Discussion

In this single‑centre study, tracheostomy was infrequently required after paediatric cardiac surgery and was predominantly used in children with prolonged and complex postoperative recovery. Our findings support the view that the need for tracheostomy in this population is more closely associated with overall disease severity and postoperative cardiopulmonary morbidity than with an isolated airway disorder. Importantly, no deaths were directly attributable to tracheostomy, and no major procedure‑related complications were observed.

The reported incidence of tracheostomy after paediatric cardiac surgery ranges from approximately 0.2% to 2.7%. ^{Reference Hoskote, Cohen, Goldman and Shekerdemian5,Reference Mastropietro, Benneyworth and Turrentine9} Previous single‑centre studies have reported rates of 1.3–2.7%, with tracheostomy generally performed after several weeks of mechanical ventilation. ^{Reference Hoskote, Cohen, Goldman and Shekerdemian5–Reference Saygı, Öztürk and Özden7} Hoskote et al. reported that cardiac failure, tracheobronchomalacia, and diaphragmatic paralysis frequently accompanied the need for tracheostomy, whereas Challapudi et al. identified ventilator‑dependent respiratory failure and tracheomalacia as common indications. ^{Reference Hoskote, Cohen, Goldman and Shekerdemian5,Reference Challapudi, Natarajan and Aggarwal8} Thus, our tracheostomy rate of 1.8% and median timing of 28.5 postoperative days are consistent with previously reported ranges.

Younger age was independently associated with the need for tracheostomy in our cohort. Haemofiltration was also independently associated with tracheostomy; however, this association likely reflects greater postoperative illness severity rather than an independent clinical predictor. Similarly, the associations of extracorporeal membrane oxygenation support and open sternum/delayed sternal closure observed in univariable analyses probably reflect the severity and complexity of the postoperative course. Because only 34 patients required tracheostomy, the number of variables that could be reliably evaluated in multivariable analysis was limited, and these associations should therefore be interpreted cautiously.

Single‑ventricle physiology was present in approximately 30% of the tracheostomy cohort. Although it was not significantly associated with the need for tracheostomy in our analysis, this physiology may contribute to a more complicated postoperative respiratory course. Previous studies have likewise demonstrated worse outcomes in selected children with single‑ventricle physiology or systemic‑to‑pulmonary shunts. ^{Reference Benneyworth, Shao and Cristea10,Reference Ortmann, Manimtim and Lachica12} Thus, the absence of a significant association in our cohort should not be interpreted as evidence that cardiac physiology lacks clinical relevance, particularly given the limited number of tracheostomy events.

Phrenic nerve injury represents another important potential contributor to prolonged ventilation after congenital cardiac surgery. Three patients who underwent tracheostomy in our cohort had diaphragmatic elevation secondary to phrenic nerve paralysis, including two with unilateral and one with bilateral paralysis. All underwent diaphragmatic plication before the subsequent decision to proceed with tracheostomy. In the entire cardiac surgical cohort, phrenic nerve paralysis was identified in 18 of 1,840 patients (0.97%). This finding suggests that diaphragmatic dysfunction contributed to prolonged respiratory support in a small subgroup, although it did not alone explain the subsequent need for tracheostomy. Tracheomalacia and subglottic stenosis were identified in a small number of patients, whereas no vocal cord paralysis was observed. In most other patients, recurrent extubation failure occurred in the context of cardiac complexity, prolonged ventilation, respiratory infection, and secretion burden, without a single structural airway abnormality accounting for the clinical course.

Extracorporeal membrane oxygenation warrants particular consideration because it reflects severe postoperative illness and may also increase procedural bleeding risk. All extracorporeal membrane oxygenation-supported patients in our tracheostomy cohort received central veno‑arterial-extracorporeal membrane oxygenation, and tracheostomy was not performed during extracorporeal membrane oxygenation support; all procedures were undertaken after extracorporeal membrane oxygenation discontinuation. Accordingly, the increased bleeding risk associated with tracheostomy during active extracorporeal membrane oxygenation support does not directly apply to our cohort. Large cardiac surgical datasets have likewise identified extracorporeal membrane oxygenation and other markers of postoperative severity as important predictors of adverse outcomes. ^{Reference Mastropietro, Benneyworth and Turrentine9–Reference Johnson, Marino, Klugman and Shamszad11}

The overall mortality observed among children who underwent tracheostomy should also be interpreted in the context of underlying disease severity. Contemporary congenital heart surgery series and national database analyses have similarly shown that tracheostomy is associated with prolonged hospitalisation, high resource utilisation, and increased in‑hospital mortality. ^{Reference Costello, Emerson and Shu16,Reference Maxwell and McMillan17} Paediatric tracheostomy studies have also demonstrated that overall mortality substantially exceeds mortality directly attributable to the tracheostomy procedure itself. ^{Reference Roberts, Powell and Begbie1–Reference Campisi and Forte4,Reference Funamura, Yuen and Kawai18,Reference Davidson, Jacob and Brown19} In our cohort, no deaths were directly related to tracheostomy. Minor complications were limited to one local wound infection that resolved with daily wound care and minimal granulation tissue in two patients treated with silver nitrate cauterisation. No accidental decannulation or cannula obstruction requiring intervention was observed. Thus, the rate of procedure‑related complications was low in this cohort; however, the small number of procedures precludes broad conclusions regarding safety.

The high frequency of documented extubation/reintubation episodes before tracheostomy further illustrates the prolonged and challenging respiratory course of these patients. At least one such episode was documented in 30 of 34 patients (88.2%). Recurrent extubation failure frequently coincided with prolonged mechanical ventilation; however, the proportion of patients with documented extubation/reintubation episodes should not be interpreted as indicating that recurrent extubation failure was the sole indication for tracheostomy in 88.2% of patients.

One notable finding was the high decannulation rate among survivors. Twenty‑four of the 27 survivors were ultimately decannulated, at a median of 62 days after tracheostomy. This rate is higher than those reported in several previous paediatric and congenital cardiac series, in which decannulation rates have generally been substantially lower. ^{Reference Roberts, Powell and Begbie1,Reference Sachdev, Chaudhari and Singh3,Reference Puchi, Lavin, Eltayeb and Billings13,Reference Çiçek and Özdemir14,Reference Costello, Emerson and Shu16} Costello et al. reported prolonged ventilator dependence after tracheostomy in children following congenital heart surgery. ^{Reference Costello, Emerson and Shu16} However, direct comparisons with these studies require caution because institutional thresholds for tracheostomy, patient selection, underlying cardiac disease, and discharge practices differ substantially.

An important characteristic of our cohort may help explain the high decannulation rate. Thirty-two of the 34 patients who underwent tracheostomy were international patients, and many families preferred not to return to their home countries with a tracheostomy in place. Consequently, some clinically stable patients remained in hospital-based accommodation and under follow-up until decannulation could be achieved. This institutional discharge practice likely contributed to the unusually high decannulation rate in our cohort and limits direct comparison with centres where children are routinely discharged home with a tracheostomy or long-term ventilatory support. Two survivors in our cohort were discharged with home mechanical ventilation, whereas one was discharged with a tracheostomy in place.

Thus, the high decannulation rate should not be interpreted as evidence that tracheostomy facilitated ventilator weaning or improved outcomes. In the absence of a comparable control group of similarly ill patients managed without tracheostomy, it is not possible to determine whether these patients would ultimately have been successfully weaned without the procedure. Our findings instead suggest that, within our institutional practice, tracheostomy served as temporary supportive airway management in selected children during prolonged postoperative recovery. This distinction is particularly important because the retrospective design does not permit causal conclusions regarding the effects of tracheostomy on ventilation duration, sedation requirements, length of stay, or survival.

The optimal timing of tracheostomy after paediatric cardiac surgery remains uncertain. Haemodynamic instability, bleeding risk, infection, organ dysfunction, the possibility of respiratory recovery, and family preferences may all delay decision‑making. ^{Reference Saygı, Öztürk and Özden7,Reference Johnson, Marino, Klugman and Shamszad11} Our institutional approach is individualised and multidisciplinary, with timing determined by the clinical course rather than by a fixed postoperative day. Although tracheostomy is generally considered after 7–10 days of prolonged ventilation, the procedure may occur considerably later depending on clinical stability, extubation attempts, secretion burden, airway findings, and the anticipated course of recovery. Day et al. similarly described tracheostomy as a supportive management option in selected children with complex cardiac disease. ^{Reference Day, Cross, Crossland, Powell and Simpson15} Thus, the median postoperative timing of 28.5 days in our cohort reflects a selective approach and does not represent an institutional policy of early tracheostomy.

This study has several limitations. Its retrospective, single‑centre design limits generalisability, and the small number of tracheostomy events restricts statistical power and the number of variables that can be included in multivariable analyses. Mortality analyses are particularly limited by the occurrence of only seven deaths, resulting in wide confidence intervals and requiring cautious, exploratory interpretation. We did not perform propensity matching or compare sedation duration, ventilator duration, or other outcomes with those of a matched non‑tracheostomy cohort; thus, no causal effect of tracheostomy on these outcomes can be inferred. Although extubation/reintubation episodes could be retrieved from patient records, prolonged mechanical ventilation, recurrent extubation failure, and the need for a secure long‑term airway were not prospectively recorded as mutually exclusive primary indications. Thus, reliable percentages for each individual primary indication could not be determined retrospectively. Long‑term quality of life, neurodevelopmental outcomes, and hospital readmissions were also beyond the scope of this study.