Mechanical ventilator support remains the cornerstone of acute respiratory distress syndrome (ARDS) management. Many harmful effects of mechanical ventilation, such as ventilator-induced lung injury (VILI), have been recognized and led to the development of a lung-protective ventilatory strategy, mainly by keeping tidal volume (VT) low. A well-known study by the ARDS Network in 2000 demonstrated that low VT (6ml/kg predicted body weight, PBW) is better than high VT (12ml/kg PBW) in terms of mortality and ventilator-free days.1 The superiority of low VT in this study may stem more from avoiding the harmful effects of high VT (12ml/kg PBW) and plateau pressure (up to 50cmH2O) than strict adherence to low VT per se. This speculation was supported by 3 futile clinical trials in low VT before the year 2000.2–4 With VT around 10ml/kg PBW and plateau pressure less than 31cmH2O in the control groups, the beneficial effects of low VT were completely abolished in these 3 studies.5 Therefore, we hypothesize that intermediate VT (7.5–10ml/kg PBW), by shunning the deleterious effects of high VT, could also be lung-protective as low VT.

Two decades after the publication of the landmark study,1 low VT adherence remains poor throughout the world.6,7 A survey of Chicago physicians found that more than 92% knew that patients with ARDS warranted low VT ventilation, but they ordered low VT in only 7% (0–14%) of their eligible patients.8 The reasons for nonadherence to low VT ventilation are complex. Some physicians are skeptical about applying evidence derived from randomized control trials to critical care practice,9,10 some may concern complications of low VT such as hypercapnia,11 air hunger sensation12 and possible self-inflicted lung injury if there is no adequate sedation or paralysis.13 Therefore, many physicians adopt a less strict version of lung protection ventilation. Intermediate VT was found in some studies to be the most commonly applied ventilator setting for patients with ARDS.14,15

Intermediate VT has been shown to be as effective as low VT for ventilated patients without ARDS in terms of mortality and other clinical outcomes.16–18 However, the role of intermediate VT in patients with ARDS has not been carefully studied. Data inferred from a study of 111 real-world patients with ARDS suggest that the mean VT of 9.5ml/kg PBW is not inferior to 6.1ml/kg PBW in terms of 28-day or 1-year mortality rate.19 In this study, we retrospectively compared the clinical outcomes of patients with ARDS who received intermediate and low VT.

Totally 786 patients were invasively ventilated for their ARDS and acute respiratory failure. Four hundred and four patients were excluded because of 20 with extreme body weights, 19 without traceable clinical outcomes, 175 invasively ventilated for less than 48h, 2 ventilated by special modes, 70 received extracorporeal membrane oxygenation during the ARDS period, 4 withdrawn from a life-sustaining machine for hospice, 102 with pre-existing terminal illnesses and 12 without ventilator settings eligible for lung-protective ventilation. Therefore, 382 patients, with a total of 6958 ventilator settings eligible for lung-protective ventilation, were analyzed. A flowchart of patients included in the final analysis was presented in Fig. 1.

Figure 1.

Flowchart of patients included in the final analysis.

(0,53MB).

The patients were classified into low (mean VT=6.7ml/kg PBW, n=76, 19.9%), intermediate (mean VT=8.9ml/kg PBW, n=204, 53.4%) and high (mean VT=11.2ml/kg PBW, n=102, 26.7%) VT groups according to their mean VT while eligible for lung protection. Their baseline characteristics, treatment, and ventilator setting variables are summarized in Table 1. Intermediate and low VT groups did not differ in their baseline APACHE II or SOFA scores. Both groups also have a comparable baseline ARDS severity distribution. The driving pressures the patients received were not significantly different.

The outcomes of the three groups are shown in Table 2. Intermediate VT has similar 28-day mortality with the low VT group (47.1% vs. 63.2%, P=0.06). Intermediate and low VT groups did not differ significantly in other outcomes, such as 90-day mortality, ventilator-free days, ventilator-dependence rate, or barotrauma rate. By Cox regression model analysis, high VT (HR 1.78, 95% CI 1.08–2.94, P=0.03), male, low CRS, liver cirrhosis, and high PEEP were independently associated with mortality at 28 days (Table 3). Intermediate VT was not independently associated with 28-day mortality (HR 1.34, 95% CI 0.92–1.97, P=0.13). Factors independently associated with 90-day mortality include high VT (HR 1.62, 95% CI 1.06–2.49, P=0.03), age, male, hypertension, liver cirrhosis, malignancy, low CRS, high FIO2 and low PaO2 (Table s1). Again, intermediate VT was not independently associated with mortality at 90 days. The collinearities of FIO2, PaO2, and CRS were excluded due to their lower than 2 VIF values (Table s2).

Intermediate and low VT cohorts matched with the propensity score were developed. Their characteristics and outcomes are presented in Table 4. Both groups did not differ in mortality (Fig. 2) and all other clinical outcomes. Multivariate analysis revealed that low CRS and high FIO2 were independently associated with mortality rate of 28- or 90-days, while intermediate VT was not (Table s3).

Figure 2.

Kaplan–Meier survival curves by day 28 (A) and day 90 (B) for propensity score-matched patients of intermediate and low VT groups.

(0,08MB).

According to several surveys,14,17 intermediate VT was commonly used in patients with or without ARDS throughout the world. However, our study is the first report on the clinical outcomes of the use of intermediate VT in patients with ARDS.

Some animal studies have confirmed that high VT contributes to VILI.26,27 A recent study found that patients ventilated with VT of 12ml/kg PBW for as short as 4 days could induce lung inflammation.28 The well-known study by the ARDS Network showed that high VT has a worse outcome than low VT.1 Our results also confirm that high VT is independently associated with mortality at 28 and 90 days in patients with ARDS. However, strict adhesion to low VT may not be necessary. According to data from our study, intermediate VT, by shunning the deleterious effect of high VT, has comparable clinical outcomes with low VT.

Low VT is not by itself the only factor in preventing VILI. Amato et al. found that VT divided by CRS, or driving pressure, is most strongly associated with survival in ARDS.29 Gattinoni et al. coined the term ‘baby lung’ to describe a fraction of the lung parenchyma that maintains normal inflation in patients with ARDS. They argued that VT should be adjusted according to the size of the baby lung and the strain it received during mechanical inflation instead of ideal body weight.30 They incorporated VT and a bundle of respiratory parameters to measure the mechanical power lung received during ventilation,31 which is considered more accurate in predicting the likelihood of VILI.32 Since VT per se is not of utmost importance in preventing VILI, more strict control of VT (i.e., ultralow VT) failed to produce additional benefit as some researchers expected.33

Strictly low VT may not be necessary for all patients with ARDS. Several ARDS phenotypes have been identified. About 55% of patients who met the Berlin definition of ARDS do not have typical pathological diffuse alveolar damage. These patients tend to have milder symptoms and shorter clinical courses.34 About 10–17% of patients with ARDS were extubated or no longer met the criteria for ARDS in less than 24h.14,35 These subsets of ARDS with their distinct clinical course may warrant personalized treatment. A recently published French trial tested personalized treatment by giving patients with ARDS with focal involvement a VT of 8ml/kg PBW and those without focal involvement a VT of 6ml/kg PBW in addition to a bundle of other related ventilatory maneuvers. The per protocol analysis showed that patients in the personalized treatment group have a survival advantage over the control group, who universally received a VT of 6ml/kg PBW.36

To classify VT as low or not, some observational studies on ARDS only counted the VT patients received in the initial few days.6,19 However, this way of counting inappropriately neglected the influences of VT patients received in subsequent days. We adopted the method of Needham et al. by counting all VT patients received for up to 28 days if their ventilator settings were considered eligible for lung protection.23 We recorded VT three times per day, more frequently than Needham (twice per day). Based on the mean of all VT patients received throughout the whole ventilation courses, we believe our categorization of VT is more accurate than many previous observational studies on ARDS.

The VT received by our patients were not randomly assigned, but were given according to the in-charge doctors’ choice. We find those receiving low VT have lower CRS (Table 1), which made higher VT inappropriate because the limitation of plateau pressure could easily be exceeded. Whereas those with higher CRS were more likely to receive intermediate, rather than low VT. This practice was in line with the global tendency toward nonadherence to low VT6–8 as we have mentioned in our introduction.

Prone positioning has been proved effective for moderate to severe ARDS patients,37 but only 12% and 20% (from intermediate and low VT group respectively, both with a median of PaO2/FIO2<150mmHg) of our patients received this adjunctive therapy. This was just another example of discrepancy between clinical trial and clinical practice. This trend was also universal. The LUNG SAFE study, involving 50 countries around the world, found prone was used in only 6% and 16% of the moderate and severe ARDS patients respectively.14 A recent survey of moderate-to-severe ARDS patients in the US found only 6% of them received prone in their early management.38 Efforts are needed to find reasons behind the widespread nonadherence.

Liver cirrhosis was found to be an independent risk factor for 28- and 90-day mortality in our patients with ARDS. This finding was in accordance with previous studies.39,40 Increased pro-inflammatory interleukine-6 and interleukine-8 in patients with decompensated cirrhosis were thought to contribute to lung injury in those at risk.41

There are several limitations to this study. First, it was a retrospective observation. The classification of VT groups was not assigned randomly. Selection bias and unrecognized confounders are possible. Second, our data were all from one center. The generalizability of our conclusion can be limited. Third, for patients with pressure-targeted ventilation, plateau pressures were not measured directly. We used the peak airway pressure or the sum of PEEP and set increment of inspiratory pressure instead.23 The plateau pressure derived by this way is prone to over-estimation.42 Driving pressure and CRS calculated from this putative plateau pressure were all subject to imprecision.

In conclusion, we found that high VT is harmful to patients with ARDS. Intermediate and low VT and have similar clinical outcomes. Our results suggested that intermediate VT is an acceptable option for ventilated patients with ARDS. This conclusion needs to be verified by randomized control trials.

The study was supported by grant number 108-CCH-IRP-012 (Changhua Christian Hospital).

SHW drafted the manuscript. CTK analyzed the data and performed statistical calculations. CYL collected clinical data from study patients. YCH gave final approval of the version to be published.

No conflict of interest to declare.