Article, Pediatrics

Cerebral oximetry with cerebral blood volume index in detecting pediatric stroke in a pediatric ED

a b s t r a c t

Background: Despite pediatric stroke awareness and pediatric stroke activation systems, recognition and imaging delays along with activation inconsistency still occur. Reliable objective pediatric stroke detection tools are need- ed to improve detection and activations. Regional cerebral oxygen saturation (r_cSO2) with cerebral Blood volume index can detect abnormal cerebral physiology.

Objective: To determine cerebral oximetry in detecting strokes in stroke alert and overall stroke patients.

Method: Left r_cSO2, right r_cSO2, and r_cSO2 side differences for stroke, location, and types were analyzed. Results: Compared with stroke alert (n = 25) and overall strokes (n = 52), r_cSO2 and CBVI were less than those in nonstrokes (n = 133; P b .0001). R_cSO2 side differences in stroke alert and overall strokes were greater than in nonstrokes (P b .0001). Lower r_cSO2 and CBVI correlated with both groups’ stroke location, left (P b .0001) and right r_cSO2 (P = .004). R_cSO2 differences greater than 10 had a 100% positive predictive value for stroke. Both groups’ r_cSO2 and CBVI side differences were consistent for stroke location and type (P b .0001). For both groups, left r_cSO2 and CBVI were greater than those of the right (P b .0001). Hemorrhagic strokes had lower bilateral r_cSO2 and CBVI than did ischemic strokes (P b .001).

Conclusions: Cerebral oximetry and CBVI detected abnormal cerebral physiology, stroke location, and type (hemor- rhagic or ischemic). R_cSO2 side differences greater than 10 or r_cSO2 readings less than 50% had a 100% positive predic- tive value for stroke. Cerebral oximetry has shown potential as a detection tool for stroke location and type in a pediatric stroke alert and nonalert stroke patients. Using cerebral oximetry by the nonneurologist, we found that the patient’s r_cSO2 side difference greater than 10 or one or both sides having less than 50% r_cSO2 readings suggests abnormal hemispheric pathology and expedites the patient’s diagnosis, neuroresuscitation, and radiologic imaging.

Introduction

Health care providers often misinterpret the signs and symptoms of acute pediatric strokes resulting in significant diagnosis and imaging de- lays [1-5] with a median time to diagnosis of 9.6 hours [3-7]. One-third

? Study presented as abstract: (1) International Stroke AHA Conference 2015 (poster), Nashville, TN, and (2) Pediatric Academic Society meeting 2015 (oral presentation).

?? Conflict of interest statement: There was no financial assistance or support, or spon-

sorship by any of the companies mentioned. The authors alone are responsible for the con- tent and writing of this manuscript.

* Corresponding author at: Department of Pediatrics, Division of Pediatric Emergency Medicine, University of Arkansas School of Medicine, Arkansas Children’s Hospital, 1 Children’s Way, Slot, Little Rock, AR 72223. Tel.: +1 501 364 2299; fax: +1 615 343 8407.

E-mail addresses: [email protected], [email protected] (T.J. Abramo).

of acute pediatric focal neurologic deficits have nonischemic patholo- gies that mimic stroke, thus impacting the health care provider’s aware- ness of pediatric stroke [2-7]. Significantly reducing the time between suspicion of a neurologic insult from a stroke and initiation of neuroim- aging is crucial in childhood stroke as stressed by the Pediatric Stroke consensus management guidelines [2-6]. A rapid, objective, noninva- sive neurologic tool that reliably detects abnormal cerebral physiology, pathology, and laterality would be ideal and would significantly in- crease Stroke recognition, decrease stroke recognition time, shorten im- aging time, and potentially improve outcomes through earlier initiation of the supportive care and neurointervention treatments.

Cerebral oximetry monitoring provides trends on Tissue hemoglobin oxygen saturation (r_cSO2), blood flow, and Oxygen extraction, among other cerebral physiological variables [8-17]. The normal pediatric

http://dx.doi.org/10.1016/j.ajem.2015.07.037

cerebral r_cSO2 (%O₂HB) range is 60% to 70%, whereas an r_cSO2 less than 50 or decreasing r_cSO2 readings indicate hypoxia, decreased perfusion, or increased oxygen extraction [8-16]. The INVOS (INVOS, Somanetics, Troy, MI) cerebral oximetry can also detect cerebral blood volume index (CBVI). Cerebral blood volume index values correspond to region- al cerebral blood flow facilitating recognition of ischemia-related cere- bral injury [12-16].

Pediatric cerebral oximetry studies have demonstrated that r_cSO2 variations correlate with abnormal cerebral physiology in various neu- rologic scenarios [8-16]. In adult and pediatric neurologic emergencies, r_cSO2 side differences or r_cSO2 readings less than 50% correlate with the laterality of abnormal cerebral physiology and radiologic pathology [12-16].

A fast objective, noninvasive neurologic tool, indicating abnormal cerebral physiology and pathology and with the potential to indicate a unilateral stroke physiology, would significantly increase stroke recog- nition, decrease stroke recognition time, shorten imaging time, and po- tentially improve outcomes through earlier initiation of the supportive care and neurointervention treatments.

The aim of this study is to analyze the diagnostic value of cerebral oximetry with or without CBVI monitoring in PED pediatric stroke and nonstroke patients for stroke location, type (hemorrhagic or ischemic), and its value in a pediatric stroke alert system and in nonalert stroke pa- tients with Altered mental status .

Materials and methods
1. Study design and participants

We conducted a prospective convenience observational trial at an urban, academic, children hospital’s pediatric level 1 trauma ED (PED) with 56000 annual visits from 2006 to 2013. Cerebral oximetry with CBVI (INVOS) with age-appropriate cerebral oximetry probes (neonate probes and pediatric probes for patients weighing b 40 kg and probes for adults weighing N 40 kg) has been an integral part of the PED cerebral assessment tool in suspected pediatric neurologic emergencies since 2006. The Vanderbilt institutional review board approved waived con- sent for review of the PED cerebral oximetry in neUrologic emergency database and the patient’s electronic medical record (EMR).

A Multidisciplinary team developed the pediatric stroke activation system in 2009 with the PED as the only site for stroke alert activation. Upon PED stroke activation, the patient had immediate left and right ce- rebral oximetry monitoring as part of their stroke monitoring order set and before any radiologic studies. High-acuity PED AMS patients assessed in the PED resuscitation room as part of their monitoring as- sessment had left and right cerebral oximetry monitoring applied by the PED staff prior to Computed tomographic scans or magnetic reso- nance imaging (MRI) studies. The patient’s cerebral oximetry monitor- ing was maintained until discharge from the PED. The normal pediatric cerebral r_cSO2 (%O₂HB) range is from 60% to 70% [8-17]. The CBVI has a range from -50 to +50 [12-16]. The positive or negative CBVI value equates to the fact that an increase or a decrease in regional cerebral blood flow has occurred [12-16].

Because of the non-MRI compatibility of the cerebral oximetry

probes, they were removed from the patient prior to the MRI studies and replace upon return to the PED. The neuroradiologist had no prior knowledge of the patient’s cerebral oximetry readings.

Study population

Investigators analyzed 2 databases for PED stroke patients who had cerebral oximetry monitoring, the PED cerebral oximetry redcap data- base (nonstroke activation) 2006-2013 and the PED stroke activation redcap database 2009-2013. The control patients were nonstroke alert and nonstroke AMS patients who had left and right r_cSO2 with or with- out CBVI monitoring and normal neuroradiologic studies.

The first population comprised 2009-2013 pediatric stroke activa- tion patients, consisting of PED patients who presented (by emergency medicine services or private transportation and no interfacility patient transfer) with signs or symptoms triggering the pediatric stroke alert activation system with cerebral oximetry monitoring (left and right r_cSO2 readings recorded every 5 seconds for 60 minutes at minimum) before radiologic studies. The second study population comprised the PED overall stroke patients from 2006 to 2013, consisting of patients with high-acuity PED AMS nontrauma, prestroke activation AMS, and nonstroke activation AMS (no trauma history or trauma-related radio- logic pathology), and strokes or nonstroke patients who had cerebral oximetry (left and right r_cSO2 readings recorded every 5 seconds for 60 minutes at minimum) prior to head computed tomographic scans or MRI plus the stroke alert activation patients (nonstroke and stroke pa- tients [2009-2013]; Table 1).

We excluded patients with a cerebrospinal fluid shunt, prior neuro- surgical procedures, new or old brain tumors, history of trauma, or any trauma-related radiologic pathology; inpatients; those who had interfacility transfers; and/or patients whose cerebral oximetry moni- toring occurred after their radiologic studies. These patients were ex- cluded due to their potential abnormal cerebral physiology and r_cSO2 readings as defined in prior cerebral oximetry studies [8-16].

The study investigators reviewed the selected patients for inclusion and exclusion criteria, EMR, radiologic reports, and final diagnosis by the pediatric neurology attending.

Statistical analysis

The primary cerebral oximetry outcome analysis was for the left and right r_cSO2 and r_cSO2 side difference values for detecting abnormal unilat- eral cerebral physiology, stroke location, and types (hemorrhagic or is- chemic) compared with the radiologic pathology in both groups. The primary statistical analysis was based on patients enrolled between 2009 and 2013 in the stroke alert activation group. Secondary statistical analysis was based on overall stroke patients enrolled between 2006 and 2013. A 0- to 60-minute mean r_cSO2 was selected for analysis. Statis- tical analysis methods for both groups used 2-sample t test, receiver op- erating characteristic (ROC) curves, stroke probability, and Youden Index to determine the optimized cutoff point. A 2-sided statistical sig- nificance level of .05 was used for all statistical tests.

Results

In the stroke alert and overall stroke groups, there were no issues with cerebral oximetry probes obtaining r_cSO2 readings of any age and no false-positive correlation for left, right, and side differences of r_cSO2 readings to radiologic pathology.

Stroke alert patient populations

From 2009 to 2013, 88 patients triggered and completed the PED stroke alert system: 25 stroke (8 hemorrhagic and 17 ischemic) patients and 63 nonstroke patients in Tables 1 and 2. All stroke alert patients had stroke activation, radiologic studies, and cerebral oximetry.

Overall stroke patient population

From 2006 to 2013, 210 PED patients who fulfilled AMS inclusion and exclusion criteria had cerebral oximetry before radiologic studies. The overall stroke population was 52 AMS strokes, 25 positive stroke alerts (hemorrhagic [n = 33] or ischemic [n = 44]), 133 AMS nonstrokes, and 63 nonstroke alerts (Tables 1 and 2).

Patients’ pertinent medical history examination showed that 36% had no medical history for the stroke alert group and 39% for the overall stroke group. The rest of the both groups’ pertinent medical history in decreasing frequency is neurologic (arteriovenous malformation

Table 1

Two stroke study populations: (1) stroke alert activation patients and (2) overall stroke population: stroke alert activation plus AMS patients

[AVM], hemorrhagic brain tumor, vascular [Middle cerebral artery], is- chemic, vertebral artery dissection, moyamoya, right internal carotid ar- tery obstruction), hematologic/oncologic (clotting disorder, postcancer: leukemia, lymphoma, nonbrain tumor), neonatal vitamin K administra- tion refusal (6 patients), Neck trauma (vertebral, internal carotid trau- ma: weight lifting), genetic/metabolic disease, congenital heart postrepair (4% for both groups), infectious disease (Aseptic meningitis, venous thrombosis), Sickle cell disease (1% for both groups), and none for rheumatologic/immunologic. The most common presenting signs/ symptoms were hemiparesis/weakness (67%), AMS (46%), and head- ache (39%). In the sickle patients there were no issues with obtaining consistent left and right r_cSO2 readings during their monitoring period. For the stroke alert population (n = 88), there were 43.7% women.

Ages ranged as follows: all stroke alert patients, 11.4 +- 5.3 years; stroke patients (n = 25), 10.1 +- 4.9 years; and nonstroke patients (n = 63),

10.1 +- 5.3 years (P = .98). The Glasgow Coma Scale was 11.6 +- 4.4 for stroke and 13.5 +- 1.9 for nonstroke (P b .001). For the 25 stroke pa- tients, 60% had right-sided stroke, with 28% being hemorrhagic (AVM) and 72% ischemic. Recent patient seizure occurred in 70% of the stroke patients and in 48% of the nonstroke patients (P = .12).

In the overall stroke group, 46.4% were female with ages as follows: all overall stroke patients (n = 210), 8.7 +- 7.9 years; stroke patients (n = 77), 6.9 +- 5.9 years; and nonstroke patients (n = 133), 7.8 +-

5.5 years (P = .53). The Glasgow Coma Scale was 11.3 +- 3.2 for stroke patients and 12.6 +- 1.7 for nonstroke patients (P b .001). For the overall stroke group, 53% had right-sided stroke and 47% had left-sided stroke, with 42% being hemorrhagic (AVM) and 58% ischemic. The seizure oc- currence was 69% in the overall stroke patients and 70% in nonstroke patients (P = .87).

The lower r_cSO2 and CBVI readings correlated with the location of the stroke as defined in the abnormal radiologic study in both stroke alert and overall stroke groups (P b .0001). In both stroke groups, there was a significantly lower left and right r_cSO2 CBVI means as compared with the nonstroke group, with a high statistical difference between the 2 groups (P b .0001). In both stroke groups, when comparing the r_cSO2 and CBVI side differences, there was a strong statistical significance

(P b .0001; Tables 1 and 2). In the 2 stroke groups, the left stroke r_cSO2 and CBVI means were significantly lower than the right r_cSO2 and CBVI means (P b .0001 Table 2).

Comparing hemorrhagic with ischemic strokes in both study groups, the lowest r_cSO2 occurred in hemorrhagic strokes and correlated with the stroke location. The lowest left and right r_cSO2 occurred within the hemorrhagic strokes (Tables 1 and 2). In comparing the r_cSO2 side differ- ences between hemorrhagic and ischemic strokes, the r_cSO2 differences were lower in hemorrhagic than in ischemic strokes (Tables 1 and 2). For both stroke groups, CBVI analysis of left (P = .001) and right stroke (P = .017) showed significantly lower readings than that of nonstroke. In both stroke groups, the side r_cSO2 differences between hemorrhage and infarct CBVI means were significant (P = .007), with greater nega- tive CBVI readings for hemorrhagic compared with ischemic strokes (Tables 1 and 2).

Hemorrhagic vs ischemic stroke analysis for the stroke alert and overall stroke patients indicated that left hemorrhagic stroke had the lowest r_cSO2 and CBVI readings (P b .001).

In Table 3, in both study populations, the stroke patients had consis-

tent wide r_cSO2 side differences from 0 to 45 minutes (P b .0001); with the greatest r_cSO2 differences occurring from 0 to 10 minutes (P b .0001). Figs. 1 and 2 show graphically the left and right r_cSO2 and CBVI trends for the stroke alert and overall stroke patients, along with 2 individual stroke

patients’ r_cSO2 and CBVI readings (ischemic and hemorrhagic strokes).

The stroke probability for stroke alert and overall stroke patients is shown in Table 4.

The stroke alert and overall stroke ROC curves for left r_cSO2, right r_cSO2, and r_cSO2 side differences are shown in Fig. 3. A further ROC anal- ysis, with exclusion of 4 hemorrhagic stroke patient outliers (1 stroke alert patient and 4 overall stroke patients) whose r_cSO2 side differences were 3.1 and whose both left and right cerebral side r_cSO2 readings were less than 40% indicating significant bilateral pathology, was done (Fig. 3). In all 3 ROC analyses, the greatest significance was found be- tween r_cSO2 side differences for detecting strokes and location (Fig. 3).

In the stroke alert and overall stroke populations, from 0 to 60 minutes, there were consistently lower r_cSO2 and CBVI in the side corresponding to

Table 2 Continuous variables
		Stroke alert patients					Overall stroke patients
	Positive stroke	Negative stroke Difference between		P		Positive stroke	Negative stroke Difference between		P
	alert (n = 25)	alert (n = 63) sides (95% CI)				(n = 77)	(n = 133) sides (95% CI)
Left r_cSO2 (O₂Hgb 15%-95%),	54.4 +- 16.0	73.0 +- 9.8 18.6 (12.7-25.6)		b.0001		54.4 +- 14.9	71.3 +- 8.4 17.42 (12.3-22.6)		b.0001
mean +- SD
Right r_cSO2 (O₂Hgb 15%-95%), mean +- SD	56.7 +- 16.4	73.7 +- 9.2	19.2 (13.1-17.3)	b.001	56.7 +- 15.6		71.4 +- 7.8	19.38 (13.8-24.9)	b.0001
Difference between sides,	33.3 +- 8.8	3.3 +- 2.3		b.0001	30.5 +- 11.9		3.3 +- 2.4		b.0001
r_cSO2 side, mean +- SD Difference between sides, r_cSO2	33.2 (27.7-38.9	3.3 (2.9-3.7)			30.5 (27.5-33.4		3.7 (3.0-3.6)
mean (CI 95%)

Stroke alert patients (CBVI) Overall stroke patients (CBVI)

Positive stroke alert (n = 23/25)

Negative stroke alert (n = 32/63)

Difference between sides (95% CI)

P Positive stroke (n = 37/77)

Negative stroke (n = 84/133)

Difference between P

sides (95% CI)

Left CBVI (-50 to +50),

mean +- SD

Right CBVI (-50 to +50), mean +- SD

Difference between CBVI side, mean +- SD

Difference between sides, mean

CBVI (CI 95%)

-3.5 +- 25.3 16.4 +- 4.1 26.7 (14.2-39.2) b.0001 -3.56 +- 25.09 16.93 +- 10.47 25.46 (13.3-37.6) b.001

7.6 +- 23.0 18.8 +- 9.5 16.1 (5.2-26.9) .016 5.9 +- 24.0 18.5 +- 8.1 14.18 (3.4-24.96) .001

35.7 +- 22.8 12.4 +- 6.9 b.0001 32.7 +- 20.2 11.4 +- 8.0 b.001

35.7 (25.9-45.6) 12.4 (9.9-14.9 32.7 (25.9-39.4) 11.4 (9.7-13.1)

Stroke (+) patients: Stroke type Stroke alert patients Overall stroke patients

Hemorrhagic (n = 8)

Infarct (n = 17)

Difference between sides (95% CI)

P Hemorrhagic (n = 33)

Infarct (n = 44)

Difference between P

sides (95% CI)

Left stroke: left r_cSO2	31.5 +- 4.6	48.8 +- 3.6	17.3 (5.7-29)	.0037	36.8 +- 6.9	46.3 +- 13.5	17.3 (5.7-29)	.02
(O₂Hgb 15%-95%), mean +- SD Left stroke: right r_cSO2	60.6 +- 3.4	80.5 +- 2.8	10.8 (6.2-21.5)	.0464	67.0 +- 17.0	75.0 +- 10.0	19.9 (11.2-28.6)	.07
(O₂Hgb 15%-95%), mean +- SD Left stroke: difference	30.5 (21.1-39.8)	31.8 (26.2-37.4)			25.5 (18.3-32.8)	25.7 (21.8-29.7)
between r_cSO2 sides, mean (CI 95%) right stroke: left r_cSO2	63.9 +- 4.2	74.7 +- 3.3	19.9 (11.2-28.6)	b.0001	59.0 +- 13.0	73.0 +- 13.0	19.9 (11.2-28.6)	b.015
(O₂Hgb 15%-95%), mean +- SD right stroke: right r_cSO2	31.9 +- 4.1	49.1 +- 3.3	17.2 (7-27.4)	b.0001	42.8 +- 8.5	55.9 +- 12.6	17.2 (7-27.4)	b.001
(O₂Hgb 15%-95%), mean +- SD right stroke: difference	27.6 (20.2-35.0)	31.4 (27.5-35.2)		.0001	21.3 (15.0-27.7)	24.2 (20.4-28.1)
between r_cSO2 sides, mean (CI 95%)

Means +- SD and P values correspond to 2-sample t tests. Categorical variables, percentages, and frequencies and the differences are compared using a Fisher exact test. The r_cSO2 side dif- ferences are calculated as the absolute value between the left and right sides.

Abbreviation: CI, confidence interval.

the patient’s stroke location. There was a significantly wide r_cSO2 and CBVI side difference between the strokes and the nonstrokes. Among all the pa- tient populations, there was a trend of lower r_cSO2 and CBVI values correlat- ing with the stroke’s pathology. Compared with ischemic stroke patients, hemorrhagic stroke patients had initially lower ipsilateral and contralateral r_cSO2 and CBVI readings that remained consistently lower.

Discussion

This is the first study to investigate specifically left and right r_cSO2 and CBVI monitoring in pediatric patients with symptoms that activated the pediatric stroke alert system or with AMS presenting to a PED caused by an acute stroke. The study has revealed some key results.

Table 3

Stroke alert and overall stroke patients’ left and right r_cSO2 and CBVI trends for 40 minutes

Difference between r_cSO2 sides Stroke alert patients Overall stroke patients

	Positive stroke alert (n = 25), mean (SD)	Negative stroke alert (n = 63)	P	Positive stroke (n = 77)	Negative stroke (n = 133)	P
0 min	37.1 +- (18.3)	3.7 +- (4.0)	b.0001	32.1 +- (18.3)	3.3 +- (3.4)	b.0001
5 min	36.6 +- (15.9)	3.5 +- (3.2)	b.0001	32.6 +- (15.9)	3.2 +- (2.7)	b.0001
10 min	34.4 +- (15.4)	4.7 +- (9.2)	b.0001	32.4 +- (15.4)	4.0 +- (6.7)	b.0001
15 min	27.5 +- (15.1)	3.3 +- (2.7)	b.0001	23.5 +- (12.6)	3.4 +- (3.2)	b.0001
20 min	29.8 +- (12.6)	3.9 +- (3.5)	b.0001	23.8 +- (13.9)	3.4 +- (3.2)	b.0001
25 min	31.9 +- (13.9)	3.2 +- (3.2)	b.0001	24.9 +- (12.5)	3.1 +- (2.9)	b.0001
30 min	31.6 +- (12.5)	3.7 +- (3.7)	b.0001	22.6 +- (12.9)	3.4 +- (3.2)	b.0001
35 min	32.7 +- (13.0)	2.7 +- (2.2)	b.0001	19.7 +- (13.0)	3.2 +- (2.9)	b.0001
40 min	36.4 +- (16.2)	3.5 +- (2.4)	b.0001	21.4 +- (16.2)	3.7 +- (3.2)	b.0001
45 min	27.7 +- (14.3)	3.6 +- (2.4)	b.0001	21.7 +- (14.3)	3.7 +- (2.6)	b.0001

Image of Fig. 1

Fig. 1. Left and right r_cSO2 and CBVI trends for 40 minutes for the stroke alert and overall stroke groups.

First, the results demonstrated immediate and consistent left and right r_cSO2 and CBVI readings in stroke patients. Second, this is the first time the stroke cerebral physiology (hemorrhagic or ischemic) has been ex- amined noninvasively. In the stroke alert and overall strokes, when comparing the stroke side with the nonstroke side, consistently, the stroke side r_cSO2 and CBVI readings were significantly lower. These r_cSO2 and CBVI findings showed the stroke side’s altered cerebral physi- ology. Third, in the stroke alert and overall stroke patients, the differ- ence in the left or right r_cSO2 and CBVI readings highly correlated with identifying radiographically confirmed stroke patients compared with nonstroke patients. Regarding r_cSO2 side differences, the lower side was highly predictive of the stroke location. Fourth, the lowest immedi- ate cerebral left or right r_cSO2 and CBVI readings and trends significantly indicate the type of stroke (hemorrhagic or ischemic). Another finding indicates that patients with left-sided stroke tended to have lower r_cSO2 and CBVI readings compared with patients with right-sided stroke. Lastly, in a small group of hemorrhagic stroke patients, results indicated that they had significantly lower r_cSO2 and CBVI readings on their con- tralateral cerebral side compared with the ischemic stroke patients’ contralateral readings. This could possibly be due to the effects of the primary hemorrhagic site expanding and causing increasing intracranial pressure on the contralateral cerebral tissue affecting the contralateral cerebral perfusion, oxygen delivery, and metabolism. For the first time, a noninvasive method suggests the effects of unilateral strokes on the contralateral cerebral physiology. The study results further dem- onstrate, as in prior studies, the abnormal effects of extravascular blood on the ipsilateral and contralateral cerebral physiology as detected by regional Tissue oxygenation cerebral oximetry.

In ischemic stroke, there is a lack of blood flow [17-24,3]. In hemor- rhagic strokes, the expanding hematoma leads to compression of re- gional tissue, vessels, and distortion and compression of the adjacent tissue leading to decreased perfusion and surrounding Ischemic injury

[17-24,3]. As this expanding hematoma or ischemic area causes in- creased pressure on the surrounding tissue, it effectively decreases re- gional tissue perfusion [19-23]. The stroke pathophysiology, either from occlusion or from vascular disruption, causes the ischemic effects on the stroke’s immediate and surrounding regional tissue producing decreased perfusion, decreased oxygen delivery, increased metabolism, and oxygen extraction/debt causing decreased regional oxygenated he- moglobin [17-24,3]. These 2 different strokes types have significant in- terplay in the surrounding regional tissue perfusion, oxygen delivery system, and tissue metabolism [19-24,3,25]. Each type of stroke effect on cerebral physiology is expected to be dynamic in respect to the na- ture of the stroke, effect on regional tissue, hemispheric and global cere- bral perfusion, and also on the effect on local, hemispheric, and global intracranial pressure [19-24,3,25]. This study’s findings demonstrate these expectations and define in the stroke patients that there were sig- nificantly lower immediate and trending r_cSO2 and CBVI readings. The difference in stroke effects on the ipsilateral and contralateral stroke’s regional cerebral tissue was demonstrated by the significantly lower ip- silateral and contralateral r_cSO2 and CBVI readings in hemorrhagic as compared with ischemic stroke. In a small group of stroke patients with either hemorrhagic or ischemic stroke, the stroke physiology af- fected not only their ipsilateral stroke’s r_cSO2 and CBVI, showing lower readings, but also their contralateral r_cSO2 and CBVI, which were signif- icantly lower compared with the nonstroke patients.

As in prior cardiovascular and neurosurgeries, the application of ce- rebral oximetry with blood volume index has demonstrated its utility in detecting cerebral ischemic and hemorrhagic events [8-16]. The appli- cation of cerebral oximetry with blood volume index in the pediatric stroke population has shown its ability to noninvasively detect changes in not only regional tissue oxidation but also changes in perfusion [8-16]. In comparing strokes with nonstrokes, the detections and trending of the r_cSO2 and CBVI showed a statistically significantly

Image of Fig. 2

Fig. 2. Left and right cerebral r_cSO2 and CBVI graph for individual hemorrhagic and ischemic stroke patient.

lower r_cSO2 and CBVI readings in strokes compared with nonstrokes. These values correlated with stroke location and stroke type identifica- tion. This study also demonstrated that hemorrhagic stroke compared with ischemic stroke has significantly lower initial r_cSO2 and CBVI mea- surements and consistently lower measurements over time. In a small subsection of hemorrhagic strokes, there were significantly lower ipsi- lateral r_cSO2 and CBVI values with concomitant decreases in contralater- al r_cSO2 and CBVI values as a possible result of the increasing intraparenchymal bleeding or expanding hematoma causing increased pressure on the contralateral cerebral hemisphere. In pediatric neuro- logic emergencies with cerebral oximetry monitoring and cerebral pa- thology, the occurrence of low ipsilateral cerebral oximetry readings and concomitant low contralateral cerebral oximetry readings is similar to what occurred in this pediatric stroke physiology [8-16]. The hemor- rhagic and ischemic stroke r_cSO2 and CBVI readings showed significantly similar patterns [8-16]. These r_cSO2 and CBVI differences between strokes and nonstrokes have demonstrated the significance of the uni- lateral stroke physiology concomitant effect on the contralateral nonstroke cerebral physiology.

Even with the increasing awareness of pediatric strokes and devel- opment of stroke alert systems, there still is a significant delay in recog- nition and imaging [2-7,21-23,20,24-26]. The key problem is the nonneurologist’s limited pediatric stroke knowledge causing misinter- pretations of child’s stroke symptoms resulting in stroke recognition delay [21-23,20,24,3,25,26]. The use of an objective detection and mon- itoring system will decrease this subjectivity, will decrease stroke recog- nition time in all patients, and has potential for decreasing imaging delays and improving the efficiency of a stroke alert system.

To the nonneurologist, the application of cerebral oximetry with or

without CBVI monitoring in pediatric patients with symptoms of neuro- logic emergencies has shown a potential benefit as an adjunct tool for streamlining differential diagnosis, indicating abnormal unilateral or

bilateral cerebral physiology, and potential increased intracranial pres- sure. This study has demonstrated that using cerebral oximetry moni- toring with or without CBVI in pediatric patients with symptoms of neurologic emergencies can immediately and objectively detect the presence of unilateral stroke with or without concomitant contralateral hemispheric abnormal physiology by correlating it with the radiologic pathology. These between-hemispheric r_cSO2 differences were consis- tent from 0 to 60 minutes, further demonstrating the utility of cerebral oximetry with or without CBVI monitoring for immediately detecting unilateral strokes The r_cSO2 and CBVI side differences have shown the Potential causes for the unilateral or bilateral hemispheric abnormal ce- rebral physiology as a possible stroke, or other cerebral pathology (epi- dural or subdural).

Cerebral oximetry with or without CBVI index readings has shown its potential as an adjunct tool in a pediatric stroke alert system as well as in the pediatric AMS patients who possibly have a stroke. These findings do not suggest or imply that cerebral oximetry will re- place or negate the necessity for radiologic imaging in neurologic emer- gencies, but has shown its potential in improving recognition time and imaging time for pediatric neurologic emergencies, especially in stroke patients [14-20].

Using cerebral oximetry by the nonneurologist, the patient’s r_cSO2 side difference less than 10 or one or both sides greater than 49% r_cSO2 readings would imply abnormal hemispheric pathology, expediting the patient’s diagnosis, neuroresuscitation endeavors, and radiologic imaging.

Limitations

To analyze the effect of active focal or Generalized seizures on the patient’s left and right r_cSO2 and CBVI readings in either the stroke alert or the overall population was not possible due to insufficient pa- tient numbers for a statistical analysis. The stroke times with these

Table 4

Probability of stroke, first 60 minutes of left and right r_cSO2 readings, and r_cSO2 side differences for stroke alert (n = 77; stroke alert [n = 25] and overall stroke [n = 52]) and negative stroke (n = 133)

Stroke alert patients (n = 88): stroke (+) (n = 25), stroke (-) (n = 63)

		P?	Sensitivity	Specificity	PPV	NPV	Youden Index
St	roke (+): r_cSO2 side differences 4.9	.03	1.0 (0.87-1)	0.86 (0.8-0.9)	0.74 (0.6-0.85)	1.0 (0.93-0.99)	0.86
	6.5	.15	0.96 (0.8-0.99)	0.97 (0.9-0.99)	0.92 (0.8-0.98)	0.98 (0.91-0.98)	0.93
	8.3	.56	0.96 (0.8-0.99)	0.98 (0.9-0.99)	0.96 (0.8-0.99)	0.98 (0.92-0.99)	0.94
	9.2	.8	0.92 (0.76-0.98)	1.0 (0.94-1)	1.0 (0.9-1)	0.96 (0.9-0.99)	0.92
	16.7	1.0	0.9 (0.7-0.96)	1.0 (0.94-1)	1.0 (0.85-1)	0.96 (0.9-0.98)	0.9
	24.9	1.0	0.81 (0.7-0.9)	1.0 (0.97-1.0)	1.0 (0.94-1.0)	0.93 (0.9-0.94)	0.8
St	roke (+): left r_cSO2 mean
b54.3%		.5	0.76 (0.6-0.9)	1 (0.94-1)	1 (0.83-1)	0.91 (0.8-0.96)	0.76
b47.6%		.7	0.6 (0.41-0.8)	1 (0.94-1)	1 (0.8-1)	0.86 (0.7-0.9)	0.6
St	roke (+): right r_cSO2 mean
b54.2%		.5	0.32 (0.2-0.52)	1 (0.94-1)	1 (0.68-1)	0.79 (0.69-0.86)	0.34
b35%		.75	0.24 (0.12-0.4)	1 (0.943-1)	1 (0.61-1)	0.77 (0.7-0.85)	0.24
Overall stroke patients (n = 220): overall stroke (+) (n = 52), stroke alert (+) (n = 25), overall stroke (-) (n = 133)
		P?	Sensitivity	Specificity	PPV	NPV	Youden Index
St	roke (+): r_cSO2 side differences
	5.6	.02	0.99 (0.93-0.99)	0.88 (0.8-0.93)	0.82 (0.72-0.9)	9.9 (0.96-1)	0.87
	6.5	.07	0.98 (0.92-0.99)	0.94 (0.89-97)	0.91 (0.8-96)	0.99 (0.96-1.0)	0.93
	7.7	.23	0.99 (0.9-0.99)	0.99 (0.95-0.99)	0.97 (0.9-0.99)	0.99 (0.96-1)	0.97
	9.2	.67	0.97 (0.9-0.99)	1.0 (0.97-1.0)	1.0 (0.95-1.0)	0.99 (0.95-0.99)	0.97
	14.4	.99	0.93 (0.9-0.97)	1.0 (0.97-1.0)	1.0 (0.95-1.0)	0.97 (0.9-0.98)	0.93
	25.0	1.0	0.74 (0.6-0.83)	1.0 (0.97-1.0)	1.0 (0.94-1)	0.88 (0.8-0.92)	0.75
St	roke (+): left r_cSO2
b50.5%		.65	0.52 (0.41-0.63)	1 (0.97-1)	1 (0.91-1)	0.8 (0.7-0.85)	0.46
b41.4%		.78	0.3 (0.21-0.42)	1 (0.97-1)	1 (0.9-1)	0.7 (0.7-0.8)	0.35
Stroke (+): right r_cSO2
b50%		.63	0.32 (0.22-0.44)	1 (0.97-1)	1 (0.86-1)	0.74 (0.66-0.79)	0.6
b33.2%		.87	0.16 (0.1-0.3)	1 (0.97-1)	1 (0.77-1)	0.69 (0.6-0.74)	0.2

Stroke alert (n = 88)	Positive likelihood ratio	Stroke alert (n = 88)	Negative likelihood ratio
Left r_cSO2 b 58.2%	47.7	Left r_cSO2 b 58.2%	0.24
Right r_cSO2 b 58.2%	27.5	Right r_cSO2 b 58.2%	0.57
Between (left or right) r_cSO2 side difference N 8.3	60.0	Between (left or right) r_cSO2 side difference N 8.3	0.04
Overall stroke (n = 210)	Positive likelihood ratio	Overall stroke (n = 210)	Negative likelihood ratio
Left r_cSO2 b 54%	71.4	Left r_cSO2 b 54%	0.44
Right r_cSO2 b 54.4	58.5	Right r_cSO2 b 54.5	0.55
Between (left or right) r_cSO2 side difference N 8.4	118.5	Between (left or right) r_cSO2 side difference N 8.4	0.05
Overall stroke minus 4 hemorrhagic stroke patients (n = 196)	Positive likelihood ratio	overall stroke minus 4 hemorrhagic stroke	Negative likelihood ratio
		patients (n = 196)
Left r_cSO2 b54%	70.3	Left r_cSO2 b 54%	0.44
Right r_cSO2 b 54.5%	54.8	Right r_cSO2 b54.5%	0.58
Between (left or right) r_cSO2 side difference N 8.3	123.3	Between (left or right) r_cSO2 side difference N 8.3	0.01

Abbreviations: PPV, positive predictive value (95% confidence interval); NPV, negative predictive value (95% confidence interval).

* Probability of a stroke.

r_cSO2 and CBVI readings could not be analyzed due to the inconsistency for the signs and symptoms times from the PED EMR.

Conclusion

This preliminary cerebral oximetry study in pediatric stroke patients demonstrated abnormal stroke hemispheric cerebral physiology. Signif- icantly lower left and or right cerebral r_cSO2 and CBVI readings were present in pediatric stroke patients compared with nonstroke patients and correlated with stroke location. The r_cSO2 side (left or right) differ- ences and r_cSO2 less than 50 objectively and consistently detected ab- normal cerebral physiology, stroke, location, and stroke types (hemorrhagic or ischemic). An r_cSO2 side difference greater than 10 and or less than 49% r_cSO2 had a 100% positive predictive value for de- tecting abnormal physiology and stroke location. Cerebral oximetry

with CBVI has shown its potential as an objective screening tool for identifying stroke’s physiology, location, and stroke types in tandem with a pediatric stroke alert system and in AMS patients. In pediatric pa- tients with signs or symptoms suggestive of stroke, AMS, or other neu- rologic emergencies, using cerebral oximetry by the nonneurologist as an adjunct screening tool should significantly improve detection of ab- normal cerebral physiology, improve stroke alert activation and stroke’s location and type, and improve imaging time.

Acknowledgments

The authors want to thank all Vanderbilt Children’s Pediatric Emer- gency Department nurses and Respiratory therapists for their active support in the stroke alert activation system.

Image of Fig. 3

Fig. 3. Left r_cSO2, right r_cSO2, difference between r_cSO2 sides, and ROC curve time interval every 5 seconds, from 0 to 10 minutes for stroke alert, overall stroke, and overall stroke patients without 4 hemorrhagic outliers with side differences of 3.1 but left and right r_cSO2 sides less than 49% r_cSO2 reading.

References

AJEM