Introduction

The arterial switch operation (ASO) has become the procedure of choice for surgical correction of transposition of great arteries (TGA, ventriculoarterial discordance) with very low mortality reported by many centers [1–3]. Due to decreasing mortality rates, the clinical focus has shifted to decreasing morbidity in pediatric cardiac surgery patients [4–6]. Length of postoperative stay (LOS) contributes significantly to cost of medical care and is independently associated with a number of morbidities including poor neurodevelopment [7, 8]. Several reports have focused on anatomical, intraoperative and postoperative risk factors affecting morbidity and LOS after ASO, but limited information is available regarding preoperative factors associated with prolonged hospital stay [9–11]. In that light, we reviewed our experience with ASO over a 10-year period (November 2001 to November 2011) in order to determine what preoperative risk factors are associated with prolonged postoperative stay.

The objective of this study was to explore associations between the preoperative management and postoperative length of stay after ASO.

Patients and Methods

Patients

The study was approved by Institutional Review Board of Miami Children’s Hospital. Given the retrospective nature of the review, individual informed consent was waived. All patients undergoing ASO at Miami Children’s Hospital between November 2001 and November 2011 were retrospectively reviewed. Data were acquired from the electronic medical record for all patients. We identified patients with postoperative stay less than 7 days (SS group) and patients with postoperative stay more than 14 days (LS group). Patients with an expected postoperative length of stay (7–14 days) were not included in the comparison. In our institution, all neonates undergoing heart surgery remain in the cardiac intensive care unit until they are ready for discharge home and that time point was used to define postoperative length of stay.

All patients were operated upon by the same two surgeons with one stage neonatal repair for all patients. Similar anesthetic and perfusion management was provided for all patients. The leadership and programmatic philosophy of the intensive care team remained unchanged during the study period.

The usual feeding practice in our Unit for babies with transposition of the great arteries is to encourage early oral (bottle/breast) feeding as soon as possible. We view oral feeding as a developmental milestone, and we focus equally on experience as well as the volume of feeds. Therefore, early oral (bottle/breast) feeds are initiated at the discretion of the attending physician with nasogastric supplementation as needed. We initiate feeds with 20 kcal per ounce (30 ml) of formula (breast milk or other standard formula based on cow’s milk). For the purpose of this study, we defined “established feeds” as any volume of enteral feeds more than 50 ml/kg per day. Typically, parenteral nutrition was discontinued once enteral feeds exceed two thirds of total fluid requirement. The presence of umbilical arterial and venous lines was not considered contraindications to enteral feeding. The need for inotropic support was a relative contraindication to feeding.

Risk Factors

We recorded and analyzed preoperative factors such as birth weight, age at surgery, performance of balloon atrial septostomy (BAS) and management the day prior to surgery including use of PGE1, inotropes, intubation status and the establishment of enteral feeds. BAS was performed routinely except in patients with a clearly unrestrictive atrial septum. The decision to feed a patient preoperatively was part of a programmatic philosophy to establish preoperative feeds but was left to the discretion of the attending physician.

Statistical Analysis

Statistical analysis was performed using the Statistical Package for Social Sciences 17.0 for windows (SPSS, Chicago, IL, USA). Due to the small sample size, all data were analyzed using nonparametric tests. Categorical data were expressed as proportions, and groups were compared using the Fisher’s exact test. Continuous variables were expressed as median with range. Comparisons between groups were performed using Mann–Whitney U test for variables with non-normal distribution. Statistical significance was defined at p ≤ 0.05.

Results

During the study period (November 2001 to November 2011), 105 arterial switches were performed at Miami Children’s Hospital (78 ASO, 25 ASO with ventricular septal defect (VSD) closure and 4 ASO with VSD closure and aortic coarctation repair). The status at the time of ASO for all 105 patients during the study period is given in Table 1. Only one postoperative death occurred. There was no difference in postoperative length of stay for patients operated on in the first 5 years versus the last 5 years (10 vs. 11.5 days) of the study period.

Table 1 Status at the time of ASO for all 105 patients during the study period
Full size table

The SS group had 25 patients and the LS group had 32 patients. Both groups (SS vs. LS) were similar in weight (3.4 ± 0.4 vs. 3.2 ± 0.6, p = 0.114), PGE1 use (48 vs. 69 %), BAS (76 vs. 59 %), age at surgery (6 vs. 7 days) and preoperative inotropes (12 vs. 38 %). The SS group had significantly higher incidence of preoperative feeding (80 vs. 31 %, p < 0.001) and less frequent intubation (12 vs. 47 %, p < 0.001). The SS group had shorter cardiopulmonary bypass time 203 ± 46 versus 237 ± 56 min (p + 0.019). Operations performed in the SS group were ASO (n = 17), ASO and VSD repair (n = 7), ASO and aortic coarctation repair (n = 1). In the LS group, there were ASO (n = 21), ASO and VSD repair (n = 4) and ASO/VSD repair/aortic arch reconstruction (n = 6).

The comparison of the variables evaluated and their association with short and prolonged postoperative stay is described in Table 2. There was no difference in use of PGE1, BAS, and on age at surgery between the two groups. The preoperative use of inotropes was more frequent in the LS group but did not reach statistical significance (12 vs. 38 %, p = 0.06).

Table 2 Comparison between SS and LS groups for the variables evaluated
Full size table

Patients in the SS group were statistically more likely to be fed preoperatively (80 vs. 31 %, p < 0.001). Preoperative feeds were present in 50 out of 105 patients at the time of ASO. All patients who were fed were not intubated, although mechanical ventilation was not considered a contraindication to feeding. The majority of these feeds were ad lib oral feeds accounting for the full caloric intake (44 out of 50). One patient received oral and complementary gavage feeds. Four patients received gavage feeds only, and one patient received oral feed plus complementary parenteral nutrition.

Patients in the SS group were also more likely to be non-intubated the day of ASO (12 vs. 47 %, p = 0.01). The impact of preoperative intubation or establishment of preoperative feeding on postoperative length of stay is demonstrated in Fig. 1.

Fig. 1
figure 1

Impact of preoperative intubation and establishment of preoperative feeding on postoperative length of stay

Full size image

The postoperative length of stay in relation to preoperative feeds and preoperative ventilation status is demonstrated in Figs. 2 and 3, respectively.

Fig. 2
figure 2

Postoperative stay in relation to establishment of preoperative feeds

Full size image
Fig. 3
figure 3

Postoperative stay in relation to preoperative ventilation status

Full size image

Comment

We have identified potentially modifiable preoperative risk factors associated with prolonged hospital stay in a large cohort of patients after ASO. Our patients had excellent operative results with mortality less than 1 % which compares favorably to other reports [12, 13]. Despite the excellent mortality outcomes, a subgroup of patients had a protracted postoperative length of stay. Identifying risk factors that can accurately define this subgroup of patients can provide benefits in two directions. First, it can allow for efforts to eliminate or reverse these factors with potential benefit to patient outcome. Postoperative LOS has been adversely associated with neurological dysfunction in children with heart disease and a number of unfavorable outcomes in cardiac as well other group of patients [7]. Second, it can also allow for risk stratification, patient selection and efficient use of healthcare resources by decreasing the degree of variance in perioperative medical care and streamlining resource allocation [14].

In this cohort of 105 consecutive patients undergoing ASO, we found that prolonged postoperative length of stay was associated with lack of enteral feeds, and the need for preoperative mechanical ventilation and is unrelated to the performance of BAS and use of PGE1.

Our findings reinforce results by other investigators with regards to BAS and use of PGE1 [1, 15]. Additionally, we have identified previously poorly described associations of preoperative feeding and non-intubated status with shorter length of stay after ASO.

Our study demonstrates that preoperative feeding has a positive association with shorter postoperative course with no adverse effects. This finding agrees with Wheeler et al. [15] who described 61 neonates after ASO and reported that lack of initiation of oral feeding is associated with prolonged stay in the cardiac intensive care unit. Enteral nutrition is challenging during the preoperative period in neonates with congenital heart disease [16]. Use of prostaglandins for ductal patency, umbilical artery catheters and vasoactive infusions can be seen as contraindications to enteral feeding. Additionally, fluid restriction, that sometimes is required in critically ill neonates as hemodynamic intervention, may result to suboptimal caloric intake and compromised metabolic reserves [17]. Furthermore, neonates undergoing cardiopulmonary bypass experience a more profound metabolic response to stress compared to older children and adults [18, 19]. These factors all may result in a suboptimal nutritional status with adverse effects on growth, wound healing and immune function. Equally important to positive nutritional balance prior to major cardiopulmonary bypass surgery is the experience of oral feeding and its associated benefits on oral-motor skills and sucking–swallowing coordination [20, 21]. Establishment of adequate oral feeding can be problematic after neonatal surgery for several reasons that include deficient oropharyngeal motor skills and feeding aversion [20, 21]. In that respect, it seems logical that oral preoperative feeding would be beneficial in both optimizing nutritional status prior to surgery and promoting developmental milestones necessary to support oral feeding after surgery.

We also found that neonates who are receiving mechanical ventilation the day prior to ASO are more likely to have a prolonged postoperative course. Our findings in that respect are similar to those reported by Dibardino et al. [1] who also, in a retrospective review of outcomes of 125 patients after ASO, concluded that preoperative mechanical ventilation is associated with longer ICU stay and longer ventilator and inotrope dependence after surgery. Other investigators have reported similar associations in infants and children after cardiac surgery [5, 6, 15, 22]. Unfortunately, due to the retrospective nature of our study, we were unable to record evidence of cardiac mixing and end-organ perfusion such as mixed venous oxygen saturation or blood lactate that will have provided a better understanding of the preoperative hemodynamics in both groups. These values were too infrequently recorded to be included in the study. Assuming comparable hemodynamics in both groups, our findings would indicate that achieving or maintaining a non-ventilated status preoperatively will have beneficial effect on lung mechanics, respiratory drive and avoidance of excessive sedation and can favorably affect the postoperative length of ventilation and hospital stay. Although this assumption cannot be proven retrospectively, our findings are in keeping with previous literature and may reflect a genuine favorable association between preoperative non-intubated status and decreased LOS. We continue, therefore, to aggressively pursue attempts to discontinue sedation and wean ventilatory support, if feasible, even for short periods prior to ASO.

In our study, preoperative use of inotropes trended toward, but did not reach, statistical significance. This is in contrast with the findings in other studies that preoperative inotropes extend the postoperative length of stay [1, 15]. This difference could be due the small size of our sample since we only included in the statistical analysis the two ends of our study population (SS vs. LS).

We did not find any association between the use of PGE1 and postoperative length of stay. Previous reports on this matter have been conflicting. Two previous studies are in keeping with our findings. Dibardino et al. found no association between preoperative prostaglandin and length of stay. Similarly, findings by Butt et al. agree with our results. In this study, which examined a group of 59 patients with TGA and intact ventricular septum, the authors showed that longer preoperative PGE1 exposure did not affect ICU or total hospital stay [23]. On the contrary, Beattie et al. [24] reported a positive association between need for PGE1 after BAS and postoperative length of stay. However, this was a small study of 26 patients and only examined patients with TGA that underwent BAS. In this study, 11 out of 26 patients required re-initiation of PGE1 after a successful septostomy, and when compared to the group who did not (8 out of 26 patients), they were found to have no difference on size of atrial communication and significantly lower systemic saturation (70.9 vs. 82.6 %) at the time of ASO despite the infusion of prostaglandins. Furthermore, 4 out of these 11 babies were found on echocardiography to have widely patent ductus arteriosus prior to re-initiation of PGE1. The authors attribute the findings to the effect of PGE1 on the pulmonary vasculature [25]. In our opinion, the findings of these studies can be constellated. Most babies with a widely patent atrial communication with or without a BAS can be weaned off PGE1. We continue, therefore, to aim to discontinue PGE1 in babies with TGA after BAS whenever possible. Inability to wean PGE1 signifies unstable hemodynamics and, in our practice, triggers efforts to categorize the patient into the small subset of patients with TGA and reactive pulmonary vasculature. For these babies, increased awareness is warranted given the association with less a favorable perioperative course [26]. In fact, in our cohort, our only mortality was a patient with pulmonary hypertension after ASO whom we were unable to wean off mechanical circulatory support.

In our study, there was no association with BAS and prolonged postoperative length of stay similarly to findings by Dibardino et al. [1]. Our strategy to perform BAS in most patients unless there was clearly unobstructed atrial communication may have masked a potentially beneficial effect to a subgroup of patients with borderline atrial restriction as reported by other investigators [27].

This study has several limitations. First, we did not describe postoperative residual anatomical lesions, low cardiac output state, and other postoperative factors that might have affected length of stay. Several reports have been published exploring these associations [3, 9, 11–13]. Furthermore, the cardiopulmonary bypass time was statistically longer in the LL stay group. This association has been previously described in the literature [28]. Our intention was to focus on preoperative management and its impact onto postoperative length of stay. Second, as we have already mentioned, we cannot exclude the possibility that our groups had different hemodynamic profiles which can be directly related to the ability to extubate and initiate enteral feeds. Confirmation of our findings by further studies is required before changes into clinical practice are to be implemented.

Finally, this is a retrospective study with a long study period with potential changes in surgical era and postoperative ICU management. However, our surgical and ICU teams remained unchanged during the study period. Furthermore, studying the entire cohort of all patients undergoing ASO, we demonstrated that there was no difference in postoperative length of stay between the first and second half of the study period.

Conclusion

Our study indicates that the patient’s preoperative condition, specifically the need for mechanical ventilation and ability to receive enteral feeding, impacts postoperative recovery. Patients who are intubated and have not yet begun to receive enteral feeds at the time of their ASO are more likely to have prolonged postoperative stays. It is unclear if prolonged stays were a result of operating on patients with worse preoperative hemodynamics or a consequence of a preoperative management strategy that did not allow for extubation and establishment of feeds prior to surgery. It follows, therefore, that aggressive measures to optimize preoperative condition such as attention to pulmonary toilet, weaning mechanical ventilation, and improving nutritional and feeding status may reduce time of postoperative recovery. Further prospective research is warranted to investigate these associations in patients with similar baseline hemodynamics undergoing ASO.