Regional cerebral oxygen saturation and postoperative delirium in endovascular surgery: a prospective cohort study

This prospective observational study was approved by our university’s institutional review board (IRB 0810-758). Written informed consent was obtained from all patients participating in the trial. The trial was registered with http://​www.​clinicalTrials.​gov (NCT02356133; date of registration, February 1, 2015). Our methodology followed the international guidelines for observational studies. The trial was conducted and reported according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) 2010 statement (Fig. 1 and Additional file 1: Table S1). We used G*Power to conduct power analysis. To calculate the required study size, we considered the results of previous studies performed in a similar population. To detect time points in mean rSO₂ values recorded during surgery, accepting a two-tailed α error of 5% and β error of 10%, 39 patients were required. Generally, the limiting values for the reference interval are the 0.025 and 0.975 fractiles of the results distribution in the population. In the present study, only high rSO₂ desaturation score was likely to be of clinical interest. Therefore, the use of the 0.05 fractile as the high reference limit made the most sense [22, 23]. We recruited 43 prospective consecutive patients with intracranial aneurysms who were scheduled for endovascular surgery with general anesthesia between May 1, 2015, and January 1, 2017, at Xuanwu Hospital. Inclusion criteria were as follows: (1) elective intracranial aneurysm embolization, (2) 35–70 years old, (3) American Society of Anesthesiologists (ASA) physical status score of II–IV, and (4) body mass index (BMI) ranging from 22 to 45 kg·m^− 2. Exclusion criteria were as follows: (1) preexisting neuropsychiatric disorders or inability to correctly perform neurocognitive tests on the patient, (2) emergency operation, (3) diagnosis of coma, (4) depression, or (5) cognitive impairment.

Patients’ demographic characteristics were recorded. After admission to the operating room (OR), patients were administered 8 ml·kg^− 1 Ringer solution with an intraoperative maintenance dose of 4 ml·kg·h^− 2. Patients were monitored with a five-lead electrocardiogram (ECG), invasive arterial pressure measurements, oxygen saturation measured by pulse oximetry (SpO₂), and rSO₂. Baseline blood pressure, heart rate, and rSO₂ were acquired 15 min after admission to the OR. Anesthetic induction was performed using intravenous etomidate (0.2 mg·kg^− 1), sufentanil (0.3 μg·kg^− 1), and rocuronium (0.5 mg·kg^− 1). Tracheal intubation was performed for mechanical ventilation when satisfactory muscle relaxation was achieved (approximately 3 min), followed by connection to an anesthesia machine (Datex-Ohmeda Avance S5; GE Healthcare, Chicago, IL, USA) for volume control ventilation with a tidal volume of 8 ml·kg^− 1, respiratory rate of 12·min^− 1, inspiration/expiration ratio of 1:2, and end-tidal carbon dioxide pressure (PETCO₂) of 35–40 mmHg. SpO₂ was maintained at 99–100% during the general anesthesia period. An infusion of 0.06 mg·kg·min^− 2 rocuronium, 0.3 mg·kg·min^− 2 propofol, 0.3 μg·kg·min^− 2 remifentanil, and 0.4–0.6% sevoflurane was administered continuously during surgery. Nimodipine (1–3 ml·h^− 1) was administered to control hypotension during the operation.

The following variables were recorded for all patients: general patient data, including age, sex, weight, height, BMI, preoperative hemoglobin, and ASA physical status; and medical history of hypertension, diabetes mellitus, coronary artery disease, and stroke. Patients who had two systemic vascular comorbidities (including history of hypertension, diabetes mellitus, coronary disease, and/or stroke) were classed as having two risk factors; those with three systemic vascular comorbidities (including history of hypertension, diabetes mellitus, coronary disease, and/or stroke) were classed as having three risk factors. During the operation, we monitored standard noninvasive vital signs, including systolic blood pressure (SBP), diastolic blood pressure (DBP), ECG data, SpO₂, and PETCO₂. In addition, we recorded anesthesia, mechanical ventilation, and operation duration. Finally, rSO₂ indices were continuously recorded in the brain during the operation using a cerebral oximeter (C2030C; CAS Medical Systems Inc., Branford, CT, USA). Briefly, bilateral rSO₂ probes were placed on the patient’s forehead and stabilized. Cerebral oxygen data were recorded bilaterally every minute. The maximum SBP (%) and DBP (%) declines were calculated as follows: ([lowest SBP recorded during the operation – SBP baseline score]/SBP baseline score) × 100. ([lowest DBP recorded during the operation – DBP baseline score]/DBP baseline score) × 100. Baseline SBP and DBP both indicate preoperative levels 15 min after entry into the OR.

To relate intraoperative rSO₂ to clinical outcomes and hemodynamic variables, we used the rSO₂ desaturation score as a measure of the degree of desaturation, where the score was calculated using the baseline and recorded rSO₂ over time [24‐26]. The rSO₂ desaturation score is expressed as a product with the units of percentage per minute, as described below. M = measured rSO₂; B = baseline rSO₂. Z is calculated as the measured rSO₂ – threshold rSO₂ according to the following definition: if M < B, then Z = M; if M ≥ B, then Z = 0. The rSO₂ desaturation score for each patient was calculated using the following formula: rSO₂ desaturation score = (∑ Z) × t, where t = total number of minutes from anesthesia induction until exit from the OR [24‐26], which accounts for both depth and duration of desaturation below the baseline of each patient. Complete rSO₂ datasets were acquired for 39 patients. For patients with bilateral cerebral rSO₂ sensors, mean rSO₂ was used for analysis.

We used the Confusion Assessment Method (CAM) delirium score to evaluate the degree of delirium [27]. Diagnosis of delirium was assessed using an algorithm based on the CAM [28]. CAM diagnostic criteria included attention dysfunction, confusion, level of consciousness, disorientation, memory loss, perception dysfunction, psychomotor excitement and retardation, volatility, and sleep-wake cycle changes [29]. Rapid CAM diagnosis of delirium requires the following four key delirium characteristics: (1) acute onset and disease fluctuations, (2) inattention, (3) disordered thinking, and (4) changes in the level of consciousness (any state of consciousness other than fully conscious). The diagnosis of delirium requires the presence of (1) and (2), accompanied by (3) or (4) or both [24]. The diagnosis of delirium was performed by three nurses who had undergone training for 1 week to ensure consistency.

IBM SPSS Statistics software (version 22.0; IBM, Armonk, NY, USA) was used to analyze the data. The patients were divided into two groups according to the incidence of delirium. The presence or absence of delirium was evaluated as a dichotomous variable. Continuous variables were expressed as mean and 95% confidence interval (95% CI) and were compared using t tests. Categorical data were expressed as frequency and percentage and were compared using chi-square tests. The distribution of data was evaluated using the Kolmogorov-Smirnov test. Significant variables from the t and chi-square tests were included in Spearman’s rank correlation analysis. Logistic regression analysis was performed to analyze the association between delirium and predictors correlated with delirium.

On the basis of the inclusion and exclusion criteria, 39 patients were analyzed in our study. Four patients were excluded for the following reasons: an underlying dementia diagnosis (n = 1), coma (n = 1), and canceled operation (n = 1). Figure 1 shows the selection procedure for the patients. A total of 38 patients were safely discharged from the hospital after surgery; one patient experienced severe delirium and died on the fifth day after surgery. Postoperative delirium occurred in 14 of 39 patients (35%). There were no different of age, height, weight, BMI, Hemoglobin, and ASA physical status in delirium group and no delirium group (Table 1). Preoperatively, there were no significant differences between the groups in terms of patient sex or history of hypertension, diabetes mellitus, coronary disease, and stroke. The number of patients with two risk factors was significantly lower in the no-delirium group (P = 0.010) (Table 2). There was no significant difference in operation duration (P = 0.147) or maximum DBP decline (P = 0.300) between groups (Table 3). Mechanical ventilation duration and anesthesia duration were significantly higher in the delirium group (mechanical ventilation duration, 143.85; 100.51–187.18; anesthesia duration, 168.46; 127.06–209.86) than in the no-delirium group (mechanical ventilation duration, 118.65; 105.18–132.13, P = 0.044; anesthesia duration, 133.46; 120.13–146.79; P = 0.023) (Table 3). The maximum SBP decline was significantly higher in the delirium group (32.98; 28.78–37.19) than in the no-delirium group (24.86; 21.78–27.93; P = 0.002) (Table 3). Furthermore, rSO₂ desaturation score was significantly higher in the delirium group (1267.44; 866.82–1668.07) than in the no-delirium group (231.95; 55.70–408.21; P = 0.001) (Table 3).

The difference in rSO₂ desaturation score between the groups was closely associated with delirium (r = 0.597; P = 0.001) (Table 4). Furthermore, maximum SBP decline was significantly correlated with delirium (r = 0.465; P = 0.003) (Table 4). The incidence of two systemic vascular risk factors was also correlated with delirium (r = 0.486; P = 0.002) (Table 4). However, anesthesia duration (in minutes) and mechanical ventilation duration (in minutes) were not correlated with delirium (anesthesia duration, r = 0.183; P = 0.264; mechanical ventilation duration, r = 0.114; P = 0.489) (Table 4).

Logistic regression analysis revealed that rSO₂ desaturation score was a significant positive predictor of postoperative delirium (OR = 1.002; P = 0.021) (Table 5), suggesting that an increase in rSO₂ desaturation was a surrogate for increased cerebral ischemia and insufficient oxygen supply, leading to an increased risk of postoperative delirium. Table 5 summarizes the results of logistic regression analysis between the potential predictors for delirium. There was no significant association between maximum SBP decline and delirium based on multivariate logistic regression (OR = 1.059; P = 0.512). There was a very weak association between the incidence of two systemic vascular risk factors and postoperative delirium; however, this did not reach statistical significance (OR = 3.593; P = 0.361) (Table 5). To determine the appropriate reference intervals for rSO₂ scores for the best monitoring of controlled hypoperfusion, we calculated the 95% prediction interval based on the one-sided high reference limit among healthy individuals. We found that the threshold rSO₂ desaturation score was 493.96, indicating that a person with a value above this reference limit had a higher risk of delirium.