Risk Factors and Causes of Readmission to the Pediatric Intensive Care Unit After Cardiac Surgery

Introduction

Advances in critical care medicine have significantly improved the outcomes of critically ill pediatric patients. Survival rates have increased worldwide, and the prognosis for children admitted to the pediatric intensive care unit (PICU) has steadily improved. However, the transition from the PICU to pediatric wards remains a challenging phase of care. Patients discharged from the PICU often have more complex and chronic conditions, making them more vulnerable than regular ward patients. ¹ This vulnerability is compounded by the need to manage limited PICU capacity, which sometimes necessitates discharge decisions based on external factors rather than solely on clinical readiness for transfer.

Unplanned readmission to the PICU, defined as an unexpected return after discharge to a general ward, stepdown unit, secondary facility, or home,^2,3 is a critical quality care metric recognized by the Society of Critical Care Medicine. These readmissions are associated with several adverse outcomes, including increased PICU length of stay (LOS), overall hospital LOS, and higher mortality risk. Additionally, they create psychological and logistical burdens for families and exacerbate bed shortages in already strained PICUs. The impact of unplanned readmissions underscores the importance of understanding and mitigating their occurrence.

Despite the clinical significance of unplanned readmissions, data on the risk factors and characteristics of readmitted patients remain limited. Reported rates of PICU readmissions vary widely, from 1.6% to 60%, with associated mortality rates ranging from 2% to 21.3%, exceeding the overall PICU mortality rate of 3% to 11%.^4–6 Factors contributing to readmission are diverse, including chronic diseases such as asthma ⁷ or neuromuscular disorders, ⁸ as well as socioeconomic disparities that heighten risks for children from underserved neighborhoods. ⁹ However, much remains unknown about the predictors and outcomes of these events.

Among the subgroups of PICU patients at risk of unplanned readmission, children undergoing cardiac surgery stand out due to their particularly poor outcomes. ⁶ The majority of cardiac surgeries performed in children aim to correct or palliate congenital heart defects—the most common congenital anomaly globally—affecting 0.8% to 1.2% of live births. ¹⁰ Children, especially newborns, undergoing cardiac surgery often require complex perioperative and postoperative care initially in the PICU in order to prevent complications^11,12 before being transferred to a ward for the next phase of care. While PICU readmissions in this population are relatively rare, they are associated with disproportionately severe consequences, including high variability in mortality rates worldwide. ¹³ Preventing unplanned readmissions in these patients is especially challenging due to the complexity and heterogeneity of their conditions.

Given the heightened risk and poor outcomes associated with unplanned readmissions in children following cardiac surgery, it is crucial to identify strategies to reduce these events. In this study, we aim to describe the clinical characteristics and PICU care of readmitted patients in this population. By comparing readmitted and nonreadmitted patients at our center, we aim to identify risk factors that can inform targeted interventions and improve care for this vulnerable group.

Material and Methods

Results

Description of the Population

From January 2015 to January 2021, 956 patients underwent cardiac surgery at our institution and were admitted to the PICU for postoperative recovery and monitoring. After PICU discharge, 23 of 956 patients (2.4%) required unscheduled PICU readmission during the same hospitalization due to complications occurring at the ward. Each readmitted patient was matched to two controls that did not require PICU readmission, totaling 46 controls. Median age on the day of the surgery was 9.0 months (1.8-58.2 months) for the readmitted patients and 8.7 months (0.8-51.0 months) in the control group (P = .719). Demographic characteristics for each group are shown in Table 1.

OPEN IN VIEWER

Table 1.

Clinical Characteristic of the 23 Readmitted Patients in ICU After PICU Discharge and Their 46 Controls.

Demographic data
Male sex—n/N, % (95% CI)	37/69	53.6 (41.9-65.4)
Age (days)—n, median (IQR; range)	69/69	271 (34-1631)
Weight at admission (kg)—n, median (IQR)	62/69	8.1 (4.0-15.3)
Hospitalization in PICU before surgery—n/N, % (95% CI)	9/69	13.0 (5.1-21.0)
Emergency surgery—n/N, % (95% CI)	4/69	5.8 (0.3-11.3)
Medical history—n/N, % (95% CI)
Preterm birth	7/69	10.1 (3.0-17.3)
Genetic disorders	11/69	15.9 (7.3-24.6)
Neurologic or neuromuscular disease	10/69	14.5 (6.2-22.8)
Chronic pulmonary disease	13/69	18.8 (9.6-28.1)
Chronic gastrointestinal disease	13/69	18.8 (9.6-28.1)
Congenital heart disease—n/N, % (95% CI)
Cyanotic heart disease	39/69	56.5 (44.8-68.2)
Transposition of great arteries (TGA)	12/69	17.4 (8.5-26.3)
Tetralogy of Fallot (TOF)	12/69	17.4 (8.5-26.3)
Ventricular septal defect	9/69	13.0 (5.1-21.0)
Atrioventricular septal defect	8/69	11.6 (4.0-19.2)
Aortic arch malformations	7/69	10.1 (3.0-17.3)
Congenital valve disease	7/69	10.1 (3.0-17.3)
Miscellaneous	14/69	20.3 (10.8-29.8)
At PICU arrival—n/N, % (95% CI)
Clinical cardiac failurea	14/69	20.3 (10.8-29.8)
Altered ventricular functionb	6/49	12.2 (3.1-21.4)
Arrythmiac	7/69	10.1 (3.0-17.3)
Vasopressor infusiond	6/69	8.7 (2.1-15.3)
Mechanical ventilation	3/69	4.4 (0.0-9.2)
Other respiratory supporte	10/69	14.5 (6.2-22.8)

Abbreviations: BiPAP, bilevel positive airway pressure; CI, confidence interval; CPAP, continuous positive airway pressure; ICU, intensive care unit; IQR, interquartile range; LVEF, left ventricular ejection fraction; PICU, pediatric intensive care unit.

^a

Clinical cardiac failure is defined as association of acute pulmonary edema and low blood pressure.

^b

Echocardiographic cardiac failure as evaluated by expert cardiologist, as LVEF might not be measured according to type of congenital heart defect. Postoperative ventricular function was considered altered if it was described as “abnormal” or estimated to be less than 50% based on the findings of the postoperative echocardiogram.

^c

Atrial fibrillation, supraventricular tachycardia, ventricular tachycardia, ventricular fibrillation, and AV block.

^d

Norepinephrine, epinephrine, dobutamine, milrinone, and vasopressin.

^e

CPAP, BiPAP, and high-flow nasal canula.

Patients Readmitted to PICU

For the readmitted group, the median time to readmission was 42.9 h (16.2-91.7). Twelve of 23 patients (52.2%) were readmitted within 48 h of their discharge from the PICU, while the remaining patients returned between 48 h and 7 days of their initial PICU discharge; 39.1% (9/23) came back to the PICU between 3 pm and 11 pm, while 17.4% (4/23) were readmitted between 11 pm and 7 am. Table 2 shows the reasons for unscheduled PICU readmission. Two of 23 patients (8.7%) were readmitted after an unplanned return to the operating room. The most common causes for readmission were management of a pericardial or pleural effusion (26.1%, 6/23) and another cardiovascular diagnosis (ie, arrhythmia) (26.1%, 6/23). Of note, one patient (2.3%) experienced a cardiac arrest outside the PICU.

OPEN IN VIEWER

Table 2.

Comparison of Characteristics, Before, During, and After Heart Surgery of Children Readmitted to the PICU After Discharge Versus Children not readmitted.

	Readmitted	Not-readmitted	P value
Patients’ characteristics
Male sex—n/N, %	8/23, 35%	29/46, 63%	.026
Weight at ICU admission (kg)—n, median (IQR)	16, 7.5 (4.1-16.2)	46, 8.4 (4.0-16.2)	.85
Characteristics before surgery—n/N, %
Cardiac
Cyanotic heart disease	12/23, 52%	27/46, 59%	.61
Cardiac failurea	6/23, 26%	8/46, 17%	.53
Echocardiographic cardiac failureb	2/14, 4%	4/35, 11%	1.00
Arrythmiasc	5/23, 22%	2/46, 4%	.037
Need for vasopressor infusion	4/23, 17%	2/46, 4%	.090
Extracardiac
Need for mechanical ventilation	1/23, 4%	9/46, 20%	.15
Need for other respiratory support	2/23, 9%	1/46, 2%	.26
Surgery characteristics
Emergency surgery—n/N, %	1/23, 4%	3/46, 7%	1.00
Duration of extracorporeal support (min)—n, median (IQR)	22, 114 (94-136)	46, 122 (87-159)	.33
Open chest after surgery—n/N, %	2/23, 9%	6/46, 13%	.71
Postoperative single-ventricle physiology—n/N, %	8/23, 35%	5/46, 11%	.024
Cyanotic physiology postsurgery—n/N, %	6/23, 26%	5/46, 11%	.16
Severe valvular regurgitation—n/N, %	5/14, 36%	10/45, 22%	.32
Characteristics after surgery
Modified PELOD-2d—n, median (IQR)	23, 9 (8-10)	46, 9 (8-12)	.70
Length of stay in PICU (day)—n, median (IQR)	23, 5.3 (2.9-10.9)	46, 5.5 (3.1-9.9)	.99
Cardiac
Altered ventricular functione—n/N, %	6/20, 30%	4/46, 9%	.056
Duration of vasopressor infusion—n, median (IQR)	9, 2.8 (0.9-4.7)	14, 2.1 (1.1-3.6)	.73
Need for vasopressor reinitiation after initial wean-off—n/N, %	3/23, 13%	2/46, 4%	.32
Arrythmias—n/N, %	5/23, 22%	12/46, 26%	.69
Extracardiac
Duration of MV after surgery (days)—n, median (IQR)	10, 0.8 (0.3-1.0)	17, 1.1 (0.2-4.0)	1.00
Need for postoperative reintubation—n/N, %	2/23, 9%	5/46, 11%	1.00
Need for postoperative noninvasive ventilation—n/N, %	15/23, 65%	30/46, 65%	1.00
Drainage of PTX/PE—n/N, %	1/23, 4%	8/46, 17%	.25
Need for RBC transfusion (mL/kg)—n, median (IQR)	8, 14.6 (10.8-20.2)	8, 17.6 (12.9-33.2)	.45
Sepsis during ICU hospitalization—n/N, %	5/23, 22%	4/46, 9%	.15
Characteristics at ICU discharge—n/N, %
Severe valvular regurgitation	7/19, 37%	10/45, 22%	.23
Known pleural effusion	5/23, 22%	2/46, 4%	.037
Chest tube in situ	1/23, 4%	0/46, 0%	.33
Internal pacemaker	2/23, 9%	3/46, 7%	1.00
Known diaphragmatic paralysis	1/23, 4%	1/46, 2%	1.00
Enteral nutrition by nasogastric tube	4/15, 27%	9/35, 26%	1.00
Treatment by sildenafil	4/23, 17%	6/46, 13%	.72

Abbreviations: ICU, intensive care unit; IQR, interquartile range; LVEF, left ventricular ejection fraction; MV, mechanical ventilation; PE, pleural effusion; PICU, pediatric intensive care unit; PTX, pneumothorax; RBC, red blood cell.

^a

Clinical cardiac failure is defined as association of acute pulmonary edema and low blood pressure.

^b

Echocardiographic cardiac failure as evaluated by expert cardiologist, as LVEF might not be measured according to type of congenital heart defect

^c

Atrial fibrillation, supraventricular tachycardia, ventricular tachycardia, ventricular fibrillation, and AV block.

^d

Modified PELOD2 was calculated according to PELOD2 without considering neurologic scoring, because consciousness is not evaluable in case of sedative medication following surgery.

^e

Postoperative ventricular function was considered altered if it was described as “abnormal” or estimated to be less than 50% based on the findings of the postoperative echocardiogram.

Risk Factors of Unplanned Readmission to PICU

Table 2 compares patients based on readmission status. After univariate regression preoperative arrhythmia, the need for pressors before the index surgery, postoperative altered ventricular function (assessed based on postoperative echocardiogram), cardiac physiology after procedure (univentricular vs biventricular physiology) and known pleural effusion at PICU discharge were all considered potential risk factors associated with unplanned PICU readmission, whereas male gender was a protective risk factor. However, after linear regression analysis, only postoperative cardiac physiology remained an independent risk factor (Figure 1). Biventricular physiology appears to be a protective factor against unplanned readmission. In our patients, 13 patients have univentricular physiology after surgery. Six were operated because of hypoplastic left heart syndrome, 4 patients have hypoplastic right heart syndrome, and 3 patients have a severe form of tetralogy of Fallot with double outlet right ventricle.

OPEN IN VIEWER

Figure 1.

Risk factors for readmission after PICU discharge in pediatric patients who underwent cardiac surgery.

Discussion

To our knowledge, this is one of the few studies that focus exclusively on pediatric patients who underwent cardiac surgery and their risk factors for readmission to the PICU. Previous studies in this area have largely examined a broader cohort of patients, combining both medical and surgical populations.^6,14,15 This approach, while informative, may obscure the unique challenges and pathways associated with postcardiac surgery patients. The physiological disruptions caused by cardiac surgery in children are profound, especially when the surgery aims to correct or palliate complex congenital heart defects, leading to significant differences in recovery trajectories and, consequently, in the causes and risk factors for readmission. By narrowing our focus to this specific population, we aim to provide a more targeted understanding of these risk factors. Additionally, this study is one of the few to specifically investigate readmissions to the PICU during the index hospitalization. Complications occurring shortly after transfer from the PICU to the ward likely differ substantially from those occurring later, such as after the patient has been discharged home.^16–18 These early complications, which arise during the same hospitalization, warrant distinct attention as they may reflect unique vulnerabilities in the immediate postoperative period or in the transition of care from the PICU to the ward.

In our cohort, the unplanned PICU readmission rate following cardiac surgery was 2.4% (23/956), with 52.2% (12/23) of these readmissions occurring within 48 h of discharge. This rate falls at the lower end of the range reported in the literature, where unplanned readmission rates for pediatric patients initially admitted to the PICU for a cardiac-related condition generally range from 2.4% to 15.9%.^6,18 The primary reasons for readmission in our cohort were cardiovascular complications, such as pleural effusion and arrhythmias, which are consistent with findings from similar studies.^14,15 These results highlight the vulnerable postoperative state of children with congenital heart disease, where even seemingly stable patients may rapidly deteriorate after transitioning from intensive care to a general ward. Furthermore, our findings emphasize the importance of close monitoring and effective handoff protocols during this critical transition period. Additionally, the high proportion of readmissions occurring within 48 h suggests that early complications often stem from conditions already present at PICU discharge. These children are likely at higher risk, and their vulnerability should be identified at the time of PICU discharge to ensure more tailored postoperative care plans, including enhanced monitoring for high-risk patients.

We identified postoperative univentricular physiology as an independent risk factor of unplanned PICU readmission. This aligns with prior research underscoring the complexity and fragility of univentricular physiology in postsurgical care. ¹⁰ Univentricular hearts pose significant challenges due to compromised cardiac output, increased arrhythmia risk, and multiorgan dysfunction. ¹² The complications observed in this cohort, such as pleural effusions and arrhythmias, reflect the inherent instability of patients with univentricular physiology, predisposing them to sudden deterioration. Structural and electrical abnormalities in these patients likely contribute to their heightened risk of rhythm disturbances and subsequent readmissions. ¹⁰ Future studies exploring the mechanisms linking univentricular physiology to postoperative complications could help refine management strategies and reduce readmission risk by tailoring interventions for this high-risk group.

The role of genetic conditions, the complexity of the surgery performed, and PICU LOS deserve consideration. These factors have been thought to impact unplanned PICU readmission in cardiac patients, despite not being extensively studied in exclusively surgical cohorts.^14,16,17 Children with Down syndrome are predisposed to congenital heart defects but may also face additional challenges during recovery, including respiratory insufficiency, impaired immune responses, and reduced physiological reserve.^7,8 The complexity of cardiac surgery can vary widely in children, and scoring systems have been used to classify procedures based on mortality risk. ¹⁹ Both genetic anomalies and very complex surgeries were underrepresented in our cohort, limiting the conclusions that can be drawn regarding their relationship to unplanned PICU readmission. While PICU LOS is typically an indicator of the child's condition and is associated with the need for critical care therapy—and can therefore reflect patient vulnerability—a longer PICU stay may also indicate a more cautious and conservative approach prior to discharge to the ward. Further investigation is warranted to better understand the impact of these factors on unplanned PICU readmission.

Developing effective monitoring strategies and early risk identification tools is crucial to minimizing unplanned PICU readmissions. Early detection enables a more rapid response from the medical team, allowing them to initiate treatment before the condition worsens. If the intervention is effective, the patient's condition can stabilize, reducing the likelihood of PICU readmission and helping to prevent the morbidity and mortality associated with severe complications. Our findings suggest that patients with risk factors such as univentricular physiology and known postoperative complications may benefit from more intensive monitoring during the transition to the ward. While tools like the Pediatric Early Warning Score (PEWS) have been integrated into electronic medical records to assist with risk stratification, ²⁰ recent studies indicate that PEWS alone may fail to predict PICU transfer or rapid escalation of care in high-resource settings. ²¹

Initially, in our center, after stage 1 surgery, children were discharged from intensive care when their clinical condition no longer requires continuous monitoring, with weaning from various organ support and/or medications that necessitated critical care for at least 48 h. They were then transferred to the cardiac surgery unit, where they remained until there was no further need for monitoring and a regular weight gain was observed. Following discharge, they were followed up externally for one week, then monthly postoperatively (including vital signs, SpO₂, auscultation, echocardiogram and echocardiography. Following this study, we changed our protocol when children were discharged home. They now are called within 24 h of official discharge, have an appointment at the clinic and/or via telemedicine every 1 to 2 weeks with a cardiac nurse practitioner and have an appointment at the clinic with a cardiologist every 4 to 6 weeks. This care requires collaboration from all involved parties (surgeons, cardiologists, intensivists, nurses, as well as parents) and follows care protocols in each department to standardize the care provided to the children as effectively as possible

Future research could explore the use of wearable devices at PICU discharge for remote vital sign monitoring, enabling early detection of postdischarge complications and facilitating timely interventions. ²² Such innovations could enhance clinical decision-making and reduce the burden of unplanned PICU readmissions. This study has several strengths, including its single-center design, which allowed for a detailed analysis of a well-defined cohort of children undergoing cardiac surgery at a tertiary pediatric hospital with standardized management protocols. Additionally, matching each readmitted patient with two controls based on demographic and clinical factors enabled a more robust comparison of similar cohorts, which is especially important in the pediatric population, where patients of different ages have vastly different physiology. The diversity of data collected in our study facilitated a comprehensive investigation of risk factors for readmission, encompassing demographics, as well as preoperative, surgical and postoperative factors. Finally, the use of our PICU database ensured extremely precise data collection, recording critical care therapies to the minute.

Limitations

However, there are limitations to consider. The low incidence of unplanned PICU readmissions (2.4%, 23/956) limited the statistical power of our analysis and may have led to an underestimation of certain risk factors (eg, valvular regurgitation). It is a relatively common postoperative complication, as observed in our population. However, it is possible that a severe grade may be a collection bias, as we considered the worst postoperative echocardiography in order to identify markers associated with the risk of readmission. Furthermore, given the number of readmitted patients, we performed a matching study with controls based on age and PELOD score who were not readmitted to the PICU. Since the patients had a similar clinical and anatomic profile, this could contribute to the particularly high frequency observed in our population. Furthermore, the single-center design may restrict the generalizability of our results to other pediatric hospitals, particularly those with different PICU capacities or patient populations. Unmeasured confounders, such as specific surgical techniques or variations in postoperative care protocols, could also have influenced the outcomes. Finally, as a retrospective study, potential biases related to missing or incomplete records may affect the accuracy of our findings.

Conclusion

In this retrospective study, we identified a 2.4% (23/956) rate of unplanned PICU readmission following pediatric cardiac surgery. Our findings demonstrate that postoperative univentricular physiology is an independent risk factor for PICU readmission, highlighting the increased vulnerability of patients with complex cardiac conditions. The most common reasons for readmission were complications such as pleural effusion and arrhythmias, underscoring the need for enhanced postdischarge monitoring as well as individualized care plans for high-risk patients. A deeper understanding of risk factors associated with unplanned PICU readmission in this population could lead to the identification of patients considered high risk prior to their transfer to the ward. These patients could potentially benefit from the development of optimized early detection systems, including the use of advanced vital sign monitoring technologies outside the PICU, to better prevent and enable early intervention for postdischarge complications. Such systems could improve patient outcomes, minimize rehospitalizations, and enhance the overall quality of care for pediatric cardiac surgery patients.

Supplemental Material