Introduction

Acute respiratory distress syndrome (ARDS) in children, although uncommon, can be a significant source of morbidity [7]. Despite the use of new treatments such as controlled-pressure ventilation and inversion of the inspiratory/expiratory ratio, permissive hypercapnia, postural changes, administration of inhaled nitric oxide, and extracorporeal membrane oxygenation (ECMO), the mortality from ARDS still is very high. In addition, children with cardiac disease or dysfunction may be particularly susceptible to deleterious cardiopulmonary interactions induced by positive-pressure ventilation and high mean airway pressure (MAP) [25].

Studies have shown high-frequency oscillatory ventilation (HFOV) to be effective in the management of ARDS [13, 19]. However, few studies have analyzed the efficacy of HFOV for ARDS in corrective cardiac surgery patients.

In our unit, HFOV is used in rescue strategies after the failure of conventional mechanical ventilation (CMV). This retrospective study aimed mainly to evaluate our experience using HFOV for patients with severe ARDS undergoing corrective cardiac surgery. Our study was reviewed and approved by the ethics committee of Fuwai Hospital.

Materials and Methods

Patient Characteristics

Between October 2008 and Augst 2012, nearly 10,843 patients with congenital heart defects (CHDs: 8,217 acyanotic and 2,626 cyanotic) were admitted to our pediatric intensive care unit (PICU) after corrective cardiac surgery. Of the 83 patients who met the criteria for ARDS, 64 treated with HFOV were included in this study.

In our PICU, the diagnosis of ARDS was established according to the criteria defined in 1994 by the American-European Consensus Conference (AECC) [5]. These criteria include acute onset, newly developed bilateral infiltrates shown on chest radiography, partial pressure of oxygen in arterial blood (PaO2)/fraction of inspired oxygen (FiO2) lower than 200, and absence of clinical evidence for left-sided heart failure. Patients with hemodynamically significant residual lesions such as left-to-right shunt, mitral valve regurgitation, and pulmonary venous obstruction were excluded from the study.

Repeated echocardiographic examinations were performed for all the included patients. These patients had no hemodynamically significant residual lesions. During the same period, six patients treated with HFOV were not included in the study. Four of these patients had residual moderate-to-severe mitral valve regurgitation after surgical correction of total endocardial cushion defect and reoperative mitral valve repair. One patient had moderate right upper pulmonary vein stenosis after surgical correction of total anomalous pulmonary venous connection. An additional patient had a residual muscular ventricular septal defect (diameter, 4 mm) after a switch operation and could not undergo a reoperation because of systemic failure.

CMV Strategy

After surgery, patients were admitted to the PICU, where they received conventional mechanical ventilation immediately. All the patients were treated with CMV in the early stages of ARDS. During this period, 64 patients transferred to HFOV after failing CMV were enrolled in this retrospective study. The remaining patients who did not require HFOV survived to discharge. All the patients underwent cardiac surgery with cardiopulmonary bypass (CPB) and experienced ARDS within 30 days of surgery.

All the patients with severe respiratory failure were managed initially with CMV using the Dräger Savina (Drägerwerk AG & Co. KGaA, Lübeck, Germany) or the PB840 (Nellcor Puritan, Bennett, Ireland). Our standard treatment for ARDS was controlled-pressure ventilation with positive end-expiratory pressure (PEEP) that allows the optimum degree of oxygenation, an FiO2 necessary to maintain a saturation higher than 85 % to 90 %, and permissive hypercapnia up to a partial pressure of carbon dioxide (PaCO2) of 55 to 65 mmHg, with maintenance of a pH higher than 7.25 to 7.30.

HFOV Strategy

The patients were converted to HFOV when faced with failure of CMV based on the following criteria: intractable respiratory failure with an OI (FiO2 × MAP [cmH2O] × 100/PaO2 [mmHg]) lower than 13 during at least 6 h or a plateau pressure exceeding 28 cmH2O despite the use of permissive hypercapnia for at least 4 h, intractable hypotension despite maximum support (mean, <60 mmHg for >4 h or <50 mmHg for >1 h), and intractable respiratory acidosis (pH 7.20 at bicarbonate [HCO3] >19 mmol/l for >4 h). High-frequency oscillatory ventilation was performed using the Sensor-Medics 3100A (Viasys Healthcare, Yorba Linda, CA, USA).

The following HFOV settings were used: an FiO2 of 0.4 to 1.0, an oscillatory frequency of 6 to 11 Hz, an inspiration time of 33 % of the respiratory cycle, a bias flow of 30 l/min, a continuous distending pressure of 18 to 35 mbar, and a pressure amplitude (ΔP) of 25 to 50 mbar. Initially, HFOV was started with continuous distending pressure (MAP) at 5 cmH2O higher than MAP on CMV and then adjusted to achieve and maintain optimal lung volume. Therefore, MAP initially was increased until an oxygen (O2) saturation exceeding 95 % was achieved. The frequency was set initially at 9 Hz, with ΔP adjusted according to PaCO2 and chest wall vibrations. If ventilation did not improve despite a maximum ΔP, the frequency could be lowered.

Weaning of HFOV was instigated if PaO2 exceeding 60 mmHg at a FiO2 lower than 0.40 and suction were well tolerated by decreasing ΔP to about 30 cmH2O and MAP to about 15 to 20 cmH2O. Weaning of the HFOV was continued on CMV. A serial chest x-ray (repeated) was used to evaluate or monitor changes.

Muscle relaxants (pipecuronium) could be given in combination with sedation to patients who had acute deterioration of gas exchange during excessive spontaneous activity. Endotracheal suctioning was performed every 4 to 12 h. Inotropes and vasopressors were used to maintain hemodynamic stability.

Data Collection and Definitions

A detailed retrospective review of all medical records was conducted. The data collected included demographics, anatomic diagnosis, perioperative variables (including the risk-adjusted classification for congenital heart surgery-1 [RACHS-1] [18], Aristotle comprehensive complexity score [17, 21, 22], preoperative hemoglobin, pulmonary hypertension, CPB time, and aortic cross-clamp time, central venous pressure [CVP], left atrial pressure, pleural drainage, red blood cell transfusion, plasma transfusion, and vasoactive inotropic score), ventilatory variables (including duration of CMV, duration of HFOV, interval between surgery and HFOV, CVP before and after HFOV, blood gas analysis, blood pressure before and after HFOV, hemodynamics, and ventilator settings), complications, and mortality.

All the patients had CVP monitoring before ICU discharge. Left atrial monitoring lines were used in 54 patients during the early postoperative days.

The definition of recurrent respiratory tract infection (RRTI) was developed by the Society of Pediatrics, Chinese Medical Association [29, 30]. The rule for the diagnosis is presented in Table 1.

Table 1 Rule for the diagnosis
Full size table

Pulmonary hypertension is defined as a mean pulmonary artery pressure of 25 mmHg or higher [3]. Direct pulmonary artery pressure was measured intraoperatively.

The vasoactive-inotropic score (VIS) was calculated as per Gaies et al. [14] as follows:

$$ \begin{aligned} {\text{VIS }} = {\text{ dopamine dose }}\left( {{\text{g}}/{\text{kg}}/{ \hbox{min} }} \right) \, & + {\text{ dobutamine dose }}\left( {{ \lg }/{\text{kg}}/{ \hbox{min} }} \right) \, + { 1}00 \times \\ {\text{epinephrine dose }}\left( {\upmu {\text{g}}/{\text{kg}}/{ \hbox{min} }} \right) \, & + 10 \times {\text{milrinone dose }}\left( {\upmu {\text{g}}/{\text{kg}}/{ \hbox{min} }} \right) \, + \\ 10,000 \times {\text{vasopressin dose }}\left( {{\text{U}}/{\text{kg}}/{ \hbox{min} }} \right) \, & + 100 \times {\text{norepinephrine dose}}\left( {\upmu {\text{g}}/{\text{kg}}/{ \hbox{min} }} \right). \\ \end{aligned} $$
(1)

Pleural drainage was recorded hourly after surgery. Bleeding of less than 1 ml/kg for more than 6 h, chest radiography, and transthoracic echocardiography were considered standard criteria for drain removal.

Patients were followed up until hospital discharge or death and classified as survived or deceased.

Statistical Analysis

Data are described as frequencies and means ± standard deviations (for data with a normal distribution). Continuous variables were compared with parametric (Student’s t test) or nonparametric (Mann–Whitney U test) tests as appropriate. Categorical variables were compared using a two-tailed Fisher’s exact test or a Chi square test.

The variables were subjected to an initial univariate analysis to screen for possible predictors of in-hospital death. To detect independent predictors of in-hospital death, the five variables (identified in the univariate analysis with a p value lower than 0.1) were subjected to multivariate analysis. The multivariate analysis used the stepwise forward Wald multiple logistic regression method to construct a model for predicting the probability of hospital death.

Values of p lower than 0.05 were considered significant. All analyses were conducted using SPSS version 20 (IBM Corporation, New York, NY, USA).

Results

Cardiac lesion types and characteristics of the patients are presented in Tables 2 and 3.

Table 2 Cardiac diagnoses
Full size table
Table 3 Clinical characteristics of patients
Full size table
Table 4 Hemodynamics before and after high-frequency oscillatory ventilation (HFOV)
Full size table

Univariate Analysis of Possible Risk Factors

After a univariate analysis, six variables showed a significant association with mortality. The patients who did not survive had a significantly higher preoperative hemoglobin level, higher CVP, higher incidence of RRTI before operation, and lower improvement of the OI within the first 24 h after initiation of HFOV. They also were more likely to have acyanotic congenital heart diseases and pulmonary hypertension (Table 5). These variables were entered into a logistic regression model.

Table 5 Univariate analysis of possible risk factors for hospital death
Full size table

Multivariate Analysis

The multivariate analysis using logistic regression analysis demonstrated that independent predictors of in-hospital mortality were pulmonary hypertension (odds ratio, 11.52; p = 0) and RRTI before surgery (odds ratio, 3.73; p = 0.048) (Table 6).

Table 6 Logistic regression analysis: independent risk factors associated with in-hospital mortality
Full size table

Discussion

The incidence of postoperative ARDS among patients with CHD is not well documented, but it may be higher than the 0.5 % reported for adult cardiac patients, partly because of longer CPB durations for patients with complicated repairs [2]. The cumulative incidence among ARDS patients treated with HFOV in this study was 0.59 % (64/10843). The overall incidence of postoperative ARDS must have been higher in our unit during this period because the ARDS patients not treated with HFOV were not included in this study (0.77 %, 83/10,843).

Cyanotic patients reflect a more complex form of CHD. This may due to a longer duration of CPB, a longer surgical procedure, and a prolonged mechanical ventilation after surgery for these patients. The higher incidence of ARDS among cyanotic patients is reasonable.

The occurrence of ARDS after cardiac surgery also is unpredictable, and little is known about risk factors for the development especially of this complication in patients. In this study, the systemic inflammatory response activated by CPB was one possible explanation.

Modern anesthetic and surgical techniques allow cardiac surgery to be performed for high-risk patients. Nevertheless, it must be recognized that apart from CPB, these patients also are exposed to other adverse conditions, including blood transfusion and pulmonary hypertension often associated with an increased risk for the development of acute lung injury (ALI)/ARDS [4, 26].

The literature contains relatively few reports about the mortality rate for ARDS after cardiac surgery or about CPB in patients with CHD. The mortality rate for ARDS patients in the general PICU is about 40 % to 70 % [12, 31]. Our results showing the mortality rate in our unit to be 30.1 % (range, 1–69.9 %) were encouraging compared with the published data.

In ARDS, the lung is filled with fluid and becomes stiff, with lung units collapsing. The main purpose of the therapy is to reinflate the collapsed alveoli and keep them open. Positive-pressure mechanical ventilation has been highlighted as a key component of treatment.

The use of ventilators without regard for lung volumes and airway pressures may perpetuate lung injury and contribute to the associated high mortality associated with these clinical conditions. As a consequence, the goal of mechanical ventilation for ALI/ARDS patients has shifted in recent years from maintaining a normal gas exchange to protecting the lung against ventilation-induced lung injury by protective lung ventilation strategies using small tidal volumes with positive end-expiratory pressure (PEEP), limited airway pressure, and low FiO2. Unfortunately, in practice, many critically ill patients with ARDS are unable to achieve oxygenation goals using conventional protective lung approaches [15], and the mortality from ARDS remains unacceptably high [8, 24].

Considering the successful use of HFOV for patients without CHD [27], we became interested in the potential application of this ventilatory mode for cardiac surgery pediatric patients with severe ARDS postoperatively. At least in theory, HFOV is an ideal tool for lung-protective ventilation because it allows effective pulmonary gas exchange with the delivery of a very small tidal volume below dead space and diminished risk of atelectrauma [16, 28] High-frequency oscillatory ventilation was used not as a first-line treatment but as rescue treatment because it was a new mode of ventilation whose safety and effectiveness had not been fully evaluated in this area. This study demonstrated that this treatment was not associated with life-threatening complications and could improve oxygen levels efficiently. One study, by Bojan et al. [6], showed a shorter duration of mechanical ventilation and ICU stay for patients treated with HFOV.

Because of the high MAPs, potential concerns include barotrauma. The incidence of pneumothorax was 34.3 % (22/64) in this study, higher than that reported for adults by Mehta et al. [23]. We postulate that this was primarily due to vulnerability of children’s lungs to high MAPs. Further study is needed to assess their lung function at follow-up assessment.

The main finding of one randomized controlled trial investigating the effect of HFOV on pediatric patient outcome [1] was that HFOV resulted in significantly improved oxygenation but did not significantly improve survival. Considering the rescue use of HFOV, this retrospective study demonstrated a beneficial effect of HFOV on mortality. This may be explained by various factors. First, the knowledge concerning lung-protective ventilation has increased significantly in recent years. Second, the diseases underlying ARDS were different.

Although ECMO is a possible therapeutic option for patients with severe ARDS, no patients were treated with ECMO during this period in our unit. Although studies of the sickest pediatric ARDS patients switched from CMV to ECMO, favoring the use of ECMO for severe respiratory failure in general PICU [10], its usage for pediatric patients with ARDS after cardiac surgery remains unpopular, especially in China due to the limited experience with ECMO and the poor outcomes of ECMO-treated pediatric patients with ARDS (the survival rate for 11 pediatric patients was 18 % (2/11) [11], the relatively high cost, and the serious complications.

This study indicates that pulmonary hypertension and RRTI are significantly and independently associated with increasing mortality. Pulmonary hypertensive pediatric patients are vulnerable to postperfusion lung injury [4, 20]. Preoperative respiratory tract infection can contribute to worse clinical outcomes [9]. It was not surprising that pulmonary hypertension and RRTI posed a significant and independent risk.

Study Limitations

First, many changes in surgical technique and ICU management occurred during the nearly 4-year study period. Second, this was a retrospective study, which might have introduced potential misclassification bias. Third, the study was based on a single center, which may have impeded the application of current results to other institutions. Additionally, the observation period was short.

In summary, HFOV was associated with improved gas exchange and a trend toward lower mortality for pediatric ARDS patients who underwent corrective heart surgery in terms of discharge survival. Pulmonary hypertension and RRTI were independent risk factors for mortality. Future studies should focus on these risk factors to lower the mortality rate for these patients.