Background

Protecting brain functions is one of the essences of anesthesia practice. There is an increasing concern about the potential effects of anesthetics on perioperative neurocognitive disorders (PND) [1]. PND is a common complication after major surgeries including postoperative delirium (POD), postoperative cognitive decline (POCD) and delayed neurocognitive recovery. Its incidence rate ranges from 10 to 50% in general population. In high-risk patients, the incidence rate could reach as high as 50–70% [2,3,4]. PND is associated with several poor prognosis, such as higher mortality, long-term cognitive decline, dementia, re-admission and prolonged length of hospitalization. It also increases the financial burdens to the public, reaching up to $ 16 billion for US health care cost every year [5,6,7,8,9,10].

It has been revealed by several studies that the risk of PND is increased by either that excessively deep anesthesia or lower level of regional cerebral oxygen saturation during the operation [11,12,13]. These findings provided evidence to support the necessities of maintaining proper depth of anesthesia and enhancing cerebral perfusion, therefore increasing the level of cerebral oxygen saturation. Various monitoring technologies have become available to monitor the cerebral function. For example, electroencephalography (EEG) is a commonly used method to monitor the depth of anesthesia [14]. Regional cerebral oxygen saturation (rSO2) monitoring detected by near infrared reflected spectroscopy (NIRS) can be used to monitor cerebral saturation and to alert cerebral ischemia [15]. Previous meta-analysis indicated that the use of cerebral monitors during surgery correlated with a reduced risk of PND [16, 17]. However, this conclusion appeared to be controversial as some recently published large randomized controlled trials showed that the use of cerebral monitors didn’t benefit the reduction of PND incidence after major surgeries [18, 19].

To better understand the effects of cerebral function monitoring on PND and to provide clearer guidance to clinicians, we conducted this systematic review to investigate the relationship between intraoperative cerebral function monitoring and the adverse clinical outcomes.

Methods

This review was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement (PRISMA) guidelines [20]. This systematic review and meta-analysis had been registered in the international prospective register of systematic reviews (CRD42019130512 https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=130512).

Search strategy

Two investigators (DL and DXC) performed a systematic search in the databases of Ovid MEDLINE, PubMed, EMBASE, Cochrane Library database, and other databases updated to April 2020. The searching keywords included “cerebral monitoring”, “electroencephalography”, “cerebral oxygenation”, “postoperative delirium”, “postoperative cognitive decline”, and “randomized controlled trial”. The search terms were modified for each database. Any conflict about search results between the two investigators (DL and DXC) was resolved by discussion and the consensus was reached. The literature search strategy is provided in Additional file 1: Material 1.

Eligibility criteria

Prior to the systematic review and meta-analysis, the inclusion criteria were predetermined by all authors. Inclusion criteria were as the following: (1) the study was randomized controlled trial (RCT), regardless of the language and status; (2) included patients were adults aged 18 years or older who underwent general anesthesia for surgery; (3) the incidence of PND under the EEG or rSO2 monitoring was compared to the PND outcome without the usage of EEG or rSO2 monitoring in the study; (4) the occurrence of PND evaluated by validated scale was reported in the study. The exclusion criteria were: (1) non-randomized studies; (2) non full-text studies; (3) ongoing studies; (4) the outcome data could not be extracted and used to analyze.

Data collection and quality assessment

Data was extracted by two investigators (DL and DXC) independently using a standardized form based on the Population Interventions Comparisons Outcomes (PICO) approach. The extracted information included the first author, year of publication, study design, sample size, outcome variables and assessment scale, summative results and conclusion. The methodological quality of the included studies was with using the Cochrane risk of bias scale, which contains seven specified domains [21]. Risk of bias were classified as high, low or unclear for each item. The methodological quality assessment was conducted by two investigators independently, and the occurred conflicts were resolved by a third investigator (QL) referring to the original article, if any.

Statistical analysis

Data analyses were performed using the Review Manager (version 5.3) software. The inspection level for the pooled data were two-sided, and P < 0.05 was regarded as statistically significant. The odds ratio (OR) and mean difference (MD) with 95% confidence interval (CI) were employed to analyze the categories and continuous data. Heterogeneity across studies was assessed with chi-square test and I2 test, and I2 > 50% or P < 0.10 was considered as significantly heterogenous. The random-effect model was adopted if the heterogeneity existed among the studies, whereas the fixed-effect model was applied if no significant heterogeneity was detected. Sensitivity analysis was conducted to assess the impact of single study to the overall analysis [22,23,24,25]. Publication bias was assessed by using the funnel plot test.

Results

Literature search

The initial search in PubMed, Ovid, EMBASE, Cochrane library, and other databases identified 3309 reports. Duplicates removal reduced the number of reports to 2630. Then, 2589 studies were further excluded after reviewing the title and abstracts. The full text of the remaining 41 studies were retrieved for evaluation, 19 out of the 41 studies were further excluded due to one or more of the following reasons: not RCT (n = 2); review (n = 6); non-general anesthesia patients (n = 3); or other studies which data could not be extracted or used to analyze (n = 8). Reviewing the reference lists of the retrieved studies did not identify any new eligible study. Finally, 22 RCTs were included in the present this review [18, 19, 26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45]. A flow diagram illustrating the literature search and trials screening process was shown in Fig. 1.

Fig. 1
figure 1

Flow diagram of the literature search and trials screening process

Full size image

Characteristics of included studies

As listed in Table 1, a total number of 6356 patients included in the 22 RCTs were enrolled in this meta-analysis [18, 19, 26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45]. Among these 22 RCTs, 10 studies used EEG as the guide for anesthesia depth [18, 26,27,28,29,30,31,32,33,34], 10 studies evaluated the effects of rSO2 monitoring on PND [19, 35,36,37,38,39,40,41,42,43] and the last 2 trials deployed both of the two monitoring [44, 45]. Ten out of the 22 trials included patients who underwent cardiac surgery [18, 19, 27, 36, 38,39,40,41, 43, 45], whereas the other 12 studies were conducted in among patients undergoing non-cardiac major surgery, including abdominal surgery, ENT surgery, hip fracture repair surgeries and others [26, 28,29,30,31,32,33,34, 37, 39, 42, 44]. The risk of bias of included studies was assessed and the result was shown in Fig. 2. Two studies were at the high level in terms of risk of bias. One of them was lacking of the methods of allocation and blinding [33], and the other had a high dropout rate [44]. Eleven studies [19, 30, 32, 34,35,36, 38, 40,41,42,43] were at unclear risk of bias due to the unclear blinding of outcome assessments (detection bias) or unclear blinding of participants and study personnel (performance bias). The remaining 9 studies rated as low in terms of the risk of bias.

Table 1 Characteristics of included studies
Full size table
Fig. 2
figure 2

The risk of bias assessment of included studies. (a, risk of bias summary: review authors’ judgements of each risk of bias item for each included study; b, risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies)

Full size image

Cerebral functional monitoring and perioperative neurocognitive disorders (PND)

Electroencephalography (EEG) guided anesthesia

Postoperative delirium (POD)

After pooling and analyzing the data from the 10 studies using EEG to guide the depth of anesthesia (n = 4451, EEG monitoring = 2214, routine care = 2237), it is noticed that in general, the EEG-guided anesthesia group had a reduced risk of POD compared to the group of routine care (OR: 0.75; 95% CI: 0.60–0.93; P = 0.008) (Fig. 3a) with a significant, heterogeneity detected among the included studies (P = 0.004, I2 = 61%). Then we divided the participants into non-cardiac or cardiac subgroups according to the types of surgeries that patients received and re-analyzed the effect of EEG on the risk of POD. The results demonstrated that in the non-cardiac surgery subgroup (8 studies, n = 3600, EEG monitoring = 1793, non-EEG group = 1807), the use of EEG-guided anesthesia correlated with a reduction of POD incidence (OR: 0.73; 95% CI: 0.57–0.95; P = 0.02), whereas in the cardiac surgery subgroup, no correlation was detected between the use of EEG and the reduction of POD incidence (2 studies, n = 541, EEG monitoring = 272, non-EEG group = 269; OR: 0.44; 95% CI: 0.05–3.54; P = 0.44). It is noted that the study conducted by Whitlock et al. was excluded from the subgroup analysis because it included both cardiac surgery and thoracic surgery without detailed information on the number of patients in the each subgroup. The funnel plot demonstrated that no publication bias existed among the included studies (Fig. 3b).

Fig. 3
figure 3

Postoperative delirium (POD) of EEG guided arm vs routine care arm. (a, forest plot of POD; b, funnel plot of POD)

Full size image

Postoperative cognitive decline (POCD)

Four studies (n = 2435, EEG monitoring = 1200, routine care = 1235) reported the incidence of POCD within 1–7 days after surgery after non-cardiac surgery, and three of them also reported the incidence of POCD 3 months after the surgery (n = 2047, EEG monitoring = 1011, routine care = 1036). It was found that EEG-guided anesthesia did not reduce the incidence of POCD in the early postoperative stage (OR: 0.61; 95% CI: 0.35–1.07; P = 0.09). However, the incidence of POCD 3 months after the surgery was reduced upon the use of intraoperative EEG monitoring (OR: 0.69; 95% CI: 0.49–0.96; P = 0.03). Only one study by Ballard and his colleagues, reported the incidence of POCD at the time of 1 year postoperatively (n = 59, EEG monitoring = 27, routine care = 32), but it did not suggest the advantages of EEG monitoring with respect to the incidence of POCD (OR: 0.27; 95% CI: 0.03–2.57; P = 0.25) (Fig. 4).

Fig. 4
figure 4

Postoperative cognitive decline (POCD) of EEG guided arm vs routine care arm

Full size image

Regional cerebral oxygen saturation (rSO2) monitoring

Postoperative delirium (POD)

Four studies including 765 patients undergoing cardiac surgery reported the outcome of incidence of POD between the rSO2 monitoring group (n = 378) and the routine care group (n = 387). All the four studies showed no difference in terms of POD between the two groups. Our meta-analysis also revealed a comparable result about the incidence of POD between the cerebral oxygenation monitoring group and the routine care group (OR: 0.74; 95% CI: 0.48–1.14; P = 0.17) (Fig. 5).

Fig. 5
figure 5

Postoperative delirium (POD) of rSO2 Monitoring arm vs routine care arm

Full size image

Postoperative cognitive decline (POCD)

Seven studies involving 805 patients analyzed the effect of rSO2 monitoring on the incidence of POCD after major surgeries compared to that of routine care (n = 805, rSO2 monitoring group, n = 411; routine care, n = 394). Among these 7 studies, Slater et al. focused on cardiac surgery [43], while Cox and his colleagues reported the result of their work in non-cardiac surgery [37]. Despite the different types of surgeries studied, both groups reported no difference found between the rSO2 monitoring group and the routine care group in terms of the incidence of POCD. However, the remaining five trials, three of which conducted in patients undertaking cardiac surgeries [36, 39, 40] and two of which performed in patients undertaking non-cardiac surgeries [35, 42], reported that the incidence of POCD in the routine care group was significantly higher than that of rSO2 monitoring group. Our meta-analysis also revealed a significant lower incidence of the POCD in the rSO2 monitoring group (OR: 0.53, 95% CI 0.39–0.73; P < 0.0001) (Fig. 6) compared to that of routine care group without heterogeneity detected (P = 0.19; I2 = 32%).

Fig. 6
figure 6

Postoperative cognitive decline (POCD) of rSO2 arm vs routine care arm

Full size image

Discussion

In the present systematic review and meta-analysis, data of 6356 patients from 22 RCTs was analyzed, including 3169 patients who received EEG or/and rSO2 monitoring, and 3187 patients who received routine care. In non-cardiac surgery patients, we found the incidence of POD significantly decreased in EEG-guided anesthesia group compared to that of routine care group. Both EEG-guided anesthesia and rSO2 monitoring were correlated with a significant lower incidence of POCD despite the types of surgery.

EEG-guided anesthesia

Twenty studies including 4976 patients assessed the association between the EEG-guided anesthesia and PND [18, 26,27,28,29,30,31,32,33,34]. Our meta-analysis showed that EEG-guided anesthesia could reduce the incidence of POD in patients undergoing non-cardiac surgeries but not cardiac surgery patients. In addition, deployment of intraoperative EEG monitoring could reduce the incidence of POCD up to 3 months after the surgery, but had no effect on the incidence of early POCD.

Prior to our study, Punjasawadwong et al. [46] and Kristen et al. [16] have performed two meta-analyses separately, each included 3 RCTs (n = 2197) and 5 RCTs (n = 2654) respectively, to evaluate the impact of EEG monitoring on POD and POCD. These two meta-analyses both reported that the EEG-guided anesthesia could reduce the incidence of POD. However, Kristen et al. did not draw a conclusion on whether EEG monitoring affect POCD due to the high heterogeneity among their included RCTs [16]. Both studies pointed out that the quality of the research evidence was moderate, and further studies should be required to clarify whether the appropriate cerebral function monitoring during surgery can reduce the incidence of PND. Recently, several studies have further explore this important issue. It is worth noting that a large-sample RCT (n = 1213) conducted by Wildes and his colleagues proposed that EEG-guided anesthesia cannot reduce the incidence of POD [18], which is inconsistent with the results from previous large-sample studies [29, 30]. In 2013, Chan et al. performed an RCT including 902 patients and revealed that the incidence rate of PND was lower in patients receiving EEG-guided anesthesia than that in patients receiving routine care [29]. In addition, Radtke and his colleagues analyzed data extracted from 1155 patients and concluded that EEG monitoring correlated with a significant reduction of POD incidence and a decreasing tendency on incidence of POCD [30]. The discrepant findings among the above-mentioned studies may attribute to the differences of their methodology and the heterogeneity of the studied population. Compared to the studies by Chan et al. or Radtke et al. [29, 30], the study conducted by Wildes et al. [18] included patients with more severe conditions as more than 30% of the patients in Wildes’ study had ASA ≥ 3, or had a history of falls, or planned cardiothoracic surgery, which might partially eliminate the effect of EEG on POD [47,48,49]. In our study, a similar conclusion that using EEG cannot reduce the incidence of POD was drawn in the subgroup of patients who underwent cardiac surgeries. For these high-risk patients, it is recommended by several clinical practice guidelines that a multi-component strategy is required to prevent the incidence of POD, indicating that a single approach of monitoring has a limited role in preventing the high-risk patients from POD [50, 51]. In our study, heterogeneity among included studies existed throughout the analysis except the subgroup analysis. Further large-scale RCTs should be conducted to confirm the conclusion.

The underlying mechanisms of the POD prevention by EEG monitoring remains unclear. One hypothesis is that the use of EEG monitoring makes it possible to avoid excessively anesthesia, therefore to specifically reduce the incidence and cumulative duration of intraoperative burst suppression. Previous studies have shown that burst suppression is an independent risk factor of POD [52, 53]. In addition, Hesse et al. have demonstrated that every incidence of burst suppression during the anesthesia maintenance was associated with a 75% increase in odds of POD [54]. Furthermore, higher incidence or longer duration of burst suppression are significantly associated with the incidence of POD [55, 56]. On the other hand, the use of EEG monitoring also reduced the dosage of general anesthetics, such as volatile agents and propofol [57, 58]. Previous studies have reported that excessively exposure to potent volatile agents might increase the incidence of POD [59]. Particularly, most of these studies were performed in geriatric patients whose aging brains were more sensitive to anesthetic agents, therefore, were more likely to experience the burst suppression and POD [60, 61].

Regional cerebral oxygenation monitoring

1380 participants from 10 studies comparing the effect of rSO2 monitoring to routine care was included in our study [19, 35,36,37,38,39,40,41,42,43]. The results indicated that intraoperative monitoring of cerebral oxygenation could reduce POCD, but have no effect on POD.

Prior to our research, Yu et al. conducted a meta-analysis to evaluate the impact of cerebral near infrared reflected spectroscopy (NIRS) monitoring on the following clinical outcomes, including cerebral oxygen desaturation events, neurological outcomes, non-neurological outcomes and socioeconomic impact. The results from the study suggested that the effects of rSO2 monitoring on POCD or POD are uncertain due to the low quality of the evidence and high heterogeneity among included studies [17].

Since the total amount of oxygen consumed by the brain is about 20% of body oxygen supply, the cerebral function is extremely vulnerable to hypoxemia. A study found that 50 to 75% of patients undergoing cardiac surgery experienced once or more rSO2 desaturations during cardiopulmonary bypass (CPB) [62], and that prolonged low rSO2 values have been associated with significantly higher risks of POCD [63, 64]. Evidence indicates that perioperative rSO2 levels below a certain level (50–60%) are associated with an increase in neurological complications and an increase in mortality. It also suggested not to interpret rSO2 levels based on absolute values rather than follow the trend analysis instead, by interpreting the relative changes of rSO2 levels with respect to an individual baseline value [65]. Therefore, rSO2 monitoring has been used to mitigate the cerebral oxygen desaturation during surgery.

NIRS is an emerging noninvasive technique of monitoring brain oxygenation and increasingly being used in various clinical settings. This provides an opportunity for early recognition of imbalances of oxygen delivery and consumption and rSO2 desaturation [66]. Clinicians can take more active treatment measures to prevent prolonged rSO2 desaturation, thereby avoiding neurological and other major complications. However, the clinical benefits of this technology have been questioned. In a multicenter prospective randomized study conducted by Deschamps et al. which included 201 patients [38], the authors found that NIRS-guided intervention can prevented the decreases of rSO2 in cardiac surgery but did not reduce the incidence of PND. These findings were consistent with two single-center RCTs conducted by Lei et al. and Slater et al. respectively [41, 43]. Although our meta-analysis suggest that rSO2 monitoring can reduce POCD, the quality of the included studies is not uniform, and the definition and evaluation methods of POCD were also different. Therefore, further and larger multi-center RCTs are needed to confirm our conclusions.

In our meta-analysis, there were two studies evaluated the effects of the combination of combined use of EEG monitoring and regional cerebral saturation monitoring on reducing PND [44, 45]. The RCT conducted by Ballard and colleagues showed that the use of EEG and rSO2 monitor can significantly reduce the incidence of POCD at1, 12 and 52 weeks postoperatively [44]. But another study conducted by Kunst et al. found that in elderly patients undergoing coronary artery bypass graft surgery (CABG), the combined use of EEG and rSO2 monitoring can reduce the incidence of POD rather than POCD [45]. Nevertheless, It was reported in both studies suggested that the combined use of EEG and rSO2 monitoring reduced the incidence of PND. This result suggested that combination of multiple monitoring approaches is better than one.

NIRS guided rSO2 monitoring can provide clinicians with information on the quality of cerebral oxygen delivery. However, NIRS cannot reflect brain function. EEG can provided objective measures assessing brain cortical function according to differently dynamic waveform, and the generation of electrical activity requires adequate cerebral perfusion and cerebral oxygen. Previous studies showed the observed EEG pattern consisted only of voltage suppression at rSO2 level was less than 20%, while the delta background activity was seen at rSO2 level was greater than 40%. The emergence of delta background activity may be a sign of cortical functional recovery [67]. In addition, Horst et al. investigated the relationship between rSO2 and EEG in preterm infants. The authors concluded that there is a significant relationship between electrocortical activity and oxygen consumption as cerebral oxygen metabolism increases with increasing EEG amplitude [68]. So by the combination of EEG data and NIRS data, the clinicians were able to potentially assess the impact of cerebral oxygen delivery on cortical function as determined via EEG patterns [67]. Both the increased brain oxygen consumption and the decreased brain electrical activity reflect the compromised cerebral oxygen delivery or perfusion, and early warning and intervention may prevent the neurological impairment. Although the potential mechanism of rSO2 on EEG under anesthesia has not been clarified, considering the possibility of intraoperative burst suppression caused by excessive anesthesia and cerebral oxygen desaturation caused by insufficient perfusion, the collaborative application of both monitoring may provide benefits for reducing the incidence of PND.

Several limitations of this meta-analysis should be acknowledged: (1) smaller number of included trials and less solidarity of the results due to the absence of adjustment variables such as age, gender and the type of surgery; (2) there were different scales to assess the incidence of POD or POCD in our included studies, such as CAM/CAM-ICU, DSM-IV, MMSE and other scales which had different specificity and sensitivity. However, we carefully read the definitions of POD and POCD in each original text to extract the data with as little heterogeneity as possible; (3) this study is based on the published articles, the publication bias is inevitable; (4) the analysis of this study is based on data at study-level, whereas the original data from individual patients was not available.

Conclusions

In conclusion, the findings of our study indicated that the use of EEG or/and rSO2 monitor correlated with a lower risk of PND. Based on this, we recommend intraoperative EEG or/and rSO2 monitoring during surgery to decrease the risk of PND. In addition, for high-risk patients, multiple monitoring approaches should be combined to optimize the anesthesia management and to prevent the incidence of PND. Further research should be conducted to verify the identified correlation between the use of EEG or/and rSO2 monitoring and PND.