Mortality Associated With Acute Respiratory Distress Syndrome 2009-2019: A Systematic Review and Meta-regression

Background: Acute respiratory distress syndrome (ARDS) is a common occurrence in an intensive care unit. The reported mortality in studies evaluating acute respiratory distress syndrome is highly variable. The adherence to ventilatory specic and adjunctive therapies is also highly variable. We investigated the mortality of ARDS since the 2009 H1N1 pandemic and examined the adherence to ventilatory specic and adjunctive therapies. Methods: We performed a systematic search in MEDLINE and EMBASE using a highly sensitive criterion from January 2009 to May 2019. We then ran a proportional meta-analysis for mortality and a meta-regression analysis using certain variables to address heterogeneity. Results: We screened 5361 citations, of which 85 fully met inclusion criteria. The weighted pooled mortality of all 85 studies published from 2009 to 2019 was 38% (95% CI 35,40). Mortality was higher in observational studies [40% (95% CI 37, 42)] compared to RCTs [35% (95% CI 30,39)], (p=0.04) Signicant variability exists in literature of reported tidal volumes, positive end expiratory pressures, plateau pressures, and use adjunctive therapies. The tidal volumes in our systematic review ranged from 5.8 to 8.9 ml/kg with a mean of 7.2 ml/kg. PEEP ranged from 4.6 to 16.1 cm H2O at the time of enrollment with a mean of 10.2 cm H2O. Reported plateau pressures ranged from 21.0 to 35.1 cm H2O, with a mean of 25.6 cm H2O. Higher reported initial PaO2/FiO2 ratios were associated with decreased mortality. A trend towards decreased mortality was seen with lower reported tidal volumes in the included studies. Conclusions: Over the last decade, the mortality in ARDS marginally and there exists signicant heterogeneity


Background
Acute respiratory distress syndrome (ARDS) accounts for approximately 10 percent of intensive care unit (ICU) admissions, with a mortality rate ranging from 35-45 percent(1). Phua et al showed that the mortality among observational studies and randomized controlled trials remained static from the implementation of the 1994 AECC de nition for ARDS to 2006 (2). Since the publication of that meta-analysis, an in uenza pandemic with a high prevalence of ARDS occurred, a new de nition of ARDS was developed (3) and several landmark randomized controlled trials evaluating various interventions were published (4)(5)(6)(7)(8)(9).
Over the last two decades lung-protective ventilation (LPV) strategies [tidal volume (Vt) 6-8 mL/kg predicted body weight and low plateau airway pressures (Pplat) (<30cmH2O)] remains the mainstay of ARDS management. Adherence to LPV strategies has consistently shown to improve patient survival (10). Use of Prone Position ventilation (PP) (9) in patients with moderatesevere ARDS is the only other therapeutic intervention that has been associated with improved survival. Use of continuous neuromuscular blocking agents (NMBA) (4,11) has been disparate results amongst the two major trials that have evaluated this question. Other interventions including use of higher positive end expiratory pressure (PEEP) (12,13), inhaled pulmonary vasodilators (14,15) and diuretics (16) improve oxygenation and duration of mechanical ventilation respectively, but their use has not been associated with a mortality bene t. Based on the current evidence early, consistent use of extracorporeal membrane oxygenation (ECMO) cannot be justi ed, and this therapy needs to be considered after failure of conventional mechanical ventilation (5,17,18). But more importantly independent of the quality or strength of evidence the application of all these therapies in ARDS remains inconsistent and is predominantly in uenced by physician comfort and discretion (1). As a result, signi cant heterogeneity is seen in the adoption and application of these evidence-based therapies in the published literature.
We conducted a systematic review and meta-analysis to investigate mortality associated with ARDS since the 2009 H1N1 pandemic. We evaluated temporal changes in mortality over the study period. Due to the presence of heterogeneity in application of conventional and adjunctive management strategies in ARDS patients, we also evaluated the impact of speci c intervention such as reported tidal volume, PEEP and use of adjunctive therapies on the reported mortality in the included studies.

Methodology Search strategy
We electronically searched MEDLINE and EMBASE (January 2009 to May 2019) using a highly sensitive strategy to identify the relevant studies. For potentially relevant articles the full text was obtained for review; for these articles, all references were inspected to supplement our search. Details of the search strategy are reported in the supplementary les. We limited our search strategy to articles published in English.

Study selection
We included both randomized controlled trials (RCT) and observational studies for the purpose of this systematic review. Using standardized criteria, two reviewers (DS and KS) reviewed titles and abstracts identi ed by the search strategy independently and in duplicate, retrieving studies that either reviewer thought relevant for full-text review. Disagreements between reviewers in study selection and data extraction were resolved by the senior author (AD). We selected observational studies and RCTs enrolling at least 50 adults with acute lung injury (ALI) / acute respiratory distress syndrome (ARDS) and reporting mortality.
We only included studies where 100% of the sample met any criteria for ARDS. We excluded reports available only in abstract form, duplicate reports and animal studies. and all reported mortality. Driving pressure was calculated using the reported PEEP and Pplat (19). The data from RCTs and observational studies that contained multiple arms were combined for analysis (if reported in means and standard deviation).
If a study contained an arm dedicated to investigating an adjunctive therapy versus control, only the control arm from that study was included, so as to pertain to all ARDS patients. Any missing data is reported in the study table in the supplementary appendix. The primary outcome was short term mortality. Short term mortality was de ned as hospital mortality where reported, as it was the most often reported. If not reported then ICU mortality, 90-day mortality, 60-day mortality, and 28/30-day mortality were substituted, in this order of preference.

Risk of bias assessment
To assess for risk of bias, two authors independently reviewed all included studies. The tool Cochrane RoB-2 was utilized to assess RCTs (20). Using this tool we evaluated ve different domains for each RCT and determined an overall risk of bias. The ROBINS-1 tool was utilized for observational studies (21). Using this tool we evaluated seven different domains for each observational study and determined an overall risk of bias. The overall risk of bias for both RCTs and observational studies was determined by the highest risk allotment in any of the categories.

Data analysis
Baseline characteristics between observational studies and RCTs were compared using student's t test and chi squared tests, for continuous and categorical variables respectively. We performed a proportion meta-analysis using random-effects models to obtain pooled estimates of mortality and 95% con dence intervals (CIs) for all observational studies and all RCTs separately (22). We used Cochran's Q statistic and I2 to test for heterogeneity among studies (23). To further explore heterogeneity among studies we formed logistic meta-regression models to evaluate the association between selected variables (PaO2/FiO2 ratios, Vt, PEEP, Pplat, driving pressure, mean age, APACHE II, SOFA, NMBA, PP, and ECMO) and mortality (24). We also ran a cumulative mortality analysis using median year of enrollment to evaluate the trend of mortality from our literature collection from 2009 to 2019 (25). A p value of £ 0.05 was considered to be statistically signi cant. Funnel plot analysis and Egger's test were done to investigate for publication bias (26,27). The statistical analysis was conducted using Stata Version 15.1 (StataCorp LP, College Station, TX.). Two sensitivity analysis were also conducted to address certain aspects of our study and their effects on our ndings. One focusing on combining both the intervention and control arms of studies which focused on one adjunctive therapy and another including only on studies reporting hospital mortality.

Study Selection
Our search strategy yielded 5361 citations after de-duplication. We reviewed full texts for 1523 articles for a detailed evaluation and included 85 articles in our qualitative assessment (  (Table 1). There were no differences in the baseline characteristics, severity of initial illness and use of ventilatory strategies between the RCT's and observational studies ( Table   1). The lack of asymmetry on the Funnel plot and the result of the Egger's test imply that publication bias did not alter the results (e- Figures 5-6).

Quality assessment
The risk of bias for RCTs was low in twelve studies, moderate in fourteen RCTs, and high in four RCTs (e- Table 2). The high risk of bias was driven mainly by deviation from intended intervention. For observational studies the risk of bias low in two studies, moderate in forty-one studies, and high in twelve studies (e- Table 3). The high and moderate risk of bias was driven mainly by confounding given the nature of study design.
There was signi cant heterogeneity among the included studies (I 2 = 92.23%, p<0.01). This heterogeneity persisted across both observational studies (I 2 = 90.21%, p<0.01) and RCTs (I 2 = 93.87%, p<0.01) (Figure 2).  Figures 9-11). Mean age, APACHE II, and SOFA also did not have any impact on mortality in our meta-regression (e- Figures 12-14). Reported adjunctive therapies were signi cantly variable in the included studies and did not have any impact on the mortality reported in our meta-regression (e- Figures 15-18). In the sensitivity analysis, which only included studies that reported hospital mortality, repeating meta-regression analysis for the same variables mentioned above produced identical results (e- Table 4).

Discussion
Our meta-analysis demonstrates a minor reduction in ARDS associated mortality since 2009. Compared to results reported by Phua et al, the cumulative reported mortality has dropped 6.9% from 44.3% to 38%(2) over the last decade. Similar to other studies, the mortality is consistently higher in observational studies compared to RCTs. The well described impact of initial severity of hypoxemia is seen in our meta-analysis and is strongly associated with mortality among the included studies.
Our study also shows that there remains signi cant heterogeneity and inconsistency in the reporting of key therapeutic interventions, markers of severity of illness and ventilatory strategies for patients with ARDS. This inconsistency in reporting makes the comparison of outcomes amongst these studies very di cult. Despite the development of the Berlin De nition (3) and a call for consistent reporting, this remains a problem among the studies that we evaluated. As seen in the LUNG SAFE study, ARDS still remains an underdiagnosed disease process and 40% of patients meeting ARDS criteria are never diagnosed(1). These ndings suggest that diagnosis of ARDS is often delayed, with a high likelihood of delay in treatment for this diagnosis. This is especially concerning given that many of the ARDS treatments with proven bene t have only demonstrated a bene t in early ARDS (4,9,29).
The signi cant variability in use of evidence-based interventions in studies evaluating patients with ARDS is of concern. The use of ventilator speci c variables, such as low tidal volume and PEEP, and adjunctive therapies, such as inhaled vasodilators, NMBAs, HFOV, PP, and ECMO, exhibit great inconsistency amongst the studies. This non-adherence to therapies with proven bene t (4,9,30) among the included studies might be a signi cant driver to the differences in outcomes reported by different studies independent of the severity of included patients. Similar to ndings in LUNG SAFE few studies mentioned the plateau pressures for the patients included in their studies and the use of adjunctive therapies is inconsistent and highly variable outside of studies evaluating a speci c intervention. LUNG SAFE study revealed that less than two-thirds of patients received low tidal volume ventilation, 82.6% of patients received a PEEP < 12 cm H 2 O, 37.8% of patients with severe ARDS received NMBAs, and only 16.3% of patients with severe ARDS were proned (1). The absence of standardization in the implementation of these therapies makes it di cult to discern their impact on outcomes. The low implementation of these therapies is surprising given a mean PaO 2 /FiO 2 ratio of 132.5 mm Hg at baseline. The overall lack of standardization and implementation of best practices observed in our study highlights the need for protocol driven ARDS management that allow the clinicians to select the most appropriate adjunctive therapies for their patient. Personalization of care to patients with ARDS may be indicated in the future as suggested by colleagues Constantin et al (31), but not before therapies with proven bene t have been systematically standardized and adopted universally. Implementation of standardized ARDS management may help to further decrease ARDS associated mortality.
Over the last decade the overall mortality reported among RCTs has remained static since the last meta-analysis(2), and remains much higher than the suggested benchmark for ARDS trials (32). Signi cant mortality bene t has only been seen with isolated interventions such as PP in PROSEVA (9). The reported mortality among large epidemiologic studies in ARDS over the last two decades has not changed much. The lower reported mortality in RCTs compared to observational studies is not surprising as RCTs usually ensure strict adherence to protocols, with the probable exclusion of patients with a poor prognosis (33,34). Cumulative mortality for all ARDS studies shows a less than 7% change in reported mortality across time.
With the application of evidence based ventilatory and non-ventilatory therapeutic interventions we had expected to see a much larger impact on mortality in the current meta-analysis. But a lack of reporting key variables and a wide variability in the reported numbers for these variables, brings to light a signi cant problem that the application of these therapies are not as widely consistent as we would hope when we are caring for ARDS patients. In many cases the intervention being studied, or cointerventions of interest are tightly accounted for in individual studies, but the complex care of these patients is not nearly enough a protocolized consistent approach that we would hope for.
Our systematic review and meta-analysis has several strengths. We conducted a comprehensive literature search using broad search terms. We included studies from institutions across ve continents, and report an international evaluation of the trend in ARDS mortality, as opposed to the recent ndings by Zhang et al (35). The inclusion criteria were carefully prede ned and carried out in a methodological fashion. Additionally, we only included studies which exclusively evaluated ARDS patients. This is the rst meta-analysis evaluating ARDS mortality trends to include a meta-regression analysis evaluating ARDS treatment modalities and their association with ARDS mortality.
Despite the strengths of our meta-analysis, there are potential limitations. First, we used hospital mortality as the primary mortality in our analysis as it was most frequently reported (46 of 85 studies). With a reported incidence of mortality of 3-15% in ARDS after discharge from the ICU (36,37), utilization of hospital mortality as our primary mortality type may have impacted our overall reported ARDS-related mortality. When hospital mortality was not available it was substituted with ICU, 90 day, 60 day, and 28/30 day mortality, in that order of preference. However, to tackle this potential inadequacy, we conducted a sensitivity analysis only including those studies reporting hospital mortality which produced identical results. Another limitation is the reporting of ventilator speci c variables only at the time of patient enrollment in the majority of included studies. Although this may not represent the overall ventilator management strategy received by patients, it provides an insight into the initial management strategy, which is associated with the greatest mortality bene t. Not all studies may have reported the use of adjunctive therapies, possibly accounting for the low numbers we discovered. Finally, we included all diagnoses criteria for ARDS in our systematic review and meta-analysis and we did not stratify results based on ARDS severity. This unlikely impacted overall mortality results, however, as there was no difference in the PaO 2 /FiO 2 ratio or severity scores at baseline for included studies.

Conclusions
Our systematic review and meta-analysis observed a minimal decline in ARDS related mortality over the last decade. We also saw that there remains a signi cant heterogeneity in reporting of both ventilator strategies and adjunctive therapies amongst the published literature. Despite established guidelines there is variability in the overall management of ARDS. Increased clinician education regarding the importance of early recognition and implementation of best practices may help to reduce ARDS related mortality (38). Future studies should evaluate standardized ARDS treatment protocols, which implement evidencebased best practices, and their impact on mortality. Financial disclosures: The authors have no nancial support to report for this manuscript.
Funding/Support: None Data availability: The data used is readily available using search strategy provided.
Ethical approval and consent to participate: Was not necessary given this was a systematic review and meta-analysis of previously conducted studies Consent for publication: No individual data presented.