Incidence, Outcome And Risk Factors of Postoperative Delirium Subtypes, A Systematic Review And Meta-Analysis

Objective: Systematic review and meta-analysis methodology was used to estimate the pooled incidence, outcome, risk factors of postoperative delirium, including three delirium subtypes: hyperactive delirium, hypoactive delirium, mixed delirium. Methonds: This systematic review and meta-analysis were conducted following the Preferred Reporting Items for Systematic Review and Meta-analyses(PRISMA) guideline. MEDLINE (cid:0) EMBASE (cid:0) CENTRAL were searched for relevant studies. Thirty-two studies from 2714 searched results with 9049 patients were enrolled in this systematic review and meta-analysis. Inclusion criteria were:1) elective surgery population;2) the incidence of delirium subtypes was recorded ;3) cohort studies.;4) language restricted to English. In addition, studies that were randomized control trials (RCT), case reports, or uncertainty in the incidence of delirium subtypes were excluded. The related information was extracted by two reviewers independently.All the analyses were conducted by the STATA (Version 16.0; Stata Corporation, College Station, TX). Result: The study we have performed showed that the highest incidence of postoperative delirium was hypoactive (14%[95%CI,12-16%]), followed by hyperactive (12%[95%CI,10-14%]), and the lowest was mixed delirium(9%[95%CI,7-11%]). Conclusion : The highest incidence of postoperative delirium was hypoactive (14%[95%CI,12-16%]), followed by hyperactive (12%[95%CI,10-14%]), and the lowest was mixed delirium(9%[95%CI,7-11%]). Therefore, it is denitely necessary to update and unify the diagnosis of delirium subtypes based on current tremendous clinical research, thus controlling and adjusting the risk factors of subtypes to reduce the incidence of postoperative delirium and improve patients' prognoses.


Introduction
Delirium is a common acute-onset organ dysfunction characterized by acute and uctuating impairment of attention and consciousness, which is associated with prolonged hospital stay [1], increased risk of death [2,3], and long-term cognitive impairment [4,5]. Based on clinical manifestations, delirium can be divided into three subtypes: hyperactive, hypoactive, and mixed delirium [6][7][8].Postoperative delirium is a relatively common and severe complication that occurs 2-5 days after surgery. It can prolong hospital stays by 2-3 days and is associated with a 30-day mortality rate of 7-10% [9]. The incidence of postoperative delirium in the general surgery population is 2-3%, but it is reported that up to 50-70% of high-risk patients will develop postoperative delirium, especially in elderly patients, who have a neurocognitive impairment and complex or emergency surgery [10]. In addition, the signi cantly increased morbidity and mortality of postoperative delirium are signi cantly associated with increased expenditures on medical resources [11]. In recent years, tremendous clinical studies have revealed the pathophysiological mechanism of postoperative delirium and provided more treatment options for reducing the occurrence of postoperative delirium. [12], but there is no corresponding comprehensive research that has reported the incidence, outcome, and risk factors of postoperative delirium.
The primary purpose of this systematic review and meta-analysis is to explore the incidence of postoperative delirium subtypes and further detect the differences in the outcomes and risk factors of each subtype of delirium.

Search Strategy and Selection Criteria
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) and registered in the prospective international register of systematic reviews(PROSPERO)(registration number is CRD42021230651 All retrieved studies were independently scanned by two investigators for eligibility. Inclusion criteria were:1) elective surgery population;2) the incidence of delirium subtypes was recorded ;3) cohort studies.;4) language restricted to English. In addition, studies that were randomized control trials (RCT), case reports, or uncertainty in the incidence of delirium subtypes were excluded.

Data Extraction And Quality Assessment
Records were managed by EndnNote X9.0 software to exclude the duplicate. The following information from eligible studies were extracted by two reviewers independently: study information; preoperative population demographics, perioperative information, postoperative dat, sample size and reported evaluation, the scale for delirium and subtype evaluation ; the frequency for delirium evaluation.
We used the NEWCASTLE-OTTAWA QUALITY ASSESSMENT SCALE (NOS scale) for quality assessment.
Minor modi cations, removing two items of the control group, were made to the NOS scale, considering our objective is to explore the incidence of delirium subtype after surgery. A total of six items in the modi ed version for quality evaluation. Details of six items are provided in Appendix 2[see additional le 2].

Statistical Analysis
The incidence of postoperative delirium subtypes was calculated by the Stata (Version 16.0; Stata Corporation, College Station, TX). Considering the heterogeneity between studies, random-effect models were used to estimate the average effect and precision, giving a more conservative estimate of the 95% con dence interval(CI). If p<0.05, the pool was considered statistically signi cant. The Cochran Q and I 2 statistics were used to assess heterogeneity between studies.
Publication bias was evaluated by visual inspection of the funnel plot, and sensitivity analyses were conducted by removing each included study at one time to obtain the remaining overall estimate of delirium subtype incidence.

Result
A total of 2714 unique studies were identi ed through systematic search from MEDLINE EMBASE and CENTRAL. After removing duplicates,2356 results were screened, and 737 published articles met inclusion criteria for full-text review, among which thirty-two studies were retained for the nal assessment and quantitative analysis. Figure 1 demonstrates the owchart of study selection.

Quality Assessment
Based on the modi ed NOS scale, twenty-six of thirty-two studies scored 4 points or more, demonstrating low or moderate risks of bias. One critical risk of bias came from selection assessment. Twenty-four studies in thirty-two did not use the MMSE to extract the patients who were in delirium before surgery, leading to selection bias. The second bias mainly came from the outcome assessment. Ttwenty-ve studies had diagnosed delirium subtypes, leaving the seven studies missing the data of mixed delirium. Other bias mainly came from the following aspects: the different evaluation scale and the follow-up period,which have been recommended by the 2017 Interpretation of the European Society of Anesthesiology Evidence-Based and Specialist Consensus Guidelines for Postoperative Delirium ,the follow-up time changed from three days to seven days [13], aiming to reduce the missing diagnosis of postoperative delirium.
Overall Meta-analysis All thirty-two studies, including 9049 patients, were enrolled in the meta-analysis for the incidence of postoperative delirium subtype. We have concluded that the highest incidence of delirium was hypoactive(14%[95%CI,12-16%]), followed by the hyperactive (12%[95%CI,10-14%], and the lowest was the mixed type(9%[95%CI,7-11%])( Figuer2-4). Heterogeneity between studies was signi cant, and I square was 92.9%, 94.1%, and 95.6% in hypoactive, mixed, and hyperactive delirium, respectively. When tested for publication bias (Figure5-7), there is an asymmetry funnel plot in the three types of delirium. A sensitivity analysis that ruled out each study at a time provided that no single affected the overall delirium rate signi cantly.

Duration of delirium
We found that the average duration of delirium was the longest in the mixed type, followed by the hypoactive type, and the hyperactive type was the shortest [15,17,[19][20][21][22][23]

Thirty-days and six-month mortality rate
Studies have shown that there is no consistent result about the thirty-days short-term mortality. Robinson T et al. have reported that the hypoactive delirium mortality rate accounted for 14% (7/50), while the mixed delirium was only 4.3% (1/23) [24]. Research by Lee A. et al. has shown that the 30-day mortality rate: the incidence of hyperactive delirium was 4.2%; hypoactive delirium was 3.8%; the mixed type was 0 (p = 0.26) [15].
In the six months long-term survival rate, the hypoactive and the mixed subtypes were the worst. Bellelli et al. have shown that the six-month survival rate of the hypoactive( 36.8%) was higher than that of the mixed type(29.1%), and the hyperactive (10.7%).And They also performed another Cox survival regression analysis, showing that after adjusting for covariates for the duration, hypoactive subtype (HR = 1.12, 95% CI, 1.011.25) and mixed delirium (HR = 1.11, 95% CI, 1.01-1.21) was independently associated with an increase in 6-month mortality [20].

Incidence of secondary intubation and accidental extubation of delirium subtypes
The enrolled studies have shown that the incidence of secondary intubation and accidental extubation were the worst in hyperactive delirium. Van den Boogaard M. et al. have demonstrated that the incidence of secondary intubation and unintentional removal of the tubes was the highest in hyperactive delirium.

Risk Factors Contributed To The Delirium Subtypes
According to a comprehensive analysis of the enrolled studies, the results have shown that the ASA score, intraoperative blood loss, duration of anesthesia, duration of operation, and the level in ammatory factors IL-6, which may be the risk factors contributed to the mixed delirium.
Additionally, we have found that pain score and type of surgery were associated with hyperactive delirium. Studies have reported that the higher the pain score, the more likely it is to develop hyperactive delirium [25]. And the incidence of hyperactive delirium was highest in cardiac surgery [16]. For hypoactive delirium, Stransky M. et al. reported that preoperative treatment with β-blockers was signi cantly associated with reducing hypoactive delirium [26], which had been reported that highest incidence in was neurological surgery [16].

Discussion
In this systematic review and meta-analysis, 32 studies involving 9049 surgical patients and quantitative analysis were used to evaluate postoperative delirium subtypes' incidence ,outcome, and risk factor. The study we have performed showed that the highest incidence of postoperative delirium was hypoactive delirium, followed by hyperactive, and the lowest incidence was the mixed type. The results are consistent with a study focusing on the incidence of delirium subtypes in adult ICU [27]. Both medical and surgical patients are more prone to hypoactive delirium, which is more likely to be overlooked in clinical practice. That may ascribe to the following reasons: Firstly, the clinical features of hypoactive are more indifferent than the hyperactive and mixed types, leading to hypoactive delirium being di cult to recognize. Secondly, the nature of the health care system. The scarcity of continuity care, the di culty of obtaining the latest records (such as drug changes, recent hospital admissions, or other risk factors such as dementia), and the delayed assessment.Thirdly, misunderstandings within the workforce. It is considered normal for older patients to be disoriented, and hyperactivity symptoms must be present to make a diagnosis; to a certain degree, hypoactive delirium is somehow bene cial to the patient, protecting them from worsening diseases [28].
To our knowledge, this is the rst quantitative analysis of the clinical outcomes and risk factors of postoperative delirium subtypes, making the results of this analysis are more accurate and more convincing than the previous qualitative studies. Regarding the clinical outcome of postoperative delirium, we have summarized that mixed delirium was the worst, including duration of mechanical ventilation, length of hospitalization, ICU stay, and delirium duration. Additionally, it is also worth noting that the highest incidence of accidental removal of the tubes and secondary intubation was hyperactive delirium. Our results indicate that the worst clinical outcome was mixed delirium. The reasons may come from the following aspects. Firstly, the mixed delirium uctuates between the hyperactive and hypoactive [19,30], making it challenging to identify. Secondly, due to the endogenous (severity of the condition, age, sex) and exogenous (surgery, drugs, anesthesia, intubation) and other factors of the patients [29], it is more likely to occur the mixed delirium, leading to prolonged mechanical ventilation time, hospital stay, ICU stay time and duration of delirium, In terms of the risk factors of postoperative delirium subtypes, our ndings suggested that mixed delirium was associated with ASA score, intraoperative blood loss, duration of anesthesia, duration of surgery, and the level of the in ammatory factor IL-6; hyperactive delirium had a relation with pain score and type of surgery; and hypoactive delirium was signi cantly related to preoperative treatment with β-blockers. Regarding other risk factors, a systematic review by Krewulak, K. D has analyzed the risk factors of delirium subtypes in adult ICU: age, gender, APACHE-disease severity score [27]. However, the study results have not shown the relationship between the risk factors and delirium subtypes. Additionally, studies have also concluded that mixed delirium was more related to drug poisoning and metabolic disorders (p<0.001). Patients with hyperactive delirium were younger, related to drug withdrawal and other systemic causes (p<0.001), and hypoactive delirium was associated with dementia, cerebrovascular and systemic infections (p<0.001) [30]. The most signi cant difference between this meta-analysis and the studies mentioned above is that we have focused on surgical patients and have found modi able risk factors during surgery. Among them, the worst outcome was mixed delirium, to which more attention should be paid. Therefore, corresponding measures should be taken to modify risk factors, such as reducing perioperative blood loss [31,32], shortening the duration of anesthesia and surgery as much as possible, controlling in ammatory reactions to reduce the level of in ammatory factors [33][34][35]. In addition, for hyperactive delirium, the risk factor is pain, for which appropriate management could be taken to relieve the pain of the patients [36][37][38][39]. For hypoactive delirium, preoperative use of β-blocker drugs can signi cantly reduce the incidence of hypoactive delirium after surgery. Whether considering the use of beta-blockers preoperative needs further investigation and depends on the actual clinical situation [40,41]. Concerning the risk factors derived from the above analysis, effective measures could be taken to prevent delirium, improve the prognosis of patients, and reduce the incidence of postoperative delirium in surgical patients.
When it comes to the heterogeneity of this article, there are several aspects for consideration. Firstly, one of the sources of heterogeneity may attribute to the different scales of the diagnosis of delirium. Among them, CAM-ICU(n=8), and DSM-IV(n=9). CAPD Scale(n=1), ICDSC(n-1), RASS(n=3). According to the latest guidelines, the gold standard for diagnosing delirium is recommended to be diagnosed according to the fth edition [8]. Secondly, another source of heterogeneity may be the frequency of the diagnostic for delirium. Among the included studies: eleven enrolled studies did not specify the frequency of use of the scale, three of the studies have reported that the frequency of use of the scale was twice a day, and the six studies were three times/day, eleven studies was once a day, what is more, the frequency of a study has reached four times/day. Although the guidelines do not specify the frequency of use of the delirium scale [17,42], a standard frequency delirium diagnostic scale is conducive to the diagnosis of delirium, especially for hypoactive delirium. Third, the source of heterogeneity may also be the difference in the location and year of the research.
This systematic review has the following limitations. Although major databases have been searched, it is still possible that some grey documents have not been explored. At the same time,this meta-analysis is quite heterogeneous after a comprehensive analysis, but we have elaborated the possible reasons for the heterogeneity.On the other hand, papers published by author Krewulak, K. D. et al., executed in October 22, 2017 and August,13,2018 ,respectively, have reported the incidence ,risk factors and outcome of delirium subtype in adult ICU.However,ours data and the enrollred aim population differs from the above studies, great signi cance thas been attached to the postoperative delirium subtypes, which still need more investigation.

Conclusion
The highest incidence of postoperative delirium was hypoactive type, followed by hyperactive delirium ,and the lowest was mixed delirium .What's more, our overall analysis of enrolled studies indicate that the worst outcome was mixed delirium. Thus, supposing adequate measures are taken to control the intraoperative risk factors. In that case, it may be bene cial to reduce the incidence of postoperative delirium, improve the prognosis of patients, and increase the long-term survival rate of patients.

Declarations
Ethical Approval and Consent to participate There is no compete interests.

No applicable
Authors' contributions QL and JF G contributed euqually to the whole process of the manuscript,including the conceptualization and methodology,QL contributed the software ,Jf G helped in inveatigation;QL contributed to writing-original draft preparation, Jf G and JS contributed to writing-review anf editing, Jf G contributed to visualization;JS caontributed to supervision and administrated the project.