Risk factors for the delayed discharge from anesthesia intensive care unit: a single-center retrospective study

Abstract

Background

A single-center retrospective study was designed to investigate the risk factors associated with delayed discharge from the Anesthesia Intensive Care Unit (AICU).

Methods

This retrospective study involved patients admitted in the AICU from January 2017 to December 2022. Risk factors for the delayed discharge from the AICU were analyzed by the binary multivariate logistic regression analysis. Nomogram was constructed to predict the risk of delayed discharge from AICU. The performance of the nomogram was assessed using the receiver operating characteristic curve and calibration curve. A decision curve analysis was also performed to determine the net benefit threshold of prediction.

Results

A total of 14,338 patients admitted in the AICU were retrospectively recruited, involving 9,271 males and 5,067 females. The incidence of delayed discharge from the AICU in the cohort was 1.54% (221/14,338). Binary multivariate logistic regression analysis showed that younger than 18 years or older than 64 years, the American Society of Anesthesiologists physical status of class III-IV, body mass index of less than 18 kg/m² or greater than 25 kg/m², preoperative complications, emergency surgeries and intraoperative massive hemorrhage were risk factors for the delayed discharge from an AICU. We utilized nomograms to visually express data analysis results. Based on receiver operating characteristic analysis, calibration plots, and decision curve analysis, we concluded that the nomogram model exhibited excellent performance. Patients undergoing spine surgeries suffered from the highest proportion of delayed discharge from the AICU, followed by those receiving orthopedic and vascular surgeries. Postoperative hemorrhage was the major cause of delayed discharge from an AICU, followed by septic shock, hypoperfusion and pulmonary insufficiency.

Conclusion

The incidence of delayed discharge from the AICU in a single-center tertiary hospital is 1.54%. It is influenced by various risk factors, including age, ASA physical status classification, BMI, preoperative complications, type of surgery and intraoperative blood loss. The nomogram model exhibits excellent performance.

Trial registration

The single-center retrospective study was approved by the Ethics Committee of Nanjing Drum Tower Hospital (No. 2021-563-01, Data: 22 November 2021) and registered on the Chinese Clinical Trial Registry (No. ChiCTR2300078251, Data: 01 December 2023).

Introduction

Managed by the Department of Anesthesiology, the anesthesia intensive care unit (AICU) provides healthcare services and monitoring for postoperative patients who are elderly or exhibit severe perioperative complications [1, 2]. The challenges of postoperative resuscitation and rescue have significantly increased for anesthesiologists, primarily due to the aging population and and the evolving spectrum of diseases. The AICU represents a distinct environment staffed by anesthesiologists, serving as a vital supplement to the intensive care treatment. It provides professional medical care for patients who are not suitable for being admitted in the postoperative anesthesia care unit (PACU) or intensive care unit (ICU).

AICU and PACU are spatially connected to the operation room, forming a well-organized frame for performing perioperative treatment, surgery and anesthesia. The role of anesthesiologists has been greatly highlighted in the AICU. The whole-process involvement of anesthesiologists in the AICU favors the rapid postoperative recovery and perioperative safety [3,4,5,6]. The theory of enhanced recovery after surgery (ERAS) further accelerates the development of AICU. Evidence-based perioperative measures largely reduce surgery stress and inflammatory response, as well as enhance the comfort of patients admitted in the AICU [7, 8]. Delays in discharge from the AICU are self-reinforcing, leading to a high risk of negative outcomes, increased medical cost and resource shortage. We performed a single-center retrospective study, aiming to explore risk factors for the delayed discharge from an AICU.

Methods

Data sources

It was a retrospective study collecting involving patients admitted in the Anesthesia Intensive Care Unit (AICU) from January 2017 to December 2022. The AICU was managed by the Department of Anesthesiology, Nanjing Drum Tower Hospital, the Affiliated Hospital of Nanjing University Medical School, providing medical services and monitoring for elderly or postoperative patients with severe perioperative complications. Clinical data and anesthesia-associated data were available from the Hospital Information System (Zesing Software, China) and the Anesthesia Information Management System (Medical System, China), respectively. The single-center retrospective study was approved by the Ethics Committee of Nanjing Drum Tower Hospital (No. 2021-563-01) and registered on the Chinese Clinical Trial Registry (ChiCTR2300078251).

Subjects

Patients remained in an AICU for greater than 24 h were considered to have a delayed discharge from AICU. All patients admitted to AICU during the study period were included in this study, and no patients were excluded. A panel of data were recorded for analyses, including the ASA physical status classification, age, sex, body mass index (BMI), preoperative underlying diseases, duration of surgery, type of surgery, methods of surgical procedures, intraoperative blood loss and outcomes. Preoperative complications include hypertension, diabetes, coronary heart disease, renal insufficiency, etc. In particular, surgical procedures were classified based on the detailed surgical specialties like general surgery, ophthalmic surgery, gastrointestinal surgery, orthopaedic surgery, neurological surgery, etc.

Management in AICU

All patients admitted to the AICU were transferred after elective or emergency surgeries, or suffered from a postoperative serious aggravation in the ward that required the anesthesia intensive care. A continuous monitoring of heart rate (HR), oxygen saturation (SpO2), blood pressure (BP), electrocardiography (ECG), body temperature and invasive arterial pressure using the Philips patient monitoring system. Patients requiring an artificial airway were managed with tracheal intubation for mechanical ventilation with the following parameters: tidal volume (TV), 6–8 ml/kg; respiratory rate (RR), 12 beats/min; inspiratory to expiratory ratio (I: E), 1:2; and end-tidal carbon dioxide partial pressure (PetCO₂), 35–45 mmHg. Low-dose fentanyl and propofol plus dexmedetomidine were given to mechanically ventilated patients for analgesia and sedation, respectively. Patients with spontaneous breathing were given an oxygen therapy using a face mask at the flow rate of 5 L/min. Low-dose fentanyl and dexmedetomidine were given for analgesia and sedation, respectively.

Statistical analysis

R software (version 4.2.2) and MSTATA software (www.mstata.com) were used to analyze the data in this research. Continuous variables with a normal distribution confirmed by the Shapiro-Wilk test were expressed as mean ± standard deviation (SD) and compared using the independent samples t-test; otherwise, they were expressed as median (interquartile range) and compared by the Mann-Whitney U test. Enumeration data were presented as absolute number or percentage, and compared by the Chi-square test or Fisher’s exact test. Variables with a significant difference of P < 0.1 in the univariate logistic regression model were introduced into the binary multivariate logistic regression model, thereafter identifying the risk factors for delayed discharge from an AICU. Incorporate the screened AICU risk factors into the Nomogram Model. All patients were randomly divided into training and internal validation cohorts according to the proportion of 7:3. In the training cohort, the least absolute shrinkage and selection operator (LASSO) logistic regression analysis was used for multivariate analysis to screen the independent risk factors and build a prediction nomogram for delayed discharge from AICU. Screening risk factors through logistic regression and LASSO regression to reduce confounding interference. The performance of the nomogram was assessed using the receiver operating characteristic (ROC) curve and calibration curve, with the area under the ROC curve (AUC) ranging from 0.5 (no discriminant) to 1 (complete discriminant). A decision curve analysis (DCA) was also performed to determine the net benefit threshold of prediction. Results with P < 0.05 were considered significant.

Results

Baseline characteristics

A total of 14,338 patients admitted in the AICU were retrospectively recruited, involving 9,271 males and 5,067 females. The incidence of delayed discharge from the AICU in the cohort was 1.54% (221/14,338). Specifically, patients at a younger or advanced age (< 18 years or > 64 years) and those with worse ASA physical status classification, much more preoperative complications, BMI outside of the normal range (< 18.5 kg/m² or > 25 kg/m²), management of an emergency surgery and greater intraoperative blood loss were more likely to suffer from a delayed discharge from the AICU (P < 0.05, Table 1). Moreover, worse perioperative prognosis and higher mortality were observed in patients with delayed discharge from an AICU.

Table 1 Baseline characteristics of patients admitted in AICU (n = 14,338)

Risk factors for delayed discharge from AICU

We conducted the univariable logistic analysis on all potential factors, included those with a P < 0.05 into binary multivariate logistic proportional hazard regression analysis to determine independent risk factors. Based on the results of this study, the risk factors associated with delayed discharge from AICU can be classified according to patient characteristics and surgical characteristics (Table 2).

Age of younger than 18 years (aOR 9.34, 95% CI 5.24–16.01) or older than 64 years (aOR 1.35, 95% CI 1.01–1.81), ASA physical status of class III (aOR 1.21, 95% CI 0.68–2.37), class IV (aOR 9.83, 95% CI 5.08–20.45) or class V (aOR 8.38, 95% CI1.20-35.66), BMI of less than 18 kg/m² (aOR 2.67, 95% CI 1.73–4.02) or greater than 25 kg/m² (aOR 1.05, 95% CI 0.75–1.45) and preoperative complications (aOR 5.43, 95% CI 3.74–8.04) were patient-related risk factors for delayed discharge from an AICU. In addition, emergency surgeries (aOR 1.83, 95% CI 1.32–2.51) and intraoperative massive hemorrhage (1000–2000 ml: aOR 3.90, 95% CI 2.51–5.86; 2000–3000 ml: aOR 13.12, 95% CI 8.32–20.17; >3000 ml: aOR 6.55, 95% CI 3.06–12.59) were surgery-related risk factors for delayed discharge from AICU.

Table 2 Multivariable logistic regression of risk factor for delayed discharge from AICU

Nomogram model establishment and validation

The baseline demographic and clinical characteristics of the study cohorts were summarized in Table 3, with a training set consisting of 10,037 patients and a testing set consisting of 4301 patients. The comprehensive analysis of baseline characteristics suggests that the training and internal test cohorts are well-balanced and suitable for predictive modeling in the study of interest. The candidate predictors were included in the original model, which were using LASSO regression analysis performed in the training cohort (Fig. 1). The final logistic model included 6 independent predictors (Age, ASA physical status, BMI, complications, type of surgery, and intraoperative blood loss) and was developed as a simple-to-use nomogram (Table 4; Fig. 2). ROC analysis revealed that AUC value of the nomogram was greater than 0.8 in both sets, indicating that this model has excellent discriminant ability (Fig. 3A, B). Calibration curves of the nomogram demonstrated a high agreement between predicted and actual probabilities in both the development and validation sets (Fig. 3C, D). In addition, DCA showed that the nomogram model is effective in clinical practice (Fig. 3E, F).

Table 3 Patient demographics and baseline characteristics of training and teat cohort

Table 4 Results of Multivariate Logistic regression for training cohort

Fig. 1

LASSO regression analysis for factors selection. (A) selection of the tuning parameter (λ) by using 10-fold via minimum and 1-SE criteria. (B) Non-zero coefficients selection by using the 10-fold cross-validation

Fig. 2

Nomogram to estimate the risk of delayed discharge patients from AICU. ASA, American Society of Anesthesiologists; BMI, body mass index; N.A., not applicable; I.A., applicable

Fig. 3

ROC curves, calibration plots and DCA of the nomogram. (A, B) The area under ROC curve of training cohort (A) and internal test cohort (B). (C, D) Calibration plot for diagnostic nomogram in training cohort (C) and internal test cohort (D). (E, F) Decision curve analysis for diagnostic nomogram in training cohort (E) and internal test cohort (F)

Distribution of surgical specialties and causes of delayed discharge from AICU

Classified by surgical specialties, patients receiving spine surgeries (7.74%) suffered from the highest proportion of delayed discharge from the AICU, followed by those treated with orthopedic (2.58%) and vascular surgeries (2.37%). Urology and neurological surgeries were the least surgical specialties leading to a delayed discharge from the AICU (Table 5).

Postoperative hemorrhage (22.17%) was the major cause of delayed discharge from AICU, followed by septic shock (18.55%), hypocirculation (18.10%) and pulmonary insufficiency (17.19%). Other causes of delayed discharge from AICU included brain dysfunction (e.g., delirium), heart failure, airway problems and severe allergies (Fig. 4).

Table 5 Patients with delayed discharge from AICU classified by surgical specialties

Fig. 4

Distribution of causes of delayed discharge from an AICU

Discussion

The Anesthesia Intensive Care Unit (AICU), managed by anesthesiologists, provides intensive care, monitoring and treatment for critically ill patients after surgery until they are recovered and transferred to a normal ward. Some of patients, unfortunately, experience a delayed discharge from the AICU for more than 24 h due to various causes, leading to the increase in medical cost and patient’s dissatisfaction. This single-center retrospective study investigated the incidence of delayed discharge from the AICU in a tertiary hospital and analyzed its risk factors. It is found that a younger or older age, higher ASA physical status classification, BMI outside of the normal range, preoperative complications, emergency surgeries and intraoperative massive hemorrhage were significantly correlated with delayed discharge from AICU.

We consistently observed a large proportion of patients at the age of 65 years and above in the AICU (43.09%), accounting for 47.15% of cases remaining in AICU for greater than 24 h [9]. The number of elderly surgical patients is annually on the rise, consisting of a high-risk group with preoperative frailty, perioperative complications and impaired organ functions. Cao et al. [10] suggested that the advanced age is mainly responsible for delayed postoperative recovery and transfer to the ward. In the present study, children and teenagers younger than 18 years were also the high-risk group of delayed discharge from AICU. A retrospective analysis further revealed that most of them were scoliosis patients, accompanying tissue and organ dysplasia at varying degrees. The prolonged duration of scoliosis surgery, coupled with significant intraoperative blood loss, substantially increased the perioperative risk, ultimately resulting in delayed discharge from the AICU [11].

The ASA physical status classification is a vital indicator in preoperative assessments, and a poor physical status is a risk factor for prolongation of postoperative recovery and unexpected transfer to ICU [12, 13]. We consistently showed that ASA Class III-IV was significantly correlated with delayed discharge from AICU. BMI is also a determinant of delayed discharge from AICU due to the negative influence of obesity on sleep apnea, delayed extubation and perioperative hypoxemia, especially in patients treated with opioids. Most studies only used dichotomies for BMI. Therefore, low body weight had been ignored. Studies have found a significant correlation between low body weight (BMI < 21 kg/m2) and delayed resuscitation, in addition to morbid obesity [14]. Further analysis reveals that the majority of low body weight patients are scoliosis patients. These patients generally have some degree of developmental and nutritional problems. And their surgery time is long, the degree of surgery is complex, and there is a lot of intraoperative bleeding, which can lead to a delay in transferring out of AICU after surgery. The present study revealed that both BMI < 18.5 kg/m² and > 25 kg/m² were risk factors for delayed discharge from AICU.

Patient’s physical condition, together with surgery-associated factors both influence the perioperative rehabilitation of AICU patients. For surgery-associated factors, surgical specialties, type of surgery, duration of surgery and intraoperative conditions are closely linked with postoperative recovery [15,16,17]. The majority of emergency surgeries are performed in critically ill patients usually with respiratory and circulatory dysfunction. The severe illnesses largely increase the incidence of delayed discharge from AICU or even the possibility of being transferred to ICU for more intensive care. In addition, a linear correlation exists between intraoperative blood loss and incidence of delayed discharge from AICU. Massive hemorrhage causes a perioperative hypoperfusion. Moreover, blood product transfusion in turn imbalances the physiological homeostasis, prolongs the pharmacokinetics and pharmacodynamics of anesthetic drugs, and influences the nervous, muscular and functional recovery, eventually prolonging AICU stay [18]. However, we did not identify a significant correlation between the duration of surgery and delayed discharge from AICU, and it may be attributed to the treatment of critical illnesses in the tertiary hospital, types of surgical specialties and high-quality postoperative care.

Complex surgeries, such as spine surgeries (7.74%), orthopedic surgeries (2.58%) and vascular surgeries (2.37%), generally need longer duration of surgery and anesthesia maintenance, more complicated methods of anesthesia, greater influences on physiological functions and higher incidence of complications, resulting in a high incidence of delayed discharge from AICU [13]. Postoperative hemorrhage (22.17%), septic shock (18.55%), hypoperfusion (18.10%) and pulmonary insufficiency (17.19%) were the most common causes of delayed discharge from AICU. The above patients usually experienced a longer admission in intensive care units because of the extensive damages to important organs and systems [19, 20]. Most postoperative recovery events are related to the cardiopulmonary system, and hypoperfusion, hypoxia and infections serve as huge challenges in perioperative management [21]. A large-scale retrospective study reported that the incidence of postoperative cardiovascular and respiratory complications in patients suffering from a delayed transfer to the ward is inflated to 40.33% [22].

A nomogram for predicting the risk of delayed discharge in AICU patients was developed and validated with excellent discrimination and calibration in the current study. We visualized these data using nomograms, which is more conducive to clinicians’ judgment and targeted treatment. The result of nomogram showed that age of younger than 18 years, ASA physical status of class IV and V, and intraoperative massive hemorrhage of 2000–3000 ml were the highest scoring factors. We retrospectively analyzed these factors and found that most of them were concentrated in adolescent spinal correction patients, who were also the group with the highest proportion of AICU delayed discharge. Because such patients have multiple preoperative complications, complex surgeries, and rapid changes in their condition, comprehensive perioperative evaluation and comprehensive management were necessary. By combining with the nomogram, personalized prevention and intervention based on risk factors can be achieved, thereby improving the treatment rate of critically ill patients and providing them with comfortable perioperative care.

Outperformed in the research of delayed discharge from AICU, our study for the first time assessed its risk factors. In addition, we introduced the binary logistic regression model and established a nomogram model. Indeed, there are some limitations in our research. First of all, it was a single-center retrospective study that lacked an external validation. Second, nomogram is based on retrospective studies, requiring further validation in prospective cohort and clinical trials. In addition, this study only analyzed all patients who entered AICU. In fact, there are certain differences in postoperative management among patients undergoing different surgeries, which is also the content that needs further analysis and research in our subsequent experiments. In future research, we will use a multicenter prospective study model to further explore the treatment and prognosis of patients with different diseases in AICU, in order to further expand our findings.

Conclusion

The incidence of delayed discharge from the AICU in a single-center tertiary hospital is 1.54%, influenced by the risk factors including age, ASA physical status classification, BMI, preoperative complications, type of surgery and intraoperative blood loss. The nomogram model exhibits excellent performance and achieves personalized prevention and intervention based on risk factors, thereby improving the treatment rate of critically ill patients and providing them with comfortable perioperative care.