Postoperative Delirium after Reconstructive Surgery in the Head and Neck Region

Postoperative delirium (POD) is an acute and serious complication following extended surgery. The aim of this study was to identify possible risk factors and scores associated with POD in patients undergoing reconstructive head and neck surgery. A collective of 225 patients was retrospectively evaluated after receiving reconstructive surgery in the head and neck region, between 2013 to 2018. The incidence of POD was examined with regards to distinct patient-specific clinical as well as perioperative parameters. Uni- and multivariate statistics were performed for data analysis. POD occurred in 49 patients (21.8%) and was strongly associated with an increased age-adjusted Charlson Comorbidity Index (ACCI) and a prolonged stay in the ICU (p = 0.009 and p = 0.000, respectively). Analogous, binary logistic regression analysis revealed time in the ICU (p < 0.001), an increased ACCI (p = 0.022) and a Nutritional Risk Screening (NRS) score ≠ 0 (p = 0.005) as significant predictors for a diagnosis of POD. In contrast, the extent of reconstructive surgery in terms of parameters such as type of transplant or duration of surgery did not correlate with the occurrence of POD. The extension of reconstructive interventions in the head and neck region is not decisive for the development of postoperative delirium, whereas patient-specific parameters such as age and comorbidities, as well as nutritional parameters, represent predictors of POD occurrence.


Introduction
Postoperative delirium (POD) is an acute, potentially life-threatening diagnosis with cerebral dysfunction, characterized by alternating sequences of disordered thinking, altered stages of consciousness, and varying degrees of inattention [1,2]. Three different forms have been scientifically described: a hyperactive (agitation, aggressiveness, and hallucination); hypoactive (reduced attention, lethargy, and apathy); as well as a mixed form of both types [3][4][5]. The development of POD is a well-known complication that occurs after major surgery and prolonged anesthesia during an intensive care unit (ICU) stay [6]. Hereby, current literature describes the incidence of POD after head and neck surgery to range between 11% and 26% [7]. Related to that, the total amount of patients who received major surgery was described to postoperatively develop delirium in 10-92% of cases [8]. POD not only causes higher costs for the healthcare system, it is also closely associated with a higher number of complications and a prolonged stay in hospital [9]. Furthermore, the occurrence of POD is linked to a higher mortality rate compared to patients without 2 of 9 delirium [10,11]. In this regard, the key to reducing the risk and severity of POD is the early identification of patients at a high risk of POD through active perioperative management and the postoperative recognition of delirium in the ICU [12][13][14]. However, delirium is still often misdiagnosed or not diagnosed at all [15].
Patients who have undergone free flap reconstructive surgery after major tumor resection in the head and neck area have a higher risk of acquiring delirium due to the surgery duration, the increased rate of malnutrition, and alcohol abuse [7,16]. While many studies have focused primarily on the complications surrounding flap surgery, relatively few have addressed medical complications such as POD in this patient population. Preliminary work on these defined factors, such as older age, male gender, tobacco consumption, the duration of surgery, blood transfusions, the type of graft, and neck dissections, could be described as potential risk factors for POD after extensive tumor surgery [6,7,[17][18][19][20].
Therefore, the aim of this study was to analyze the occurrence of POD, including further factors and scores, in patients who have undergone reconstructive surgery using a microvascular or distant graft in the head and neck area.

Patient Selection and Data Collection
This retrospective monocentric study was conducted at the Department of Cranio-and Maxillofacial Surgery, University Hospital Regensburg, Germany. A total of 494 patients who were reconstructed using free and distant flaps in the oral cavity and head and neck region, between 2013 and 2018, were evaluated. From this collective, 225 patients had a complete delirium documentation as well as a comprehensive survey of other factors mentioned below and were used for the final examination. Pre-, peri-and postoperative data as possible risk factors for POD were collected for the analysis, and included: the degree of comorbidity; nicotine and alcohol abuse; diagnosis for surgical intervention; the region of surgery; previous surgeries in the head and neck; the type and size of reconstruction; distinct flap type; flap success; duration of surgery; tracheostomy passed; wound healing disorders; duration of ICU stay; and Nutritional Risk Screening (NRS) score after sedation. The degree of comorbidity was evaluated using the age-adjusted Charlson Comorbidity Index (ACCI) as previously described in the literature and patients were classified according to the score ( Figure A1) [21]. A successful graft was defined as a functioning graft with no signs of graft loss up to 6 weeks after surgery. The diagnosis of POD, represented by a disturbance of attention, awareness, and cognition, that developed over a short period of time and whose severity fluctuated over the course of a day, was made by both an experienced nurse and the attending physician [2]. The diagnosis was determined in the ICU on the basis of the RASS score and the NRS score. Patients diagnosed with alcohol withdrawal delirium were not included in the study.

Statistical Analysis
Statistical data were collected using IBM SPSS Statistics 26 (IBM Corp. SPSS for Mac, Armonk, NY, USA). Univariate analysis was performed using chi-square test to compare different groups of outcome parameters in the presence or absence of certain pre-and perioperative risk factors. Uni-as well as multivariate correlations/regression analysis between the occurrence of delirium and the individual variables were carried out. Multivariate regression analysis was based on the binary logistic regression method. The significance level was considered as p < 0.05.

Clinicopathological Characteristics of the Patient Cohort
The present retrospective cohort study on patients having received pedicled or microvascular flap transfer in the head and neck region comprises 225 cases. Table 1 shows patient characteristics of the entire cohort. In all, 77 patients were female (34.2%), whereas the majority of the cohort was comprised by male patients (65.8%). POD was diagnosed in 49 patients (21.8%). A total of 99 patients (44.0%) had a positive history of nicotine and alcohol abuse, the mean age was 62.4 years (range: 20-89 years). The most common diagnosis for surgical intervention was oral squamous cell carcinoma (OSCC; 84.4%) followed by osteoradionecrosis (ORN; 10.7%). The majority of patients received reconstructive surgery based on microvascular flap transfer (88.0%), whereas pedicled transplants such as the pectoralis major myocutaneous flap was applied in 10.8% of cases. Additionally, the ACCI was retrospectively calculated for each patient. Hereby, the mean ACCI was 3 points with a range between 0 and 11 points ( Figure 1A).

Clinicopathological Characteristics of the Patient Cohort
The present retrospective cohort study on patients having received pedicled or microvascular flap transfer in the head and neck region comprises 225 cases. Table 1 shows patient characteristics of the entire cohort. In all, 77 patients were female (34.2%), whereas the majority of the cohort was comprised by male patients (65.8%). POD was diagnosed in 49 patients (21.8%). A total of 99 patients (44.0%) had a positive history of nicotine and alcohol abuse, the mean age was 62.4 years (range: 20-89 years). The most common diagnosis for surgical intervention was oral squamous cell carcinoma (OSCC; 84.4%) followed by osteoradionecrosis (ORN; 10.7%). The majority of patients received reconstructive surgery based on microvascular flap transfer (88.0%), whereas pedicled transplants such as the pectoralis major myocutaneous flap was applied in 10.8% of cases. Additionally, the ACCI was retrospectively calculated for each patient. Hereby, the mean ACCI was 3 points with a range between 0 and 11 points ( Figure 1A).

Association of Clinicopathologic Characteristics with Diagnosis of POD
Univariate correlation analysis revealed a highly significant correlation between the occurrence of POD and elevated ACCI in our retrospective cohort (p = 0.009). The diagnosis of POD significantly prolonged patients' stay in the ICU (Table 1, Figure 2B). Furthermore, a trend to develop POD was observed for patients with a positive history of nicotine and alcohol abuse (p = 0.093) and a pathologic Nutritional Risk Screening (NRS) score (p = 0.052; Table 1).
Univariate correlation analysis revealed a highly significant correlation between the occurrence of POD and elevated ACCI in our retrospective cohort (p = 0.009). The diagnosis of POD significantly prolonged patients' stay in the ICU (Table 1, Figure 2B). Furthermore, a trend to develop POD was observed for patients with a positive history of nicotine and alcohol abuse (p = 0.093) and a pathologic Nutritional Risk Screening (NRS) score (p = 0.052; Table 1).

Correlation of Flap-Related Parameters with Diagnosis of POD
As a major aspect of the present cohort study, flap-related parameters were evaluated with regards to the development of POD. Hereby, pedicled and microvascular transplants showed no significant difference in terms of correlation with POD (p = 0.185; Table 1, Figure 3B). Chi-square-test showed no significant correlation for duration of flap surgery, the size of reconstruction, and a performed tracheostomy in univariate statistics (p = 0.568, p = 0.642, and p = 0.660, respectively; Figure 2A, Table 1). Furthermore, the development of POD did not correlate with flap loss or impaired wound healing after reconstructive surgery in the head and neck area (p = 0.568 and p = 0.642, respectively; Table 1, Figure  3A).

Binary Logistic Regression Analysis of POD Diagnosis and Clinicopathological Parameters
In addition to univariate statistics, binary logistic regression was applied to analyse the occurrence of POD with regards to different clinicopathological parameters. Hereby, in line with previous results, time in the ICU was highly significantly correlating with the

Correlation of Flap-Related Parameters with Diagnosis of POD
As a major aspect of the present cohort study, flap-related parameters were evaluated with regards to the development of POD. Hereby, pedicled and microvascular transplants showed no significant difference in terms of correlation with POD (p = 0.185; Table 1, Figure 3B). Chi-square-test showed no significant correlation for duration of flap surgery, the size of reconstruction, and a performed tracheostomy in univariate statistics (p = 0.568, p = 0.642, and p = 0.660, respectively; Figure 2A, Table 1). Furthermore, the development of POD did not correlate with flap loss or impaired wound healing after reconstructive surgery in the head and neck area (p = 0.568 and p = 0.642, respectively; Table 1, Figure 3A). occurrence of POD and elevated ACCI in our retrospective cohort (p = 0.009). The diagnosis of POD significantly prolonged patients' stay in the ICU (Table 1, Figure 2B). Furthermore, a trend to develop POD was observed for patients with a positive history of nicotine and alcohol abuse (p = 0.093) and a pathologic Nutritional Risk Screening (NRS) score (p = 0.052; Table 1).

Correlation of Flap-Related Parameters with Diagnosis of POD
As a major aspect of the present cohort study, flap-related parameters were evaluated with regards to the development of POD. Hereby, pedicled and microvascular transplants showed no significant difference in terms of correlation with POD (p = 0.185; Table 1, Figure 3B). Chi-square-test showed no significant correlation for duration of flap surgery, the size of reconstruction, and a performed tracheostomy in univariate statistics (p = 0.568, p = 0.642, and p = 0.660, respectively; Figure 2A, Table 1). Furthermore, the development of POD did not correlate with flap loss or impaired wound healing after reconstructive surgery in the head and neck area (p = 0.568 and p = 0.642, respectively; Table 1, Figure  3A).

Binary Logistic Regression Analysis of POD Diagnosis and Clinicopathological Parameters
In addition to univariate statistics, binary logistic regression was applied to analyse the occurrence of POD with regards to different clinicopathological parameters. Hereby, in line with previous results, time in the ICU was highly significantly correlating with the

Binary Logistic Regression Analysis of POD Diagnosis and Clinicopathological Parameters
In addition to univariate statistics, binary logistic regression was applied to analyse the occurrence of POD with regards to different clinicopathological parameters. Hereby, in line with previous results, time in the ICU was highly significantly correlating with the diagnosis of POD after reconstructive surgery (p < 0.001; Table 2). Additionally, an elevated ACCI, as well as a NRS score = 0 correlated significantly with POD (p = 0.022 and p = 0.005, respectively; Table 2). While patients with nicotine and alcohol abuse were more likely to develop POD, no significant correlation was observed for flap loss, impaired wound healing, or the type of flap transfer (Table 2).

Discussion
The incidence of POD in our study was 21.8% and had a similar diagnostic rate as previously reported in the literature [17,20]. Any delirium that occurred was treated symptomatically according to the current guidelines of the German Society for Anaesthesia and Intensive Care Medicine [22]. At the beginning, patients were supported by orientation and supportive measures, such as maintaining a day-night rhythm. If sedation or analgesia was required, the selective α2-agonist dexmedetomidine was used, as patients showed a significantly shorter duration of delirium in previous studies [23,24]. For the attenuation of autonomic sympathetic hyperactivity, α-blockers such as clonidine were used. In the case of productive psychotic symptoms, haloperidol, risperidone, and olanzapine were applied in low doses.
A prolonged postoperative stay in the ICU was linked with a higher incidence rate of POD and is in agreement with the results of Kolk et al. [19]. Likewise, a high Charlson score and thus a higher comorbidity as well as older age had a positive influence on POD. In contrast to Yamagata et al., Shah et al., Zhu et al., and Densky et al., and in agreement with Booka et al., we were not able to demonstrate a higher incidence of POD after prolonged surgical procedure [6,7,16,17,20]. In this regard, even extended free flap reconstructive surgery, which are not unusual to last longer than 6 h, did predominantly not result in patients showing signs of POD. Therefore, the choice of graft cannot be considered as a risk factor for POD in our patient cohort, which is a contradictory finding in comparison to previous studies [19]. In detail, we could not detect any difference in patients' frequency to develop a POD between microvascular flaps and pedicled alternatives such as the pectoralis flap.
A pathological NRS score showed up in our results with an increased incidence of POD. Although malnutrition cannot be easily remedied shortly before the usually urgent surgery, it represents an opportunity for improvement in care, provided that the patient remains an inpatient for a certain period of time pre-operatively.
The ACCI, cited over 8800 times in the literature, represents an extensively validated model for the assessment of comorbidities [25,26]. Especially for tumors in the head and neck region, the ACCI has proven to be a valid prognostic indicator to predict the outcome of cancer patients [27]. However, some modifications may be necessary for applicability in head and neck surgery. For example, AIDS as a heavily weighted parameter of comorbidity is not necessarily up-to-date any more, since it can no longer be compared with distant metastatic tumor disease in terms of mortality and thus in its severity-at least in developed countries [28]. Furthermore, there are other scores for determining comorbidity such as the Elixhauser comorbidity score or the comorbidity-polypharmacy score [29,30]. Besides, of all criticism, the ACCI potently predicted the occurrence of POD after major reconstructive surgery in the head and neck area and thereby is, at least in our view, a valuable and easily accessible tool to assess comorbidity and screen for higher patients' risk to develop POD.
Of course, one must bear in mind that delirium diagnosis is partly subjective in nature and depends on the experience of the nurse or the attending physician. Although 494 operated patients were initially considered in the observed period, only 225 patients remained, since the majority of patients did not have sufficient delirium documentation, which therefore leads to the fact that POD could not be 100 per cent confirmed or excluded. Accordingly, the retrospective conception represents the greatest weakness of this study, since it was necessary to rely absolutely on the electronical and paper-based documentation of the cases and questionable entries thereby could not be included in the study. Furthermore, there was no individual risk assessment of delirium in the patients preoperatively, which could certainly show an early hint for clinicians to be aware of an increased patient's risk to develop POD. In addition, there was a profound history of noxious substance abuse in our patient collective, which may well occur more frequently with the primary diagnoses treated here. However, this was not infrequently incorrect due to possible misreporting of the amount of nicotine and alcohol consumption and, moreover, need not correspond to the normal population. Compared with the delirium cohorts from other surgical specialties, however, the incidence was similar [31,32].

Conclusions
The degree of morbidity of each individual patient should be considered with greater caution as a major predictor for developing POD after reconstructive flap surgery in the head and neck region. In contrast, as a fundamental result of the present retrospective cohort study, the sole and especially the extent of flap surgery in the head and neck do not inevitably correlate with the occurrence of POD. This should be kept in mind for planning extensive reconstructive interventions in the head and neck region.  Informed Consent Statement: Based on retrospective analysis, a fully anonymized set of clinical data, and in accordance with the Ethics Committee's decision, signing informed consent was not required.
Data Availability Statement: Data can be obtained by scientists that conducted the work independently from the industry, on request. Data are not stored on publicly available servers.

Conflicts of Interest:
The authors declare no conflict of interest.