specifically, a PRx threshold of 0.5 was associated with increased mortality, and a threshold of 0.0 was associated with unfavourable outcomes.
STARSHIP Paper 1 Brief Summary:
Impaired cerebrovascular autoregulation (elevated PRx) is independently associated with higher mortality and unfavourable neurological outcomes at 12 months. PRx threshold of 0.5 was associated with increased mortality, and a threshold of 0.0 was associated with unfavourable outcomes.
This paper is a secondary analysis of the STARSHIP trial cohort, focusing on interpreting the PRx thresholds and neurological outcomes in relation to two other measures of autoregulatory function:
- Optimal CPP (CPPopt)
- Lower Limit of Autoregulation (LLA)
The investigators used Dynamic High Frequency Monitoring of intracranial pressure (ICP) and arterial blood pressure (ABP) using the ICM+ software. These were then used to calculate PRx. We’ll explore the two derived values, Optimal CPP (CPPopt) and the lower limit of autoregulation (LLA), later. They are calculated dynamically using the patient’s recent PRx data to interrogate their autoregulatory function.
What is PRx?
Pressure Reactivity Index (PRx) is a widely used measure in adult neurocritical care which assesses the state of cerebrovascular autoregulation. It is a correlation coefficient that associates the relationship between slow changes in MAP and ICP, using 5 minutes of preceding data in 10-second averages, and is updated every minute to provide a dynamic assessment of cerebrovascular autoregulation. This mitigates the impacts of the faster respiratory and cardiac frequency-based changes and uses ICP as a cerebral blood volume surrogate.
When Cerebral Autoregulation is intact, the cerebral vascular resistance changes to preserve CBF, counteracting changes in the MAP.
For example, a falling MAP causes arteriolar dilatation, reducing resistance to stabilise CBF; however, this will increase the cerebral blood volume and thus the ICP, being represented by a negative PRx. A positive PRx therefore occurs when autoregulation is failing, and reductions in MAP are associated instead with falling cerebral blood flow, blood volume and ICP.
The magnitude of the PRx does not necessarily correspond to the magnitude of changes in ICP (e.g. at a PRx of 0.3, you don’t necessarily get larger ICP changes to increasing MAPs than at 0.1), but instead the closeness of association of the changes in MAP to ICP and more positive values indicate a worse degree of autoregulatory impairment.
Who were the Patients?
This was a secondary analysis of the STARSHIP cohort. This multicentre trial recruited 135 patients from 10 UK PICUs. The inclusion criteria were:
- Children less than 16 years old
- Admitted to PICU with a GCS ≤8 and neuroimaging confirmed TBI-related pathology
- Required invasive ICP and ABP monitoring for clinical indications.
153 patients were consented, with 18 excluded for incomplete data or other reasons. 11 patients had incomplete follow-up at 12 months, with a total of 124 patients used in the research database.
What was the Exposure?
The team examined various exposures related to dynamic CPP thresholds (CPPopt and LLA) based on autoregulatory status in this secondary analysis. They measured how long the cerebral perfusion pressure (CPP) was outside these calculated ‘safe ranges’ as well as the relation to the fixed CPP threshold of 50mmHG recommended by BTF.
To assess the magnitude of the time outside the safe range of cerebral autoregulation, the investigators considered both the total AND hourly dose below the targeted value (CPPopt or LLA)
This was calculated as the product of the deviation in CPP (i.e. CPPopt-CPP) * time below to give the total dose. The hourly dose was calculated as the average of that divided by the amount of valid data. In addition, they investigated the percentage time with CPP below the target (ptime).
Optimal Cerebral Perfusion Pressure(CPPopt)
This is a derived value from the PRx measurements of a patient, typically derived over the preceding 8 hours of recordings. When plotted against CPP, the PRx values tend to form a U-shaped curve, with increasingly deranged (more positive PRx) cerebral autoregulation as CPP moves farther to each end.
The CPPopt is defined as the lowest point in this U-shaped curve and the position at which changes in ABP were least correlated with ICP changes, and conceptually, the cerebrovascular autoregulatory system is performing optimally for the patient in that time frame.
The Lower Limit of Autoregulation (LLA)
This is another derived value from the PRx, with the clinical correlate being the point at which the arterioles are maximally distended (see the above figure in the Autoregulation section). The authors view this value as a “critical safety threshold“, a point at which further reductions in CPP will have linear reductions in cerebral blood flow and a point at which cerebral autoregulation is poorly functioning. For this study, the authors used the value of PRx of (+)0.2, where further reductions in CPP result in higher PRx (i.e. the left side of the U-shaped curve).
What were the Outcomes?
Primary outcomes
The primary outcome in the cohort was mortality post-discharge and was available for the full 135 patient dataset.
Secondary outcomes
The researchers looked at how exposure affected patients’ outcomes, which were measured using the GOS-E Peds score, a score that stratifies a patient’s recovery between Upper Good Recovery (1) and Death (8). These scores were then dichotomised into ‘favourable’ (1-4) or ‘unfavourable’ (5-8). This information was available for 124 patients.
Prognostic risk scores were calculated using age, motor GCS, pupil reactivity, ISS, and Rotterdam score alongside conditions from pre-admission: hypoxia, hypotension, and cardiac arrest based on the IMPACT scoring system, to re-classify the patients into three roughly equal groups of low, intermediate, and high risk of unfavourable outcome.
The GOS-E Peds thresholds were then set as ≤2, ≤4 and ≤6, respectively, for the groups (corresponding to Lower Good Recovery, Lower Moderate Disability and Lower Severe Disability.
What were the results?
Primary outcome
Due to the small non-survivor group (10 patients, 7.4% of the cohort, and 8% of those with 12-month follow-up included in the subsequent analysis), the investigators focused on the secondary outcome of the dichotomised 12-month outcomes.
Secondary outcomes
Whilst CPPopt deviations were not associated with changes in GOS-E Peds outcome at 12 months, deviations from LLA were. The investigators demonstrated that percentage time (ptime) and hourly dose of CPP below LLA were significantly associated with unfavourable outcome (p 0.04 and 0.04, respectively).
As a dynamic marker representative of the limits of cerebral autoregulation, the investigators also looked at how the LLA changed over time throughout the patient’s admission. In all patients, the LLA changed dynamically over time, with those experiencing a worse outcome experiencing a greater increase in LLA by day 5 (p= 0.003) before returning to match the favourable cohort by around day 8.
Additionally, the investigators examined not only the time spent with a CPP below the LLA but also the LLA’s height throughout a patient’s admission.
All age groups had a relevant percentage of time where the LLA was above the Brain Trauma Foundation recommended CPP threshold of 50 (37% for 0-2 years, 38.2% for 2-8 years, 48% for >8 years).
Lastly, the investigators investigated the predictive value of CPPopt and LLA on the outcome. The CPPopt displayed low Area Under Curve (AUC) with confidence intervals crossing 0.5, suggesting no prognostic relevance.
However, LLA hourly dose (AUC 0.7, 0.51-0.73) and ptime (AUC 0.68, 0.54-0.75) suggested univariable predictive value, and when added to prognostic models of clinical parameters + LLA hourly dose or ptime, improved the predictive performance (AUC 0.72 without -> 0.77 with either hourly dose or ptime). This means that both the LLA hourly data and ptime could each do a decent job of predicting the outcome. With the additional prognostic model, the AUC (a measure of prediction accuracy) increased by 0.05, which is a significant number of additional correct predictions.
What were the limitations?
This was a retrospective observational study and, as such, identifies associations but cannot prove causal links.
Whilst this is the most extensive multicentre prospective observational study of PRx and PRx-based autoregulation CPP targets, it remains a small study size with insufficient numbers for age-stratified analysis.
There was a lack of specific information on intervention strategies and the degree to which CPP deviations may have been unavoidable despite optimal care. Most PRx-derived targets are validated in adult cohorts, so the parameters used for CPPopt and LLA estimation were based on adult data, adjusting for paediatric physiology.
The authors note specifically that the ideal PRx threshold for designation of LLA as the “onset of cerebrovascular autoregulation impairment” is a necessary step, but currently unknown. The value of 0.2 was based in part on a large single-centre cohort, which was utilising mortality data and explored values between 0.2 and 0.3, with the value for STARSHIP designated as 0.2 to correspond to the onset of impairment.
CASP checklist
Does the study address a clearly focused issue?
Yes.
Was the cohort recruited in an acceptable way? (inclusions, exclusions, representation)
Yes. 165 patients were identified for inclusion, and 153 consented (92.7%). 18 patients (11.7%) were excluded, primarily due to incomplete data (11/18)
Were the exposures and outcomes accurately measured to minimise bias?
Yes. The CPPopt and LLA measurements were calculated through high-frequency dynamic monitoring of ICP and ABP. The yield, or percentage of time with valid data, was 88%.
Outcomes were measured using a well-validated tool for assessing post-TBI disability at an appropriate period of follow-up.
Were confounding factors identified and accounted for?
Yes: The patients were stratified based on initial severity and had their GOS-E Peds thresholds adjusted at study design to adjust for this confounder. The authors do note that the deviations from CPP targets are occurring in patients already undergoing Brain Trauma Foundation guideline intervention and thus may be unavoidable even under optimal clinical care.
Was there sufficient follow-up of the patients?
Yes. Of the 135 patients with sufficient data, all had their mortality at discharge recorded, and the 124 patients (91%) included in the further analysis had GOS-E Peds scores recorded at 12 months.
The validation cohort for GOS-E Peds utilised 3- and 6-month follow-up for developmentally appropriate assessment in children, supporting a 12-month follow-up for the STARSHIP cohort.
What were the results?
Whilst CPPopt deviations were not associated with outcome at 12 months, deviations from LLA were.
Percentage time and hourly dose of CPP below LLA were significantly associated with unfavourable outcomes.
These metrics showed prognostic value both alone and when added to a multivariate model involving initial clinical presentation.
LLA changed dynamically over time, with those experiencing a worse outcome diverging from the favourable cohort with higher values. In addition, all age groups had clinically significant periods where LLA exceeded the Brain Trauma Foundation’s upper threshold recommendation of 50mmHg
Do you believe the results?
Yes.
Can the results be applied to our local population?
Yes. This was an observational study across the United Kingdom, with 9/10 of the units being major trauma centres. There were no trial-specific modifications to management practices, which generally aligned with national consensus targets and intervention.
Do the results fit with the other evidence available?
There is growing adult evidence of the benefit of PRx-based monitoring and the validity of derived values in guiding treatment. In adults, CPP below both CPPopt and LLA has shown associations with worse outcomes[18]. However, there have been few and usually small retrospective studies in children, below 67 patients, and these almost exclusively investigated CPPopt.[19, 20, 21, 22]
STARSHIP’s results align with the other large cohort investigating PRx thresholds in children[23]. This 196-child South African single-centre study showed higher PRx in patients with a poor outcome and suggested a PRx threshold of 0.25 as the best predictor for mortality.
Current BTF guidelines for targeted CPP thresholds in children[10] are extrapolations from adult data and whilst there are a few studies suggesting autoregulation-based targets improve prognostication, the optimal CPP thresholds remain understudied.
What did the authors conclude, and what can we take away from this study?
Dynamic High Frequency Monitoring is important, and it is likely to evolve in the neurocritical PICU. Without additional invasive monitoring beyond what is traditionally used, it allows a more complete personalised assessment of a patient’s PRx, CPPopt and LLA.
There is a U-shaped risk curve for PRx with CPP. In addition, in adult studies, this is flattened at the higher end. This suggests a higher risk with a CPP lower than cerebral autoregulatory thresholds than with a higher CPP.
Deviations from a patient’s LLA are a better predictor of outcome than fixed targets. If validated, this may help with clinically relevant long-term prognostication.
LLA is higher in many cases than the Brain Trust Foundation’s recommended CPP of 50mmHg for all age groups in 35-50% of the time. This suggests a potential need to target higher CPPs in our patients and should support targeting the upper end of the BTF guideline 40-50mmHg threshold for paediatric patients.
Further prospective interventional studies will be needed to validate the clinical utility of Lower Limit of Autoregulation (LLA) based targets.
“You discuss Peter’s case with your consultant, in the context of the STARSHIP paper, Peter’s age and your clinical impression of Peter’s ICP lability, suggesting that he might be experiencing cerebral autoregulatory failure.
Without the ability to calculate his PRx and Lower Limit of Autoregulation at the bedside without real time high resolution physiological monitoring, nor a clear evidence base for the optimal PRx to target, you agree that the prudent thing to do in Peter’s case is to aim for the upper range of the Brain Trauma Foundation guidance and aim for a fixed CPP minimum of 50mmHg.
Peter’s ICP lability seems to settle in the subsequent days, and he is extubated a week post-injury without further neurosurgical intervention. He returns to the unit 12 weeks later to thank the staff with his mother, but Peter will need ongoing follow-up to determine the long-term impact of his neurological sequelae.”
Our thanks to Dr Shruti Agrawal, who acted as the senior reviewer for this article
About PICSTAR
PICSTAR is a trainee-led research network open to all doctors, nurses and allied health trainees within Paediatric Intensive Care. We are the trainee arm of the Paediatric Critical Care Society – Study Group (PCCS-SG) and work with them on research, audit and service evaluation.
If you would like to join PICSTAR and get involved in projects, have ideas you would like to propose or get advice/mentorship via PCCS-SG, don’t hesitate to contact us at picstar.network@gmail.com. See their website for more: https://pccsociety.uk/research/picstar/
References
- https://pubmed.ncbi.nlm.nih.gov/16049060/ Epidemiology of traumatic brain injury in children receiving intensive care in the UK, Parslow et al 2005
- https://pubmed.ncbi.nlm.nih.gov/22025592/ Disability 3, 12, and 24 Months After Traumatic Brain Injury Among Children and Adolescents. Rivara FP et al 2011
- https://pubmed.ncbi.nlm.nih.gov/38529909/ Long-Term Outcomes of Pediatric Traumatic Brain Injury Following Inpatient Rehabilitation, Horn et al 2024
- https://pubmed.ncbi.nlm.nih.gov/35061977/ A Population-Based Study of Global Outcome After Moderate to Severe Traumatic Brain Injury in Children and Adolescents, Olsen M et al 2022
- https://pubmed.ncbi.nlm.nih.gov/30470686/ Patterns of Moderate and Severe Injury in Children After the Introduction of Major Trauma Networks, Jones S et al 2019
- https://pubmed.ncbi.nlm.nih.gov/29476389/ Age and Mortality in Pediatric Severe Traumatic Brain Injury: Results From an International Study, Sarnaik A et al 2018
- https://pubmed.ncbi.nlm.nih.gov/32425879/ Injury Causes and Severity in Pediatric Traumatic Brain Injury Patients Admitted to the Ward or Intensive Care Unit: A Collaborative European Neurotrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) Study, Riemann L et al 2022
- https://dontforgetthebubbles.com/targeted-temperature-management-in-paediatric-traumatic-brain-injury/ Targeted Temperature Management in Paediatric Traumatic Brain Injury – Don’t Forget the Bubbles
- https://dontforgetthebubbles.com/managing-raised-intracranial-pressure-in-severe-traumatic-brain-injury-the-basics/ Managing raised intracranial pressure in severe traumatic brain injury – the basics – Don’t Forget the Bubbles
- https://braintrauma.org/coma/guidelines/pediatric
- https://pubmed.ncbi.nlm.nih.gov/39996125/ Status of cerebrovascular autoregulation relates to outcome in severe paediatric head injury: STARSHIP, a prospective multicentre validation study, Agrawal et al 2025
- https://pubmed.ncbi.nlm.nih.gov/9218290/ Continuous assessment of the cerebral vasomotor reactivity in head injury, Czosnyka et al, 1997
- https://pubmed.ncbi.nlm.nih.gov/40165332/ Unlocking the potential of high-resolution multimodality neuromonitoring for traumatic brain injury management: lessons and insights from cases, events, and patterns Bögli SY et al 2025
- https://pubmed.ncbi.nlm.nih.gov/4814394/ The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care, Baker et al 1974
- https://pubmed.ncbi.nlm.nih.gov/18684008/ Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics, Steyerberg et al 2008
- https://pubmed.ncbi.nlm.nih.gov/18684008/ Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics
- https://pmc.ncbi.nlm.nih.gov/articles/PMC3325553/ Validity of a Pediatric Version of the Glasgow Outcome Scale–Extended, Beers et al 2012
- https://pubmed.ncbi.nlm.nih.gov/40399968/ Cerebral perfusion pressure targets after traumatic brain injury: a reappraisal, Bögli SY et al 2025
- https://pubmed.ncbi.nlm.nih.gov/34268644/ Association of outcomes with model-based indices of cerebral autoregulation after pediatric traumatic brain injury, Appavu et al 2021
- https://pubmed.ncbi.nlm.nih.gov/37270434/ ICP, PRx, CPP, andΔ CPPopt in pediatric traumatic brain injury: the combined effect of insult intensity and duration on outcome, Svedung Wettervik et al 2023
- https://pubmed.ncbi.nlm.nih.gov/36804198/ Cerebral pressure autoregulation and optimal cerebral perfusion pressure during neurocritical care of children with traumatic brain injury, Velle et al 2023
- https://pubmed.ncbi.nlm.nih.gov/39702837/ An open source autoregulation-based neuromonitoring algorithm shows PRx and optimal CPP association with pediatric traumatic brain injury, van Twist et al 2025
- https://pubmed.ncbi.nlm.nih.gov/36790173/ Cerebrovascular Pressure Reactivity Has a Strong and Independent Association with Outcome in Children with Severe Traumatic Brain Injury*, Smith CA et al 2023
