Determinants of Length of Stay after Neonatal Cardiac Surgery Using Path Analysis

Abstract

After neonatal cardiac surgery, families, and the health care team strive for exclusive oral feedings before hospital discharge. With the hypothesis that exclusive oral feedings would reduce the length of stay (LOS), a multidimensional path analysis was used to examine a cross-section of 280 neonates from 2009 to 2013. Buttigieg, Abela, and Pace’s theoretical framework of structural and process-related determinants of LOS was modeled with hypothesis-driven correlation and directionality. The recursive path model had a good global and local fit with outcome variances of 26% for exclusive oral feeding and LOS. In the full cohort and model groups (single and biventricular), when controlling for covariances: sepsis, birth distance, necrotizing enterocolitis, genetic differences, specialty consults, the age at which neonatal cardiac surgery occurred (β = .23, p ≤ .001) and the duration of postoperative intubation (β = .47, p ≤ .001) more significantly influenced the LOS than intermediate mediation of exclusive oral feedings at discharge.

Keywords

pediatric cardiology path analysis length of stay enteral nutrition

Length of stay (LOS) in the hospital after pediatric cardiac surgery is a key performance indicator of best practice and hospital efficiency (Buttigieg et al., 2018). Additionally, LOS helps families anticipate the projected postoperative course for their child. During this postoperative period, the health care team focuses on neonatal feeding and nutrition, care planning with consultants, and family education to help move each patient toward discharge home with a primary goal of exclusive oral feedings. Many determinants of LOS in neonates after cardiac surgery have been evaluated, such as route and timing of feedings, days of intubation, distance from birth location to the surgical hospital, postoperative infections, necrotizing enterocolitis (NEC), genetic abnormalities, and the number of consulting specialists (Erickson et al., 2019; Wong et al., 2015). These determinants have been explored by various statistical approaches, including a deep learning neural network from a pediatric cardiology cohort (2005–2009) (Jalali et al., 2020). Still, to our knowledge, path analysis modeling has not been used recently to examine determinants of LOS in a pediatric cardiology cohort after neonatal cardiac surgery.

Congenital heart disease (CHD) is the most frequent congenital disability in the United States, with approximately one in 100 live births having some form of CHD (Hoffman & Kaplan, 2002). Of these, nearly 25% will have critical CHD requiring cardiac surgery intervention within six weeks of birth (Oster et al., 2013). CHD surgical admissions cost $40,000 per week, and the most critical surgeries have LOS of more than 30 days (Chan et al., 2015). Additionally, hospital-acquired infections may increase hospitalization charges to approximately $185,000 (Tweddell et al., 2019). Along with the significance of financial costs, prolonged neonatal LOS also results in family emotional challenges, traumatic stress, and potential hospital lifestyle changes (Alzawad et al., 2021). Pediatric cardiology patients often require enteral feeding support to achieve adequate growth after surgery, and we hypothesized that this was a primary determinant in prolonging the LOS in the study hospital.

Despite the multidisciplinary input from speech and occupational therapists, family, and the clinical team, inpatient nurses are the primary health care providers who monitor progression toward exclusive oral feedings. Poor progression to exclusive oral feeding after CHD surgery can be due to overstimulation from the intensive care environment, postoperative laryngeal complications, developmental delays, and reduced stamina after cardiac surgery (Jones et al., 2021). In the hospital setting, the focus is primarily on increasing stamina and developing a plan to enhance oral skills and nutrition.

Feeding challenges may impact the need for supplemental enteral nutrition at home for adequate growth of the neonate. Significant short- and long-term implications remain for the neonate and the family’s quality of life when enteral nutrition is required (Jones et al., 2021). In a recent single-site cohort study reviewing neonates who had a gastrostomy tube (g-tube) placed after cardiac surgery, only 19.1% (22/137) progressed and went on to not needing an enteral tube at one year of age (Mahdi et al., 2022). Additionally, neonates with g-tubes have higher risks of infections, emergency room utilization, and hospital readmissions in the first year of life (Khalil et al., 2017). Enteral nutrition provided by a g-tube has neurodevelopmental implications, with recent literature associating neurodevelopmental impairments with lower cognitive, linguistic, and motor scores at the two-year assessment compared to those exclusively orally fed (Holst et al., 2019).

Analysis of Pediatric Cardiac Hospital Length of Stay

The majority of previous publications evaluate LOS in pediatric cardiology statistically through comparing means, calculating analysis of variance distributions, utilizing Kaplan-Meier survival analysis, and developing multivariable regression models (Azhar & Aljefri, 2018). While these methods are critical for quantifying outcomes, these statistical approaches fail to provide a theoretical base or propose hypothetical weights through direct or indirect effects, so conclusions based on results must be assumed. As an extension of multiple regression modeling, path analysis adds the benefit of decomposing all variables in a model to evaluate each parameter, meaning that significant correlations between LOS and determinant variables are the sum of all direct and mediator indirect effects (Kline, 2016). A weighted, theoretically based path analysis model with correlation and proposed directionality provides significant strength for the analysis of LOS in a new way.

Theoretical Framework

Buttigieg et al.’s (2018) theoretical framework for evaluating complex hospital LOS includes three significant domains: structure, process, and outcomes. Within the structure domain, the following three additional areas further define variables: (a) health care systems’ characteristics such as clinical pathways and efficiency of support services; (b) patients’ characteristics such as demographics, medical history, comorbidities, and severity of the condition; and (c) social and family characteristics. The process domain defines clinical caregiver characteristics such as communication, multidisciplinary approach, discharge planning, leadership, and professional groups. The outcome domain is defined as overall LOS and related complications, ultimately leading to family satisfaction (Buttigieg et al., 2018).

The purpose of this study was to use path analysis to theoretically model the impact of exclusive oral feedings at neonatal discharge from the hospital with additional structure and process-related determinants on neonatal hospital LOS after cardiac surgery. We hypothesized that in a path analysis model, a longer LOS would be impacted directly by not achieving exclusive oral feedings at discharge and longer intubation and indirectly by an (a) higher rate of sepsis, (b) farther distance at birth from the study hospital, (c) no preoperative feedings, (d) later postoperative feedings, (e) higher rate of NEC, (f) higher incidence of major genetic syndromes, (g) older age at cardiac surgery, and (h) higher number of specialty consults.

Methods

This study used a retrospective, correlational design utilizing path analysis on an existing dataset to evaluate determinants of LOS among neonates with CHD who underwent cardiac surgery.

Participants and Sampling

A convenience sample of all neonates admitted to a children’s hospital between 2009 and 2013 was included in this study. Initial data analysis focused on patients admitted before and after developing a fetal health delivery center in March 2011. Participant inclusion criteria include children (a) with a diagnosis of CHD, (b) born from January 2008 to March 2013, (c) admitted to a Midwest tertiary freestanding children’s hospital, and (d) who underwent critical cardiac surgery in the first six weeks of life. Children were excluded if they had mortality before neonatal discharge or had only a cardiac catheterization procedure.

Data Collection and Measures

The available sample for this study was 280 neonates. As a rule of thumb for path analysis, a minimum sample size of over 200 participants is needed for adequate power with at least a 5:1 ratio of participants to variables (Kyriazos, 2018). This was achieved with a participant-to-variable ratio of 5.3:1. Variables were obtained for this secondary analysis from a de-identified patient dataset from a pediatric cardiac delivery location at a freestanding children’s hospital. De-identified variables available from the original data set included: (a) neonatal LOS (days), (b) age at cardiac surgery procedure (days), (c) birth location to surgical hospital (miles), (d) length of intubation (days), (e) gender (male/female), (f) gestational age (weeks), (h) birth weight (kg), (i) prenatal detection of CHD (yes/no), (j) single ventricle (SV) CHD (yes/no), (k) biventricular CHD (yes/no), (l) maternal age (years), (m) maternal marital status (single/married), (n) Medicaid/state insurance (yes/no), (o) sepsis (yes/no), (p) fed preoperatively (yes/no), (q) NEC (yes/no), (r) first enteral feeding postoperatively (days), (s) route of feedings on discharge (all by mouth, nasogastric (NG) tube and by mouth, g-tube and by mouth, all NG tube or g-tube), (t) major genetic difference (yes/no), and (u) and the number of specialty consults.

Structure: Patient Characteristics

Type of surgery

While the Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery Score and the Risk Adjustment in Congenital Heart Surgery scores have been used in another LOS analysis, the most critical higher scores have not always correlated with longer LOS (Azhar & Aljefri, 2018). Neonates with staged SV surgeries had more extended postoperative stays than biventricular or definitive surgeries (McHugh et al., 2016; Moore & Gephart, 2021). We stratified the type of neonatal surgery for grouping after a path modeling the complete cohort to understand the impact on the LOS for neonates with more complex SV cardiac surgeries.

Specialty consults

The number of specialty consults, beyond pediatric cardiac surgery and cardiology during the neonatal stay, was a proxy for the complexity of congenital and postoperative issues. This would also be categorized as “Severity of Condition” (Buttigieg et al., 2018). This potentially impacts the discharge process as consults with additional care providers impact workflow on discharge timing. Standard of care consults, such as physical, occupational, speech therapy, and social work, were not counted in the total.

Genetic differences

Neonates with genetic abnormalities have increased postoperative mortality and morbidity (Morris et al., 2014). The comorbidities associated with genetic conditions likely impact LOS (Buttigieg et al., 2018). Significant genetic differences included documented Down Syndrome, 22q11 deletion, and Heterotaxy Syndrome. Variants of unknown significance were excluded from the genetic difference category and neonates were counted as not having a genetic abnormality.

Birth distance

Farther birth hospital distance from the cardiac surgical hospital has also been associated with increased mortality for neonates with the most complex type of CHD, SV CHD but was not found to be associated with LOS (Tregay et al., 2015). The birth distance was calculated as the difference in miles from the center of the zip code where the neonate was born compared with the zip code of the study surgical hospital.

Structure: Health Care Systems’ Characteristics

Age at surgery

The ideal timing of surgery has also been considered a modifiable factor that could impact outcomes. However, recommendations for the ideal timing of neonatal cardiac surgery remain inconsistent, are typically based on retrospective review data, and are usually surgery-specific (Backer, 2015; Kumar et al., 2014). Age at surgery was calculated as the difference between the day of surgery and the neonate’s birthday.

Preoperative feeding

Preoperative enteral feeding of neonates before cardiac surgery has been associated with reduced LOS, fewer hemodynamic issues, more stable postoperative findings, and a shorter time to wean off parenteral nutrition (Martini et al., 2021; Scahill et al., 2017). The measure of preoperative feedings was defined as any enteral feeding the neonate received prior to cardiac surgery, excluding oral care. These included (a) breastfeeding, (b) breast milk via NG tube, (c) expressed breast milk by mouth, (d) formula via NG tube, or (e) formula by mouth. Of note, during the study period, neonates with biventricular heart disease were fed routinely preoperatively, but the standard of care for neonates with SV CHD was not to routinely offer oral or enteral feedings before cardiac surgery.

Postoperative feeding

In a review of predictors of LOS using stepwise regression, postoperative feeding difficulties were found to be associated with a longer LOS (Azhar & Aljefri, 2018). Additionally, delayed postoperative feeding has been associated with a longer length of intubation (Jones et al., 2021). The measure of postoperative feeding was calculated in days by the difference between the surgery day and the day the neonate received any volume of enteral feeding. This included all feeding routes such as breastfeeding, breast milk via tube, expressed breast milk by bottle, formula by tube, or formula by mouth but did not include oral care only.

Sepsis

Infection in the postoperative period after cardiac surgery has been associated with an increased LOS upwards of 23 days (Tweddell et al., 2019), including intensive care unit days (Shao et al., 2020). Sepsis was defined as an infection treated with intravenous antibiotics for a specific length of treatment (more than a 48-hour rule-out of infectious disease).

Necrotizing enterocolitis (NEC)

During neonatal hospitalizations, NEC in the pediatric cardiac population, even nonsurgically treated, has been associated with increased LOS (Arbra et al., 2018). NEC was defined as a treated course of no formula or breast milk (NPO) time, an associated intravenous antibiotic duration of more than 48 hours, and NEC course documented in the electronic medical record (not NEC rule-out), with or without modified Bell staging for NEC in neonates documented.

Process: Clinical Caregiver’s Characteristics

Length of intubation

After pediatric cardiac surgery, the extubation process requires communication between the multidisciplinary intensive care team and the cardiac intensivist. Time to extubation of fewer than 8 hours (fast track) and more than 8 hours was evaluated in 717 neonates (mean age four months) requiring cardiac surgery; there was a lower LOS for those extubated quicker (7 vs. 10 days) and lower mortality, with no difference in reintubation (Murin et al., 2021). Additionally, the length of intubation was associated with an extended stay via proxy through the number of intensive care days (Azhar & Aljefri, 2018). Intubation was defined as the time from the date of cardiac surgery to the first extubation documented in the electronic medical record.

Exclusive oral feedings at discharge

Feeding postoperatively is an essential variable for evaluating LOS and is often the final clinical goal of discharge planning. There is an association with longer intensive care LOS in children who receive enteral feedings compared to children who are fed orally (Azhar & Aljefri, 2018; Mahdi et al., 2022). Exclusive oral feedings were defined as all neonatal nutrition was successfully taken in by mouth with no use of enteral nutrition after hospital discharge via gastrostomy, NG, or nasojejunal tubes. The study hospitals’ care standard of care for neonates with SV CHD included that they are discharged primarily with surgically placed enteral tubes (g-tubes) and not NG tubes. Neonates with biventricular CHD were more likely to be discharged home with an NG tube if they required enteral feeding support.

Procedures

This de-identified patient dataset was obtained after Children’s Mercy Hospital Institutional Review Board approval #14040178 via a nonhuman subjects retrospective review. The determinants were chosen from the available dataset and used to evaluate the LOS among neonates after pediatric cardiac SV and biventricular surgical repair for path analysis. The determinants included significant genetic differences, birth distance, age at cardiac surgery, number of specialty consults, length of intubation, receiving exclusive oral feedings at discharge, postoperative sepsis, and NEC.

Model Specification

The proposed path analysis was modeled for hospital LOS with standardized coefficients for the entire cohort, biventricular, and SV CHD surgery groups. Length of intubation and exclusive oral feedings at discharge were placed in the path model as mediators from the other variables on LOS based on the theoretical framework process domain. These two variables focused more on communication, discharge planning, leadership, and a multidisciplinary approach to care than on the structure-related variables from Buttigieg et al.’s (2018) framework.

The variables were organized in the path analysis model with hypothesized direct and indirect effects. Similar to other modeling analyses (Jalali et al., 2020), the eight variables selected for inclusion in this study were derived from previous literature and theoretical correlation and directionality from Buttigieg et al.’s (2018) theoretical model of hospital LOS and published outcomes (Erickson et al., 2019).

Effect sizes for standardized path coefficients included small = 0.2, medium = 0.5, and large = 0.8 for direct paths, small = 0.1, medium = 0.3, large = 0.5 for nondirected arches, and small = 0.01, medium = 0.09, and large = 0.25 for indirect paths (Hu & Bentler, 1999). Model parameter estimations were reported through standardized regression weights and statistical findings using maximum likelihood in SPSS AMOS (2017). The model chi-square test, including degrees of freedom and p-value, was used to evaluate if the null hypothesis represented a correct model (Kline, 2016). Alternative models with varying directionality of variables on the entire cohort were tested. Further evaluation via model groups of SV and biventricular CHD surgeries compared the neonates’ LOS with primary biventricular surgeries and those with SV physiology.

Length of intubation also had significant kurtosis over the threshold, so a two-step log transformation was used to secure normal distribution and maintain outcome variables’ multivariate normality. A low rate of substantial outliers (7/280, 2.5%) was identified in the outcome variable of LOS with a mean of 36.17 days but was not removed from the analysis. Three cases (1%) had a missing variable and were deleted stepwise to maintain a complete data set at 280 (Kline, 2016). The data were likely missing completely at random since one variable was missing from the distance of delivery, intubation length, and LOS.

Model Testing

Standardized root mean square residual was reported as an absolute fit index, and root mean square error of approximation and its 90% confidence interval were reported as a parsimony index (Kline, 2016). The proposed model was evaluated through a comparative fit index for incremental fit. Estimation was assessed for multivariate normality, including kurtosis ±10 and skewness ±3 (Sterne et al., 2009). Bootstrapping maximum-likelihood and Bollen-Stine were completed to provide a nonparametric estimate of indirect effects through a reference distribution. Using AMOS, 1,000 bootstrap resamples were conducted for the path analysis to evaluate indirect path coefficients and categorical variables (Kline, 2016).

Results

Demographics of the cohort, path analysis model variables, and comparisons for SV and biventricular CHD surgery groups are presented in Table 1. The route of feeding at discharge varied across the cohort, with 64.3% (180/280) of neonates receiving exclusive oral feedings at hospital discharge, 19.3% (54/280) of neonates receiving oral and NG feedings, 5.0% (14/280) receiving oral and gastric tube feedings, and 11.4% (32/280) received only gastric tube feedings.

Table 1.

Cohort Demographic and Descriptive Statistics with Comparisons of Groups by Surgery Types.

	Overall Cohort(n = 280)	Single Ventricle(n = 115)	Biventricular(n = 165)	p-Value
Continuous Variables (Mean ± SD)
Gestational age	38.4 ± 1.62	38.4 ± 1.5	38.4 ± 1.7	.916
Birthweight (kg)	3.26 ± 0.55	3.22 ± 0.52	3.29 ± 0.58	.295
Maternal age (years)	26.9 ± 5.57	26.8 ± 5.9	27.0 ± 5.38	.859
Neonatal LOS (days)	36.1 ± 28.1	45.75 ± 23.8	29.6 ± 27.3	<.001
Age cardiac surgery (days)	15.1 ± 7.33	14.7 ± 7.1	15.5 ± 7.5	.322
Birth distance from the surgical hospital (miles)	72.8 ± 93.4	60.5 ± 95.0	82.1 ± 92.2	.058
Length of intubation post-op (days)	5.76 ± 7.07	9.25 ± 12.7	4.9 ± 13.8	.008
First feeding post-surgery (days)	5.03 ± 6.47	4.93 ± 8.31	5.1 ± 4.8	.839
Specialists: Number of specialty consults	1.41 ± 1.82	1.28 ± 1.72	1.51 ± 1.91	.301
Categorical variables (frequency [%])
Female	100 (35.7)	45 (39.1)	55 (33)	.003
Prenatal detection	101 (36.1)	68 (59.1)	33 (20)	<.001
Parents married	189 (67.5)	69 (60)	120 (73)	.025
Medicaid insurance	138 (49.6)	75 (65.2)	73 (44)	.001
Exclusive oral feeds at discharge	180 (64.3)	74 (64.3)	106 (64.2)	.799
Feeding pre-surgery	193 (68.9)	79 (68.7)	114 (69.1)	.944
Necrotizing enterocolitis	24 (8.5)	8 (7.0)	16 (9.7)	.422
Sepsis	43 (15.5)	14 (12.2)	29 (17.6)	.219
Major genetic anomaly	37 (13.2)	10 (8.7)	27 (16.4)	.083

Note: CHD = congenital heart disease; CI = confidence intervals; LOS = length of stay; SV = single ventricle; SD = standard deviation.

Model Assessment

The model was assessed and found to be theoretically identified, with degrees of freedom greater than zero, and all disturbances were scaled. The model assessment included a chi-square χ² = 16.73, degrees of freedom (df) = 16, and p-value = .400, which is nonsignificant, so the null hypothesis was retained. The model’s fit indices included a comparative fit index = 1.0, root mean square error of approximation = .01 with a 95% confidence interval [.00, .06], and standardized root mean square residual = .0288. These findings indicate a good global fit with low standardized root mean square residual, high comparative fit index, and the overall, low, and high 95% confidence interval of root mean square error of approximation in the range of an adequate fit. Additionally, no standardized residual covariances were >2 or <–2, indicating a good local fit. There were no findings of extreme collinearity (ρ > .999) between variables (Schisterman et al., 2017). A full correlation matrix with associated variances and covariances is available in Supplementary Resource 1.

Table 2 denotes standardized and unstandardized parameter estimates of the final path model. All variances were significant, with four significant direct path coefficients (Hu & Bentler, 1999). An alternative model was evaluated with intubation having covariances with all other exogenous variables and no direct effect on LOS but mediated through the need for tube feeding. The entire cohort alternative model had a low variance for LOS (1%) and a weak global fit, so the proposed model was retained.

Table 2.

Parameter Estimates of the Final Path Model (n = 280).

Direct Path Coefficients			Indirect Path Coefficients
Path	Unstand. (SE)	r	Variable	Unstand. (SE)	r
PO ← Feed post	−0.01 (0.00)	−.13	LOS ↔Age surgery	−0.18 (0.44)	−.038
PO ← Birth distance	0.00 (0.00)†	−.15	LOS ↔NEC	0.58 (2.95)	.005
PO ← Genetic anomaly	−0.07 (0.09)	−.05	LOS ↔Feed pre	0.01 (1.09)	.000
PO ← Feeding pre	0.00 (0.06)	.00	LOS ↔Feed post	0.04 (0.17)	.008
PO ← Sepsis	0.05 (0.08)	.04	LOS ↔Specialists	0.46 (1.38)	.023
PO ← Specialists	−0.10 (0.02)‡	−.39	LOS ↔Genetics	0.30 (2.17)	.003
PO ← NEC	−0.13 (0.10)	−.08	LOS ↔Sepsis	−0.23 (2.39)	−.002
IT ← Age at surgery	−0.15 (0.11)	−.08	LOS ↔Distance	0.00 (0.01)	.009
LOS ← PO	−4.4 (3.77)	−.06	Variances
LOS← Age at surgery	1.10 (0.25)‡	.23	Variable	Unstand. (SE)	R ²
LOS ← IT	1.23 (0.13)‡	.47	Birth distance	8760.6 (741)‡	—
			Sepsis	0.13 (0.01)‡	—
			Genetics	0.11 (0.01)‡	—
			Specialists	3.36 (0.28)‡	—
			Feeding post	41.72 (3.53)‡	—
			Feeding pre	0.21 (0.02)‡	—
			NEC	0.08 (0.01)‡	—
			Age at surgery	53.49 (4.53)‡	—
			ePO	0.17 (0.01)‡	.26
			eIT	180.76 (15.3)‡	.01
			eLOS	910.74 (77.1)‡	.26

Note: IT = intubation days; LOS = length of stay; NEC = necrotizing enterocolitis; PO = feeding by mouth; SE = standard error. R² = variance. r = standardized regression weight. Unstand. = unstandardized. †p ≤ .01. ‡p ≤ .001.

The final recursive path model in Figure 1 had a good global and local fit and an overall explanation of variance (R²) of 26% for LOS. Additionally, 26% of the variability around the mean for LOS was explained with a medium effect size when the decomposition of other variables was controlled. The low variance of 1% for intubation indicated there were variables for intubation length that were not accounted for in the model. The significant direct and indirect coefficient regression weights for the entire cohort of n = 280, neonates with SV CHD at n = 115, and those with biventricular CHD at n = 165 are additionally presented in Figure 1.

Model Groups

Across the path model’s three groups (the overall cohort, SV, and biventricular), there were only two significant determinants of exclusive oral feedings (birth distance and the number of specialists) and two significant determinants of LOS (age at cardiac surgery and intubation days). The LOS for neonates with SV was indirectly negatively influenced through intubation days from the age at cardiac surgery with a small effect size (β = –.07, p = .01). For neonates with SV, CHD age at surgery was found to have a direct negative influence on the length of intubation (β = –.25, p = .001). There was a significant influence of exclusive oral feedings at discharge in neonates after biventricular surgery with a lower incidence of NEC (β = .10, p ≤ .001) and when they were fed sooner postoperatively (β = –.14., p = .03).

Discussion

This study evaluated the proposed relationships between neonatal hospital structure and process characteristics that may influence hospital LOS after complex cardiac surgery using a path analysis model. An acceptable theoretical fit of the proposed model was delineated. An overall good model variance was explained at 26%, with a medium effect size and adequately powered model variable ratio. Surprisingly, we found that exclusive oral feedings at discharge did not independently significantly impact hospital LOS when all other variables and directionality in the path model were considered. Several findings in this study support the utilization of path analysis to understand areas of impact in complex processes such as hospital discharge after neonatal cardiac surgery. This study aimed to provide new insights for preoperative LOS counseling with families and postoperative interventions for nursing and health care teams to reduce postoperative LOS after pediatric cardiac surgery.

The age that neonatal cardiac surgery occurred and the duration of postoperative intubation more significantly influenced the LOS when the following covariances were controlled: sepsis, birth distance from the surgical hospital, NEC, genetic differences, preoperative feedings, postoperative feedings, the number of specialty consults with intermediate mediation of exclusive oral feedings at discharge, and age that neonatal cardiac surgery occurred. Our findings indicated that an outcome of LOS was modeled using path analysis with the most significant determinants on the outcome found to be the neonates’ age at the time of their cardiac surgery and the associated length of intubation days, and not the route of feedings at discharge as hypothesized.

The study’s resulting demographics and incidence of determinants of LOS were comparable (Table 1) with similar cohorts of patients with CHD (Ferguson et al., 2017; Karamlou et al., 2018). Despite finding no statistically significant differences in the majority of demographic and model variables between the two groups, neonates with SV did tend to have a higher frequency of these morbidity determinants, likely prompting the longer LOS in the SV and biventricular groups (45.75 vs. 29.56 days, p < .001). In the entire cohort and both SV and biventricular groups, a higher rate of exclusive oral feedings by mouth at discharge was found when the neonates had fewer specialty consults and were born farther away from the hospital. These were potential markers of less severe CHD with improved neonatal postoperative recovery and fewer comorbidities.

Despite the path analysis not showing a significant correlation with exclusive oral feeding at discharge with LOS, supporting neonatal discharge planning with oral feedings is critical for neonatal developmental and nutritional outcomes. Preoperative feeding rates and the 64% rate of exclusive oral feedings on discharge found in this study are similar to findings in other cohorts in high-risk pediatric CHD studies (O’Neal Maynord et al., 2021). There was a stronger association between receiving exclusive oral feeds at discharge in neonates with biventricular CHD who were fed sooner postoperatively and had a lower incidence of NEC. Postoperative feeding clinical pathways are also increasingly focused on neonates with complex CHD (Anderson et al., 2019). These pathways may help to improve the number of days for the neonate to receive enteral nutrition after neonatal CHD surgery.

Previous studies found correlations between higher rates of exclusive oral feedings at discharge and a shorter length of intubation postoperatively (Kurtz et al., 2019). The finding of 1% variance likely means that measures that could improve the understanding of intubation on LOS and the impact on exclusive oral feeding at discharge are missing. The number of days of intubation was also evaluated as a mediator variable in the model, with the hypothesis that longer intubation days would result in a higher level of postoperative support. This assumption was confirmed with the finding that the youngest neonates at the time of surgery for SV CHD were intubated longer, mediating a significant result in extended LOS. However, this was not found in younger neonates with biventricular CHD. Clinical pathway evaluation of the health care process for extubation practices in intensive care settings may be needed to understand these differences adequately. Focused fast-track extubation after neonatal cardiac surgery has been implemented safely with shortened LOS and less nursing and intensive care staff resource utilization (Murin et al., 2021). Evaluating these programs’ focus areas may provide insights to assess related to shortening LOS and understanding missed measures for oral feedings and intubation days.

Older age at cardiac surgery was a structure-related determinant found to have a significant association with a longer LOS in the path analysis model. There were no differences in surgical timing at a mean of about two weeks of age for both SV and biventricular surgery groups. This study period ended in 2014, and clinical practices for the ideal age at neonatal surgery are currently more closely aligned to being completed in the first week of life (Smith et al., 2022). The ideal timing of SV and biventricular cardiac surgery is complex and may be impacted by numerous factors such as the neonates’ clinical symptoms or process-related determinants such as operating room scheduling practices and prenatal detection. Neonates with SV CHD had a higher prenatal detection rate, indicating a possible earlier arrival to a pediatric surgery center for surgical care. The older age at the surgery with LOS would need to be repeated in a path analysis model with individual subgroups of surgery type to understand the association with the current, comparable literature (Smith et al., 2022).

There are several limitations to this study. First, this study used a single site. However, the sample’s demographics are comparable to the general population, increasing the generalizability of these findings (Ferguson et al., 2017). Second, the length of time needed to obtain a volume of over 200 patients in pediatric cardiology for an adequately powered path analysis was over four years. During this timeframe, there may have been changes in the standard of care and provider preferences in this population that may have impacted LOS that were not accounted for in this study. Although a medium effect size was found on the percent variance explained for the model, other determinants may still need to be added to future path analysis to improve the overall understanding of the LOS with a goal of a higher variance explained result. Finally, this study used a de-identified dataset with associated limitations on data manipulation and explanations for outlying data. However, due to only a few data outliers, there was a low need for data manipulation to maintain the multivariate normality and data quality used in path analysis.

In conclusion, this study evaluated determinants of LOS after neonatal cardiac surgery using a novel path analysis approach. We hypothesized from the research literature and clinical experience that neonates receiving exclusive oral feedings at discharge would have shorter LOSs. However, our hypothesis was not supported. We found that through path analysis modeling, receiving exclusive oral feedings did not significantly affect LOS while accounting for other process and structure-related variables. Considering multiple covarying determinants, the length of intubation and age at surgery more significantly impacted the LOS after pediatric cardiac surgery. In addition to the potential clinical impact of this study on inpatient throughput after neonatal cardiac surgery, health care teams can utilize multidimensional path modeling with hypothesized correlation and directionality to evaluate many different types of complex processes for new insights.

Supplemental Material

sj-pdf-1-wjn-10.1177_01939459221129037 – Supplemental material for Determinants of Length of Stay after Neonatal Cardiac Surgery Using Path Analysis

Supplemental material, sj-pdf-1-wjn-10.1177_01939459221129037 for Determinants of Length of Stay after Neonatal Cardiac Surgery Using Path Analysis by Lori A. Erickson, Amy Ricketts, Tara Swanson, Julie Weiner, Usman A. Hasnie, Katherine Bonessa, Janelle Noel-Macdonnell and Cynthia L. Russell in Western Journal of Nursing Research

Footnotes

Acknowledgements

The first author is a Jonas Foundation Nursing scholar. We would like to thank Children’s Mercy Ward Family Heart Center staff, specifically the Research team, for their support and persistence for this research. Additionally, UMKC’s Dr. Jacob Marszalek for sparking the interest to use path analysis as an advanced model for understanding outcomes.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. The first author received a Mary Pudelko Bubacz grant for Nursing Excellence from the UMKC Women’s Council for this research.

ORCID iD

Lori A. Erickson

Supplemental Material

Supplemental material for this article is available online.

References

Alzawad

Marcus Lewis

Ngo

Thomas

(2021). Exploratory model of parental stress during children’s hospitalisation in a paediatric intensive care unit. Intensive and Critical Care Nursing, 67, 103109. https://doi.org/10.1016/j.iccn.2021.103109

AMOS SPSS Corporation. (2017). [Software] Version 25. IBM.

Anderson

J. B.

Brown

D. W.

Lihn

Mangeot

Bates

Van Bergen

A. H.

Rudd

Hanke

Tweddell

Lannon

(2019). Power of a learning network in congenital heart disease. World Journal for Pediatric and Congenital Heart Surgery, 10(1), 5. https://doi.org/10.1177/2150135118815023

Arbra

C. A.

Oprisan

Wilson

D. A.

Ryan

R. M.

Lesher

A. P.

(2018). Time to reintroduction of feeding in infants with nonsurgical necrotizing enterocolitis. Journal of Pediatric Surgery, 53(6), 1187–1191. https://doi.org/10.1016/j.jpedsurg.2018.02.082

Azhar

A. S.

Aljefri

H. M.

(2018). Predictors of extended length of hospital stay following surgical repair of congenital heart diseases. Pediatric Cardiology, 39(8), 1688–1699. https://doi.org/10.1007/s00246-018-1953-1

Backer

C. L.

(2015). Timing is everything; Or is it? The Journal of Thoracic and Cardiovascular Surgery, 149(1), 211–212. https://doi.org/10.1016/j.jtcvs.2014.10.051

Buttigieg

S. C.

Abela

Pace

(2018). Variables affecting hospital length of stay: A scoping review. Journal of Health Organization and Management, 32(3), 463–493. https://doi.org/10.1108/jhom-10-2017-0275

Chan

Kim

Minich

L. L.

Pinto

N. M.

Waitzman

N. J.

(2015). Surgical volume, hospital quality, and hospitalization cost in congenital heart surgery in the United States. Pediatric Cardiology, 36, 205–213. https://doi.org/10.1007/s00246-014-0987-2

Erickson

Bonessa

Ricketts

Hasnie

Noel-MacDonnell

Jayaram

Weiner

Swanson

(2019). Associations of enteral tube feedings at discharge for neonates with critical heart disease. World Journal for Pediatric and Congenital Heart Surgery, 10(2), NP1–NP76. https://doi.org/10.1177/2150135119830271

10.

Ferguson

L. P.

Gandiya

Kaselas

Sheth

Hasan

Gabra

H. O.

(2017). Gastrointestinal complications associated with the surgical treatment of heart disease in children. Journal Pediatric Surgery, 52(3), 414–419. https://doi.org/10.1016/j.jpedsurg.2016.10.052

11.

Hoffman

J. L.

Kaplan

(2002). The incidence of congenital heart disease. Journal for American College of Cardiology, 39(12), 1890–1900.

12.

Holst

L. M.

Serrano

Shekerdemian

Ravn

H. B.

Guffey

Ghanayem

N. S.

Monteiro

(2019). Impact of feeding mode on neurodevelopmental outcome in infants and children with congenital heart disease. Congenital Heart Disease, 14(6), 1207–1213. https://doi.org/10.1111/chd.12827

13.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55.

14.

Jalali

Lonsdale

Peck

Gupta

Kutty

Ghazarian

Jacobs

Rehman

Ahumada

L. M.

(2020). Deep learning for improved risk prediction in surgical outcomes. Scientific Reports, 10(1), Article 9289. https://doi.org/10.1038/s41598-020-62971-3

15.

Jones

C. E.

Desai

Fogel

J. L.

Negrin

K. A.

Torzone

Willette

Fridgen

J. L.

Doody

L. R.

Morris

Engstler

Slater

N. L.

Medoff-Copper

Smith

Harris

B. D.

Butler

S. C.

(2021). Disruptions in the development of feeding for infants with congenital heart disease. Cardiology in the Young, 31(4), 589–596. https://doi.org/10.1017/S1047951120004382

16.

Karamlou

Peyvandi

Federman

Goff

Murthy

Kumar

S. R.

Southwest Congenital Cardiac

(2018). Resolving the Fontan paradox: Addressing socioeconomic and racial disparities in patients with a single ventricle. Journal Thoracic Cardiovascular Surgery, 155(4), 1727–1731. https://doi.org/10.1016/j.jtcvs.2017.11.103

17.

Khalil

S. T.

Uhing

M. R.

Duesing

Visotcky

Tarima

Nghiem-Rao

T. H.

(2017). Outcomes of infants with home tube feeding: Comparing nasogastric vs. gastrostomy tubes. Journal of Parenteral and Enteral Nutrition, 41(8), 1380–1385. https://doi.org/10.1177/0148607116670621

18.

Kline

R. B.

(2016). Principles and practice of structural equation modeling (4th ed.). The Guilford Press.

19.

Kumar

T. K.

Charpie

J. R.

Ohye

R. G.

Hirsch-Romano

J. C.

Donohue

J. E.

Sood

Wilkinson

D. A.

Nelson

Mitchell

Goldberg

Gaies

M. G.

(2014). Timing of neonatal cardiac surgery is not associated with perioperative outcomes. Journal of Thoracic Cardiovascular Surgery, 147(5), 1573–1579. https://doi.org/10.1016/j.jtcvs.2013.07.020

20.

Kurtz

J. D.

Chowdhury

S. M.

Woodard

F. K.

Strelow

J. R.

Zyblewski

S. C.

(2019). Factors associated with delayed transition to oral feeding in infants with single ventricle physiology. The Journal of Pediatrics, 211, 134–138. https://doi.org/10.1016/j.jpeds.2019.02.030

21.

Kyriazos

T. A.

(2018). Applied psychometrics: Sample size and sample power considerations in factor analysis (EFA, CFA) and SEM in general. Psychology, 9(8), 2207–2230. https://doi.org/10.4236/psych.2018.98126

22.

Mahdi

E. M.

Tran

N. N.

Ourshalimian

Sanborn

Alquiros

M. T.

Squillaro

Lascano

Herrington

Kelley-Quon

L. I.

(2022). Factors impacting long-term gastrostomy tube dependence in infants with congenital heart disease. Journal of Surgical Research, 270, 455–462. https://doi.org/10.1016/j.jss.2021.09.023

23.

Martini

Beghetti

Annunziata

Aceti

Galletti

Ragni

Donti

Corvaglia

(2021). Enteral nutrition in term infants with congenital heart disease: Knowledge gaps and future directions to improve clinical practice. Nutrients, 13(3), 932. https://www.mdpi.com/2072-6643/13/3/932

24.

McHugh

K. E.

Pasquali

S. K.

Hall

M. A.

Scheurer

M. A.

(2016). Impact of postoperative complications on hospital costs following the Norwood operation. Cardiology in the Young, 26(7), 1303–1309. https://doi.org/10.1017/S1047951115002498

25.

Moore

E. F.

Gephart

S. M.

(2021). State of the science of care coordination, rurality, and well-being for infants with single ventricle heart disease in the Interstage period, an integrative review. Heart Lung: The Journal of Cardiopulmonary and Acute Care, 50(5), 720–729. https://doi.org/10.1016/j.hrtlng.2021.04.014

26.

Morris

S. A.

Ethen

M. K.

Penny

D. J.

Canfield

M. A.

Minard

C. G.

Fixler

D. E.

Nembhard

W. N.

(2014). Prenatal diagnosis, birth location, surgical center, and neonatal mortality in infants with hypoplastic left heart syndrome. Circulation, 129(3), 285–292. https://doi.org/10.1161/circulationaha.113.003711

27.

Murin

Weixler

V. H. M.

Romanchenko

Schulz

Redlin

Cho

Sinzobahamvya

Miera

Kuppe

Berger

Photiadis

(2021). Fast-track extubation after cardiac surgery in infants: Tug-of-war between performance and reimbursement? Journal Thoracic Cardiovascular Surgery, 162(2), 435–443. https://doi.org/10.1016/j.jtcvs.2020.09.123

28.

O’Neal Maynord

Johnson

Slaughter

J. C.

Killen

S. A. S

. (2021). A multi-interventional nutrition program for newborns with congenital heart disease. The Journal of Pediatrics, 228, 66–73.E2. https://doi.org/10.1016/j.jpeds.2020.08.039

29.

Oster

M. E.

Lee

K. A.

Honein

M. A.

Riehle-Colarusso

Shin

Correa

(2013). Temporal trends in survival among infants with critical congenital heart defects. Pediatrics, 131(5), e1502–1508. https://doi.org/10.1542/peds.2012-3435

30.

Scahill

C. J.

Graham

E. M.

Atz

A. M.

Bradley

S. M.

Kavarana

M. N.

Zyblewski

S. C.

(2017). Preoperative feeding neonates with cardiac disease: Is the necrotizing enterocolitis fear justified? World Journal for Pediatric and Congenital Heart Surgery, 8(1), 62–68. https://doi.org/10.1177/2150135116668833

31.

Schisterman

E. F.

Perkins

N. J.

Mumford

S. L.

Ahrens

K. A.

Mitchell

E. M.

(2017). Collinearity and causal diagrams: A lesson on the importance of model specification. Epidemiology, 28(1), 47–53. https://doi.org/10.1097/ede.0000000000000554

32.

Shao

P. L.

Chen

M. Y.

M. H.

Wang

J. K.

Huang

L. M.

Chiu

S. N.

(2020). Nosocomial severe bacterial infection after cardiac surgery for complex congenital heart disease in heterotaxy syndrome. The Pediatric Infectious Disease Journal, 39(8), e163–e168. https://doi.org/10.1097/INF.0000000000002672

33.

Smith

A. H.

Shin

A. Y.

Tabbutt

Banerjee

Zhang

Borasino

Elhoff

J. J.

Gaynor

J. W.

Ghanayem

N. S.

Pasquali

S. K.

Louis

J. D.

Shashidharan

Ruppe

Schumacher

K. R.

Gaies

Costello

J. M.

(2022). Age at surgery and outcomes following neonatal cardiac surgery: An analysis from the Pediatric Cardiac Critical Care Consortium. The Journal of Thoracic and Cardiovascular Surgery, 1–11. https://doi.org/10.1016/j.jtcvs.2022.05.029

34.

Sterne

J. A. C.

White

I. R.

Carlin

J. B.

Spratt

Royston

Kenward

M. G.

Wood

Carpenter

J. R.

(2009). Multiple imputation for missing data in epidemiological and clinical research: Potential and pitfalls. BMJ, 338, b2393. https://doi.org/10.1136/bmj.b2393

35.

Tregay

Wray

Bull

Franklin

R. C.

Daubeney

Barron

D. J.

Brown

Knowles

R. L.

(2015). Unexpected deaths and unplanned re-admissions in infants discharged home after cardiac surgery: A systematic review of potential risk factors. Cardiology in the Young, 25(5), 839–814. https://doi.org/10.1017/S1047951114002492

36.

Tweddell

Loomba

R. S.

Cooper

D. S.

Benscoter

A. L.

(2019). Health care-associated infections are associated with increased length of stay and cost but not mortality in children undergoing cardiac surgery. Congenital Heart Disease, 14(5), 785–790. https://doi.org/10.1111/chd.12779

37.

Wong

J. M.

Cheifetz

I. M.

Ong

Nakao

Lee

J. H.

(2015). Nutrition support for children undergoing congenital heart surgeries: A narrative review. World Journal for Pediatric and Congenital Heart Surgery, 6(3), 443–454. https://doi.org/10.1177/2150135115576929

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.44 MB