Incidence of and Risk Factors for Emergence Delirium and Postoperative Delirium in Neurosurgical Patients- A Prospective Cohort Study
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.333461
Source of Support: None, Conflict of Interest: None
Keywords: Emergence delirium, incidence, neurosurgery, postoperative delirium, spectrum
Emergence delirium (ED) after surgery presents in the postoperative period with features of decreased attention and awareness, as well as change in baseline cognition. Manifestations of ED include incoherent speech and inappropriate behavior, such as screaming; hitting; biting; self-removal of monitoring equipment, cables, invasive catheters, and tracheal and drainage tubes; attempts to get off the bed; or lethargic and deeply sedated., On the other hand, postoperative delirium (POD) without identifiable etiology is not temporally related to emergence from anesthesia. These patients often emerge smoothly and may be lucid in the post anesthesia care unit (PACU); however, they may develop a fluctuating mental status, most commonly, between postoperative days one and three. The overall incidence of ED in the adult population varies from 5% to 21% and of POD from 5% to 50%., The consequences of delirium include difficulty in patient care requiring additional material and human resources, prolonged hospital stay and increased cost, and high in-hospital and long-term mortality (4%–17%). Younger age, recent smoking, sevoflurane anesthesia, postoperative pain of numerical rating scale score (NRS) ≥5, and presence of tracheal tube and urinary catheter are identified as risk factors for ED in patients undergoing nonneurological surgeries.,, On the other hand, advanced age, history of delirium, pre-existing cognitive impairment, history of alcoholism, and preoperative use of narcotics and benzodiazepines have been identified as risk factors for POD.,,
Apart from general risk factors for ED/POD, specific risk factors in neurosurgical patients that have been identified as contributing to ED/POD include frontal brain manipulation, lobectomy and disconnection surgeries for epilepsy, cerebrospinal fluid (CSF) loss, and perioperative electrolyte disturbances.,,,,, We hypothesized that incidence of delirium is higher after neurosurgeries than after nonneurosurgical procedures reported previously in the literature and that ED and POD are strongly associated as a spectrum of clinical syndrome.
The primary objective of this study was to evaluate the incidence of ED and POD in patients undergoing neurosurgical procedures. The secondary objectives were (1) to examine the relationship between ED and POD and (2) to identify perioperative risk factors of ED and POD in the neurosurgical population.
Study population and setting
This is a prospective cohort study conducted at the National Institute of Mental Health and Neurosciences, a tertiary neurosciences university hospital in southern India. After obtaining approval of the institute ethics committee, written informed consent was obtained from consecutive adult patients scheduled to undergo elective neurosurgical procedures at our hospital over a 10-month period from February 2018 to November 2018 for inclusion into this study. For logistical reasons, we recruited only first case of the day from each operating room, as we did not have a dedicated research assistant to collect (preoperative and follow-up) data for surgeries performed in the later part of the day in this nonfunded project. We excluded children (age < 18 years), patients with preoperative Glasgow Coma Scale score < 15, and patients with preoperative delirium from our study. This study is registered with the Clinical Trials Registry of India (CTRI/2018/06/014669).
Study conduct and data collection
No preoperative sedation was administered to any patient before the completion of baseline delirium assessment. All study participants were assessed for pre-existing delirium using Riker's Sedation-Agitation Score (SAS) [Supplement 1] and Confusion Assessment Method (CAM) [Supplement 2]. Standard practices of anesthesia and postoperative care were provided for all patients. The attending anesthesiologists were free to select anesthetic drugs and techniques of their choice.
Anesthetic details were retrieved from the electronic or manual anesthesia record. Other data collected were patient comorbidities (diabetes mellitus, hypertension, ischemic heart disease, psychiatric illness, smoking, alcohol abuse, and electrolyte disturbances); details of preoperative medications (psychiatric medications, benzodiazepines, corticosteroids, and anti-Parkinson's medications); anesthetic/analgesic medications and intraoperative details (propofol, inhalational agent in the form of minimum alveolar concentration [MAC] hours, dexmedetomidine, corticosteroid, presence of urinary catheter, blood loss, blood transfusion details, fluid intake/output, core body temperature at extubation, hemodynamic variations, and use of muscle relaxants); and postoperative details (SAS, ED and its treatment, duration of PACU stay, postoperative pain, nausea and vomiting, hemodynamic variations, desaturations, seizures, and neurological deficits). Data of intraoperative opioids were expressed as dose of fentanyl alone as well as morphine equivalents (in milligram) when both fentanyl and morphine were administered. Intraoperative surgical data collected were site of surgery (frontal, parietal, occipital, temporal, and infratentorial surgery); opening of dura and cisterns; and intraoperative complications, such as brain swelling, venous air embolism, prolonged temporary clipping (>10 min] of vessel, vascular injury, intraoperative rupture of aneurysm, and prolonged brain retraction.
ED was assessed in the PACU during the first hour of the postoperative period using SAS, and the highest score was considered for analysis. Patients with SAS of 4 were labeled as normal or having no ED., POD was evaluated using CAM to identify presence of delirium at predetermined time points—after 12 h, 48 h, and 72 h of surgery apart from the baseline assessment before the surgery. We a priori identified the following parameters as putative risk factors for ED and POD based on prior literature and what we considered as clinically important: cranial surgery as compared with spinal; older patients as compared with younger; male gender; higher American Society of Anesthesiologists (ASA) physical status grade; prolonged duration of surgery; smoking (>5 pack years); alcohol abuse; perioperative use of benzodiazepines or steroids; preoperative dementia; dysnatremia; opioid-only intraoperative analgesia as compared with nonopioid adjuvants, such as dexmedetomidine; inhalational anesthesia (sevoflurane, desflurane, and isoflurane) as compared with total intravenous anesthesia; longer MAC-hours; use of pharmacological burst suppression; higher blood loss and transfusions; intraoperative hyper or hypotension; intraoperative desaturation; frontal surgical lesions; CSF loss from cisternal opening or external ventricular drain; and prolonged temporary clipping.
Statistical analysis and sample size estimation
Data were collated offline on a Microsoft Excel spreadsheet (version 2007) and analyzed using R software (version 3.5.3). We report incidences of ED and POD as number and percentages and predictors as proportions or means ± standard deviations as applicable. Univariate and multivariate analyses were performed to identify potential predictors of ED and POD.
We aimed to recruit all eligible patients into the study during the study period. Considering an incidence of ED in nonneurosurgical population as 15% and the incidence of ED after cranial surgery as 29%, the sample size was estimated to be 60, 84, and 107 for a power of 80%, 90%, and 95% respectively. As our ratio of cranial to spine surgery is about 65:35, sample size was inflated by (100/65) to 92, 129, and 165 for a power of 80%, 90%, and 95% to observe a similar incidence of ED. Further to accommodate any possible loss of patients because of sedation and ventilation resulting in inability to assess delirium, we planned to inflate our sample size to 200.
ED and POD prediction modeling
Univariate binary logistic regressions were conducted to identify individual predictors of ED and POD. Significant predictors at P < 0.2 were selected for prediction modeling. Due to potential correlations between predictors of ED/POD and need to incorporate ED as a predictor of POD in the same multivariate model, confirmatory factor analysis (CFA) was chosen as the multivariate modeling technique. The model was estimated using diagonally weighted least squares estimator and robust sandwich type standard error calculation, with no constraints placed on any of the model parameters. The CFA was conducted using lavaan package of R., Odds ratios are presented as results. P <0.05 was chosen as significance level for multivariate tests.
A total of 465 patients undergoing elective cranial and spinal neurosurgery were screened during the study period. Of them, 238 patients were found eligible based on our selection criteria and consented for participation into the study. The flow of the patients into the study is depicted in the flow diagram. [Figure 1]
All the assessments of our study outcomes during the study period were completed by 200 neurological patients, and data of these patients were analyzed. Of these, 157 underwent cranial surgery and 43 underwent spine surgery. The mean age of the patients was 44 ± 14 years, and 133 patients were males in this study.
The incidences of ED and POD are described in [Table 1]. Eighty-two of the 200 neurosurgical patients (41%) developed ED. Patients with ED manifested with self-removal of intravenous lines, urinary catheters, random spitting, and delayed ability for neurological assessment. All patients required more than one health care worker to hold and restrain the patient to the bed, and few of them required pharmacological intervention and urgent imaging to exclude postsurgical complication.
The incidence of POD in all neurosurgical patients was 20% (N = 40). Those patients who developed POD required physical restraints in bed, and the occurrence of POD resulted in delayed discharge from high dependency unit to the ward. Patients with POD needed intravenous haloperidol bolus or dexmedetomidine infusion to control agitation. The incidence of ED and POD occurring in the same patient as a delirium spectrum was 15% (N = 30).
Predictors of ED and POD
The results of univariate logistic regression for predicting ED and POD are tabulated in [Table 2]. The significant predictors for ED at P < 0.2 were type of surgery, age, ASA score, burst suppression, blood loss, intraoperative hypertension, prolonged temporary clipping, and type of inhalational agent. We excluded burst suppression from the model because of a smaller sample (4 in patients with ED and two in patients without ED). Visual analog score (VAS) for pain was assessed in all patients. Average VAS score was 4 and pain was not a significant predictor of ED or POD after univariate logistic regression. Putative predictors for POD at P < 0.2 were gender, dose of fentanyl, inhalational anesthesia, type of surgery, age, and ASA score. For including inhalational agent as a multicategory variable, it was dummy coded into three variables for sevoflurane, desflurane, and isoflurane. As sevoflurane had the highest frequency of use in the data set, desflurane and isoflurane were modeled with sevoflurane as the reference category. The path diagram of multivariate model is shown in [Figure 2], with model fit statistics in the legend. The direction of arrows shows direction of prediction. The results of the model are tabulated in [Table 3]. Patients undergoing spine surgeries were found to have 44% less risk of ED than cranial surgeries (P = 0.032). Males were associated with 2.5 times higher risk of POD than females (P = 0.005), and presence of ED was associated with 1.8 times higher risk of POD (P < 0.001). None of the other factors were significantly associated with ED or POD.
This study evaluated the incidence of ED, POD and their relationship. The study also identified the risk factors for development of ED and POD in elective adult neurosurgical patients. This is probably the first study which attempted to investigate if ED and POD are related as a spectrum. The occurrence of ED and POD coexisting as a continuous spectrum was 15% among all neurosurgical patients. Incidences ED–POD spectrum were 7% after spine surgeries and 17% after cranial surgeries. Demonstration of ED and POD coexisting as a continuum of delirium syndrome is a novel finding of this observational study. This study also confirms that ED is a strong predictor of POD after neurosurgery.
In our study, incidences of ED and POD in neurosurgical patients were 41% and 20%, respectively. ED has been sparsely reported in neurosurgical patients even though it is a commonly witnessed clinical finding after general anesthesia. A recent study reported an incidence of ED of 36% in adult patients undergoing craniotomy for brain tumor resection. The incidence of ED in patients with brain tumor in our study was similar at 40%. However, while frontal location of tumor was a predictor for ED in their study, it was not a risk factor in our study. This difference could be because of the potential contribution of pre-existing delirium in patients with frontal lesions in their study, which was not excluded unlike done in our study.
The incidence of POD varies from 5% to 15% in different studies., Within certain high-risk groups, such as hip fracture involving elderly patients, the range is 16% to 62%, with an average of 35%.,, The incidence of POD in geriatric (>65 years) patients undergoing spine surgery is 21.4% with highest incidence on the first postoperative day. Currently, there are no available data on the incidence of POD in adult neurosurgical patients of all age groups. The mean age of patients in the present study was 44 years and the incidence of POD was 20%, with majority of the patients undergoing cranial surgery. There are no studies in the literature analyzing ED and POD as a part of a disease spectrum.
Various risk factors have been identified as contributing to ED and POD. Predictors for ED in adults after brain tumor surgery were male gender, preoperative use of antidepressant drugs or benzodiazepines, frontal surgery, balanced anesthesia of long duration, and presence of endotracheal tube after surgery. In our study, spine surgery was found to have a 44% lesser risk for ED than cranial surgery, indicating that the cranial surgery itself is a risk factor for ED. Other conventional factors were not found to be statistically significant. Male gender and ED were the only independent predictors of POD in this study. The strong association between ED and POD suggests that ED–POD is a continuing spectrum of postoperative delirium syndrome, especially in those undergoing cranial surgery. In our study, advanced age, male gender, higher ASA grade, and cranial surgery were identified as predisposing risk factors for ED and POD on univariate analysis. Similarly, inhalational anesthesia, opioids, intraoperative hypertension, blood loss, prolonged temporary vascular occlusion (TVO), and occurrence of ED (for POD) were precipitating risk factors.
Self-harm; violence against health care providers; and pulling lines, catheters, and monitoring cables were consequences of ED in other studies, which were similar to our study., Pulmonary complications, such as respiratory depression, reintubation, and acute respiratory distress syndrome, are described as complications of POD. But none of our patients with POD developed respiratory complications during the study period.
This study is not without limitations. First, although our total sample size was large, our spine surgery sample was small, reflecting a skewed distribution of subgroups of neurosurgical population, which could have influenced our findings. In addition, our sample size was not powered to identify the risk factors of ED or POD. Second, SAS was assessed by different observers (attending anesthesiologists); hence, there could be some variations in the interpretation of SAS score. Third, the conduct of anesthesia was not standardized (left to the discretion of individual anesthesiologists), which might have influenced our findings. Fourth, we recruited only the first case of the day because of nonavailability of research personnel daily for data collection of second case late in the evening. Hence, selection bias could have influenced the results of this study. Fifth, study participants with preoperative delirium are not included in the study. All patients are not assessed for preoperative cognitive decline. Such an exclusion, may have eliminated most of the high-risk neurosurgical patients for development of ED or POD. These limitations should be considered when interpreting our findings.
Neurosurgical patients with predisposing (male gender) and precipitating (presence of ED) risk factors are at high risk for development of POD. Education of the health care team and preoperative identification of risk factors and appropriate management plans to decrease the occurrence of ED and POD is essential to reduce occurrence and impact of delirium.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]