Ventilatory Parameters in Obstetric Patients With COVID-19 and Impact of Delivery

Abstract

Background

Current evidence on obstetric patients requiring advanced ventilatory support and impact of delivery on ventilatory parameters is retrospective, scarce, and controversial.

Research Question

What are the ventilatory parameters for obstetric patients with COVID-19 and how does delivery impact them? What are the risk factors for invasive mechanical ventilation (IMV) and for maternal, fetal, and neonatal mortality?

Study Design and Methods

Prospective, multicenter, cohort study including pregnant and postpartum patients with COVID-19 requiring advanced ventilatory support in the ICU.

Results

Ninety-one patients were admitted to 21 ICUs at 29.2 ± 4.9 weeks; 63 patients (69%) delivered in ICU. Maximal ventilatory support was as follows: IMV, 69 patients (76%); high-flow nasal cannula, 20 patients (22%); and noninvasive mechanical ventilation, 2 patients (2%). Sequential Organ Failure Assessment during the first 24 h (SOFA₂₄) score was the only risk factor for IMV (OR, 1.97; 95% CI, 1.29-2.99; P = .001). Respiratory parameters at IMV onset for pregnant patients were: mean ± SD plateau pressure (PP), 24.3 ± 4.5 cm H₂O; mean ± SD driving pressure (DP), 12.5 ± 3.3 cm H₂O; median static compliance (SC), 31 mL/cm H₂O (interquartile range [IQR], 26-40 mL/cm H₂O); and median Pao₂ to Fio₂ ratio, 142 (IQR, 110-176). Respiratory parameters before (< 2 h) and after (≤ 2 h and 24 h) delivery were, respectively: mean ± SD PP, 25.6 ± 6.6 cm H₂O, 24 ± 6.7 cm H₂O, and 24.6 ± 5.2 cm H₂O (P = .59); mean ± SD DP, 13.6 ± 4.2 cm H₂O, 12.9 ± 3.9 cm H₂O, and 13 ± 4.4 cm H₂O (P = .69); median SC, 28 mL/cm H₂O (IQR, 22.5-39 mL/cm H₂O), 30 mL/cm H₂O (IQR, 24.5-44 mL/cm H₂O), and 30 mL/cm H₂O (IQR, 24.5-44 mL/cm H₂O; P = .058); and Pao₂ to Fio₂ ratio, 134 (IQR, 100-230), 168 (IQR, 136-185), and 192 (IQR, 132-232.5; P = .022). Reasons for induced delivery were as follows: maternal, 43 of 71 patients (60.5%); maternal and fetal, 21 of 71 patients (29.5%); and fetal, 7 of 71 patients (9.9%). Fourteen patients (22.2%) continued pregnancy after ICU discharge. Risk factors for maternal mortality were BMI (OR, 1.10; 95% CI, 1.006-1.204; P = .037) and comorbidities (OR, 4.15; 95% CI, 1.212-14.20; P = .023). Risk factors for fetal or neonatal mortality were gestational age at delivery (OR, 0.67; 95% CI, 0.52-0.86; P = .002) and SOFA₂₄ score (OR, 1.53; 95% CI, 1.13-2.08; P = .006).

Interpretation

Contrary to expectations, pregnant patient lung mechanics were similar to those of the general population with COVID-19 in the ICU. Delivery was induced mainly for maternal reasons, but did not change ventilatory parameters other than Pao₂ to Fio₂ ratio. SOFA₂₄ score was the only risk factor for IMV. Maternal mortality was associated independently with BMI and comorbidities. Risk factors for fetal and neonatal mortality were SOFA₂₄ score and gestational age at delivery.

Graphical Abstract

FOR EDITORIAL COMMENT, SEE PAGE 473

Take-home Points.

Study Question: What are ventilatory parameters for obstetric patients with COVID-19 and how does delivery impact them?

Results: Lung mechanics of pregnant patients were similar to those of the general ICU population with COVID-19, and induced delivery did not improve ventilatory parameters associated with mortality in patients with ARDS.

Interpretation: Recommended ventilatory settings for general patients with ARDS also should be used for pregnant patients with ARDS, and delivery should not be indicated to improve maternal lung mechanics.

Evidence on obstetric patients requiring invasive mechanical ventilation (IMV) and the impact of delivery on maternal respiratory mechanics and oxygenation is scarce, controversial, and limited to retrospective studies.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 Before the COVID-19 pandemic, most studies included myriad causes of respiratory failure leading to IMV, either obstetric (preeclampsia, hemorrhage, and so forth) or nonobstetric (pneumonia, sepsis, and so forth), with only 10 to 71 patients included.5, 6, 7 ^{, 10} Effects on respiratory parameters after delivery proved different in these two categories, preventing us from generalizing conclusions for specific diseases such as pneumonia. Evidence supporting use of other respiratory support, for example, high-flow nasal cannula (HFNC) or noninvasive mechanical ventilation (NIMV), in obstetric patients was even poorer and was based on case reports or series of no more than four patients.12, 13, 14, 15, 16

During the second wave of the pandemic, pregnant patients were particularly affected, presenting an increased risk of ICU admission and IMV compared with nonpregnant reproductive-age women.¹⁷ To date, few case reports and retrospective case series—primarily from high-income countries—have described obstetric patients with COVID-19 requiring IMV.1, 2, 3, 4 ^{, 8 , 9 , 11 , 18} Evidence-based decisions supporting terminating or continuing pregnancy in patients with pneumonia requiring any kind of ventilatory support remains up for debate. Given this scenario, we launched a prospective multicenter binational (Argentina and Colombia) study including pregnant and postpartum patients with COVID-19 requiring any type of advanced ventilatory support in the ICU. Our primary objective was to describe maternal ventilatory parameters and the impact of delivery on them. Our secondary objective was to describe risk factors for IMV and maternal, fetal, and neonatal mortality.

Study Design and Methods

We followed the Strengthening the Reporting of Observational Studies in Epidemiology statement checklist for this study.¹⁹ This was a multicenter, binational (Argentina and Colombia), prospective, cohort study conducted in 20 ICUs in seven provinces in Argentina and one ICU in Colombia (16 medical-surgical ICUs and five obstetric ICUs). Pregnant or postpartum (< 42 days) patients admitted to the ICU between March 20 and December 31, 2021, because of COVID-19, confirmed by polymerase chain reaction testing of nasopharyngeal specimens requiring any type of advanced respiratory support—HFNC, NIMV, or IMV—were included.

Demographic data collected included level of education (years), comorbidities (Charlson comorbidity index score,²⁰ Acute Physiology and Chronic Health Evaluation [APACHE] II score,²¹ and Sequential Organ Failure Assessment during the first 24 h [SOFA₂₄] score²²), ICU length of stay (LOS), hospital LOS, maternal mortality, and fetal or neonatal mortality (spontaneous or induced abortion, fetal death, and 28-d mortality), hypertension, and obesity.²³ Obstetric history included antepartum or postpartum admission; gestational age on admission to the ICU, at delivery (weeks), or both; preterm birth (< 37 weeks) classified as late or early preterm (32-36.6 weeks and < 32 weeks, respectively)²⁴; minimal vs standard prenatal care (minimum one vs five visits for term pregnancies)²⁵; parity; frequency and type of fetal monitoring; route of delivery (vaginal or C-section); and reason for induced delivery, if occurred (maternal, fetal, or both). Also collected were respiratory support on admission (nasal prongs, nonrebreathing masks [NRMs], HFNC, NIMV, and IMV), type of maximum ventilatory support (HFNC, NIMV, and IMV), tracheostomy status, and prone positioning status. Pao ₂ to Fio ₂ ratio was registered on admission, before IMV (< 2 h), after intubation (days 0, 2, and 7) and before (< 2 h) and after (< 2 h and at 24 h) delivery. For those requiring IMV, ventilatory support before IMV, peripheral oxygen saturation before intubation, and mechanical ventilation parameters (positive-end expiratory pressure, plateau pressure [PP], driving pressure [DP], and static compliance) after intubation (days 0, 2, and 7) and before (< 2 h) and after (< 2 h and at 24 h) delivery were recorded. Neuromuscular blocker and vasopressor treatments were recorded. Treatments received included steroid type and dose (per the Randomized Evaluation of COVID-19 Therapy- RECOVERY-trial: for pregnant patients, prednisolone 40 mg or hydrocortisone 80 mg bid, and for postpartum patients, dexamethasone 6 mg/d for 10 days²⁶), convalescent plasma, tocilizumab, and anticoagulation. ICU complications were detailed as follows: ARDS,²⁷ preeclampsia,²⁸ septic shock,²⁹ acute renal failure requiring dialysis, pulmonary embolism or DVT,³⁰ and acquired infections.

Statistical Analysis

Categorical variables are presented as number (percentage) and continuous variables are presented as mean ± SD or median (interquartile range [IQR]). Variables were compared with the Student t test, Wilcoxon test, χ ² test, or Fisher exact test. Multiple comparisons of parametric and nonparametric variables were analyzed with repeated-measures analyses of variance and Friedman tests, respectively, whereas both post hoc analyses were carried out with Bonferroni correction. A P value of ≤ .05 was considered significant. A binary logistic regression was built to find independent predictors of IMV and maternal, fetal, and neonatal mortality. Variables with P value of ≤ .20 were included in univariate analysis. The multivariate model included variables with a P value of ≤ .05 on Wald test, confounding effects (variation coefficient, ≥ 20%), or both. The model was calibrated with the Hosmer-Lemeshow goodness-of-fit test to evaluate discrepancy between observed and expected ICU mortality and IMV. SPSS version 15 software (SPSS, Inc.) was used for analysis.

Ethical Considerations

This study was approved by the institutional review board of the Argentinian Society of Critical Care Medicine (Identifier: PRIISABA-5012). It was performed in accordance with ethical standards laid down in the 1964 Declaration of Helsinki and later amendments and with provincial laws regarding research on human beings. Patient information was de-identified before analysis. Informed written consent was waived; instead, written information regarding the study was given to patients and relatives.

Results

Population Characteristics and Outcomes

Forty-three ICUs were enrolled, but 22 were excluded for not recruiting patients. Twenty-one ICUs recruited 91 pregnant or postpartum patients, 73 patients (80%) from the public health sector and 18 patients (20%) from the private health sector. General and obstetric characteristics and differences between survivors and nonsurvivors are presented in Table 1 . Organ dysfunction was frequent and was associated with differential mortality (Fig 1 ).

Table 1.

Characteristics and Outcomes of 91 Pregnant or Postpartum Patients With COVID-19 Requiring Advanced Respiratory Support in 21 ICUs in Argentina and Colombia During 2021

Variable	All	ICU Status		P Value
		Survivors	Nonsurvivors
No. of patients	91	75 (82.4)	16 (17.6)	. . .
Age, y	31 ± 6	31.5 ± 6	31.7 ± 7	.90
Level of education, y	11 ± 3.5	11 ± 3.5	10 ± 3.6	.43
Public health sector	73 (80.2)	57 (76)	16 (100)	.04
No comorbiditya	74 (81.3)	65 (86.6)	9 (56.25)	.005
BMI	30 ± 6	28.9 ± 5.7	33.2 ± 7.6	.013
Obesity	36 (39.5)	26 (34.6)	10 (62.5)	.05
Obesity class				.021b
I (30-34.9 kg/m2)	17 (18.7)	16 (21.3)b	1 (6.25)
II (35-39.9 kg/m2)	15 (16.5)	8 (10.6)	7 (43.75)
III (≥ 40 kg/m2)	4 (4.4)	2 (2.7)	2 (12.5)
Vaccination (one dose)c	8/75 (10.6)	5/62 (8)	3/13 (23)	.28
APACHE II score	13 (7-16)	12 (6-18)	14 (9.5-14.5)	.31
APACHE II risk, %	14 (5.2-21)	12.6 (5.1-23)	15 (7.9-19.3)	.41
SOFA24 scored	4 (3-6)	4 (3-6)	5 (4-7)	.077
LOS, d
ICU	15 (8-28)	16 (8-31)	11 (9-15)	.12
Hospital	22 (11-36)	23 (13-37)	12 (10-17)	.01
Obstetric characteristics
Antepartum admission	63/91 (69.2)	54/75 (72)	9/16 (56.25)	.21
Parity	2 (1-3)	2 (1-3)	2 (1-3)	.65
Minimal prenatal caree	84 (92)	70 (93)	14 (87.5)	.43
Standard prenatal caref	43 (47)	38 (55)	5 (31)	.36
Gestational age on admission, wk	31 ± 5	31 ± 4	29.5 ± 7	.28
Gestational age at delivery, wk	32.5 ± 3	32.3 ± 3	30.9 ± 6.5	.24
Births, wk	74	61	13	.79
Term, ≥ 37	7 (9.5)	6 (9.8)	1 (7.7)	. . .
Late preterm, 32-36.6	40 (54)	33 (54.1)	7 (53.8)	. . .
Early preterm, < 32	27 (36.5)	22 (36.1)	5 (38.5)	. . .
Route of delivery				.68
Cesarean section	70 (96)	56	13
Vaginal	3 (4)		0

Data are presented as No. (%), mean ± SD, or median (interquartile range), unless otherwise indicated. APACHE = Physiology and Chronic Health Evaluation; LOS = length of stay; SOFA₂₄ = Sequential Organ Failure Assessment during the first 24 h.

^aCharlson comorbidity index score of 0.

^bP = .01 obesity class I vs class II.

^cNo patient received a full vaccination schedule; only eight patients in total received one dose.

^dWithin the first 24 h of ICU admission.

^eMinimal prenatal care: at least one visit for term pregnancies.

^fStandard prenatal care: at least five visits for term pregnancies.

Figure 1.

Bar graph showing organ dysfunction during the first 24 h in the ICU and its corresponding ICU mortality among 91 pregnant or postpartum patients with COVID-19.

Sixteen maternal deaths (17.5%) occurred and 14 fetal or neonatal losses (15.4%) occurred; however, only five of these deaths or losses were both maternal and fetal. BMI and presence of at least one comorbidity (Charlson comorbidity index score, ≥ 1) were the only risk factors for maternal mortality (OR, 1.10 [95% CI, 1.006-1.204; P = .037] and OR, 4.15 [95% CI, 1.212-14.20; P = .023], respectively). Of the 14 fetal or neonatal losses, two were spontaneous abortions, four fetal deaths, and eight were neonatal deaths. Of the four fetal deaths, three were nonviable fetuses (two at 23 weeks and one at 26.1 weeks), and one at 35 weeks was the fetus of a severely ill patient (APACHE II score, 22; Pao ₂ to Fio ₂ ratio, 87; SOFA₂₄ score, 5). The threshold for fetal viability in Argentina is considered to be approximately 26 weeks.³¹ Risk factors for fetal or neonatal mortality were gestational age at delivery and SOFA₂₄ score (Table 2 ).

Table 2.

Multivariate Analysis of Variables Associated With Fetal or Neonatal Mortality Among 91 Pregnant or Postpartum Patients Admitted to the ICU Because of COVID-19 in 21 ICUs in Argentina and Colombia

Variable	OR (95% CI)a	P Value
Sequential Organ Failure Assessment during the first 24 h	1.53 (1.13-2.08)	.006
Gestational age at delivery (wks)	0.67 (0.52-0.86)	.002

^aHosmer and Lemeshow test significance, 0.82.

Respiratory Support, Ventilatory Parameters, and Impact of Delivery

Respiratory support on admission and differences between survivors and nonsurvivors is illustrated in Figure 2 . Of the 36 patients admitted receiving HFNC treatment, 24 patients (66.6%) required IMV, and of the 28 patients admitted receiving NRM, 21 patients (75 %) eventually required IMV (P = .48). Respiratory support on admission did not affect days receiving IMV or ICU LOS (Table 3 ). Among patients receiving noninvasive oxygen support on admission (n = 49), the only independent risk factor for IMV was SOFA₂₄ score (OR, 1.97; 95% CI, 1.29-2.99; P = .001), even after adjusting for APACHE II score and BMI.

Figure 2.

Bar graph showing type of respiratory support on admission for 91 pregnant or postpartum patients with COVID-19 and differences between survivors and nonsurvivors. ^aP = .71 for comparison of type of respiratory support on admission between SV and NonSV groups. HFNC = high-flow nasal cannula; IMV = invasive mechanical ventilation; NIMV = noninvasive mechanical ventilation; NonSV = nonsurvivors; NP = nasal prong; NRM = nonrebreathing mask; SV = survivor.

Table 3.

No. of Days of IMV and ICU LOS According to Ventilatory Support on Admission Among Obstetric Patients With COVID-19 Requiring IMV as Maximum Ventilatory Support in 21 ICUs in Argentina and Colombia During 2021

Variable	Type of respiratory support on admission					PValue
	HFNC	NRM	NP	NIMV	IMV
No. of patients	24	21	2	2	20	. . .
IMV duration, d	10 (10-20)	14 (5-23)	18 (11-32)	18 (7-28)	21 (11-35)	.22
ICULOS, d	14 (11-27)	18 (11-32)	21 (18-24)	26 (15-36)	30 (18-43)	.07

Data are presented as median (interquartile range), unless otherwise indicated. HFNC = high-flow nasal cannula; IMV = invasive mechanical ventilation; LOS = length of stay; NIMV = noninvasive mechanical ventilation; NP = nasal prong; NRM = nonrebreathing mask.

Sixty-nine patients required IMV as maximum ventilatory support, 16 of whom died (23.2%). Forty-seven of these patients were pregnant at IMV onset (day 0); mechanical ventilation characteristics and differences between survivors and nonsurvivors over the first week are presented in Table 4 and Figure 3 . Delivery did not improve ventilatory parameters other than oxygenation (Table 5 , Fig 4 ).

Table 4.

Ventilatory and Oxygenation Parameters Among 47 Pregnant Patients Requiring IMV Because of COVID-19 and Differences Between Survivors and Nonsurvivors in 21 ICUs in Argentina and Colombia During 2021

Variable	All (n = 47)	Survivors (n = 38)	Nonsurvivors (n = 9)	P Value
Pao2 to Fio2 ratio before IMV (< 2 h)	121 (86.5-162.2)	123.5 (87.7-174)	102.5 (90-129)	.20
SpO2 before IMV, %	90 (88-94)	91 (90-95)	89 (88-90)	.055
Pao2 to Fio2 ratio on IMV day 0 (< 2 h)	142 (110-176)	147.5 (120-176)	120 (100-142)	.17
Pao2 to Fio2 ratio
Day 2	200 (139-232.2)	203 (157-230)	144 (105-237)	.31
Day 7	183 (141.75-235.5)	186 (168-236)	100 (80-188)	.08
PEEP, cm H2O
Day 0	11.1 ± 2.7	11 ± 2.4	11.8 ± 3.8	.45
Day 2	11 ± 3	10.7 ± 2.6	11.8 ± 4	.31
Day 7	10.2 ± 2.7	9.9 ± 2.4	11.8 ± 3.5	.10
PP, cm H2O
Day 0	24.3 ± 4.5	23.7 ± 4.5	26.8 ±3.9	.062
Day 2	23.8 ± 6.4	22.7 ± 5.7	29.2 ± 7.1	.011
Day 7	25.1 ± 5.9	23.7 ± 4.7	31.8 ± 7.2	.001
SC, mL/cm H2O
Day 0	31 (26-40)	32.5 (26-41)	27 (23-38)	.48
Day 2	33 (23.2-42)	33 (24-42)	27 (21-40)	.75
Day 7	29 (23.7-38)	30 (24-38)	29 (21-34.5)	.83
DP, cm H2O
Day 0	12.5±3.3	12.1 ± 3.5	14.1 ±2.1	.10
Day 2	12.6 ± 4.3	12.1 ± 4	15.3 ± 4.8	.069
Day 7	13.8 ± 4.8	12.7 ± 3.7	19.3 ± 4.1	.000

Data are presented as No. (%), mean ± SD, or median (interquartile range), unless otherwise indicated. DP = driving pressure; IMV = invasive mechanical ventilation; PEEP = positive-end expiratory pressure; PP = plateau pressure; SC = static compliance; SpO₂ = peripheral oxygen saturation.

Figure 3.

A-D, Box-and-whisker plots showing mechanical ventilation and oxygenation parameters over the first week after intubation for 47 patients admitted before delivery because of COVID-19 and differences between survivors and nonsurvivors. A, Plateau pressure. B, Driving pressure. C, Static compliance. D, Pao₂ to Fio₂ ratio. ^aDifference between survivors (blue) and nonsurvivors (red) (P < .05) that were significant for plateau pressure at days 2 and 7, and driving pressure at day 7. ^bDifferences in Pao₂ to Fio₂ ratio over the first week for the entire group (gray) (P = .000): post hoc Bonferroni significant differences between day 0 vs day 2 and day 0 vs day 7, P = 000.

Table 5.

Ventilatory and Oxygenation Parameters Before (< 2 h) and After (< 2 h and 24 h) Delivery Among 47 Pregnant Patients With COVID-19 Requiring Invasive Mechanical Ventilation in 21 ICUs in Argentina and Colombia During 2021

Variable	Before Delivery (< 2 h)	After Delivery (< 2 h)	After Delivery (24 h)	P Value
PP, cm H2O	25.6 ± 6.6	24 ± 6.7	24.6 ± 5.2	.59
DP, cm H2O	13.6 ± 4.2	12.9 ± 3.9	13 ± 4.4	.69
SC, mL/cm H2O	28 (22.5-39)	30 (24.5-44)	30 (24.5-44)	.058
Pao2 to Fio2 ratio	134 (100-230)	168 (136-185)	192 (132-232.5)	.022a
PEEP, cm H2O	11 ± 3	12 ± 4	11 ± 2	.8

Data are presented as No. (%), mean ± SD, or median (interquartile range), unless otherwise indicated. DP = driving pressure; PEEP = positive-end expiratory pressure; PP = plateau pressure; SC = static compliance.

^aIn the post hoc Bonferroni analysis, differences were observed between the subgroup before delivery and the subgroup at 24 h after delivery (P = .025).

Figure 4.

A-D, Box-and-whisker plots showing mechanical ventilation and oxygenation parameters before (< 2 h) and after (< 2 h and 24 h) delivery among 47 patients with COVID-19 admitted before delivery. A, Plateau pressure. B, Driving pressure. C, Static compliance. D, Pao₂ to Fio₂ ratio. Comparisons of variables before (< 2 h) and after (< 2 h and 24 h) delivery: ^aD, Pao₂ to Fio₂ ratio = P = .022 for the entire group (P = .025 between before [< 2 h] and at 24 h of delivery in the post hoc Bonferroni analysis).

Twenty patients required HFNC as maximum respiratory support, 12 of whom were admitted and continued to receive it until discharge, whereas three and five patients, respectively, required HFNC after trials of nasal prongs and NRM. Patients continuing to receive HFNC vs those requiring IMV differed in, respectively, BMI (mean ± SD, 27.2 ± 3.5 kg/m² vs 30.2 ± 6 kg/m²; P = .010), APACHE II score (median, 7 [IQR, 5.5-12.5] vs 14 [IQR, 9-19]; P = .001), and SOFA₂₄ score (median, 2.5 [IQR, 2-4] vs 5 [IQR, 4-7]; P = .000). No maternal, fetal, or neonatal deaths occurred among the HFNC group.

Four patients required NIMV during ICU stay: two were receiving NIMV on admission and treatment failed in both, which required IMV; two patients were admitted receiving NRM, but required NIMV as maximum respiratory support. The latter two patients received NIMV for 5 and 9 days, respectively, were admitted before delivery (at 28 and 35.3 weeks), and successfully continued their pregnancies after ICU discharge. Both showed obesity class II (BMI, 38.7 kg/m² and 39.1 kg/m²), but no other comorbidities according to Charlson comorbidity index score. Pao ₂ to Fio ₂ ratio on admission was 150 and 250, and SOFA₂₄ score was 3 for both patients.

Obstetric and Delivery Characteristics

Twenty-eight patients (31%) delivered at a mean ± SD of 34 ± 2.7 weeks and were admitted to ICU after delivery, whereas 63 patients (69%) were admitted to the ICU before delivery (at a mean ± SD of 29.2 ± 4.9 weeks) and delivered there at a mean ± SD of 31 ± 4.4 weeks (P = .002 for gestational age at delivery for patients admitted after delivery vs before delivery). Fetal monitoring in ICU consisted mainly of intermittent fetal heart rate monitoring (18/21 ICUs [86%]) every 8 to 12 h and Doppler ultrasound (15/21 [71.4%]) every 2 to 3 days; only one ICU performed continuous cardiotocography. Reasons for induced delivery were: maternal (43/71 [60.5%]), maternal and fetal (21/71 [29.5%]), and fetal (7/71 [9.9%]). Among the 63 patients admitted before delivery, 43 patients (68.3%) required delivery during ICU stay, 14 patients (22.2%) were discharged pregnant, two patients (3.2%) experienced spontaneous abortion, and fetal death occurred in four patients (6.3%). Most patients admitted before delivery who required delivery gave birth during the first 24 to 48 h of ICU admission; population median was 3 days (IQR, 1-6.75 days). For the 14 patients who continued pregnancy after ICU discharge, maximum respiratory support in the ICU was HFNC in eight patients (57%), IMV in four patients (29%), and NIMV in two patients (14%); median Pao ₂ to Fio ₂ ratio on admission was 157.5 (IQR, 120-240) vs 180 (IQR, 100-240) for those requiring ICU delivery (P = .9).

ICU Management and Complications

Of 69 patients requiring IMV, 64 patients (92.7%) received neuromuscular blockers, 25 patients (36.2%) required tracheostomy, and 49 patients (71%) required prone positioning, 33 patients (67%) of whom were admitted before delivery with a median of 3 proning sessions (IQR, 1-3 sessions) for a median duration of 30 h (IQR, 21-48 h) each. Twenty-six of the 49 patients (53%) requiring prone positioning received pressure-point protection. Fifteen of the 49 patients (30.6%) who underwent prone positioning demonstrated skin lesions.

In the entire population, 46 patients (50.5%) experienced some type of hospital-acquired infection, 35 patients (38.5%) showed septic shock, 13 patients (14.3%) showed acute renal failure requiring dialysis, and 15 patients (16.4%) demonstrated preeclampsia. DVT or pulmonary embolism were documented in two patients (2.2%).

All patients (n = 91) received steroids, 61 patients (67%) per RECOVERY trial types and dosages.²⁵ However, 30 patients (33%) received nonstandardized steroids, dosages, or both. No patient received convalescent plasma or tocilizumab. Sixty-two patients (68%) received vasopressors, and 35 patients (38.5%) were treated with anticoagulants.

Discussion

To our knowledge, this is the first prospective study evaluating pregnant and postpartum patients admitted to the ICU with respiratory failure requiring advanced ventilatory support that sequentially analyzed the impact of delivery on lung mechanics and oxygenation. The use of HFNC did not prevent intubation. The only risk factor for IMV was SOFA₂₄ score. Ventilatory parameters in pregnant patients were similar to those in general ICU populations with COVID-19. Delivery was induced mainly for maternal reasons, but did not change ventilatory parameters other than Pao ₂ to Fio ₂ ratio. Twenty-two percent of patients were discharged pregnant with good outcomes. Risk factors for maternal mortality were BMI and presence of comorbidities. Risk factors for fetal or neonatal mortality were SOFA₂₄ score and gestational age at delivery.

Population Characteristics and Outcomes

Ninety-one pregnant or postpartum patients with COVID-19 requiring advanced ventilatory support in the ICU were included in this multicenter, binational, prospective cohort study. Most patients were treated in the public health sector and none were vaccinated fully; this correlates with the time frame of the study, which was performed when vaccination was not recommended for pregnant patients.¹⁸ Although in Argentina and Colombia the health-care system comprises public and private sectors, the quality of care is comparable in both sectors.^{32 , 33} Although most patients demonstrated no comorbidities, the presence of at least one comorbidity was associated independently with maternal mortality, as was described for the general population with COVID-19.^{34 , 35} Forty percent of patients presented obesity, also associated independently with increased maternal mortality. As noted in previous studies, pregnant patients with COVID-19 experience higher maternal, fetal, and neonatal mortality compared with pregnant patients without COVID-19.³⁶ They also have higher ICU and IMV requirements compared with nonpregnant patients with COVID-19, even those with no comorbidities or obesity.³⁷

In this study, mortality in patients requiring IMV (23.2%) was considerably lower than in the general ICU population with COVID-19 (37%-57%),^{34 , 35} which may be explained by younger age and recovery capacity of these patients. However in this cohort, maternal mortality numbers were between those described for obstetric patients in high-income countries (0%-8.4%)^{1 , 4 , 8 , 9 , 11 , 18 , 37 , 38} and upper and lower middle-income countries (78%-100%),^{2 , 3} which may be related to health budgets, patient characteristics such us underlying diseases, and health-care variables such as quality of care, accessibility, or system overload, among others.

Patients were severely ill on admission according to the SOFA₂₄ score, similar to critically ill obstetric patients with other community-acquired pneumonias.³⁹ Respiratory and cardiovascular dysfunctions were the most frequent, and their associated mortality coincided with previous reports (18% and 16%, respectively).³⁹ However, renal dysfunction, although less frequent, was associated with the highest mortality in this group (23%) and exhibited higher mortality compared with the general critically ill obstetric population without COVID-19 (15%).³⁹ The proportion of obstetric patients with COVID-19 requiring dialysis was 14.3%, compared with 3.3% among critically ill obstetric patients in general,³³ which may explain why this particular group showed worse outcomes associated with renal dysfunction.

Respiratory Support, Ventilatory Parameters, and Impact of Delivery

The most frequently used types of ventilatory support on admission were HFNC and NRM, followed by IMV; however, neither type of respiratory support nor oxygenation parameters on admission had an impact on maternal mortality. Furthermore, use of HFNC did not prevent IMV requirement compared to use of NRM, contrary to what was reported in the last meta-analysis.⁴⁰ Discrepancies between our results and those from the meta-analysis may be related to sample size issues or different populations evaluated. In our study, 20 patients (22%) were able to continue to receive HFNC as maximum respiratory support, with excellent maternal, fetal, and neonatal outcomes. These patients were less severely ill on admission and showed lower BMI than those requiring IMV. SOFA₂₄ score was the only risk factor for IMV among patients admitted with noninvasive respiratory support. This may provide additional information to consider before intubating these patients.

Most patients required IMV as maximum respiratory support and most were pregnant when intubated. To our knowledge, this is the first prospective study describing obstetric patients with community acquired pneumonia requiring IMV and the impact of delivery on ventilatory parameters. Current evidence available is from retrospective case series of patients with respiratory failure resulting from myriad obstetric and nonobstetric causes,5, 6, 7 ^{, 10} and limited retrospective case series of patients with COVID-19.1, 2, 3, 4 ^{, 8 , 9 , 11 , 18 , 38} Pregnancy is characterized by decreased chest wall compliance because of a gravid uterus, commonly believed to impact PP or DP as well as positive-end expiratory pressure settings, leading some to suggest permitting, for example, higher PP (35 cm H₂O).⁴¹ However, one relevant finding of our study is that pregnant patient lung mechanics were similar to those from the general ICU population with COVID-19.^{34 , 35} Therefore, because potential differences were not evident, we need to dig deeper for the explanation. In patients with acute lung injury, the impact of the chest wall on respiratory mechanics was described insufficiently and, excluding those with intraabdominal hypertension,⁴² it might have been overestimated.⁴³ Changes in respiratory mechanics seem to be determined by abnormalities in lung vs chest wall mechanics.⁴⁴ Finally, although in pregnant patients intraabdominal pressure can be higher than in nonpregnant critically ill patients, presence of intraabdominal hypertension is rare.⁴⁵ These findings have clear clinical implications because they may guide health-care professionals to adjust ventilator settings for pregnant patients with nonobstetric causes of respiratory failure to those proven to reduce mortality in general ICU population, such as PP of < 30 cm H₂O.⁴⁶

One of the most controversial issues regarding pregnant patients with respiratory failure is the impact of delivery on maternal lung mechanics and oxygenation. To date, the evidence available comes from four retrospective case series that indicate some improvement in oxygenation after delivery, with only one describing improvement in DP.^{5 , 7 , 10 , 18} Tomlinson et al¹⁰ found decreased oxygen requirements after delivery in 10 patients receiving IMV (nine with pneumonia). Lapinsky et al⁷ described improvement in oxygenation index after delivery in three of 10 patients and static compliance in five of 10 patients, most of whom were ventilated for nonobstetric causes. Hung et al⁵ recorded improvement in oxygenation and peak pressure after delivery, but only in the subgroup of patients with obstetric causes of respiratory failure. Finally, Péju et al¹⁸ described improvement in oxygenation and inconsistent improvement in DP after delivery. This is the first prospective study evaluating impact of delivery on maternal lung mechanics and oxygenation. In pregnant patients with COVID-19, delivery did not change PP or DP, which are respiratory parameters strongly associated with mortality in patients with ARDS^{34 , 47}; instead, it improved only Pao ₂ to Fio ₂ ratio, which is not associated independently with mortality. Our results for DP are not consistent with those of the COVIDPREG study¹⁸; however, their own limitations noted large interindividual variability. Furthermore, its retrospective nature could have increased missing data and reporting bias. In most of these patients, delivery was chosen for maternal reasons, as described by Eman et al,² highlighting another key finding of this study: unsupported beliefs lead to the removal of the fetus under the assumption that this will automatically improve maternal parameters.^{5 , 6}

Obstetric and Delivery Characteristics

After maternal stabilization, interrupting pregnancy for fetal reasons when fetal alterations persist is completely warranted because continuing pregnancy would not benefit the fetus. Notably, in this population, nonviable fetuses died, with only one death occurring at 35 weeks, but this was because of an extremely adverse maternal condition. Fetal and neonatal mortality was associated independently with gestational age and SOFA₂₄ score. Risking the fetus without significantly improving maternal conditions could be eliminated with these results. In this study, most deliveries were performed within 48 h after admission, meaning that SOFA₂₄ score was readily available and potentially might have saved fetuses. Physicians should strive for equipoise to obtain maximum benefit for both mother and child.

ICU Interventions and Complications

Evidence about prone positioning during pregnancy is scarce.^{11 , 18 , 48} In this study, 33 patients required prone positioning for a median of 30 h during pregnancy. Prone positioning was deemed feasible, resulting in only skin complications, which might have been related to inadequate use of pressure point protection. This information may encourage physicians to use prone positioning—so beneficial in treating patients with ARDS—in this particular group of patients as well.⁴⁹

In terms of treatment, although all patients received steroids, one-third did not receive an evidenced-based formula or dose, per the RECOVERY trial.²⁶ None received tocilizumab. In most patients receiving anticoagulants, thrombosis was not documented, underscoring common, unsatisfactory medical treatment for pregnant patients.^{2 , 18}

Strengths and Limitations

Strengths of this study are in the number of obstetric patients included and its prospective design with consecutive enrollment. Community-acquired pneumonia requiring advanced respiratory support during pregnancy is uncommon, but the pandemic affected obstetric patients in particular, increasing the number of cases to evaluate. The prospective design allowed us to minimize missing data, to improve data quality, and to measure variables before outcomes occur, contributing to causal pathways. It also afforded us the opportunity to gather sequential data related to impact of delivery on maternal respiratory parameters, the first study ever to do so.

Limitations of this study are its observational nature, which could have introduced some bias and confounding, although the latter was controlled by regression analysis. Selection bias also might have occurred because of overrepresentation of patients treated in the public health sector. Response bias is possible because of inclusion of qualified ICUs more prone to participate in studies. Generalizability could be uncertain. Although evaluating the impact of maternal hypercapnia on fetal outcomes would have provided relevant information, we opted not to request daily Pco ₂ given pandemic workload considerations. Finally, neonatal morbidity was not measured; instead, we focused entirely on fetal and neonatal mortality, because it is a nonbiased outcome.

Interpretation

Maternal mortality in patients with COVID-19 requiring advanced ventilatory support was associated independently with BMI and presence of comorbidities, which are outside the control of critical care practitioners. Pregnant patient lung mechanics and ventilator settings were similar to those in general ICU populations with COVID-19, supporting the use of evidenced-based ventilator settings proven to reduce mortality in nonobstetric patients with ARDS. Ventilatory parameters associated with mortality in ARDS patients were not improved by inducing delivery. Fetal and neonatal mortality was associated independently with gestational age at delivery and SOFA₂₄ score. Given what we know now, the clinical practice of inducing delivery in pregnant patients with pneumonia to improve maternal conditions should be abandoned. Gestational age and SOFA₂₄ score should be considered in the decision-making process, thereby improving both maternal and fetal outcomes, as well as considering the psychological impact of losing a child.

Funding/Support

The authors have reported to CHEST that no funding was received for this study.

Financial/Nonfinancial Disclosures

None declared.