- Email: [email protected]
Skin pigmentation interferes with the clinical measurement of regional cerebral oxygen saturation
BJA Advance Access published October 27, 2014 British Journal of Anaesthesia Page 1 of 5 doi:10.1093/bja/aeu335
Skin pigmentation interferes with the clinical measurement of regional cerebral oxygen saturation X. Sun 1*, J. Ellis 1, P. J. Corso 1, P. C. Hill1, F. Chen2 and J. Lindsay3 1
Department of Cardiac Surgery, 2 Medstar Cardiovascular Research Network, and 3 Department of Cardiology, Medstar Washington Hospital Center, 110 Irving Street, Room Suite 1E RM1223, Washington, DC 20010, USA * Corresponding author. E-mail: [email protected]
Editor’s key points
† The authors studied preoperative rSO2 values in a cardiac surgery population. † Skin pigmentation was found to independently influence preoperative rSO2. † Low preoperative rSO2 values were associated with worse mortality in all racial groups.
Methods. From our ongoing departmental registry, 3282 consecutive patients underwent cardiac surgery between 2010 and 2012 and their pre-induction measurements of rSO2 were available. Of these, 2096 identified themselves as Caucasian (Cauc) and 1186 as African-American (AA). Pre-induction rSO2, clinical and operative features were compared. Results. Clinical and operative details of these patients differed widely between the two populations. High-risk features were more common in AA patients, but no difference in mortality was observed (4.8% in AAs vs 4.7% in Caucs, P¼0.87). Preprocedure rSO2 was systematically higher in Cauc (65.5% vs 53.3%, P,0.001). After multivariate linear regression adjustment, AA ethnicity proved to be associated independently with low rSO2 [odds ratio (OR) 28.28, 95% confidence interval (CI) 29.12 to 27.44, P,0.001]. Multivariate logistic regression analysis showed that preprocedural rSO2 was independently associated with operative mortality both in the Cauc group (OR 0.97, 95% CI 0.96– 0.99, P¼0.001) and in the AA group (OR 0.97, 95% CI 0.95 –0.99, P¼0.01). Conclusions. AAs have a lower rSO2 than Caucs as measured by the INVOS 5100C cerebral oximeter. Reasonably, this could be attributed to attenuation of the NIR light by skin pigment. Despite this limitation, in both ethnic groups, lower preoperative rSO2 was predictive of greater operative mortality. Keywords: cardiovascular anaesthesia; regional cerebral oxygen saturation; skin pigmentation Accepted for publication: 15 June 2014
Since Food and Drug Administration approval in 1993, devices utilizing near-infrared (NIR) spectroscopy have been used to assess regional intracerebral oxygen saturation (rSO2) during anaesthesia and surgery.1 2 The cranium and extracranial soft tissue are relatively transparent to light in the NIR region of the spectrum. Thus, rSO2 can be assessed using differences in the intensity of transmitted and received light as delivered in specific wave lengths. NIR light transmitted in the wave lengths for oxyhaemoglobin and deoxyhaemoglobin are absorbed to different degrees in their passage through tissue and from the differences between these measurements, the percentage of oxyhaemoglobin in the superficial layers of the frontal cortex can be estimated. This technique has found clinical utility in monitoring deviation in rSO2 from a baseline level during anaesthesia and surgery. Episodes of declining rSO2 levels may be identified and corrective measures undertaken to improve oxygen delivery to the cerebral circulation.3 – 5
Steady-state rSO2 levels have been thought to be subject to wide individual variability and, as a consequence, of little clinical use. Heringlake and colleagues5 recently published data suggesting that low preoperative rSO2 levels are associated with worse clinical outcomes in cardiac surgery and thus are, in fact, useful for assessing surgical risk. Unpublished data from this institution using only Caucasian (Cauc) patients are consistent with these observations. African-American (AA) patients make up an important minority of our practice. We have been concerned that skin pigmentation in this group may interfere with the measurement of rSO2 and compromise the association of rSO2 with surgical outcome. The predominant skin pigment in AAs, melanin, is known to attenuate NIR light transmission and thus to impede NIR spectroscopy signals.4 Accordingly, it is recommended that melanin-containing hair be avoided when the transmitting and receiving optodes are placed.4 Although
& The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
† Near-infrared spectroscopy is increasingly being used to assess regional cerebral oxygen saturation (rSO2).
Background. Devices utilizing near-infrared (NIR) spectroscopy have been used to assess regional intracerebral oxygen saturation (rSO2) during anaesthesia for a decade. The presence of wide differences among individuals reduces their applicability to steady-state measurements. Current devices may not adequately account for variations in skin pigmentation.
BJA melanin in the skin has not been regarded as a problem, our experience with the large number of AAs in our surgical practice has suggested that this assumption may not be true, and that melanin in the skin may, in fact, interfere with baseline measurements made for this purpose. The aim of this analysis was two-fold. First, it was intended to identify and quantify any systematic difference associated with skin pigmentation in the estimation of steady-state rSO2 by means of NIR spectroscopy. Secondly, we sought to determine if any detected pigment-related interference had clinical relevance with regard to the preoperative risk estimation in cardiac surgery. For these purposes, we utilized a large, consecutive series of patients undergoing cardiac surgery in which rSO2 was measured before the induction of anaesthesia.5
Patients The records of all patients included in the computerized registry of cardiac surgical procedures performed in the Division of Cardiac Surgery between January 1, 2010, and December 31, 2012, were examined. The 3282 patients who underwent rSO2 assessment before cardiac surgery were included in this analysis, provided they identified themselves as either Cauc or AAs. The study was approved by the institutional review board.
Data collection and entry As is routine in our department, baseline patient and clinical characteristics together with the details of the operative procedure were prospectively collected as part of routine clinical care and entered into the registry by the staff of the data-coordinating centre. Variables were defined and sourced according to the Society of Thoracic Surgeons National Cardiac Surgery Database Guidelines and Definitions (http://www.STS.org/STS national database).
Operative procedures Operations were performed by experienced cardiac surgeons. Selection of patients and choice of the operative procedure were at the discretion of the responsible surgeon.
Cerebral oxygen saturation rSO2 was measured before the induction of anaesthesia by means of the INVOS 5100C cerebral oximeter (Semantics Corporation, Troy, MI, USA). Each patient was receiving 100% oxygen through a face mask to which an oxygen flow of 3– 5 litre min21 was being delivered. After cleaning the patient’s skin with an alcohol pad, paired transmitting and receiving optodes were positioned on either side of the patient’s forehead. Each optode was covered to prevent ambient light from reaching it.
Statistical analysis Preoperative characteristics are presented as proportions and percentage for dichotomous variables. Differences between ethnic groups were tested by means of the x 2 analysis or
Page 2 of 5
Fisher’s exact test. Continuous variable (e.g. rSO2) measurements are presented as the mean and standard deviation and tested for differences by means of Student’s t-test for normally distributed data or the Wilcoxon rank-sum test for nonparametric data. A two-sided P-value of ,0.05 was chosen as the level of significance. To examine the effect of ethnic group on rSO2, a multivariate linear regression model was used to adjust for preoperative factors that were potentially confounding. Included as candidate variables were those with a univariate difference P,0.15. From this model, parameter estimates and 95% confidence interval (CI) were calculated. To evaluate the association of baseline rSO2 and operative mortality in each ethnic group, a multivariable logistic regression model was constructed. Candidate variables were chosen among both preoperative and intraoperative factors. Odds ratio (OR) with 95% CI are reported. All analyses were performed with SAS version 9.1 (SAS Institute, Cary, NC, USA).
Results Clinical patient characteristics Clinical characteristics of these patients are shown in Table 1. Of the 3282 patients undergoing preprocedural rSO2 measurement, 1186 (36.1%) identified themselves as AA and 2096 (63.9%) as Cauc. There are striking contrasts between the ethnic groups. Importantly, many of the differences suggest the likelihood of a greater risk of operative mortality in AA patients. Reflecting this preponderance, the STS mortality risk score was significantly higher in AAs (1.2 in Caucs vs 1.6 in AAs, P,0.001).
Procedural details Selected details of the operative procedures are compared in Table 2. AAs underwent more complex operations than Caucs. They less often underwent elective surgery (25.7% vs 34.5%, P,0.001) and more often required cardiopulmonary bypass (80.8% vs 70.5%, P,0.001). Thus, both baseline clinical characteristics and the nature of the operative procedure predict a higher operative mortality for AA patients. Surprisingly, no difference was observed (4.8% in AA patients vs 4.7% in Cauc patients, P¼0.87).
Preoperative cerebral oxygen saturation Table 3 reports the group mean and standard deviation for each pair of optodes both before operation and at the time of skin closure. At each of the optode pairs and at each of the two time points, the saturation was significantly lower in AAs by 12% points. Figure 1 demonstrates that the difference between the two ethnic groups occurred across all levels of rSO2 values. Moreover, the normal distribution of these values in each ethnic group is noteworthy. Table 4 reports the association between rSO2 level and preprocedural clinical characteristics (Table 1) as determined by multivariate linear regression analysis. After adjustment for
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Methods
Sun et al.
BJA
Skin pigmentation interferes with the clinical measurement of rSO2
Table 1 Preoperative patient characteristics. LV, left ventricular; IABP, intra-aortic balloon pump; SD, standard deviation Variables
AA patients (n51186)
Cauc patients (n52096)
P-value
Patients characteristics and comorbid factors Age (median, minimum – maximum)
63 (18 – 92)
66 (20 –96)
,0.01
Female (n, %)
515 (43.2)
534 (25.5)
,0.01
BMI (kg m22)
30.0 (7.0)
29.1 (6.4)
,0.01
Active or former smoker (n, %)
317 (26.7)
363 (17.3)
,0.01
Diabetes mellitus (n, %)
598 (50.4)
693 (33.1)
,0.01
Dyslipidaemia (n, %)
820 (69.1)
1606 (76.67
,0.01
Sleep apnoea (n, %)
56 (4.2)
134 (5.6)
STS mortality risk score (median, 25th –75th)
1.6 (0.8 – 3.7)
0.09
1.2 (0.6 –2.6)
,0.01
Prior cardiovascular diseases 1039 (87.6)
1571 (75.0)
Cardiogenic shock (n, %)
Hypertension (n, %)
24 (2.0)
46 (2.2)
0.75
Active infective endocarditis (n, %)
36 (3.0)
32 (1.5)
,0.01 ,0.01
,0.01
325 (27.4)
355 (17.0)
502 (42.3)
812 (38.7)
74 (6.2)
204 (9.7)
Preoperative IABP support (n, %)
77 (6.5)
128 (6.1)
LV ejection fraction ,35% (n, %)
194 (16.4)
236 (11.3)
,0.01
LV ejection fraction [mean (SD), %]
47.9 (13)
50.6 (11.5)
,0.01
Peripheral arterial disease (n, %)
225 (19.0)
340 (16.2)
0.05
Cerebrovascular disease (n, %)
277 (13.2)
213 (18.0)
,0.01
Preoperative stroke (n, %)
127 (6.1)
163 (13.7)
,0.01
11 (0.8)
17 (0.6)
,0.01
19 (1.4)
97 (4.1)
,0.01
36.1 (5.2)
38.7 (4.8)
,0.01
1.7 (1.9)
1.1 (0.8)
,0.01
Preoperative atrial fibrillation (n, %)
0.04 ,0.01 0.90
Prior cerebrovascular diseases
Carotid stenosis (≥80%) (n, %) Carotid surgery (n, %) Preoperative laboratory findings Haematocrit [mean (SD), %] Serum creatinine concentration [mean (SD), mg dl21]
Table 2 Operative details
Table 3 Cerebral oxygen saturation (rSO2) AA patients Cauc patients P-value (n51186) (n52096)
AA patients (n51186)
Cauc patients (n52096)
P-value
Elective vs non-elective (n, %)
305 (25.7)
724 (34.5)
,0.01
Pre-induction, left [mean (SD), %]
53 (13.3)
66 (12.2)
,0.01
On-pump surgery (n, %)
969 (81.7)
1528 (72.9)
,0.01
Pre-induction, right [mean (SD), %]
53 (13.3)
65 (12.3)
,0.01
Pre-induction [mean (SD), %]
53 (12.6)
65 (11,6)
,0.01
Skin closure, left [mean (SD), %]
49 (11.4)
62 (11.3)
,0.01
Skin closure, right [mean (SD), %]
49 (11.4)
62 (11.5)
,0.01
Skin closure [mean (SD), %]
49 (10.9)
62 (11.0)
,0.01
Procedure category (n, %) Isolated CABG (n, %) CABG+valve (n, %)
,0.01 730 (72.0)
1299 (70.7)
59 (5.8)
153 (8.3)
Isolated valve (n, %)
202 (19.9)
375 (20.4)
Aortic surgery (n, %)
23 (5.3)
93 (3.9)
57 (4.8)
98 (4.7)
Postoperative 30 day mortality (n, %)
0.87
these baseline characteristics, AA ethnicity remained significantly associated with lower rSO2 values (parameter estimate 28.58, 95% CIs 29.40, 27.76, P,0.001).
Operative mortality To investigate the clinical relevance of the attenuation by skin pigment in the use of the rSO2 measurement as a
predictor of operative risk, we used two separate multivariate logistic regression models, one for each ethnic group. Both included the preoperative rSO2. Raw 30 day mortality was adjusted for differences in baseline and operative characteristics.
Page 3 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Heart failure (n, %) Myocardial infarction (n, %)
BJA
Sun et al.
Table 5 reports the independent predictors of mortality for the two ethnic groups. The preoperative rSO2 was equally powerful as a predictor of adjusted operative in both ethnic groups.
300
Patient count
African-American Caucasian 200
100
20
40
60 Pre-rSO2 (%)
80
100
Fig 1 Preoperative rSO2 by ethnic group. The number of patients associated with each level of rSO2 is depicted. Notice the normal distribution of the data in each ethnic group with that in AAs being systematically lower.
Table 4 Multivariate linear regression adjusted associations between preoperative factors and preoperative rSO2 Variables
Regression coefficient
95% CI
Age
20.05
20.09 to 20.02 ,0.01
Male gender AA race BMI
2.57 28.28 0.15
P-value
1.71 to 3.42
,0.01
29.12 to 27.44 ,0.01 0.09 –0.21
,0.01
Active or former smoker
23.67
24.64 to 22.71 ,0.01
Diabetes mellitus
22.24
23.06 to 21.42 ,0.01
Haematocrit
0.60
0.52 to 0.69
,0.01
Ejection fraction
21.99
23.20 to 20.77
Preoperative atrial fibrillation
23.39
24.78 to 22.01 ,0.01
0.01
Serum creatinine
21.13
21.89 to 20.42 ,0.01
These data are consistent with the notion that skin pigment attenuates the transmission of NIR light and thus, potentially disturbs the estimation of cerebral oxygen saturation. Steady-state (preoperative) rSO2 measurements were lower in the AA group than in the Cauc group by 12% (Tables 3 and 4 and Fig. 1). After multivariate analysis, this difference between the ethnic groups proved to be independent of other potential covariates. The figure demonstrates that this difference in both groups is normally distributed across the observed range of rSO2. This relationship is consistent with the assumption that the concentration of skin pigment varies from individual both between and within ethnic groups. Thus, our analysis supports the notion that ethnicity, and by inference skin pigment, is an independent predictor of the level of rSO2. It is therefore reasonable to consider the possibility that skin pigmentation in the path of the NIR light attenuates the signal.
Cerebral oxygen saturation and operative risk It has been recognized that individual variation in steady-state measurements of rSO2 with NIR spectroscopy limits its application for clinical purposes. Thus, this technology has found its primary clinical application in monitoring rSO2 during surgical procedures. Since such monitoring is intended to identify deviation from baseline values, variability in individual steadystate measurements are relatively unimportant. Our findings suggest that skin pigmentation partially accounts for individual variability in the use of NIR spectroscopy to measure rSO2. To our knowledge, Heringlake and colleagues5 were the first to explore the use of steady-state NIR spectroscopy for preoperative risk assessment. These investigators found a strong association between lower preoperative rSO2 and higher operative mortality. The findings of the current observations support this suggestion. We found that the attenuation of the NIR signal by skin pigmentation did not detract from the ability of that measurement to predict operative mortality in AAs. It should be pointed out that Heringlake and colleagues5 did not report the proportion of patients of African ethnicity in their analysis, but that proportion was likely to have been low since their work was conducted in northern Germany, a geographical region in which it seems probable that few members of their
Table 5 Multivariate adjusted associations between pre-induction rSO2 and operative mortality Cauc group OR
AA group 95% CI
P-value
OR
95% CI
P-value
Age
1.05
1.03 –1.07
,0.01
1.03
1.04 – 1.05
0.02
Male
0.81
0.51 –1.28
0.36
0.85
0.49 – 1.45
0.55
LV ejection fraction,35%
0.27
0.17 –0.44
,0.01
0.48
0.26 – 0.89
0.02
Elective surgery vs non-elective surgery
0.63
0.37 –1.06
0.08
0.33
0.13 – 0.85
0.02
Preoperative rSO2 (mean)
0.97
0.96 –0.99
,0.01
0.97
0.95 – 0.99
0.01
On-pump surgery
3.44
1.70 –6.94
,0.01
2.50
0.97 – 6.40
0.06
Page 4 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
0
Discussion
BJA
Skin pigmentation interferes with the clinical measurement of rSO2
cohort were of such descent. Thus, our data supporting the clinical relevance of its application to AA patients seem to remove at least some of the doubt regarding its application to that group.
Despite this interference with NIR light transmission, low preoperative cerebral oxygen saturation was a predictor of a higher operative mortality in both AAs and Caucs.
Limitations of the study
Authors’ contributions
The data have certain strengths: (i) a large number of unselected subjects in clinical practice were studied; (ii) the study population included a large number of AAs providing adequate statistical power for crucial comparisons. Nevertheless, interpretation of the findings should be undertaken with caution. It must be acknowledged that the fundamental assumption may be in error. The difference in rSO2 measurements between these two ethnic groups may not reflect differences in skin pigmentation. Indeed, other unrecorded biological factors (e.g. skull thickness) could play a role. Moreover, current instrumentation allows sampling of cerebral tissue from only a limited segment of the brain, the frontal cortex. That region is quite vulnerable to hypoxic –ischaemic injury and may be an appropriate area for sampling, but cerebral oxygen saturation need not be the same in other areas. Lastly, as with any retrospective cross-sectional study, there are substantial differences between studied groups apart from the rSO2. Furthermore, relevant but unmeasured covariates may exist. Thus, statistical adjustment may not be adequate to eliminate the unwanted influences of these differences.
P.J.C. and J.E. proposed the study idea; X.S. proposed the study idea and analysed the data with F.C. and J.L. X.S and J.L. wrote the manuscript; J.L. edited the manuscript. All authors have approved the manuscript.
In a steady-state preoperative setting, because of a greater concentration of melanin in the skin, AAs have lower cerebral oxygen saturation when compared with Caucs as measured by current instrumentation using NIR light. This difference can arguably be attributed to interference in the transmission of the NIR light by the greater amount of skin pigment in AAs, but this assumption is not proven.
None declared.
Funding The study was conducted by the Department of Cardiac Surgery; no other funding involved.
References 1 Young AE, Germon TJ, Barnett NJ, Manara AR, Nelson RJ. Behavior of near-infrared light in the adult human head: implications for clinical near-infrared spectroscopy. Br J Anaesth 2000; 84: 38 – 42 2 Murkin JM, Arango M. Near-infrared spectroscopy as an index of brain and tissue oxygenation. Br J Anaesth 2009; 103(Suppl. 1): i3–13 3 Ferrari M, Mottola L, Quaresima V. Principles, techniques and limitations of near infrared spectroscopy. Can J Appl Physiol 2004; 29: 463–87 4 Watzman HM, Kurth CD, Montenengro LM, Rome J, Steven JM, Nicolson SC. Arterial and venous contributions to near infrared cerebral oximetry. Anesthesiology 2000; 93: 947– 53 5 Heringlake M, Garbers C, Med C, et al. Preoperative cerebral oxygen saturation and clinical outcomes in cardiac surgery. Anesthesiology 2011; 114: 58 –69
Handling editor: A. R. Absalom
Page 5 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Conclusions
Declaration of interest
Skin pigmentation interferes with the clinical measurement of regional cerebral oxygen saturation X. Sun 1*, J. Ellis 1, P. J. Corso 1, P. C. Hill1, F. Chen2 and J. Lindsay3 1
Department of Cardiac Surgery, 2 Medstar Cardiovascular Research Network, and 3 Department of Cardiology, Medstar Washington Hospital Center, 110 Irving Street, Room Suite 1E RM1223, Washington, DC 20010, USA * Corresponding author. E-mail: [email protected]
Editor’s key points
† The authors studied preoperative rSO2 values in a cardiac surgery population. † Skin pigmentation was found to independently influence preoperative rSO2. † Low preoperative rSO2 values were associated with worse mortality in all racial groups.
Methods. From our ongoing departmental registry, 3282 consecutive patients underwent cardiac surgery between 2010 and 2012 and their pre-induction measurements of rSO2 were available. Of these, 2096 identified themselves as Caucasian (Cauc) and 1186 as African-American (AA). Pre-induction rSO2, clinical and operative features were compared. Results. Clinical and operative details of these patients differed widely between the two populations. High-risk features were more common in AA patients, but no difference in mortality was observed (4.8% in AAs vs 4.7% in Caucs, P¼0.87). Preprocedure rSO2 was systematically higher in Cauc (65.5% vs 53.3%, P,0.001). After multivariate linear regression adjustment, AA ethnicity proved to be associated independently with low rSO2 [odds ratio (OR) 28.28, 95% confidence interval (CI) 29.12 to 27.44, P,0.001]. Multivariate logistic regression analysis showed that preprocedural rSO2 was independently associated with operative mortality both in the Cauc group (OR 0.97, 95% CI 0.96– 0.99, P¼0.001) and in the AA group (OR 0.97, 95% CI 0.95 –0.99, P¼0.01). Conclusions. AAs have a lower rSO2 than Caucs as measured by the INVOS 5100C cerebral oximeter. Reasonably, this could be attributed to attenuation of the NIR light by skin pigment. Despite this limitation, in both ethnic groups, lower preoperative rSO2 was predictive of greater operative mortality. Keywords: cardiovascular anaesthesia; regional cerebral oxygen saturation; skin pigmentation Accepted for publication: 15 June 2014
Since Food and Drug Administration approval in 1993, devices utilizing near-infrared (NIR) spectroscopy have been used to assess regional intracerebral oxygen saturation (rSO2) during anaesthesia and surgery.1 2 The cranium and extracranial soft tissue are relatively transparent to light in the NIR region of the spectrum. Thus, rSO2 can be assessed using differences in the intensity of transmitted and received light as delivered in specific wave lengths. NIR light transmitted in the wave lengths for oxyhaemoglobin and deoxyhaemoglobin are absorbed to different degrees in their passage through tissue and from the differences between these measurements, the percentage of oxyhaemoglobin in the superficial layers of the frontal cortex can be estimated. This technique has found clinical utility in monitoring deviation in rSO2 from a baseline level during anaesthesia and surgery. Episodes of declining rSO2 levels may be identified and corrective measures undertaken to improve oxygen delivery to the cerebral circulation.3 – 5
Steady-state rSO2 levels have been thought to be subject to wide individual variability and, as a consequence, of little clinical use. Heringlake and colleagues5 recently published data suggesting that low preoperative rSO2 levels are associated with worse clinical outcomes in cardiac surgery and thus are, in fact, useful for assessing surgical risk. Unpublished data from this institution using only Caucasian (Cauc) patients are consistent with these observations. African-American (AA) patients make up an important minority of our practice. We have been concerned that skin pigmentation in this group may interfere with the measurement of rSO2 and compromise the association of rSO2 with surgical outcome. The predominant skin pigment in AAs, melanin, is known to attenuate NIR light transmission and thus to impede NIR spectroscopy signals.4 Accordingly, it is recommended that melanin-containing hair be avoided when the transmitting and receiving optodes are placed.4 Although
& The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: [email protected]
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
† Near-infrared spectroscopy is increasingly being used to assess regional cerebral oxygen saturation (rSO2).
Background. Devices utilizing near-infrared (NIR) spectroscopy have been used to assess regional intracerebral oxygen saturation (rSO2) during anaesthesia for a decade. The presence of wide differences among individuals reduces their applicability to steady-state measurements. Current devices may not adequately account for variations in skin pigmentation.
BJA melanin in the skin has not been regarded as a problem, our experience with the large number of AAs in our surgical practice has suggested that this assumption may not be true, and that melanin in the skin may, in fact, interfere with baseline measurements made for this purpose. The aim of this analysis was two-fold. First, it was intended to identify and quantify any systematic difference associated with skin pigmentation in the estimation of steady-state rSO2 by means of NIR spectroscopy. Secondly, we sought to determine if any detected pigment-related interference had clinical relevance with regard to the preoperative risk estimation in cardiac surgery. For these purposes, we utilized a large, consecutive series of patients undergoing cardiac surgery in which rSO2 was measured before the induction of anaesthesia.5
Patients The records of all patients included in the computerized registry of cardiac surgical procedures performed in the Division of Cardiac Surgery between January 1, 2010, and December 31, 2012, were examined. The 3282 patients who underwent rSO2 assessment before cardiac surgery were included in this analysis, provided they identified themselves as either Cauc or AAs. The study was approved by the institutional review board.
Data collection and entry As is routine in our department, baseline patient and clinical characteristics together with the details of the operative procedure were prospectively collected as part of routine clinical care and entered into the registry by the staff of the data-coordinating centre. Variables were defined and sourced according to the Society of Thoracic Surgeons National Cardiac Surgery Database Guidelines and Definitions (http://www.STS.org/STS national database).
Operative procedures Operations were performed by experienced cardiac surgeons. Selection of patients and choice of the operative procedure were at the discretion of the responsible surgeon.
Cerebral oxygen saturation rSO2 was measured before the induction of anaesthesia by means of the INVOS 5100C cerebral oximeter (Semantics Corporation, Troy, MI, USA). Each patient was receiving 100% oxygen through a face mask to which an oxygen flow of 3– 5 litre min21 was being delivered. After cleaning the patient’s skin with an alcohol pad, paired transmitting and receiving optodes were positioned on either side of the patient’s forehead. Each optode was covered to prevent ambient light from reaching it.
Statistical analysis Preoperative characteristics are presented as proportions and percentage for dichotomous variables. Differences between ethnic groups were tested by means of the x 2 analysis or
Page 2 of 5
Fisher’s exact test. Continuous variable (e.g. rSO2) measurements are presented as the mean and standard deviation and tested for differences by means of Student’s t-test for normally distributed data or the Wilcoxon rank-sum test for nonparametric data. A two-sided P-value of ,0.05 was chosen as the level of significance. To examine the effect of ethnic group on rSO2, a multivariate linear regression model was used to adjust for preoperative factors that were potentially confounding. Included as candidate variables were those with a univariate difference P,0.15. From this model, parameter estimates and 95% confidence interval (CI) were calculated. To evaluate the association of baseline rSO2 and operative mortality in each ethnic group, a multivariable logistic regression model was constructed. Candidate variables were chosen among both preoperative and intraoperative factors. Odds ratio (OR) with 95% CI are reported. All analyses were performed with SAS version 9.1 (SAS Institute, Cary, NC, USA).
Results Clinical patient characteristics Clinical characteristics of these patients are shown in Table 1. Of the 3282 patients undergoing preprocedural rSO2 measurement, 1186 (36.1%) identified themselves as AA and 2096 (63.9%) as Cauc. There are striking contrasts between the ethnic groups. Importantly, many of the differences suggest the likelihood of a greater risk of operative mortality in AA patients. Reflecting this preponderance, the STS mortality risk score was significantly higher in AAs (1.2 in Caucs vs 1.6 in AAs, P,0.001).
Procedural details Selected details of the operative procedures are compared in Table 2. AAs underwent more complex operations than Caucs. They less often underwent elective surgery (25.7% vs 34.5%, P,0.001) and more often required cardiopulmonary bypass (80.8% vs 70.5%, P,0.001). Thus, both baseline clinical characteristics and the nature of the operative procedure predict a higher operative mortality for AA patients. Surprisingly, no difference was observed (4.8% in AA patients vs 4.7% in Cauc patients, P¼0.87).
Preoperative cerebral oxygen saturation Table 3 reports the group mean and standard deviation for each pair of optodes both before operation and at the time of skin closure. At each of the optode pairs and at each of the two time points, the saturation was significantly lower in AAs by 12% points. Figure 1 demonstrates that the difference between the two ethnic groups occurred across all levels of rSO2 values. Moreover, the normal distribution of these values in each ethnic group is noteworthy. Table 4 reports the association between rSO2 level and preprocedural clinical characteristics (Table 1) as determined by multivariate linear regression analysis. After adjustment for
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Methods
Sun et al.
BJA
Skin pigmentation interferes with the clinical measurement of rSO2
Table 1 Preoperative patient characteristics. LV, left ventricular; IABP, intra-aortic balloon pump; SD, standard deviation Variables
AA patients (n51186)
Cauc patients (n52096)
P-value
Patients characteristics and comorbid factors Age (median, minimum – maximum)
63 (18 – 92)
66 (20 –96)
,0.01
Female (n, %)
515 (43.2)
534 (25.5)
,0.01
BMI (kg m22)
30.0 (7.0)
29.1 (6.4)
,0.01
Active or former smoker (n, %)
317 (26.7)
363 (17.3)
,0.01
Diabetes mellitus (n, %)
598 (50.4)
693 (33.1)
,0.01
Dyslipidaemia (n, %)
820 (69.1)
1606 (76.67
,0.01
Sleep apnoea (n, %)
56 (4.2)
134 (5.6)
STS mortality risk score (median, 25th –75th)
1.6 (0.8 – 3.7)
0.09
1.2 (0.6 –2.6)
,0.01
Prior cardiovascular diseases 1039 (87.6)
1571 (75.0)
Cardiogenic shock (n, %)
Hypertension (n, %)
24 (2.0)
46 (2.2)
0.75
Active infective endocarditis (n, %)
36 (3.0)
32 (1.5)
,0.01 ,0.01
,0.01
325 (27.4)
355 (17.0)
502 (42.3)
812 (38.7)
74 (6.2)
204 (9.7)
Preoperative IABP support (n, %)
77 (6.5)
128 (6.1)
LV ejection fraction ,35% (n, %)
194 (16.4)
236 (11.3)
,0.01
LV ejection fraction [mean (SD), %]
47.9 (13)
50.6 (11.5)
,0.01
Peripheral arterial disease (n, %)
225 (19.0)
340 (16.2)
0.05
Cerebrovascular disease (n, %)
277 (13.2)
213 (18.0)
,0.01
Preoperative stroke (n, %)
127 (6.1)
163 (13.7)
,0.01
11 (0.8)
17 (0.6)
,0.01
19 (1.4)
97 (4.1)
,0.01
36.1 (5.2)
38.7 (4.8)
,0.01
1.7 (1.9)
1.1 (0.8)
,0.01
Preoperative atrial fibrillation (n, %)
0.04 ,0.01 0.90
Prior cerebrovascular diseases
Carotid stenosis (≥80%) (n, %) Carotid surgery (n, %) Preoperative laboratory findings Haematocrit [mean (SD), %] Serum creatinine concentration [mean (SD), mg dl21]
Table 2 Operative details
Table 3 Cerebral oxygen saturation (rSO2) AA patients Cauc patients P-value (n51186) (n52096)
AA patients (n51186)
Cauc patients (n52096)
P-value
Elective vs non-elective (n, %)
305 (25.7)
724 (34.5)
,0.01
Pre-induction, left [mean (SD), %]
53 (13.3)
66 (12.2)
,0.01
On-pump surgery (n, %)
969 (81.7)
1528 (72.9)
,0.01
Pre-induction, right [mean (SD), %]
53 (13.3)
65 (12.3)
,0.01
Pre-induction [mean (SD), %]
53 (12.6)
65 (11,6)
,0.01
Skin closure, left [mean (SD), %]
49 (11.4)
62 (11.3)
,0.01
Skin closure, right [mean (SD), %]
49 (11.4)
62 (11.5)
,0.01
Skin closure [mean (SD), %]
49 (10.9)
62 (11.0)
,0.01
Procedure category (n, %) Isolated CABG (n, %) CABG+valve (n, %)
,0.01 730 (72.0)
1299 (70.7)
59 (5.8)
153 (8.3)
Isolated valve (n, %)
202 (19.9)
375 (20.4)
Aortic surgery (n, %)
23 (5.3)
93 (3.9)
57 (4.8)
98 (4.7)
Postoperative 30 day mortality (n, %)
0.87
these baseline characteristics, AA ethnicity remained significantly associated with lower rSO2 values (parameter estimate 28.58, 95% CIs 29.40, 27.76, P,0.001).
Operative mortality To investigate the clinical relevance of the attenuation by skin pigment in the use of the rSO2 measurement as a
predictor of operative risk, we used two separate multivariate logistic regression models, one for each ethnic group. Both included the preoperative rSO2. Raw 30 day mortality was adjusted for differences in baseline and operative characteristics.
Page 3 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Heart failure (n, %) Myocardial infarction (n, %)
BJA
Sun et al.
Table 5 reports the independent predictors of mortality for the two ethnic groups. The preoperative rSO2 was equally powerful as a predictor of adjusted operative in both ethnic groups.
300
Patient count
African-American Caucasian 200
100
20
40
60 Pre-rSO2 (%)
80
100
Fig 1 Preoperative rSO2 by ethnic group. The number of patients associated with each level of rSO2 is depicted. Notice the normal distribution of the data in each ethnic group with that in AAs being systematically lower.
Table 4 Multivariate linear regression adjusted associations between preoperative factors and preoperative rSO2 Variables
Regression coefficient
95% CI
Age
20.05
20.09 to 20.02 ,0.01
Male gender AA race BMI
2.57 28.28 0.15
P-value
1.71 to 3.42
,0.01
29.12 to 27.44 ,0.01 0.09 –0.21
,0.01
Active or former smoker
23.67
24.64 to 22.71 ,0.01
Diabetes mellitus
22.24
23.06 to 21.42 ,0.01
Haematocrit
0.60
0.52 to 0.69
,0.01
Ejection fraction
21.99
23.20 to 20.77
Preoperative atrial fibrillation
23.39
24.78 to 22.01 ,0.01
0.01
Serum creatinine
21.13
21.89 to 20.42 ,0.01
These data are consistent with the notion that skin pigment attenuates the transmission of NIR light and thus, potentially disturbs the estimation of cerebral oxygen saturation. Steady-state (preoperative) rSO2 measurements were lower in the AA group than in the Cauc group by 12% (Tables 3 and 4 and Fig. 1). After multivariate analysis, this difference between the ethnic groups proved to be independent of other potential covariates. The figure demonstrates that this difference in both groups is normally distributed across the observed range of rSO2. This relationship is consistent with the assumption that the concentration of skin pigment varies from individual both between and within ethnic groups. Thus, our analysis supports the notion that ethnicity, and by inference skin pigment, is an independent predictor of the level of rSO2. It is therefore reasonable to consider the possibility that skin pigmentation in the path of the NIR light attenuates the signal.
Cerebral oxygen saturation and operative risk It has been recognized that individual variation in steady-state measurements of rSO2 with NIR spectroscopy limits its application for clinical purposes. Thus, this technology has found its primary clinical application in monitoring rSO2 during surgical procedures. Since such monitoring is intended to identify deviation from baseline values, variability in individual steadystate measurements are relatively unimportant. Our findings suggest that skin pigmentation partially accounts for individual variability in the use of NIR spectroscopy to measure rSO2. To our knowledge, Heringlake and colleagues5 were the first to explore the use of steady-state NIR spectroscopy for preoperative risk assessment. These investigators found a strong association between lower preoperative rSO2 and higher operative mortality. The findings of the current observations support this suggestion. We found that the attenuation of the NIR signal by skin pigmentation did not detract from the ability of that measurement to predict operative mortality in AAs. It should be pointed out that Heringlake and colleagues5 did not report the proportion of patients of African ethnicity in their analysis, but that proportion was likely to have been low since their work was conducted in northern Germany, a geographical region in which it seems probable that few members of their
Table 5 Multivariate adjusted associations between pre-induction rSO2 and operative mortality Cauc group OR
AA group 95% CI
P-value
OR
95% CI
P-value
Age
1.05
1.03 –1.07
,0.01
1.03
1.04 – 1.05
0.02
Male
0.81
0.51 –1.28
0.36
0.85
0.49 – 1.45
0.55
LV ejection fraction,35%
0.27
0.17 –0.44
,0.01
0.48
0.26 – 0.89
0.02
Elective surgery vs non-elective surgery
0.63
0.37 –1.06
0.08
0.33
0.13 – 0.85
0.02
Preoperative rSO2 (mean)
0.97
0.96 –0.99
,0.01
0.97
0.95 – 0.99
0.01
On-pump surgery
3.44
1.70 –6.94
,0.01
2.50
0.97 – 6.40
0.06
Page 4 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
0
Discussion
BJA
Skin pigmentation interferes with the clinical measurement of rSO2
cohort were of such descent. Thus, our data supporting the clinical relevance of its application to AA patients seem to remove at least some of the doubt regarding its application to that group.
Despite this interference with NIR light transmission, low preoperative cerebral oxygen saturation was a predictor of a higher operative mortality in both AAs and Caucs.
Limitations of the study
Authors’ contributions
The data have certain strengths: (i) a large number of unselected subjects in clinical practice were studied; (ii) the study population included a large number of AAs providing adequate statistical power for crucial comparisons. Nevertheless, interpretation of the findings should be undertaken with caution. It must be acknowledged that the fundamental assumption may be in error. The difference in rSO2 measurements between these two ethnic groups may not reflect differences in skin pigmentation. Indeed, other unrecorded biological factors (e.g. skull thickness) could play a role. Moreover, current instrumentation allows sampling of cerebral tissue from only a limited segment of the brain, the frontal cortex. That region is quite vulnerable to hypoxic –ischaemic injury and may be an appropriate area for sampling, but cerebral oxygen saturation need not be the same in other areas. Lastly, as with any retrospective cross-sectional study, there are substantial differences between studied groups apart from the rSO2. Furthermore, relevant but unmeasured covariates may exist. Thus, statistical adjustment may not be adequate to eliminate the unwanted influences of these differences.
P.J.C. and J.E. proposed the study idea; X.S. proposed the study idea and analysed the data with F.C. and J.L. X.S and J.L. wrote the manuscript; J.L. edited the manuscript. All authors have approved the manuscript.
In a steady-state preoperative setting, because of a greater concentration of melanin in the skin, AAs have lower cerebral oxygen saturation when compared with Caucs as measured by current instrumentation using NIR light. This difference can arguably be attributed to interference in the transmission of the NIR light by the greater amount of skin pigment in AAs, but this assumption is not proven.
None declared.
Funding The study was conducted by the Department of Cardiac Surgery; no other funding involved.
References 1 Young AE, Germon TJ, Barnett NJ, Manara AR, Nelson RJ. Behavior of near-infrared light in the adult human head: implications for clinical near-infrared spectroscopy. Br J Anaesth 2000; 84: 38 – 42 2 Murkin JM, Arango M. Near-infrared spectroscopy as an index of brain and tissue oxygenation. Br J Anaesth 2009; 103(Suppl. 1): i3–13 3 Ferrari M, Mottola L, Quaresima V. Principles, techniques and limitations of near infrared spectroscopy. Can J Appl Physiol 2004; 29: 463–87 4 Watzman HM, Kurth CD, Montenengro LM, Rome J, Steven JM, Nicolson SC. Arterial and venous contributions to near infrared cerebral oximetry. Anesthesiology 2000; 93: 947– 53 5 Heringlake M, Garbers C, Med C, et al. Preoperative cerebral oxygen saturation and clinical outcomes in cardiac surgery. Anesthesiology 2011; 114: 58 –69
Handling editor: A. R. Absalom
Page 5 of 5
Downloaded from http://bja.oxfordjournals.org/ at University of Washington on October 27, 2014
Conclusions
Declaration of interest