Comparison of conventional mechanical ventilation and high-frequency oscillatory ventilation in congenital diaphragmatic hernias: a systematic review and meta-analysis

Outcomes of conventional mechanical ventilation (CMV) and high-frequency oscillatory ventilation (HFOV) in patients with congenital diaphragmatic hernia (CDH) were compared through a systematic review and meta-analysis. Outcome measures included mortality and incidence of chronic lung disease (CLD). Odds ratio (OR) and 95% confidence interval (95%CI) were evaluated. Subgroup analyses were performed according to the strategy for applying HFOV in CDH patients. Group A: CMV was initially applied in all CDH patients, and HFOV was applied in unstable patients. Group B: chronologically analyzed. (CMV and HFOV era) Group C: CMV or HFOV was used as the initial MV. Of the 2199 abstracts screened, 15 full-text articles were analyzed. Regarding mortality, 16.7% (365/2180) and 32.8% (456/1389) patients died in CMV and HFOV, respectively (OR, 2.53; 95%CI 2.12–3.01). Subgroup analyses showed significantly worse, better, and equivalent mortality for HFOV than that for CMV in group A, B, and C, respectively. CLD occurred in 32.4% (399/1230) and 49.3% (369/749) patients in CMV and HFOV, respectively (OR, 2.37; 95%CI 1.93–2.90). The evidence from the literature is poor. Mortality and the incidence of CLD appear worse after HFOV in children with CDH. Cautious interpretation is needed due to the heterogeneity of each study.


Data extraction
Systematical extraction of the data was performed regarding study design, study period, mortality, subgroup, and the incidence of CLD from the included studies.A data extraction form was used to record the results.

Subgroup analysis
All studies included in this meta-analysis compared the outcomes of CMV and HFOV.However, heterogeneity was observed regarding the strategy for applying HFOV.Some studies applied HFOV in selected patients who showed persistent unstable vital signs after being initially ventilated with CMV.Others applied HFOV at the first timing of MV and selected either CMV or HFOV.In these studies, CMV was not routinely used as the initial MV strategy.The others compared the historical outcomes of CMV and HFOV.The outcomes of CMV before introducing HFOV were compared with the outcomes of HFOV.Herein, subgroup analyses were performed according to the strategy for applying HFOV in CDH patients.In group A, CMV was initially applied in all CDH patients, and HFOV was applied in unstable patients showing poor outcomes in CMV.In group B, the results of CMV and HFOV were chronologically analyzed.The use of HFOV was more recent than the use of CMV in all studies in group B. In group C, either CMV or HFOV was used as the initial MV, which is different from group A, which used CMV as the routine initial MV strategy.

Statistical analysis
Statistical analysis was performed using R × 64 4.1.2.Pooled estimates of odds ratios (ORs) with 95% confidence intervals (95%CIs) were evaluated using standard methods.Both fixed and random-effects models were used.Heterogeneity between studies was assessed using the I 2 statistic.The forest plot and funnel plot were used to estimate the overall result and publication bias, respectively.

Results
Of 2199 studies found by defined search strategy, 1636 abstracts were reviewed, and 88 full-text articles were examined.Of these, 15 studies were included in this study [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] (Fig. 1).There was one prospective randomized clinical trial.Table 1 summarizes the presentation center, patient enroll timing, and main results of the study.Mortality rate were reported in all studies and the incidence of CLD was reported in 6 papers among the 15 included studies.Various countries, including the United Kingdom, the United States, France, and Canada conducted the studies.Patients were included from 1981 to 2017.Most of them were retrospective studies and one paper was prospective RCT.

Mortality
The pooled result of all enrolled studies for mortality showed significantly higher mortality in HFOV (Fig. 2).In total, there were 456 deaths in HFOV (n = 1389) and 365 deaths in CMV (n = 2180).The ORs were 2.53 for the fixed model and 2.16 for the random-effect model, respectively; 95%CIs did not contain the value 1.In I 2 statistics, heterogeneity was observed to be 86%, higher than the 50% cut-off, indicating high heterogeneity between studies.
In subgroup analysis, Group A had statistically significantly higher mortality in HFOV group with OR 3.18 in fixed model.(Fig. 3) Group B had significantly lower mortality in the HFOV group with OR 0.22 in fixed model.In group C, OR was 1.69 in fixed model, but it was not statistically significant.The funnel plot for overall mortality is reported in Fig. 4.

The incidence of CLD
Regarding the incidence of CLD, the OR and 95%CI were 2.37 and 1.93-2.90,respectively, for the fixed model.In other words, the incidence of CLD was significantly higher in HFOV (Fig. 5).The funnel plots for the incidence of CLD are presented in the supplemental materials.

Discussion
CDH that needs not only various medical supports according to the underlying condition of the patients but also requires surgical treatment of diaphragmatic defect is a complex disease that shows high mortality rate even with modern developed perinatal care.The mortality rate is known to reach 40-100% in low-, middle-income countries whereas reported as 30% in developed countries 2,19 .Careful perinatal management is required, especially for respiratory system, pulmonary HTN, and cardiac dysfunction, which are critical for the patient's survival.
The most important point of the respiratory management is to use management that can prevent ventilatorinduced lung injury 20 .The introduction of gentle ventilation contributed to the improvement of survival rate.Delayed repair and extracorporeal membrane oxygenation contributed to increase the survival rate by 50% until the early 1990s, and permissive hypercapnia improved the survival rate by an additional 10% 19 .Mechanical www.nature.com/scientificreports/ventilation strategies such as HFOV and HFJV have been introduced in clinical practice and are being applied for the unstable CDH patients 21,22 .This meta-analysis compares the outcome of CMV and HFOV of CDH in over three thousand new-born children with CDH.This study shows that the evidence for the comparison of HFOV and CMV for CDH is scarce.In subgroup analysis, the patients with HFOV showed better survival in group B in which showed chronological analysis of the outcomes of CMV and HFOV.www.nature.com/scientificreports/A pooled result of all enrolled studies for mortality is higher in HFOV than in CMV.(OR: 2.53, 95%CI 2.12-3.01)Since there are heterogeneities regarding the application of HFOV all through the enrolled studies, subgroup analysis was performed.Group A in which applied HFOV in the unstable patients who were initially applied CMV showed statistically significant higher mortality in HFOV (OR: 3.18, 95%CI 2.63-3.84).Group B in which analyzed the outcomes of CMV before the introduction of HFOV and of HFOV after the introduction of HFOV showed statistically significant better survival in HFOV.Group C in which applied either of CMV or HFOV at the timing of the application of MV did not show statistically significant result.The outcome of HFOV  www.nature.com/scientificreports/could be explained with the characteristics of subgroups.The mortality of group A might be higher in HFOV because the HFOV was selectively applied in poor patients with the use of CMV.More accurate results could be elicited if a randomized controlled study was completed comparing the effectiveness of HFOV and CMV in unstable patients with CDH.Further, those patients in group A where CMV did not treat respiratory acidosis properly may have a benefit from ECMO.HFOV often delays ECMO and therefore survival rate could drop.Early ECMO may have better results 23 .Since group B is a comparison of different periods before and after the introduction of HFOV, other advanced perinatal care other than HFOV may have affected the outcome, resulting in a low mortality rate in HFOV.There are many developments in the care of CDH not only in the area of mechanical ventilation strategy.Surfactant use, fraction of inspired oxygen, the concept of gentle ventilation, the importance of the management of pulmonary hypertension 24 .The improvement in mortality of CDH could also be related improving prenatal care.Also, in the younger era without protocols concerning to standard of care and ventilator settings, cares were not limited as recommended in a standardized protocol 25 .Group C showed no significant difference, which can be explained as follows.The use of HFOV may have no additional benefit compared to CMV in patients who are stable with CMV alone.Unstable patients could benefit with HFOV whereas either HFOV or CMV could offer similar results in relatively stable patients.Of interest a study published in 2016 by Snoek (Group C in this analysis), suggests that HFOV caused overinflation and had a longer ventilation period, which may have caused ventilator-induced lung injury by more tracheal suction.HFOV could have limitations than CMV, which can cause deterioration in vulnerable patients 26 .In the VICI Trial high initial ventilator settings were allowed and there was also a trend to a later initiation of rescue ECMO after initial application of HFOV. 4 A future study concerning to HFOV should be done with lower settings of HFOV and early ECMO as a rescue treatment.Chronic lung disease is known to occur in 30-50% of CDH survivors 20 .It is known that the incidence of CLD is higher when ECMO is used and patch repair is performed 27 , but there is insufficient evidence that HFOV reduces CLD incidence 28 .In this meta-analysis, the incidence of CLD was higher in HFOV.Because the number of the enroll studies was small, subgroup analysis like mortality could not be performed.As in mortality, it may have occurred more in HFOV due to baseline characteristic differences between groups more than the difference of the MV modality.Therefore, it seems difficult to reach a clear conclusion.
This study is limited by the heterogeneity of indications for using HFOV or CMV.The evidence from the systematic review is poor due to low level of evidence (a single randomized controlled trial).Also, in further studies risk stratifications should be done by some prenatal information like lung volumes or liver position, or analysis should be done in only isolated CDH patients without other major malformations or syndromes.One study suggested to make a distinction between mild moderate and severe CDH 29 .
On the basis of the poor existing evidence, mortality and the incidence of CLD appear worse after HFOV compared to CMV.Cautious interpretation of this result is needed and only a well-designed prospective study involving large number of patients could address this issue.

Figure 1 .
Figure 1.Diagram of workflow in the systematic review.

Table 1 .
Overall results of meta-analysis comparing conventional mechanical ventilation and high-frequency oscillatory ventilation.