Introduction

Unilateral phrenic nerve injury is a known complication of pediatric cardiac surgery [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. The overall incidence is approx. 2%, and higher for complex operations performed in early infancy [9, 10]. However, using direct percutaneous stimulation of the phrenic nerve before and after surgery Mok et al. [7] demonstrated an incidence of 10%. The weakness of the intercostal musculature and the horizontal orientation of the rib cage aggravates the paradoxical motion of the paralyzed diaphragm in children [11, 12]. Plication of the diaphragm helps facilitate weaning from mechanical ventilation [3, 4, 5, 6, 8, 9, 10]. Bilateral diaphragm paralysis (BDP) is very rare and causes thoracoabdominal asynchrony. Only a few case reports have been published [13, 14, 15, 16, 17, 18]; the earlier ones recommend the use of tracheostomy [17, 18]. To our knowledge, there are no relatively large series of bilateral diaphragm paralysis. Over the past 10 years our team treated nine patients with bilateral diaphragm paralysis [19]. This retrospective analysis highlights the early recovery of the diaphragm in some patients and the effectiveness of conservative management.

Materials and methods

Study population

Between January 1995 and December 2004 a total of 3,214 children underwent cardiac surgery at the Schneider Children's Medical Center of Israel. One-half were under 1 year old, including 514 neonates (16%) and 1,093 children aged 1 month to 1 year. Cardiopulmonary bypass was performed in 2,731 patients (85%) and closed heart surgery in 483. BDP developed in nine patients (0.28%), who constituted our study group (age range was 12 days–16 months). BDP occurred in 0.77% of the neonates in this group (4/514), 0.36% of the children aged 1 month–1 year (4/1093) and 0.062% of those aged over 1 year (1/1607). Patients' characteristics are summarized in Table 1. The incidence of BDP among children requiring cardiopulmonary bypass was 0.25% (6/2,731), and that among children requiring closed heart surgery 0.41% (2/483). The time from surgery to diagnosis was 3–10 days (median 3). One patient (no. 2) required one operative revision for bleeding, and one (no. 4) had two revisions; in patient 2 the phrenic nerve was permanently damaged. One patient (no. 9) required three operative revisions to restore blood flow to the right pulmonary artery.

Table 1 Preoperative and operative data (TOF tetralogy of Fallot, VSD ventricular septal defect, TAPVD total anomalous pulmonary venous drainage, BDP bilateral diaphragm paralysis, RPA right pulmonary artery)
Full size table

Data were collected from the departmental computerized database which includes daily follow-up. The variables examined were type of cardiac surgery, complications during surgery, interval from surgery to diagnosis of BDP, inotropic support, duration of mechanical ventilation, and complications in the intensive care unit (ICU).

Definitions

  • Diaphragm paralysis: BDP was suspected in the presence of thoracoabdominal asynchrony, i.e., movement of the abdomen relative to the chest during inspiration. The diagnosis was made by ultrasonography [20] and confirmed by fluoroscopy.

  • Inotropic support provided during the 24 h following surgery was categorized according to Dagan et al. [21], as follows:

    • Grade 3 (high inotropic support): Any dose of adrenaline, dopamine, or dobutamine at a dose of 15 μg/kg per minute or higher, or both dopamine and dobutamine at a dose of 10 μg/kg minute or higher.

    • Grade 2 (medium): Dopamine or dobutamine alone at a dose of 6–14 μg/kg per minute, or dopamine and dobutamine together at a dose below 10 μg/kg per minute.

    • Grade 1 (low): Dopamine or dobutamine alone at a dose of 5 μg/kg per minute or less.

  • Bleeding following surgery was considered significant if surgical revision was required.

  • Time to diagnosis: The period between performance of surgery and diagnosis of BDP.

  • Bacteremia: Positive blood culture with clinical signs of infection.

  • Line infection: Positive blood culture with fever, and cessation of fever after line removal.

Management protocol

Patients with fluoroscopy-confirmed BDP were managed with the following stepwise protocol:

  • Maintain mechanical ventilation to keep the child comfortable and use of hypnotic agents, if necessary.

  • Institute early feeding by nasogastric tube.

  • Discontinue all invasive lines and maintain one long peripheral line.

  • Conduct a continuous positive airway pressure (CPAP) trial every 24 h; confirm clinical findings o diaphragm movement by ultrasound.

  • If the diaphragm starts to recover, start a weaning process by increasing the time of CPAP and rest with ventilatory support; start with short periods of 20 min, and increase the duration of CPAP every 4 h to up to 4–6 h before extubation.

  • If only one side of the diaphragm recovers, perform plication of the other.

  • After extubation, use noninvasive mechanical ventilation if there is increased respiratory distress.

Results

Table 2 summarized the ICU course of the patients. All patients were placed on mechanical ventilation for 14–62 days (median 25). The duration of ICU stay was 20–70 days (median 30). Diaphragms started to recover after 12 days–6 weeks. In 6 patients both hemidiaphragms started to recover at almost the same time. The other three patients had plication of one hemidiaphragm because of either surgical trauma (patient no. 4) or delay in recovery (nos. 3 and 9). In patient 4 the need for repeated plication increased the ventilation time. Six patients had seven episodes of infection, including three patients with a line infection. All patients were extubated and discharged home without the need for respiratory support. The overall stay in hospital was 21–150 days (Table 2). Only one patient (no. 3) had minor subglottic stenosis due to prolonged stridor after reintubation, diagnosed by bronchoscopy. Patient no. 4 died 3 years later of cardiac problems.

Table 2 Postoperative data (LOHS length of hospital stay, UTI urinary tract infection; inatropic support: 1 use of dopamine or dobutamine ≤ 5 μg kg− min−, 2 a combination of the above, 3 use of any dose of adrenaline, dopamine or dobutamine ≥ 15 μg kg− min−, or both dopamine and dobutamine at a dose of ≥ 10 μg kg−,min−1)
Full size table

Evaluation of the data in Table 2 suggests the presence of two patient subgroups. Postoperatively, patient nos. 2, 3, 4, and 9 were unstable hemodynamically and required high-dose inotropic support, leaving the chest open. The recovery of the diaphragm was relatively late, and the duration of mechanical ventilation was extensive (50–62 days). The other five patients required only low-dose inotropic support. In two of these the sternum was left open as part of our routine policy (for truncus arteriosus repair and total anomalous pulmonary venous drainage) [22]. This less severely ill subgroup required mechanical ventilation for a shorter duration of 14–25 days.

Discussion

Some authors recommend early plication for the treatment of phrenic nerve paralysis following cardiac surgery [3, 5, 6, 8, 9, 10]. A prospective study by Van Onna et al. [10] reported 17 cases of diaphragm paralysis among 867 cardiac surgery patients; one (0.11%) was bilateral, and 14 were treated with plication. The median time to diagnosis was 14 days. In a larger retrospective study from the Hospital for Sick Children in Toronto Leeuw et al. [8] found 170 cases of diaphragm paralysis after 10,395 cardiothoracic operations during a 12-year period, for an annual prevalence of 1.6%. Plication required in 40% of the affected patients, and in only 15 patients did the diaphragm recover by the time of discharge. The paralysis was bilateral in 10 patients (0.1%), but the authors provided no details about these patients other than that they required longer mechanical ventilation, and that “they may benefit from a more aggressive approach which may include earlier diaphragm plication.” Four of the ten patients in their series had diaphragm plication.

Most studies have found the incidence of phrenic nerve paralysis after cardiac surgery to be approx. 2% [9, 10], although the true incidence may be higher. One prospective study using direct percutaneous stimulation of the phrenic nerve found an incidence of 10% [7]. There are two potential reasons for this discrepancy between clinical and neurophysiological diagnoses. First, children over 2 years of age tolerate phrenic nerve injury well [8]. Second, the mechanism of injury varies and can include thermal injury [23] and traction or cutting of the phrenic nerve, creating a wide spectrum of clinical presentations.

Infants with phrenic nerve injury have more complications than older children owing to major differences in the respiratory physiology. Because of their weak intercostal muscles and the horizontal orientation of the ribs infants rely mainly on diaphragmatic excursion for respiration. In unilateral diaphragm paralysis the marked cranial displacement of the flaccid diaphragm in the supine position and the mediastinal shift to the nonaffected side result in a significant reduction in the functional residual capacity of the affected and normal lung (paradoxical movement), leading to alveolar collapse and the formation of atelectasis. In BDP paradoxical movement is absent, and plication can only maintain lung volume. Schoenfeld et al. [13] found no difference between the intra-abdominal and intrathoracic pressure before and after diaphragm plication and concluded that bilateral plication has no effect on diaphragm function. In contrast to unilateral diaphragm paralysis, the patients with bilateral diaphragm paralysis pose a dilemma for the clinician. To our knowledge, our series is the largest describing the clinical course and the possibility of using conservative management.

The experience with BDP comprises only a few case reports [13, 14, 15, 16, 17]. Two of the children had peripheral neuropathy [13] and brainstem hemorrhage [14]. Stewart et al. [17] described their experience with four children following mustard operation. All presented with marked paradoxical movement of the upper abdominal wall, which retracted inward and upward rather than outward during inspiration. Tracheostomy was performed in all cases, and weaning from the ventilator was gradual. Diaphragm recovery time was 4–15 weeks; in two patients only one hemidiaphragm recovered. The authors recommend tracheostomy for BDP for three reasons: to avoid potential catastrophe, to permit early oral intake, and to simplify the weaning process.

In our series of nine patients we opted for a more conservative approach. Our prevalence rate was relatively high: 0.28%, compared to 0.11% and 0.1% in two other series [8, 10]. However, the median time to diagnosis was only 4 days compared to 15 days in the other series. The short period to extubation (less than 20 days following surgery) in more than one-half our patients was probably attributable to their minor degree of phrenic trauma, such as traction. In all these patients both hemidiaphragms started to recover at almost the same time. The other four patients had prolonged mechanical ventilation of 50–62 days. This subgroup had a complicated operative and postoperative course. Two required revision for bleeding with the sternum left open and heavy inotropic support. Plication was carried out in three cases: in one at an early stage due to surgical trauma and in two because only one hemidiaphragm recovered spontaneously. Even in this group, however, recovery had already begun by 7 weeks. Our method of management was based on the observation that function in one diaphragm in children is required to plicate the other. Using our method we gained both diaphragm function and lung function by preventing paradoxical movement. We are not aware of any experience with early plication of both diaphragms.

Mechanical ventilation with translaryngeal airway intubation was not associated with any catastrophic events. We were able to wean the children successfully, and only two required more than one extubation trial. Only one patient had any respiratory complication (mild subglottic stenosis). We used the more conservative approach of mechanical ventilation by nasotracheal tube, taking into account the major complications of short-term tracheostomy.

Hoch et al. [16] recently reported an 8-month-old child with BDP who was treated with noninvasive ventilation. After 24 h the child was able to tolerate the nasal mask, and the respiratory symptoms disappeared. Noninvasive ventilation was commenced after 11 weeks of mechanical ventilation and one diaphragm plication. We used noninvasive mechanical ventilation in three patients, but only when signs of distress developed and only for a transition time of a few days. Two of them tolerated it well, and one patient failed to synchronize with the ventilator. Recently a group from the Children's Hospital of Los Angeles reported a 23-month-old child with multiple extubation trials before diagnosis of BDP [24]. They opted for tracheostomy, and after 7 weeks the diaphragm began to function. An earlier diagnosis of BPD is dependent on familiarity with the clinical course. Earlier recognition can avoid unnecessary extubation.

In conclusion, most cases of BDP last for a relatively short period, with both hemidiaphragms recovering concomitantly. Children with an eventful postoperative course may be expected to require more prolonged mechanical ventilation. In our series maximum recovery was 7 weeks. A well experienced staff can maintain prolonged mechanical ventilation with only minor complications. By using a slow weaning process the need for repeated intubation can be minimized, while still allowing for the possibility of a conservative approach.