This study demonstrates that TCD can be used to quickly and easily determine the “brain shock index” that, with excellent sensitivity and specificity, predicts poor outcome in children with cerebral malaria. TCD was first used to evaluate the cerebrovascular hemodynamics of 50 Kenyan children with CM in 1996, with the hopes of gaining an improved understanding of the pathophysiologic contributors to brain injury in the disease [34]. Abnormal flows were reported in 30% of children, with perturbations associated with poor outcomes [30]. Our group has subsequently performed TCD on > 1200 children with the disease (with findings published for n = 616). [23, 24, 35] Based on well-defined alterations to TCD waveform morphology and measured flow velocities in all vessels of the Circle of Willis, five distinct phenotypes of deranged cerebral blood flow have consistently emerged (Fig. 1). The low flow phenotype is associated with the highest rate of death or neurodisability (RR 2.1, 95% CI 1.0-3.4). We have also found clear evidence of significant disruption of the cerebral energy metabolism in children with CM and the low flow phenotype. The median cerebrospinal fluid lactate: pyruvate ratio (LPR) in the low flow subset of children was 85 [IQR: 73–184], with values < 40 being considered normal [36]. Together, our studies show that TCD can be used as a “stethoscope into the brain” of children with CM to identify poor cerebral perfusion that leads to cerebral metabolic failure and worse outcomes.

However, as complete TCD studies evaluating all intracranial vessels, examinations performed for these prior reports require significant technical acquisition skills that limit the generalizability of its use as a neurodiagnostic tool in this setting. The BSI simplifies the approach by requiring assessment of only two easily visualized vessels (the MCA and Ex-ICA).

We also found the BSI advantageous to the traditional diagnostic TCD in these critically ill pediatric patients as it eliminated the need to interpret complex waveforms and to determine the number of standard deviations the measured flow velocities fell from age normative values. This allowed for rapid scanning and a BSI determination time of 4 min or less for most children in our cohort, making it feasible to use during emergent cerebral resuscitation events.

Further, use of the BSI may overcome an important obstacle the current TCD criteria for low flow has if this bedside tool were to be used in the future to guide therapeutic interventions for high risk children with CM. CM associated seizures are extremely common, prolonged, and refractory. [37] Thus, a majority of children presenting with CM will receive multiple doses of diazepam and/or phenobarbital prior to TCD study. Sedative agents such as these will reduce cerebral metabolic demand, and in doing so, reduce cerebral blood flow velocities by 21–47%[38]. This flow metabolism coupling is a normal physiologic response but, using our current diagnostic criteria, is still identified as a low flow phenotype. Distinguishing pathologic reductions in cerebral perfusion pressure that may benefit from therapeutic intervention from those with low flow that is simply physiologic is paramount. In pathologic alterations to the cerebrovascular hemodynamics, MCA flow will reduce, but Ex-ICA flow is maintained resulting in a “low” BSI. In flow-metabolism coupling, both the MCA and Ex-ICA flow will be low, resulting in a “normal” BSI. [27, 28] Future trials should confirm the ability of the BSI to make this important distinction.

The basis for measuring the BSI arises from the Monroe-Kellie doctrine. This doctrine states that the space of the cranial cavity is fixed so that the volume of its components (brain matter, blood, and cerebrospinal fluid) are interdependent. When cerebral edema occurs, the volume of the brain tissue increases and cerebral blood flow decreases. [39] MCA flows on TCD have been shown to negatively correlate with both the degree of cerebral edema on neuroimaging and with invasively measured ICP, whereas Ex-ICA flow is generally maintained. [27, 29, 40, 41] Therefore, when the ratio of flow between the MCA and Ex-ICA is low, a significant impairment in cerebral perfusion is present.

Given the pathophysiologic tenants of its measurement, we believe the BSI may have broader applicability as a screening and neuromonitoring tool in other critically ill pediatric patients with conditions other than CM. In particular, the BSI is most likely to be helpful in those at risk of or with diffuse brain injury and global cerebral edema from traumatic brain injury (abusive head trauma or diffuse axonal injury), hypoxic ischemic encephalopathy, neurologic infection, diabetic ketoacidosis with impaired consciousness, liver failure with hepatic encephalopathy, and for those undergoing extracorporeal support.

It remains less clear how the BSI would perform in those with focal neurologic illness or injury (traumatic brain injury (TBI) with focal intracranial hemorrhages, stroke, arteriovenous malformation rupture). In this setting, it may be possible that the BSI ipsilateral to the impacted side or significant side to side variability in the measured BSIs would be the most clinically meaningful. It is unlikely that the BSI would be helpful when impaired cerebral perfusion has a cardiovascular cause such as septic or cardiogenic shock as the Ex-ICA flow would be low in these situations and the measured BSI would be “normal”.

Last, while frequently disrupted in children with various forms of acute neurologic illness or injury, intact autoregulation may blunt the reductions in BSI as ICP increases and CPP decreases to a certain degree, resulting in a change to the clinically relevant threshold. [5, 6, 42] Future studies from other settings that include patients with heterogeneous diagnoses and those with both global and diffuse brain injury and various degrees of impairment to autoregulation should be done to explore if the determination of the BSI is a useful adjunct to clinical care in these situations.

There are limitations to our work. Participants were enrolled only when a TCD technician was available to perform the admission study, so the data presented here are not from consecutive patients. Also, based on operator availability, some patients had one or two TCD examinations missed. If all TCD studies had been performed, the values describing the relationship between the BSI and outcome may have differed slightly from what we report.

Another limitation is that invasive ICP monitoring was not available at any of the sites to allow us to determine if the BSI correlates directly with measured ICP/CPP. In future trials, if a strong relationship is identified, the BSI may not only be used as a prognostic biomarker but also to determine if/when invasive monitoring is indicated or to guide therapies to ensure ICP/CPP goals are met without monitoring.

The final limitation of the study is that no therapeutic interventions were undertaken for those with a BSI ≤ 1.1. It therefore remains unclear if treatment of a critically low BSI would improve rates of neurologic morbidity or mortality or if these were the moribund patients in the cohort. Future studies can also explore if treatment improves the BSI and/or has an impact on short and long-term outcomes.