Related factors associated with the prognosis of children undergoing liver transplantation under the enhanced recovery after surgery nursing concept

Abstract

This study aimed to investigate factors associated with the clinical outcomes of patients who underwent pediatric liver transplantation (LT) and received enhanced recovery after surgery (ERAS) nursing. A cohort of 104 pediatric patients was studied at our hospital. Data on 8 indicators and 2 clinical outcomes, including length of hospital stay (LOS) and 30-day readmission rates, were collected. Linear and logistic regression analyses were employed to examine the associations of the 8 indicators with hospital-LOS and readmission risks, respectively. The predictive value of these indicators for the outcomes was determined using the receiver operating characteristic (ROC) curve, decision curve analysis, and importance ranking through the XGBoost method. A comprehensive model was developed to evaluate its predictive accuracy. Regression analyses identified donor age, donor gender, and intensive care unit (ICU)-LOS of recipients as significant predictors of hospital LOS (all P < .05), whereas no indicators were significantly associated with readmission risk. Further, ROC analysis revealed that 3 indicators provided superior prediction for 28-day hospital LOS compared to the median LOS of 18 days. ICU-LOS demonstrated the highest clinical net benefit for predicting 28-day hospital-LOS. Multivariable regression analysis confirmed the independent predictive value of donor age and ICU-LOS for the hospital-LOS (all β > 0, all P < .05). Although the comprehensive model incorporating donor age and ICU-LOS showed stable predictive capability for hospital-LOS, its performance did not significantly exceed that of the individual indicators. In pediatric LT, hospital LOS warrants greater emphasis over readmission rates. Donor age and ICU-LOS emerged as independent risk factors associated with prolonged hospital LOS.

1. Introduction

Pediatric liver transplantation (LT) is a vital treatment for a variety of acute and chronic end-stage liver diseases and genetic metabolic disorders.^[1] Its implementation across numerous transplant centers has yielded favorable outcomes. The perioperative management of pediatric LT, distinguished from adult procedures, integrates aspects of transplantation surgery, pediatrics, nursing, nutrition, and intensive care.^[2] The standardized perioperative approach involving these multidisciplinary fields is crucial for improving pediatric LT prognoses.^[3] Enhanced recovery after surgery (ERAS) encompasses a collection of evidence-based strategies to reduce physical and psychological traumatic stress responses in patients undergoing operation, thereby reducing postoperative complications, shortening hospital stays, and facilitating patient recuperation.^[4,5] This approach has succeeded in multiple surgical disciplines, including colorectal, upper gastrointestinal, and urogenital surgical procedures.

Currently, the adoption of ERAS in LT is emerging. Preliminary research indicates that ERAS protocols significantly improve perioperative safety and patient satisfaction, shorten postoperative hospitalization, and lower readmission rates.^[6,7] However, LT procedures are inherently complex, necessitating extensive perioperative nursing care, leading to extended hospital stays and readmission risks.^[8] With ERAS still under evaluation in this field, particularly in pediatric LT, this study examined the integration of ERAS within pediatric LT nursing practices and delved into the clinical characteristics involved. Although the favorable impacts of ERAS have been demonstrated in comparison to traditional care, the continuous evaluation and monitoring of clinical outcomes for patients under ERAS nursing are paramount. Moreover, this research sought to identify critical factors influencing the clinical outcomes of patients who have benefitted from ERAS nursing interventions.

2. Materials and methods

2.1. General information

A retrospective analysis was conducted on pediatric patients who underwent LT guided by ERAS principles at our hospital from August 2019 to August 2022. Based on a significance level of α = 0.05, a power (1-β) of 0.85, and an effect size of 0.7, the required sample size was determined to be 86. Considering the missing data, the sample size was increased by 20% based on the initial calculations. Consequently, 104 children, including 57 boys and 47 girls, were enrolled in our study.

The age distribution of the participants ranged from 4 months to 14 years, and they were categorized into 4 age groups: 43 children were under 1 year old, 33 children were 1 to 3 years old, 19 were in the 3 to 6 years category, and 9 children were > 6 years. The primary etiology for LT included biliary atresia (n = 86), cirrhosis (n = 6, with 5 cases of cryptogenic cirrhosis and 1 case of hepatitis B cirrhosis), and other varied causes (n = 12) such as propionic acidemia (two cases), Booga syndrome (one), biliary tract infection (one), hepatolenticular degeneration (two), hepatoblastoma (one), hepatic glycogen accumulation (one), acute liver failure (one), arginine succinic acid lyase deficiency (one), Langerhans cell hyperplasia (one), and ornithine amino formyltransferase deficiency (one). The Medical Ethics Committee considered this study observational, positing no additional risk to the participants, thereby waiving the requirement for informed consent from the children or their families. However, the data analysis conducted was approved by the Medical Ethics Committee.

2.2. ERAS protocol

All 104 children underwent the LT operation and received perioperative care adhering to the ERAS framework. As detailed below, the ERAS protocol was meticulously applied throughout the perioperative phase.

During the preoperative phase, the ERAS protocol encompassed preoperative education, nutritional management, psychological guidance, and guidelines on preoperative fasting. Specifically, preoperative education involved nurses providing health education to the children, their family members, and both, including preoperative assessment, surgical methods and outcomes, intraoperative and postoperative care, and complication prevention strategies. Nutritional management entailed conducting nutritional risk assessments within 24 hours of admission and evaluating nutritional issues based on the “ABCDE” framework (anthropometric, biochemical, clinical, dietary, environmental and family information). Enteral nutrition support was provided to malnourished children. Psychological guidance was offered in the surgery waiting area through toys, children books, and cartoons to mitigate fear, unfamiliarity, and distress. Nurses are encouraged to help ease the children anxiety. Preoperative fasting instructions specified avoiding solid food for 6 hours, breast milk for 4 hours, and clear fluids for 2 hours.

During the intraoperative phase, the ERAS protocol emphasized body temperature regulation and the prevention of pressure injuries. Body temperature management involves using a circulating water temperature-sensing blanket to maintain temperatures at 36 to 40˚C, thereby preventing intraoperative hypothermia. Blood transfusions and infusions were warmed using a fan heater set to 36 to 40˚C. The children body temperatures were dynamically monitored intraoperatively. The abdominal cavity was rinsed with a warm 0.9% sodium chloride solution. To prevent pressure injuries, a head ring stabilized the head while the arms were extended; the knee and hip joints were flexed correctly. Cylindrical cotton pads were strategically placed beside the trunk, outside the lower extremities, and between the legs. The cotton pads were positioned lower than the chest and abdominal area to avoid interference with the operation. Inflatable latex gloves alleviated foot pressure, and wide tape supported the overall positioning. Nurses checked for limb compression every 30 minutes intraoperatively.

In the postoperative stage, the ERAS protocol included respiratory care, nutritional support, sedative analgesic nursing, skin nursing, and early extubation. Respiratory care included vigilant airway management during endotracheal intubation, using strong adhesive tape for secure tube fixation. Nutritional support commenced within the 24 to 48 hour postoperative period, with appropriate selection of pacifiers for oral feeding and gentle back-tapping post-feeding to prevent vomiting. For tube feeding, the children heads were elevated by 30° to 45° to minimize the risk of vomiting and reflux. An enteral nutrition pump was used for those with gastroesophageal reflux or feeding intolerance. Sedative and analgesic care involved continuous monitoring of vital signs, consciousness level, pupil dilation, respiratory rate, and oxygen saturation. Utilizing suitable sedation and pain management strategies, methods such as tactile stimulation, kangaroo care, and music therapy were employed to alleviate pain and guarantee the postoperative safety of children. Intensive care unit (ICU) skincare included daily chlorhexidine wipes, keeping the buttocks area clean and dry, using soft cotton cloths or special wipes, and applying moisturizing oil to protect the skin. Early extubation focused on meticulously monitoring the volume and characteristics of surgical drainage fluid. Once potential complications such as bleeding and biliary fistula were excluded, the drainage tube was removed from the operative area if the drainage volume fell <10 mL·kg⁻¹·d⁻¹. Similarly, the catheter was extracted after clamping if the drainage volume was <5 mL·kg⁻¹·d⁻¹.

2.3. Data collection

Data were extracted from electronic medical records, encompassing variables such as gender, age, donor gender, donor age, primary etiology, operation time, blood loss volume, anhepatic phase, length of the stay in the ICU, total hospital-length of hospital stay (LOS), and records of 30-day readmission post-discharge.

2.4. Statistical analysis

Statistical analyses were performed using SPSS version 25.0 and R software. Quantitative data fitting a normal distribution were expressed as means and standard deviations, and their differences across multiple groups were evaluated using ANOVA. The medians (interquartile ranges) were used for non-normally distributed data, and non-parametric tests for analysis. Categorical variables were presented as frequencies, and differences among groups were assessed using the χ² test. The association of clinical variables with the LOS and 30-day readmission risk were analyzed using univariable linear regression and logistic regression analyses, respectively. The efficacy of significant variables in predicting clinical outcomes was determined through the receiver operating characteristic (ROC) analysis, and their clinical net benefit was analyzed using decision curve analysis (DCA). The XGBoost method prioritized these variables based on their predictive value for clinical outcomes. Further, the independent association of related variables with clinical outcomes was assessed through multivariable regression analysis. Subsequently, a comprehensive model incorporating these independent variables was established. Before establishment, patients were divided into training and validation sets in a 6:4 ratio. The model, initially created using the training set, was evaluated for predictive accuracy through ROC analysis. Its performance was later confirmed in the validation set to ensure its reliability. Statistical significance was set at P < .05.

3. Results

3.1. Clinical profile

Among the 104 pediatric patients, surgery duration varied from 320 to 590 minutes, averaging 409.24 ± 54.78 minutes. Blood loss ranged from 40 to 500 mL, with a median of 125 (100–175) mL. The anhepatic phase duration ranged from 20 to 70 minutes, with a median of 30 (25–40) minutes. ICU stays varied from 38 to 2417 hours, with a median duration of 67 (63–88.75) hours. Postoperative hospital stays ranged from 10 to 190 days, with a median of 18 (14–23.75) days. Within 30 days after discharge, 23 patients (22.11%) were readmitted.

3.2. Clinical characteristics of children

An analysis was conducted to identify the differences in clinical characteristics across various age groups (Table 1). The incidence of biliary atresia as the primary cause was significantly lower in children older than > 6 years compared to younger age groups (P < .001). The intraoperative blood loss was observed to be greater in children between 3 and 6 years and those > 6 years, in contrast to those under 1 year and those aged 1 to 3 years (P = .002). The length of ICU-LOS was significantly longer for the group older than 6 years compared with other age groups (P = .021). Moreover, the frequency of no readmission within 30 days post-discharge was higher in the 3 to 6 years age group than in other groups (P = .029).

Table 1.

Clinical characteristics of children of different ages.

Clinical characteristics	<1 year old group (n = 43)	1~3 yr old group (n = 33)	>3~6 yr old group (n = 19)	>6 yr old group (n = 9)	P value
Sex					.456
Male	21 (48.84)	18 (54.55)	11 (57.89)	7 (77.78)
Female	22 (51.16)	15 (45.45)	8 (42.11)	2 (22.22)
Donor sex					.771
Male	26 (60.47)	16 (48.48)	10 (52.63)	5 (55.56)
Female	17 (39.53)	17 (51.52)	9 (47.37)	4 (44.44)
Donor age (yr)	35.00 ± 8.53	31.97 ± 11.10	39.21 ± 10.80	39.67 ± 6.78	.037
Primary etiology					<.001
Biliary atresia	42 (97.67)	27 (81.82)	14 (73.68)	3 (33.33)
Cirrhosis	1 (2.33)	1 (3.03)	2 (10.53)	2 (22.22)
Other	0 (0.00)	5 (15.15)	3 (15.79)	4 (44.44)
Operation time (min)	412.35 ± 52.44	403.18 ± 54.85	412.89 ± 59.94	408.89 ± 61.94	.893
Blood loss (mL)	100 (80–150)	100 (80–150)	180 (120–200)	200 (125–245)	.002
Anhepatic phase (min)	30 (25–35)	30 (27.5–40)	30 (20–45)	30 (22.5–35)	.641
ICU stay time (h)	67 (64–90)	70 (63.5–100)	63 (55–67)	81 (62–88.5)	.021
Postoperative hospital stay (d)	20 (16–30)	17 (14.5–22.5)	16 (13–20)	16 (13–48)	.224
30-day readmission after discharge					.029
Yes	12 (27.91)	8 (24.24)	0 (0.00)	3 (33.33)
No	31 (72.09)	25 (75.76)	19 (100.00)	6 (66.67)

ICU = intensive care unit.

3.3. Significant variables associated with the clinical outcomes

Subsequently, the study examined variables potentially impacting clinical outcomes, namely postoperative hospital stay length and the risk of 30-day readmission (Table 2). Linear regression analysis revealed significant associations of donor age (β = 0.706, P = .008), ICU stay length (β = 0.047, P < .001), and donor gender (β = −11.726, P = .029) with postoperative hospital stay length. However, no variable significantly predicted the risk of 30-day readmission post-discharge (all P > .05), suggesting a greater relevance of hospital stay length over 30-day readmission risk. Therefore, hospital stay length was prioritized as the primary dependent variable for subsequent analyses.

Table 2.

The association analyses between variables and clinical outcome of patients.

	Hospital-LOS [β (95%CI)]	P	30-day readmission [OR (95%CI)]	P
Donor age	0.706 [0.186, 1.226]	.008	0.978 [0.933, 1.025]	.347
Anhepatic phase	0.160 [−0.377, 0.698]	.559	1.006 [0.961, 1.054]	.785
Blood loss	0.054 [−0.021, 0.129]	.161	1.005 [0.999, 1.011]	.113
Operation time	0.075 [−0.022, 0.172]	.132	0.997 [0.988, 1.006]	.456
Length of ICU stay	0.047 [0.026, 0.069]	<.001	1.003 [0.996, 1.010]	.435
Donor gender	−11.726 [−22.234, −1.218]	.029	1.434 [0.566, 3.629]	.447
Primary etiology
Cirrhosis	13.899 [−9.079, 36.878]	.236	0.756 [0.083, 6.881]	.804
Other	−2.434 [−19.204, 14.335]	.776	1.889 [0.511, 6.985]	.341
Patient age
1~3 yr old	−4.394 [−16.983, 8.195]	.494	0.827 [0.293, 2.335]	.719
>3~6 yr old	−8.316 [−23.301, 6.669]	.277	0.618 [0.023, 1.098]	.999
>6 yr old	8.111 [−11.829, 28.051]	.425	1.292 [0.278, 6.012]	.744
Patient gender	−0.429 [−11.180, 10.322]	.938	0.727 [0.283, 1.870]	.509

The biliary atresia was set as a reference group regarding Primary etiology. The <1-yr-old group was set as a reference group regarding the Patient age. The male patient was set as the reference group regarding the Donor gender and the Patient gender. Linear regression analysis was used to explore the association between variables and postoperative length of hospital stay (LOS). Logistic regression analysis was used to explore the association between variables and 30-day readmission risk after discharge.

ICU = intensive care unit, LOS = length of hospital stay.

The predictive capabilities of significant variables (donor age, ICU stay length, donor gender) for hospital stay length were further evaluated. Given the median LOS of 18 days, their predictive value for this duration was initially examined. Donor gender exhibited a higher area under the curve (AUC) value (AUC = 0.604) and clinical net benefit in predicting an 18-day LOS (Fig. 1A). Considering 28-day LOS as a standard clinical outcome, their efficacy in predicting this outcome was assessed, revealing the length of ICU stay as having the highest predictive value and clinical net benefit for a 28-day LOS (Fig. 1B). Donor age, ICU stay length, and donor gender were more effective in predicting a 28-day LOS than the median 18-day LOS. These results further highlighted the significance of these variables in predicting longer LOS.

Figure 1.

Evaluation of the clinical significance of variables influencing length of hospital stay (LOS). (A) The predictive accuracy and clinical net benefit of 3 variables in predicting an 18-day LOS using ROC (left) and DCA. (B) The prediction performance and clinical net benefit of 3 variables for a 28-d LOS employing ROC analysis (left) and DCA. (C) The variable importance ranking in predicting an 18-d or 28-d LOS utilizing the XGBoost method. ROC = receiver operating characteristic.

Furthermore, the importance of these 3 variables in predicting an 18-day or 28-day LOS was determined using the XGBoost method (Fig. 1C). The importance ranking was consistent for both durations, with the length of ICU stay being the most critical predictor.

3.4. Establishing a comprehensive model for predicting LOS

Our analyses underscored the significant roles of donor age, length of ICU stay, and donor gender in predicting the LOS. A multivariable linear regression analysis incorporating these variables revealed that (Table 3) donor age (β = 0.563, P = .027) and ICU stay length (β = 0.045, P < .001) independently predicted LOS. In contrast, donor gender did not show a significant independent association with LOS.

Table 3.

The multivariable linear regression analysis regarding LOS as a clinical outcome.

Name	β	Lower 95%	Upper 95%	P
(Intercept)	4.394	−15.208	23.996	.661
Donor age	0.563	0.071	1.055	.027
Length of ICU stay	0.045	0.024	0.065	<.001
Donor gender	−7.124	−16.984	2.737	.160

ICU = intensive care unit, LOS = length of hospital stay.

Subsequently, a comprehensive model was developed based on donor age and ICU stay length to predict a 28-day LOS. Before model creation, patients were divided into training and validation sets in a 6:4 ratio. The comprehensive model, incorporating donor age and ICU stay length, was developed using logistic regression analysis with the training set, and its predictive accuracy was subsequently evaluated. This evaluation was then confirmed with the validation set. The findings (Table 4) reveal that the model achieved an AUC of 0.663 for predicting a 28-day LOS in the training set, closely matching that of the validation set AUC of 0.647. This observation reflects the model consistent predictive reliability for a 28-day LOS. However, the comprehensive model predictive capability did not significantly surpass that of using donor age (AUC = 0.631) or ICU stay length alone (AUC = 0.680).

Table 4.

The performance analysis of the comprehensive model based on Donor age and Length of ICU stay for predicting the 28-d LOS.

	Training set (60%)	Validation set (40%)
AUC (SD)	0.663 (0.067)	0.647 (0.036)
Cutoff (SD)	0.207 (0.031)	0.207 (0.031)
Accuracy (SD)	0.670 (0.011)	0.648 (0.034)
Sensitivity (SD)	0.636 (0.091)	0.727 (0.091)
Specificity (SD)	0.712 (0.015)	0.591 (0.106)
Positive predictive value (SD)	0.384 (0.027)	0.373 (0.039)
Negative predictive value (SD)	0.826 (0.026)	0.830 (0.022)

ICU = intensive care unit, LOS = length of hospital stay.

4. Discussion

Advancements in surgical techniques and anesthesia have progressively enhanced LT outcomes. However, patient outcomes post-LT remains a concern. Wilson GC et al reported a 30-day post-LT readmission rate of 37.9%, with variations across hospitals ranging from 26.3% to 50.8%.^[9] The focus on perioperative management has intensified, with ERAS being recognized as a practical approach to improving LT perioperative management.^[10] Tinguely P et al reported that ERAS implementation in LT settings reduced overall complications, decreased ICU LOS, and enhanced short-term patient prognoses.^[11] In this study, all patients underwent ERAS protocols during the perioperative period, resulting in a 30-day post-discharge readmission rate of 22.11% for patients with LT. The readmission rate of patients is comparatively lower than previously reported figures, potentially attributable to the implementation of ERAS. Furthermore, Sachar Y et al implemented a quality improvement initiative involving 5 Plan-Do-Study-Act cycles, which effectively reduced the hospital-LOS; the proportion of patients discharged within 10 days increased from 18.4% to 60%, and the median ICU-LOS decreased from 3.4 to 1.9 days.^[12] These outcomes underscore the significant potential and importance of applying ERAS protocols and management strategies to enhance patient clinical outcomes during the perioperative period.

The study commenced by explaining baseline differences across various age groups among patients receiving ERAS nursing care, revealing age-related variations in donor age, primary etiology, blood loss, ICU-LOS, and 30-day readmission risk. Consequently, factors such as ICU-LOS and 30-day readmission were found to be associated with the patient age. While Keeling et al observed a negative correlation between age and LT prognosis in children,^[13] Byun et al determined that the prognosis for children undergoing LT was consistent across different ages, with infancy presenting no greater risk than older childhood.^[14] Our findings indicated an increased readmission rate in the < 1-year-old group, possibly due to their more precarious health status. Moreover, those with age > 6 years exhibited a more extended ICU stay compared to other age groups. These insights suggest a need for enhanced ERAS-focused clinical nursing for these patient groups, with further optimization of ERAS protocols based on patient characteristics as a vital approach to improving prognosis.

Despite improvements in the prognosis of patients with LT under ERAS management, multiple factors influence clinical outcomes. Yoshiya et al found that an extended ICU-LOS post-LT correlated with a worse short-term prognosis.^[15] Similarly, Rana et al developed a scoring system predicting hospital-LOS after LT, highlighting ICU-LOS (odds ratio [OR]: 1.75, 95% confidence interval [CI]: 1.58–1.95) as a significant determinant.^[16] Our study corroborates that ICU-LOS is independently associated with hospital-LOS positively and demonstrates strong predictive capability for ICU-LOS. These findings emphasize the critical impact of prolonged ICU-LOS on extended hospital-LOS, advocating for intensified nursing care for ICU patients.

Moreover, our research confirmed the donor age independent association with hospital-LOS, highlighting its predictive value and importance for patient clinical outcomes. Our study further validates the significance of donor age on patient clinical outcomes. Houben et al discovered an increase in graft loss risk by 1.4% annually, with donor ages ranging from 18 to 49 years. Donor age had the most substantial impact in patients with hepatitis C-related cirrhosis and the least in patients with hepatocellular carcinoma.^[17] While older donor age is identified as a risk factor for graft failure, it does not significantly affect liver function. However, the incidence of various liver diseases escalates with advancing age.^[18] The disparity in age between donor and recipient is considerably linked to recipient prognosis, with a reduced long-term survival observed in recipients < 45 years facing a donor age group of ≥ 10 years, a contrast not observed in older recipients (age > 60 years).^[19] Furthermore, no significant variations were observed in post-transplant complications or survival rates among recipients from different donor obesity groups.^[20] Identifying additional factors affecting recipient clinical outcomes is crucial for further research.

The ERAS protocol encompasses numerous stages and demands cooperation across multiple disciplines, presenting challenges for its widespread adoption.^[21] However, its potential to enhance the prognosis of patients with LT cannot be ignored. This study contributions include investigating ERAS in pediatric LT nursing, providing valuable insights into factors influencing postoperative outcomes, and highlighting the importance of comprehensive perioperative management. However, limitations, including its retrospective nature, confinement to a single institution, and the absence of a comparative control group, constrain the ability to gauge ERAS effectiveness against conventional care or alternative strategies. The generalizability of our findings is limited, and results from multicenter studies would offer more robust evidence. Data from a single center limits the broader applicability of our findings.

5. Conclusion

This research focused on patients who underwent LT and received ERAS nursing care throughout the perioperative period. Our analysis identified significant age-related variations in several clinical indicators. The hospital-LOS is a more critical measure of clinical outcome than readmission rates in patients with LT. Donor age and ICU-LOS of recipients were independently positively associated with the hospital-LOS. These 2 factors demonstrated a favorable performance for predicting the hospital-LOS, particularly for the 28-day. The ICU-LOS was more effective than donor age in predicting hospital-LOS, providing a more significant clinical net benefit. Despite developing a comprehensive model integrating donor age and ICU-LOS, there is no substantial enhancement in predictive accuracy over individual variables. This outcome further emphasizes the significant impact of donor age and ICU-LOS on the hospital-LOS for LT recipients.

Author contributions

Conceptualization: Xin-Bin Zhou.

Data curation: Xin-Bin Zhou, Wei-Ming Qian.

Formal analysis: Xin-Bin Zhou, Qin Xu.

Investigation: Li Chen.

Methodology: Qin Xu.

Supervision: Wei-Ming Qian.

Writing – original draft: Xin-Bin Zhou, Qin Xu, Li Chen.

Writing – review & editing: Wei-Ming Qian.