A review of medical records was approved by the hospital committee on clinical investigation and publication was approved by the institutional ethics committee chair (protocol number 4482/AO/18). All patients in our center who underwent a surgical procedure for CHD between February 1996 and March 2019 via MICS were included.

Patients requiring a surgical procedure for CHD repair are regularly evaluated for MICS in our institution. A thorough discussion is held with patients and families on the various surgical approaches available and a specific informed consent process is followed if a minimally invasive strategy is chosen.

The surgical approach is chosen based on numerous demographic characteristics including age, sex and patient preferences. When feasible, we also employ a “sex-differentiated approach” in which we attempt to honor the preferences of the different genders based on cosmetic considerations [8].

Patients are all managed under the same postoperative protocol. This includes early extubation (<6 hours after surgery) and early discharge from the intensive care unit (ICU <24 hours) when appropriate. Postoperative pain is controlled by intravenous nonsteroidal anti-inflammatory medications (ketorolac 0.2 mg/kg) every 6 hours. Upon discharge, all patients are then followed up at 1 month, 1 year, and 3 years after surgery, at which point an assessment of cosmesis is included in the clinical evaluation. In patients who underwent mini-thoracotomies, the postoperative evaluations include a specific assessment of the presence of scoliosis, any limitations of shoulder mobility, and any sensory deficits in mammary development.

Cardiopulmonary bypass (CPB) via direct aortic cannulation and bicaval cannulation under mild hypothermia (rectal temperature 34°C) is the standard, and mainly used in small infants. Vacuum-assisted venous drainage (between –30 and –50 mmHg maximum pressure) is routinely utilized to reduce the caliber of the venous cannula that is necessary. In some cases, the inferior vena cava cannula is inserted into the chest through a separate 5 mm skin incision below the primary incision. This site is then used for chest drainage at the end of the operation.

In patients with a body weight above 15 kg a remote cardiopulmonary bypass strategy with peripheral cannulation is used. In such cases, arterial and venous femoral vessels are surgically isolated through a 1.5 to 2 cm incision in the inguinal fold. The right common femoral artery is cannulated directly while the femoral vein is accessed under 2D-echo guidance with the Seldinger technique to position the tip of the venous cannula at the inferior vena cavo-atrial junction. The venous return of the upper part of the body is generally drained by cannulating the superior vena cava. The internal jugular vein is the preferred access site for venous drainage from the upper body in case of partial anomalous pulmonary venous return (PAPVR) [12]. Near-infrared spectroscopy with a cerebral oximeter (INVOS 5100, Medtronic, Dublin, Ireland) is used to continuously monitor oxygen saturation in the lower extremities during CPB, when peripheral arterial cannulation is used.

Intracardiac repair in patients with atrial septal defect (ASD) is achieved in simple CHD by using induced ventricular fibrillation (IVF) with an epicardial lead. In more complex procedures aortic cross-clamping (CC) with cardioplegic arrest if preferred.

All patients weighing more than 5 kg are monitored with transesophageal echocardiography, particularly when attempting to wean from CPB, to visualize any retained air in the left side of the heart. For patients weighing less than 5 kg, intraoperative epicardial echocardiography is performed.

MS, RAMT, and RLMT are used variably depending on the anatomic lesions and can be employed for the surgical correction of simple CHDs such as ASD, ventricular septal defects (VSD), or PAPVR. Generally, MS or RAMT, are used for more complex CHD lesions such as partial atrioventricular septal defects (pAVSD) or procedures on the left ventricular outflow tract.

Continuous variables are expressed as median ± interquartile range (IQR), Categorical variables are presented as absolute and relative frequencies (%). All continuous variables were normally distributed as tested by Shapiro-Wilk test and graphically represented by histograms and Q-Q plots. Variables were analyzed with regard to the surgical access (MS, RAMT and RLMT). Between-group comparisons were performed using one-way analysis of variance (ANOVA) followed by Bonferroni’s correction for continuous variables. Fisher’s exact test or Pearson’s chi-square were used for nominal variables. Multivariate linear or logistic regression was used to assess the correlation between possible risk factors and the continuous or binomial dependent variable, respectively. The level of significance was set at p-value ≤ 0.05. Statistical analyses were performed with STATA 15.0 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC).

A total of 1,002 patients were included. Of these, 455 (45%) underwent MS, 356 (36%) underwent RAMT and 191 (19%) underwent RLMT (Fig. 1). No case required conversion to full median sternotomy (FMS). Genders were equally distributed in the MS and RLMT group (54% and 50%, respectively), while females were predominant in the RAMT group (85%) (p < 0.001). Median age at surgery was 2 years (IQR 0.4–5) in the MS group, and 8 years both for RAMT (IQR 4–20) and RLMT (IQR 5–14) (p < 0.001). Weight at surgery was 11 kg in the MS group (IQR 6–18), 27 kg in the RAMT group (IQR 16–55) and 28 kg in the RLMT group (IQR 18–52) (p < 0.001) (Tab. 1).

The percentage of minimally invasive procedures in our center (based on the total number of major surgeries) was 16.1% during the first decade (1996–2007) and 22.3% during the second part of the study (2008–2019). Between 1996–1997, 55% of ASD closures were performed through MICS. Since 1998 all ASD patients have been treated with MICS.

MS was the MICS incision of choice until 2009 and was particularly dominant during the period 2005–2009 (Fig. 2). RAMT was equally distributed throughout the study period. RLMT has become increasingly more prevalent approach since 2015. There was a significant correlation between the type of surgical access and the year in which the surgery was performed (p < 0.001).

Atrial septal defect was the most common diagnosis in the RAMT and RLMT groups (338/356, 95% and 97/191, 51%, respectively) (p < 0.001). VSD was most common (48%) in the MS group. As for complex CHD, pAVSD was prevalent in less than 10% of the population regardless of the surgical access, whereas PAPVR was present in 50/191 (26%) of the RLMT cases (Tab. 1). Other diagnoses were less commonly approached via MICS (Fig. 1S, Appendix A).

Central cannulation was commonly used in MS and RAMT (438/455, 96%; 253/356, 71%, respectively), while peripheral cannulation was utilized in the majority of the RLMT group (183/191, 96%) (p < 0.001). Median CC time ranged between 37 min to 48 min in the three groups, and time of IVF ranged between 14 min to 17 min (p < 0.001). Blood transfusions were required in 210/455 (46%), 37/356 (10%) and 8/191 (4%) patients in the MS, RAMT and RLMT groups, respectively (p < 0.001). Median ICU stay was 1 day for RAMT (IQR 1–1) and RLMT (IQR 1–1), and 2 days for MS (IQR 1–2) (p < 0.001). Median length of stay was 5 days for the RAMT (IQR 5–6) and MS (IQR 4–7) and was 4 days for RLMT (IQR 4–5) (p < 0.001) (Tab. 1).

No peri-operative or in-hospital deaths occurred. Overall, 111 patients (11%) had at least one postoperative complication, with post-cardiotomy syndrome being the most common (40/1,002, 3.9%). Temporary arrhythmia was the second most common complication (2.6%) and was most common in the RLMT group (4.7%). Other minor complications included pneumonia, pneumothorax, postoperative bleeding requiring re-operation, mechanical ventilatory support for more than 7 days, neurological complications, wound dehiscence and low cardiac output syndrome, each having an incidence lower than 1% (Tab. 2).

In order to control for any confounding effect of the different diagnoses and confirm our results, a multivariate logistic regression analysis of only patients who underwent ASD closure was performed (Tab. 3). No difference in the percentage of postoperative complications was noted between MS, RAMT and RLMT when adjusted for age and gender. Similarly, no difference was found in the CC time. CPB time was shorter in RLMT than in the MS and RAMT (p = 0.015 and p = 0.001, respectively).

Meta-analysis of the existing literature demonstrated that our cohort of ASD patients had a shorter CPB time (Hedge’s g: 1.96; 95% CI: 1.10–2.81, p < 0.01), IVF time (Hedge’s g: 2.31; 95% CI: 0.25–4.38, p = 0.03) and ICU stay (Hedge’s g: 1.32; 95% CI: 0.00–2.64, p = 0.05) (Fig. 3). No differences were noted in CC time, post-operative complications and pericardial effusions (Fig. 2S, Appendix A).

This study has several limitations. First, it is a retrospective examination of a heterogeneous cohort of patients with multiple diagnoses who underwent different procedures at different ages and at different times in our institutional experience. The population of patients treated with RLMT is proportionately smaller and only short-term results were analyzed. This limits the ability to draw definitive conclusions about long-term efficacy and safety. The population of patients treated with FMS was derived from a historical cohort from previous studies and this may have determined an era effect. However, previously published data from our experience on this topic reveal no significant complications such as arrhythmias at midterm follow-up [29].