Impact of proning with and without inhaled pulmonary vasodilators and neuromuscular blocking agents in COVID acute respiratory distress syndrome

Abstract

BACKGROUND

A major cause of mortality in the coronavirus disease 2019 (COVID-19) pandemic was acute respiratory distress syndrome (ARDS). Currently, moderate to severe ARDS induced by COVID-19 (COVID ARDS) and other viral and non-viral etiologies are treated by traditional ARDS protocols that recommend 12-16 hours of prone position ventilation (PPV) with neuromuscular blocking agents (NMBA) and a trial of inhaled vasodilators (IVd) if oxygenation does not improve. However, debate on the efficacy of adjuncts to PPV and low tidal volume ventilation persists and evidence about the benefits of IVd/NMBA in COVID ARDS is sparse. In our multi-center retrospective review, we evaluated the impact of PPV, IVd, and NMBA on outcomes and lung mechanics in COVID ARDS patients with moderate to severe ARDS.

AIM

To evaluate the impact of PPV used alone or in combination with pulmonary IVd and/or NMBA in mechanically ventilated patients with moderate to severe ARDS during the COVID-19 pandemic.

METHODS

A retrospective study at two tertiary academic medical centers compared outcomes between COVID ARDS patients receiving PPV and patients in the supine position. PPV patients were divided based on concurrent use of ARDS adjunct therapies resulting in four subgroups: (1) PPV alone; (2) PPV and IVd; (3) PPV and NMBA; and (4) PPV, IVd, and NMBA. Primary outcomes were hospital and intensive care unit (ICU) length of stay (LOS), mortality, and venovenous extracorporeal membrane oxygenation (VV-ECMO) status. Secondary outcomes included changes in lung mechanics at 24-hour intervals for 7 days.

RESULTS

Total 114 patients were included in this study. Baseline respiratory parameters and Sequential Organ Failure Assessment scores were significantly worse in the PPV group. ICU LOS and LOS were significantly longer for patients who were proned, but no mortality benefit or difference in VV-ECMO status was found. Among the subgroups, no difference in primary outcomes were found. In the secondary analysis, PPV was associated with a significant improvement in arterial oxygen partial pressure (PaO₂)/fractional inspired oxygen (FiO₂) (P/F) ratio from day 1 to day 4 (P < 0.05) and higher driving pressures day 5 to day 7 (P < 0.05). The combination of PPV and IVd together resulted in improvements in P/F ratio from day 1 to day 7 and plateau pressure on day 4 and day 6 (P < 0.05). PPV with NMBA was not associated with improvements in any of the secondary outcomes. The use of all three rescue therapies together resulted in improvements in lung compliance on day 2 (P < 0.05) but no other improvements.

CONCLUSION

In mechanically ventilated patients diagnosed with moderate to severe COVID ARDS, PPV and PPV with the addition of IVd produced a significant and sustained increase in P/F ratio. The combination of PPV, IVd and NMBA improved compliance however this did not reach significance. Mortality and LOS did not improve with adjunct therapies. Further research is warranted to determine the efficacy of these therapies alone and in combination in the treatment of COVID ARDS.

Key Words: Acute respiratory distress syndrome; COVID; Prone position ventilation; Neuromuscular blocking agents; Pulmonary vasodilators; Mechanical ventilation; Plateau pressure; Driving pressure; Peak end expiratory pressure

Core Tip: In this retrospective study at two tertiary academic medical centers during the initial coronavirus disease 2019 surge, we examine trends in ventilator mechanics and outcomes of 114 acute respiratory distress syndrome (ARDS) patients receiving three adjunct therapies (prone positioning, neuromuscular blockade and inhaled vasodilations). We found a significant improvement in arterial oxygen partial pressure/fractional inspired oxygen ratio with the addition of inhaled vasodilators while proning and in lung compliance with the addition of continuous neuromuscular blockade among others. Our groups were not large enough to detect a difference in mortality or length of stay. However, this study provides a large amount of data and multi-day trends to further our understanding of the physiologic response to multiple adjunct therapies for ARDS in combination.

INTRODUCTION

The acute respiratory distress syndrome (ARDS) continues to be a major cause of mortality worldwide. During the recent coronavirus disease 2019 (COVID-19) pandemic intensivists across the country witnessed an unprecedented uptick in ARDS, often refractory to standard ventilation protocols[1]. During the initial COVID-19 surge, professional guidelines suggested leveraging established protocols for traditional ARDS including implementation of mainstay treatments such as lung protective mechanics, trialing prone position ventilation (PPV) for refractory hypoxemia, as well as pharmacological interventions such as neuromuscular blockade and inhaled vasodilators (IVd) in the management of ARDS induced by COVID-19 (COVID ARDS)[2-4].

Trials of PPV has become standard of care in moderate to severe ARDS[4]. Multiple randomized control trials have shown PPV for more than 12 hours reduces mortality and improves the ratio of arterial oxygen partial pressure (PaO₂) to fractional inspired oxygen (FiO₂), a common measure used to assess lung injury severity[5-7]. Continuous neuromuscular blocking agents (NMBA) are typically reserved for decreasing ventilator desynchrony, work of breathing, and the alveolar fluid accumulation in moderate to severe ARDS; however, research on their outcomes in ARDS remains mixed[8,9]. IVd therapy’s preferential dilation of well-ventilated alveoli improves the ventilation-perfusion ratio in ARDS however has failed to demonstrate a mortality benefit[10,11]. In the most recent guidelines from the European Society of Intensive Care Medicine guidelines on ARDS, prone positioning is recommended for at least 12 hours in patients with moderate to severe ARDS. However, current guidelines recommend against continuous NMBA for non-COVID ARDS and indeterminate for COVID ARDS. IVd are not included in recent guidelines but often trialed in clinical practice[5].

Data specifically for COVID ARDS adjunct therapies has shown considerable crossover with other forms of ARDS. PPV has been shown to improve respiratory physiology with similar efficacy in patients who have non-COVID ARDS and reduces 30-day to 90-day mortality[12-14]. Research on the efficacy of NMBA in COVID ARDS remains mixed showing no mortality benefit in these patients[15]. Similarly, when examining IVd therapy in COVID ARDS, a handful of retrospective studies have shown a minimal increase in the PaO₂/FiO₂ (P/F) ratio with undetermined clinical significance[16-18]. Despite recommendations for the use of adjunct therapies in moderate to severe ARDS, in clinical practice the use of these therapies remains inconsistent with 67% of non-survivors receiving no adjunct therapies in the first 48 hours of ARDS onset[19].

Clinicians often use these therapies in combination with incremental adjustments based on illness severity and response to early interventions; however, data on this subject is extremely limited. Some studies have shown improvements in oxygenation with concomitant use of IVd when given to patients with severe hypoxemia prior to PPV but no studies have examined the impact of all three of these therapies (e.g., PPV, NMBA, and IVd) used alone or in conjunction on clinical and respiratory physiology outcomes over an extended period[20,21].

This multi-center retrospective study aims to describe the physiologic changes to lung mechanics and outcomes of PPV with and without IVd and NMBA to further advance evidence-based approaches in the use of adjunct therapies in mechanically ventilated patients with moderate to severe ARDS. This study is unique in evaluating the combined effects of prone ventilation with inhaled pulmonary vasodilators [epoprostenol sodium (iEPO)] and NMBA (cisatracurium) in patients with COVID ARDS, a cohort with distinct clinical features such as hypercoagulability and endothelial dysfunction. Previous studies have largely focused on individual interventions; however, this study advances existing knowledge by exploring the synergistic effects of these interventions. These findings have the potential to guide future work and treatment strategies for COVID ARDS and inform clinical management protocols.

MATERIALS AND METHODS

Patients

Patients admitted to the intensive care unit (ICU) at two tertiary urban hospitals in Washington D.C. with a confirmed diagnosis of COVID-19 by severe acute respiratory coronavirus 2 polymerase chain reaction (PCR) were reviewed from March 1, 2020 to June 30, 2020. Inclusion criteria involved patients with a clinical diagnosis of COVID-19 pneumonia who met criteria for moderate to severe ARDS as modeled by the Berlin Criteria of a P/F ratio of < 200. Patients excluded from the study involved those who were mechanically ventilated for less than 24 hours, lacked nursing documentation of ventilation positioning, or had a confirmed bacterial pneumonia superinfection during their ICU stay.

Study design

The study was designed to evaluate the impact of IVd and NMBA on COVID ARDS using PPV as the standard backbone therapy. The control group included mechanically ventilated patients in the supine position and the treatment group included mechanically ventilated patients who received PPV at any point during their hospitalization. The treatment group was further subdivided into combinations of adjunct therapy used with PPV (i.e., PPV alone, PPV and IVd, PPV and NMBA, and PPV, IVd and NMBA). Primary outcomes examined clinical outcomes such as ICU length of stay (LOS), hospital LOS, hospital mortality, and progression to venovenous extracorporeal membrane oxygenation (VV-ECMO). Secondary outcomes analyzed changes in lung mechanics including P/F ratio, plateau pressure, driving pressure, tidal compliance, and oxygen index. To be included into a treatment subgroup, therapies either alone or in combination had to be delivered for at least 24 hours and implemented within 60 hours of initial mechanical ventilation.

Interventions and data collection

Standard of care treatment protocols for COVID ARDS, including criteria for the use of PPV, NMBA, and IVd therapy were reviewed and consistent between the two participating sites at the time of inclusion. These protocols were developed by critical care experts and based on evidence-driven ARDS therapies at the time. ARDS patients followed ARDSnet protocol with strict adherence to low tidal volume ventilation. For adjuncts, the guidelines at the time recommended initiating PPV if the P/F ratio was < 150 at the time of intubation, administering NMBA as a last resort if ventilator desynchrony persists, and giving IVd therapy in refractory hypoxemia cases when the P/F ratio is < 150 or if intermittent or sustained recruitment maneuvers had been attempted. However, determination of when to implement these adjunct therapies was primarily left to the discretion of the intensivists at the time.

A data registry using a secure Research Electronic Data Capture database was created for all ICU patients who met criteria. Data collection involved a medical chart review that extracted time of PPV initiation with subsequent position changes, and medications administered such as IVd and NMBA. The standard prone position protocol for patients requiring mechanical ventilation was implemented at our centers with thorough preparation to ensure patient safety and comfort. The patient is carefully turned to the prone position by the nursing team with the respiratory therapist at bedside, ensuring the head remains in a neutral position. The chest and abdomen rested on a firm surface, with the arms placed either beside the body or flexed at 90-degree angles above the head to minimize pressure on the joints. A pillow or soft cushion is placed under the pelvis to maintain a slight pelvic tilt, and the legs are slightly flexed to reduce strain on the knees and lower back. Respectively, patients who were delivered the IVd, iEPO, via continuous nebulization at a 50 ng/kg/min dose and the NMBA, cisatracurium, dosed at 0.15 mg/kg as an initial bolus followed by continuous infusion to achieve an adequate level of paralysis (i.e., determined by 2 out of 4 twitches on train of four monitoring), were included in the database. Those who received a bolus or as needed (pyronaridine) NMBA did not qualify for the PPV and NMBA group specifically to avoid accounting for use of NMBA in cases of desynchrony. For our primary outcomes, ICU and hospital LOS, hospital mortality, and VV-ECMO status were collected. To control for confounding factors, international classification of diseases-10 codes matched to each patient were used to extract data on demographics and comorbidities. Severity of illness was electronically calculated from the highest Sequential Organ Failure Assessment (SOFA) score prior to ICU admission and missing data was factored in using established guidelines from Lambden et al[22]. For our secondary outcomes, day zero baseline respiratory and ventilator parameters were extracted within 24 hours of intubation for the control group and within 24 hours of PPV initiation. To determine lung mechanic fluctuations over a continuous period, secondary outcomes following day zero baseline ventilator parameters before PPV initiation were tracked in 24-hour intervals for seven consecutive ICU days using daily morning arterial blood gas values, or values closest to this time. A random selection of 10% of charts underwent secondary review for quality control.

Statistical analysis

Patient characteristics were presented using descriptive statistics such as mean, standard deviation, and proportions. Comparison of group means for continuous variables were conducted using one-way analysis of variance (ANOVA). In the case of highly skewed variables, data was presented using median, 25^th and 75^th percentiles, and a comparison was conducted using a Kruskal-Wallis test. Proportions were compared using a χ² squared or a Fisher’s Exact test if there was a small sample size.

For our primary outcomes, ICU and hospital LOS, mortality, and VV-ECMO status were compared among the treatment subgroups using χ² square or a Fisher’s exact test for groups with a small sample size, and mortality rates and LOS values were presented using mean and standard deviations for each group. Multivariate logistic regression analysis was conducted for controlling for baseline characteristics of patients using age and SOFA score prior to ICU admission. Results of logistic regression were presented using odds ratios and their 95%CI. For our secondary outcomes, changes in P/F ratio, plateau pressure, driving pressure, tidal compliance, and oxygen index over time were compared between and within the treatment subgroups using a repeated measures ANOVA. Multivariate repeated measures ANOVA was completed to control for baseline patient characteristics such as age, gender, and race. Day of initiation of mechanical ventilation was treated as a dependent variable and post intervention values were recorded at 24-hour intervals over a seven-day course, where interactions between groups and time were used as independent variables. Descriptive statistics were presented using mean, standard deviation, median, 25^th and 75^th percentile, and range. If the dependent variable was skewed and model residuals did not satisfy normality assumption, then appropriate log transformations were used prior to conducting repeated measures ANOVA. All analysis was done using R version 4.0.0 except repeated measures ANOVA on SAS version 9.4.

RESULTS

Total 136 patients had a positive COVID-19 PCR test and were mechanically ventilation between March 1, 2020 to June 30, 2020. After applying our exclusion criteria, 114 patients were left. The supine group consisted of 25 patients and the treatment group had 89 patients. Within the treatment subgroups, majority of patients received PPV (40.5%, n = 36) or PPV and IVd (40.5%, n = 36), followed by PPV, IVd, and NMBA (13.4%, n = 12), and PPV and NMBA (5.6%, n = 5) respectively (see Figure 1 for more detail). The mean age of participants was 59 (SD = 14) and 58% (n = 66) identifying as male. Majority of participants identified as either Black (37%, n = 71) or another race (37%, n = 31) and 26% (n = 30) identified as Hispanic ethnicity. Comorbidities were matched well between groups, with hypertension (67%, n = 78) and heart disease (60%, n = 70) being the most common. There was no statistically significant difference in gender, race, or age; however, the PPV group had significantly more Hispanic patients (P = 0.021) and higher body mass index (BMI) scores (P = 0.012). Discharge status was similar between groups. See Table 1 for a full summary of demographic characteristics.

Figure 1 Flowchart of patients analyzed for the study. Intervention group was further subdivided (gray) by therapy combination for secondary analysis. ARDS: Acute respiratory distress syndrome; COVID: Coronavirus disease; IVd: Inhaled vasodilators; NMBA: Neuromuscular blocking agents; PCR: Polymerase chain reaction; P/F: Arterial oxygen partial pressure/fractional inspired oxygen; PPV: Prone position ventilation; SARS-CoV-2: Severe acute respiratory coronavirus 2.

Table 1 Demographics and comorbidities, n (%).

Characteristic	All	Control	Prone position ventilation ± rescue therapy	P value
n	114	25	89	-
Age (years) (mean ± SD)	58.32 ± 13.44	62.48 ± 14.84	57.18 ± 12.89	0.081
Sex
Female	25 (21.55)	14 (56.0)	35 (38.46)	0.116
Male	91 (78.45)	11 (44.0)	56 (61.54)
Race
African American	73 (69.93)	19 (76.0)	54 (59.34)	0.208
White	6 (5.17)	0 (0.00)	6 (6.59)
Other	37 (31.9)	6 (24.0)	31 (34.07)
Ethnicity
Hispanic or Latino	30 (25.86)	2 (8.0)	28 (30.77)	0.021a
Non-Hispanic or Latino	86 (74.14)	23 (92.0)	63 (69.23)
Body mass index (median, interquartile range)	30.7 (27.5, 39.2)	28.3 (24.3, 33.0)	32.8 (28.4, 40.0)	0.012a
Comorbidities
Immune disorders	11 (9.48)	1 (4.0)	10 (10.99)	0.291
Kidney disease	13 (11.21)	1 (4.0)	12 (13.19)	0.197
Liver disease	17 (14.66)	4 (16.0)	13 (14.29)	0.830
Diabetes	72 (62.07)	15 (60.0)	57 (62.64)	0.830
Heart disease	70 (60.34)	14 (56.0)	56 (61.54)	0.616
Hypertension	78 (67.24)	16 (64.00)	62 (68.13)	0.697
Discharge status
Expired	59 (50.86)	12 (48.0)	47 (51.65)	0.722
Inpatient rehab	10 (8.77)	1 (4.0)	9 (9.89)
Acute Care Hospital	16 (14.0)	4 (16.0)	12 (13.19)
Skilled nursing facility, long-term acute care	5 (4.31)	1 (4.0)	4 (4.40)
Home	23 (19.82)	7 (28.0)	16 (17.58)
Other	3 (2.58)	0 (0.00)	3 (3.30)

^aStatistically significant value (P < 0.05).

For our overall primary outcomes, higher SOFA score and age were both independently associated with mortality (P < 0.001, P < 0.001) and VV-ECMO status (P < 0.05, P < 0.001) and SOFA scores were significantly lower in the supine group compared to the PPV group at baseline (P < 0.001). See Table 2 for more details. However, when controlling for SOFA and age, no significant difference was elicited in mortality rates and VV-ECMO status between the control group and the PPV group (P = 0.747). ICU LOS and hospital LOS were significantly longer for patients who received PPV compared to the control group by an average of 9.9 days (P < 0.001) and 9.7 days (P = 0.003), respectively. See Tables 2 and 3 for more detail. Among the treatment subgroups, the PPV and IVd group had the highest pre-ICU SOFA score but this was not significantly different from the other groups (P = 0.089). In addition, there was no difference in ICU LOS, hospital LOS, mortality, or VV-ECMO status between subgroups. See Table 4 for more detail.

Table 2 Clinical features, baseline ventilator settings and hard outcomes of patients that received either intervention in combination with prone position ventilation compared to controls, n (%).

Characteristic	All	Control	Prone position ventilation ± rescue therapy	P value
n	114	25	89	-
Highest Sequential Organ Failure Assessment score (mean ± SD)	13.8 ± 3.2	11.7 ± 3.5	14.4 ± 2.9	< 0.001a
Baseline respiratory parameters (mean ± SD)
FiO2	73.6 ± 21.4	55 ± 15.6	80.2 ± 19.2	< 0.001a
PaO2	85.9 ± 35.0	105.7 ± 45.7	80.3 ± 29.2	0.002a
PaO2/FiO2	121.6 ± 58.8	175.7 ± 75.0	105.3 ± 41.4	< 0.001a
SpO2	94.8 ± 4.5	97.2 ± 3.5	94.0 ± 4.5	0.001a
Baseline ventilator parameters, median (interquartile range)
Positive end expiratory pressure (cm H2O)	11.1 ± 6.3	10.8 ± 7.9	11.2 ± 4.9	0.771
Tidal volume (mL)	425.7 ± 64.8	418.4 ± 65.6	429.2 ± 64.8	0.496
Plateau pressure (cm H2O)	25.2 ± 5.5	24.3 ± 6.1	25.6 ± 5.1	0.360
Driving pressure (cm H2O)	14.6 ± 4.4	15.2 ± 4.8	14.3 ± 4.2	0.470
Tidal compliance (mL/cm H2O)	32.4 ± 10.1	31.0 ± 9.0	33.1 ± 10.6	0.478
Peak inspiratory pressure (cm H2O)	31.0 ± 7.3	28.0 ± 6.1	32.6 ± 7.4	0.009a
O2 index	14.0 ± 9.0	9.7 ± 11.2	16.4 ± 6.6	0.002a
Average number of times patient was proned during hospitalization	-	-	1.3	-
Average time from intubation to first prone (hours)	-	-	25.7	-
Average duration of first proning session (days) (n = 89)	-	-	2.5	-
Average duration of second proning session (days) (n = 44)	-	-	2.6	-
Average duration of third proning session (days) (n = 12)	-	-	1.5	-
Primary outcomes
Mortality	59 (50.8)	12 (48.0)	47 (51.65)	0.747
Venovenous extracorporeal membrane oxygenation	21.9 (25)	8 (2.0)	76 (19)	0.052a
Intensive care unit LOS (mean ± SD)	15.2 ± 12.8	7.4 ± 7.0	17.3 ± 13.2	< 0.001a
Hospital LOS (mean ± SD)	20.2 ± 15.7	12.6 ± 14.1	22.3 ± 15.6	0.003a

^aStatistically significant value (P < 0.05).

FiO₂: Fractional inspired oxygen; LOS: Length of stay; PaO_2: Arterial oxygen partial pressure.

Table 3 Logistical and linear regression of primary outcomes adjusted for baseline Sequential Organ Failure Assessment score and age.

Primary outcomes	Control vs treatment subgroup	Odds ratio (95%CI)	P value	SE
Mortality	PPV	0.82 (0.19-3.36)	0.78	-
	PPV and IVd	0.57 (0.15-2.04)	0.39	-
	PPV and NMBA	0.8 (0.07-7.28)	0.84	-
	PPV, IVd and NMBA	2.15 (0.51-9.40)	0.3	-
Venovenous extracorporeal membrane oxygenation	PPV	2.33 (0.03-22.63)	0.43	-
	PPV and IVd	1.03 (0.12-10.36)	0.98	-
	PPV and NMBA	8 (0.55-139.75)	0.13	-
	PPV, IVd and NMBA	1.04 (0.17-8.90)	0.97	-
Intensive care unit LOS	PPV	-	0.028	0.24
	PPV and IVd	-	< 0.001a	0.23
	PPV and NMBA	-	< 0.001a	0.38
	PPV, IVd and NMBA	-	< 0.001a	0.24
Hospital LOS	PPV	-	0.04	0.25
	PPV and IVd	-	< 0.001a	0.23
	PPV and NMBA	-	0.007	0.4
	PPV, IVd and NMBA	-	< 0.001a	0.03

^aStatistically significant value (P < 0.05).

IVd: Inhaled vasodilators; LOS: Length of stay; NMBA: Neuromuscular blocking agents; PPV: Prone position ventilation.

Table 4 Clinical features, ventilator settings and outcomes of treatment sub-group analysis.

Characteristic	PPV		PPV + IVd		PPV + NMBA	PPV + NMBA + IVd		P value
Total	36		36		5	12		-
Highest Sequential Organ Failure Assessment score pre-ICU (mean ± SD)	14.0 ± 2.7		15.2 ± 2.4		12.2 ± 2.9	14.2 ± 4.2		0.089
Respiratory parameters < 24 hours prior to the first proning session (mean ± SD)
FiO2	76.6 ± 20.0		85.3 ± 18.5		70.0 ± 7.0	81.2 ± 21.6		0.237
PaO2	82.4 ± 31.3		73.6 ± 21.4		73.6 ± 21.4	73.6 ± 21.4		0.701
PaO2/FiO2	112.0 ± 68.1		101.3 ± 58.5		101.0 ± 9.3	61.6 ± 55.1		0.241
SpO2	94.0 ± 4.6		93.9 ± 4.8		95.6 ± 3.5	93.9 ± 3.4		0.896
Ventilator parameters < 24 hours prior to the first proning session (mean ± SD)
positive end expiratory pressure, cm H2O	10.7 ± 2.2		12.6 ± 8.8		8.7 ± 2.5	12.8 ± 2.2		0.528
Tidal volume (mL)	413.5 ± 65.0		440.5 ± 70.3		428.7 ± 44.0	443.3 ± 64.0		0.570
Plateau pressure (cm H2O)	24.5 ± 6.1		25.9 ± 4.1		26.5 ± 4.9	27.8 ± 5.8		0.643
Driving pressure (cm H2O)	13.2 ± 12.3		14.3 ± 3.9		18.0 ± 0.0	15.8 ± 4.7		0.373
Tidal compliance (mL/cm H2O)	34.8 ± 12.3		35.5 ± 9.0		29.5 ± 10.60	24.0 ± 6.7		0.185
Peak inspiratory pressure	32.7 ± 7.0		33.5 ± 5.8		24.2 ± 9.9	35.1 ± 9.6		0.111
O2 index	16.8 ± 8.3		15.6 ± 5.5		14.2 ± 2.0	20.5 ± 3.1		0.537
Average duration of intervention
First proning session duration (days)	2.3		3.1		1.0	2.0		-
Total duration of IVd (days)	-		2.9		-	3.7		-
Total duration of NMBA (days)	-	-		3.4		2.7	-
Average time from intubation to first prone (hour)	23.2	18.8		23.5		43.4	-
Primary outcomes
Mortality	21 (58.33)		20 (55.56)		2 (40.00)	4 (33.33)		0.439
Total LOS (mean ± SD)	15.2 ± 10.7		15.8 ± 10.8		32.2 ± 34.2	17.4 ± 8.0		0.618
ICU LOS (mean ± SD)	19.5 ± 11.8		20.6 ± 13.1		34.2 ± 20.0	34.2 ± 34.1		0.737

FiO₂: Fractional inspired oxygen; ICU: Intensive care unit; IVd: Inhaled vasodilators; LOS: Length of stay; NMBA: Neuromuscular blocking agents; PaO_2: Arterial oxygen partial pressure; PPV: Prone position ventilation.

In the secondary outcome analysis, the PPV group compared to the control group had a significantly higher FiO₂ requirement (P < 0.001) and O₂ index (P = 0.002), and lower PaO₂ (P = 0.002), SpO₂ (P = 0.001) and P/F ratio (P < 0.001) at baseline. Peak inspiratory pressure was on average 4.6 cm H₂O higher (P = 0.009) in the PPV group. There was no significant difference in baseline positive end expiratory pressure, tidal volume, driving pressure or tidal compliance between the two main groups. In terms of the treatment subgroups, baseline respiratory and ventilator parameters were similar between all four groups (Table 4). When tracked for seven days, PPV was found to be associated with improvements in P/F ratio by 16.3% on day 1 (P = 0.001), 23.7% on day 2 (P = 0.01), 15.0% on day 3 (P = 0.005), and 16.9% on day 4 (P = 0.005). PPV alone was also associated with increased driving pressure on day 5, day 6, and day 7 (P < 0.05), but did not have a statistically significant impact on plateau pressure or lung compliance. The combination of PPV and IVd resulted in significant improvements in P/F ratio ranging from 14.4% to 26.8% that were sustained for 7 consecutive days (P < 0.05). PPV and IVd also increased plateau pressure by a mean of 2.6 cm H₂O on day 4 (P = 0.02) and 2.7 cm H₂O on day 6 (P = 0.03). The use of PPV, IVd, and NMBA, resulted in an overall improvement in lung compliance with a significant mean increase of 11.3 mL/cm H₂O on day 2 (P = 0.02). There were no other significant improvements in other secondary outcomes. The complete seven-day patterns of all secondary outcome variables by intervention group can be visualized in Table 5 and Figure 2.

Figure 2 Trends in secondary outcomes including ventilator parameters 7 days following initiation of prone ventilation. IVd: Inhaled vasodilators; NMBA: Neuromuscular blocking agents; P/F: Arterial oxygen partial pressure/fractional inspired oxygen; PPV: Prone position ventilation.

Table 5 Secondary outcomes from day 0 to day 7 after initiation of intervention per treatment subgroup.

Intervention	Day	Arterial oxygen partial pressure/fractional inspired oxygen Ratio			Plateau pressure			Compliance			Driving pressure			Oxygen index
		Mean	∆ From day 0	P value	Mean	∆ From day 0	P value	Mean	∆ From day 0	P value	Mean	∆ From day 0	P value	Mean	∆ From day 0	P value
PPV	0	121.6	-	-	23.4	-	-	39.4	-	-	12.0	-	-	13.1	-	-
	1	141.5	16.3	0.00a	24.3	0.9	0.31	39.1	-0.3	0.88	12.7	0.7	0.42	10.4	-2.6	0.04a
	2	150.4	23.7	0.00a	24.0	0.6	0.56	37.8	-1.7	0.55	12.8	0.8	0.43	10.3	-2.7	0.09
	3	139.8	15.0	0.05a	23.9	0.5	0.66	35.2	-4.3	0.16	13.2	1.2	0.30	11.9	-1.1	0.53
	4	142.2	16.9	0.05a	24.3	0.9	0.45	36.2	-3.3	0.30	13.6	1.7	0.16	11.4	-1.7	0.39
	5	142.4	17.1	0.07	24.2	1.8	0.17	35.3	-4.1	0.19	14.5	2.5	0.04a	12.7	-0.3	0.87
	6	140.1	15.2	0.12	24.7	1.2	0.34	35.1	-4.4	0.18	14.9	2.9	0.02a	12.2	-0.8	0.69
	7	140.9	15.9	0.12	24.8	1.4	0.29	35.0	-4.4	0.18	15.1	3.1	0.01a	11.6	-1.1	0.61
PPV and IVd	0	109.3	-	-	25.0	-	-	36.3	-	-	13.6	-	-	11.5	-	-
	1	124.9	14.4	0.01a	26.3	1.2	0.13	38.1	1.8	0.49	13.1	-0.5	-	14.0	2.6	0.05
	2	136.1	24.6	0.00a	26.1	1.1	0.26	38.1	1.8	0.55	12.2	-1.3	0.51	13.4	2	0.22
	3	133.3	22.0	0.01a	26.9	1.9	0.08	34.3	-2	0.52	14.2	0.6	0.18	13.5	2.1	0.25
	4	129.8	18.8	0.03a	27.7	2.6	0.02a	34.2	-2.1	0.51	15.2	1.7	0.55	13.8	2.4	0.22
	5	136.1	24.5	0.01a	27.1	2.0	0.09	36.6	0.3	0.92	15.1	1.6	0.14	13.0	1.6	0.44
	6	134.2	22.8	0.02a	27.8	2.7	0.03a	32.7	-3.6	0.28	16.3	2.8	0.18	13.6	2.2	0.30
	7	138.5	26.8	0.01a	27.4	2.3	0.07	32.5	-3.8	0.26	16.0	2.4	0.02a	12.6	1.2	0.58
PPV and NMBA	0	108.3	-	-	25.9	-	-	35.9	-	-	-	-	-	11.8	-	-
	1	115.6	6.8	0.54	29.7	3.8	0.13	29.2	-6.7	0.33	18.7	2.3	0.40	12.3	0.5	0.86
	2	115.6	6.8	0.65	27.7	1.8	0.54	32.3	-3.7	0.66	18.3	1.9	0.53	10.2	-1.7	0.65
	3	115.6	6.8	0.70	27.1	1.2	0.71	23.1	-12.8	0.15	18.6	2.2	0.50	12.0	0.2	0.96
	4	119.5	10.3	0.61	26.8	0.9	0.79	34.9	-1.1	0.91	19.4	3	0.38	9.7	-2.2	0.64
	5	120.4	11.3	0.60	24.8	-1.0	0.77	35.8	-0.1	0.99	16.8	0.6	0.86	11.2	-0.6	0.91
	6	130.6	20.6	0.40	24.6	-1.2	0.74	36.4	0.5	0.96	16.8	0.6	0.86	10.2	-1.6	0.75
	7	127.6	17.8	0.47	24.6	-1.2	0.74	36.4	0.5	0.96	17.0	0.8	0.82	11.0	-0.8	0.87
PPV, IVd and NMBA	0	119.5	-	-	25.9	-	-	28.8	-	-	14.7	-	-	14.4	-	-
	1	102.0	-3.0	0.78	27.6	1.7	0.28	36.7	7.9	0.06	14.4	-0.3	0.86	18.0	3.6	0.19
	2	98.6	-6.2	0.64	25.8	-0.1	0.97	40.1	11.3	0.02a	12.9	-1.8	0.32	18.9	4.4	0.19
	3	119.5	-10.9	0.46	26.4	0.6	0.78	38.0	9.2	0.08	13.2	-1.6	0.43	20.5	6	0.11
	4	119.5	23.8	0.21	24.6	-1.3	0.55	35.5	6.6	0.22	12.9	-1.8	0.40	14.7	0.3	0.95
	5	122.7	16.8	0.39	24.4	-1.4	0.52	37.1	8.3	0.13	13.7	-1	0.63	13.4	-1	0.81
	6	123.9	18.0	0.40	25.7	-0.2	0.94	37.4	8.6	0.12	15.0	-0.1	0.96	11.6	-2.8	0.52
	7	127.2	21.1	0.34	23.4	-2.4	0.29	37.7	8.9	0.12	13.0	-2.1	0.34	11.0	-3.5	0.44

^aStatistically significant value (P < 0.05).

IVd: Inhaled vasodilators; NMBA: Neuromuscular blocking agents; PPV: Prone position ventilation.

DISCUSSION

To our knowledge, this is the first report to describe lung mechanics and outcomes among ARDS patients whom received different combinations of PPV, IVd and NMBA in COVID ARDS. In this retrospective multi-center study, we found that PPV alone and PPV in addition to IVd led to a significant improvement in P/F ratio during the days immediately following initiation of proning. The combination of PPV, IVd, and NMBA significantly improved tidal compliance on day 2, with an overall trend showing improvement in ventilator parameter over the 7 days; however, only day 2 reached significance. In comparison to patient’s whom remained supine, the PPV group had longer ICU and hospital LOS but no associated improvement in mortality or progression to VV-ECMO status compared to the supine group after adjusting for age and SOFA score. Analysis of baseline ventilatory parameters for the proned group showed significantly worse FiO₂ requirements, peak inspiratory pressure, and O₂ index, and lower PaO₂, SpO₂, and P/F ratio compared to the supine group which was a major confounder for this group. Additionally, no difference in ICU and hospital LOS, mortality, and VV-ECMO status was found among the treatment subgroups (e.g., PPV, PPV and IVd, PPV and NMBA, PPV, IVd, and NMBA).

Our hard outcome findings vary from other studies that have shown a mortality benefit in COVID ARDS patients who received PPV and those who have ARDS secondary to other infectious agents[23-25]. This difference in findings is may be explained by inadequate control of other contributing factors besides age and SOFA score, such as baseline ventilation parameters (e.g., O₂ index and driving pressure), that have been shown to predict 28-day mortality in COVID ARDS and ARDS patients, which were significantly worse in our treatment group[26-28]. Additionally, BMI and ethnicity varied significantly and were not controlled for in our analysis. At our center, patients with a P/F ratio less then 150 were selectively chosen for PPV, and likely had worse ventilation parameters secondary to disease progression and fibrosis development. One retrospective analysis showed COVID ARDS patients were mechanically ventilated for longer and given more adjunct therapies before VV-ECMO initiation compared to traditional ARDS patients[29]. Given this context and considering our treatment group had an inferior baseline clinical status, similar mortality and VV-ECMO rates may suggest a subclinical benefit that prevented further clinical deterioration.

Interestingly our subgroups of proned patient’s that were administered either IVd, NMBA or both had similar 7 day ventilator and respiratory characteristics suggesting the use of adjunct therapies may not impact overall clinical outcomes despite their recommended use in refractory cases. This is in line with other studies that have shown use of NMBA may increase 28-day mortality, prolong ICU stay, and delay discontinuation of mechanical ventilation in patient with COVID-19[15]. One study even noted NMBA was used for longer durations and at increasing rates for COVID ARDS patients compared to traditional ARDS patients in the LUNGSAFE study without a noticeable difference in extubation rates[30]. Results remain mixed on the efficacy of IVd in COVID ARDS patients, with one meta-analysis showing no mortality benefit or impact on hospital LOS[17]. In contrast, a small retrospective study did show a 29% mortality reduction in patients who responded to iEPO when it was initiated within an hour of PPV compared to those who did not respond[31]. However, other studies have found patients administered iEPO and PPV had longer hospital stays and on average patients with COVID ARDS stayed 6.5 days longer than those with ARDS[32]. This was likely attributed to the nature of the disease which has a multitude of processes involved that may contribute to progression and timing of patient illness[13]. In comparison to our findings, we may not have elicited a mortality benefit due to the delayed timing of IVd and PPV administration, in which these patients may have already been severely hypoxic at the time of initiation (e.g., P/F ratio in the low 100 seconds) and because of variation of therapy initiation due to provider discretion[33,34]. Further investigation is warranted regarding optimal timing of employing these adjunct therapies to impact clinical outcomes.

Many of the patients in our study were proned for days at a time at our center during this early phase of the pandemic. PPV alone had a significant improvement in P/F ratio from day 1 to day 4 but not in the last 3 days. In the landmark PROSEVA trial, patients with severe ARDS who were proned for at least 16 hours had a reduction in mortality[24]. In contrast, our patients on average received PPV for 60 hours (2.5 days) due to adaptations of standard protocols to meet the clinical needs of decompensating or desaturating patients. One recent randomized trial investigating the effects of prolonged prone positioning in patients with ARDS due to COVID-19 suggested that while prolonged prone positioning was associated with a longer ICU stay and increased use of neuromuscular blockers, it also led to greater muscular impairment. These findings highlight the need to balance the benefits of prone positioning with potential risks and to determine optimal session durations[35]. Similarly, other studies have found COVID ARDS survivors had higher P/F ratio when a goal directed PPV therapy was applied with a median PPV duration of 70.8 hours (2.95 days), in which PPV was used at higher rates and for longer durations (e.g., 43 hours vs 28 hours) in patients with COVID ARDS vs traditional ARDS[36,37]. In this regard, patients with COVID ARDS in our study required PPV 70% of the time compared to 16% of the time for patients with traditional ARDS in the LUNG SAFE study which examined ARDS treatment in 459 ICUs around the world[37]. Additionally, other studies have shown mechanically ventilated COVID ARDS patients who received prolonged PPV (at least 24 hours) compared to intermittent PPV laid out in the PROSEVA trial had reduced mortality suggesting intermittent supination may not be as efficacious due to fluctuations between dorsal and ventral atelectasis[13,38]. This caveat may be another reason our study did not elicit a mortality benefit as well. Questions about prolonged PPV and its impact on clinical outcomes in COVID ARDS still need to be further studied.

Other notable secondary outcomes include a significant increase in driving pressure for patients who received PPV on day 5, day 6, and day 7 and a significant improvement in P/F ratio from day 1 through 7 in patients who received both PPV and IVd, with an increase in plateau pressure by a mean of 2.6 cm H₂O on day 4 and day 6. Additionally, the combination of PPV, IVd, and NMBA significantly improved tidal compliance, a measure of lung elasticity, on day 2, with an overall trend showing improvement in ventilator parameter; however, this did not reach significance. These finding are in line with other studies showing IVd has an additive benefit in oxygenation in COVID ARDS patients when combined with PPV, showing clinically significant improvements in P/F ratio up to 10% with better response in patients who have lower P/F ratio[21,39,40]. One potentially reasoning for the improvement in P/F ratio could be the augmentation effects of iEPO on COVID ARDS pathophysiology including relaxing vascular smooth muscle, increasing pulmonary vasodilation, and stabilizing platelets[41]. Additionally, we found a significant increase of plateau pressure by a mean of 2.6 cm H₂O on day 4 with a meta-analysis suggesting the optimal value for plateau pressure should be around 27 cm H₂O on the first day of mechanical ventilation and below 32 cm H₂O in the first 3 days in patient with ARDS, which correlates with the range of all our sub-treatment groups excepts for the PPV and NMBA group, showing short-term mortality was higher in patients who had pressures above this range[42].

Literature examining the combined benefits of PPV, IVd and NMBA is very limited and ours appears to be the first that examined pulmonary mechanics in patients receiving all three in conjunction. This group was the sickest at baseline with an average P/F ratio of 61 and SOFA score of 12.2. Our study found no significant changes in lung mechanics in this cohort after receiving all three within 24 hours of each other but did note a positive trend in lung compliance in the first 4 days suggesting the combined use of these therapies might help reduce elastic resistance in lung tissue, a measure related to work or breathing and is useful in understanding disease progression and ventilator management[43,44]. A study by Grasselli et al[45], showed patients with COVID ARDS had 28% higher static compliance than traditional ARDS patients and suggested this difference in pathophysiology could be due to subpleural interstitial lung edema that leaves a preserved lung compliance during the early phase of the viral infection. The study also hypothesized compliance reduces later in the disease course due to an increase in lung inflammation from the severity of the infection that causes non-aerated portions of the lung parenchyma to increase. This may be why driving pressure is increased from day 5, day 6, and day 7 in the PPV alone group as lung compliance decreases (driving pressure = tidal volume/respiratory system compliance) in later disease stages[46]. However, other studies have shown a pattern of compliance that did not fluctuate throughout the initial two weeks attributing this difference to well documented impairment of hypoxic vasoconstriction due to endothelial damage[47,48]. More research examining the underlying pathology difference between traditional ARDS and COVID ARDS hypoxia is needed to guide subsequent administration of adjunct therapies.

Our study is the first to our knowledge to evaluate the impact of PPV in combination with adjunctive therapies IVd and NMBA, on COVID ARDS patient outcomes and longitudinal lung mechanics. The strength of our study is our ability to tease out the different treatment modalities using detailed chart reviews for type and duration of therapy. Additionally at our centers there was a single uniform protocol applied to dictate trials of adjunctive therapies. However, there are many limitations to our findings. The retrospective nature of our study does not allow us to control for other potentially confounding factors and the relatively small sample size among treatment subgroups hinder the power behind our conclusions concerning certain combinations of adjunct therapies. Additionally, our control group was small and limits the statistical power that would potentially show a difference in outcomes. Despite accounting for age and SOFA score, this did not adequately capture the variation of disease severity that can contribute to COVID ARDS outcomes such as severity of lung disease and initial pulmonary mechanics which were both significantly worse in proned patients. In the context of the novelty of COVID-19 during the initial surge, despite existence of standard care protocols at both hospitals, provider driven changes likely based on disease progression and patient factors heavily dictated timing and administration of adjunct therapies that we were not able to account for in our analytical model. Our study also predates the routine use of dexamethasone therapy for COVID-19 and does not factor in its impact on lung mechanics.

CONCLUSION

In mechanically ventilated patients diagnosed with moderate to severe COVID ARDS, PPV and PPV with the addition of IVd was associated with a significant and sustained increase in P/F ratio. The combination of PPV, IVd and NMBA improved compliance however this did not reach significance. Outcomes were similar among subgroups however this is likely a reflection of our small sample size. Lung mechanics in COVID ARDS appeared to respond to proning, IVd and NMBA similar to other causes of ARDS. Clinicians could consider time limited trial of IVd and NMBA after an initial failed proning session and if no improvement is seen in ventilator parameters or P/F ratio, consultation for VV-ECMO is likely warranted. More research needs to be done to better stratify the efficacy of how and when to implement ARDS adjunctive therapies in mechanically ventilated patients with moderate to severe ARDS to optimize outcomes and lung mechanics.

ACKNOWLEDGEMENTS

We would like to acknowledge Sameer Desale, M.Sc. whom is a biomedical statistician and performed the statistical analysis of this work.