Minimally invasive surgery (MIS) for infants and children continues to grow. MIS was first introduced for the treatment of congenital diaphragmatic hernia (CDH) in 1995; Silen et al. [1] used thoracoscopy, while van der Zee and Bax [2] used laparoscopy. However, most pediatric surgeons have hesitated to apply MIS to CDH because of the associated fragile respiratory status and pulmonary hypertension (PH). Nevertheless, thoracoscopic repair under high-frequency oscillatory ventilation (HFOV) [3] or after extracorporeal membrane oxygenation (ECMO) therapy [4] have been reported recently, and MIS appears to be gaining added acceptance for CDH repair without any fixed selection criteria being established [47].

We introduced thoracoscopic CDH repair (TR) for CDH in 2007 and reviewed our patients to define safe indications for TR use because to date there are no established indications.

Patients and methods

Treatment protocol and selection criteria for thoracoscopic CDH repair

We reviewed 26 consecutive CDH patients diagnosed prenatally or in the immediate postnatal period (i.e., less than 6 h after birth) who were treated at our institution between January 2007 and December 2009. Of these, five preoperative deaths were excluded.

All patients were managed according to a standard protocol described previously [8, 9]. Briefly, immediately after birth or after admission to the neonatal intensive care unit (NICU), the infant was intubated and artificially ventilated with HFOV (Humming V®, Metran Inc., Japan). Initial HFOV settings were FiO2, 1.0; frequency, 15 Hz; stroke volume (SV), 15 mL; and mean airway pressure (MAP), 15 cmH2O. Nitric oxide (NO) was introduced if post-ductal SpO2 was less than 75% or echocardiography (EC) showed PH. HFOV settings were monitored to maintain maximum SV of 10 mL/kg and maximum MAP of 25 cmH2O. Fentanyl (5–10 μg/kg per h) was used for continuous deep sedation and vasopressors (dopamine 5 μg/kg per min and dobutamine 5–10 μg/kg per min) were used for maintaining systemic blood pressure above 50 mmHg. HFOV settings were decreased gradually to maintain post-ductal SpO2 at more than 75%. HFOV was converted to conventional mechanical ventilation (CMV) when FiO2 < 0.4, SV < 4 mL/kg, and MAP was 12 mmHg. Initial settings for CMV were FiO2, 0.4–0.6; peak inspiratory pressure, 15–20 mmHg; positive end-expiratory pressure (PEEP), 5 mmHg; and respiratory rate, 40/min. Surgery was performed once there was spontaneous closure of the patent ductus arteriosus (PDA) or dominant left-to-right shunting, and marked increase in ipsilateral pulmonary arterial blood flow confirmed by EC. TR was indicated if patients met both the following selection criteria.

  1. 1.

    Cardiopulmonary status stable for more than 10 min in the decubitus position in NICU under CMV or HFOV with/without NO as a marker for tolerance of surgery.

  2. 2.

    Tolerance of manual ventilation with/without NO to allow transfer to the operating room.

Patients who did not satisfy both these criteria had conventional open repair (OR) in NICU.

Thoracoscopic CDH repair

Patients were placed in the decubitus position. Single-lung ventilation was not required. A 5 mm port was placed initially in the anterior axillary line in the 4th intercostal space for the scope. Insufflation pressure was initiated at 4 mmHg, then increased up to 6–8 mmHg. Two additional 5 mm working ports were placed under direct vision; one below the tip of the scapula and one in the mid-clavicular line at the level of the first port. Herniated organs could be reduced with relative ease using insufflation pressure of 6–8 mmHg and thoracoscopic instruments. After reduction of hernia contents, insufflation pressure was decreased to 5–6 mmHg, and primary repair was performed with interrupted 3-0 Ethibond® (Ethicon Inc., Japan). No drain was placed.

Data collection and statistical analysis

Patient demographic data including gender, gestational age at birth, birth weight and side of defect (left/right) were obtained from medical records. Diagnosis (prenatal/postnatal) and severity as represented by requirement for HFOV and NO, and contents of herniated organs (i.e., liver, stomach) were also recorded. Operation data including age at repair and duration of surgery (operative time) were noted. Outcome variables were compared according to actual survival rates, number of days postoperative ventilation was required, type of complications, and recurrence.

Categorical data were analyzed using the chi-squared test. Continuous data were analyzed using the Student or Welch t test. Data were represented as mean ± SD and p < 0.05 was considered significant.


There were 13/21 OR cases (62%) and 8/21 TR cases (38%). All OR cases did not satisfy our selection criteria completely: actually, all 13 did not tolerate manual ventilation, while 2 did tolerate the decubitus position. Patient demographics are summarized in Table 1. Gender, gestational age at birth, birth weight, and side of defect were similar between 2 groups. Prenatal diagnosis was significantly less in TR (3/8 vs. 11/13 or 37.5 vs. 84.6%, p = 0.026). Table 2 shows clinical data related to severity of CDH. HFOV was required in 4 TR patients (50%) and all OR patients on day 0 of life (p = 0.004), and in 3 TR patients (38%) and in all OR patients intraoperatively (p = 0.001); both were statistically significant. NO requirement was also significantly less in TR patients; 2 TR patients (25%) versus 10 OR patients (77%) on day 0 (p = 0.019), and 1 TR patient (13%) versus 10 OR patients (77%) (p = 0.004) intraoperatively. In 3 TR patients who required HFOV intraoperatively, preoperative HFOV settings were FiO2, 0.6–0.7; SV, 15–17 mL; and MAP, 15 mmHg. No TR patients who required HFOV had cardiopulmonary compromise intraoperatively. Stomach and liver herniation were significantly less in TR patients than OR patients. TR patients had surgery significantly earlier than OR patients (1.43 ± 0.79 vs. 2.46 ± 1.67 days of life, p = 0.027). Three TR patients were converted to OR because of bleeding from short gastric vessels in 1, and technical difficulty due to antero-lateral diaphragmatic defect in 2. Mean operation times were 164 ± 78 min in TR and 143 ± 22 min in OR, which were not statistically significant. Synthetic patch repair was not required in any patient. There was only one postoperative complication in a TR patient who developed pneumothorax soon after TR and required chest-tube placement. Recurrence of hernia occurred in only one TR patient 20 days after TR, which was corrected by OR. One OR patient died postoperatively due to deteriorating PH.

Table 1 Patients’ demographics in TR and OR patients
Table 2 Clinical data of TR and OR patients


In order to perform TR safely for CDH, we reviewed our successfully treated CDH patients to assess the validity of our selection criteria. In our protocol for CDH management [8, 9], timing of surgery is determined by spontaneous closure of the PDA or when shunting becomes left-to-right dominant, and there is marked increase in ipsilateral pulmonary arterial blood flow confirmed by EC, all of which indicate stabilization of PH. With this in mind, we chose our selection criteria based on factors most likely to influence the success of surgery itself and reflect cardiopulmonary stability. Cardiopulmonary instability in the decubitus position in NICU would suggest that the operative position for TR cannot be tolerated, and inability to use manual ventilation would suggest that the patient cannot be transferred to the operating room; thus unless both are satisfied, TR would be contraindicated. In support of this, none of the conversions to OR were for cardiopulmonary compromise, so we feel confident that the logic behind our selection criteria is sound. A combination of our CDH protocol [8, 9] and selection criteria would appear to be reliable indicators for TR.

There are few reports about indications for TR. Yang et al. [5] proposed selection criteria for TR from anatomic and physiologic aspects: (1) stomach in the abdomen, (2) minimal ventilator support with low peak inspiratory pressure (<24 mmHg), and (3) no evidence of PH at the time of surgery. In their series, only one patient required NO, but was weaned preoperatively. Similar criteria were also proposed by Guner et al. [6]. Although no cases required conversion in their series, from our experience, stomach herniation and NO use are not contraindications to TR if PH is controlled even though HFOV may be required. Kim et al. [4] also attempted to establish criteria for TR based on preoperative cardiopulmonary stability during gentle ventilation in the absence of evidence of PH. Although there were two preoperative ECMO patients, ECMO was not considered a contraindication to TR.

Physiologically, preoperative cardiopulmonary stability can only be achieved by using HFOV, NO, and ECMO in CDH accompanied with severe PH. Surprisingly, Cho et al. [7] used TR to treat unselected CDH patients, and in a consecutive series, only 1 of 29 patients was converted to OR and it was for inability to reduce the liver. They used ECMO in 6.9% of patients for stabilization, but none of their patients required HFOV and NO intraoperatively during TR. We can only guess our patients were more unstable because more than half our patients had OR, and it is difficult to imagine most patients could be treated using TR without HFOV or NO.

Conversion rates in the recent literature range from 3.4 to 20% [4, 7, 10, 11]. In our study, 3 of 8 patients (37.5%) were converted to OR, a rate that is higher than found in the literature. All three were early TR cases and probably reflect our limited experience. Requirement for synthetic patch repair is a valid reason for conversion to OR usually because of technical difficulty and more serious PH [4, 5, 12]. But in our study none of our patients was converted to OR because of a large hernia defect. In fact, patients with large defects were found not to satisfy both criteria for TR, an indication of the possible superiority of our selection criteria. For large defects we prefer OR because TR has technical limitations and use Toldt’s fascia flap repair [13] rather than patch repair.

TR is a feasible and safe surgical approach in selected patients. Further investigation of the surgical stress associated with TR and the results of long-term follow-up are warranted for thoroughness. A trial of TR in CDH patients failing to meet our selection criteria could expand and qualify our indications for TR and further contribute to the establishment of definitive indications for TR; in particular, simply extending the preoperative stabilization period could allow TR to be used more often.