Non-Invasive Methods for Intracranial Pressure Monitoring in Traumatic Brain Injury Using Transcranial Doppler: A Scoping Review

Abstract

Intracranial pressure (ICP) monitoring is necessary for managing patients with traumatic brain injury (TBI). Although gold-standard methods include intraventricular or intraparenchymal transducers, these systems cannot be used in patients with coagulopathies or in those who are at high risk of catheter-related infections, nor can they be used in resource-constrained settings. Therefore, a non-invasive modality that is more widely available, cost effective, and safe would have tremendous impact. Among such non-invasive choices, transcranial Doppler (TCD) provides indirect ICP estimates through waveform analysis of cerebral hemodynamic changes. The objective of this scoping review is to describe the existing evidence for the use of TCD-derived methods in estimating ICP in adult TBI patients as compared with gold-standard invasive methods. This review was conducted in accordance with the Joanna Briggs Institute methodology for scoping reviews, with a main search of PubMed and Embase. The search was limited to studies conducted in adult TBI patients published in any language between 2012 and 2022. Twenty-two studies were included for analysis, with most being prospective studies conducted in high-income countries. TCD-derived non-invasive ICP (nICP) methods are either mathematical or non-mathematical, with the former having slightly better correlation with invasive methods, especially when using time-trending ICP dynamics over one-time estimated values. Nevertheless, mathematical methods are associated with greater cost and complexity in their application. Formula-based methods showed promise in excluding elevated ICP, exhibiting a high negative predictive value. Therefore, TCD-derived methods could be useful in assessing ICP changes instead of absolute ICP values for high-risk patients, especially in low-resource settings.

Introduction

Intracranial pressure (ICP) monitoring is often recommended in the management of patients with severe traumatic brain injury (TBI).¹ Such monitoring allows for the titration of specific medical therapies, the assessment of response to said therapies, and the timing of necessary surgical interventions such as decompressive craniectomies.² In fact, intracranial hypertension is independently associated with increased morbidity and mortality in TBI, although there is still a lack of level one evidence about the role of ICP monitoring in patient outcomes.^3,4 In addition, ICP monitoring is one of the main variables necessary for obtaining more advanced functional data such as cerebral autoregulation (e.g., estimates of pressure-reactivity index [Prx], and optimal cerebral perfusion pressure [CPP]) and brain compliance (e.g., the RAP index, cerebral compensatory reserve index, and mean ICP pulse amplitude), among others.^5
–7

Currently, the gold-standard technique for estimating ICP requires invasive methods, particularly intraventricular and intraparenchymal transducers.^1,8 Because of its invasive nature, invasive ICP monitoring may face challenges in its widespread use because of several factors such as patient comorbidities (e.g., coagulopathies in the early post-TBI phase), device availability in resource-limited settings such as low- or middle-income countries (LMICs), physician experience with the necessary devices, the risk of catheter-related infections (particularly with prolonged use, with an incidence of 9.2% for intraventricular and 0.8% for intraparenchymal⁾⁵², and controversies surrounding proper indications for ICP monitoring, placement, and catheter withdrawal timing.^9
–11 Given the aforementioned drawbacks of invasive ICP monitoring, non-invasive modalities may be a key alternative, especially in low-resource contexts, in which cost is of paramount concern. Such modalities are indeed cost effective, and are also safe, as similarly deployable at bedside as invasive methods are, useful in emergent settings, and available for patients for whom there is uncertainty about contraindications for invasive ICP monitoring.¹² The literature describes many different methods for non-invasive ICP (nICP) estimation, including optic nerve sheath diameter (ONSD) measurement assessed by ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI); transcranial doppler (TCD)-derived indices (e.g., pulsatility index [PI] and flow velocities [FV]), and the measurement of pupil size along with other dynamic pupillary variables, such as neurological pupilar index (NPi), latency, constriction velocity, and dilation velocity.^13

–16 This article, however, focuses exclusively on characterizing the evidence regarding TCD's ability to estimate nICP.

Review Questions

The objective of this scoping review is to describe the extent and type of evidence regarding non-invasive methods for ICP monitoring in traumatic brain injury (TBI) using TCD as compared with standard, invasive methods in the adult population. Applying the population, concept, and context (PCC) framework, the following specific questions were formulated.

Which methods are available for non-invasive ICP monitoring using TCD?

What evidence exists for the accuracy and comparison of non-invasive ICP monitoring using TCD versus invasive monitoring (IPCi) for ICP estimation?

Methods

This scoping review was conducted in accordance with the Joanna Briggs Institute (JBI) methodology for scoping reviews.

Inclusion criteria, participants.

This scoping review considered studies including patients >18 years old across the entire TBI-severity spectrum, who underwent non-invasive ICP monitoring using TCD and also required diagnostic invasive ICP monitoring for ICP estimation. All studies in the pediatric population (< 18 years) were excluded.

Concept

The concept of this scoping review was to review studies that investigated nICP monitoring by TCD in adult patients with all degrees of TBI (mild, moderate, and severe) as compared with the analysis derived from gold-standard invasive methods. Topics in this concept include but are not limited to device features, methodological details, variables derived from said methods, the diagnostic accuracy of each method in detecting intracranial hypertension, the reliability of these methods, and the sensitivity and specificity of a specific TCD method in the diagnosis of intracranial hypertension.

Context

This scoping review did not consider the specific race, gender, or geographic location of participants in the selected studies. Given that the anatomy and pathophysiology of TBI within the pediatric population differ substantially from those of their adult counterparts, exclusion was determined solely by participant age, with only studies conducted in the adult population (> 18 years of age) being included.

Types of sources

The present scoping review assessed both experimental and quasi-experimental study designs including randomized controlled trials, non-randomized controlled trials, before-and-after studies, and interrupted time-series studies. In addition, analytical observational studies including prospective and retrospective cohort studies, case-control studies, and analytical cross-sectional studies were considered for inclusion. This review also considered descriptive observational study designs including case series, individual case reports, and descriptive cross-sectional studies for inclusion. Qualitative studies that focus on qualitative data were also considered, including, but not limited to, designs such as phenomenology, grounded theory, ethnography, qualitative description, action research, and feminist research. In addition, systematic reviews that met the inclusion criteria were also considered, depending on the research question.

Search strategy

An initial search in EMBASE and PubMed was undertaken, aimed at locating published studies in the adult population between January 2012 and June 2022 so as to obtain the most updated evidence and technological advances on the subject. Additionally, studies published in any language were included, as the available and useful literature is in a variety of languages. Studies that contained non-invasive monitoring with techniques other than TCD were excluded. Studies containing invasive or non-invasive ICP monitoring for the diagnosis of intracranial hypertension from etiologies other than TBI were also excluded. A detailed search strategy for both databases is contained in Supplementary Table S1.

Source of evidence screening/selection

The initial Embase and PubMed search yielded 608 studies. All identified citations were collated and uploaded into Covidence, and 35 duplicated studies were removed. Studies were screened by two independent researchers (K.M. and S.V.) and one collaborator (O.A.F.). After examining 573 titles and abstracts for inclusion, 520 irrelevant studies were removed, 53 full-text studies were assessed for eligibility, and 31 studies were excluded for reasons described in Figure 1. The results of the search are reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) checklist.²¹

FIG. 1.

Extraction methodology

Results

After reviewing and applying inclusion and exclusion criteria, 22 studies were included for final analysis. Table 1 provides the characteristics of the included publications. Supplementary Table 2 provides the extracted information based upon the formulated research questions.

Table 1.

Characteristics of Included Publications

Characteristic	n
Year of publication
2012	4
2013	1
2016	3
2017	2
2018	1
2019	3
2020	2
2021	4
2022	2

Study design
Prospective cohort study	18
Non-randomized experimental study	1
Diagnostic test accuracy	3

Country
United States of America	3
Italy	1
United Kingdom	8
France	1
Belgium	1
Germany	1
Lithuania	1
Turkey	1
Nepal	1
China	1
South Korea	1
Multinational	2
• United Kingdom and Uruguay	1
• Belgium, France, Italy, Spain, United Kingdom, United States of America, and United Arab Emirates	1

Participant age
>18 years	22

Type of invasive monitoring
Intraprenchymal	9
Intraventricular	3
Intraparenchymal and Intraventricular	8
Not specified	2

Table 2.

TCD Methods for ICP Estimation and Their Correlation With Invasive ICP

First author, year of publication	Country	Type of study	No. of participants	Technique/ Method of nICP estimation	Method of invasive monitoring	nICP vs. ICP monitoring	Conclusions
Kim et al., 2013²⁹	United States	Diagnostic test accuracy study	90 (44 TBI)	nICP calculation based on FV and ABP based on the concepts of semi-supervised machine learning (with MOCAIP algorithm) Type of measurement: Intermittent, unilateral	Intraventricular (for TBI) Type of ICP measurement: Continuous	nICP semi-supervised method showed a high predictive ability (AUC = 0.92) and correlation (r = 0.55) for intracranial hypertension (defined as ICP >30 mmHg)	The semi-supervised method is more accurate and clinically useful than the supervised or PI-based method for the detection of intracranial hypertension.
Kashif et al., 2012²⁸	United Kingdom	Diagnostic test accuracy study	37	nICP calculation based on FV and ABP in an electric circuit analog model (nICP Heldt) Type of measurement: Intermittent, bilateral (25 patients) and unilateral (12 patients)	Intraparenchymal	Across all TCD records: Sensitivity 83%, specificity 70%, AUC 0.83 (>20 mmHg), and correlation between nICP and ICPi was high (R = 0.90) Across bilateral TCD records: Correlation between nICP and ICPi was moderate to high (R = 0.76) On a patient-record basis (45 records): sensitivity 90%, Specificity 80%, AUC 0.88 (>20 mmHg), and correlation between nICP and ICPi was high (R = 0.90)	There was a moderate-to-high correlation between the proposed dynamic model of nICP and ICPi.
Youm et al., 2022²⁰	South Korea	Cohort study	50	TCD (transorbital): Flow velocity in the CRA and PCiA PSV, MFV, EDV, and PI in the CRA and PCiA were estimated Type of measurement: Intermittent, bilateral	Intraventricular Type of ICP measurement: Intermittent	CRA PSV: p = 0.639 EDV: p = 0.232 PI: p = 0.246 PCiA PSV: p = 0.666 EDV: p = 0.495 PI: p = 0.458	CRA and PCA Doppler variables showed poor correlation with invasive ICP.
Rasulo et al., 2017²⁴	Italy	Cohort study	38 (20 TBI)	A 2 MHz ultrasound probe was placed over the acoustic temporal window for insonation of the M1/M2 section of the MCA at a depth ranging from 45 to 55 mm (ipsilateral to the side of ICP bolt placement) ICP_TCD was calculated using the following equations: CPPe = MAP x FVdia/FVm +14 ICP_TCD = MAP - CPPe The sensitivity was expressed as the probability that a patient with high ICPi (>20 mmHg) would also have a high ICP_TCD value, and the specificity was expressed as the probability that a patient with normal ICPi(≤20 mmHg) would also have a normal ICP_TCD value	Intraventricular (type of measurement not specified); and Intraparenchymal	In 5% of the measurements, the ICP_TCD was lower than the corresponding ICPi reading, whereas in 95% of the measurements, the ICP_TCD was higher. With dichotomized ICPi (≤20 mmHg vs. >20 mmHg), the sensitivity of ICP_TCD was 100%, and all patients with ICPi >20 mmHg also had ICP_TCD >20. 2% increase in bias for unit of ICP (p = 0.59), meaning a significant correlation between the measurements provided by the two techniques There was an overestimation of 6.2 mmHg (95% CI 5.08-7.30 mmHg) for ICP_TCD compared with ICPi, with an agreement range from -5.6 to 18 mmHg. The estimated best threshold was at ICPi of 24.8 mmHg corresponding to a sensitivity 100% and a specificity of 91.2%	ICP_TCD is a reliable tool for excluding intracranial hypertension when compared with ICPi for all values of ICPi >20 mmHg, being the best threshold of ICPi of 24.8 mmHg. Although the great majority of patients had TBI, specific and individual patient subgroup analyses (by excluding them from aSAH and ICH patients) were not made.
Sawicki et al., 2021³³	United Kingdom	Cohort study	96	Recordings made on ICM+ software Mean arterial pressure aligned with flow velocity (FV) and cardiac cycles = Difference Index (DI) (formula on page 3 original article) Type of measurement: Not specified. PI estimation	Intraparenchymal	DI values were higher during ICP plateau wave (p < 0.0001), AUC = 0.918 PI values also differed with ICP plateau wave (p < 0.0001)	Both DI and PI, although the former slightly better than the latter, were correlated with ICP plateau waves and may reflect the hemodynamic repercussions of ICP crisis on CPP.
Zhou et al., 2019²³	China	Cohort study	73	Arterial and venous TCD: FVd, FVm and FVs MCA flow velocities using transtemporal window FVd, FVm, and FVs SS flow velocities were recorded by 2 MHz linear probes on the straight sinus through an occipital and transforaminal bone window. Type of measurement (arterial TCD): Intermittent, bilateral Type of measurement (venous TCD): Not specified	Intraventricular (type of measurement not specified); and intraparenchymal	ICP vs. arterial and venous TCD data were presented only describing maximum, minimum, and mean (SD) values, p values, and by correlation coefficients (r): Venous TCD correlation with ICPi: SS FVs (cm/s): r = 0.8662, p < 0.0001 SS FVd (cm/s): r = 0.5538, p < 0.0001 SS FVm (cm/s): r = 0.9298, p < 0.0001 Arterial TCD correlation with ICPi: MCA FVs (cm/sec): r = 0.8724, p < 0.0001 MCA FVd (cm/sec): r = 0.7681, p < 0.0001 MCA FVm (cm/sec): r = 0.9, p < 0.0001	Among all arterial and venous TCD variables, straight sinus MFV seems to be the best correlated variable with invasive ICP measurements.
Martin et al., 2019³⁶	France	Cohort study	54	TCD: 2-MHz probe, transtemporal window (PI was used) Type of measurement: Not specified	Intraparenchymal	ICP of >20 mmHg for at least 5 min TCD-PI: AUC ROC = 0.67 (95% CI, 0.52-0.81) Sensitivity and NPV of 100% for a maximal PI cutoff of 0.9	TCD-PI exhibited a weak correlation with invasive ICP values.
Glenn et al., 2012³⁴	United States	Cohort study	99	MCA, ACA, PCA, ICA, and VA were insonated bilaterally, along with the BA using the general method described by Aaslid et al. Recordings were performed using a 2-MHz pulsed probe. Type of measurement: Bilateral	Intraventricular (type of measurement not specified)	Overall, the right middle cerebral artery (RMCA) PI had the strongest Pearson correlation with ICP, r = 0.45. No PI significantly predicted an ICP >20 mmHg.	The correlations between PI and ICP can be best described as moderate to nil. Thus, the clinical usefulness of TCD as a predictor of ICP is also weak.
de Riva N, 2012⁴⁰	United Kingdom	Cohort study	20	- Recordings on ICM+ software - Pulsatility index (PI) - Type of measurement: Intermittent	Intraparenchymal	PI and ICP correlation: R = 0.7 Linear model 95% prediction margin for ICP of 21 mmHg	PI was moderately correlated with ICP; however, PI itself has a complex relationship with multiple hemodynamic variables (i.e. heart rate, cerebral arterial compliance, cerebral perfusion pressure, among others).
Zweifel et al., 2012³⁸	United Kingdom	Cohort study	290	FV in the middle cerebral artery was monitored using TCD with a 2-MHz probe Type of measurement: Not specified The signals were digitized using A/D converter sampled at frequency 50 Hz Data acquisition and analysis were done using a laptop PC running ICM+ software	Intraventricular (type of measurement: Intermittent) Intraparenchymal	Correlation analysis (Spearman) performed across all patients revealed highly significant but rather poor association between PI and ICP of 0.31 (p < 0.001) The 95% prediction interval of ICP values for a given PI was more than ±15 mmHg The diagnostic value of PI to assess ICP AUC ranged from 0.62 (ICP >15 mmHg) to 0.74 (ICP >35 mmHg) ROC analysis showed that the predictive value of PI can indeed be relatively high, with AUC values >0.75; however, this is the case only for highly elevated ICP (>35 mmHg)	High ICP values (> 35 mmHg) are detected by TCD using the estimation of PI; however, for detecting certain critical clinical thresholds (ICP >20 mmHg), PI is certainly not a reliable diagnostic tool.
Cardim et al., 2016³²	United Kingdom	Cohort study	40	Recordings on ICM+ software A mathematical “black-box” model based on interaction between TCD and ABP (nICP_BB) Based on diastolic FV (nICP_FVd): nCPP = ABP x (FVd/FVm) +14 mmHg nICP = ABP - nCPP Based on critical closing pressure: (nICP_CrCP) CrCP = ICP + WT Equation 2 (article page 13) nICP = ABP - nCPP Based on TCD-derived PI (nICP_PI): nICP = 4.47 x PI +12.68 mmHg Type of measurement: Intermittent, bilateral	Intraparenchymal	“General” correlations (Pearson): nICP_BB: 0.39 nICP_FVd: 0.39 nICP_CrCP: 0.35 nICP_PI: 0.15 nICP_Av (average of first 3 methods): 0.47 “Time domain” correlations (R): nICP_BB: 0.48 ± 0.4 nICP_FVd: -0.28 ± 0.69 nICP_CrCP: 0.18 ± 0.56 nICP_PI: 0.61 ± 0.35 “Changes” in ICP ≥7 mmHg correlations (R): nICP_BB: 0.68 nICP_FVd: -0.32 nICP_CrCP: 0.28 nICP_PI: 0.19	In general, all methods showed significant but not strong correlation with invasive ICP; however, nICP_BB has the strongest relationship with invasive ICP measurement “numerically” (absolute values), whereas nICP_PI has the strongest relationship with invasive ICP measurement in “time trends”, reflecting ICP dynamics.
Schmidt et al., 2016³⁰	United Kingdom	Cohort study	137	Recordings on ICM+ software 2-MHz pulsed Doppler device Type of measurement: Continuous and unilateral, in the ipsilateral MCA (brain lesion side) nICP_BB was used for non-invasive ICP estimation	Intraparenchymal	The absolute difference between ICP and nICP (ΔICP) was on average 4.9 ± standard deviation (SD) of 3.3 mmHg The SD of differences ICP–nICP was 5.6 mmHg The 95 % CI of ICP prediction ranged from −9.6 to 12.3 mmHg In recordings with highly dynamic ICP signals (90th percentile value was at least 9 mmHg above its 10th percentile value), ICP and nICP correlated in time on average by R = 0.51 ± 0.4	nICP has a moderate correlation with ICP, and is observed specifically with dynamic ICP signals (correlation in time).
Cardim et al., 2016¹⁵	United Kingdom and Uruguay	Cohort study	27	Recordings on ICM+ software FV was monitored exclusively from the M1 segment of the MCA through the temporal window with a 2-MHz probe. Type of measurements: Not specified A mathematical “black-box” model (Schmidt et al): Based on interaction between TCD and ABP (nICP_BB) Based on diastolic FV (nICP_FVd) (Czosnyka et al): nCPP = ABP x (FVd/FVm) +14 mmhg nICP = ABP - nCPP Based on critical closing pressure (nICP_CrCP) (Varsos et al): CrCP = ICP + WT Equation 2 (article's page 3) nICP = ABP - nCPP Based on TCD-derived pulsatility index (nICP_PI) nICP = 4.47 x PI +12.68 mmHg	Intraparenchymal	Correlation with invasive ICP >35 mmHg The best correlation was with nICP_PI: AUC ROC 0.81 (95% CI 0.70-0.91) Correlation with changes (delta) in invasive ICP: The best correlation was with nICP_BB (R = 0.44) and nICP_CrCP (R = 0.48)	In patients with intracranial hypertension, nICP_PI showed the best correlation “numerically” with invasive ICP, whereas the best correlation with changes in time domain were exhibited by nICP_BB and nICP_CrCP.
Robba et al., 2017²²	United Kingdom	Cohort study	64 (45 TBI)	Arterial TCD: Intermittent and bilateral on the MCA through the temporal window using a 2 MHz transducer with head elevated to 30 degrees The final values of flow velocities were calculated by averaging the 2 measured values Venous TCD: Intermittent and bilateral on the straight sinus through an occipital and transforaminal bone window using a 2 MHz transducer at a depth of 50 to 80 mm	Intraparenchymal	The correlation coefficient between the FVsv method and ICP averaged per patient showed a correlation with ICP of R = 0.72 The best FVsv cutoff values for prediction of intracranial hypertension were 38.50 cm/sec The formula of the derived models that best fit the data was: nICP_FVsv = 0.38 x FVsv +0.0005 (mmHg) Arterial TCD did not show correlation with invasive ICP monitoring compared with venous TCD	FVsv by venous TCD showed a good correlation with invasive ICP values, being more in terms of the cutoff given, but also with the aforementioned formula of derived models.
Schmidt B, 2018³¹	Germany	Cohort study	41 (20 TBI)	2-MHz pulse doppler device Type of measurement: Continuous and unilateral, in the ipsilateral MCA (brain lesion side) “nICP_BB-like” Signal transformation (DIR) of the ABP signal to ICP, controlled and continuously updated by parameters (TCD-characteristics) 3 calibration methods (adjustment): 1. nICP matrix (cnICP) 2. DIR ABP -> ICP (anICP) 3. Calculated nICP itself (snICP)	Intraventricular (type of measurements not specified); and intraparenchymal	Mean differences and SD for each method: 1. cnICP: -3.4 ± 5.68 mmHg and 5.4 mmHg 2. anICP: +1.0 ± 7.22 mmHg and 5.6 mmHg 3. snICP: +0.7 ± 6.98mmHg and 5.5 mmHg Changes (delta) in ICP (compared with “raw nICP”) - “Raw nICP”: 6.5 mmHg 1. cnICP: 5.4 mmHg 2. anICP: 6.5 mmHg 3. snICP: 6.5 mmHg	The cnICP calibration method showed the best results, with a significantly decreased delta ICP compared with ICPi (see text for further details).
Sari et al., 2020³⁷	Turkey	Cohort study	52 (16 with decompressive craniectomy)	The TCD measurements of MCA were taken bilaterally and continuously in the temporal window using a traditional 2 MHz transducer. Pathological PI values were considered as PI >1.0.	Intraventricular (type of measurement: Continuous) Intraparenchymal	Correlation between ICPi and PI: r = 0.091 (mean of 3 days) Correlation between ICPi and TCD Vmax: r = -0.213 (mean of 3 days)	TCD measured PI and Vmax are poorly correlated with iICP. TCD studies were not performed on patients post decompressive craniectomy.
Robba et al., 2020³⁵	Belgium	Cohort study	100 (30 TBI)	Insonation of the temporal window on both sides and an echo-color Doppler device with a 2-MHz transducer. The TCD measurements were performed bilaterally and intermittently on MCA; the MFV in the MCA was recorded 1 cm after the internal carotid artery bifurcation. PI was considered abnormal if >1.2 eICP and eCPP were calculated using a validated formula: 1. eCPP = (MAP x FVd/FVm) +14, then: 2. eICP = MAP-eCPP	Intraventricular (type of measurement: Intermittent) Intraparenchymal	There was a significant moderate correlation between invasive ICP and MCA PI (r = 0.60, p < 0.001) and eICP (r = 0.66, p < 0.001). The AUC to predict intracranial hypertension (ICP >20 mmHg) was 0.79 [95% CI: 0.63-0.96] for PI and 0.83 [95% CI: 0.69-0.98] for eICP. To predict intracranial hypertension, PI >1.10 and eICP >20 mmHg, had 61% sensitivity and 80% specificity, and 67% sensitivity and 87% specificity, respectively.	The non-invasive ICP estimated by PI and eICP methods were significantly correlated with invasive ICP; however, the correlation was moderate to weak.
Cardim et al., 2019²⁶	United Kingdom	Cohort study	100	Insonation of the temporal window on both sides and an echo-color Doppler device with a 2-MHz transducer. The TCD measurements were performed bilaterally and intermittent on MCA; the MFV in the MCA was recorded 1 cm after the internal carotid artery bifurcation. nICP based on “ICP_TCD”, as follows: 1. CPP non-invasive = (MAP x FVd/FVm) +14, then: 2. ICP_TCD = MAP-CPP non-invasive	Intraventricular (type of measurement: Intermittent) Intraparenchymal	There was no correlation between ICP_TCD and invasive ICP (R = −0.17; 95% CI: −0.35, 0.03; p = 0.097). The mean difference between invasive ICP and ICP_TCD in the overall population was 1.26 mmHg, with 95% prediction interval (limits of agreement) of −27.58 mmHg and 30.10 mmHg, and SD, ±14.42 mmHg. The AUC was 34.5% (95% CI: 23.1%-45.9%), with 0% sensitivity and 74.4% specificity for ICP_TCD to detect invasive ICP >20 mmHg.	nICP estimation, by means of TCD_ICP formula, is not a method to reliably detect intracranial pressure >20 mmHg.
Lucinskas et al., 2021¹⁹	Lithuania	Non-randomized experimental study	7	2-MHz ultrasonic transducer Type of measurement: Continuous The locations of the intracranial ophthalmic artery (IOA) and extracranial ophthalmic artery (EOA) segments were determined using the edge of the internal carotid artery as a reference point “Blood flow factor” (BFF): Integral of the difference of logarithms of Doppler signal intensity P, obtained from the IOA and EOA (Equation 2, Appendix 2 from the article)	Intraparenchymal	Non-invasive ICP monitoring method (TCD) showed a strong positive correlation (r = 0.94) with invasively monitored ICP data The total mean and SD of the measured differences between ICPi and nICP are 0.086 ± 1.34 mmHg Calibration equations and BFF formulas are depicted on pages 8-9 of the original article.	Ophthalmic artery Doppler has a good correlation with invasive ICP measurements up to 1 h without recalibration.
Singer et al., 2021²⁷	United States	Cohort study	135 (66 TBI)	Transtemporal window, MCA insonation. Mean flow velocities, PI and Lindegaard ratio Type of measurement: Intermittent and bilateral Two different machine brands (Spencer and Sonosite) were used	Not specified.	TCD using Sonosite device: Average MCA PSV (right+left) correlation coefficient and p value: 0.302 and 0.019 Average MCA EDV (right+left) correlation coefficient and p value: 0.195 and 0.114 Average MCA PI (right+left) correlation coefficient and p value: 0.129 and 0.565 TCD using Spencer device: Average MCA flow velocity (right+left) correlation coefficient and p value: 0.304 and 0.041 Average MCA PI (right+left) correlation coefficient and p value: 0.195 and 0.44	MCA PSV had a better correlation with ICP than MCA PI, although it tends to be low. Sonosite appears to correlate better with ICP vs. Spencer. Two different machine brands were used, creating a potential confounding variable; by variables assessed, Sonosite > Spencer.
Shrestha et al., 2021⁴⁰	Nepal	Cohort study	26	2-MHz probe, transtemporal window MCA PSV estimation Type of measurement: Not specified	Not specified	For invasive ICP >15 cmH2O: Correlation (Pearson) of MCA PSV: 0.715 (p < 0.000) AUC ROC = 0.893 (95% CI = 0.83-0.96) MCA PSV of 39.6 cm/sec: Sensitivity 82.1%, specificity 84.4%	MCA PSV show a moderate association with invasive ICP values.
Rasulo et al., 2022³⁴	Belgium, France, Italy, Spain, United Arab Emirates, United States, and United Kingdom	Diagnostic test accuracy study	262 (135 TBI)	A low-frequency pulsed 2-MHz ultrasound probe was placed over the acoustic temporal window for insonation of the M1/M2 section of MCA at a depth ranging from 45 to 55 mm Type of measurement: Intermittent and bilateral CPPe: MAP^*FVd/FVm +14 ICP_TCD: MAP - CPPe	Intraventricular (type of measurement: Intermittent) Intraparenchymal	Three thresholds of invasive ICP: Above 20 mmHg: Optimal ICP threshold: 21.5 mmHg, sensitivity: 65.1%, specificity: 74.4%, NPV: 91.3%, AUC: 71.5 Above 22 mmHg: Optimal ICP threshold: 20.5 mmHg, sensitivity: 70.4%, specificity: 71.9%, NPV: 95.6%, AUC: 75.6 Above 25 mmHg: Optimal ICP threshold: 20.5 mmHg, sensitivity: 70.4%, specificity: 71.9%, NPV: 98.6%, AUC: 75.6	TCD is a reliable non-invasive method for excluding intracranial pressure >20 mmHg, and has a better detection capacity for even higher pressures (i.e >25 mmHg).

ABP: arterial blood pressure (in those studies, mean arterial pressure); anICP, signal transformation-estimated ICP; AUC, area under the curve; CPPe, non-invasively-estimated cerebral perfusión pressure; CI, confidence interval; CRA, central retinal artery; cnICP, matrix method ICP estimation; DIR ,signal transformation; eICP, estimated formula-based non-invasive intracranial pressure; EDV, end-diastolic velocity; FV, flow velocity; ICP, intracranial pressure; ICPi, invasive ICP; ICP_TCD, transcranial doppler-estimated intracranial pressure, nICP, non-invasive intracranial pressure, nICP_BB, “black box” method; nICP_CrCP, critical closing pressure method; FVdia (FVd), diastolic flow velocity; FVm, mean flow velocity; FVs, systolic flow velocity; FVsv, straight sinus systolic flow velocity; nICP, non-invasive intracranial pressure; MAP, mean arterial pressure; MCA, middle cerebral artery; MFV, mean flow velocity; PCA, posterior cerebral artery; PCiA, posterior ciliary artery; PSV, peak systolic velocity; PI, pulsatility index; ROC, receiver operating characteristic; SS, straight sinus; TCD, transcranial Doppler; VA, vertebral artery; Vmax, maximal velocity.

Non-invasive ICP monitoring using TCD-derived methods

Among the studies analyzed, the reported technical settings were generally similar in terms of probes, frequencies, insonation windows, and vessels chosen for insonation. In general, a 2 MHz transducer was used with a transtemporal window to insonate the middle cerebral artery (MCA) (Table 2). Five studies included other windows and vessels in addition to the aforementioned. Two used transorbital windows and the central retinal and posterior ciliary arteries,^32,33 one used basilar artery (BA) insonation,³⁴ and two included venous TCD methods (with a transforaminal window for straight sinus flow velocities and indices).^19,47 A few studies reported complementary data, such as insonation depth,^19,35 transducer site positioning (injury vs. side of ICP monitoring),^35,36 and, for bilateral measurements, a final values estimation.¹⁹ With the exception of that by Robba and coworkers, no study reported patient positioning when describing the technical aspects of TCD measurement (Table 2).¹⁹ In addition, there were scarce descriptions of probe placement (band or frame-holding vs. in-place holding by sonographer), and, for non-simultaneous bilateral measurements, there were no inter-measurement time descriptions.¹⁷ Although detailed description and analysis of the different kinds of biases is outside of the scope of this review, it is interesting to mention that, given the operator and technology-dependent nature of ultrasound devices, Singer and coworkers used two different machine brands for TCD studies, with different variables and results for each one of them (Table 2).³⁸

According to the extracted information, nICP estimation using TCD-derived methods can be divided into mathematical techniques and non-mathematical ones. Non-mathematical methods use TCD variables in isolation, such as the PI, resistance index (RI), or flow velocities (systolic, diastolic, and mean), which are often automatically estimated by the device. Conversely, mathematical methods extract those TCD variables (using standardized software) and associate them with pre-established or newly designed algorithms derived from complex operational models, allowing for the simultaneous, real-time integration and analysis of these variables among themselves and with other systemic (e.g., mean arterial pressure) and/or multimodal monitoring-derived (e.g., ICP signals) variables.¹⁵

In this review, the identified mathematical methods were $n I C P_{H e l d t}$ ,³⁹ semi-supervised learning,⁴¹ the “black-box” model $(n I C P_{B B})$ ,⁴⁰ the “black-box-like” model,³⁶ the diastolic flow velocity-based model ( the critical closing pressure-based model ( $n I C P_{C r C P}$ ), the TCD-derived PI $(n I C P_{P I})$ ,⁴² the difference index (DI) (mean arterial pressure aligned with flow velocity and cardiac cycles),⁴³ and the “blood flow factor” (BFF).³² It is worth mentioning that some of these mathematical models generate formulas that can be used at bedside, such as those applied by Rasulo and coworkers, Robba and coworkers, and Cardim and coworkers.^{17,19,35,44,45} Each method description and its corresponding employed variables are summarized in Table 2.

nICP by TCD versus ICPi for IPC estimation

The reviewed studies included TBI patients with both high ICP values (≥ 20 mm Hg at minimum) and lower ICP values (≤ 15 mm Hg). The term “elevated ICP” (defined as ICP >20 mm Hg for at least 5 min, >22 mm Hg for at least 10 min, or otherwise not specified) was interchangeably used with the term “intracranial hypertension” across the studies. The vast majority of the articles described the (ICPi) monitoring method comparator (either intraventricular and/or intraparenchymal) (Table 2); however, only two of them made any distinction regarding the direct comparison of TCD with either invasive method. Additionally, these same two studies were the only two to disclose the site of the injury and the site of the ICP probe placement on the side indicated for insonation.^35,36 There were no specifications about the type of injury (e.g., contusion vs. subdural hematoma) for ICPi and nICP estimation, and one of them did not distinguish between traumatic and non-traumatic diseases (e.g., ischemic stroke).³⁶ Most of the studies were conducted in Europe (especially in the United Kingdom) and the United States, with a low proportion of studies performed in LMICs (Table 1).

Kashif and coworkers³⁹ performed a diagnostic test accuracy study in 37 patients using an nICP calculation based on FV and arterial blood pressure (ABP) in an electric circuit analog model (. In this study, sensitivity was reported as 83% and specificity was reported as 70%, and there was a high correlation (r = 0.90) for ICPi values >20 mm Hg.

Kim and coworkers⁴¹ used a semi-supervised machine learning model showing a high predictive ability (area under the curve [AUC] = 0.92) and correlation (r = 0.55) for ICPi >30 mm Hg, outperforming a PI-based method studied before by the same group.

Cardim and coworkers,⁴² on behalf of Cambridge's Neurophysics Research Group, performed a prospective study in 40 TBI patients with four different nICP methods, establishing different correlations with global ICPi, with the time domain, and in ICPi variations (change in ICP >7 mm Hg) (Table 2). In general, all methods showed significant but not strong correlation with invasive ICP. However, $n I C P_{B B}$ exhibited the strongest relationship with ICPi in absolute values, whereas $n I C P_{P I}$ had the strongest relationship in the time domain.

Cardim and coworkers,¹⁵ in a cohort study of 27 patients assessing the relationship between ICPi plateau waves and TCD using the same four methods described, found that the best correlation with ICPi >35 mm Hg was with $n I C P_{P I}$ (receiver operating characteristic [ROC] = 0.81). Further, they found that the best correlation with changes (delta) in ICPi was with $n I C P_{B B}$ (r = 0.44) and $n I C P_{C r C P}$ (r = 0.48).

Given the theoretical relationship between hemodynamic changes in the venous circulation and global ICP, Robba and coworkers¹⁹ developed a prospective cohort trial in 64 patients (45 with TBI) aimed at comparing arterial (MCA) and venous (straight sinus) FV, both measured by TCD and ICPi methods. The straight sinus FV ( $F V_{s v}$ ) method showed a good correlation with ICPi (r = 0.72). The best $F V_{s v}$ cutoff values for prediction of intracranial hypertension were 38.50 cm/sec. It is of note that arterial TCD did not show correlation with ICPi as compared with venous TCD. A derived formula for nICP estimation with $F V_{s v}$ is described in Table 2.

In a cohort study, Zhou and coworkers⁴⁷ assessed both arterial and venous TCD for nICP estimation in 73 TBI patients. This work identified that straight sinus MFV seems to be the most highly correlated with Invasive intracranial pressure (ICPi) measurements, even over other arterial TCD variables (r = 0.9298, p < 0.0001).

Rasulo and coworkers,³⁵ in a cohort study of 38 patients (20 with TBI), used a validated nICP formula (Table 2), and found that $n I C P_{T C D}$ was a reliable tool for excluding intracranial hypertension when compared with ICPi for all values of ICPi >20 mm Hg. The most effective cutoff threshold for elevated ICPi was 24.8 mm Hg.

Schmidt and coworkers³⁶ conducted a prospective study including 41 patients (20 with TBI) using a “black box-like” model with different calibration methods (Table 2). The cnICP calibration method showed the best results, with a significantly decreased delta ICP compared with ICPi. Nevertheless, it was not specified if the craniotomy patients had TBIs or other pathologies (e.g., ischemic stroke), which limits the interpretation in this population regarding each method.

Cardim and coworkers,¹⁷ using a prospective study of 100 TBI patients in the United Kingdom, identified that there was no correlation between $n I C P_{T C D}$ and ICPi (r = -0.017). Further, $n I C P_{T C D}$ exhibited a 0% sensitivity and a 74.4% specificity in detecting ICPi >20 mm Hg.

Regarding isolated PI correlation with ICPi, Martin and coworkers⁴⁸ conducted a prospective cohort study in 54 patients with TBI and found a weak correlation between these two variables (ROC = 0.67), despite a sensitivity and negative predictive value (NPV) of 100%. The maximal PI cutoff was found to be 0.9.

Sari and coworkers,³⁷ in a cohort study of 52 TBI patients, determined that both PI and maximal flow velocity from the MCA were poorly correlated with ICPi measurements; however, subgroup analysis was not performed on patients with decompressive craniectomy (16 patients), and PI values were considered abnormal in absolute numerical terms (PI >1).

Sawicki and coworkers,⁴³ in another study assessing PI correlation with ICPi using a commercial software and a mathematically obtained index (the DI) in 96 patients with TBI, showed that both variables (DI and PI) were correlated with ICP plateau waves (defined as a sudden rise in ICP >40 mm Hg that lasted at least 5 min) (p < 0.0001).

Zweifel and coworkers⁴⁶ performed a prospective cohort trial in 290 TBI patients assessing the reliability of the PI for ICP estimation. This study found a poor association between PI and ICP (r = 0.31; p < 0.001). The predictive value of PI was indeed relatively high, with AUC values >0.75; however, this was only the case in highly elevated ICP values (> 35 mm Hg).

De Riva and coworkers,⁴⁹ in another prospective cohort trial performed in a small sample (20 TBI patients), found a moderate correlation between PI and ICPi (r = 0.7) and a linear model 95% prediction margin for ICP of 21 mm Hg. This needs to be interpreted in context with other variables that have a significant influence on absolute values of PI (Table 2).

Glenn and coworkers³⁴ analyzed flow velocities and indices from anterior and posterior circulation vessels in 99 patients in a prospective cohort study, concluding that the right MCA PI had the strongest Pearson correlation with ICPi (r = 0.45) in comparison with other vessels and variables. Nevertheless, this correlation was ultimately very weak in absolute terms for predicting elevated ICPi.

Singer and coworkers,³⁸ in a prospective study of 135 patients (66 with TBI), analyzed flow velocities and PI using two different brands of ultrasound machines. They concluded that the MCA peak systolic velocity (PSV) had a better correlation with ICPi than PI (with some differences between each ultrasound machine). In all, however, correlations were weak overall.

Shrestha and coworkers⁵⁰ studied MCA flow velocities and their correlation with ICPi in a cohort of 26 patients in an LMIC. Here, they noted a good MCA PSV correlation with ICPi (r = 0.715, p < 0.000) and a PSV threshold of 39.6 cm/sec with 82.1% sensitivity and 84.4% specificity for ICP values >15 cm H2O (>11 mm Hg)

Lucinskas and coworkers,³² in a randomized experimental study, considered a mathematical model to establish a BFF derived from indices of internal and external segments of the ophthalmic artery. This model had a strong positive correlation (r = 0.94) with ICPi and low measured differences between the two methods (Table 2).

Youm and coworkers,³³ similarly to Lucinskas and coworkers,³² but without a mathematical model, assessed TCD variables in the central retinal artery (CRA) and posterior ciliary artery (PCA) in a cohort study in 50 TBI patients. They noted a poor correlation for each parameter with ICPi measurements.

Finally, Rasulo and coworkers⁴⁴ published the results of the most recently published study analyzed, the IMPRESSIT-2, a prospective multi-center, international diagnostic test accuracy study in 100 patients (20 with TBI), that used a validated formula employed in other studies.^17,35,42,45 They established a correlation with three different ICPi values and three different times, with sensitivity, specificity, NPV, and AUC values increasing with ICPi. This shows that TCD is a reliable non-invasive method for excluding ICP >20 mmHg and that is has better detection capacity for even higher pressures (e.g., >25 mmHg) (Table 2).

Discussion

TCD is a reliable method for estimating nICP, with some mathematical and formula-based methods showing slightly greater correlations with ICPi than isolated, non-mathematical ones. This is particularly true in cases of increased ICP and when considering a serial assessment (trends) rather than obtaining one-time absolute values. Although more and larger controlled studies are needed, some studies analyzed here suggest that TCD could be a useful tool mainly for excluding elevated ICP and detecting ICP variations (so-called “delta ICP”) in patients with TBI, particularly in scenarios in which invasive monitoring is infeasible or contraindicated.

TCD uses a device composed of a 2-MHz handheld transducer that is placed on the surface of the scalp. Arterial information is gleaned via the placement of a probe with a range-gated ultrasound Doppler instrument.³⁰ An ultrasonic beam transmitted by the probe crosses the skull at pre-specified locations and is reflected back from flowing erythrocytes in the blood vessels. These erythrocytes move at different speeds, and the resultant Doppler signal obtained is a combination of different frequency components. Insonation of cerebral arteries can be achieved using four different windows: temporal, orbital, foraminal, or submandibular. Spectral analysis then presents three-dimensional Doppler data in a two-dimensional format.⁵¹ The time vector is represented on the horizontal scale while velocity (frequency shift) is displayed on the vertical scale. Color brightness represents the signal intensity translated into variables, such as mean cerebral blood flow velocity (CBFV), PI, PSV, RI, and end-diastolic velocity (EDV), which allow for the assessment of cerebral hemodynamic changes from different etiologies (e.g., TBI, vasospasm, and aneurysmal subarachnoid hemorrhage [aSAH]), among others.³¹ Measuring these changes in relation to cerebral blood flow (CBF) provides, through different models, an indirect estimation of ICP.

The rationale behind the use of TCD variables for nICP estimation rests in its ability to reflect hemodynamic changes that take place in the cerebrovascular circuitry in response to local changes and/or changes in other components, such as the cerebrospinal fluid (CSF), brain parenchyma, or venous system. These changes subsequently influence ICP.²² As such, CPP estimation is in fact a “driving force” for CBF and has a physiological basis in assessing ICP changes indirectly, given their close relationship (as in complex vasodilation and vasoconstriction cascades).^22
–24A detailed explanation of ICP waveform physiology is outside the scope of this review, but it is well known that ICP pulse amplitude and waveform are shaped by (1) changes in the pulsatile component of the cerebral blood volume in the amplitude (and in P2 waveform), (2) synchronic systolic arterial pressure peak (P1), and (3) synchronic venous blood outflow (P3).^25,26 As a consequence, methods that consider either the calculation of an indirect CPP via TCD or the transformation of ABP and TCD signals may represent a physiological correlation with ICPi.¹⁵ Some of these methods deserve to be analyzed deeply in response to this scoping review's questions.

Mathematical methods have the advantage of considering different variables for nICP estimation that are calculated and depicted both instantly and continuously, allowing for detection of ICP fluctuations in specified time frames.^39,42,43 On the other hand, their complexity in terms of model design as well as the need to interpret algebraic and physical formulas and invest in software, computational resources, and expertise to do so limit their widespread use in the neurocritical care setting.^32,40
–42 This may even reverse the cost effectiveness in LMICs touted at the outset. In addition, because of the numerous variables considered in each model, some may individually represent different components of ICP. An example of this scenario is exhibited in a study using four different methods, in which $n I C P_{B B}$ had the strongest relationship with invasive ICP measurement “numerically,” and $n I C P_{P I}$ had the strongest relationship with invasive ICP measurement in “time trends,” which reflect ICP dynamics.⁴² Another point to be considered is the fact that some models are designed using different cerebral vessels than the MCA (e.g., branches of the ophthalmic artery).³² Given that the vast majority of studies did not include descriptions of the type of injuries (e.g., subdural hematomas, contusions, traumatic SAH), site of the injury, or ICP probe placement side, it is not possible to attribute differences in measurements to those variables.⁴⁰ We must also consider the fact that some patients may exhibit features suggestive of a more complex condition such as intracranial compartment syndrome.^27,28 An additional caveat is the type of measurements performed (continuous vs. intermittent and unilateral vs. bilateral).⁴⁰ This indeed has an enormous influence on the final values and results of the TCD studies. For example, inter-measurement times or angle of insonation errors resulting from the time-consuming task of maintaining the probe in a steady position could have unpredictable effects on the published results.¹⁷ Finally, an issue that has not been considered so far in the reviewed studies is the fact that ABP signals were not specified as being obtained by invasive (arterial line) or non-invasive (arm cuff) means. In the case of the former, no specifications regarding the height of the transducer placement (at the level of the heart vs. at the level of the ear) were made, and such differences may impact the values obtained when incorporating ABP signals into nICP algebraic models.²⁹

With respect to the non-mathematical methods for nICP estimation, it must be stated that the classical and mistaken concept of the PI as an isolated marker of intracranial hypertension (even with consideration of flow velocity changes) needs to be abandoned. Although some studies showed correlations of PI with ICPi, they seem to be weak, because of the complex relationship that PI itself has with multiple hemodynamic variables such as heart rate, cerebral arterial compliance, and CPP.^34,48 Therefore, PI is not simply an indirect marker of distal cerebrovascular resistance.⁴⁹ Lastly, the methodological aspects of two studies need to be considered. The first is related to the estimation units of ICP (cm H2O instead of mm Hg) in the study by Shrestha and coworkers, which was not specified by the authors, even when mentioning that “invasive intracranial pressure monitoring” was used.⁵⁰ The second is related to different TCD machines (and, by consequence, variables) that were employed in the study by Singer and coworkers³⁸ which, although not typically an objective of a scoping review such as theirs, may condition biases related to accuracy and performance of a given device.

As a final point, the results of the recently published trial by Rasulo and coworkers using a validated and easily applicable nICP formula exhibited a very high value for real-world daily clinical practice, a setting in which crucial decisions must be made in ICP screening, ICPi monitoring insertion, and early detection of potentially severe TBI patients.^20,44 However, only time and the dissemination of the results of this work will be able to validate this assertion. Future studies should consider the complex issues behind TCD analysis and interpretation, specific trial designs (namely randomized control trials), the implications of comparing nICP with either intraventricular or intraparenchymal probes, proper definitions of intracranial hypertension (in physiological parameters rather than by absolute ICP values), and the relationship of TCD variables with “numerically exclusive” ICP properties that are not “numerically exclusive,” such as ICP waveform changes.

Limitations

This scoping review has several limitations. First, only studies in patients ≥18 years of age were included, potentially leaving out valuable information from patients ≥16 years who are also considered part of the “adult population”. Second, as a matter of a scoping review design and according to the PRISMA-ScR recommendations,²¹ in-depth statistical analyses or risk-of-bias assessments (usually performed in systematic reviews) were not conducted in our study. This represents a significant limitation for data interpretation in terms of diagnostic accuracy and nICP-ICPi correlation comparison between studies. Third, narrative reviews, which may contain valuable information from other sources in a clear format and consider expert opinions in both technical and data interpretation from TCD studies, were not included in our search. Fourth, during the review process, we did not consider studies related to indirect markers of ICP changes (e.g., TCD-derived autoregulation indices) because of their huge implications in other pathophysiological processes. Undoubtedly, we believe that they deserve a detailed analysis in the context of nICP estimation in TBI patients in future studies. Finally, we only included studies in TBI patients. As such, the analyses and conclusions derived from this work cannot be extrapolated to other neurocritical care patient populations (e.g., those with aSAH, ischemic stroke, or intracerebral hemorrhage).

Conclusion

This review serves as an informative tool for understanding the different methods for bedside ICP monitoring derived from TCD in patients with TBI. We hope that this allows for the consideration of this non-invasive technique as a proven means of assessing ICP changes rather than ICP values, with the most utility being in settings where invasive devices are not available and/or are considered to be of high risk because of patient factors.

Transparency, Rigor, and Reproducibility Summary

This scoping review was conducted in accordance with the JBI methodology for scoping reviews.²¹ Although a protocol was written before beginning the review, the protocol was not published because of project time constraints. A total of two databases (PubMed and Embase) were searched, with initial search strategy formation being undertaken in PubMed. The final searches were completed in June of 2022, yielding 608 studies for review. All identified citations were collated and uploaded into Covidence, and 35 duplicated studies were removed. Studies were screened by two independent researchers (K.M. and S.V.) and one collaborator (O.A.F.). After examining 573 titles and abstracts for inclusion, 520 irrelevant studies were removed, 53 full-text studies were assessed for eligibility, and 31 studies were excluded for reasons described in Figure 1. Any conflicts that arose in the selection process were resolved by a tie-breaking vote from the third member who did not participate in the initial vote. The results of the search are reported using the PRISMA-ScR checklist.²¹ Data were extracted from included studies by all three researchers using a data extraction tool developed in the protocol, and modified after initial review of the articles, to present data in the clearest manner (Table 2). After two researchers completed extraction, consensus was achieved by mediation between the two researchers involved. In accordance with JBI methodology for scoping reviews specifically, included studies were not reviewed for quality or risk of bias.

Footnotes

Acknowledgments

We thank the non-invasive ICP monitoring international consensus participants for their comments on this article based on their experience in the field:

Yasser Abulhassan, Amos Adeleye, Amelia Ain, Marcel Aries, Rafael Badenes R, Pierre Bouzat, Sérgio Brasil, Luis Bustamante, Danilo Cardim D, Marek Czosnyka, Thomas Geeraerts, Daniel Godoy, Mohammad Hirzallah, Bhagavatula Indiradevi, Manuel Jibaja, Pier G. Lochner, Julio Mijangos Mendez, David Menon, Thangaraj Munusamy, Wellingson Paiva, Juan Pinedo, Frank Rasulo, Chiara Robba, Andrés Rubiano, Diana Sanchez, Aarti Sarwal A, Hamisi Shabani, Gentle Shrestha, Dhaval P Shukla, Gene Sung, Fabio Silvio Taccone, Abenezer Tirsit, Franly Vasquez, Sebastián Vásquez, Yu Lin Wong and Getaw Worku.

Authors' Contributions

Karol Martínez-Palacios and Sebastián Vásquez-García were equally responsible for methodology and writing – original draft preparation. Olubunmi A. Fariyike was responsible for software, visualization, writing – reviewing and editing. Chiara Robba was responsible for writing – reviewing and editing. Andrés M. Rubiano was responsible for conceptualization, supervision, writing – reviewing and editing.

Funding Information

There are no sources of funding to report.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

References

Evensen

, Eide

. Measuring intracranial pressure by invasive, less invasive or non-invasive means: limitations and avenues for improvement. Fluids Barriers CNS. 2020 May 6;17(1):34; doi: 10.1186/s12987-020-00195-3.

Carney

, Totten

, O'Reilly

, et al. Guidelines for the Management of Severe Traumatic Brain Injury, Fourth Edition. Neurosurgery. 2017 Jan;80(1); doi: 10.1227/NEU.0000000000001432.

Vik

, Nag

, Fredriksli

, et al. Relationship of “dose” of intracranial hypertension to outcome in severe traumatic brain injury. J Neurosurg. 2008 Oct;109(4):678–84; doi: 10.3171/JNS/2008/109/10/0678.

Chesnut

, Temkin

, Carney

, et al. A trial of intracranial-pressure monitoring in traumatic brain injury. N Engl J Med. 2012 Dec 27;367(26):2471–81; doi: 10.1056/NEJMoa1207363.

Hall

, O'Kane

The best marker for guiding the clinical management of patients with raised intracranial pressure-the RAP index or the mean pulse amplitude?. Acta Neurochir (Wien). 2016 Oct;158(10):1997–2009; doi: 10.1007/s00701-016-2932-z.

Kazimierska

, Kasprowicz

, Czosnyka

, et al. Compliance of the cerebrospinal space: comparison of three methods. Acta Neurochir (Wien). 2021 Jul;163(7):1979–89; doi: 10.1007/s00701-021-04834-y.

Czosnyka

, Pickard

. Monitoring and interpretation of intracranial pressure. J Neurol Neurosurg Psychiatry. 2004 Jun;75(6):813–21; doi: 10.1136/jnnp.2003.033126.

Hawthorne

, Piper

. Monitoring of intracranial pressure in patients with traumatic brain injury. Front Neurol. 2014; 5:121; doi: 10.3389/fneur.2014.00121.

Feng

, Yang

, Jiang

Real-world appraisal of intracranial pressure monitoring. Lancet Neurol. 2021 Jul;20(7):502–3; 10.1016/S1474-4422(21)00164-2.

10.

Tavakoli

, Peitz

, Ares

, et al. Complications of invasive intracranial pressure monitoring devices in neurocritical care. Neurosurg Focus. 2017 Nov;43(5):E6; doi: 10.3171/2017.8.FOCUS17450.

11.

Robba

, Graziano

, Rebora

, et al. Intracranial pressure monitoring in patients with acute brain injury in the intensive care unit (SYNAPSE-ICU): an international, prospective observational cohort study. The Lancet Neurology. 2021 Jul 1;20(7):548–58; doi: 10.1016/S1474-4422(21)00138-1.

12.

Robba

, Bacigaluppi

, Cardim

, et al. Non-invasive assessment of intracranial pressure. Acta Neurol Scand. 2016 Jul;134(1):4–21; doi: 10.1111/ane.12527.

13.

Raval

, Shen

, Lau

, et al. Comparison of Three Point-of-Care Ultrasound Views and MRI Measurements for Optic Nerve Sheath Diameter: A Prospective Validity Study. Neurocrit Care. 2020 Aug;33(1):173–81; doi: 10.1007/s12028-019-00881-7.

14.

Sekhon

, Griesdale

, Robba

, et al. Optic nerve sheath diameter on computed tomography is correlated with simultaneously measured intracranial pressure in patients with severe traumatic brain injury. Intensive Care Med. 2014 Sep;40(9):1267–74; doi: 10.1007/s00134-014-3392-7.

15.

Cardim

, Robba

, Bohdanowicz

, et al. Non-invasive Monitoring of Intracranial Pressure Using Transcranial Doppler Ultrasonography: Is It Possible?. Neurocrit Care. 2016 Dec;25(3):473–91; doi: 10.1007/s12028-016-0258-6.

16.

Chen

, Gombart

, Rogers

, et al. Pupillary reactivity as an early indicator of increased intracranial pressure: The introduction of the Neurological Pupil index. Surg Neurol Int. 2011; 2:82; doi: 10.4103/2152-7806.82248

17.

Cardim

, Robba

, Czosnyka

, et al. Noninvasive Intracranial Pressure Estimation With Transcranial Doppler: A Prospective Observational Study. J Neurosurg Anesthesiol. 2020 Oct;32(4):349–53; doi: 10.1097/ANA.0000000000000622.

18.

Bellner

, Romner

, Reinstrup

, et al. Transcranial Doppler sonography pulsatility index (PI) reflects intracranial pressure (ICP). Surg Neurol. 2004 Jul;62(1):45–51; discussion 51; doi: 10.1016/j.surneu.2003.12.007.

19.

Robba

, Cardim

, Tajsic

, et al. Ultrasound non-invasive measurement of intracranial pressure in neurointensive care: A prospective observational study. PLoS Med. 2017 Jul;14(7):e1002356; doi: 10.1371/journal.pmed.1002356.

20.

Godoy

, Rabinstein

. How to manage traumatic brain injury without invasive monitoring? Curr Opin Crit Care. 2022 Apr 1;28(2):111–22; doi: 10.1097/MCC.0000000000000914.

21.

Tricco

, Lillie

, Zarin

, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Ann Intern Med. 2018 Oct 2;169(7):467–73; doi: 10.7326/M18-0850.

22.

Czosnyka

, Matta

, Smielewski

, et al. Cerebral perfusion pressure in head-injured patients: a noninvasive assessment using transcranial Doppler ultrasonography. J Neurosurg. 1998 May;88(5):802–8; doi: 10.3171/jns.1998.88.5.0802.

23.

Schmidt

, Klingelhöfer

, Schwarze

, et al. Noninvasive prediction of intracranial pressure curves using transcranial Doppler ultrasonography and blood pressure curves. Stroke. 1997 Dec;28(12):2465–72; doi: 10.1161/01.str.28.12.2465.

24.

Rosner

, Rosner

, Johnson

. Cerebral perfusion pressure: management protocol and clinical results. J Neurosurg. 1995 Dec;83(6):949–62; doi: 10.3171/jns.1995.83.6.0949.

25.

Carrera

, Kim

, Castellani

, et al. What shapes pulse amplitude of intracranial pressure?. J Neurotrauma. 2010 Feb;27(2):317–24; doi: 10.1089/neu.2009.0951.

26.

Czosnyka

, Czosnyka

Origin of intracranial pressure pulse waveform. Acta Neurochir (Wien). 2020 Aug;162(8):1815–7; doi: 10.1007/s00701-020-04424-4.

27.

Sahuquillo

, Poca

, Arribas

, et al. Interhemispheric supratentorial intracranial pressure gradients in head-injured patients: are they clinically important?. J Neurosurg. 1999 Jan;90(1):16–26; doi: 10.3171/jns.1999.90.1.0016.

28.

Rubiano

, Figaji

, Hawryluk

. Intracranial pressure management: moving beyond guidelines. Curr Opin Crit Care. 2022 Apr 1;28(2):101–10; doi: 10.1097/MCC.0000000000000920.

29.

Depreitere

, Meyfroidt

, Güiza

. What Do We Mean by Cerebral Perfusion Pressure?. Acta Neurochir Suppl. 2018; 126:201–3; doi: 10.1007/978-3-319-65798-1_41.

30.

D'Andrea

, Fabiani

, Cante

, et al. Transcranial Doppler ultrasound: Clinical applications from neurological to cardiological setting. Journal of Clinical Ultrasound. 2022; 50(8):1212–23; doi: 10.1002/jcu.23344.

31.

Wang

, Li

, Tan

. Transcranial Doppler Ultrasound for Monitoring the Cerebral Hemodynamic Changes and Prognosticating Outcomes in Venoarterial Extracorporeal Membrane-Oxygenated Patients. Int J Clin Pract. 2022; 2022:2912477; doi: 10.1155/2022/2912477.

32.

Lucinskas

, Deimantavicius

, Bartusis

, et al. Human ophthalmic artery as a sensor for non-invasive intracranial pressure monitoring: numerical modeling and in vivo pilot study. Sci Rep. 2021 Feb 26;11(1):4736; doi: 10.1038/s41598-021-83777-x.

33.

Youm

, Lee

, Park

. Comparison of transorbital ultrasound measurements to predict intracranial pressure in brain-injured patients requiring external ventricular drainage. J Neurosurg. 2022 Jan 1;136(1):257–63; doi: 10.3171/2021.1.JNS204218.

34.

Glenn

, Sherma

, McArthur

, et al. The linear relationship between transcranial Doppler pulsatility indices and intracranial pressure is influenced by traumatic brain injury and vasospasm. Acta Neurochir Suppl. 2012; 114:11–5; doi: 10.1007/978-3-7091-0956-4_3.

35.

Rasulo

, Bertuetti

, Robba

, et al. The accuracy of transcranial Doppler in excluding intracranial hypertension following acute brain injury: a multicenter prospective pilot study. Crit Care. 2017 Feb 27;21(1):44; doi: 10.1186/s13054-017-1632-2.

36.

Schmidt

, Cardim

, Weinhold

, et al. Comparison of Different Calibration Methods in a Non-invasive ICP Assessment Model. Acta Neurochir Suppl. 2018; 126:79–84; doi: 10.1007/978-3-319-65798-1_17.

37.

Sari

, Bolukbasi

, Kahraman Özlü

, et al. Decompressive craniectomy in traumatic brain injury: Transcranial Doppler sonography used as a guide. Ulus Travma Acil Cerrahi Derg. 2020 May;26(3):418–24; doi: 10.1474/4/tjtes.2020.04640.

38.

Singer

, Wallen

, Jalbert

, et al. Efficacy of Noninvasive Technologies in Triaging Traumatic Brain Injury and Correlating With Intracranial Pressure: A Prospective Study. J Surg Res. 2021 Jun;262:27–37; doi: 10.1016/j.jss.2020.12.042.

39.

Kashif

, Verghese

, Novak

, et al. Model-based noninvasive estimation of intracranial pressure from cerebral blood flow velocity and arterial pressure. Sci Transl Med. 2012 Apr 11;4(129):129ra44; doi: 10.1126/scitranslmed.3003249.

40.

Schmidt

, Czosnyka

, Smielewski

, et al. Noninvasive Assessment of ICP: Evaluation of New TBI Data. Acta Neurochir Suppl. 2016; 122:69–73; doi: 10.1007/978-3-319-22533-3_14.

41.

Kim

, Hamilton

, Pineles

, et al. Noninvasive intracranial hypertension detection utilizing semisupervised learning. IEEE Trans Biomed Eng. 2013 Apr;60(4):1126–33; doi: 10.1109/TBME.2012.2227477.

42.

Cardim

, Robba

, Donnelly

, et al. Prospective Study on Noninvasive Assessment of Intracranial Pressure in Traumatic Brain-Injured Patients: Comparison of Four Methods. J Neurotrauma. 2016 Apr 15;33(8):792–802; doi: 10.1089/neu.2015.4134.

43.

Sawicki

, Placek

, Łysoń

, et al. Change in Blood Flow Velocity Pulse Waveform during Plateau Waves of Intracranial Pressure. Brain Sci. 2021 Jul 29;11(8):1000; doi: 10.3390/brainsci11081000.

44.

Rasulo

, Calza

, Robba

, et al. Transcranial Doppler as a screening test to exclude intracranial hypertension in brain-injured patients: the IMPRESSIT-2 prospective multicenter international study. Crit Care. 2022 Apr 15;26(1):110; doi: 10.1186/s13054-022-03978-2.

45.

Robba

, Pozzebon

, Moro

, et al. Multimodal non-invasive assessment of intracranial hypertension: an observational study. Crit Care. 2020 Jun 26;24(1):379; doi: 10.3171/jns.1998.88.5.0802.

46.

Zweifel

, Czosnyka

, Carrera

, et al. Reliability of the blood flow velocity pulsatility index for assessment of intracranial and cerebral perfusion pressures in head-injured patients. Neurosurgery. 2012 Oct;71(4):853–61; doi: 10.1227/NEU.0b013e3182675b42.

47.

Zhou

, Li

, Ye

, et al. Ultrasound measurements versus invasive intracranial pressure measurement method in patients with brain injury: a retrospective study. BMC Med Imaging. 2019 Jul 5;19:53; doi: 10.1186/s12880-019-0354-0.

48.

Martin

, Lobo

, Bitot

, et al. Prediction of Early Intracranial Hypertension After Severe Traumatic Brain Injury: A Prospective Study. World Neurosurg. 2019 Jul;127:e1242–8; doi: 10.1016/j.wneu.2019.04.121.

49.

de Riva

, Budohoski

, Smielewski

, et al. Transcranial Doppler pulsatility index: what it is and what it isn't. Neurocrit Care. 2012 Aug;17(1):58–66; doi: 10.1007/s12028-012-9672-6.

50.

Shrestha

, Shrestha

, Ghale

, et al. Correlation between Invasive Intracranial Pressure Monitoring and Optic Nerve Sheath Diameter in Patients with Traumatic Brain Injury. Kathmandu Univ Med J (KUMJ). 2021; 19(74):221–4.

51.

Kalanuria

, Nyquist

, Armonda

, et al. Use of Transcranial Doppler (TCD) ultrasound in the Neurocritical Care Unit. Neurosurg Clin N Am. 2013 Jul;24(3):441–56; doi: 10.1016/j.nec.2013.02.005.

52.

Dimitriou

, Levivier

, Gugliotta

: Comparison of complications in patients receiving different types of intracranial pressure monitoring: a retrospective study in a single center in Switzerland. World Neurosurg, 89:641–646, 2016

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB