Introduction
Both acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are life-threatening conditions that are usually associated with substantial morbidity [
1,
2], mortality [
3], and financial costs [
4]. Conventional mechanical ventilation (CMV) is still considered the cornerstone of treatment for these patients. However, although mechanical ventilation can initially sustain life, it may cause further lung injury [
5‐
8].
To avoid ventilator-induced lung injury, lung-protective ventilation has been recommended, which focused on avoiding cyclic alveolar collapse and re-expansion, preventing alveolar excess distension, and achieving and maintaining alveolar recruitment [
9‐
11]. High-frequency oscillation is an alternative mechanical ventilation method that delivers very small tidal volumes at high frequencies (3 to 15 Hz) using an oscillatory pump [
12]. High-frequency oscillatory ventilation (HFOV) can not only avoid over-distension of alveoli by delivering small tidal volumes but can also prevent end-expiratory alveolar collapse and maintain alveolar recruitment by applying a constant airway pressure [
13‐
15]. Therefore, HFOV theoretically achieves all goals pursued by lung-protective ventilation strategies [
11,
16,
17].
However, no more than six randomized controlled trials (RCTs) in adult ARDS patients have been published on the safety and efficacy of HFOV as an initial treatment strategy. Three previous trials comparing HFOV with CMV suggested that HFOV improved both oxygenation and survival in adults with ARDS [
18‐
20], but two recent larger-scale RCTs presented different or even opposite results [
21,
22]. Therefore, this approach remains an unproven and controversial therapy for adults with ARDS [
23‐
26].
Two Cochrane reviews examining the effect of HFOV on mortality in ALI/ARDS patients have been published. The earlier one found only two small RCTs and was not powerful enough to draw definitive conclusions [
27]; the later study [
28] concluded that HFOV might improve survival, which was not completely consistent with the conclusions of two recently published large RCTs [
21,
22]. Neither of the above two Cochrane reviews focused on the effect of HFOV in unique adults with ARDS. Since two large scale RCTs comparing HFOV with CMV as an initial treatment for unique adult ARDS patients have been recently published, we performed a meta-analysis of RCTs, to systematically review the efficacy and safety of HFOV in the population of adult ARDS patient compared with CMV.
Methods and materials
Ethics statement
We performed a meta-analysis of published RCTs comparing HFOV with CMV for ALI/ARDS in unique adult patients. All analyses were based on published data extracted from the six eligible studies, which have been approved by the Institutional Review Committee on Human Research. An additional file shows this in more detail (see Additional file
1). Additionally, as described in the six primary studies, all patients (or their representatives) enrolled in these six trials have provided written informed consent before any study-related procedure was performed. Therefore, the present meta-analysis does not present any further problems in relation to ethics or conflicts of interest.
Literature search and identification of the publications
To identify all published RCTs comparing HFOV with CMV in adult ARDS patients, a search of electronic databases (including PubMed, MedLine, Springer Link, Elsevier Science Direct, ISI web of knowledge, and EMBASE) was carried out with the following terms: acute respiratory distress syndrome; acute lung injury; high frequency oscillation ventilation. There were no language restrictions. This search was conducted through July 2013, with no additional time limits. The reference lists of the identified primary studies and relevant meta-analyses were also searched for additional studies.
To be included in the present meta-analysis, studies had to be RCTs comparing HFOV with CMV, enrolling unique adult ALI/ARDS patients, and reporting at least one of the following outcomes of interest: ICU mortality; 28- or 30-day mortality; hypoxemia and ventilation failure; duration of mechanical ventilation; and the incidence of barotrauma or hypotension. All of the candidate articles were independently read and checked for the inclusion criteria by two investigators (XG and GW). Disagreements were resolved through consensus. The methodological quality of the included studies was evaluated according to the Cochrane handbook 5.1.0 for randomized controlled trials [
29]. Given that blinding of physicians, patients or related family members was impossible in those trials, we compared whether rescue treatments were equally applied in the treatment group and control group, and performed quality assessment according to the following five aspects, including random sequence generation, allocation concealment, incomplete outcome data, selective reporting, and others.
Information was extracted independently by two investigators (XG and GW) from all eligible studies. The required items included in the data form were as follows: (1) basic information about the primary study, including the first author’s name, year of publication, sample size of the study, single or multicenter design, the definition of ALI or ARDS used, and the overall risk of bias; (2) clinically relevant primary outcomes, including ICU mortality, 28- or 30-day mortality, hypoxemia (including oxygenation failure and refractory hypoxemia diagnosed in the primary eligible studies) and ventilation failure (including ventilation failure, acidosis, and refractory acidosis diagnosed in the primary eligible studies), and duration of mechanical ventilation; and (3) the incidence of barotrauma or hypotension. The lists from the two investigators were compared, and disagreements about the extracted data were resolved by consensus.
Statistical analyses
All meta-analysis were performed by random-effects models (the DerSimoniane and Laird method). We reported continuous outcomes using standardized mean differences with the 95% CI, and binary outcomes were presented as relative risk with the 95% CI. The Z-test and chi-square test were used to generate the P-value for the continuous outcomes and for the binary outcomes, respectively. P <0.05 was considered statistically significant.
The presence of heterogeneity between studies was tested with the chi-square-based
Q-test and quantified with the
I2 statistic (25 to 49% for low heterogeneity, 50 to 74% for moderate heterogeneity, and 75 to 100% for high heterogeneity) [
30]. Sensitivity analysis was performed to evaluate the influence of the individual trial on the pooled effect. Potential publication bias was investigated by funnel plots and was formally evaluated with Egger’s linear regression test and Begg’s adjusted rank correlation test. All statistical analyses were performed with Stata software (version 11.0; StataCorp LP, College Station, TX, USA) using two-sided p values.
P <0.05 was considered statistically significant.
Discussion
To the best of our knowledge, the present study is the first meta-analysis examining the effect of HFOV in unique adults with ARDS, although two cognate Cochrane reviews [
27,
32] have been published based on studies including mutually exclusive groups of patients, and the more recent one [
32] combined the results from adult and pediatric patients. Importantly, the efficacy of HFOV is likely associated with the age of the patient, as Arnold
et al. reported that patients older than five years had dramatically increased mortality compared with patients younger than five years, when treated with HFOV [
33]. Additionally, the earlier Cochrane review [
27] found only two small RCTs and was not powerful enough to draw definitive conclusions. Although the later one [
32] included eight randomized controlled trials with 419 patients, it included two studies investigating the combination effect of HFOV and additional interventions (prone positioning [
34] and tracheal gas insufflation [
35]), which would complicate the results of meta-analysis.
In the present meta-analysis of mortality, we showed that HFOV did not significantly reduce mortality at 30 or 28 days compared with CMV. This finding contrasts sharply with experimental studies in animals in which benefits of high-frequency oscillation were observed [
36]. Our results may suggest that the benefits of HFOV cannot be translated directly from animal models to adult ARDS patients, probably because there is great heterogeneity in the recruitability of the lung [
37] and because the well-controlled conditions of animal studies are often difficult to replicate in human clinical trials. Our results are also at variance with those of the latest Cochrane review of HFOV in 2013 [
32], which showed HFOV significantly reduced in-hospital or 30-day mortality compared with conventional ventilation. This may be simply because the present meta-analysis enrolled two more large multicenter trials and recruited more than three times the number of patients recruited in the previous meta-analysis. Our results are consistent with those of two recently published large-scale RCTs, which are known as the OSCILLATE trial [
21] and OSCAR trial [
22], respectively. Sensitivity analysis showed that in the meta-analysis of mortality at 30 or 28 days, after each study was excluded from the overall meta-analysis, similar results were obtained. This suggests that our results of 30- or 28-day mortality are valid.
As moderate heterogeneity was revealed in the overall meta-analysis of mortality, and considering that the tidal volume and plateau airway pressure might be the sources of heterogeneity, we performed subgroup analysis according to the tidal volume (≤8 ml/kg predicted body weight) and according to the plateau airway pressure (≤35 cmH2O) in the control group, respectively. Although the heterogeneity remained moderate after the stratification by tidal volume, the heterogeneity was significantly reduced when the data were stratified depending on the control group’s plateau airway pressure. It appears that the plateau airway pressure is one of the most possible sources of heterogeneity. The results from these subgroup analyses all suggest that HFOV did not significantly reduce mortality at 30 or 28 days in adult ARDS patients. This indicates that the results of this meta-analysis are stable.
Although the meta-analysis of ICU mortality demonstrated that HFOV did not significantly affect ICU mortality compared with CMV, the sensitivity analysis showed that the OSCAR trial [
22] drastically affected the pooled results of our meta-analysis of ICU mortality. After the OSCAR trial was excluded, the pooled results suggested that the application of HFOV would significantly increase ICU mortality compared with CMV. Additionally, the beneficial effect of HFOV on mortality could have been underestimated because one study [
21] enrolled 40% of the patients included in our meta-analysis of ICU mortality, and in that study more than 10% of patients in the control group crossed over to receive HFOV. Thus, the result of this meta-analysis of the risk of death in ICU was not stable, and further clinical trials are needed to determine whether HFOV is as effective as CMV for the improvement of ICU survival in adult ARDS patients.
The present meta-analysis showed that HFOV significantly reduced the risk of hypoxemia compared with CMV. This result is consistent with our meta-analysis of the PaO
2/FiO
2 ratio on day 1 which demonstrated that the application of HFOV significantly improved the PaO
2/FiO
2 ratio on the first day after the initiation of mechanical ventilation (data not shown). However, the sensitivity analysis showed whenever the OSCILLATE trial [
21] or the study by Derdak
et al. [
18] was excluded, the pooled results of the other two studies suggested that HFOV did not significantly reduce the incidence of oxygenation failure. This may be simply because the other two trials enrolled too few patients to identify a significant difference. Thus, it is still possible that the application of HFOV significantly improves oxygenation and reduces the risk of hypoxemia.
There are several limitations of the present meta-analysis. First, the six enrolled studies were published between 2002 and 2013, which may have caused moderate heterogeneity of the control group and therefore complicated the results of this meta-analysis. Second, the sample size of the included trials ranged from 28 to 795, which could have influenced the precision of the pooled effect-estimates. Lastly, we only evaluated the effect of HFOV on the incidence of barotrauma and hypotension, therefore, the evaluation of the safety of HFOV might not be comprehensive enough.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors conceived the study and contributed to the study design. XLG and GNW performed the literature review and data extraction. YWY and DHS performed statistical analysis. GNW, DHS and YS analyzed and interpreted the data. XLG, DHS and YWY drafted the manuscript. XLG, GNW and YS were responsible for the revision of the manuscript for important intellectual content. YS supervised the study. All authors have read and approved the manuscript for submission.