Pediatric Critical Care

High-Frequency Oscillatory Ventilation in Pediatric Acute Lung Injury: A Multicenter International Experience* Jordan S. Rettig, MD1; Craig D. Smallwood, RRT1; Brian K. Walsh, RRT1,2; Peter C. Rimensberger, MD3; Thomas E. Bachman, MSHA4; Casper W. Bollen, MD PhD5; Els L. Duval, MD, PhD6; Fabienne Gebistorf, MD, PhD3; Dick G. Markhorst, MD, PhD7; Marcel Tinnevelt, BSN5; Mark Todd, RRT8; David Zurakowski, PhD1; John H. Arnold, MD1

*See also p. 2697. Department of Anesthesiology, Perioperative and Pain Medicine, Division of Critical Care Medicine, Boston Children's Hospital, Harvard Medical School, Boston, MA. 2 Children’s Medical Center of Dallas, Dallas, TX. 3 Neonatal and Pediatric Intensive Care Unit, Department of Pediatrics, University Hospital of Geneva, Geneva, Switzerland. 4 Czech Technical University in Prague, Kladno, Czech Republic. 5 Department of Pediatric Intensive Care, Wilhelmina Children’s Hospital, University Medical Center Utrecht, Utrecht, The Netherlands. 6 Pediatric Intensive Care, Queen Paola Children's Hospital, Antwerp, Belgium. 7 Division of Pediatric Intensive Care, Department of Pediatrics, VU University Medical Center, Amsterdam, The Netherlands. 8 The Hospital for Sick Children, Toronto, ON, Canada. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ccmjournal). Supported, in part, by CareFusion to compensate the individuals who performed the chart reviews who are not listed as co-authors. Work was performed at all centers listed, except for the Czech Technical University in Prague. Dr. Rettig’s institution received grant support from CareFusion. Dr. Smallwood disclosed a financial relationship with the American Association of Respiratory Care. His institution received grant support from Economedtrx and had a financial relationship with Hill-Rom Medical. Dr. Walsh received grant support from Econometrica (per patient fee paid for each complete data set). Dr. Rimensberger received support for travel (support by several educational conferences for travel and accommodation, when invited as speaker). Dr. Bachman disclosed that he is also the principal of a clinical research consultancy, Economedtrx Inc. (ongoing relationship with CareFusion the manufacturer of the High frequency oscillatory ventilation system the centers used in this study. CareFusion was not involved with this project, other than as disclosed in the article. That is, they provided support to the participating center for data collection). Dr. Gebistorf’s institution received grant support (funds were provided to compensate the individuals who performed the chart reviews). Dr. Markhorst’s institution received grant support from Economedtrx (funds were provided to compensate the individuals [not listed as author] for their work on data retrieval). Dr. Todd’s institution received grant support from Economedtrx. Dr. Arnold consulted for Medical Conservation Devices (compensation [< $500] for 1

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved. DOI: 10.1097/CCM.0000000000001278

2660

www.ccmjournal.org

participation on conference calls) and provided expert testimony for Post & Schell, LLC (malpractice case). His institution received grant support from Economedtrx (funds were provided to compensate the individuals who performed the chart reviews [who are not listed as co-authors]) and received grant support from Maquet (Research support). The remaining authors have disclosed that they do not have any potential conflicts of interest. Address requests for reprints to: Jordan S. Rettig, MD, Division of Critical Care Medicine, Department of Anesthesiology, Perioperative, and Pain Medicine, Boston Children’s Hospital, 300 Longwood Avenue, Bader 634, Boston, MA 02115. E-mail: [email protected]

Objective: We aim to describe current clinical practice, the past decade of experience and factors related to improved outcomes for pediatric patients receiving high-frequency oscillatory ventilation. We have also modeled predictive factors that could help stratify mortality risk and guide future high-frequency oscillatory ventilation practice. Design: Multicenter retrospective, observational questionnaire study. Setting: Seven PICUs. Patients: Demographic, disease factor, and ventilatory and outcome data were collected, and 328 patients from 2009 to 2010 were included in this analysis. Interventions: None. Measurement and Main Results: Patients were classified into six cohorts based on underlying diagnosis. We used univariate analysis to identify factors associated with mortality risk and multivariate logistic regression to identify independent predictors of mortality risk. An oxygenation index greater than 35 and immunocompromise exhibited the greatest predictive power (p < 0.0001) for increased mortality risk, and respiratory syncytial virus was associated with lowest mortality risk (p = 0.003). Differences in mortality risk as a function of oxygenation index were highly dependent on primary underlying condition. A trend toward an increase in oscillator amplitude and frequency was observed when compared with historical data. Conclusions: Given the number of centers and subjects included in the database, these findings provide a robust description of current December 2015 • Volume 43 • Number 12

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Pediatric Critical Care practice regarding the use of high-frequency oscillatory ventilation for pediatric hypoxic respiratory failure. Patients with severe hypoxic respiratory failure and immunocompromise had the highest mortality risk, and those with respiratory syncytial virus had the lowest. A means of identifying the risk of 30-day mortality for subjects can be obtained by identifying the underlying disease and oxygenation index on conventional ventilation preceding the initiation of high-frequency oscillatory ventilation. (Crit Care Med 2015; 43:2660–2667) Key Words: acute respiratory distress syndrome; acute hypoxic respiratory failure; acute lung injury; high-frequency ventilation; mechanical ventilation; pediatrics

H

igh-frequency oscillatory ventilation (HFOV) is a form of mechanical ventilation that has demonstrated benefit in children with moderate and severe acute respiratory distress syndrome (ARDS) (1, 2). HFOV has become a standard mode of ventilatory support in the PICU (1, 3). Only two randomized trials of HFOV in pediatric patients have been published (4, 5). Despite limited clinical trial data, HFOV is being used as a rescue mode of ventilation when conventional lung protective ventilatory modalities have failed (1). HFOV uses a constant distending pressure (the mean airway pressure [MAP]) coupled with a sinusoidal flow oscillation, to achieve adequate alveolar ventilation using smaller tidal volumes than delivered during conventional mechanical ventilation (CMV). This allows the use of higher MAPs to aggressively pursue alveolar recruitment while avoiding repetitive opening and collapse of atelectasis-prone lung (atelectrauma). The use of tidal volumes at or less than deadspace reduces the need for high peak airway pressures in the poorly compliant lung and reduces the prevalence of volutrauma and barotrauma. Therefore, HFOV provides lung protective ventilation with lower peak-to-trough pressure amplitudes while simultaneously preventing alveolar collapse. Biotrauma has also become an important consideration because it may exacerbate lung injury, promote translation of inflammatory mediators, and contribute to the systemic inflammatory response associated with acute lung injury. In several models of lung injury, HFOV has been associated with lower concentrations of inflammatory cytokines than CMV (6, 7). Clinical trials have demonstrated that HFOV produces rapid improvement in oxygenating efficiency with minimal hemodynamic side effects (3, 8). In pediatric patients with acute lung injury and air leak, HFOV was associated with improvement in arterial to alveolar oxygenation (PaO2/PAO2) ratio and oxygenation index (OI) (4). The patients treated with HFOV had a lower prevalence of lung injury, defined by any use of supplemental oxygen at 30 days, than patients managed with CMV. A recently published meta-analysis of eight randomized controlled trials in adults and children suggested that HFOV may decrease mortality risk and has a lower prevalence of treatment failure (8). In 2013, the Oscillation in ARDS study group conducted a randomized controlled trial of HFOV when compared with that of CMV and reported that the use of HFOV had no significant Critical Care Medicine

effect on 30-day mortality risk in patients with ARDS who required mechanical ventilation (9). Simultaneously, the Oscillation for Acute Respiratory Distress Syndrome Treated Early study group reported that the early application of HFOV, when compared with a CMV strategy, increases mortality risk (10). These studies provide insight into adult practices, but it remains unclear how relevant adult studies looking at nonrescue application of HFOV are to pediatric practice (2, 3, 11). We have previously reported a multicenter HFOV experience in 10 tertiary care PICUs (1). That study was able to broadly reflect practices and outcomes in pediatric patients who received HFOV. In the present study, we surveyed three North American and four European centers to describe the evolution of HFOV clinical practice and outcomes in the past 10 years.

MATERIALS AND METHODS Seven high-volume tertiary PICUs with significant experience using HFOV provided data regarding practice patterns and outcomes with the use of a single HFOV device (model 3100A or 3100B; CareFusion, Yorba Linda, CA). The participating centers were Boston Children’s Hospital (Boston, MA), Children’s Medical Center of Dallas (Dallas, TX), Queen Paola Children’s Hospital (Antwerp, Belgium), University Hospitals of Geneva (Geneva, Switzerland), University Medical Center Utrecht (Utrecht, Netherlands), Universiteit Medical Center (Amsterdam, Netherlands), Hospital for Sick Children (Toronto, ON). The data collection tool was a paper-based detailed case report form for any pediatric patient treated with HFOV during the study period from 2009 to 2010, which was 10 years after the prior study (1). Patients were identified by chart review; in some institutions, the respiratory care departments kept their own databases. All patients receiving HFOV during the study period were counted. The case report form included information about the etiology of acute lung injury, comorbidities, conventional ventilator management prior to HFOV and after HFOV, arterial blood gas data, ventilator settings on HFOV for the first 72 hours, and patient outcomes. The research ethics committee at each of the participating centers approved this study. Respiratory therapists, physicians, or nurses at each institution completed the case report forms. Case report forms were collated, and the first and the second authors checked for any missing data points or data that seemed out of physiologic range, which may have been entered by error. In those cases, the primary institution was contacted to confirm data. If there were still missing or unresolved data points, then those patients were excluded from the study. The data were entered into a database for statistical analysis in Microsoft Excel (v14.0.6129.5000; Microsoft, Redmond, WA). Patients were stratified to one of five distinct categories based upon primary diagnosis and a category for acute treatment failure: 1. History of lung disease (previous history of chronic lung disease before initiation of HFOV). 2. Acute treatment failure (patient did not tolerate HFOV and received < 3 hr). www.ccmjournal.org

2661

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Rettig et al

Table 1.

Patient Demographics Stratified by Cohort History of Lung Disease

Variable

No. of patients

Acute ­Treatment Failure

­Respiratory Syncytial Virus

Cyanotic Heart Disease

­Immunocompromised

No Prior Lung Disease

53

12

36

28

23

176

15

4

4

1a

38a

10

(1–100)

(2–24)

(0–9)

(17–84)

(1–55)

52

95

53

(50–58)

(69–108)

(46–89)

4.1a

15.3a

Age (mo)  Median  IQR

(7–48)

Height (cm)  Median

70

 IQR

47

78

(56–98)

(42–52)

(60–97)

8.7

3.9

(5.1–14.5)

(2.6–13.0)

(4.0–13.2)

(3.3–5.8)

(11.0–27.0)

(3.3–17.0)

 Male

24 (45)

4 (33)

14 (39)

19 (68)

11 (48)

88 (50)

 Female

29 (55)

8 (67)

22 (61)

9 (32)

12 (52)

88 (50)

Weight (kg)  Median  IQR

6.0

8.5

Gender, n (%)

IQR = interquartile range. Age and weight were significantly lower in cyanotic heart disease group (p < 0.01) and higher in immunocompromised group (p < 0.01) than those in no prior lung disease group (Mann-Whitney U test) (ap < 0.01).

3. Respiratory syncytial virus (RSV) positive. 4. Congenital cardiac anomaly with right-to-left shunting. 5. Immunocompromise (patients with documented congenital or acquired immunodeficiency disorder or those on immunosuppressive agents). 6. No prior lung disease (including remainder of subjects without previous history of lung disease). A total of 361 case report forms were collected from the collaborating institutions, which represented all of the patients managed with HFOV during the study period. A total of 12 cases were excluded due to missing outcome data, ventilatory data, and diagnosis. In 21 cases involving patients treated with HFOV more than once, a single case report form was generated for the first HFOV run. Therefore, a total of 328 case report forms were included in the analysis. The centers all had experience with HFOV and the single device used (CareFusion 3100A and 3100B). The center sizes ranged from six beds to 48 beds, with an average total ICU admission rate during the study period of 2,583 patients (range, 550–5723). Table 1 provides a detailed patient description of the 328 patients included in our analysis. All subjects were categorized into one of six predetermined categories based on their primary diagnosis, or acute treatment failure, which was determined from physician documentation in the medical record. Disease groups were compared using the Mann-Whitney U test for median and interquartile range in patient age, height, weight, conventional ventilation duration and follow-up, and the Pearson chi-square test for gender. Conventional settings immediately preceding HFOV such as positive end-expiratory pressure (PEEP), pH, Fio2, PaO2, OI, and at initiation of HFOV 2662

www.ccmjournal.org

(such as deltaP, power, frequency, and bias flow) are summarized by means and sd with groups compared by analysis of variance with post hoc Dunnett t tests using no prior lung disease as the reference category (12). The relationship between deltaP (cm H2O) and frequency (Hz) with respect to patient’s weight was analyzed using linear and nonlinear regression models with a power function of the general form y = axb, providing the best fit to the data as judged by R2 both for the current study and for the Arnold 2000 study (13). Disease groups were compared in terms of prevalence of 30-day mortality risk using logistic regression. Independent predictors of 30-day mortality risk based on continuous covariates, such as pH and OI at initial HFOV, were modeled using multivariable logistic regression with the backward selection procedure and the likelihood ratio test to assess significance. The probability of mortality risk using maximum likelihood estimation was derived for each disease group based on a range of OI, adjusted for age and gender (14). Receiver operating characteristic curve analysis based on the Youden J-index was applied to identify the optimal cutoff value of OI at the start of HFOV for assessing mortality risk, and on the basis of this cutoff value and the specific disease group, we constructed a multivariable predictive algorithm (15). Among survivors, univariate and multivariable logistic regressions were used to determine predictive CMV factors immediately preceding HFOV of the need for respiratory support or supplemental oxygen with the goal of developing a risk stratification model. Power analysis indicated that the sample sizes for the disease groups provided 80% power to detect a 30% difference in OI between each group when compared with patients with no prior lung disease December 2015 • Volume 43 • Number 12

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Pediatric Critical Care

Table 2. Conventional Ventilation Setting Immediately Preceding High-Frequency Oscillatory Ventilation Variable

Ppeak, cm H2O

History of Lung Acute Treatment Disease Failure

Respiratory ­Syncytial Virus

Cyanotic Heart Disease

Immunocompromised

No Prior Lung Disease

29 ± 7

30 ± 2

31 ± 5

28 ± 8

31 ± 5

Positive end-expiratory pressure, cm H2O

9 ± 3

10 ± 3

8 ± 3

7 ± 3

12 ± 3

Mean airway pressure, cm H2O

16 ± 4

17 ± 3

16 ± 3

15 ± 6

19 ± 4

17 ± 5

Respiratory rate, breaths/min

33 ± 8

31 ± 7

38 ± 9b

41 ± 12b

34 ± 8

31 ± 11

Tidal volume, mL/kg

5.8 ± 1.8

6.8 ± 3.9

6.2 ± 3.0

Fio2

0.85 ± 0.19

0.96 ± 0.05

pH

7.22 ± 0.16

7.33 ± 0.32 46 ± 5

31 ± 7 a

10 ± 4

7.2 ± 3.4

6.1 ± 1.5

6.3 ± 4.3

0.69 ± 0.24

a

0.74 ± 0.26

0.88 ± 0.16

0.82 ± 0.21

7.29 ± 0.11

a

7.21 ± 0.12

7.25 ± 0.12

7.21 ± 0.13

61 ± 17

62 ± 17

64 ± 20

60 ± 24

72 ± 47

68 ± 35

71 ± 31

123 ± 103

80 ± 46

98 ± 79

23 ± 18

29 ± 13

23 ± 11

39 (4–71)

5 (0–28)

CO2, mm Hg

69 ± 23

PaO2, mm Hg

64 ± 25

124 ± 162a

75 ± 26

PaO2/Fio2, %

81 ± 51

126 ± 161

126 ± 75

Oxygenation index

25 ± 17

26 ± 12

19 ± 12

1 (0–17)

20 (4–72)

Conventional mechanical ventilation, hr  Before HFOV  After HFOV

10 (2–84) 117 (82–246)

161 (29–492)

34 (0–108)

7 (0–125) 96 (1–216)

50 (0–118)

84 (0–217)

Data are mean ± sd, except duration of conventional mechanical ventilation, which is median (interquartile range). In the acute treatment failure group, the PaO2 was elevated. In the respiratory syncytial virus group, RR and pH were significantly higher and Fio2 was decreased. In the cyanotic heart disease group, RR was significantly higher. In the immunocompromised group, PEEP was significantly higher. HFOV = high-frequency oscillatory ventilation. All values were compared with no prior lung disease group (ap < 0.05; bp < 0.01).

based on the F-test in analysis of variance (version 7.0; nQuery Advisor; Statistical Solutions, Saugus, MA). Statistical analysis was performed using IBM/SPSS Statistics (version 21.0; IBM, Armonk, NY). Two-tailed p values of less than 0.05 were considered statistically significant.

RESULTS Patient Description Table 1 provides subject demographics for each of the six discrete categories for those patients who were included in the analysis. Comparisons with the referent cohort (no prior lung disease) were done, and statistically significant differences are denoted by values of p less than *0.05 and **0.01. Initiation of HFOV Table 2 shows the CMV settings just prior to the initiation of HFOV. The acute treatment failure group had a statistically significantly higher PaO2 than the other groups (p < 0.05). The RSV group showed a higher respiratory rate (p < 0.01), lower FIO2 (p < 0.05), and a higher pH (p < 0.05) prior to initiation of HFOV. Patients with congenital heart disease had a lower PEEP (p < 0.05), higher respiratory rate (p < 0.01), and longer duration of CMV (p < 0.01). The immunocompromised group was on a higher PEEP (p < 0.05). The OI at the time of initiation of HFOV Critical Care Medicine

for RSV, immunocompromised patients, and those with no prior lung disease is presented in Figure 1. In each category, greater than 70% of patients had an OI greater than 16 preceding initiation of HFOV, with the majority (> 60%) having an OI greater than 24. Centers did not consistently report peripheral capillary oxygen saturation (SpO2), so that was not included in the analysis. Response to HFOV Table 3 shows the initial HFOV settings for each of the disease categories. For the entire cohort, the MAP was increased from 16.8 ± 4.6 cm H2O on CMV to 23.0 ± 7.5 cm H2O at initiation of HFOV. Patients with a history of lung disease had higher bias flow (p < 0.01). Patients who had acute treatment failure were placed on a higher frequency (p < 0.05). Among this group, six of 12 patients died in the 3-hour period. Patients who had RSV were on lower power (p < 0.01), lower Fio2 (p < 0.01), and had a higher pH (p < 0.01). The immunocompromised patients had a higher bias flow (p < 0.01) and higher deltaP (p < 0.05). To assess change in HFOV strategy over time, we examined deltaP by weight and frequency by weight when compared with data reported in a prior publication describing practice in the late 1990s (1). As shown in Supplemental Figure 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/B411) and Supplemental Figure 2 (Supplemental Digital Content 2, http://links.lww.com/CCM/B412), there is a trend toward higher www.ccmjournal.org

2663

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Rettig et al

Figure 1. Oxygenation index (OI) at time of initiation of highfrequency oscillatory ventilation (HFOV). This shows the OI for patients in representative disease categories at the time of initiation of HFOV. The bars are color coded by the range of OI, and the percentages above the bars represent the percentage of patients from that disease category with that range of OI. In all categories, greater than 70% of patients had an OI greater than 16 preceding initiation of HFOV, with the majority (> 60%) having an OI greater than 24. RSV = respiratory syncytial virus.

pressure amplitudes (deltaP) and higher frequency (Hz) in our cohort when compared with those in the prior data set (1). Our power regression analyses yielded a statistically significant fit of the model and coefficients of determination: R2 and p values were R2 = 0.476 (p < 0.001) and R2 = 0.289 (p < 0.001) for the deltaP-weight and frequency-weight associations. Quantification of Mortality Risk Including all categories, the overall mortality risk in this study was 38.1%. The patients with immunocompromise showed the highest mortality risk. An OI greater than 35 and Table 3.

immunocompromise exhibited the greatest predictive power (p < 0.001) for mortality risk when compared with no prior lung disease. RSV was associated with survival (p = 0.003). Once we separated the patients into their disease groups, we were able to create a model to predict 30-day mortality risk by looking at both disease category and OI at the initiation of HFOV (Fig. 2). Multivariable logistic regression analysis confirmed that both OI obtained during CMV immediately preceding HFOV (likelihood ratio = 9.61 on 1 df; p = 0.002) and diagnosis (likelihood ratio test = 18.76 on 4 df; p = 0.001) are significant predictors of 30-day mortality risk. Subgroup analysis revealed a significantly increasing risk of 30-day mortality for higher levels of OI for patients with history of lung disease (p = 0.03), patients with cyanotic heart disease (p = 0.03), and immunocompromised patients (p = 0.04) (Fig. 2). Patients with RSV (p = 0.16) and those patients with no prior lung disease (p = 0.22) did not show a statistically significant increase in mortality risk with higher OI. Further analysis with adjustment for OI during CMV preceding HFOV initiation confirmed that when compared with patients with no prior lung disease, the risk of mortality was significantly higher in patients with cyanotic heart disease (p = 0.034) and those who were immunocompromised (p < 0.001). Patients with RSV were at a significantly lower risk of 30-day mortality than those with no prior lung disease (p = 0.041) independent of the level of OI. No significant differences were detected between patients with and without a history of lung disease (p = 0.20). We evaluated whether or not there was a time period within which OI may start to separate survivors from non-survivors. Not surprisingly and as reported by a number of authors, the overall OI in the survivors was lower (1, 5) than in the nonsurvivors. Initially, both survivors and nonsurvivors had improvement in OI, but at approximately 48 hours, the OI increased

Initial High-Frequency Oscillatory Ventilation Settings History of Lung Disease

Acute ­Treatment Failure

­Respiratory Syncytial Virus

Cyanotic Heart Disease

Immunocompromised

No Prior Lung Disease

DeltaP, cm H2O

64 ± 19

49 ± 31

65 ± 20

51 ± 24

70 ± 13a

55 ± 25

Power

3.7 ± 2.0

3.3 ± 1.4

2.3 ± 2.1

4.1 ± 1.8

5.3 ± 1.7

4.1 ± 2.2

9 ± 2

11 ± 3

10 ± 2

9 ± 2

8 ± 1

9 ± 2

28 ± 8

25 ± 8

23 ± 6

22 ± 4

29 ± 6

23 ± 7

Variable

Frequency, Hz Bias flow, L/min

b

a

b

b

Fio2

0.91 ± 0.15

0.87 ± 0.22

0.68 ± 0.24

0.82 ± 0.26

0.90 ± 0.16

0.83 ± 0.23

pH

7.31 ± 0.16

7.28 ± 0.29

7.37 ± 0.15

7.32 ± 0.20

7.29 ± 0.19

7.26 ± 0.15

CO2, mm Hg

53 ± 20

53 ± 23

50 ± 21

47 ± 20

60 ± 27

53 ± 19

PO2, mm Hg

75 ± 54

67 ± 47

77 ± 58

97 ± 107

75 ± 31

91 ± 73

PaO2/Fio2, %

86 ± 60

97 ± 114

135 ± 95

122 ± 128

90 ± 52

117 ± 103

Oxygenation index

43 ± 36

41 ± 24

29 ± 21

37 ± 29

37 ± 17

33 ± 26

b b

Data are mean ± SD. In the history of lung disease group, the bias flow was significantly higher. The frequency was higher in the acute treatment failure group. In the RSV group, power and Fio2 was lower while the pH was higher. In the immunocompromised group, both the deltaP and bias flow were elevated. RSV, respiratory syncytial virus. All value were compared with no prior lung disease group (ap < 0.05; bp < 0.01).

2664

www.ccmjournal.org

December 2015 • Volume 43 • Number 12

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Pediatric Critical Care

DISCUSSION Despite limited clinical trial data, HFOV is used as a rescue mode of ventilatory support in the PICU (2, 3, 11). A recent analysis of PICU practices by Gupta et al suggested that HFOV might not be effective; however, the conclusions and methods presented in the analysis by Gupta et al (16) have been meet with skepticism (17–19). In the current study, we have described recent HFOV practices, which we feel are representative of the evolution of practice in North America and Western Europe. This study was designed to provide a 10-year follow-up to the original study by Arnold et al (1). A decade is an appropriFigure 2. Prediction of 30-d mortality by oxygenation index (OI) and diagnostic category. Probability of ate amount of time to watch 30-d mortality curve for each diagnostic category plotted against oxygenation index that was obtained on practice evolve. More recently, conventional ventilation immediately preceding initiation of high-frequency oscillatory ventilation. Those some institutions have created diagnostic categories in which increased OI significantly increased risk of 30-d mortality are shown as *p < 0.05. Multivariable logistic regression controlling for OI revealed significantly higher risk of mortality in the their own HFOV guidelines. immunocompromised and cyanotic heart disease groups (†p < 0.05) than patients with no prior lung disease The ways in which guidelines (LD). The respiratory syncytial virus (RSV) group demonstrated a significantly lower risk of mortality affect practice is interesting, (†p < 0.05) than those with no prior history of LD. No differences were detected between patients with LD and those without a history of LD. but not the goal of this article. So we used data prior to these changes as better representation of the evolution of the steadily in nonsurvivors than in survivors (Fig. 3). This trend use of HFOV in pediatric patients. The main findings of this was evident for patients with lung disease and patients with no study are that the immunocompromise group had the highprior lung disease. est mortality risk and the RSV group the lowest. Eighty percent of patients with RSV were initiated on HFOV with an OI Risk of Chronic Lung Disease We defined chronic lung disease as ongoing need for any respi- greater than 16 and 61% with OI greater than 24, indicating ratory support or supplemental oxygen among survivors at 30 a rescue strategy versus a primary (nonrescue) HFOV strategy, which likely would have shown HFOV initiation at a lower days, regardless of disposition at that time. The two indepenOI (Fig. 1). When compared with a decade ago, we observed dent multivariate predictors of the need for respiratory supa trend toward higher Hertz and higher deltaP strategy on port were low pH and high OI on CMV immediately preceding HFOV. Finally, the probability of 30-day mortality risk can be initiation of HFOV. Receiver operating characteristic analysis estimated by identifying the underlying condition and comrevealed cutpoints of less than 7.3 for pH and greater than 25 puting OI on CMV prior to the initiation of HFOV. for OI as being significant predictors of requirement of respiraAnalysis of the CMV settings prior to initiation of HFOV tory support for all patients or any use of supplemental oxygen showed predictable trends, including a lower PEEP and lonat 30 days following initiation of HFOV. Among all patients ger duration of CMV in the congenital heart disease group. who survived and had a pH less than 7.3 and OI greater than We believe this is reflective of the tendency to limit negative 25, there was an 80% probability that they required any respi- cardiopulmonary interaction by limiting airway opening presratory support at 30 days when compared with patients who sure in ventilated cardiac patients. It is also possible that these exhibit pH greater than 7.3 and OI less than 25, in which case patients were transitioned to HFOV to improve alveolar venthe probability is 20%. Univariate analysis indicated diagnosis tilation rather than to enhance oxygenation, which may also and the following five CMV variables as predictors of need for explain the lower PEEP prior to HFOV. The immunocomrespiratory support: decreased pH (p = 0.006), decreased PaO2 promised patients were managed with higher PEEP, which is (p = 0.009), decreased P/F ratio (p = 0.003), increased Fio2 likely reflective of the secondary lung injury caused by their disease process. Overall, the RSV group had lower Fio2 and (p = 0.008), and increased OI (p = 0.006). Critical Care Medicine

www.ccmjournal.org

2665

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Rettig et al

Figure 3. Oxygenation index (OI) in survivors versus nonsurvivors. A, OI over time for survivors (solid line) and nonsurvivors (dashed line) of those subjects who had a history of lung disease. B, OI over time for survivors (solid line) and nonsurvivors (dashed line) of those subjects who did not have a history of lung disease. Analysis of variance results demonstrated that survivors exhibited a statistically significant decrease in OI continued until 72 hr of treatment. Nonsurvivors demonstrated decreased OI from initiation to 48 hr. At 72 hr, the nonsurvivors demonstrated an increased OI, which was similar to the OI at initiation of HFOV. Error bars represent sem.

higher pH at the time of transition to HFOV, showing a possible trend toward using earlier HFOV in this patient population. Although not statistically significant, the RSV group did show a trend toward shorter duration of CMV prior to HFOV. Interestingly, the acute treatment failure group started with a significantly higher PaO2, perhaps suggesting that if the primary derangement is not oxygenation, HFOV may not be the ideal modality. Once HFOV was initiated, the acute treatment failure group was on a higher frequency. From our retrospective analysis, it is difficult to precisely identify factors related to treatment failure in the six who survived. It is possible that they developed adverse hemodynamic effects, hypercarbia or a combination that would discourage the clinician from altering HFOV strategy or persisting with the mode for an extended trial. The six who died during the 3-hour period were likely too severely ill to be rescued by any ventilator modality. The OSCILLATE study group findings demonstrated that early use of HFOV might increase mortality risk (10). In Figure 1, we demonstrate that in representative groups, greater 2666

www.ccmjournal.org

than 70% of patients had an OI greater than 16 preceding initiation of HFOV, with the majority (> 60%) having an OI greater than 24. This implies that in current pediatric practice, HFOV is still being used as a rescue strategy and not as a primary mode of ventilatory support. Therefore, the OSCILLATE data, which focuses on primary application of HFOV in adults, may not be applicable to current pediatric practice. Our data demonstrate a higher Hertz and higher deltaP strategy than data from the past decade (1). This change in practice is likely related to evolution of our understanding of the lung protective effects of HFOV. Meyer et al (20) and Liu et al (6) used an animal model to demonstrate less lung injury and reduced levels of inflammatory mediators using a higher Hertz strategy. Hager et al (21) examined factors determining Vt during HFOV in both a test lung model and patients using a hot wire anemometer placed in series with a Sensormedics 3100B high-frequency ventilator. In patients with ARDS, they reported that increasing frequency by 2 Hz was associated with a 23.1% (± 6.3%) reduction in delivered Vt, while that increasing deltaP by 10 cm H2O was associated with only a 5.6 (± 4.5%) increase in delivered Vt (18). This demonstrates that higher Hertz results in lower tidal volumes, and an increase in deltaP does not contribute as significantly to increasing Vt. Our data demonstrate that centers in North America and Western Europe are using higher Hertz, higher deltaP strategies, which may represent an attempt to provide enhanced lung protection with HFOV (1). In our cohort, an OI greater than 35 and immunocompromise exhibited the greatest predictive power (p < 0.0001) for mortality risk. Furthermore, we were able to determine probability of 30-day mortality risk by disease category and OI at initiation of HFOV. This emphasizes the relevance of underlying disease process in predicting response to HFOV and overall mortality risk. It is clear from our data that certain patient groups treated with HFOV, such as patients with RSV, maintain a low mortality rate. The strengths of this study are that we have complete HFOV data on all patients included in the analysis. Furthermore, we have included both small and large centers from North America and Western Europe and therefore believe that our data provide a robust description of current practice. The major weakness is that this is a retrospective, questionnaire study. Comparison data and criteria for changes from CMV to HFOV are not available. We also did not collect hemodynamic data. In addition, there are more cases from North America than from Europe. This may potentially skew the study to better represent North American practices.

CONCLUSIONS The present study provides an update on current practices surrounding application of HFOV in pediatric subjects over the past decade. On the basis of these trends, in this study, we observed that the mortality risk might be linked to disease category rather than the use of HFOV itself. Patients with immunocompromise had the highest mortality risk, and those with December 2015 • Volume 43 • Number 12

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

Pediatric Critical Care

RSV had the lowest. HFOV continues to be used as a rescue mode of ventilation in the pediatric population. The regression models we have described, which predict 30-day mortality risk and need for respiratory support, may assist in determining ideal ventilatory modality and overall prognosis.

ACKNOWLEDGMENTS We thank the respiratory care departments at their respective institutions for help in gathering information on the study patients.

REFERENCES

1. Arnold JH, Anas NG, Luckett P, et al: High-frequency oscillatory ventilation in pediatric respiratory failure: A multicenter experience. Crit Care Med 2000; 28:3913–3919 2. Randolph AG: Management of acute lung injury and acute respiratory distress syndrome in children. Crit Care Med 2009; 37:2448–2454 3. Arnold JH, Hanson JH, Toro-Figuero LO, et al: Prospective, randomized comparison of high-frequency oscillatory ventilation and conventional mechanical ventilation in pediatric respiratory failure. Crit Care Med 1994; 22:1530–1539 4. Samransamruajkit R, Prapphal N, Deelodegenavong J, et al: Plasma soluble intercellular adhesion molecule-1 (sICAM-1) in pediatric ARDS during high frequency oscillatory ventilation: A predictor of mortality. Asian Pac J Allergy Immunol 2005; 23:181–188 5. Derdak S, Mehta S, Stewart TE, et al; Multicenter Oscillatory Ventilation For Acute Respiratory Distress Syndrome Trial (MOAT) Study Investigators: High-frequency oscillatory ventilation for acute respiratory distress syndrome in adults: A randomized, controlled trial. Am J Respir Crit Care Med 2002; 166:801–808 6. Liu S, Yi Y, Wang M, et al: Higher frequency ventilation attenuates lung injury during high-frequency oscillatory ventilation in sheep models of acute respiratory distress syndrome. Anesthesiology 2013; 119:398–411 7. Ronchi CF, dos Anjos Ferreira AL, Campos FJ, et al: High-frequency oscillatory ventilation attenuates oxidative lung injury in a rabbit model of acute lung injury. Exp Biol Med (Maywood) 2011; 236:1188–1196

Critical Care Medicine

8. Sud S, Sud M, Friedrich JO, et al: High frequency oscillation in patients with acute lung injury and acute respiratory distress syndrome (ARDS): Systematic review and meta-analysis. BMJ 2010; 340:c2327 9. Young D, Lamb SE, Shah S, et al; OSCAR Study Group: High-frequency oscillation for acute respiratory distress syndrome. N Engl J Med 2013; 368:806–813 10. Ferguson ND, Cook DJ, Guyatt GH, et al; OSCILLATE Trial Investigators; Canadian Critical Care Trials Group: High-frequency oscillation in early acute respiratory distress syndrome. N Engl J Med 2013; 368:795–805 11. Slee-Wijffels FY, van der Vaart KR, Twisk JW, et al: High-frequency oscillatory ventilation in children: A single-center experience of 53 cases. Crit Care 2005; 9:R274–R279 12. Cabral HJ: Multiple comparisons procedures. Circulation 2008; 117:698–701 13. Seber GAF, Wild CJ. Nonlinear Regression. New York: John Wiley; 2003:21–45 14. Katz MH. Multivariable analysis. A Practical Guide for Clinicians and Public Health Researchers. Third ed. New York: Cambridge University Press; 2011:140–161 15. Zhou XH, Obuchowski NA, McClish DK. Statistical Methods in Diagnostic Medicine. Second ed. New York: John Wiley; 2011:36–55 16. Gupta P, Green JW, Tang X, et al: Comparison of high-frequency oscillatory ventilation and conventional mechanical ventilation in pediatric respiratory failure. JAMA Pediatr 2014; 168:243–249 17. Essouri S, Emeriaud G, Jouvet P: It is too early to declare early or late rescue high-frequency oscillatory ventilation dead. JAMA Pediatr 2014; 168:861–862 18. Kneyber MC, van Heerde M, Markhorst DG: It is too early to declare early or late rescue high-frequency oscillatory ventilation dead. JAMA Pediatr 2014; 168:861 19. Rimensberger PC, Bachman TE: It is too early to declare early or late rescue high-frequency oscillatory ventilation dead. JAMA Pediatr 2014; 168:862–863 20. Meyer J, Cox PN, McKerlie C, et al: Protective strategies of high-frequency oscillatory ventilation in a rabbit model. Pediatr Res 2006; 60:401–406 21. Hager DN, Fessler HE, Kaczka DW, et al: Tidal volume delivery during high-frequency oscillatory ventilation in adults with acute respiratory distress syndrome. Crit Care Med 2007; 35:1522–1529

www.ccmjournal.org

2667

Copyright © 2015 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.