Development and validation of a post-operative delirium prediction model for patients undergoing abdominal surgery: A retrospective, observational, single-center study

doi:10.21203/rs.3.rs-2698392/v1

Background

Postoperative delirium (POD) is considered as a relatively common and serious problem after major abdominal surgery procedures. It is presumed to be preventable in most of cases. The purpose of this study was to develop and evaluate a POD prediction model for patients undergoing abdominal surgery.

Methods

From July 2019 to December 2019, patients underwent elective abdominal surgery in our hospital were retrospectively analyzed, and their demographics, pre-operative evaluation, intra-operative and anesthesiologic factors were recorded. Based on the results of the multivariate regression analyzes using P < 0.05 and P < 0.001 as two significance level, we obtained 2 different prediction models comprising of 10 and 4 factors respectively. After factorizing the risk of overfitting and cross-validation, we proposed a final POD prediction model consisting of 4 predictors. From January 2021 to December 2021, 346 more qualifying patients were enrolled for the external validation of the 4-factor model. The study was retrospectively registered on the World Health Organization International Clinical Trials Registry Platform (WHO-ICTRP) with ID ChiCTR2100047405.

Results

After screening, 838 patients were included as the training cohort and 10.9% (91/838) of the patients manifested POD. Those patients who developed POD were more likely to be aged more than 60 years (OR = 1.345, P =0.005), with history of diazepam usage (OR = 3.622, P =0.003), history of cerebrovascular disease (OR = 2.150, P = 0.012) and intraoperative positive fluid balance (OR = 1.41, P <0 .001). The optimum cut-off point of the predicted probability that maximized the sum of sensitivity and specificity was 0.12. The fitting set AUC was 0.703 (95%CI: 0.637–0.753). The cross validation set AUC was 0.684 (SD = 0.068) and the external validation AUC of the model was 0.63 (95%CI: 0.511–0.758), quite closed to that of the fitting set, which indicated that the selected model was robust.

Conclusions

The 4-factor POD prediction model shows good prediction efficiency and can prompt for prophylactic intervention in patients at risk for POD.

Trial registration

: A retrospective registration of the study has been submitted to the World Health Organization International Clinical Trials Registry Platform (WHO-ICTRP) with Registration ID ChiCTR2100047405 (18/06/2021).

Postoperative delirium

abdominal surgery

prediction model

anesthesia

nomogram.

As a relatively common and severe phenomenon after major abdominal surgery procedures, postoperative delirium (POD) poses adverse impacts on postoperative recovery [1–3]. According to previous reports, POD after major surgical procedures ranges from 17 to 61%, and it affects the patient's health and social statuses since POD may cause cognitive impairment, decreased functional independence, dementia, morbidity mortality, prolonged hospitalization time, increased caregiver physical as well as financial burdens [4–5]. In line with previous reports, we noticed that POD is typically manifested during the first 7 postoperative days [6–9]. Nevertheless, after an initial POD episode, post-episode treatment or intervention has little effect on severity, duration, or likelihood of recurrence [10–11]. However, it is presumed that POD can be prevented in 30–40% of cases, which further emphasizes the importance of a more accurate estimation and early identification system of high-risk groups which can prompt for effective prevention and management of POD [12–14].

In order to optimize postoperative outcomes and prevent POD, it is henceforth critical to identify preoperatively the high-risk factors for POD [15–16]. Previous literature has reported several risk factors associated with POD, such as advanced age, precious history of dementia or psychiatric comorbidity, cognitive deficiencies, uncertain polypharmacy, postoperative electrolyte and metabolic disorders and cerebrovascular events [17–21]. Nevertheless, advanced age has been most commonly reported independent risk factor in past reports. In the past, there have been scholars that have tried to build POD prediction models based on the related high-risk factors, however due lack of substantial clinical data or feasibility, those models were not included in routine clinical practice. Therefore, in this study, we have investigated the incidence and risk factors associated with the incidence of POD after abdominal surgery, and developed and validated a POD prediction model.

2.1 Patients Characteristics

In this study, from July 2019 to December 2019, 1000 consecutive patients scheduled for elective abdominal surgery in Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine were retrospectively analyzed. Another group of patients among the same hospital between January 2021 and December 2021 were enrolled to be analyzed as external validation set. The eligible patients were screened. The Inclusion Criteria were as follows: (1) Age ≥ 18 years old; (2) ASA I-IV; (3) Patients undergoing elective surgery; (4) No allergy to conventional anesthetics or contra-indication. The exclusion criteria were as follows: (1) Patients in critical condition (ASA grade V), severe liver and kidney disease; (2) Unplanned secondary surgery due to post-operation complication; (3) Patients who had undergone general anesthesia within three months before operation; (4) Patients who developed delirium preoperatively.

2.2 Data collection

Demographics, preoperative and intraoperative factors were investigated as possible factors for POD development. Demographics included age, gender, body mass index (BMI), level of literacy, alcohol consumption (estimated weekly intake ≥ 200 mL of ethanol), smoking history (at least 10 cigarettes a day for the past 3 years), quality of sleep, history of diazepam use, history of psychiatric disorders, history of cerebrovascular illness and anxiety score rating. Possible pre-operative factors associated with POD prediction included American Society of Anesthesiologists (ASA) physical status classification /score, preoperative blood tests (albumin count, hemoglobin, blood glucose, total bilirubin, serum urea, serum creatinine, serum sodium level, serum phosphorus level and white blood cell count). Intraoperative operative factors included location of surgery, intensive care unit (ICU) admission, duration of surgery, intraoperative Rocuronium bromide, intraoperative Cisatracurium besylate, intraoperative Sufentanil, intraoperative Remifentanyl, total intraoperative fluid infusion, intraoperative blood transfusion, intraoperative blood loss, intraoperative urine output and intraoperative fluid balance during anesthesia.

2.3 Diagnosis of POD

All patient records were carefully scrutinized for consult and progress notes determining clinical manifestations of post-operative delirium. In all cases where the post-operative patient presented with disturbance of consciousness and a change in cognition over a short period of time, a neurological consult was ordered and POD was defined and diagnosed according to the Confusion Assessment Method (CAM), [22].

2.4 Ethics Approval

This retrospective study was approved by the Ethics Committee of Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine. A retrospective registration of the study has been submitted to the World Health Organization International Clinical Trials Registry Platform (WHO-ICTRP) with Registration ID ChiCTR2100047405 (18/06/2021). Written informed consent from all patients involved in the study was exempted due to the retrospective nature of the study.

2.5 Statistical methods

Continuous variables were presented as mean and standard deviation and categorical variables were expressed as counts and proportions. For continuous data, 2-sample t tests were used to compare the difference of two groups, and for categorical factors Chi-square tests were adopted. Factors with P-value less than 0.05 and 0.001 in univariate analysis were respectively entered into the multivariate logistic regression model to fit 2 different predicting models for POD. A backward stepwise variable selection procedure would be performed to select the optimal multivariate logistic regression model with smallest AIC (Akaike information criterion) to avoid overfitting. The point estimates and 95% CIs for the adjusted odds ratios of independent risk factors, as well as the adjusted P-values of covariate coefficients, were provided through the multivariate logistic regression model. The cut-off point of the predicted values was selected to maximize the sum of sensitivity and specificity. In addition, in order to evaluate the prediction performance of the constructed model, we used cross-validation to correct the bias on the basis of the training set, and the mean AUC of the 100 validations was reported. The external validation was conducted using validation set. In addition, we constructed a nomogram on the basis of determinants identified in the multivariate logistic model using rms package in R Project v.3.6.1. The model discrimination performance is evaluated by AUC, and the model calibration is evaluated by calibration curve. All the analyses were performed by R 3.6.1. (The R Foundation for Statistical Computing, Vienna, Austria. http://www.r-project.org) and Statistical Product and Service Solutions (SPSS) 19.0 (IBM, Armonk, NY, USA) software whereby statistical difference was defined by P < 0.05 (2-sided).

3.1 Description of study population

Of the 1000 consecutive patients who underwent elective abdominal surgery that were initially observed for the study, a final total of 838 patients were included in the study as training cohort. 162 patients were excluded due to reasons such as post-op complications leading to second surgery (n=18), severe liver or kidney failure (n=22), ASA Grade V (n=4), history of previous multiple surgeries (n=44), Other complications such as atrial flutter, lung infection and so on (n=36) and incomplete patient history (n=38).

In terms of patient demographics, 364 patients were males (43.4%) while 474 were females (56.6%). The mean age was 65 years (range, 18–86 years).

As external validation set (346 patients), 194 patients were males (56.1%) and 152 were females (43.9%). The mean age was 64 years (range, 19–87 years).

3.2. Incidence of POD

Of the 838 participants, 10.9% (91/838) were diagnosed with POD. The incidence of POD was typically observed 2 days after surgery (range, 1–3 days) and more than 95% of the patients were diagnosed within 3 days. All 91 patients were relieved of POD, with the mean duration of the POD of 3 days. Of the 346 participants as external validation set, 6.65% (23/346) were diagnosed with POD.

3.3. Factors associated with development of POD

3.3.1. Univariate analysis for patients undergoing abdominal surgery with POD

The patients’ demographic characteristics, pre-operative and intraoperative factors between two groups were compared, and the results are shown in Table 1.

The univariate analysis showed that 11 factors: age, history of diazepam usage, history of cerebrovascular disease, history of psychiatric disease, preoperative blood glucose, location of surgery, duration of surgery, intra-operative sufentanyl, total intraoperative fluid intake, intra-operative blood transfusion and intraoperative positive fluid balance had significantly statistical difference (P<0.05).

3.3.2. Multivariate analysis for patients undergoing abdominal surgery with POD

The statistically significant factors in univariate analysis were included as independent variables, and POD was included as the dependent variable in the binary logistic regression analysis. Since the intraoperative positive fluid balance and total intraoperative fluid infusion are strongly correlated with a correlation of 0.883 (P value <10^-12), with reference to previous literature, only positive fluid balance was included in the model [23-25]. By factorizing the significance level into 2 subsets, one P<0.05 and the other P<0.001, we obtained two models. The model with P<0.05 comprised of 10 factors: age, history of diazepam usage, history of cerebrovascular illness, history of psychiatric disease, preoperative blood glucose, location of surgery, duration of surgery, use of sufentanil, intraoperative blood transfusion and intraoperative positive fluid balance. On the other hand, the model with P<0.001 demonstrated that 4 factors including the advanced age (OR = 1.345, P =0.005), history of diazepam usage (OR=3.622, P =0.003), history of cerebrovascular disease (OR=2.150, P=0.012) and the intraoperative positive fluid balance (OR = 1.41, P <0 .001) were independent risk factors for POD development (Table 2).

3.4 Development and validation of the POD Prediction Model

To avoid overfitting, and depending on the result of external validation AUC, the 4-factor model had distinct advantages over the 10-factor model, we chose the selected 4-factor model as our final POD prediction model. The point estimates and 95% CIs for the adjusted odds ratios of covariates, as well as the adjusted P-values of covariate coefficients of the selected model, were reported in Table 2-3.

The AUC based on the prediction of the fitting set was 0.703 (95%CI: 0.637-0.753), indicating a good prediction effect. The optimum cut-off point of the predicted probability that maximized the sum of sensitivity and specificity was 0.12. The mean cross-validation AUC of the selected model was 0.684 (SD=0.068), and the external validation AUC of the selected model was 0.634 (95%CI: 0.511- 0.758) and quite closed to the AUC based on the fitting set (0.703). Moreover, the model’s callibration evaluated by the fitting set calibration curve also indicated that the selected model was robust and performed well in prediction. (Figure 1)

By the constructed prediction model, the predicted probability of a patients experiencing POD is exp(h)/(1+exp(h)), where h = -5.052 + 0.296 age/10 + 1.287 I(history of diazepam usage) + 0.765 I(history of cerebrovascular illness) + 0.349 total intraoperative fluid difference/1000. Here I(·) was the indicator function that took 1 if the condition in the brackets was true and 0 otherwise. The nomogram of the prediction model was given in Figure 2.

POD is a complication that occurs more commonly following major surgeries and is often associated with considerable postoperative morbidity. While previous literature suggests an incidence rate varying around 17–61%, in our study we noted a POD incidence rate of 10.9% (91/838) after elective abdominal surgeries, slightly lower than previous literature. Nevertheless, since POD has been previously associated with higher post-operative morbidity and mortality rates in terms of cognitive decline, identification of risk factors associated with POD during the preoperative assessment may prompt for prophylactic strategies to reduce its incidence with the ultimate goal of improving patient outcomes.

Based on past literature and theory, it is presumed that POD evolves and manifests as a result of a combination of the predisposing and precipitating factors [26, 27]: As mentioned before, advanced age, pre-history of cognitive deficiencies, previous episodes of dementia, psychological or emotional disruptions, multiple or specific comorbidities such as cerebrovascular events and alcohol abuse have been identified as common predisposing risk factors. Nevertheless, other studies have identified higher ASA physical status, non-elective surgery and higher postoperative pain scores as risk factors for the development of POD [28–35]. Diabetes Mellitus (DM) is yet the most frequent comorbidity or metabolic disease that has been associated with POD and the underlying mechanism has often been linked to the polyneuropathy associated with advanced DM but further clarity about the mechanism behind its correlation with POD is required [36–37].

During the initial statistical evaluation of the study, using P < 0.05 as the determining statistical criterion, we shortlisted 10 factors that showed correlation with the incidence of POD: age, history of diazepam usage, history of cerebrovascular illness, history of psychiatric disease, preoperative blood glucose, location of surgery, duration of surgery, use of sufentanil, intraoperative blood transfusion and intraoperative positive fluid balance. When comparing our results with previously reported POD models, we found that factors such as duration of surgery, as reported by Ida et al, could increase the incidence of POD and the probable reason behind was hypothesized to be larger extent of systemic inflammation due to longer exposure to surgical stress but this needs further research [38]. Factors such as age and hyperglycemic conditions have been reported before. Furthermore, we found that fluid infusion and fluid intake-outtake balance during surgery were interesting factors contributing to POD. In models of POD prediction for cardiopulmonary and neurosurgeries, the intraoperative fluid balance was a common factor that showed clinical and statistical significances. The impact of intravenous isotonic and hypotonic maintenance fluid on the risk of delirium in adult postoperative patients was investigated by Nagae et al and they found out that fluid infusion was indeed related to POD [39]. Positive fluid balance during surgery was significantly related to postoperative delirium, the reason is that fluid overload affects hemoglobin and hematocrit levels and electrolyte balance. Smulter et al. believe that increasing the interstitial space with the formation of edema might interfere with brain function [23]. Nevertheless, we also found that the amount of sufentanil were also factors contributing to the incidence of POD. Various aspects of anesthesia such as depth, type and pharmacy have been investigated before but no concrete conclusion has been reached as to how POD can be effectively avoided during anesthesia [5, 40]. However, since the amount of sufentanil are very subjective factors, varying from anesthesiologist to anesthesiologist, institution to institution, patient to patient, we decided that it would be rational not to include subjective variants in the model. Hence, the first initial model comprised of 10 factors (age, history of diazepam usage, history of cerebrovascular illness, history of psychiatric disease, preoperative blood glucose, location of surgery, duration of surgery, use of sufentanil, intraoperative blood transfusion and positive fluid balance). In order to improve the performance of the model and avoid overfitting, we set a second set of statistical criterion and used P < 0.001 to define the differential factors associated with the incidence of POD and 4 factors were obtained: age (P = 0.005), history of diazepam usage (P = 0.003), history of cerebrovascular disease (P = 0.012) and intraoperative fluid balance (P < 0.001). When comparing the prediction ability of both the 10-factor and 4-factor models, the results were basically similar. To avoid overfitting, we chose the selected 4-factor model as our final POD prediction model.

In externally validated POD prediction models, advanced age, pre-history of cognitive deficiencies, functional and sensory aptitudes have been the most common variables. Yet, the lack of uniformity and standardization in the definition of the variables led to inconsistencies in the application and comparison of the prediction models, whereby the handling of critical data such as functional, physical or cognitive impairments were vaguely defined and assessed. In an attempt to decrease such nuances, some POD delirium prediction model excluded cognitive variables and instead used hearing impairment as a predictive variable [41–42]. However, previous research and clinical practice actually demonstrates that cognitive deficiency is one of the most common underlying factors for delirium development [43–45]. However, the lack of more precise and reliable tools to evaluate subtler cognitive decline limits the appropriate inclusion of cognitive deficiency in prediction model. There are previous models that have used the evaluation criteria for dementia but in the case of POD, these criteria are not precise and accurate enough.

A good clinical model is based on its efficacy at predicting risk and its feasibility for clinical application. In this study, by factorizing the results of the multivariate analysis, and clinical significance, the developed POD prediction model comprised of 4 predictors: age, history of diazepam use, history of cerebrovascular illness and intraoperative positive fluid balance. Receiver operating characteristic curve (ROC) analysis revealed an AUROC of 0.703, indicating a good prediction effect.

Nevertheless, this study has some limitations. First, this study was a single-center study, which restricted its extrapolation of the results. Second, the purpose of this retrospective study was to identify the potential perioperative risk factors associated with POD; yet the lack of uniformity and accuracy between certain clinical variables affects the overall prediction effect of the model. Henceforth, we are looking forward to validating the POD prediction model in a prospective trial, with objective to be able to improve the sensitivity and specificity of detecting patients at high risk of POD and implementing prophylactic strategies to decrease the burden that POD brings on the disease course of the patients.

The 4-factor POD prediction model (age + history of diazepam use + history of cerebrovascular illness + total intraoperative fluid difference) shows good prediction efficiency and can prompt for prophylactic intervention in patients at risk for POD.

Abbreviation	Full name
POD	Postoperative Delirium
OR	Odds Ratio
CI	Confidence Interval
AUC	Area under the Curve
SD	Standard Deviation
WHO-ICTRP	World Health Organization International Clinical Trials Registry Platform
ASA	American Society of Anesthesiologists
BMI	Body Mass Index
ICU	Intensive Care Unit
CAM	Confusion Assessment Method
AIC	Akaike Information Criterion
DM	Diabetes Mellitus

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of Ruijin Hospital affiliated to Shanghai Jiao Tong University School of Medicine ( (2021) Clinical Ethics Approval No.86). A retrospective registration of the study has been submitted to the World Health Organization Clinical Trials - International Registry Platform (WHO-ICTRP) with Registration ID ChiCTR2100047405 (18/06/2021). Written informed consent from all patients involved in the study was exempted due to the retrospective nature of the study. This study complied with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

This work was supported by grants from the National Natural Science Foundation of China (No: T2293734).

Authors’ Contributions

Z-H Hand M-K B performed the experiments and drafted the manuscript. X-Y X participated in acquisition of data, C-R B and L T analyzed and interpreted the data. Y L was in charge of the study design and critically revised the manuscript. All authors approved the final manuscript for publication.

Acknowledgements

We acknowledge all health-care workers involved in the diagnosis and treatment of patients in Ruijin Hospital. We also appreciate Dr Jian Li (Clinical Research Center, Ruijin Hospital, Shanghai, China) for critical review for manuscript.

McDonald SR, Heflin MT, Whitson HE, Dalton TO, Lidsky ME, Liu P, et al. Association of integrated care coordination with postsurgical outcomes in high-risk older adults: the perioperative optimization of senior health (POSH) initiative. JAMA Surg (2018) 153 (5): 454-462.
Chen CC, Li HC, Liang JT, Lai IR, Purnomo JDT, Yang YT, et al. Effect of a modified hospital elder life program on delirium and length of hospital stay in patients undergoing abdominal surgery: a cluster randomized clinical trial. JAMA Surg (2017) 152 (9): 827-834.
Goldberg TE, Chen C, Wang Y, Jung E, Swanson A, Ing C, et al. Association of delirium with long-term cognitive decline: a meta-analysis. JAMA Neurol (2020) 77 (11): 1373-1381.
Sadlonova M, Vogelgsang J, Lange C, et.al. Identification of risk factors for delirium, cognitive decline, and dementia after cardiac surgery (FINDERI-find delirium risk factors): a study protocol of a prospective observational study. BMC Cardiovasc Disord. 2022;22(1):299.
Janssen TL, Alberts AR, Hooft L, Mattace-Raso F, Mosk CA, Laan L. Prevention of postoperative delirium in elderly patients planned for elective surgery: systematic review and meta-analysis. Clin Interv Aging (2019) 14:1095-1117.
Ahrens E, Tartler TM, Suleiman A, et.al. Dose-dependent relationship between intra-procedural hypoxaemia or hypocapnia and postoperative delirium in older patients. Br J Anaesth. 2023;130(2):e298-e306.
Jiang JL, Zhang L, He LL, et.al. Volatile Versus Total Intravenous Anesthesia on Postoperative Delirium in Adult Patients Undergoing Cardiac Valve Surgery: A Randomized Clinical Trial. Anesth Analg. 2023;136(1):60-69.
Ackenbom MF, Zyczynski HM, Butters MA, et.al.Postoperative delirium in older patients after undergoing pelvic organ prolapse surgery. Int Urogynecol J. 2023;34(1):201-209.
Li T, Li J, Yuan L, et.al. Effect of Regional vs General Anesthesia on Incidence of Postoperative Delirium in Older Patients Undergoing Hip Fracture Surgery: The RAGA Randomized Trial. JAMA. 2022;327(1):50-58.
Burry L, Mehta S, Perreault MM, JS Luxenberg JS, Siddiqi N, Hutton B,et al. Antipsychotics for treatment of delirium in hospitalised non-ICU patients. Cochrane Database Syst Rev (2018) 6 (6):CD005594.
Friedman JI, Soleimani L, McGonigle DP, Egol C, Silverstein JH. Pharmacological treatments of non-substance-withdrawal delirium: a systematic review of prospective trials. Am J Psychiatry (2014) 171(2):151–159.
Inouye SK, Westendorp RG, Saczynski JS. Delirium in elderly people. Lancet (2014) 383(9920):911–922.
Deeken F, Sánchez A, Rapp MA, Denkinger M, Brefka S, Spank J, et al. Outcomes of a Delirium Prevention Program in Older Persons After Elective Surgery: A Stepped-Wedge Cluster Randomized Clinical Trial. JAMA Surg (2022) 157(2):e216370.
Humeidan ML, Reyes JC, Mavarez-Martinez A, Roeth C, Nguyen CM, Sheridan E, et al. Effect of cognitive prehabilitation on the incidence of postoperative delirium among older adults undergoing major noncardiac surgery: the neurobics randomized clinical trial. JAMA Surg (2021) 156 (2): 148-156.
Staveski SL, Pickler RH, Khoury PR, et.al. Prevalence of ICU Delirium in Postoperative Pediatric Cardiac Surgery Patients. Pediatr Crit Care Med. 2021;22 (1):68-78.
Hughes CG, Boncyk CS, Culley DJ,et.al. American Society for Enhanced Recovery and Perioperative Quality Initiative Joint Consensus Statement on Postoperative Delirium Prevention. Anesth Analg. 2020; 130 (6):1572-1590.
Huang H, Li H, Zhang X,et.al. Association of postoperative delirium with cognitive outcomes: A meta-analysis. J Clin Anesth. 2021;75:110496.
Zhou Q, Zhou X, Zhang Y, et.al. Predictors of postoperative delirium in elderly patients following total hip and knee arthroplasty: a systematic review and meta-analysis. BMC Musculoskelet Disord. 2021;22(1):945.
Hiraki M, Tanaka T, Ishii H.et.al. A clinical risk analysis of early post-operative delirium after laparoscopic colorectal cancer surgery in elderly patients: a retrospective study. Int J Colorectal Dis. 2021;36(7):1461-1468.
de Jong L, Rijckevorsel VA, Raats JW, Klem TM, Kuijper TM, Roukema GR. Delirium after hip hemiarthroplasty for proximal femoral fractures in elderly patients: risk factors and clinical outcomes. Clin Interv Aging (2019) 14:427–435.
Huang J, Sprung J, Weingarten TN. Delirium following total joint replacement surgery. Bosn J Basic Med Sci (2019) 19 (1):81–85. doi: 10.17305/bjbms.2018.3653
Inouye SK, van Dyck CH, Alessi CA, et al. Clarifying confusion: the confusion assessment method. A new method for detection of delirium. Ann Intern Med. 1990; 113 (12): 941–948.
Smulter N, Lingehall HC, Gustafson Y, et al. Disturbances in Oxygen Balance During Cardiopulmonary Bypass: A Risk Factor for Postoperative Delirium. J Cardiothorac Vasc Anesth. 2018;32(2):684-690.
Mailhot T, Cossette S, Lambert J, et al. Delirium After Cardiac Surgery and Cumulative Fluid Balance: A Case-Control Cohort Study. J Cardiothorac Vasc Anesth. 2019;33(1):93-101.
Nguyen DN, Huyghens L, Parra J, et al. Hypotension and a positive fluid balance are associated with delirium in patients with shock. PLoS One. 2018;13(8):e0200495.
Wittmann M, Kirfel A, Jossen D, et al. The Impact of Perioperative and Predisposing Risk Factors on the Development of Postoperative Delirium and a Possible Gender Difference. Geriatrics (Basel). 2022;7(3):65.
Eschweiler GW, Czornik M, Herrmann ML, et al. Presurgical Screening Improves Risk Prediction for Delirium in Elective Surgery of Older Patients: The PAWEL RISK Study. Front Aging Neurosci. 2021;13:679933.
Kim, EMi, Li G, Kim M. Development of a Risk Score to Predict Postoperative Delirium in Patients With Hip Fracture. Anesthesia & Analgesia (2020). 130 (1), 79–86.
Wang Y, Yu H, Qiao H. Risk Factors and Incidence of Postoperative Delirium in Patients Undergoing Laryngectomy. Otolaryngology–Head and Neck Surgery 2019, Vol. 161(5) 807–813.
Ziman N, Sands LP, Tang C,et al. Does postoperative delirium following elective noncardiac surgery predict long-term mortality? Age Ageing. 2020;49(6):1020-1027.
Lee SJ, Sung TY. Emergence agitation: current knowledge and unresolved questions. Korean J Anesthesiol. 2020;73(6):471-485.
Jin Z, Hu J, Ma D. Postoperative delirium: perioperative assessment, risk reduction, and management. Br J Anaesth. 2020;125(4):492-504.
O'Gara BP, Gao L, Marcantonio ER, et al. Sleep, Pain, and Cognition: Modifiable Targets for Optimal Perioperative Brain Health. Anesthesiology. 2021;135(6):1132-1152.
Tang C, Hu Y, Zhang Z, et al. Dexmedetomidine with sufentanil in intravenous patient-controlled analgesia for relief from postoperative pain, inflammation and delirium after esophageal cancer surgery. Biosci Rep. 2020;40(5):BSR20193410.
Culley DJ, Flaherty D, Fahey MC, Rudolph JL, Javedan H, Huanget C, et al. Poor Performance on a Preoperative Cognitive Screening Test Predicts Postoperative Complications in Older Orthopedic Surgical Patients. Anesthesiology (2017) 127 (5):765–774.
Xing H, Zhou W, Fan Y, et al. Development and validation of a postoperative delirium prediction model for patients admitted to an intensive care unit in China: a prospective study. BMJ Open. 2019;9:e030733.
Chaiwat O, Chanidnuan M, Pancharoen W, et al. Postoperative delirium in critically ill surgical patients: incidence, risk factors, and predictive scores. BMC Anesthesiology. 2019;19(1):39.
Ida M, Takeshita Y, Kawaguchi M. Preoperative serum biomarkers in the prediction of postoperative delirium following abdominal surgery. Geriatr Gerontol Int (2020) 20(12):1208-1212.
Nagae M, Egi M, Furushima N, Okada M, Makino S, Mizobuchi S, et al. The impact of intravenous isotonic and hypotonic maintenance fluid on the risk of delirium in adult postoperative patients: retrospective before-after observational study. Journal of anesthesia (2019) 33 (2) : 287-294.
Li X, Wang Y, Liu J, Xiong Y, Chen S, Han J, et al. Effects of perioperative interventions for preventing postoperative delirium: A protocol for systematic review and meta-analysis of randomized controlled trials. Medicine (Baltimore) (2021) 100(29):e26662.
Visser L, Prent A, Banning LBD, et al. Risk Factors for Delirium after Vascular Surgery: A Systematic Review and Meta-Analysis. Ann Vasc Surg. 2021;76:500-513.
Sanson G, Khlopenyuk Y, Milocco S, et al. Delirium after cardiac surgery. Incidence, phenotypes, predisposing and precipitating risk factors, and effects. Heart Lung. 2018;47(4):408-417.
Huang H, Li H, Zhang X, et al. Association of postoperative delirium with cognitive outcomes: A meta-analysis. J Clin Anesth. 2021;75:110496.
Duning T, Ilting-Reuke K, Beckhuis M, et al. Postoperative delirium - treatment and prevention. Curr Opin Anaesthesiol. 2021;34(1):27-32.
Kang T, Park SY, Lee JH, et al. Incidence & Risk Factors of Postoperative Delirium After Spinal Surgery in Older Patients. Sci Rep. 2020;10(1):9232.

TABLE 1 A, B, C Comparison of the demographics, preoperative and intra-operative characteristics for patients undergoing abdominal surgery with and without POD.

A. Univariate analysis of the demographics for patients undergoing abdominal surgery with and without POD.

Parameters	No POD Group (n=747)	POD Group (n=91)	P Value
Demographic Data
Sex
Male	319 (42.7%)	45 (49.5%)	0.27
Female	428 (57.3%)	46 (50.5%)
Age (years old)	58.7± 12.9	63.8± 11.7	<0.001
BMI	22.8± 3.26	23.2± 3.35	0.21
Level of Literacy			0.37
< 6 years of schooling	187 (25.0%)	29 (31.9%)
< 12 years of schooling	383 (51.3%)	43 (47.3%)
> 12 years of schooling	177 (23.7%)	19 (20.9%)
Alcohol Consumption			0.15
No	635 (85.0%）	83 (91.2%）
Yes	112 (15.0%）	8 (8.8%）
Cigarette Smoking			0.28
No	619 (82.9%）	80 (87.9%）
Yes	128 (17.1%）	11 (12.1%）
Quality of Sleep			0.51
Good	554 (74.2%）	64 (70.3%）
Poor	193 (25.8%）	27 (29.7%）
History of Diazepam usage			<0.001
No	728 (97.5%）	82 (90.1%）
Yes	19 (2.5%）	9 (9.9%）
History of Psychiatric Disorders			0.03
No	727 (97.3%）	84 (92.3%）
Yes	20 (2.7%）	7 (7.7%)
History of Cerebrovascular Disease			<0.001
No	679 (90.9%）	71 (78.0%）
Yes	68 (9.1%）	20 (22.0 %)

B. Univariate analysis of the preoperative characteristics for patients undergoing abdominal surgery with and without POD.

Parameters	No POD Group (n=747)	POD Group (n=91)	P Value
Pre-operative Evaluation
ASA Score			0.69
1 (I-II)	656 (87.8%）	78 (85.7%）
2 (III-IV)	91 (12.2%）	13 (14.3%）
Albumin Level (g/L)			0.36
Normal (≥35)	654 (87.6%）	76 (83.5%）
Low (<35)	93 (12.4%）	15 (16.5%）
Hemoglobin Count (g/L)			0.59
Normal (≥120)	494 (66.1%）	57 (62.6%）
Low (<120)	253 (33.9%）	34 (37.4%）
Blood Glucose (mmol/L)	5.8± 2.0	6.5± 2.9	0.03
Total Bilirubilin Count (umol/L)	22.6± 39.0	28.8± 49.7	0.25
Serum Urea Count (mmol/L)	5.5± 8.21	5.1± 1.63	0.19
Serum Creatinine (umol/L)	73.7± 18.7	73.7± 17.8	0.99
Sodium Level (mmol/L)	141± 6.5	141± 2.9	0.41
Phosphorus Level (mmol/L)	1.1± 0.2	1.1± 0.2	0.68
White Blood Cell Count (x10⁹/L)	6.0± 3.3	5.6± 1.7	0.05

C. Univariate analysis of the intra-operative characteristics for patients undergoing abdominal surgery with and without POD.

Parameters	No POD Group (n=747)	POD Group (n=91)	P Value
Intra-operative Parameters
Location of Surgery			0.04
Upper Abdomen	622 (83.3%）	84 (92.3%）
Lower abdomen	125 (16.7%）	7 (7.7%）
ICU Admission			0.05
No	484 (64.8%）	49 (53.8%）
Yes	263 (35.2%）	42 (46.2%）
Duration of Surgery			0.004
<3h	266 (35.6%）	18 (19.8%）
≥3h	481 (64.4%）	73 (80.2%）
Intra-operative Rocuronium Bromide (mg/Kg)	1.2± 0.5	1.3± 0.6	0.08
Intra-operative Sufentanyl (ug/Kg)	0. 9±0.2	1.0± 0.3	0.01
Intra-operative Remifentanyl (mg/Kg)	0. 02±0.03	0.02± 0.01	0.96
Total Intraoperative Fluid intake (ml)	3490±1570	4040±1300	<0.001
Intra-operative Blood Perfusion			0.01
No	463 (62.0%）	43 (47.3%）
Yes	284 (38.0%）	48 (52.7%）
Total Intraoperative Blood Loss (ml)	200 [5 to 8000]	200 [20 to 1600]	0.33
Total Intraoperative Urine Output (ml)	600 ±445	695 ± 464	0.07
Intraoperative Positive Fluid Balance (ml)	2570 (± 989)	2970 (± 1060)	<0.001

TABLE 2 Multivariate logistic analysis of factors associated with POD

Factor	Estimate	Std. Error	Z Value	P Value	OR	95% CI
Factor	Estimate	Std. Error	Z Value	P Value	OR	Low	High
Age	0.030	0.011	2.778	0.005	1.345	1.098	1.668
Use of Diazepam	1.287	0.440	2.922	0.003	3.622	1.464	8.375
History of Cerebrovascular Disease	0.765	0.306	2.498	0.012	2.150	1.156	3.863
Intraoperative Positive Fluid Balance	0.000	0.000	3.640	0.000	1.417	1.172	1.713

TABLE 3 Evaluation of different prediction models

Number of factors

Factors

Sensitivity

Specificity

AUC

External Validation AUC

10

age+history of diazepam usage+ history of psychiatric disorders+history of cerebrovascular illness+preoperative blood glucose+site of surgery+duration of surgery+intraoperative use of sufentanil+ intraoperative fluid balance+intraoperative positive blood transfusion

0.611

0.708

0.706

0.601

4

age+history of diazepam usage+history of cerebrovascular illness+ intraoperative positive fluid balance

0.556

0.754

0.703

0.634

No competing interests reported.

Supplementarytable.docx

Development and validation of a post-operative delirium prediction model for patients undergoing abdominal surgery: A retrospective, observational, single-center study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Trial registration

Figures

1 Background

2 Methods

2.1 Patients Characteristics

2.2 Data collection

2.3 Diagnosis of POD

2.4 Ethics Approval

2.5 Statistical methods

3 Results

4 Discussion

5 Conclusions

Abbreviations

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1