Abstract
Objectives:
The use of extracorporeal membrane oxygenation (ECMO) therapy has increased in the adult population. Studies from the H1N1 influenza pandemic suggest that ECMO deployment in pregnancy is associated with favorable outcomes. With increasing numbers of pregnant women affected by COVID-19 and potentially requiring this life-saving therapy, we sought to compare comorbidities, costs, and outcomes between pregnancy and non-pregnancy associated ECMO therapy among reproductive-aged female patients.
Study Design:
We used the 2013-2019 National Readmissions Database. Diagnosis and procedural coding were used to identify ECMO deployment, potential indications, comorbid conditions, and pregnancy outcomes. The primary outcome was in-hospital mortality during the patient’s initial ECMO stay. Secondary outcomes included length of stay and hospital charges/costs, occurrence of thromboembolic or bleeding complications during ECMO hospitalization, and mortality and readmissions up to 330 days following ECMO stay. Univariate and multivariate regression models were used to model the associations between pregnancy status and outcomes.
Results:
The sample included 324 pregnancy-associated hospitalizations and 3,805 non-pregnancy associated hospitalizations, corresponding to national estimates of 665 and 7,653 over the study period, respectively. Pregnancy-associated ECMO had lower incidence of in-hospital death (adjusted odds ratio (AOR): 0.56, 95% confidence interval (CI): 0.41-0.75) and bleeding complications (AOR: 0.67, 95% CI: 0.49-0.93). Length of stay was significantly shorter (adjusted rate ratio (ARR): 0.86, 95% CI: 0.77-0.96) and total hospital costs were less (ARR: 0.83, 95% CI: 0.75-0.93). Differences in the incidence of thromboembolic events (AOR: 1.04, 95% CI: 0.78-1.38) were not statistically significant.
Conclusion:
Pregnancy-associated ECMO therapy had lower incidence of in-hospital death, bleeding complications, total inpatient cost, and length of stay when compared to non-pregnancy associated ECMO therapy without increased thromboembolic complications. Pregnancy-associated ECMO therapy should be offered to eligible patients.
Keywords: Extracorporeal Membrane Oxygenation, Pregnancy, Respiratory Insufficiency, Cardiac failure
INTRODUCTION
Extracorporeal Membrane Oxygenation (ECMO) is a system of prolonged cardiopulmonary support that improves survival in patients with cardiogenic shock or respiratory failure refractory to pharmacologic therapy and mechanical ventilation. Though evidence to date from randomized control trials has been mixed,1, 2 there is consensus in the critical care community that ECMO therapy provides a survival benefit compared to standard care in appropriately-selected patients, likely by allowing time for organ recovery or acting as a bridge to transplantation.
Pregnancy is a unique physiologic state with altered maternal cardiopulmonary physiology and increased thrombotic and bleeding risks. Utilization of ECMO in pregnancy has markedly increased since the H1N1 influenza pandemic of 2009.3 Whether or not physiologic differences in pregnancy affect ECMO outcomes is poorly understood. One meta-analysis notes a 75% survival rate in obstetrical patients who received ECMO for H1N1-related acute respiratory distress syndrome (ARDS).4 Emerging data from patients with COVID-19 infection is similarly reassuring.5 Knowledge of short- and long-term outcomes of ECMO deployed in obstetric practice, however, remains limited. Using a large, nationally-representative U.S. dataset, our objectives were to describe outcomes of ECMO use during and following pregnancy and to compare outcomes of pregnancy-associated ECMO deployment to ECMO deployment in non-pregnant reproductive aged female patients at a population level.
METHODS
Our analysis used data from the Nationwide Readmissions Database (NRD), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality from 2013-2019 inclusive.6 The NRD is an all-payor administrative database comprising discharge and readmission data from acute inpatient stays at non-federal hospitals throughout the United States. Using weighting and stratification by region and hospital type, the database is designed to produce nationally representative estimates of hospitalizations and readmissions. Patients can be followed across hospitalizations within a single state and calendar year (January to December period). Data available include demographics (including age, sex, and the quartile of income in the patient’s home ZIP code), hospital characteristics (size, academic versus community), diagnosis-related grouping (DRG), International Classification of Disease (ICD) diagnosis and procedure codes, discharge disposition, and length of stay and inpatient charges. Cost-to-charge ratio files are provided to convert inpatient charges to estimated cost.
In 2013, twenty-one U.S. states contributed to the database. In 2019, this had risen to thirty U.S. states, representing 61.8% of the national population and comprising 60.4% of all hospitalizations. Diagnosis and procedural coding data were used to identify and define the study population. In October 2015, the United States adopted use of the 10th edition of the ICD for diagnosis and procedural coding. For observations with discharge dates before October 2015, 9th Edition coding was used and 10th Edition coding was used for subsequent discharges. Receipt of ECMO was identified by ICD-9 procedure code 39.65 and ICD-10 procedure code 5A15223 (for discharges before October 1, 2018) and 5A1522F, 5A1522G, and 5A1522H (for discharges on or after October 2018). The population was then limited to female patients who were aged 12 through 55 to create a reproductive-aged female cohort. Each patient’s first ECMO-associated hospitalization was identified and treated as the index hospitalization for the purposes of all subsequent analyses. Next, the ICD diagnosis and procedure codes on the index hospitalization record were used to identify pregnancy-associated hospitalizations, based on criteria from HCUP. Pregnancy hospitalizations were further characterized as being antepartum, delivery, or postpartum hospitalizations, using previously-described criteria which we updated for use of ICD-10-CM codes.7 8 The original publication limited application of postpartum coding criteria to only the principal or second diagnosis coded on the claim. We relaxed these criteria to reflect that ECMO-associated hospitalization and critical illness may have altered sequencing of coding. Patients who had an early, abnormal pregnancy (e.g., early pregnancy loss or ectopic pregnancy) were excluded. The presence of comorbid conditions which might confound the association between pregnancy status and outcome were identified by adaptation of a validated obstetric-specific comorbidity index to exclude obstetric-specific comorbid conditions (e.g., preeclampsia), and by adjustment of criteria developed for an international version of ICD-10 for the United States implementation.9, 10 Diagnosis and procedural coding were similarly used to identify potential indications for ECMO deployment (such as acute respiratory distress syndrome, heart failure, pulmonary hypertension, and amniotic fluid embolism). Outcomes analyzed included in-hospital mortality from index hospital stay, length of stay, total estimated hospital charges and costs and an occurrence of thromboembolic or bleeding complication. A complete list of all ICD and DRG-based coding criteria is included as a Supplementary Table (Supplementary Table 1).
Unadjusted comparisons were made using weighted regression models to accommodate the stratified and weighted nature of the NRD data. Covariate-adjusted comparisons of outcomes based on pregnancy status were made using weighted logistic, Poisson, and gamma-log link regression models as appropriate, adjusting for demographic, hospital, potential indications for ECMO, and comorbid conditions. We selected covariates a priori based on potential clinical confounding relationship with the outcomes of interest. These variables included age, ZIP code income quartile, primary payor, facility size and teaching/rural status, potential reasons for ECMO, and comorbid conditions.
We evaluated longitudinal outcomes via two approaches. First, all outcomes (including readmission) occurring within 30- or 60-days following discharge from the initial ECMO-related stay were analyzed. Patients discharged in December were excluded for 30-day outcomes, as the NRD only captures readmission dates at the month level of detail. Similarly, discharges in November and December were excluded for 60-day outcomes. Second, among patients who survived their initial ECMO hospitalization stay, we analyzed both inpatient death and all-cause readmission as time-to-event variables and portrayed these outcomes using Kaplan Meier curves. Weighted Cox proportional hazard regressions were used for these outcomes. We assumed conservatively that patients were discharged on the last day of the month when calculating potential follow-up time. Thus, the longest follow-up possible in this analysis was 11 months (or approximately 330 days) for patients who were discharged from their ECMO encounter in January. There were few missing data (<1% for expected payor and ZIP code income); imputation using modal values was thus used given the complexity involved in implementation of multiple imputation with weighted, stratified data. A two-sided alpha level of 0.05 was pre-specified as statistically significant. Dataset construction was performed in SAS System, Version 9.4 (SAS Systems, Cary, NC) and statistical analysis in Stata Statistical Software, Version 17.0 (StataCorp, College Station, Texas). Because the NRD is a Limited Data Set as defined by the United States Health Insurance Portability and Accountability Act, this study was deemed to be exempt from review by the Duke University Health System Institutional Review Board.
RESULTS
There were 4,129 female patients aged 12 to 55 with ECMO deployment recorded in the NRD, of which 324 were pregnancy associated and 3,805 were non-pregnancy associated (Figure 1). This sample corresponds to a national estimate of 8,317 hospitalizations, of which 665 (7.9%) were pregnancy-associated; the remainder of statistics presented in this analysis reflect these weighted national estimates. Patients hospitalized with pregnancy-associated ECMO hospitalizations were younger, lived in ZIP codes with lower median income, and had a higher likelihood of Medicaid insurance coverage than non-pregnancy associated ECMO hospitalizations (Table 1). The percentage of patients with diagnosis codes consistent with influenza, aspiration, transfusion-related acute lung injury (TRALI), and ARDS were all similar between pregnancy and non-pregnancy-associated ECMO hospitalizations. Patients with pregnancy-associated ECMO deployment were more likely to have diagnosis codes reflecting acute heart failure and less likely to have diagnosis codes reflecting pneumonia, and pulmonary hypertension; amniotic fluid embolism codes were, as expected, only found among pregnancy-associated patients. With respect to comorbid conditions, rates of alcohol abuse, asthma, chronic kidney disease, ischemic heart disease, cystic fibrosis, diabetes, lupus, and chronic hypertension were lower in the pregnancy-associated hospitalizations.
Figure 1:
Derivation of Study Population
Table 1:
Baseline Demographic and Clinical Characteristics
| Overall (N=4,129) (Weighted N = 8,317) |
Not Pregnancy Associated (N=3,805) (Weighted N = 7,653) |
Pregnancy Associated (N=324) (Weighted N = 665) |
p | |
|---|---|---|---|---|
| Weighted Mean (Standard Deviation) or N (%) |
||||
| Patient Demographics | ||||
| Patient age (years) | 36.9 (5.5) | 37.4 (5.5) | 30.8 (3.3) | <0.001 |
| ZIP Code Median | 0.002 | |||
| Household Income | ||||
| Quartile 1 (Lowest) | 2,513 (30.5) | 2,246 (29.6) | 267 (40.6) | |
| Quartile 2 | 2,101 (25.5) | 1,972 (26.0) | 128 (19.5) | |
| Quartile 3 | 1,991 (24.2) | 1,850 (24.4) | 141 (21.5) | |
| Quartile 4 (Highest) | 1,634 (19.8) | 1,512 (19.9) | 122 (18.5) | |
| Primary Payer | <0.001 | |||
| Medicare | 1,134 (13.6) | 1,094 (14.3) | 40 (6.0) | |
| Medicaid | 2,753 (33.1) | 2,411 (31.5) | 343 (51.5) | |
| Private insurance | 3,755 (45.2) | 3,492 (45.7) | 263 (39.6) | |
| Self-pay | 350 (4.2) | 344 (4.5) | * | |
| Other | 33 (0.4) | 30 (0.4) | * | |
| Hospital Characteristics | 284 (3.4) | 274 (3.6) | * | |
| Hospital Bed size | 0.02 | |||
| Small | 228 (2.7) | 222 (2.9) | * | |
| Medium | 830 (10.0) | 788 (10.3) | 42 (6.3) | |
| Large | 7,260 (87.3) | 6,643 (86.8) | 617 (92.8) | |
| Hospital location and teaching status | 0.70 | |||
| Metropolitan non-teaching | 304 (3.7) | 278 (3.6) | 26 (3.9) | |
| Metropolitan teaching | 7,996 (96.1) | 7,357 (96.1) | 639 (96.1) | |
| Non-metropolitan hospital | 17 (0.2) | 17 (0.2) | * | |
| Potential Indications for ECMO | ||||
| Adult respiratory distress syndrome | 7,166 (86.2) | 6,610 (86.4) | 555 (83.5) | 0.18 |
| Amniotic fluid embolism | 69 (0.8) | 0 (0.0) | 69 (10.4) | <0.001 |
| Acute heart failure | 3,883 (46.7) | 3,518 (46.0) | 365 (54.9) | 0.003 |
| Pulmonary Hypertension | 1,354 (16.3) | 1,273 (16.6) | 81 (12.1) | 0.03 |
| Influenza | 782 (9.4) | 733 (9.6) | 49 (7.4) | 0.25 |
| Pneumonia (excluding influenza) | 3,301 (39.7) | 3,088 (40.4) | 213 (32.1) | 0.005 |
| Aspiration | 1,106 (13.3) | 1,027 (13.4) | 79 (11.9) | 0.44 |
| TRALI | 87 (1.0) | 75 (1.0) | 12 (1.9) | 0.17 |
| Categorical | 0.19 | |||
| ARDS only | 499 (6.0) | 461 (6.0) | 38 (5.7) | |
| Acute heart failure and/or pulmonary hypertension only | 3,291 (39.6) | 3,054 (39.9) | 237 (35.6) | |
| ARDS and acute heart failure/pulmonary hypertension | 652 (7.8) | 581 (7.6) | 71 (10.7) | |
| Neither ARDS nor acute heart failure/pulmonary hypertension | 3,875 (46.6) | 3,556 (46.5) | 319 (47.9) | |
| Comorbid Conditions | ||||
| Alcohol abuse | 290 (3.5) | 290 (3.8) | * | <0.001 |
| Asthma | 1,541 (18.5) | 1,451 (19.0) | 90 (13.6) | 0.03 |
| Cardiac valve disease | 1,215 (14.6) | 1,139 (14.9) | 76 (11.5) | 0.12 |
| Congestive Heart failure | 1,385 (16.7) | 1,269 (16.6) | 116 (17.5) | 0.71 |
| Ischemic Heart disease | 1,046 (12.6) | 1,018 (13.3) | 28 (4.2) | <0.001 |
| Chronic kidney disease | 1,050 (12.6) | 996 (13.0) | 54 (8.1) | 0.02 |
| Congenital heart disease | 570 (6.9) | 534 (7.0) | 37 (5.5) | 0.38 |
| Cystic fibrosis | 279 (3.4) | 272 (3.6) | * | 0.02 |
| Pre-gestational diabetes | 1,528 (18.4) | 1,496 (19.5) | 33 (4.9) | <0.001 |
| Drug abuse | 780 (9.4) | 731 (9.5) | 49 (7.4) | 0.19 |
| HIV Disease | 60 (0.7) | 60 (0.8) | * | 0.11 |
| Chronic hypertension | 2,979 (35.8) | 2,855 (37.3) | 124 (18.6) | <0.001 |
| Obesity | 2,175 (26.2) | 2,018 (26.4) | 157 (23.7) | 0.37 |
| Sickle Cell Disease | 68 (0.8) | 63 (0.8) | * | 0.92 |
| Systemic lupus erythematosus | 234 (2.8) | 229 (3.0) | * | 0.02 |
| Tobacco Use Disorder | 1,373 (16.5) | 1,286 (16.8) | 87 (13.1) | 0.15 |
P-values by weighted linear regression for continuous variables and weighted chi2 test for binary/categorical variables
Missing values in Zip Income (20 observations), and primary payer (7 observations)
N ≤ 10; value suppressed by data vendor limitations
ECMO: Extracorporeal membrane oxygenation; TRALI: Transfusion related acute lung injury; ARDS: Acute respiratory distress syndrome; HIV: Human immunodeficiency virus
Outcomes of the initial ECMO-related hospitalization generally favored pregnancy-associated hospital stays (Table 2). Specifically, pregnancy-associated hospitalizations were associated with lower probability of in-hospital mortality, shorter lengths of stay, and lower inpatient charges and costs. The probability of a thromboembolic event did not differ based on pregnancy status, while pregnancy-associated hospitalizations were associated with lower probability of a bleeding complication.
Table 2:
Clinical and Economic Outcomes for Initial ECMO-Related Hospitalization
| Overall (N=4,129) (Weighted N = 8,317) |
Not Pregnancy - associated (N=3,805) (Weighted N = 7,653) |
Pregnancy- associated (N=324) (Weighted N = 665) |
p | Unadjusted Odds Ratio or Rate Ratio (95% Confidence Interval) |
Adjusted* Odds Ratio or Rate Ratio (95% Confidence Interval) |
Adjusted Incremental Difference** (95% Confidence Interval) |
|
|---|---|---|---|---|---|---|---|
| Weighted Mean (Standard Deviation) or N (%) |
|||||||
| Died in Hospital | 3,244 (39.0) | 3,052 (39.9) | 192 (28.9) | <0.001 | 0.61 (0.46, 0.81) | 0.56 (0.41, 0.75) | −12.25 (−17.99, −6.50) |
| Length of stay | 31.3 (16.3) | 31.7 (16.6) | 27.1 (11.8) | 0.003 | 0.86 (0.77, 0.95) | 0.86 (0.77, 0.96) | −4.57 (−7.61, −1.52) |
| Total inpatient charges ($1000s) | 856.8 (442.1) | 866.8 (442.9) | 741.3 (429.0) | 0.01 | 0.86 (0.75, 0.98) | 0.81 (0.73, 0.91) | −162.05 (−245.81, −78.29) |
| Total inpatient cost ($1000s) | 219.0 (101.4) | 221.3 (102.3) | 192.9 (89.5) | 0.02 | 0.87 (0.77, 0.98) | 0.83 (0.75, 0.93) | −36.80 (−56.63, −16.98) |
| Thromboembolic event | 2,825 (34.0) | 2,605 (34.0) | 220 (33.0) | 0.75 | 0.96 (0.73, 1.26) | 1.04 (0.78, 1.38) | 0.80 (−4.92, 6.52) |
| Bleeding Complications | 2,109 (25.4) | 1,977 (25.8) | 132 (19.9) | 0.03 | 0.71 (0.52, 0.97) | 0.67 (0.49, 0.93) | −6.64 (−11.62, −1.66) |
P-values by weighted linear regression for continuous variables and weighted chi2 test for binary/categorical variables
Missing values in Length of stay (7 observations), Total charges (33 observations), and Total cost (33 observations)
Outcomes adjusted for patient age, ZIP code income quartile, primary payor, facility size and teaching/rural status, potential reasons for ECMO, and comorbid conditions. All comparisons are made comparing pregnancy-associated hospital stays to non-pregnancy associated hospital stays.
Results of adjusted analysis, reformatted to reflect average percentage point difference in probability of event (for death, thromboembolic event, and bleeding complications) and average difference in length of stay and inpatient costs and charges.
In covariate-adjusted analyses (Table 3), these differences persisted. Pregnancy-associated hospitalizations were associated with lower adjusted odds of in-hospital mortality, lower inpatient charges and costs, and lower probability of a bleeding complication with no difference in probability of a thromboembolic event. These trends were consistent when including events from readmissions within 30 and 60 days of initial discharge (Supplementary Table 2). Given the age difference between pregnancy and non-pregnancy associated hospitalization we also compared the in-hospital mortality rate among only those patients age 18-40 (inclusive) and results were also consistent (adjusted odds ratio 0.60, 95% confidence interval 0.45 to 0.80). Among patients who survived their initial ECMO-related hospitalization, there were no differences noted in survival or hospital readmissions up to 11 months post initial hospitalization (Figure 2). This finding was consistent in both unadjusted and covariate-adjusted analyses (adjusted hazard ratio for in-hospital mortality 1.00 with 95% confidence interval from 0.4 to 2.50; for readmission the adjusted hazard ratio was 1.07 with a 95% confidence interval from 0.80 to 1.41).
Table 3:
Outcomes of index ECMO hospitalizations which included a delivery
| Overall (N=163) (Weighted N = 332) |
|
|---|---|
| Delivery type | |
| Spontaneous vaginal | 35 (10.8) |
| Operative vaginal | 16 (4.9) |
| Cesarean | 272 (84.3) |
| Stillbirth | 43 (13.0) |
| Preterm Birth | 143 (43.1) |
| Severe Maternal Morbidity | |
| Acute myocardial infarction | 11 (3.2) |
| Aneurysm | * |
| Acute renal failure | 152 (45.7) |
| Adult respiratory distress syndrome | 262 (79.0) |
| Amniotic fluid embolism | 58 (17.6) |
| Cardiac arrest/ventricular fibrillation | 100 (30.2) |
| Conversion of cardiac rhythm | 95 (28.7) |
| Disseminated intravascular coagulation | 112 (33.6) |
| Eclampsia | * |
| Heart failure/arrest during surgery or procedure | 11 (3.5) |
| Puerperal cerebrovascular disorders | 38 (11.4) |
| Pulmonary edema / Acute heart failure | 95 (28.5) |
| Severe anesthesia complications | * |
| Sepsis | 119 (36.0) |
| Shock | 247 (74.5) |
| Sickle cell disease with crisis | * |
| Air and thrombotic embolism | 56 (16.8) |
| Blood products transfusion | 81 (24.3) |
| Hysterectomy | 36 (10.8) |
| Temporary tracheostomy | 55 (16.5) |
| Ventilation | 188 (56.5) |
| Discharge destination | |
| Routine | 88 (26.5) |
| Transfer to short-term hospital | 21 (6.4) |
| Other transfer | 40 (11.9) |
| Home health care | 80 (24.1) |
| Died | * |
| Length of stay (cleaned) | 23.4 (10.8) |
| Total Costs ($ Thousands) | 156.4 (60.6) |
P-values by weighted linear regression for continuous variables and weighted chi2 test for binary/categorical variables
N ≤ 10; value suppressed by data vendor limitations
Missing values in Delivery Type (5 observation) Total Cost (2 observations)
Figure 2:
Post-discharge outcomes of Extracorporeal membrane oxygenation-related hospitalizations, including inpatient death (Panel A) and readmissions (Panel B)
Among pregnancy-associated ECMO hospitalizations, most (50%) involved a delivery procedure, while 40.1% were postpartum hospitalizations and 9.5% were antepartum hospitalizations (Figure 3). Most patients who had a delivery during their ECMO-related hospitalization were delivered by cesarean section (84.3%) (Table 3). Rates of severe maternal morbidity were high, including rates of ARDS and shock of 79.0% and 74.5%, respectively. The preterm birth rate was 43.1%, and the stillbirth rate was 14.0%. In-hospital mortality for these admissions was 30.6%, and the average length of stay was nearly 11 days.
Figure 3:
Pregnancy Outcomes
DISCUSSION
Principal findings:
In this national study of ECMO deployment among reproductive-aged female patients, we found that pregnancy-associated ECMO deployment was associated with lower rates of mortality and bleeding complications during the initial ECMO hospitalization, lower average length of stay, lower inpatient costs and similar rates of thromboembolism when compared with their non-pregnant counterparts. These findings remained consistent in both unadjusted and covariate-adjusted analysis. Among patients who survived their initial ECMO course, rates of all readmissions and readmissions ending in mortality were similar when comparing pregnancy-associated ECMO use to non-pregnancy-associated ECMO use.
Results in the Context of What is Known:
Our findings strengthen those previously reported by single centers. High 30-day survival rates for pregnancy-associated ECMO have been reported previously −75% and 77.8% respectively in two series.11, 12 Our study also reports a non-pregnancy associated ECMO mortality of 39.9%; similar to the 2016 ELSO International Report mortality of 42%.13 The only other nationally-representative study of ECMO in pregnancy – using the U.S. National Inpatient Sample (NIS) – reported similarly increased survival in pregnancy-associated ECMO, although interestingly demonstrated an increase in thromboembolic events in pregnancy patients not reflected in non-pregnancy counterparts.14 This difference in our findings may reflect improvements in anticoagulation regimens for critically ill patients in our more contemporary sample (2013-19 for this analysis vs. 1994-2014 for the NIS study).
Clinical Implications:
Despite the hypothesis that the physiologic changes of pregnancy may increase the risk of ECMO deployment, these results suggest that pregnancy confers no excess risk to outcomes from ECMO deployment with respect to non-pregnant patients. Currently, use of ECMO in pregnancy achieves at least equivalent if not superior outcomes when compared to ECMO deployment in non-pregnant female patients. In settings in which the supply of ECMO is limited and care must be rationed – such as a global pandemic – these data strengthen the assertion that current or recent pregnancy should not be considered a de-prioritizing factor for ECMO deployment; indeed, these findings should encourage prioritization of pregnant and postpartum patients for deployment of ECMO.
Research Implications:
Where this study is limited to diagnosis and procedure based clinical information, future research may include individual clinical data collection for more thorough study of maternal and neonatal outcomes. It is highly likely forthcoming data from the period 2020 to present (including cases associated with the COVID-19 pandemic) will demonstrate a marked increase in use of ECMO in pregnancy, further elucidating maternal outcomes.15 In addition, several centers are actively linking maternal and neonatal records for data collection, which should allow for improved study of the relationship of maternal ECMO to neonatal outcomes 16. Continued submission of individual clinical data to the Extracorporeal Life Support Organization data registry and multicenter retrospective and prospective data-sharing will be necessary to determine if indications or thresholds for ECMO should differ in pregnancy.
Strengths and Limitations:
The strengths of our study include a large nationally representative database designed to evaluate hospital admission. This allows us to analyze the relatively rare maternal need for ECMO therapy. We were able to capture ECMO based upon both ICD-9 and ICD-10 procedure codes (through the classification transition) leading to one of the larger analysis performed on a pregnancy-associated population. This is also one of the first studies to compare the economic impact of ECMO in pregnancy associated and non-pregnancy associated cohorts. Finally, through the use of the NRD, we are able to assess longer-term outcomes than only the ECMO deployment hospitalization.
The primary and overarching limitation of this study relates to the limited clinical data available in administrative datasets such as the NRD. Unlike with detailed patient chart review or in multicenter clinical data registries, the only available clinical data in the NRD is that provided in diagnosis and procedures codes. A number of desirable data elements (e.g., certainty regarding criteria for ECMO deployment, ECMO settings, laboratory tests) cannot be obtained from the NRD. Additionally, we are limited by the lack of temporal data within the inpatient hospitalization (we cannot reliably determine onset of clinical conditions during the hospital stay). Finally, we acknowledge and affirm that not all pregnant patients identify as women. Based on how pregnancy-related diagnosis codes are restricted to patient's reported sex in administrative databases such as the NRD, this study is limited to a female population.
Conclusions:
In conclusion, this study gives novel economic and clinical data by comparing pregnancy-associated and non-pregnancy associated hospitalizations. Despite the unique physiology of pregnancy, pregnant and postpartum patients appear to have equivalent or superior outcomes from ECMO deployment in cardiopulmonary failure when compared with non-pregnant patients, even when matched for demographics and comorbidities. These results strongly suggest that ECMO should continue to be offered in appropriately selected patients in pregnancy and the postpartum. Further research is needed to identify if differing clinical criteria may be beneficial for ECMO deployment in pregnancy and the postpartum.
Supplementary Material
KEY POINTS.
Pregnancy-related ECMO use was compared to non-pregnant use.
Outcomes were equal or favored pregnancy-related deployment.
These data may be useful when considering ECMO use in pregnancy
ACKNOWLEDGEMENTS
The authors appreciate the HCUP Data Partners who contribute data to the NRD. A complete list of partners can be found at (www.hcup-us.ahrq.gov/hcupdatapartners.jsp). Work contained in this manuscript were made possible by the following grants from the National Institutes of Health (TL1-TR002555 [JJF]). Data acquisition was also supported by funding from the Foundation for Women and Girls with Blood Disorders to JJF.
Dr. Federspiel is supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UL1TR002553. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Data acquisition for this project was supported by grant funding for the Foundation for Women and Girls with Blood Disorders.
Footnotes
The authors report no conflicts of interest.
This paper was presented in preliminary form at the 2022 Society for Maternal Fetal Medicine meeting (held virtually January 31-February 5, 2022).
REFERENCES
- 1.Peek GJ, Mugford M, Tiruvoipati R, et al. Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial. Lancet 2009;374:1351–63. [DOI] [PubMed] [Google Scholar]
- 2.Combes A, Hajage D, Capellier G, et al. Extracorporeal Membrane Oxygenation for Severe Acute Respiratory Distress Syndrome. N Engl J Med 2018;378:1965–75. [DOI] [PubMed] [Google Scholar]
- 3.Mosier JM, Kelsey M, Raz Y, et al. Extracorporeal membrane oxygenation (ECMO) for critically ill adults in the emergency department: history, current applications, and future directions. Crit Care 2015;19:431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Saad AF, Rahman M, Maybauer DM, et al. Extracorporeal Membrane Oxygenation in Pregnant and Postpartum Women With H1N1-Related Acute Respiratory Distress Syndrome: A Systematic Review and Meta-analysis. Obstet Gynecol 2016;127:241–7. [DOI] [PubMed] [Google Scholar]
- 5.O'neil ER, Lin H, Shamshirsaz AA, et al. Pregnant and Peripartum Women with COVID-19 Have High Survival with Extracorporeal Membrane Oxygenation: An Extracorporeal Life Support Organization Registry Analysis. Am J Respir Crit Care Med 2022;205:248–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nationwide Readmissions Database. In: Healthcare Cost and Utilization Project (HCUP), ed.: Agency for Healthcare Research and Quality,. [PubMed] [Google Scholar]
- 7.Callaghan WM, Creanga AA, Kuklina EV. Severe maternal morbidity among delivery and postpartum hospitalizations in the United States. Obstet Gynecol 2012;120:1029–36. [DOI] [PubMed] [Google Scholar]
- 8.Kuklina EV, Whiteman MK, Hillis SD, et al. An enhanced method for identifying obstetric deliveries: implications for estimating maternal morbidity. Matern Child Health J 2008;12:469–77. [DOI] [PubMed] [Google Scholar]
- 9.Bateman BT, Mhyre JM, Hernandez-Diaz S, et al. Development of a comorbidity index for use in obstetric patients. Obstet Gynecol 2013;122:957–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Bliddal M, Moller S, Vinter CA, Rubin KH, Gagne JJ, Pottegard A. Validation of a comorbidity index for use in obstetric patients: A nationwide cohort study. Acta Obstet Gynecol Scand 2020;99:399–405. [DOI] [PubMed] [Google Scholar]
- 11.Naoum EE, Chalupka A, Haft J, et al. Extracorporeal Life Support in Pregnancy: A Systematic Review. J Am Heart Assoc 2020;9:e016072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Moore SA, Dietl CA, Coleman DM. Extracorporeal life support during pregnancy. J Thorac Cardiovasc Surg 2016;151:1154–60. [DOI] [PubMed] [Google Scholar]
- 13.Thiagarajan RR, Barbaro RP, Rycus PT, et al. Extracorporeal Life Support Organization Registry International Report 2016. ASAIO J 2017;63:60–67. [DOI] [PubMed] [Google Scholar]
- 14.Taha B, Guglielminotti J, Li G, Landau R. Utilization and Outcomes of Extracorporeal Membrane Oxygenation in Obstetric Patients in the United States, 1999-2014: A Retrospective Cross-Sectional Study. Anesth Analg 2021. [DOI] [PubMed] [Google Scholar]
- 15.Clemenza S, Zullino S, Vacca C, et al. Perinatal outcomes of pregnant women with severe COVID-19 requiring extracorporeal membrane oxygenation (ECMO): a case series and literature review. Arch Gynecol Obstet 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hurst JH, Liu Y, Maxson PJ, Permar SR, Boulware LE, Goldstein BA. Development of an electronic health records datamart to support clinical and population health research. J Clin Transl Sci 2020;5:e13. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.