Patient characteristics and outcomes associated with adherence to the low PEEP/FIO2 table for acute respiratory distress syndrome

It remains uncertain how best to set positive end-expiratory pressure (PEEP) for mechanically ventilated patients with the acute respiratory distress syndrome (ARDS). Among patients on low tidal volume ventilation (LTVV), we investigated if further adherence to the low PEEP/FIO2 (inspired oxygen fraction) table would be associated with better survival compared to nonadherence. Patients with ARDS, admitted directly from the Emergency Department to our 20-bed Medical Intensive Care Unit (ICU) from August 2016 to July 2017, were retrospectively studied. To determine adherence to the low PEEP/FIO2 table, PEEP and FIO2 12 h after ICU admission were used, to reflect ventilator adjustments by ICU clinicians after initial stabilization. Logistic regression was used to analyze hospital mortality as an outcome with adherence to the low PEEP/FIO2 as the key independent variable, adjusted for age, APACHE II score, initial P/F ratio and initial systolic blood pressure. 138 patients with ARDS were analysed. Overall adherence to the low PEEP/FIO2 table was 75.4%. Among patients on LTVV, nonadherence to the low PEEP/FIO2 table was associated with increased mortality compared to adherence (adjusted odds ratio 4.10, 95% confidence interval 1.68–9.99, P = 0.002). Patient characteristics at baseline were not associated with adherence to the low PEEP/FIO2 table.

Although clinical equipoise currently exists for setting PEEP, some technique for PEEP titration is still needed. The most convenient and user-friendly method remains following the low PEEP/FIO2 table, which was used in the landmark ARMA trial of low versus high tidal volume ventilation 18 . In contrast to quality improvement efforts directed at studying and improving adherence to LTVV, there has been little effort directed at adherence to the low PEEP/FIO2 table. We hypothesize that among patients on LTVV, further adherence to the low PEEP/ FIO2 table would be associated with better survival compared to nonadherence. We therefore aimed to study the patient characteristics and clinical outcomes of adherence to the low PEEP/FIO2 table.

Methods
Participants and setting. We conducted a retrospective cohort study of patients with ARDS admitted directly from the Emergency Department to our 20-bed Medical ICU from August 2016 to July 2017. We included patients over the age of 21 years old, who were intubated in the Emergency Department prior to ICU transfer, and who had LTVV i.e. tidal volume < 7 ml/kg ideal body weight 2 . Patients were excluded if they were transferred from other locations, as any potential delays in ICU admission might adversely influence patient survival 19 . Patients who were intubated in the ICU after admission were also excluded, as it is uncertain if intubation was delayed 20,21 . Our Ethics Review Board (National Healthcare Group Domain-Specific Review Board) approved the study (National Healthcare Group Domain-Specific Review Board approval number 2018/00223). As the study is a retrospective observational one, the need for patient consent was waived. All procedures and analyses were performed in accordance with relevant guidelines and regulations.
General clinical care. Patients were initiated on flow-triggered, descending-ramp, volume assist-control as the default mode, using either the Puritan-Bennett 840 or Puritan-Bennett 980 ventilator (Medtronic, MN). Low tidal volumes were used, targeting plateau pressures of less than 30 cmH2O 2 . Recruitment manoeuvers, inhaled nitric oxide, neuromuscular paralysis and prone positioning were seldom used (due to inadequate staffing and protocols). Analgesia was titrated to achieve a Critical-Care Pain Observation Tool score of 0-2 and sedation was titrated to achieve a Richmond Agitation-Sedation Scale score − 2 to 0. Daily assessment for awakening and spontaneous breathing trials were done. Noradrenaline was the preferred vasopressor, targeting a mean arterial pressure of at least 65 mmHg 22,23 . Sepsis was treated with early, broad-spectrum antibiotics and source control.
Data collection and definitions. Clinical parameters and arterial blood gas measurements were obtained at the time of ICU admission and at 12 h after ICU admission. Patient outcomes were determined till hospital death or discharge. To determine adherence to the low PEEP/FIO2 table (Table 1), PEEP and FIO2 12 h after ICU admission were used 24 , rather than at the time of ICU admission, to reflect ventilator adjustments by ICU clinicians after initial stabilization. To determine the recommended PEEP, FIO2 values were rounded up to the nearest 10%. For instance, an FIO2 of 55% was taken as 60% on the PEEP/FIO2 table. Applied PEEP at 12 h after ICU admission within the ranges recommended by the low PEEP/FIO2 table was considered adherent, while PEEP above or below the recommended ranges was considered non-adherent. Statistical analysis. Proportions, means and medians were compared using Fisher exact, Student t, and Wilcoxon rank-sum tests respectively. We examined the association of adherence to the low PEEP/FIO2 table with age, gender, body-mass index, Acute Physiology and Chronic Health Evaluation (APACHE) II score, primary diagnosis, comorbid conditions, initial arterial oxygen partial pressure to inspired oxygen fraction (P/F) ratio, initial systolic blood pressure, use of vasopressors within the first 24 h of ICU admission, ICU/hospital mortality, ICU/hospital length-of-stay and ventilator-free days through day 28 25 . Logistic regression was used to analyze hospital mortality as an outcome with adherence to the low PEEP/FIO2 as the key independent variable, adjusted for age, APACHE II score, initial P/F ratio and initial systolic blood pressure (adjustment was determined a priori to account for key baseline prognostic factors). For logistic regression analysis, PEEP adherence was additionally coded as a 3-level indicator variable (PEEP lower than recommended; PEEP as recommended; PEEP higher than recommended), with PEEP as recommended being the reference level. Logistic regression was

Results
138 patients with ARDS were analysed: mean age 64.0 ± 26.9 years, 31.2% female, mean APACHE II score 27.9 ± 9.9, mean P/F ratio 145 ± 66 mmHg ( Table 2). Distribution of PEEP and FIO2 at 12 h after ICU admission is shown in Fig. 1. Overall adherence to the low PEEP/FIO2 table was 75.4% (Table 3). Of 34 patients in the non-adherent group, 26 (76.5%) had PEEP higher than recommended and these patients also fell into the high PEEP/FIO2 When PEEP adherence was coded as a 3-level indicator variable (PEEP lower than recommended; PEEP as recommended; PEEP higher than recommended), only PEEP higher than recommended was associated with increased hospital mortality in the whole cohort (adjusted odds ratio 4.51, 95% confidence interval 1.68-12.2, P = 0.003; Table 4) and in patients with moderate-to-severe ARDS (adjusted odds ratio 4.27, 95% confidence interval 1.42-12.8, P = 0.010; Table 5).

Discussion
The main finding of our study is that among patients on LTVV, nonadherence to the low PEEP/FIO2 table was associated with increased mortality compared to adherence, adjusted for baseline critical illness severity, oxygenation and blood pressure (odds ratio 4.10, 95% confidence interval 1.68-9.99, P = 0.002). In particular, applied PEEP higher than that recommended by the low PEEP/FIO2 table was associated with increased mortality (adjusted odds ratio 4.51, 95% confidence interval 1.68-12.2, P = 0.003). Patient characteristics at baseline were not associated with adherence to the low PEEP/FIO2 table.
Possible explanations for the association of mortality with PEEP/FIO2 table non-adherence could be the deleterious pulmonary and cardiovascular effects of high PEEP. High PEEP may cause overdistension of alveoli 7 , possibly leading to biotrauma and barotrauma, though we did not observe the latter in our patients. In addition, high PEEP can increase intrathoracic pressure, impede venous return, reduce cardiac output, and cause systemic hypotension 8 . In our patients, these might have occurred. Patients who were not adherent to the low PEEP/FIO2 table generally received higher PEEP than recommended and a greater proportion required vasopressors in the first 24 h of ICU admission. Given that association does not mean causation, one needs to consider whether reverse causation was possible i.e. a patient who has higher mortality drove the use of higher PEEP. We feel that this possibility is slim since none of the baseline prognostic factors were associated with nonadherence to the low PEEP/FIO2 table.
Other possible explanations for the mortality difference between the adherent and non-adherent groups were considered. Firstly, based on peripheral oxygen saturation/FIO2 (S/F ratio) changes, we classified patients with decreased S/F ratio over the first 12 h as deteriorating, and patients with stable or improved S/F ratio as non-deteriorating. The proportion of deteriorating patients was similar between the non-adherent and the adherent groups (6/34 [17.7%] versus 15/104 [14.4%], P = 0.783). As such, the higher mortality in the non-adherent group could not be explained by a higher proportion of deteriorating patients. Secondly, among 85 patients with plateau pressure measured at 12 h, between the adherent and non-adherent groups, the mean driving pressure was not significantly different (13.3 versus 11.2 cmH2O, P = 0.307) and could not explain the difference in mortality. Thirdly, although the absolute difference was large, pneumonia as the etiology of ARDS was not statistically different between the adherent and non-adherent groups (P = 0.062). The increased hospital mortality for patients who demonstrated nonadherence persisted after adjustment for age, APACHE II score, initial P/F ratio, initial systolic blood pressure and pneumonia as the ARDS etiology (odds ratio 5.90, 95% confidence interval 2.14-16.2, P = 0.001).
In our study, applying PEEP lower than that recommended by the low PEEP/FIO2 table was not significantly associated with mortality for a few reasons. Firstly, all patients except one had PEEP of at least 5 cmH2O and severe decruitment in most patients would be unlikely. Secondly, it may be possible that low PEEP may be sometimes beneficial, for instance to avoid excessive right ventricular afterload and to reduce the risk of acute right ventricular failure. Thirdly, only 5.8% of patients had PEEP lower than recommended by the low PEEP/FIO2 table, compared to triple the proportion of patients (18.8%) with PEEP higher than recommended by the low PEEP/FIO2 table, which meant that the power to detect a significant association for the former group of patients The lack of association between baseline patient characteristics and adherence to low PEEP/FIO2 table is not surprising, given clinical equipoise over PEEP optimization. In our ICU, while there has been broad consensus over the need to limit tidal volume and driving pressure, and guidelines to keep peripheral oxygen saturation between 90-96%, there has been no firm recommendation about PEEP titration. Esophageal balloon catheters and electrical impedance tomography are unavailable, leaving only three options for clinicians: pressure-volume curve based titration, lung ultrasound and the low PEEP/FIO2 table. Anecdotally, the low PEEP/FIO2 table was the most convenient tool and as expected, most people followed it. www.nature.com/scientificreports/ Our observational study differs from prior randomized trials of high versus low PEEP in ARDS. Our study does not suggest that low PEEP is better than high PEEP, nor does it suggest that personalization of PEEP using other methods cannot not used. Rather, our results suggest that if one sets PEEP using the low PEEP/FIO2 table, then adherence is associated with reduced hospital mortality, supporting the use of the low PEEP/FIO2 table to guide PEEP setting for mechanically ventilated patients with ARDS. In addition, our results highlight the need to study adherence to the low PEEP/FIO2 table as a quality assurance metric.
If nonadherence to the PEEP/FIO2 table is found, several methods could be used to improve adherence. Manual methods include implementation of a management protocol for clinicians 27,28 or an order set driven by respiratory therapists 24 . Semi-automated methods include using a computer-assisted oxygen advisor 29 or a closed-loop system 30 . As can be surmised, prospective studies demonstrating the clinical impact of improving adherence to the PEEP/FIO2 are needed.    . Although this higher proportion would predispose any observed association of adherence and mortality towards the null, our analysis still turned in a significant association. Thirdly, given the distribution of PEEP and FIO2 in our cohort, which reflected general use of the low PEEP/FIO2 table among our ICU clinicians, we could not investigate adherence to the high PEEP/FIO2 table. Fourthly, our results may not apply to patients who are severely obese 31 or who have raised intra-abdominal pressure 32 . Fifthly, we did not have reliable recordings of peak inspiratory pressure and inspiratory flow and could not compute the mechanical power for further investigation.

Conclusions
In conclusion, adherence to the low PEEP/FIO2 table was associated with better survival compared to nonadherence, among mechanically ventilated patients with ARDS who received LTVV. This suggests that for PEEP setting, in lieu of more sophisticated methods, guidance using the low PEEP/FIO2 table remains clinically meaningful 17 . Similar studies should be performed in other ICUs to confirm our results. Upon confirmation, quality assurance for mechanical ventilation among patients with ARDS should include steps to monitor and improve adherence to the low PEEP/FIO2 table.

Data availability
No consent to share data could be obtained.
Received: 20 February 2021; Accepted: 1 July 2021 Table 5. Logistic regression for hospital mortality, by adherence to low PEEP/FIO2 table, for moderate-tosevere ARDS. Total number of patients with moderate-to-severe ARDS (initial P/F ratio ≤ 200 mmHg) = 105. APACHE Acute Physiology and Chronic Health Evaluation, FIO2 Inspired oxygen fraction, PEEP Positive endexpiratory pressure, P/F ratio Ratio of arterial oxygen partial pressure to inspired oxygen fraction.