Introduction

Delirium is common in elderly surgical patients, including those with Parkinson’s disease (PD), and is predictive of cognitive decline (Davis et al. 2012; Golden et al. 1989; Robinson et al. 2009; Serrano-Duenas and Bleda 2005). While little is known regarding the underlying pathophysiology of delirium, inflammation, sympathetic tone and neurodegeneration are thought to play important roles (Androsova et al. 2015). Surgery under general anesthesia often leads to delirium in elderly patients, particularly those with impaired cognition, suggesting that underlying neuropathology may play a role in the development of postoperative delirium (Gleason et al. 2015; Pervin et al. 2015). Early identification of patients at a greater risk for postoperative delirium would allow for targeted intervention prior to surgery (van Eijk et al. 2010), which may help reduce both the frequency and severity of adverse consequences (Gleason et al. 2015).

Clinical features of α-synuclein-related cognitive disorders, such as dementia with Lewy bodies (DLB) or dementia in PD (PDD), are very similar to delirium and are characterized by visual hallucinations, fluctuations, and rapid eye movement (REM) sleep behavior disorder (RBD) (Golden et al. 1989). In addition, a recent clinical study reported that LB pathology in patients undergoing gastrectomy was associated with postoperative delirium (Sunwoo et al. 2013), suggesting that LB pathology may contribute to postoperative delirium.

Olfactory dysfunction is a well-known marker of underlying Lewy body (LB) pathology commonly seen in patients with PD, DLB, RBD, and even incidental LB disease, and can be used to predict cognitive decline in PD (Baba et al. 2012; Stephenson et al. 2010; Yoon et al. 2015). However, despite these associations, there have been no studies examining the role of olfaction in the context of postoperative delirium in PD.

Here, we hypothesized that low olfaction may be indicative of more severe LB pathology in PD, making these patients more vulnerable to postoperative delirium. To address these issues, we examined the possible associations between baseline olfactory function and the development of postoperative delirium in patients with PD.

Methods

Study participants (n = 144) were selected from the database of Ajou movement registry who fulfilled the following selection criteria: (1) PD ≥60 years of age (2) who took olfactory function test using a Cross-Cultural Smell Identification Test (CCSIT) (within 2 years of disease onset) (3) who had elective surgery under general anesthesia at Ajou University Hospital, Korea. (from October 2006 and June 2013). Most frequent types of operations were orthopedic 68 % (98/144) followed by general 18.7 % (27/144), intracranial 7.6 % (11/144) and vascular 5.5 % (8/144) surgery.

Change in cognition (MMSE, CDR) and UPDRS in patients were evaluated annually as screening tools to detect dementia. Practically, when a patient had evidence of dementia (a score <26 on the Mini-Mental State Examination or a score >0.5 on the Clinical Dementia Rating scale) in a follow-up period, we evaluated several cognitive functions and diagnosed Parkinson’s disease with dementia based on the criteria for probable Parkinson’s disease with dementia (Emre et al. 2007). Patients were excluded if they had (1) loss of follow up (more than 6 months before surgery) (n = 23) or PD dementia before surgery (n = 32) or, (2) high fever within 3 days of the operation (n = 2), (3) laboratory abnormalities, such as hepatic or renal dysfunction (n = 3), (4) significant cerebral lesions observed during magnetic resonance imaging (MRI) and/or computed tomography (CT) scans that may have explained the presence of cognitive disturbances (n = 4), or (5) intracranial surgery (n = 11). Following these exclusions, a total of 77 patients with PD enrolled in this study.

Odor identification was assessed at baseline using the Cross-Cultural Smell Identification (CCSI) test. The CCSI is a widely used test involving scratch-and-sniff of 12 microencapsulated odorants, with a forced choice of 4 alternatives per item; a high score indicates good olfactory performance (Yoon et al. 2015). Demographic information related to age, sex, PD duration, motor unified PD rating scale (UPDRS) score, mini-mental state examination (MMSE) score, clinical dementia rating (CDR) assessed at the last follow-up before surgery, onset of delirium after surgery, and intensive care unit (ICU) admissions were obtained from the patients’ medical records. Preoperative medical comorbidity was assessed using charlson comorbidity index (Charlson et al. 1987).

Patients with postoperative delirium (n = 17) were identified among patients referred to the Department of Neurology or Psychiatry for delirium; all were compatible with the diagnostic criteria of delirium based on the Confusion Assessment Method. Age-, sex-, and PD duration-matched patients without postoperative delirium (n = 17) were randomly selected among the remaining cohort of patients for use as a control group. Ethical approval was given by the local ethics committee, and written informed consent was obtained from each patient.

Statistics

Continuous variables were compared using Fisher’s exact test and categorical variables were examined using the Mann–Whitney U test. Comparisons were made between PD patients with and without delirium after controlling for age, gender, and disease duration. Logistic regression analysis was used to identify independent predictors of postoperative delirium. Independent variables included sex, age, PD duration, operation time, and ICU admission. All statistical analyses were performed using SPSS version 13.0 software. A two-tailed P < 0.05 was considered statistically significant.

Results

Preoperative patient demographics and surgical data are summarized in Tables 1 and 2. No significant differences were observed for age, sex, motor UPDRS, parkinsonian medication, MMSE and CDR (at last follow-up), operation time, pre and post-operative medications or ICU admissions between patients with and without postoperative delirium. Baseline CCSI scores were lower in PD patients with postoperative delirium (4.4 ± 1.5) than those in the unaffected group (6.8 ± 2.4, P < 0.005; Fig. 1).

Table 1 Preoperative demographic data between PD with and without postoperative delirium
Full size table
Table 2 Surgical data between PD with and without postoperative delirium
Full size table
Fig. 1
figure 1

Cross Cultural Smell Identification (CCSI) scores in PD patients with delirium compared with patients without delirium. Box-and-whisker plot revealed that CCSI score was significantly decreased in patients with delirium compared with those without delirium (p < 0.001). The black horizontal line in each box represents the median, with the boxes representing the interquartile range and the whiskers representing the maximum and minimum. Outliers are recorded individually. *p < 0.005

Full size image

Multivariate logistic regression analysis identified olfaction (odds ratio [OR] 0.25; 95 % confidence interval [CI] 0.07–0.86; P = 0.03) and operation time (OR 1.02; 95 % CI 1.00–1.02; P = 0.04) as significant predictors of the development of postoperative delirium (Table 3).

Table 3 Multiple logistic regression analysis to identify independent factor for postoperative delirium
Full size table

ROC curve for estimating diagnostic value of CCSIT for predicting delirium is shown in Fig. 2. The area under ROC curve (AUC) is 0.822 (95 % CI 0.653–0.931, p < 0.001). An optimal cut-off value of CCSIT score of 4 or less yielded a sensitivity of 58.8 %, specificity of 88.2 %, and positive predictive value of 83.3 %. The diagnostic values of CCSIT score at each cut-off value were shown in Table 4.

Fig. 2
figure 2

Receiver operating characteristic (ROC) curves for Cross Cultural Smell Identification (CCSI) scores in predicting postoperative delirium. The area under ROC curve (AUC) is 0.822 (95 % CI 0.653–0.931, p < 0.001)

Full size image
Table 4 Predictive value of CCSIT score for postoperative delirium in PD
Full size table

Discussion

This is the first study to investigate the predictive value of baseline olfactory function for the development of postoperative delirium in patients with PD. Impaired olfaction may be indicative of more severe subclinical LB in PD patients, making them more vulnerable to delirium. The greater risk of delirium in PD patients with impaired olfaction was independent of age, PD duration, MMSE, CDR, operation time and ICU admission.

The management of delirium in patients with PD is often difficult. The most common treatment for delirium is the use of anti-dopaminergic medications; however, these medications are known to exacerbate motor symptoms in patients with both PD and DLB (Pervin et al. 2015). Early identification of high-risk individuals may enable targeted prevention strategies such as cholinesterase inhibitor (Gamberini et al. 2009; van Eijk et al. 2010), which may reduce some of the adverse impacts of postoperative delirium, including the higher frequency and duration of ICU stays, and greater chance of discharge to nursing facilities or dementia (Davis et al. 2012; Gleason et al. 2015).

While olfaction is useful at diagnosing PD, it is also associated with non-motor symptoms of PD, particularly cognition (Baba et al. 2012; Morley et al. 2011; Yoon et al. 2015). These findings suggest that the pathogenic mechanism of olfactory dysfunction may be different from nigrostriatal dopaminergic denervation, indicating more wide spread LB pathology (Bohnen and Muller 2013; Bohnen et al. 2010). Actually, olfaction appears to be associated with cholinergic deficits and altered cortical brain metabolism in PD, and is predictive of dementia susceptibility (Baba et al. 2011, 2012; Muller and Bohnen 2013). Given that delirium group showed low olfaction and require slightly higher dosage of PD medications to control their symptoms (i.e., more levodopa equivalent dose), it is possible that brain was more affected in this group than delirium-free group. Consistent with the data presented here, Brown et al. showed that olfaction could be used to predict postoperative delirium in elderly patients undergoing cardiac surgery, suggesting that olfactory dysfunction may be indicative of a subclinical cholinergic deficit (Brown et al. 2015).

There are some limitations to this study. First, with retrospective analysis using registry, the sample size of the present study was too small to draw a solid conclusion. Thus, the present study should be viewed as hypothesis generating. Second, one may think that MMSE score (25.9) is rather low for intact cognition, thus there is high chance of including dementia. But, in all of patients, Korean version of the Mini-Mental State Examination (K-MMSE) were above the 16th percentile for the age and education appropriate norm (Han et al. 2008) and no evidence of abnormal activities of daily living, judged clinically and by a Korean instrumental activities of daily living (Cho et al. 2008). It is possible that low educational status (<6 years) and old age may affect MMSE score in our patients. However, we were unable to perform a full neuropsychological examination in many cases; therefore, some patients may have already had mild cognitive impairment or very mild dementia at the time of surgery. Third, although dopamine transporter imaging and cardiac 123I-MIBG scintigraphy was used to determine the underlying PD pathology in 29 of 34 study subjects, this study was not autopsy-proven data, meaning that we could not completely exclude the influence of AD-like pathology. Fourth, several important factors, such as type of surgery, anesthetic drug (Lange et al. 2015), details of nursing care, presence of obstructive sleep apnea (Kaw et al. 2006), non-parkinsonian medication use, were not analyzed, which need to be included in future studies. Finally, we only assessed a single component of olfactory function using a brief test, and some odorants included in the CCSI test seemed unfamiliar to elderly Korean people. In addition, we did not perform olfactory testing at regular intervals or shortly prior to surgery as scores may change in some patients, especially, those with mild impairment at onset of PD (Meusel et al. 2010).

PD is a heterogeneous syndrome involving a wide range of systems including dopaminergic, cholinergic, noradrenergic, and serotoninergic neurons. The impaired olfactory function may be reflective of more severe LB burden, indicating specific involvement of the cholinergic system in these patients. Given both the ease of administration and the low cost of the CCSI test, preoperative evaluation of olfactory function may useful in non-demented PD. Finally, given the increasing awareness that delirium itself is an independent risk factor for dementia (Davis et al. 2012), further large and longitudinal studies examining the impact of postoperative delirium on cognitive decline in PD are warranted.