A Multicenter Pilot Study Assessing Regional Cerebral Oxygen Desaturation Frequency During Cardiopulmonary Bypass and Responsiveness to an Intervention Algorithm

Abstract

BACKGROUND

The purpose of this multicenter pilot study was to: (1) determine the frequency of regional cerebral oxygen saturation (rSco₂) desaturations during cardiac surgery involving cardiopulmonary bypass (CPB); (2) evaluate the accuracy of clinician-identified rSco₂ desaturations compared with those recorded continuously during surgery by the near-infrared spectroscopy (NIRS) monitor; and (3) assess the effectiveness of an intervention algorithm for reversing rSco₂ desaturations.

METHODS

Two hundred thirty-five patients undergoing coronary artery bypass graft and/or valvular surgery were enrolled at 8 US centers in this prospective observational study. NIRS (Invos™ 5100C; Covidien) was used to monitor rSco₂ during surgery. The frequency and magnitude of rSco₂ decrements >20% from preanesthesia baseline were documented, and the efficacy of a standard treatment algorithm for correcting rSco₂ was determined. The data from the NIRS monitor were downloaded at the conclusion of surgery and sent to the coordinating center where the number of clinician-identified rSco₂ desaturation events was compared with the number detected by the NIRS monitor.

RESULTS

The average rSco₂ obtained at baseline (mean ± SD, 61% ± 11%; 99% confidence interval, 57%–65%) and during CPB (62% ± 14%; 57%–67%) was not different. However, rSco₂ after separation from CPB (56% ± 11%; 53%–60%) was lower than measurements at baseline and during CPB (P < 0.001). During CPB, rSco₂ desaturations occurred in 61% (99% confidence interval, 50%–75%) of patients. The area under the curve for product of magnitude and duration of the rSco₂ was (mean ± SD, 145.2; 384.8% × min). Clinicians identified all patients with an rSco₂ desaturation but identified only 340 (89.5%) of the 380 total desaturation events. Of the 340 clinician-identified rSco₂ desaturation events, 115 resolved with usual clinical care before implementation of the treatment algorithm. For the remaining 225 events, the treatment algorithm resulted in resolution of the rSco₂ desaturation in all but 18 patients.

CONCLUSIONS

This multicenter pilot study found that 50% to 75% of patients undergoing cardiac surgery experience one or more rSco₂ desaturations during CPB. Nearly 10% of desaturation events were not identified by clinicians, suggesting that appropriate alarming systems should be adopted to alert clinicians of such events. The intervention algorithm was effective in reversing clinically identified rSco₂ desaturations in the majority of events.

Near-infrared spectroscopy (NIRS) is used increasingly during cardiac surgery to monitor regional cerebral oxygen saturation (rSco₂) as an indicator of cerebral perfusion adequacy, particularly during cardio-pulmonary bypass (CPB).^1,2 Unlike pulse oximetry, NIRS monitoring does not distinguish arterial from venous blood. Because approximately 70% to 75% of cerebral blood volume is venous blood, rSco₂ provides an indicator of O₂ supply versus-demand balance.³ Case reports and observational studies have linked episodes of rSco₂ desaturation during coronary artery bypass graft (CABG) surgery with risk for stroke and postoperative cognitive dysfunction.^1,2 These studies, although, have limitations that confound the ability to interpret whether decrements in rSco₂ during surgery are clearly linked with adverse neurologic outcomes.^1,2 Low rSco₂ may merely serve as an epiphenomenon identifying patients with cerebral vascular disease or limited cardio-pulmonary reserve who are at high risk of complications.⁴ Importantly, limited data are available on the effectiveness of interventions aimed at correcting rSco₂ desaturations and whether such interventions lead to improved patient neurologic outcomes.^5–10

Many clinicians, administrators, and health care payers may desire a higher level of evidence than what is currently available on the benefits of NIRS monitoring before it is adopted for routine use in patients undergoing cardiac surgery. Such evidence would ideally come from multicenter prospective studies that compare the effect of interventions to correct rSco₂ desaturations with the effect of current standard of care on the frequency of neurologic complications. As a first step to achieve this goal, in this pilot study, we assembled a group of investigators at 8 US institutions to: (1) determine the frequency of rSco₂ desaturations during cardiac surgery with CPB; (2) compare the accuracy of clinician-identified rSco₂ desaturations with that of continuous NIRS monitoring during surgery; and (3) assess the effectiveness of an intervention algorithm for reversing rSco₂ desaturations. We hypothesized that rSco₂ desaturations are frequent during CPB and that a multicenter network of investigators can successfully follow an intervention algorithm and accurately report their data to a central data-coordinating center. As a proof of concept construct, the intent of the study was not to link desaturations or their correction to specific outcomes. However, these data may be of value to clinicians for establishing reference ranges of rSco₂ and for evaluating potential interventions for treating rSco₂ desaturations.

METHODS

The procedures of this study were approved by The Johns Hopkins Medical Institutions Human Research Review Board. The procedures further met with the approval of the IRBs of 7 other enrolling sites: Beth Israel Deaconess Hospital, Harvard Medical School, Boston, MA; Brigham and Women’s Hospital/Cape Cod Hospital, Harvard Medical School, Hyannis, MA; Duke University School of Medicine, Durham, NC; Emory University School of Medicine, Atlanta, GA; Mayo University School of Medicine, Rochester, MN; University of Virginia School of Medicine, Charlottesville, VA; and University of Maryland School of Medicine, Baltimore, MD. All study procedures were performed only after receiving individual written informed patient consent.

Data Collection and Patient Monitoring

From November 8, 2013, to March 4, 2015, 239 patients were enrolled at the 8 study sites; this analysis focuses on 235 patients with nonmissing demographic and medical history data. Male and female patients were eligible for inclusion if they were >50 years old and undergoing primary or reoperative CABG and/or cardiac valve surgery. Enrolled patients at each research site were assigned a sequential 3-digit study number and a 4-character alpha code. Study data were collected on a paper case report form provided by the coordinating center (The Johns Hopkins University School of Medicine). The collected data included preoperative demographics and medical history. Medical conditions such as hypertension, prior stroke or transient ischemic attack, congestive heart failure, and other diseases were based on diagnosis by the patient’s primary medical team.

The patients received institutionally standard anesthesia and surgery care, including direct arterial blood pressure monitoring. Before the induction of general anesthesia, the patients were connected to an NIRS monitor (Invos™ 5100C; Covidien, Boulder, CO). The forehead was cleansed with an alcohol swab, and 2 self-adhesive probes were attached according to manufacturer’s guidelines. Baseline rSco₂ readings were obtained, while the patients were breathing room air. Nonpulsatile CPB flow was maintained between 2.0 and 2.4 L/min/m². The patients were managed with α-stat pH management. There was no standardized algorithm for blood transfusion.

rSco₂ Data Collection and Intervention Algorithm

An episode of rSco₂ desaturation was defined as a >20% decline from baseline detected from either the right or the left sensor. During surgery, the time of any rSco₂ desaturation was recorded on the study data sheet. Management of these rSco₂ desaturations followed an algorithm as proposed by Murkin et al.¹¹ and Denault et al.¹² (Fig. 1). Clinicians and perfusionists were instructed on the study procedures and the treatment algorithm for rSco₂ desaturation. They then recorded each sequential intervention and the time when the rSco₂ desaturation was corrected. Subsequent rSco₂ desaturations were similarly treated and recorded when they occurred. At the conclusion of the surgery, rSco₂ data files from the NIRS monitor were downloaded to a USB driver.

Figure 1.

Treatment algorithm for regional cerebral oxygen saturation (rSco₂) desaturations (>20% decline from baseline) during cardiopulmonary bypass. Adopted from Denault et al.¹² CPB = cardiopulmonary bypass; MAP = mean arterial pressure.

Study demographic and other data were recorded onto the case report forms, which was then transferred to an Excel spreadsheet (Microsoft, Inc., Redmond, WA) format. Once completed, the electronic case report form was uploaded onto an Internet, study-specific, encrypted Web site supported by MedDium, Inc. (Bethesda, MD, https://www.meddium.com/). The rSco₂ data from the NIRS monitor were also uploaded to the same Web site. Only deidentified data for a patient assigned a study number and alpha code were uploaded. The study data could be reviewed only by authorized personnel at the coordinating center.

Sample Size Calculation and Data Analysis

The primary outcome of this pilot study was the combined frequency of rSco₂ desaturations from all enrolling centers. The frequency of this event is not known and was the aim of this study. We sought to enroll 28 patients at each site or a total of 225 patients. We added an additional 14 patients to this sample size as a buffer in case any protocol was incomplete. This sample size was chosen to be similar to that reported by Murkin et al.¹¹ who observed rSco₂ desaturation in 56% of 200 patients. In that study, the interventions to correct the rSco₂ desaturations were effective in 80.4% of episodes. Thus, we expected 126 episodes of rSco₂ desaturation during surgery in this pilot study.

Demographic, prior medical, and preoperative data are presented as mean ± SD or median (range) for continuous variables and frequencies for categorical variables and compared using 1-way analysis of variance, Kruskal-Wallis, or χ² tests across the study sites. Pairwise comparisons after significant overall test were performed considering multiple comparisons using the Tukey method. The overall proportion of desaturation events was estimated for the entire sample and for each site. Ninety-nine percent confidence intervals (CI) are reported for all estimates; they were calculated using generalized linear models with site-specific robust variance estimates to account for potential nonindependence of observations within sites. rSco₂ at baseline, the start of CPB, and the end of CPB was summarized and compared using generalized linear model with time (start and end of CPB versus baseline) as the primary independent variable. The model was estimated using generalized estimating equations with exchangeable working correlation structure to account for within-patient correlation of rSco₂ levels over time. We evaluated the accuracy of clinician-identified rSco₂ desaturations by comparing the number and the frequency of investigator-reported rSco₂ saturations to the number recorded continuously by the NIRS monitor during surgery. The severity of rSco₂ desaturations was determined as the area under the curve of magnitude and duration of rSco₂ desaturations. The latter was determined off-line at the coordinating center from the uploaded rSco₂ data. The number and specific type of each sequential intervention for rSco₂ desaturation were determined. Compliance with the intervention algorithm for rSco₂ desaturation was defined as a clinician’s successful correction of each rSco₂ desaturation episode identified by retrospective review of the patient’s NIRS monitor data file.

RESULTS

Data were successfully collected from 235 patients. The medical and operative characteristics of the enrolled patients are shown in Table 1 and the Supplemental Table 1 (Supplemental Digital Content, http://links.lww.com/AA/B408). There were differences among the enrolling sites in some characteristics, including patient age, race, history of congestive heart failure, type of surgical procedure, and duration of CPB and aortic cross-clamping.

Table 1.

Patients’ Medical History and Operative Data

Characteristic	n = 235	P value between sites
Agea (y)	68.3 ± 9.1	0.036
Male/femaleb	161 (68.5)/74 (31.5)	0.117
Heighta (cm)	171.5 ± 9.6	0.792
Weighta (kg)	85.6 ± 20	0.043
Raceb		<0.001
Caucasian	204 (86.8)
African American	25 (10.6)
Asian	3 (1.3)
Prior strokeb	17 (7.2)	0.160
Prior transient ischemic attackb	11 (4.7)	0.154
Hypertensionb	180 (76.6)	0.800
Diabetesb	72 (30.6)	0.921
Insulinb	25 (10.6)	0.922
Chronic obstructive pulmonary diseaseb	31 (13.2)	0.806
Congestive heart failureb	63 (26.8)	<0.001
Tobacco smokingb		0.414
Current	20 (8.5)
Former	108 (46.0)
Atrial fibrillationb	49 (20.9)	0.213
Prior myocardial infarctionb	57 (24.3)	0.249
Medicationb
Aspirin	181 (77.0)	0.319
Clopidogrel	30 (12.8)	0.108
Prasugrel	1 (0.4)
Ticagrelor	3 (1.3)
Coumadin	19 (8.1)	0.862
β-adrenergic receptor blockers	171 (72.8)	0.081
Angiotensin-converting enzyme inhibitor	78 (33.2)	0.524
Surgical procedureb
CABG surgery	129 (54.9)	0.003
1-vessel	24 (10.2)	0.038
2-vessel	27 (11.5)
3-vessel	49 (20.9)
4-vessel	27 (11.5)
5-vessel	2 (0.9)
Aortic valve replacement	85 (36.2)	0.777
Mitral valve replacement/repair	61 (26.0)	0.008
Tricuspid valve replacement/repair	13 (5.5)	0.020
Prior cardiac surgery	24 (10.2)	0.006
Duration of CPB (min)c	102.0 (31.0–329)	0.002
Duration of aortic cross-clamping (min)c	74.0 (0–242)	0.110
Patients at each enrolling siteb
Beth Israel Deaconess Hospital	28 (11.9)
Cape Cod Hospital	28 (11.9)
Duke University Hospital	28 (11.9)
Emory University Hospital	28 (11.9)
Johns Hopkins  Hospital	37 (15.7)
Mayo Clinic	28 (11.9)
University of  Maryland Hospital	30 (12.9)
University of Virginia	28 (11.9)

CABG = coronary artery bypass graft; CPB = cardiopulmonary bypass.

^aMean ± SD.

^bNumber (%).

^cMedian (range).

Baseline rSco₂ and results obtained during and after separation from CPB are shown in Figure 2. Average baseline rSco₂ for the left, right, and the average of the left and right frontal lobes, respectively, were (mean ± SD; 99% CI) 61% ± 11% (57%–65%), 61% ± 12% (58%–65%), and 61% ± 11% (57%–67%). During CPB, the average rSco₂ measurements were 62% ± 14% (57%–66%), 62% ± 15% (57%–67%), and 62% ± 14% (57%–67%) for the left, right, and the average of the left and right sides, respectively. After CPB, these results were 56% ± 11% (53%–60%), 56% ± 12% (52%–60%), and 56% ± 11% (53%–60%). The average rSco₂ during CPB was no different than baseline (P = 0.338), but rSco₂ after separation from CPB was lower than that at baseline and that recorded during CPB after adjustment for site (P < 0.001). rSco₂ values differed among enrolling sites. The average rSco₂ was lower at Duke University Hospital than that at the Mayo Clinic before (P < 0.001), during (P = 0.001), and after CPB (P = 0.004). The average rSco₂ also was lower at Duke University Hospital than that at The Johns Hopkins Hospital before CPB (P = 0.001). Thirty-one (15%; 99% CI, 5%–41%) patients had baseline left-sided rSco₂ <50%, and 34 (15%; 99% CI, 6%–37%) had baseline right-sided rSco₂ <50%, while breathing room air.

Figure 2.

Regional cerebral oxygen saturation (rSco₂) at baseline, the start of cardiopulmonary bypass (CPB), and the end of CPB. The data (mean ± SD) were combined from all enrolling sites.

The frequency of rSco₂ desaturations during CPB for all patients, as well as that observed at each enrolling site, is shown in Table 2. One or more rSco₂ desaturations occurred in 61% (99% CI, 50%–75%) of patients during CPB. A total of 380 desaturation events were detected from data downloaded from the NIRS monitor at the conclusion of surgery. The frequency of desaturation varied among enrolling sites (P = 0.023). After adjusting for multiple comparisons, we identified a difference in the frequency of rSco₂ desaturation between the Mayo Clinic and Duke (P = 0.004), Emory University and the Mayo Clinic (P = 0.027), and the Mayo Clinic and the University of Virginia (P = 0.032). Clinicians detected all patients with at least 1 rSco₂ desaturation during CPB, but they identified 340 (89.5%) of the 380 desaturation events.

Table 2.

The Number and Frequency of One or More rSco₂ Desaturations (>20% Decline from Baseline) and the Area Under the Curve (AUC) for Cumulative Duration of Desaturations (% × min) Occurring During Cardiopulmonary Bypass

	Total rSco2desaturations	Unilateral desaturation	Bilateral desaturation	Left AUCa	Right AUCa	Left and right AUCa
Total number of patients with at least one event	138 (61%)	193	187	140.7 (358.0)	149.7 (462.9)	145.2 (384.8)
Beth Israel Deaconess	14 (50%)	20	19	74.3 (164.6)	30.8 (66.6)	52.6 (97.7)
Cape Cod Hospital	15 (54%)	14	13	141.8 (321.9)	138.2 (303.3)	140.0 (308.4)
Duke University	18 (64%)	26	28	106.2 (288.2)	90.0 (134.2)	98.1 (193.4)
Emory University	23 (82%)	31	36	173.2 (281.8)	137.1 (187.9)	155.1 (202.1)
Johns Hopkins	23 (79%)	34	25	199.2 (313.5)	239.4 (459.6)	219.3 (328.3)
Mayo Clinic	12 (43%)	25	19	107.6 (174.6)	71.1 (142.8)	89.4 (156)
University of Maryland	15 (54%)	18	24	219.3 (754)	235.4 (819.7)	227.3 (785.3)
University of Virginia	18 (64%)	25	23	101.1 (218.8)	250.0 (811.1)	175.5 (510.1)

^aMean (SD).

rSco₂ = Regional cerebral oxygen saturation

The effectiveness of each intervention for reversing rSco₂ desaturation is shown in Table 3. There was spontaneous resolution of the desaturation in 115 events. For the remaining 225 events, treating hypotension was the most effective strategy for correcting rSco₂ desaturation. Desaturations remained unresolved during CPB in 18 (5%; 99% CI, 3%–10%) of the 340 events. Of the patients with resolved desaturations, 1 intervention was needed in 70 events, 2 interventions in 57 events, 3 interventions in 28, and >3 interventions were needed in 65 events.

Table 3.

Efficacy of Interventions to Correct Decrements in Regional Cerebral Oxygen Saturation (rSco₂) of >20% from Baseline for the 340 Clinician-Identified Events

	Intervention-corrected rSco2 desaturation
Treat hypotension	67 (29.8%)
Increase Fio2%	35 (15.6%)
Normalize CPB flow	32 (14.2%)
RBC transfusion	31 (13.8%)
Decrease CPB “Sweep Speed”	25 (11.1%)
Deepen anesthesia	24 (10.7%)
Adjust CPB cannula	18 (8.0%)
Reposition head to midline	6 (2.7%)

There was resolution of 115 desaturation events with usual clinical care and not the treatment algorithm. The data are listed as the number of desaturation of events that were not spontaneously corrected with percentage based on the event effectiveness versus all events that responded to any intervention (n = 225). In 18 (8%) patients, the rSco₂ was unresolved despite interventions.

CPB = cardiopulmonary bypass; RBC = red blood cell.

DISCUSSION

In this study, we observed that the average rSco₂ during CPB was no different from that at baseline. However, average rSco₂ decreased after separation from CPB compared with values measured at baseline and during CPB. Desaturation in rSco₂ occurred in 61% (99% CI, 50%–75%) of patients. NIRS monitoring recorded 380 desaturation events, of which clinicians identified 340 (89.5%). In all but 5% (99% CI, 3%–10%) of patients, there was either resolution of rSco₂ desaturations with usual clinical care or they were corrected by the intervention algorithm.

Many clinicians have embraced rSco₂ monitoring as a means for judging the adequacy of cerebral perfusion during CPB in an effort to reduce the risk for adverse neurologic events. Currently, only a low level of evidence supports the use of rSco₂ monitoring for this purpose. In a previously published systematic review, our group identified 43 articles involving 6399 patients that evaluated the relationship between rSco₂ monitoring and adverse neurologic outcomes after cardiac surgery.¹ The identified citations included 13 case reports and 27 observational studies. These reports provided anecdotal evidence that rSco₂ desaturations can be indicative of rare CPB cannula malposition, allowing for timely surgical adjustments. Although the data were not consistent, some observational studies linked rSco₂ desaturations during CPB and postoperative cognitive dysfunction or stroke. Interpretation of these studies, however, is confounded by small sample sizes, the use of a limited psychometric testing battery in many of the studies, the lack of patient testing after hospital discharge, use of historical controls, the failure to risk adjust in the data analysis, and the use of monitors from multiple manufacturers.

In our systematic review, we identified 2 studies in which patients were prospectively randomized to an intervention to treat rSco₂ desaturations during CPB.^11,13 Poor adherence to the intervention protocol in the study by Slater et al.,¹³ however, limits interpretation of those results. Murkin et al.¹¹ randomly assigned 200 patients undergoing CABG surgery to 2 groups. In one, patients with rSco₂ desaturations >30% below baseline were treated with a stepped intervention algorithm. These patients were compared with a control group for whom rSco₂ data were collected but not displayed to clinicians. In that study, the patients in the control group had longer rSco₂ desaturations (P = 0.014), a higher frequency of a composite organ complication end point (death, ventilation >48 hours, stroke, myocardial infarction, return for re-exploration [P = 0.048], and longer hospitalization in the intensive care unit [P=0.029]) than did patients in the intervention arm. Episodes of rSco₂ desaturation were observed in 56 of 100 patients in the intervention group, and the success rate of the rSco₂ desaturation intervention was 80.4%. The study by Murkin et al. did not have adequate power to detect a difference in neurologic complications. Moreover, the sample size calculations assumed a complication rate in the control arm of 40%, whereas the observed rate was 30%.

A higher level of evidence than is currently available seems desirable to justify routine rSco₂ monitoring for the approximately 600,000 patients who undergo CABG and/or valvular surgery annually in the United States.¹⁴ Its use may be justified given the devastating consequences of stroke and cognitive complications after cardiac surgery, including high costs of care for affected patients. For example, postoperative stroke alone adds an average of 15 days to the length of hospitalization after cardiac surgery, increases hospital costs by >$35,000, and increases the likelihood of discharge to an extended care facility after surgery.^15–17 In this study, we demonstrated the frequency of rSco₂ desaturations, the feasibility of multicenter collection of rSco₂ monitoring data, and the application of an intervention algorithm for desaturations. These results would be useful in the design of a comparative effectiveness trial that evaluates the cost-benefit relationship of rSco₂ monitoring in patients undergoing cardiac surgery.

There is no universal definition of rSco₂ desaturation. Widely reported definitions are based largely on older data obtained from patients undergoing carotid endarterectomy surgery. In those studies, decrements in rSco₂ between 5% and 20% from baseline or values <50% ipsilateral to carotid artery cross-clamping were associated with reductions in transcranial Doppler–measured cerebral blood flow velocity, electroencephalogram slowing, alterations in somatosensory-evoked potentials, or mental status changes when surgery was performed under regional anesthesia.¹ In this study, we defined rSco₂ desaturation as a >20% decline from baseline but did not use a threshold value such as decrements <50%. Many patients have baseline rSco₂ <50% that may result from factors other than acute reduction in cerebral oxygenation, such as low cardiac output, pulmonary disease, cerebral vascular disease, and anemia. Indeed, in this study, baseline rSco₂ <50% was observed on the left side in 13% of patients and on the right side in 14% of patients, before induction of anesthesia. We observed a baseline averaged rSco₂ of 61% ± 11% (99% CI, 57%–65%). In a single-center study of 250 CABG surgery patients, the reported ranges of rSco₂ before cardiac surgery varied between 47% and 83%.⁵ In a larger prospective study of 1178 patients, preoperative room air rSco₂ ranged from 57% to 67% for survivors but 47% to 62% for nonsurvivors after CABG surgery.⁴ We did observe differences in the baseline rSco₂ value among enrolling sites that might be explained by differences in patient age, race, or preoperative medical conditions among the centers. Differences in the frequency of desaturation and greater responsiveness to interventions in our study compared with those reported by Murkin et al.¹¹ might be explained, in part, by the different definitions of desaturation (>20% from baseline versus >30% decrement).

In our study, clinicians were not blinded to rSco₂ data. We observed that there was resolution of the clinician-identified rSco₂ desaturation in 115 (34%) of 340 events with usual clinical care. We did not include a threshold duration in our definition of desaturation. Thus, many of these events with resolution before implementation of the treatment algorithm were likely of short duration. Clinicians might have further intervened with corrective maneuvers (e.g., increasing CPB flow after transient reduction) before institution of the algorithm. Furthermore, our intent was to simulate a usual clinical environment of NIRS monitoring to enhance the external validity of our findings. Thus, we did not specify a protocol for setting the alarms on the NIRS monitor. Regardless, the treatment of hypotension was the most effective intervention at reversing rSco₂ desaturations. In the randomized study by Murkin et al., raising blood pressure was effective in 62% of cases, and increasing CPB flow was the second most effective intervention.¹⁸ We did not record which medications were used to raise blood pressure. Drugs with vasoconstrictive properties potentially could influence rSco₂ readings indirectly as a result of their effects on the scalp circulation. This potential effect is related to the imprecise subtraction algorithms used by NIRS monitors, whereby light absorption from superficial tissue is subtracted from absorption in deeper tissue to yield frontal lobe rSco₂^.1,19,20 Hypothetically, vasoconstriction of scalp blood vessels could reduce rSco₂ readings regardless of any benefit of higher blood pressure on cerebral blood flow. However, in this investigation, we observed correction of rSco₂ desaturation with raised blood pressure. Regardless of these theoretical concerns, investigations during carotid endarterectomy surgery found that changes in rSco₂ were 87.5% sensitive and 100% specific for identifying changes in transcranial Doppler blood flow velocity during selective clamping of the internal carotid artery. At the same time, rSco₂ monitoring had 0% sensitivity and specificity for identifying laser Doppler blood flow velocity of the scalp during selective clamping of the external carotid artery.²¹

We observed that 57% (i.e., 193 of 340) of clinician-detected rSco₂ desaturations were unilateral. This observation may explain why 3% of desaturation events responded to a repositioning of the head to the midline. Presumably, head position away from the midline might impede cerebral venous return and reduce cerebral perfusion pressure. Malposition of CPB venous or arterial cannulae might also explain some cases of unilateral rSco₂ desaturation. Repositioning of CPB cannulae was effective in 8% of cases. The most effective interventions for rSco₂ desaturation collectively were those aimed at improving cerebral oxygen delivery included increasing the FIo₂% of gas delivery, avoiding the cerebral vasoconstrictive effects of hypocapnia by decreasing gas flow to the membrane oxygenator (i.e., decreasing “sweep speed”), ensuring appropriate CPB flow, and increasing oxygen-carrying capacity with red blood cell transfusion. These efforts were effective in 55% of desaturation events that were not spontaneously corrected with usual clinical care.

In this study, we did not relate rSco₂ desaturations or their treatment to patient outcomes. The small sample size and limited funding available did not permit our team to collect such outcomes. Regardless, this study was meant as a pilot to establish feasibility for performing such a study. Finally, our definition of desaturation did not include the duration that rSco₂ was >20% below baseline. Although short and long desaturations were considered equally, we did quantify severity by calculating area under the curve of events.

In conclusion, in this multicenter pilot study, we found that between 50% and 75% of patients undergoing cardiac surgery experience one or more rSco₂ desaturations during CPB. Nearly 10% of desaturation events were not identified by clinicians, suggesting that appropriate alarming systems should be adopted to alert clinicians of such events. Acute rSco₂ desaturation events either spontaneously corrected with usual clinical care or in response to the intervention algorithm in all but 18 (5%) of 340 clinician-identified events.