If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Characteristics differed by race among patients treated with extracorporeal support.
•
Mortality for White patients decreased significantly.
•
Mortality for Black and Asian patients remained stable.
•
Non-White race was linked to increased complications and resource utilization.
•
Rates of non-home discharge were comparable across race.
Abstract
Background
Racial disparities in extracorporeal membrane oxygenation (ECMO) outcomes in patients with a broad set of indications are not well documented.
Methods
Adults requiring ECMO were identified in the 2016–2019 National Inpatient Sample. Patient and hospital characteristics, including mortality, clinical outcomes, and resource utilization were analyzed using multivariable regressions.
Results
Of 43,190 adult ECMO patients, 67.8% were classified as White, 18.1% Black, and 10.4% Hispanic. Although mortality for Whites declined from 47.5 to 41.0% (P = 0.002), it remained steady for others. Compared to White, Asian/Pacific Islander (PI) race was linked to increased odds of mortalty (AOR = 1.4, 95% CI = 1.1–2.0). Black race was associated with increased odds of acute kidney injury (AOR = 1.4, 95%-CI: 1.2–1.7), while Hispanic race was linked to neurologic complications (AOR 21.6; 95% CI 1.2–2.3). Black and Hispanic race were also associated with increased incremental costs.
Conclusions
Race-based disparities in ECMO outcomes persist in the United States. Further work should aim to understand and mitigate the underlying reasons for such findings.
With a growing body of literature supporting its efficacy, extracorporeal membrane oxygenation (ECMO) has seen a dramatic rise in utilization over the past decade.
Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial.
Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc bayesian analysis of a randomized clinical trial.
Despite increased adoption and advances in cannula and circuit technology, ECMO remains a resource intensive therapy that is associated with significant mortality, complications, and cost.
To investigate economic constraints, several groups have previously examined the association of disadvantaged socioeconomic status and safety net hospitals on outcomes of extracorporeal life support in limited cohorts and indications. In general, socioeconomic disadvantage at patient and hospital level alike, have been linked to inferior clinical and financial outcomes for ECMO.
Worse outcomes including increased mortality and length of stay (LOS) have similarly been noted in patients of color requiring ECMO after congenital cardiac surgery
Nonetheless, examination of race-based outcomes of ECMO across various indications remains generally lacking.
The present study investigated the presence of racial disparities in outcomes of ECMO in a large national cohort. We hypothesized non-White race to be associated with increased mortality, complications, and length of stay despite risk adjustment for other variable factors.
2. Methods
The present study was a cross-sectional analysis of the 2016–2019 National Inpatient Sample (NIS). The NIS is the largest, publicly available, all-payer inpatient database in the United States and provides accurate estimates for 97% of all US hospitalizations through application of robust survey weights.
All adults (≥18 years) hospitalizations for ECMO were identified using previously reported International Classification of Diseases 10th revision procedure codes.
Patients receiving preoperative ECMO or in the setting of heart/lung transplantation were excluded (12.5%). Records missing key data such as age, sex, race/ethnicity and mortality were further excluded (12.1%). Patients who self-reported their race/ethnicity as “other” were not included due to lack of granularity. Based on previously defined methodology,
patients were categorized into the following groups for indication of ECMO: postcardiotomy, cardiogenic shock, respiratory failure, and mixed cardiopulmonary failure.
Patient and hospital characteristics of interest were defined according to the NIS data dictionary and included age, sex, self-reported race/ethnicity, income quartile, insurance status and hospital teaching status were considered. Non-White race or ethnicity was defined as any race/ethnicity other than White and encompassed Black, Hispanic, Asian or Pacific Islander (PI). The Elixhauser Comorbidity Index, a validated composite of 30 comorbidities, was used to quantify the burden of chronic conditions. Individual comorbidities were tabulated using ICD-10 diagnosis codes. Complications included thromboembolic (disseminated intravascular coagulation, deep venous thrombosis and pulmonary embolism), neurologic (stroke, intracranial hemorrhage, seizure and other central nervous system complication), new acute kidney injury (AKI), and infectious (sepsis, septicemia, bacteremia, and central line bloodstream infection). Hospitalization costs were calculated by application of hospital specific cost-to-charge ratios to overall charges and inflation adjustment to the 2019 Personal Healthcare Index. To account for hospital and provider experience, hospitals were categorized into volume tertiles based on total number of ECMO patients per year.
The primary outcome of interest was in-hospital mortality while complications, length of stay (LOS), hospitalization costs, and non-home discharge, were secondarily evaluated. Categorical and continuous variables are reported as proportions and means with standard deviations (SD) or medians with interquartile range (IQR) if non-normally distributed. The Pearson's chi-squared test and adjusted Wald test were used to compare categorical and continuous variables, respectively. We assessed the significance of temporal trends using Cuzick's nonparametric trend test (nptrend).
Multivariable models were fit to evaluate the association of race with outcomes of interest. We used the least absolute shrinkage and selection operator (LASSO) to select model covariates. Briefly, LASSO is a penalized regularization technique that minimizes overfitting
and thus, enhances generalizability. The Akaike and Bayesian Information Criteria as well as receiver operating characteristics (C-statistic) were used to optimize models. Regression outputs are reported as adjusted odds ratios (AOR) with and β coefficients with 95% confidence intervals (CI), as appropriate. The final prediction model for mortality included the following variables: age, race/ethnicity, sex, Elixhauser Comorbidity Index, income quartile, insurance status, ECMO indication, year, congestive heart failure, coronary artery disease, hypertension, cardiac arrythmias, drug use disorder, tobacco use, peripheral vascular disease, renal failure, diabetes, smoker, liver disease, pulmonary circulation disorder, cancer, chronic lung disease, hospital bed size, hospital geographic region, and hospital volume tertile. All statistical analyses were performed using Stata version 16.1 with α set at 0.05 (StataCorp, College Station, TX). The University of California, Los Angeles Institutional Review Board deemed this study exempt from full review (#17–00112, 7/26/2017).
3. Results
Among an estimated 29,410 ECMO patients who met study criteria, 67.8% were classified as White, 18.1% Black, 10.4% Hispanic, and 3.7% as Asian/PI (Fig. 1). Compared to other groups, Black patients had the lowest median age (49 years; IQR = 36–60) and were most commonly female (42.6%). While the distribution of Elixhauser Comorbidity Index was comparable across groups, rates of coronary artery disease (25.6%) and congestive heart failure (53.5%) were highest among Asian/PI patients. Relative to their counterparts, Black patients were more frequently dialysis dependent and hypertensive, while Hispanic patients were more likely to be diagnosed with diabetes (Table 1). Asian/PI patients were more likely to have coronary artery disease but less frequently smoked. Notably, Black patients least frequently required ECMO for postcardiotomy syndrome (15.5%) while Hispanics most commonly were placed on support for respiratory failure (51.2%). Patients classified as Asian/PI were more likely to be privately insured, while Black and Hispanic patients most frequently had Medicaid coverage.
Fig. 1Consort diagram of patient selection criteria.
Of all groups, Asian/PI patients faced the highest rate of in-hospital mortality (53.0%). Notably, mortality for Whites decreased significantly from 47.5% in 2016 to 41.0% in 2019 (nptrend = 0.002). In contrast, the mortality rate of Black (nptrend = 0.07) and Asian (nptrend = 0.7) patients remained stable over the study period.
In addition, utilization of tracheostomy and blood transfusions were lowest among Whites and highest among Hispanics and Asian/PIs (Table 2). Patients of Hispanic and Asian/PI race had significantly increased rates of neurologic events (Hispanic 14.1%; Asian/PI 10.2%). Black patients most frequently experienced new-onset AKI (35.1%), while Asians/PI more frequently developed infectious complications (41.8%; Table 2). Black, Hispanic, and Asian races had higher median hospitalization costs and longer stays compared to White race (Table 3). Nonetheless, rates of non-home discharge were comparable across groups (Table 2). Unadjusted rates of in-hospital mortality, adjunct therapies and complications stratified by ECMO indication are reported in Supplementary Table 1.
Table 2Unadjusted rates of in-hospital mortality, adjunct therapies and complications stratified by race. All frequencies reported as %. PI=Pacific Islander, AKI = acute kidney injury.
Following adjustment for patient and hospital characteristics, Asian/PI race was linked to increased odds of mortality (AOR = 1.4, 95% CI = 1.1–2.0; Ref = White) while Black and Hispanic race/ethnicity did not alter the odds of death. As shown in Fig. 2, Black race was associated with increased odds of AKI (AOR = 1.4, 95% CI: 1.2–1.7), while Hispanic ethnicity was linked to higher odds of neurologic complications (AOR = 1.6; 95% CI = 1.2–2.1), with White as reference. Risk adjustment also revealed Black race to be associated with increased incremental hospitalization costs (β coefficient = +$16,000; 95% CI = 2000–31,000) and a 3-day increment in LOS (95% CI = 0.9–5.1; Fig. 3). Hispanic ethnicity was also linked to increased incremental cost (β coefficient = +$22,000; 95% CI = 2000–42,1000) but not duration of stay (β coefficient = +2.0; 95% CI = −0.4 – 4.4 days; Fig. 4).
Fig. 2Association of race with odds of in-hospital mortality, neurologic complications, and acute kidney injury (AKI). PI=Pacific Islander.
Fig. 3Risk adjusted cost in United States Dollars in days with 95% confidence intervals by age in years at admission and stratified across race. PI=Pacific Islander.
Fig. 4Risk adjusted length of stay (LOS) in days with 95% confidence intervals by age in years at admission and stratified across race. PI=Pacific Islander.
Despite the drastic increase in ECMO utilization over the past decade, the technology remains expensive and continues to be associated with high mortality and complication rates. Minority groups have been found to have suboptimal standard health outcomes in the United States across a multitude of conditions and procedures,
warranting investigation into race-based disparities among patients receiving ECMO. The present study is the first to use a national, all-payer cohort to assess race and ethnicity-based disparities in ECMO outcomes and resource utilization. Over the study period, mortality for White patients decreased significantly but remained stable for Black and Asian/PI patients. While Asian/PI race was associated with higher odds of mortality, Black race and Hispanic ethnicity were consistently linked to increased odds of complications and resource utilization. These findings warrant further discussion.
The present study noted race-based disparities in mortality following ECMO across various indications. Specifically, rates of mortality decreased among Whites, but remained stable among Black and Asian/PI patients. The reduction in mortality and complications following ECMO has been well documented. For example, Sanaiha and colleagues have shown a decline in mortality for ECMO admissions from 2008-2014.
Several factors including technical advances in cannula/pump technology, increased center expertise and optimized patient selection have been suggested to contribute to the overall decline. The disparate improvements in outcomes over the study period may be attributable to differences in center-deployment and utilization of the most optimized practices among vulnerable populations. However, given the increased burden of comorbidities and likely delayed presentation of socioeconomically disadvantaged patients, the impact of patient-level factors on trends in outcomes cannot be ignored. Nonetheless, the dissemination of standardized practices and patient selection criteria may ameliorate disturbing disparities in ECMO outcomes observed in the present, nationally representative data.
Despite risk-adjustment, the present study also found non-white race/ethnicity to be associated with inferior clinical outcomes following ECMO, including mortality and complications. Specifically, Asian race was linked to higher odds of death. This finding is consistent with previous reports of heath disparities, including rates of coronary artery disease, stroke, and diabetes, among Asian American subgroups.
While other races/ethnicities were not linked to mortality, Black race was associated with increased odds of AKI. Similarly, Hispanic ethnicity linked to higher odds of neurologic complications, with White as reference. The reason for these findings is likely multifactorial. First, patients of low socioeconomic status, including minority race and ethnicity, have been shown to present to the hospital with more advanced disease progression and uncontrolled comorbidities such as diabetes. The complex hospital course of patients of color may reflect a lack of access to preventative and primary care compounded by medical mistrust leading to the under management of chronic conditions prior to admission.
Race, medical mistrust, and segregation in primary care as usual source of care: findings from the exploring health disparities in integrated communities study.
However, the influence of clinical and biologic factors cannot be examined in the present work. Additionally, structural racism, the institutional, cultural, and historical forces that adversely affect people of color, likely plays a role in the observed disparities.
These systemic forces not only lead to disparate health risks among minority individuals but also chronically disadvantage health systems that predominately serve historically marginalized communities. A study by Gandjian et al. found safety net hospitals were associated with worse survival and clinical outcomes in ECMO patients compared to non-safety net centers.
Since White patients in our cohort were more likely to have been transferred into hospitals for ECMO, it is plausible that there are race- and ethnic-based differences in management of critically ill patients who qualify for ECMO. The present data demonstrates the need for racially/ethnically sensitive research to optimize outcomes for patients of all racial and ethnic backgrounds.
Furthermore, non-White race/ethnicity was associated with significantly increased resource use following ECMO from a health systems perspective. Several studies have shown that, in addition to poor outcomes, non-white patients accumulate greater healthcare expenditure.
Notably, we observed greater rates of blood transfusion and tracheostomy among non-white patients. These interventions are costly and contribute to the increased resource use of non-white patients. Furthermore, several groups have shown complications to be associated with increased resource use during index hospitalization and following discharge. While the present study cannot capture long term costs, the impact of these sequela cannot be ignored. For example, a study by Mehta et al. showed that tracheostomies contribute to over 200,000 dollars in expenditure one-year following discharge.
Given the resource intensive nature of ECMO, efforts to reduce complications and resource use associated with this life saving modality are necessary.
The present study has several limitations owing to its design and use of administrative data. First, the retrospective nature of this study precludes any causal conclusions. Although the NIS provides a large number of variables for all US hospital admissions, it does not include physiologic parameters, lab values, duration of cardiopulmonary resuscitation, and functional status; these factors may affect survival and were unable to be incorporated. In addition, the NIS is unable to capture patient with multiple indications for ECMO, such as cardiogenic shock as a result of respiratory failure secondary to idiopathic pulmonary fibrosis leading to pulmonary hypertension. This may significantly affect predicted mortality, and a single indication for ECMO may not adequately characterize our patient population. Another limitation is that ICD-10 coding does not specify ECMO cannulation strategy (i.e. veno-venous vs. veno-arterial) and was thus unable to be evaluated in our analysis. Moreover, not including patients who self-reported their race/ethnicity as “other” may have led to differential exclusion.
5. Conclusions
In summary, we found mortality declined over the study period for White patients but remained stable for Black and Asian/PI patients. Following risk adjustment, Asian/PI race was linked to increased odds of mortality while Black race and Hispanic ethnicity were associated with higher odds of complications and resource utilization. Our findings warrant more robust clinical guidelines for ECMO use, additional research to mitigate outcome disparities, and policy efforts to reduce resource use associated with this life saving intervention.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of competing interest
The authors have no financial disclosures or conflicts of interest to report.
Acknowledgments
None.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation for severe adult respiratory failure (CESAR): a multicentre randomised controlled trial.
Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome and posterior probability of mortality benefit in a post hoc bayesian analysis of a randomized clinical trial.
Race, medical mistrust, and segregation in primary care as usual source of care: findings from the exploring health disparities in integrated communities study.
00583-9&id=gr1.jpg){kind=link}
00583-9&id=gr2.jpg){kind=link}
00583-9&id=gr3.jpg){kind=link}
00583-9&id=gr4.jpg){kind=link}