The relationship of tidal volume and driving pressure with mortality in hypoxic patients receiving mechanical ventilation.

PURPOSE: To determine whether tidal volume/predicted body weight (TV/PBW) or driving pressure (DP) are associated with mortality in a heterogeneous population of hypoxic mechanically ventilated patients.
METHODS: A retrospective cohort study involving 18 intensive care units included consecutive patients ≥18 years old, receiving mechanical ventilation for ≥3 days, with a PaO2/FiO2 ratio ≤300 mmHg, whether or not they met full criteria for ARDS. The main outcome was hospital mortality. Multiple logistic regression (MLR) incorporated TV/PBW, DP, and potential confounders including age, APACHE IVa® predicted hospital mortality, respiratory system compliance (CRS), and PaO2/FiO2. Predetermined strata of TV/PBW were compared using MLR.
RESULTS: Our cohort comprised 5,167 patients with mean age 61.9 years, APACHE IVa® score 79.3, PaO2/FiO2 166 mmHg and CRS 40.5 ml/cm H2O. Regression analysis revealed that patients receiving DP one standard deviation above the mean or higher (≥19 cmH20) had an adjusted odds ratio for mortality (ORmort) = 1.10 (95% CI: 1.06-1.13, p = 0.009). Regression analysis showed a U-shaped relationship between strata of TV/PBW and adjusted mortality. Using TV/PBW 4-6 ml/kg as the referent group, patients receiving >10 ml/kg had similar adjusted ORmort, but those receiving 6-7, 7-8 and 8-10 ml/kg had lower adjusted ORmort (95%CI) of 0.81 (0.65-1.00), 0.78 (0.63-0.97) and 0.80 0.67-1.01) respectively. The adjusted ORmort in patients receiving 4-6 ml/kg was 1.26 (95%CI: 1.04-1.52) compared to patients receiving 6-10 ml/kg.
CONCLUSIONS: Driving pressures ≥19 cmH2O were associated with increased adjusted mortality. TV/PBW 4-6ml/kg were used in less than 15% of patients and associated with increased adjusted mortality compared to TV/PBW 6-10 ml/kg used in 82% of patients. Prospective clinical trials are needed to prove whether limiting DP or the use of TV/PBW 6-10 ml/kg versus 4-6 ml/kg benefits mortality.

Introduction

Previous literature suggests that modifying the ventilator parameters, tidal volume/predicted body weight (TV/PBW) and/or driving pressure (DP), improves survival from acute respiratory distress syndrome (ARDS). However, methodological shortcomings of that literature and limitations of the diagnostic criteria for ARDS described below prevent straightforward translation of these results into population-based quality improvement.

The optimal specific approach to titrating ventilator parameters remains unclear. The most widely-recommended approach, low tidal volume ventilation (LTVV) specifically targets TV/PBW 4–6 ml/kg if tolerated by the patient, but may range up to 8 ml/kg [1, 2]. Clinical trials that support the use of LTVV largely focused on patients with ARDS and compared LTVV with high control tidal volumes ranging 10–15 ml/kg [1-8]. Tidal volumes this high are rarely clinically employed [9], and few data are available to compare LTVV with intermediate tidal volumes (ITVV) such as in the range of 8–10 ml/kg commonly used in current practice.

A meta-analysis of LTVV clinical trials suggested that driving pressure (DP), rather than TV/PBW, is the modifiable ventilator parameter independently associated with mortality in patients with ARDS [10], but prospective trials of DP-limiting ventilator management have not been performed. A clinical practice guideline endorsed by multiple international professional societies recommends LTVV for patients with ARDS [2]. Some have recommend LTVV for virtually all mechanically ventilated patients [11-13]. Others recommend a DP-limiting ventilator strategy [14, 15].

Much of the previous literature on modifiable ventilator parameters focused on patients with ARDS. However, the diagnosis of ARDS may be difficult to operationalize in clinical practice. ARDS is conceptually a clinical-pathological entity [16] diagnosed using clinical-radiological criteria [17] for which clinicians and researchers have demonstrated limited inter-rater reliability and accuracy [18-22]. Furthermore, current data suggest that limiting TV/PBW or DP may also benefit patients without ARDS [11–13, 15, 23–25], suggesting that making the diagnosis of ARDS is not essential to deciding whether to limit TV/PBW or DP. In fact, an observational study including 459 ICUs in 50 countries showed that after adjusting for potentially confounding variables, clinical recognition of ARDS did not significantly influence the choice of TV/PBW, or whether DP was measured [9]. Population-based quality improvement efforts would be greatly simplified if the necessity to distinguish patients with ARDS from others with similar hypoxic ventilatory failure could be circumvented.

Our goal has been to implement evidence-based ventilator practice in a large healthcare system to improve survival of mechanically ventilated patients. But in order to do so, we needed to better understand the relationship between modifiable ventilator parameters and mortality in a heterogeneous population of hypoxic mechanically ventilated patients. The specific aims of our study were: 1) To determine whether DP, TV/PBW, or subcategories of TV/PBW (4–6, 6–7, 7–8, 8–10, and >10 ml/kg) were independently associated with adjusted hospital mortality in hypoxic adult patients receiving ≥3 days of mechanical ventilation, regardless of whether they met ARDS diagnostic criteria, and 2) to establish discrete thresholds for optimal TV/PBW and/or DP.

Methods

Study design

This retrospective cohort study was performed as part of a quality improvement project, and was determined by our institution’s Research Determination Committee to not require Institutional Review Board review. The study setting included the medical/surgical, cardiovascular, neurological, transplant and trauma intensive care units (ICUs) of 18 acute care hospitals within a large healthcare system in the southwestern United States between February 1, 2017 and January 31, 2019.

Participants

Consecutive patients were included based on the inclusion criteria: ≥18 years of age, received volume control mechanical ventilation for at least three days, had a PaO2/FiO2 ratio ≤300 mmHg while receiving PEEP ≥5 cmH2O during the first 24 hours of mechanical ventilation, met criteria for calculation of an APACHE IVa score, had height recorded (with which to calculate PBW). Patients with less than three ventilator days were excluded in order to focus the analysis on patients more likely to accrue lung injury from prolonged mechanical ventilation. All patients were followed-up until hospital discharge or death.

Variables

The main outcome variable was hospital mortality. The two predictor variables of interest were the modifiable ventilator parameters: TV/PBW and DP. Potential confounding variables included age, PaO2/FiO2 ratio, PaCO2, respiratory system compliance (CRS), APACHE IVa® predicted hospital mortality, hospital site, and the annual quarter in which the patient was admitted.

Data sources

We used previously described bioinformatics [26, 27] embedded within a Cerner Millenium® electronic medical record (EMR) and an honest broker system to collect de-identified clinical and ventilator parameters on all patients receiving mechanical ventilation. These data included: age, gender, height (cm), PaO2 (mmHg), PaCO2 (mmHg), FiO2, positive end-expiratory pressure (PEEP cmH20), set tidal volume (ml), and plateau pressure (cmH20). The first complete set of data for each patient obtained within 24 hours of intubation was used in the analysis described below. We used the APACHE IVa® severity scoring system (Cerner Corp, Kansas City MO) to enumerate hospital mortality, predicted hospital mortality and ventilator days.

Study size

We calculated that 1248 patients were needed per TV/PBW stratum (for instance comparing patients receiving 4-6ml/kg to those receiving 8-10ml/kg) to provide 80% power to discern a 5% difference in mortality, assuming baseline mortality of approximately 25%.

Statistical analysis

The following values were calculated for each patient: PaO2/FiO2, PBWmen = 50 + 0.91(cm of height—152.4); PBWwomen = 45.5 + 0.91(cm of height—152.4), TV/PBW, CRS = TV/ (P -PEEP) (ml/cmHO) and DP = TV/C (cmHO).

The APACHE IVa® severity scoring system (Cerner Corp, Kansas City MO) incorporated chronic health conditions, 115 discrete admission diagnostic categories, and 27 clinical variables, including age, vital signs, Glasgow Coma Scale score, FiO2, PaO2, PaCO2, arterial pH, urine output, creatinine, bilirubin, albumin, glucose, white blood cell count and hematocrit, with a reported a discriminant accuracy of 88% for predicting hospital mortality [28].

We performed three step-wise forward multiple logistic regression (MLR) analyses to investigate the association between TV/PBW, DP and hospital mortality with adjustment for confounders. The first MLR incorporated TV/PBW and all potential confounders listed above; the second included DP, TV/PBW and all confounders. Next we segregated patients into five strata of TV/PBW: (4–6, 6–7, 7–8, 8–10, and >10 mL/kg), chosen a-priori based on their relation to the design of multiple previous clinical trials [1–5, 11–13, 23, 29], and performed the third MLR forcing the five TV/PBW strata into the model as nominal variables, including DP and all potential confounders. We used this MLR to calculate the adjusted odds ratio for mortality (ORmort) with 95% confidence intervals for each TV/PBW strata. We used the TV/PBW 4–6 ml/kg strata as the referent, based on the proposition that this strata (strict LTVV) theoretically represents best practice. In a post-hoc analysis, we combined three strata that had similar ORmort including tidal volumes in the 6–10 ml/kg range and used it as the referent to calculate the adjusted ORmort in the 4–6 ml/kg stratum.

We used the significant ORmort for a 1 standard deviation (SD) increase in DP to calculate a threshold DP (mean DP + 1 SD) above which adjusted mortality was significantly increased.

Post-hoc MARS and sensitivity analysis

When the relationship between TV/PBW and mortality was observed to be non-linear based on the first MLR analysis described above, we performed post-hoc multivariate adaptive regression splines (MARS) analysis of the relationship between TV/PBW and adjusted mortality.

We used STATA® Version 15 (Statacorp, College Station, TX) for all statistical analyses.

Results

Our CDS system identified 21,851 discrete episodes of mechanical ventilation received by 20,703 patients during the 2-year study period. 20,057 (96.9%) of these patients received volume control mechanical ventilation, and 14,320 (72.4%) had a PaO2/FiO2 ratio ≤300 mmHg during the first 24 hours of mechanical ventilation. 5,658 (39.5%) of patients with a PaO2/FiO2 ≤300 mmHg accrued ≥3 ventilator days. Three hundred and forty-seven (6.2%) of these did not have a height recorded, and 144 (2.6%) failed to satisfy criteria for APACHE IVa® mortality prediction and were excluded ().

Flow diagram of cohort inclusion/exclusion criteria.

[Abbreviations: APACHE: Acute physiology and chronic health evaluation, APRV: Airway pressure release ventilation, MLR: Multiple logistic regression, PC: Pressure control, PPLAT: Plateau pressure, PS: Pressure support, DP: Driving pressure].

Clinical characteristics of the 5,167 study patients are shown in . The mean age was 61.9 years, and 42.4% were women. The most common admission diagnoses were pneumonia (30.0%), non-pulmonary sepsis (10.8%), cardiopulmonary arrest (10.2%), respiratory failure not due to pneumonia (8.8%), and non-cardiovascular surgery (8.7%). The mean APACHE IVa® score was 79.3, and mean PaO2/FiO2 ratio was 166 mmHg. The mean applied TV/PBW was 7.25 ml/kg. PPLAT was recorded in 4,490/5,167 (86.9%) patients, from which CRS and DP were calculated, yielding means of 20.7 cmH2O, 40.5 ml/cmH2O, and 13.7 cmH2O, respectively. The median ventilator length of stay was five days (IQR: 4–9 days), and overall hospital mortality was 28.4% (95%CI: 27.1–29.6%). The distribution of TV/PBW received is shown in

shows the model resulting from the first MLR analysis of the relationship between TV/PBW and hospital mortality. Age, APACHE predicted hospital mortality, PaO2/FiO2, and admission to two particular hospitals out of the 18 participating in the study were significantly associated with mortality, but TV/PBW taken as a continuous variable, was not.

Post-hoc MARS analysis confirmed a significant U-shaped relationship between TV/PBW and adjusted mortality, which was down-sloping (slope -0.17, p = 0.001) as TV/PBW increased from 4 to 7.1 ml/kg, and up-sloping (slope 0.20, P = 0.001) as TV/PBW increased above 7.1 ml/kg.

shows the results of second MLR analysis, of the relationship between DP, TV/PBW and hospital mortality. DP was found to be significantly associated with mortality after adjustment for significant confounders with an ORmort of 1.10 (95% CI: 1.06–1.13; p = 0.009) for each one standard deviation (SD) increase in DP. Seven hundred twenty-six patients received DP one standard deviation or greater above the mean (≥19 cmH2O), with an ORmort of 1.15 (95%CI: 1.01–1.30), representing an estimated 33 deaths attributable to excessive DP. Age, APACHE predicted hospital mortality, PaO2/FiO2, and admission to two individual hospitals were also significant in this MLR model, which had Nagelkerke’s pseudo-R2 of 0.143.

shows the results of third MLR analysis of the relationship between five strata of TV/PBW, DP and all significant confounders. Patients receiving 4–6 ml/kg were used as a referent for comparison with the other groups–ie. the OR for each of the other strata were compared to the 4–6 ml/kg strata. Patients receiving 6–7 ml/kg and 7–8 ml/kg had significantly lower adjusted ORmort (0.81 p = 0.05, and 0.78 p = 0.03 respectively. Patient’s receiving 8–10 ml/kg had adjusted ORmort of 0.80 with p = 0.06. The stratified analysis results from are illustrated in . Since the adjusted ORmort for TV 6–7, 7–8 and 8–10 ml/kg were all similar, we combined them together in post-hoc analysis, and using them as the referent group, the comparative adjusted ORmort for patients receiving 4–6 ml/kg was 1.26 (95% CI: 1.04–1.52; p = 0.02).

Adjusted odds ratio for mortality for five strata of TV/PBW (4–6, 6–7, 7–8, 8–10 and >10 ml/kg).

The median TV/PBW for each strata is plotted on the X-axis. Adjusted ORmort with 95% CI error bars plotted on the Y-axis for the second through fifth strata. [The first stratum (4–6 ml/kg) is the referent for the calculation of OR for each of the other strata].

Discussion

We report a large observational cohort study designed to examine the relationship between modifiable ventilator parameters and hospital mortality in hypoxic patients receiving mechanical ventilation. Our study cohort comprised about 25% of all patients receiving mechanical ventilation in our healthcare system, with a crude mortality of 28%, comparable to that of mild ARDS [17, 30]. Our patients were also similar to patients with ARDS in terms of age, PaO2/FiO2, CRS, TV and PPLAT [9, 30]. A recent epidemiologic study suggests that more than 50% receiving mechanical ventilation with hypoxic respiratory failure, such as those included in our study, meet Berlin criteria for ARDS [30].

We adopted a pragmatic approach to patient selection, breaking with the convention of using ARDS as a selection criteria. ARDS is conceptually a clinical-pathological entity [16], but is currently diagnosed using imperfect clinical-radiological criteria [17] for which clinicians have demonstrated poor accuracy and low inter-rater reliability [18, 19]. Previous studies have shown that even researchers with expertise in ARDS have only moderate agreement when applying the clinical diagnosis of ARDS [20-22]. In contrast, the PaO2/FiO2 ratio criteria we used to select our study patients was easy to accurately extract from the EMR and represents the only ARDS criteria independently associated with mortality [17]. This approach to patient selection makes our study unique in the context of related literature and eases translation of our results into population-based quality improvement.

In our first two multivariate analyses, we found that DP rather than TV/PBW was the modifiable ventilator parameter independently associated with survival. But our stratified analysis and multivariate adaptive regression splines analysis showed a significant U-shaped relationship between TV/PBW and adjusted mortality. Patient receiving 4–6 ml/kg had similar mortality to those receiving >10 ml/kg. But patients receiving 6–7, 7–8 and 8–10 ml/kg all had comparative adjusted ORmort about 0.80. These findings raise two provocative hypotheses. The current range of LTVV (4–8 ml/kg) might be composed of substrata with distinctly different mortality effects: TV in the range of 6-8ml/kg may be safer than 4–6 ml/kg. Intermediate tidal volume ventilation (ITVV) might not be inferior to LTVV. Disregarding previous convention, we might hypothesize the optimal TV/PBW range to be 6–10 ml/kg based on these observational findings. Using TV 6–10 ml/kg as the referent group, the comparative adjusted ORmort for patients receiving 4–6 ml/kg, such as targeted in the ARMA study, is 1.26 (95% CI: 1.04–1.52; p = 0.02).

A U-shaped relationship between TV/PBW and mortality was first posited by Eichacker and colleagues in their 2002 meta-analysis of ARDS trials testing LTVV [31], and was later described in a cohort of 5,183 mechanically ventilated patients that found that those receiving ≤6 ml/kg and those receiving >10 ml/kg had increased mortality (OR 1.23 and 1.14 respectively) compared to patients receiving 6–10 ml/kg (p = 0.09) [32]. A U-shaped relationship was not demonstrated in a much smaller cohort study (N = 485) that was not specifically powered to do so [33].

A U-shaped relationship between TV/PBW and mortality has sound physiological explanation [31]. Tidal volumes that are too high lead to alveolar overdistention, ventilator-induced lung injury, systemic inflammation, ventilator non-triggering, and eccentric respiratory muscle injury [1, 34–37]. Tidal volumes that are too low can cause hypercarbic acidosis, increased work of breathing, and patient-ventilator dyssynchrony [38-40]. The later can manifest as strenuous inspiratory efforts and double-triggering, either of which can paradoxically lead to alveolar overdistention [40-43]. Insufficient TV/PBW can also cause atelectrauma, increased respiratory rate (stress frequency) and increased sedation requirements [44]. The use of LTVV in the subset of ARDS patients with relatively preserved CRS has been shown to be associated with increased mortality [45].

Current evidence supporting LTVV has several important limitations. Clinical trials used to support LTVV [1, 3–8] did not perform comparative analysis of TV/PBW substrata comprising LTVV, and used tidal volumes (10–15 ml/kg PBW) in their control groups that were significantly higher than the intermediate tidal volumes commonly used in clinical practice at the time [31, 32, 46, 47]. Several authors have posited that the apparent benefit of LTVV in these trials might be solely attributable to the injuriously high tidal volumes and PPLAT received by control patients rather than to any specific benefit of LTVV [31, 48]. Deans and colleagues analyzed 2,587 patients who met enrollment criteria for the landmark ARMA trial [1], but were excluded for technical reasons such as lack of consent; these patients went on to receive conventional tidal volume management, yet achieved mortality of 31.7%—comparable to the 31.0% mortality of study patients that received LTVV and significantly lower than the 39.8% mortality experienced by study patients randomized to the control group [45]. In 2017, an international consensus of critical care societies published the results of a meta-analysis of nine randomized controlled trials comparing LTVV (4–8 ml/kg PBW) to “traditional” tidal volumes (10–15 ml/kg PBW) in patients with ARDS. They found no difference in mortality (risk ratio: 0.87–95% CI: 0.70–1.08) in their primary analysis [2] and provided no data comparing LTVV to ITVV, yet “strongly recommended” LTVV. Less than 5% of patients in a large epidemiological study of ARDS [9] and less than 3% of patients in our cohort received tidal volumes in the range >10 ml/kg used in the control groups of the studies included in this meta-analysis, so it’s relevance to current practice would be questionable even if it’s findings had been significant.

As early as 2002, Eichacker posited that ITVV (with limited PPLAT) was not inferior, and might be superior to LTVV, based on a patient data-level meta-analysis of ARDS clinical trials [31]. The only previous randomized controlled trial we are aware of that specifically compared LTVV to ITVV included 961 patients without ARDS, and showed no difference in mortality, ventilator days, length of stay, or pulmonary complications [29].

Although we cannot calculate the proportion of our cohort with ARDS, it is noteworthy that 20 years after the ARMA trial supporting LTVV 4-6ml/kg [1], only 14.7% of our patients received it, compared to the majority 82% receiving 6–10 ml/kg. Several previous reports similarly show that although most clinicians agree that LTVV should be used in patients with ARDS, only 7–19% of their ARDS patients receive it [49, 50]. An observational study including 459 ICUs in 50 countries showed that tidal volumes of 6–10 ml/kg were used in approximately 75% of ARDS patients, four times more often than 4-6ml/kg was used, and that clinical recognition of ARDS did not significantly influence the TV/PBW administered [9]. Our study supports this practical approach to ventilator management, even though it sometimes incorporates non-recommended ITVV. It is possible that clinicians are correctly choosing higher range LTVV and ITVV ventilation to avoid complications of strict 4–6 mg/kg LTVV.

Our observation that DP is related to mortality supports and extends the findings of a prior clinical trial meta-analysis [10], by showing that the relationship between DP and mortality holds in a heterogenous group of hypoxic patients, regardless of whether they have ARDS, even though clinical measurement of CRS by respiratory care practitioners in the presence of spontaneous inspiratory efforts could introduce significant error into the measurement or calculation of PPLAT, CRS and DP [14]. The later concern has previously been posited as an unresolved barrier to clinical implementation of DP-limited mechanical ventilation [42]. It is worth noting that DP equals TV/CRS−therefore titration of DP can be seen as a form of precision medicine in which TV is matched to an individual patient’s respiratory system compliance.

Several studies provide context for our finding of a relationship between DP and mortality in hypoxic ventilated patients. A meta-analysis of clinical trials and observational studies [51] and three subsequent cohort studies [15, 30, 52] showed a significant relationship between DP and mortality in patients with ARDS. In patients without ARDS, a quasi-experimental trial [25], and two cohort studies [15, 24] showed that DP was associated with mortality. Of note, the latter study showed a significant relationship only in non-ARDS patients with PaO2/FiO2 <300 mmHg. One recent observational study that did not require hypoxia as an inclusion criterion showed no relationship between DP and mortality in non-ARDS patients [52]. Reported threshold values for DP associated with increased mortality are in a relatively narrow range of 14–19 cmH2O [10, 30, 53], consistent with our results. Taken together, considerable data support the contention that reducing DP <19 cmH2O would make a reasonable target to reduce mortality in mechanically ventilated patients.

Limitations

Residual bias/confounding remains a major threat to the validity of our observational study. The heterogeneity of our cohort and the real-world shortcomings of clinical data used in our study reduce internal validity, but increase the translational utility of our findings. Our exclusion of patients with less than three ventilator days could have introduced bias if such patients are especially prone to ventilator induced lung injury. We used only a single set of ventilator parameters per patient to describe ventilator management–a practical decision based on the large amount of clinical data we accessed. Our simplified calculation of CRS did not take into account spontaneous patient effort. Elevated DP could be a marker of more severe underlying lung injury rather than a cause of increased mortality, notwithstanding our statistical adjustment for CRS, PaO2/FiO2 and other confounders. We used APACHE IVa® predicted hospital mortality as an independent variable in our MLR, which has reduced discriminant accuracy in mechanically ventilated patients [54]. Although our MLR model was superior to APACHE IVa®, it had only a modest pseudo-R2 for predicting mortality. We therefore suspect that lowering DP may have only a modest effect on mortality of mechanically ventilated patients, in whom myriad other factors likely influence survival. Prospective clinical trials are needed to further investigate whether ITVV is superior to strict LTVV, and to demonstrate clinical efficacy of DP-limiting ventilator strategies.

Conclusions

Driving pressures ≥19 cmH2O were associated with increased adjusted hospital mortality in our retrospective cohort. We intend to use this finding to inform development of clinical decision support focused on limiting DP rather than achieving strict LTVV in our healthcare system. Tidal volumes of 4–6 ml/kg were used in less than 15% of patients and were associated with increased adjusted mortality. Tidal volumes of 6–10 ml/kg were used in 82% of patients and had significantly reduced adjusted mortality compared to that associated with tidal volumes 4–6 ml/kg. We hypothesize that ITVV may not be inferior to strict LTVV in patients with hypoxic respiratory failure.

(XLSX)

Click here for additional data file.

14 Jun 2021

PONE-D-21-17095

The Relationship of Tidal Volume and Driving Pressure with Mortality in Hypoxic Patients Receiving Mechanical Ventilation.

PLOS ONE

Dear Dr. Raschke,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jul 29 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Corstiaan den Uil

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"This study was funded in part by grant 2196 from the Flinn Foundation. SP was funded by NIH

(HL126140, HL151254, AG059202, AI135108, HL140144, HL128954) and PCORI (DI-

2018C2-13161, EADI-16493, CER-2018C2-13262) at the time of writing this manuscript."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"No. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: In a retrospective review of over 5000 patients and involving 18 intensive care units, the authors establish the relationship between two mechanical ventilation parameters: driving pressure (DP) and tidal volume/predicted body weight (TV/PBW) and mortality. The study concluded that: (1) driving pressure is significantly associated with mortality, consistent with the findings of Amato et al, and therefore needs to be considered when titrating ventilator settings and (2) that low-tidal volume ventilation (LTVV) may not be the optimal mechanical ventilation strategy for all patients but rather that intermediate-tidal volume ventilation (ITVV) may lead to superior patient outcomes. I applaud the authors on their study and in challenging the concept of LTVV, which has long-deserved additional reflection given the inability to replicate the results of the initial ARMA study, and in the absence of longitudinal change in ARDS-related mortality.

Major Comments:

The authors demonstrate that both DP and VT/PBW significantly modify mortality risk, however they did not provide data comparing DP to the (five) VT/PBW strata. This is important because if the (average) driving pressures were similar (but with different VT/PBW), this suggests that the compliance of the patients was different. This might suggest that the differences in mortality may have been modified by underlying patient lung disease severity/compliance rather than the VT/PBW.

Minor Comments:

- These study results suggest that the ARMA trial did not establish that 6mL/kg was optimal but that 12mL/kg was sub-optimal. In support of the findings of this study, it would be worth including a mention of the article published by Deans et al (Crit Care Med 2005 – PMID: 15891350) in the discussion as that study retrospectively analyzed the ARMA data and demonstrated that the mortality rates between the group of patients receiving 6mL/kg and those who were excluded from the study and receiving then-conventional ventilation (ITVV) had similar mortality rates. [They also demonstrated that the impact of tidal volumes on mortality was related to lung compliance or - in essence, but not in name - driving pressure.]

- Recommend avoiding the use of contractions

- Although ITVV was defined as an abbreviation (on the "Abbreviations" page - 3), its first use in the text is in the discussion (page 17). Recommend defining it in the text prior to its use.

Reviewer #2: In this retrospective cohort, Dr Raschke et al studied over five thousand patients mechanically ventilated with PF ratios below 300 mHg in 18 ICUs in the southwestern United States. They found that tidal volume had a U-shaped relationship with hospital mortality and that driving pressure was also associated with hospital mortality even after adjusting for disease severity. I have some concerns and suggestions as detailed below.

1) My major concern is that the increased mortality in low tidal volume (<6mL/Kg PBW) be due to residual confounding. For example, patients with higher PEEP values due to hypoxemia might end up receiving lower tidal volumes to limit plateau pressure to less than 30 cmH2O as per protective ventilation protocols. I suggest that the authors include PEEP in the multivariable adjustment.

2) Respiratory system compliance can be a marker of the underlying severity of lung disease, but also varies according to patient lung size. Normalizing compliance to PBW (in mL/cmH2O/Kg PBW) helps take into account patient size. I suggest that the authors use normalized compliance in their multivariable adjustment.

3) Sample size was computed to find differences between groups. What groups? Please explain.

4) Why did the authors use ANOVA to compare the different TV/PBW strata? I believe that adding the strata to the multivariable logistic regression (as dummy variables) would have been a more natural choice.

5) Please explain in further detail how the driving pressure threshold of 19 cmH2O was found.

6) Please add the unit for the PF ratio (mmHg)

7) Is tidal volume still significantly associated with survival when driving pressure is in the model? This information is relevant because – if not – it would be simpler to target protection in terms of driving pressure.

8) Please discuss why the findings are in disagreement with those of Needham et al (BMJ 2012). In that prospective cohort study, they found that tidal volume had a linear relationship with survival. Modeling tidal volume with cubic splines did not improve the relationship.

9) How is it possible to reconcile the authors’ finding to those from Prevent trial (Simonis JAMA 2018) in which they tested a strategy with low vs intermediate tidal volumes in patients without ARDS?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Michaela Kollisch

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

21 Jul 2021

Reviewer #1: In a retrospective review of over 5000 patients and involving 18 intensive care units, the authors establish the relationship between two mechanical ventilation parameters: driving pressure (DP) and tidal volume/predicted body weight (TV/PBW) and mortality. The study concluded that: (1) driving pressure is significantly associated with mortality, consistent with the findings of Amato et al, and therefore needs to be considered when titrating ventilator settings and (2) that low-tidal volume ventilation (LTVV) may not be the optimal mechanical ventilation strategy for all patients but rather that intermediate-tidal volume ventilation (ITVV) may lead to superior patient outcomes. I applaud the authors on their study and in challenging the concept of LTVV, which has long-deserved additional reflection given the inability to replicate the results of the initial ARMA study, and in the absence of longitudinal change in ARDS-related mortality.

Major Comments:

The authors demonstrate that both DP and VT/PBW significantly modify mortality risk, however they did not provide data comparing DP to the (five) VT/PBW strata. This is important because if the (average) driving pressures were similar (but with different VT/PBW), this suggests that the compliance of the patients was different. This might suggest that the differences in mortality may have been modified by underlying patient lung disease severity/compliance rather than the VT/PBW.

We agree with the reviewer and further note that TV/PBW is mathematically associated with driving pressure (DP= TV/CRS). Both are ways of expressing the TV, either in relation to the patients body size, or in relation to their respiratory compliance (CRS). This relationship was demonstrated by Amato and colleagues in their retrospective analysis of ARDSnet clinical trial data.

We calculated DP for each of the five increasing strata of TV/PBW and they were positively associated as the reviewer suspected they would be: 12.8, 12.8, 13.7, 15.2, and 16.7 cmH2O respectively. However, compliance did not appear to be associated with increasing TV/PBW, with compliances of: 38.7, 43.3, 40.3, 37.4 and 40.7 ml/cmH2O respectively in the five strata of increasing TV/PBW.

Compliance was adjusted for in all three of our MLRs and MARS analyses and therefore the associations we reported should not have been confounded by differences in compliance. We used the following additional covariates in our MLR to also adjust for severity of illness: PaO2/FiO2 ratio, PaCO2, and APACHE IVa predicted hospital mortality. But it is likely that these few variables do not eliminate confounding as suggested by the reviewer . We added the statement: “Residual bias/confounding remains a major threat to the validity of our observational study” to the discussion of our study limitations.

Minor Comments:

- These study results suggest that the ARMA trial did not establish that 6mL/kg was optimal but that 12mL/kg was sub-optimal. In support of the findings of this study, it would be worth including a mention of the article published by Deans et al (Crit Care Med 2005 – PMID: 15891350) in the discussion as that study retrospectively analyzed the ARMA data and demonstrated that the mortality rates between the group of patients receiving 6mL/kg and those who were excluded from the study and receiving then-conventional ventilation (ITVV) had similar mortality rates. [They also demonstrated that the impact of tidal volumes on mortality was related to lung compliance or - in essence, but not in name - driving pressure.]

The article by Deans was fascinating and we greatly appreciate the reference We have added it to support our discussion in several places, focusing on the equalivalent outcomes in study patients receiving LTVV and in patients excluded for technical reasons. We also thought the association Deans found between compliance and the response to LTVV could help explain the U-shaped relationship between TV/PBW and mortality that we describe and also added that to the discussion.

- Recommend avoiding the use of contractions. Agreed and corrected throughout.

- Although ITVV was defined as an abbreviation (on the "Abbreviations" page - 3), its first use in the text is in the discussion (page 17). Recommend defining it in the text prior to its use. Agreed and corrected. ITVV is now defined in the introduction.

Reviewer #2: In this retrospective cohort, Dr Raschke et al studied over five thousand patients mechanically ventilated with PF ratios below 300 mHg in 18 ICUs in the southwestern United States. They found that tidal volume had a U-shaped relationship with hospital mortality and that driving pressure was also associated with hospital mortality even after adjusting for disease severity. I have some concerns and suggestions as detailed below.

1) My major concern is that the increased mortality in low tidal volume (<6mL/Kg PBW) be due to residual confounding. For example, patients with higher PEEP values due to hypoxemia might end up receiving lower tidal volumes to limit plateau pressure to less than 30 cmH2O as per protective ventilation protocols. I suggest that the authors include PEEP in the multivariable adjustment.

We agree that confounding remains a major concern in our observational study. We selected potential confounders carefully when we developed the methods of our study. One factor in that consideration was that independent variables in the model should be mathematically independent of each other. Therefore we did not include plateau pressure or PEEP as independent variables, but instead used driving pressure (which was calculated as Pplat minus PEEP).

In specific response to the reviewers concern, we calculated mean PEEP values in the five TV/PBW strata and they ranged from 6.2 to 7.9 mmHg – a difference of no more than 1.7mmHg between any two TV/PBW strata. Multiple clinical trials, including the ARDSnet trial, have failed to demonstrate a relationship between PEEP and mortality in patients with and without ARDS (1-3). Therefore, it seems unlikely that a difference in PEEP levels equal to, or less than 1.7 mmHg between groups would have a significant effect on the mortality in our study. We therefore respectfully declined reanalyzing our data based on this post-hoc consideration. It is our preference in general to stay true to the analysis we planned a-priori, while recognizing it’s limitations.

We therefore added the statement: “Residual bias/confounding remains a major threat to the validity of our observational study findings” to the discussion of our study limitations.

1) The National Heart, Lung, and Blood Institute ARDS Clinical Trials Network. Higher versus Lower Positive End-Expiratory Pressures in Patients with the Acute Respiratory Distress Syndrome. N Engl J Med 2004; 351:327-33. DOI: 10.1056/NEJMoa032193

2) Writing Committee and Steering Committee for the RELAx Collaborative Group. Effect of a Lower vs Higher Positive End-Expiratory Pressure Strategy on Ventilator-Free Days in ICU Patients Without ARDS: A Randomized Clinical Trial. JAMA. 2020;324(24):2509–2520. DOI:10.1001/jama.2020.23517

3) Walkey AJ, Del Sorbo, L, Hodgson CL. Higher PEEP versus Lower PEEP Strategies for Patients with Acute Respiratory Distress Syndrome. A Systematic Review and Meta-Analysis. Annals American Thoracic Society 2017;17 Ann Am Thorac Soc Vol 14, Supplement 4, pp S297–S303 https://doi.org/10.1513/AnnalsATS.201704-338OT

2) Respiratory system compliance can be a marker of the underlying severity of lung disease, but also varies according to patient lung size. Normalizing compliance to PBW (in mL/cmH2O/Kg PBW) helps take into account patient size. I suggest that the authors use normalized compliance in their multivariable adjustment.

We considered this suggestion with caution since one of the aims of our study was to compare selecting TV based on PBW (TV/PBW) versus selecting TV based on compliance (CRS), since driving pressure = TV/CRS. Although CRS is not mathematically independent from driving pressure, we tried to avoid further mathematical-linking of variables in our model except when compelled.

On review of landmark literature, we found that CRS not CRS/PBW, was used by the ARDS definition task force (4), the ARDSnet trial group (5), and the LUNG-SAFE investigators (6) among others, in their multivariate regression models. We do not find consensus in the literature that CRS/PBW is superior to CRS for such modeling. Again, we respectfully refer to our practice of restraint in reanalyzing data based on post-hoc considerations, and holding true to the analysis planned a-priori.

4) The ARDS Definition Task Force*. Acute Respiratory Distress Syndrome: The Berlin Definition. JAMA. 2012;307(23):2526–2533. doi:10.1001/jama.2012.5669

5) Amato, MBP, Meade MO, Slutsky AS, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med 2015; 372:747-755. DOI: 10.1056/NEJMsa1410639

6) Bellani G, Laffey JG, Pham T, et al. Epidemiology, Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units in 50 Countries. JAMA. 2016;315(8):788–800. doi:10.1001/jama.2016.0291

3) Sample size was computed to find differences between groups. What groups? Please explain.

We changed this section of methods in response to the reviewer’s suggestion: “We calculated that 1248 patients were needed per TV/PBW stratum (for instance comparing patients receiving TV/PBW <6ml/kg to those receiving 8-10ml/kg) to provide 80% power to discern a 5% difference in mortality, assuming baseline mortality of approximately 25%.”

4) Why did the authors use ANOVA to compare the different TV/PBW strata? I believe that adding the strata to the multivariable logistic regression (as dummy variables) would have been a more natural choice.

We appreciate the reviewer pointing out this statistical error. We performed the analysis as the reviewer recommended after reconsulting with our statistician, who agreed that the use of MLR was preferable to ANOVA. This analysis is identified in the methods section as the third MLR analysis. This caused our main statistical expression of mortality risk to change from observed/expected mortality to adjusted odds ratio for mortality. It changed the significance of some of our between-strata comparisons, although our conclusions remained generally robust. We subsequently had to redo table 4 and figure 3 using the new analysis, and adjust multiple parts of the paper, from the abstract through the conclusions.

5) Please explain in further detail how the driving pressure threshold of 19 cmH2O was found.

We clarified this per the reviewer’s suggestion in the methods and results sections of the manuscript. We simply chose 1 standard deviation above the mean DP. The odds ratio for mortality for a 1 SD increase in DP was 1.10; 95% CI: 1.06 -1.13, p=0.009. 726/4490 of our cohort patients who had DP measured fell into this group, which seemed a reasonably-sized target group for clinical decision support.

6) Please add the unit for the PF ratio (mmHg)

Done.

7) Is tidal volume still significantly associated with survival when driving pressure is in the model? This information is relevant because – if not – it would be simpler to target protection in terms of driving pressure.

TV was not significantly associated with survival in either of our first two MLR models – with or without driving pressure in the model. Therefore we concluded that we were going to limit driving pressure as our quality-improvement target. We state this in our conclusions: “We intend to use these findings to inform development of clinical decision support focusing on limiting DP rather than achieving strict LTVV in our healthcare system”.

We did not expect or hypothesize a U-shaped relationship between TV/PBW and mortality a priori, and therefore our first two MLR analyses were not designed to detect such an association. We would likely have had to add-in quadratic transformations of TV/PBW in order to do so with MLR. But once we observed the U-shaped relationship, we felt it was worth further investigation. We performed a separate post-hoc MARS analysis, which statistically confirmed a significant U-shaped relationship. Thus, although TV/PBW was not a significant predictor in either of the first two MLR models (with or without DP), that doesn’t mean it’s not related to mortality. Our MLR models could not have detected a significant U-shaped relationship the way they were originally designed. It was only demonstrable when a post hoc statistical test designed to complex relationships was performed, and when TV/PBW strata were forced into the third MLR model as dummy variables (as you suggested).

We don’t expect that our observational trial will appreciably change the widespread use of TV/PBW anytime soon, and we think that the hypothesis that intermediate TV may not be inferior to LTVV will remain clinically important and interesting to clinicians in the foreseeable future. We note in our discussion that a prospective clinical trials will be needed to confirm the hypotheses generated by our study.

8) Please discuss why the findings are in disagreement with those of Needham et al (BMJ 2012). In that prospective cohort study, they found that tidal volume had a linear relationship with survival. Modeling tidal volume with cubic splines did not improve the relationship.

We read Needhams study with great interest and appreciate the reference. The primary reason Needham might have failed to detect a U-shaped relationship is that their study was not powered to do so. Their sample size was selected to provide 85% power to compare two equally-sized patient groups with a hazard ratio of 0.7. Notably, their study had N=485 as compared to Eichacker’s study N=5,183 which showed the U-shaped relationship observed in our study (N=5,167).

There were other differences between the studies including Needham’s use of multiple TV/PBWs per patient over time vs. our use of a single TV/PBW per patient, and Needham’s choice of long-term (2-year) mortality as the main outcome variable vs our use of hospital mortality. Needham used Cox proportional hazards modeling vs MLR - of note, all their covariates were chose “a priori” (including CRS – rather than CRS/TBW). We have mentioned and referenced Needham’s study in our discussion section.

9) How is it possible to reconcile the authors’ finding to those from Prevent trial (Simonis JAMA 2018) in which they tested a strategy with low vs intermediate tidal volumes in patients without ARDS?

In our paper, we discuss that current recommendations strongly support the use of low tidal volumes in ARDS and perhaps for all mechanically ventilated patients, but epidemiological studies show that clinicians commonly use intermediate tidal volumes. In this context, our findings and those of the PREVENT trial are complimentary. Both studies failed to show that LTVV as currently recommended is superior to ITVV. Therefore, we referenced the PREVENT trial as supporting our hypothesis. Both papers support the common clinical practice of often employing ITVV for patients instead of strict LTVV.

Respectfully and appreciatively,

Robert A Raschke MD MS

Corresponding Author.

Submitted filename: Responsetoreviewerstevecurry.docx

Click here for additional data file.

26 Jul 2021

The Relationship of Tidal Volume and Driving Pressure with Mortality in Hypoxic Patients Receiving Mechanical Ventilation.

PONE-D-21-17095R1

Dear Dr. Raschke,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Corstiaan den Uil

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

30 Jul 2021

PONE-D-21-17095R1

The Relationship of Tidal Volume and Driving Pressure with Mortality in Hypoxic Patients Receiving Mechanical Ventilation.

Dear Dr. Raschke:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Corstiaan den Uil

Academic Editor

PLOS ONE

Table 1

Characteristics of 5,167 study patients.

	Mean	SD
	(unless otherwise noted)	(unless otherwise noted)
Female gender	42.4%
Age (years)	61.9	16.9
PBW (kg)	64.3	10.9
APACHE IVa ® score	79.3	28.0
ICU admission diagnosis categories:	N
Pneumonia with or without sepsis	1548 (30.0%)
Sepsis (non-pulmonary)	559 (10.8%)
Cardiorespiratory arrest	526 (10.2%)
Pulmonary (not pneumonia)	453 (8.8%)
Surgery (non-cardiovascular)	448 (8.7%)
Cardiology	345 (6.7%)
Neurology	343 (6.6%)
Surgery (cardiovascular)	321 (6.2%)
Trauma	238 (4.6%)
Toxic/metabolic	176 (3.4)
Gastrointestinal	137 (2.7%)
All other	73 (1.4%)
Pulmonary parameters:
PaO2/FiO2 (mmHg)	165.6	70.0
CRS (ml/cmH2O)	40.5	32.1
Ventilator settings/parameters
Tidal volume (ml)	457.4	74.9
TV/PBW (ml/kg)	7.25	1.34
PEEP (cmH2O)	7.0	3.0
	5.0 (median)	IQR: 5–10
PPLAT (cmH2O)	20.7	6.3
DP (cmH2O)	13.7	5.4
Outcomes
Hospital mortality	28.4%
APACHE IVa® Obs/Exp mortality	0.980
Ventilator days	5 (median)	IQR: 4–9

Table 2

Significant independent variables in the first MLR model:	Odds ratio* (95% CI)	P value
Age	1.16 (1.09–1.24)	<0.001
APACHE predicted mortality	1.99 (1.86–2.14)	<0.001
PaO 2 /FiO 2	0.87 (0.81–0.94)	<0.001
Admission to hospital X	1.64 (1.32–2.04)	<0.001
Admission to hospital Y	0.56 (0.36–0.86)	0.008

*Odds ratios are associated with a one standard deviation (SD) increment in the given variable, except in the case of admission to hospitals X or Y. Values used for SD: Age, 16.9 years; risk of death 26%; PaO2/FiO2, 70.

Table 3

Significant independent variables in the second MLR model:	Odds ratio* (95% CI)	P value
Driving pressure	1.10 (1.06–1.13)	0.009
Age	1.17 (1.09–1.26)	<0.001
APACHE predicted mortality	1.98 (1.90–2.05)	<0.001
PaO 2 /FiO 2	0.87 (0.80–0.94)	0.002
Admission to hospital X	1.74 (1.54–1.95)	<0.001
Admission to hospital Y	0.53 (0.44–0.68)	0.006

*Odds ratios are associated with a one standard deviation (SD) increment in the given variable, except in the case of admission to hospitals X or Y. Values used for SD: Age, 16.9 years; risk of death 26%; PaO2/FiO2, 70; DP, 5.42 cmH2O.

Table 4

Adjusted odds ratio for mortality in five strata of TV/PBW.

	Strata of TV/PBW (ml/kg)
	4–6	6–7	7–8	8–10	>10
N	760	1672	1554	1026	155
Median TV/PBW	5.8	6.5	7.5	8.6	11.0
Multivariate ORmort mortality (95% CI)	1.00*	0.81	0.78	0.80	1.03
		(0.65–1.00)	(0.63–0.97)	(0.63–1.01)	(0.67–1.59)
P-value from MLR (comparing adjusted ORmort to that of the 4–6 ml/kg)		0.05**	0.03**	0.06	0.90

Abbreviations. TV/PBW: Tidal volume/predicted body weight, ORmort: Odds ratio for mortality, CI.: Confidence interval

*TV/PBW 4-6ml/kg is the referent group for the OR analysis.

** Patients receiving 6–7 and 7–8 ml/kg had significantly lower adjusted mortality than those receiving 4–6 ml/kg TV/PBW.