Incidence and perioperative risk factors for postoperative delirium in the intensive care unit among adult surgical patients – Retrospective cohort study

Abstract

Background and Aims:

Postoperative delirium (POD) is a transient but serious complication that affects cognition and recovery. It may develop immediately after anaesthesia or following an otherwise uneventful emergence. As POD is associated with increased mortality and prolonged hospitalisation, identifying perioperative risk factors is essential. This study aimed to evaluate anaesthetic factors influencing POD during postoperative intensive care unit (ICU) stay.

Methods:

After ethics approval, we retrospectively reviewed ICU-admitted patients who underwent propofol- or desflurane-based general anaesthesia between January and December 2020. Patients who were intentionally sedated or mechanically ventilated postoperatively were excluded. Of 1,040 eligible patients, the POD was assessed using the Confusion Assessment Method for the ICU. Patients were classified into POD and non-POD groups. Demographics, surgical variables, and anaesthetic factors were compared using the Chi-square test, goodness-of-fit test, and Student’s t-test. Multivariate logistic regression was used to identify significant risk factors (P < 0.05).

Results:

POD occurred in 43 patients (4.1%). Affected patients were older, in poorer preoperative health, and had longer anaesthesia and surgery times. Univariate analysis showed associations between POD and age, American society of Anesthesiologists-Physical Status (ASA-PS), dementia, surgery duration, and intraoperative tracheostomy. Multivariate analysis identified age, ASA-PS, surgical site, anaesthesia time, and tracheostomy as independent predictors.

Conclusion:

Prolonged anaesthesia was found to be a modifiable risk factor for POD. Identifying at-risk patients and minimising anaesthesia time may help reduce POD incidence. Persistent delirium beyond 72 hours postoperatively warrants further evaluation.

INTRODUCTION

Delirium is an acute disturbance of attention and cognition that commonly affects older adults and is a frequent, serious, and often under-recognised complication during hospitalisation. When delirium occurs following surgery and anaesthesia, it is termed postoperative delirium (POD), a prevalent and resource-intensive complication of surgical care.[1] Epidemiological studies report delirium in approximately 10%–30% of hospitalised patients overall,[2] with higher rates observed among elderly or critically ill surgical populations. POD is associated with adverse outcomes: affected patients often experience prolonged hospital stays, accelerated cognitive and functional decline, and increased mortality.[2]

The development of POD is multifactorial. Established vulnerability factors include advanced age, baseline cognitive impairment, and severe acute illness, while precipitating factors encompass surgical stress, systemic inflammation, and perioperative medications such as sedatives and analgesics.[1] These predisposing and precipitating factors interact to trigger delirium in susceptible individuals. Recognising high-risk patients is thus essential; contemporary perioperative guidelines emphasise the importance of comprehensive preoperative risk stratification and the implementation of multicomponent preventive strategies by a multidisciplinary team.[1]

The present retrospective cohort study aimed to determine the incidence of POD and to identify independent perioperative risk factors among adult surgical patients admitted to the intensive care unit (ICU). Using a PECO framework, where the population comprised postoperative ICU patients (P), the exposures included patient characteristics and perioperative factors such as anaesthesia duration (E), the comparator was the absence of such exposures (C), and the outcome was the development of POD (O), we employed multivariate logistic regression to explore potential associations. The study was conducted and reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for observational research.

METHODS

This retrospective observational study was approved by the local ethics committee of Tohoku University Hospital (Approval No. 2020-1-1192, dated 22 March 2021). The study was registered with the UMIN Clinical Trial Registry under registration number UMIN000044416 on 3 June 2021, accessible at https://center6.umin.ac.jp/cgi-open-bin/ctr/ctr_view_his.cgi). As the study involved no investigational interventions and only utilised de-identified retrospective data, the local ethics committee waived the requirement for informed consent. Information about this retrospective study was made available to participants on the institution’s website before the study’s commencement. The study was conducted in accordance with the principles outlined in the Declaration of Helsinki (2013 revision) and Good Clinical Practice guidelines.

All surgical cases transferred to the ICU following propofol- or desflurane-based general anaesthesia at our hospital between January 2020 and December 2020 were included. By our routine clinical practice, delirium was assessed in all ICU patients by trained nursing staff during every shift. Patients were evaluated using the Confusion Assessment Method for the ICU (CAM-ICU) at ICU admission, every 8 hours throughout their ICU stay, and upon discharge from the ICU.[3] To ensure assessment consistency, periodic evaluations of inter-rater reliability were conducted to ensure consistency.

A retrospective review of consecutive cases was conducted manually. Data were extracted from anaesthetic records, ICU nursing charts, and clinical records. Cross-verification procedures were implemented to ensure data accuracy, and the handling of missing or inconsistent data was documented. The sole exclusion criterion was the use of intentional sedation or mechanical ventilation at the end of anaesthesia. Patients were classified into two groups: those diagnosed with POD at any point during their ICU stay (Group POD) and those who did not develop POD throughout their ICU stay (Group non-POD).

All statistical analyses were conducted using SAS software, version 9.3 (SAS Institute Inc., Cary, NC, USA) at an independent biostatistics and data centre (STATZ Institute, Inc., Tokyo, Japan). Patient demographics, clinical characteristics, and the incidence of POD during the ICU stay were collected and compared between groups. As this was a retrospective study, a sample size calculation was not performed. Continuous variables, including age, body mass index, body surface area, operative time, and anaesthesia time, were analysed using an unpaired Student’s t-test, assuming normal distribution without formal normality testing. Categorical variables such as gender, history of dementia, history of hypertension, type and site of surgery, timing of operation theatre entry and discharge, anaesthesia method, type of general anaesthetic, intraoperative airway management, and use of intraoperative depth of anaesthesia monitoring were analysed using the Chi-square test. The goodness-of-fit test was used for comparison of the American Society of Anesthesiologists Physical Status (ASA-PS) classification. To identify independent risk factors associated with POD, multivariate logistic regression analysis was conducted using a stepwise forward selection method. Variables with P < 0.05 in univariate analysis were entered into the model. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported. No subgroup analyses were conducted. P < 0.05 was considered statistically significant.

RESULTS

A total of 1,406 patients underwent propofol- or desflurane-based general anaesthesia during the study period. Of these, 366 patients were excluded from the analysis because they were transferred from the operation theatre to the ICU under sedation and/or mechanical ventilation. The remaining 1,040 patients were included in the study, of whom 43 (4.1%) were diagnosed with POD (Group POD), while 997 (95.9%) did not develop POD (Group non-POD) [Figure 1]. Most patients in Group POD exhibited delirium on postoperative day 1 [Figure 2]. The morphometric and demographic characteristics, as well as the surgical and anaesthetic backgrounds of each group, are summarised in Table 1. Patients in Group POD were significantly older (P < 0.001) and had poorer preoperative physical status (P < 0.001) compared to those in Group non-POD. Patients who developed POD had significantly longer durations of anaesthesia (P = 0.021) and surgery (P = 0.027). The proportion of POD cases on each specific postoperative day is shown in Figure 3.

Figure 1.

STROBE flow diagram. STROBE = Strengthening the Reporting of Observational Studies in Epidemiology, POD = Postoperative delirium, n = Number of patients

Figure 2.

Occurrence of postoperative delirium (The number of patients with delirium during their ICU stay on each specified postoperative day is shown). ICU = Intensive care unit

Table 1.

Characteristics of the patients

	Group POD n=43	Group non-POD n=997	P
Demographics
Age (years)	71 (15)	61 (15)	<0.001
Gender (male/female)	24/19	532/465	0.752
BMI (kg/m2)	22.5 (3.4)	23.6 (7.5)	0.366
BSA (m2)	1.6 (0.2)	1.7 (0.4)	0.236
ASA-PS (1/2/3/4)	1/25/16/1	108/677/210/2	<0.001
History of dementia	3	16	0.010
History of hypertension	26	489	0.143
Surgical Factors
Operative time (minutes)a	343 (228)	282 (173)	0.027
Type of surgery (Elective/Emergency)	37/6	910/87	0.240
Surgical site (Neuro/Head and Neck/Thorax/Cardiac/Body Surface/Upper Abdomen/Lower Abdomen/Thorax and Abdomen/Joints/Spine/Delivery)b	9/4/8/6/2/8/6/0/0/0/0	100/38/188/70/20/251/236/36/20/37/1	0.051
Time of OT entry (06:00–18:00/18:00–06:00)	43/0	979/18	0.374
Time of OT discharge (06:00–18:00/18:00–06:00)	24/19	673/324	0.110
Anaesthetic Factors
Anaesthesia time (minutes)c	432 (234)	365 (181)	0.021
Anaesthesia method (general & regional/general)	16/27	530/467	0.040
Type of general anaesthetic (propofol/desflurane)	30/13	814/183	0.051
Intraoperative airway management (SGA/intubation/tracheostomy)	1/40/2	26/968/3	0.003
Use of intraoperative depth of anaesthesia monitoring	23	718	0.009

Data are presented as mean (standard deviation) or as the number of patients, unless otherwise specified. a=Time from skin incision to wound closure. b=Surgical site classification was based on the Japanese Society of Anesthesiologists Perioperative Information Management System (JSA-PIMS). c=Time from connection to a closed-circuit general anaesthesia machine to disconnection, as defined by the Japanese Ministry of Health, Labour, and Welfare. POD=Postoperative delirium; BMI=Body mass index; BSA=Body surface area; ASA-PS=American Society of Anesthesiologists Classification of Physical Status; OT=Operation theatre; SGA=Supraglottic airway

Figure 3.

Proportion of delirious patients (The proportion of patients with delirium, based on data from all 43 patients in Group POD, is shown for each specific postoperative day during their ICU stay). ICU = Intensive care unit, POD = Postoperative delirium

Univariate analysis identified several factors associated with POD [Table 2]. Similarly, multivariate logistic regression analysis using a stepwise regression showed that age (OR: 1.83; 95%CI: 1.35, 2.49) (P < 0.001), ASA-PS classification (OR: 2.53; 95% CI: 1.36, 4.72) (P = 0.003), neurosurgery (OR: 1.97; 95% CI: 1.25, 3.10) (P = 0.003), anaesthesia time (OR: 1.12; 95% CI: 1.01, 1.23) (P = 0.032), and intraoperative airway management with tracheostomy (OR: 3.81; 95% CI: 1.49, 9.72) (P = 0.005) were significantly associated with a higher incidence of POD in ICU [Table 3].

Table 2.

Univariate analysis of postoperative delirium during the intensive care unit stay

	Variables	Number of Patients	Number of Events	Event (%)	OR	(95% CI)	P
Demographics
Age		1040	43	4.1	1.75	(1.33, 2.32)	<0.001
Gender	Female#	484	19	3.9
	Male	556	24	4.3	1.05	(0.77, 1.43)	0.752
BMI		1040	43	4.1	0.96	(0.89, 1.03)	0.222
BSA		1040	43	4.1	0.30	(0.07, 1.32)	0.111
ASA-PS		1040	43	4.1	2.43	(1.42, 4.15)	0.001
History of dementia	No#	1021	40	3.9
	Yes	19	3	15.8	2.14	(1.14, 4.05)	0.019
History of hypertension	No#	525	17	3.2
	Yes	515	26	5.0	1.26	(0.92, 1.72)	0.146
Surgical Factors
Operative time		1040	43	4.1	1.11	(1.01, 1.22)	0.028
Type of surgery	Elective#	947	37	3.9
	Emergency	93	6	6.5	1.30	(0.84, 2.03)	0.245
Surgical site	Non-neurosurgical#	931	34	3.7
	Neurosurgical	109	9	8.3	1.54	(1.05, 2.26)	0.026
Time of operation theatre discharge	18:00–06:00#	343	19	5.5
	06:00–18:00	697	24	3.4	0.78	(0.57, 1.06)	0.114
Anaesthetic Factors
Anaesthesia time		1040	43	4.1	1.11	(1.02, 1.21)	0.022
Anaesthesia method	General & Regional#	546	16	2.9
	General	494	27	5.5	1.38	(1.01, 1.90)	0.043
Types of general anaesthetics	Propofol#	844	30	3.6
	Desflurane	196	13	6.6	1.39	(0.99, 1.94)	0.055
Intraoperative airway management	SGA or intubation#	1035	41	4
	Tracheostomy	5	2	40	4.02	(1.62, 9.97)	0.003
Use of intraoperative depth of anaesthesia monitoring	No	299	20	6.7	1.50	(1.10, 2.04)	0.010
	Yes#	741	23	3.1

. # Reference category. OR=Odds ratio; CI=Confidence interval; BMI=Body mass index; BSA=Body surface area; ASA-PS=American Society of Anesthesiologists Classification of Physical Status; SGA=Supraglottic airway

Table 3.

Multivariate logistic regression model predicting postoperative delirium during the intensive care unit stay

Variables		OR (95% CI)	P
Age		1.83 (1.35, 2.49)	<0.001
ASA-PS		2.53 (1.36, 4.72)	0.003
Surgical site	Non-neurosurgical# vs Neurosurgical	1.97 (1.25, 3.10)	0.003
Anaesthesia time		1.12 (1.01, 1.23)	0.032
Intraoperative airway management	SGA or intubation# vs tracheostomy	3.81 (1.49, 9.72)	0.005

^#Reference category. OR=Odds ratio; CI=Confidence interval; ASA-PS=American Society of Anesthesiologists Physical Status, SGA=supra glottic airway

DISCUSSION

This study identified prolonged anaesthesia duration as a modifiable risk factor for the development of postoperative ICU delirium. Several non-modifiable predisposing and precipitating factors were also found to be associated with the risk. Current evidence does not support specific recommendations regarding anaesthetic agents, dosages, regional/neuraxial blockade as the primary anaesthesia method, or prophylactic medications to reduce POD risk.[4] However, our findings helped clarify the risk factors that anaesthesiologists should consider in preventing POD, as anaesthesia duration is a controllable variable.

Higher delirium rates are commonly observed in geriatric, postsurgical, and ICU patients. In our cohort, only 43 of the 1,040 patients were diagnosed with POD, resulting in an ICU delirium rate of 4.1%. Although relatively low, this rate was considered reasonable given the unique characteristics of our study population. Patients who were intentionally sedated or mechanically ventilated at ICU admission were excluded, as our focus was on anaesthesia-related precipitating factors. Van den Boogaard et al.[5] demonstrated in a multinational cohort study that deeper levels of sedation significantly increased the apparent prevalence of delirium when assessed using the CAM-ICU. The interaction between sedation and delirium assessment was likely even more significant when deep sedation, such as general anaesthesia, was administered without subsequent sedation weaning. Additionally, ICU-specific environmental factors, including mechanical ventilation, medications, and critical illness, contributed to sleep disruption, which were known causes of POD. Sedative agents, such as propofol, benzodiazepines, and opioids, were associated with reduced rapid eye movement sleep. Patient-ventilator asynchrony and hyper-assistance during pressure support ventilation resulted in greater sleep fragmentation compared to assisted control ventilation in sedated patients.[6] The COVID-19 pandemic also influenced the characteristics of our study population. As part of the regional COVID-19 treatment strategy, our ICU prioritised COVID-19 patients, leading to an atypical bed-control system. As a result, scheduled surgeries requiring prolonged ICU stays, such as cardiac and major vascular surgeries, were sometimes performed at other hospitals. Additionally, certain postoperative ICU cases were directly transferred to general wards depending on the ICU’s availability. These conditions reduced the prevalence of high-risk POD surgeries in our cohort, contributing to the low POD incidence. Our structured POD screening system may have also influenced the low POD incidence. Given the fluctuating nature of POD, scheduled screenings play a crucial role in accurate diagnosis. Kappen et al.[7] systematically reviewed POD screening protocols and found that structured screening once versus twice per day increased POD incidence from 20.0% to 36.0%. However, when screening was performed three times per day, as in our study, the incidence unexpectedly decreased to 5.0%. Thus, our structured screening approach is proposed as a significant preventive strategy for POD in high-risk patients. The reported delirium incidence in elderly postoperative patients varies significantly based on diagnostic criteria, screening frequency, and assessor training. Current guidelines recommend screening at least twice daily until postoperative day 3.[1]

Regarding the initial postoperative recovery period, POD typically occurs within 72 hours after surgery.[8] Consistent with previous reports, our findings indicated that POD prevalence gradually decreased after postoperative day 3 [Figure 2], though a substantial number of patients remained delirious. Figure 3 illustrates the proportion of delirious patients in the ICU on each postoperative day. Among the 43 POD patients, seven remained in the ICU until postoperative day 5, and two still exhibited delirium. One patient experienced postoperative ileus, and another patient with preoperative hearing loss developed aphasia following cerebral infarction. Both postoperative pain and sensory deprivation are known POD risk factors.[4] When delirium persists beyond 72 hours postoperatively, clinicians should reassess POD risk factors and consider additional interventions. The COVID-19 pandemic also influenced patient management and discharge practices. Contrary to current standard recommendations,[1] 13 patients in Group POD were discharged to the general ward while still exhibiting delirium. Of these, four were later diagnosed with delirium by a psychological specialist, but only one had POD at ICU discharge. Among POD patients discharged from the ICU, delirium recurrence in the ward was not statistically significant (P = 0.811). Family intervention is a key component of non-pharmacological POD prevention, and restricted family visitation policies during COVID-19 may have contributed to delirium recurrence after ICU discharge.

This study had several limitations. First, it was conducted during the COVID-19 pandemic, which resulted in significant deviations from standard medical practices. The social and medical environment during this period differed from normal conditions, and cases that would have required ICU management under normal circumstances may not have been included. As a result, the true incidence of postoperative ICU delirium in our study was likely underestimated compared to the non-pandemic era. Second, follow-up beyond the ICU setting was limited, as our primary focus was on postoperative ICU delirium. Our findings emphasised the importance of early and sustained delirium management, including screening at-risk patients, addressing modifiable precipitating factors, and implementing preventive strategies until at least postoperative day 3. However, POD is formally defined as delirium occurring within 1 week postoperatively or until hospital discharge, meeting Diagnostic and Statistical Manual of Mental Disorders-5 diagnostic criteria.[8] Some of our patients experienced recurrent delirium during ward recovery, which was beyond the scope of our study. Third, although our regression analysis included key POD risk factors, unmeasured confounders may still exist. Additionally, despite the relatively large sample size, this study was conducted at a single academic centre with a retrospective design, which may limit generalisability. Nonetheless, our study’s major strength lay in its structured POD screening system, which systematically collected and evaluated delirium signs from ICU admission to discharge by using a validated structural delirium assessment tool. This rigorous screening approach enhanced the reliability of our findings and underscored the importance of consistent, structured POD screening in clinical practice.

CONCLUSION

Prolonged anaesthesia and advanced age are significant risk factors for postoperative delirium in intensive care unit; thus, minimising both operative and anaesthetic duration in elderly patients may help mitigate this risk. If delirium persists beyond 72 hours postoperatively, other underlying causes should be considered, as the incidence of postoperative delirium typically declines after this period.

Authors contributions

KK: Concepts, Design, Definition of intellectual content, literature search, investigation, data acquisition, data analysis, statistical analysis, manuscript preparation, manuscript review.

SH: Concepts, Design, Definition of intellectual content, manuscript editing, manuscript review.

Statement on data sharing

De-identified data may be requested with reasonable justification from the authors (email to the corresponding author) and shall be shared after approval as per the authors’ Institution policy.

Disclosure of use of artificial intelligence (AI)-assistive or generative tools

The authors confirm that no AI tools or language models (LLMs) were used in the writing or editing of the manuscript, and no images were manipulated using AI.

Declaration of use of permitted tools

The Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) is freely available and not copyrighted.

Presentation at conferences/CMEs and abstract publication

This article was previously presented as a meeting abstract at Euroanaesthesia 2023 on June 5, 2023.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

This research was supported by the SHISEIKAI Scientific Award.