Race is not Associated with Medical Emergency Team Evaluation Prior to In-Hospital Cardiac Arrest

Study population

A retrospective cohort study utilizing the American Heart Association (AHA) Get With The Guidelines®-Resuscitation (GWTG-R) registry was conducted. The GWTG-R program is provided by the AHA. This national registry is a large, voluntary, multi-center, prospective database of IHCA which has been previously described in detail.¹⁴ Hospitals participating in the registry submit clinical information regarding the medical history, hospital care, and outcomes of consecutive patients hospitalized for cardiac arrest using an online, interactive case report form and Patient Management Tool—PMT™ (IQVIA, Parsippany, New Jersey). All participating institutions were required to comply with local regulatory and privacy guidelines and, if required, to secure institutional review board approval. Because data were used primarily at the local site for quality improvement, sites were granted a waiver of informed consent under the common rule. IQVIA serves as the data collection (through their PMT) and coordination center for the American Heart Association/American Stroke Association GWTG-R programs.

Within the registry, cardiac arrest is defined as either pulselessness or a pulse with inadequate perfusion requiring chest compressions and/or defibrillation of ventricular fibrillation or pulseless ventricular tachycardia. Quality improvement hospital personnel enroll patients with a cardiac arrest without do-not-resuscitate orders who undergo cardiopulmonary resuscitation (CPR). The registry utilizes standardized Utstein-style definitions to report patient variables and outcomes across sites.^15,16 Data accuracy is ensured by thorough training of research personnel and standardized software. This research did not meet the definition of human subjects research by Children’s Hospital of Philadelphia and was deemed exempt from human subjects research (IRB 21-019305).

Cohort Design

The study cohort was obtained from hospitals that submitted data on adults 18 years of age or older to the GWTG-R registry between January 1, 2000 and May 10, 2021. The GWTG-R registry cardiopulmonary arrest (CPA) and MET databases were linked and the cohort was derived from hospitals with 10 or more IHCAs that reported both CPA and MET data. Given the primary exposure of Black or White race, this study excluded patients of other races or patients for whom race information was incomplete or missing. The cohort was further confined to only include index IHCAs that occurred after the date of the site’s first MET evaluation as a surrogate of when the MET was instituted at a given hospital. We further excluded patients with a cardiac arrest not on a general inpatient unit (defined as general inpatient area, rehab, skilled nursing, or mental health unit) or a telemetry unit (defined as adult coronary unit, telemetry unit, or step-down unit) as MET evaluations do not typically occur in these locations. Similarly to a previous study analyzing missed opportunities for MET evaluations, we further delineated the cohort to include patients who had a MET evaluation on a general inpatient unit or telemetry unit, transferred to the ICU after MET evaluation, and afterward had an IHCA within 24 hours of transfer to assign credit for the MET evaluation.¹² The primary cohort was additionally restricted to exclude patients with missing vital signs or a modified early warning score (MEWS) < 3.

Study variables and outcomes

The primary exposure variable was race (Black vs. White) among patients with acute physiologic decline (MEWS ≥ 3) within 24 hours of cardiac arrest. Race was classified as identified in the patient’s medical record. The primary outcome was MET evaluation (Yes vs. No). Patient data included demographic characteristics (age, sex), illness category (medical cardiac, medical non-cardiac, surgical-cardiac, surgical non-cardiac), cardiac arrest location (general inpatient unit, telemetry unit, or ICU), pre-existing conditions (acute central nervous system [CNS] non-stroke event; acute stroke; baseline CNS depression; congestive heart failure [current or prior]; diabetes mellitus; renal, respiratory, or hepatic insufficiency; hypotension or hypoperfusion; major trauma; metastatic or hematologic malignancy; metabolic or electrolyte abnormality; myocardial infarction [current or prior]; sepsis or pneumonia); initial rhythm of cardiac arrest (ventricular fibrillation, pulseless ventricular tachycardia, asystole, pulseless electrical activity, bradycardia, or other/unknown), interventions in place before the cardiac arrest (antiarrhythmic infusions, arterial catheter, dialysis, electrocardiogram monitor, supplemental oxygen, vascular access, vasoactive agent), whether the event was witnessed, timing of the cardiac arrest (weekday [7:00 AM-10:59 PM], weeknight [11:00 PM-6:59 AM], weekend, or holiday), event duration, medications administered during arrest (epinephrine, vasopressors other than epinephrine bolus, sodium bicarbonate, calcium chloride or gluconate, atropine, lidocaine, amiodarone), and MEWS.

Given that vital sign abnormalities may signify that a patient warranted a MET evaluation, we included patients with a MEWS ≥ 3 within 24 hours of the IHCA. The MEWS is a bedside evaluation tool based on five physiologic parameters: systolic blood pressure, heart rate, respiratory rate, temperature, and level of consciousness for a maximum score of 14.¹⁷ Specific values for assignment of the MEWS are outlined in Supplemental Table 1. A MEWS > 4 signifies a high risk for death.¹⁸ For this analysis, MEWS was calculated as a composite of blood pressure, heart rate, respiratory rate, and temperature for a maximum MEWS of 11 (level of consciousness is not available in the database). As METs serve to intervene on patients before cardiac arrest (i.e., before they are at high risk of death), we incorporated a more sensitive cut off for MEWS (≥ 3), which may signify acute physiologic decline and consideration for MET evaluation.

Hospital characteristics included hospital teaching status, hospital bed number, geographic region, and hospital racial composition. Hospital teaching status was defined as major teaching (fellowship program), minor teaching (only a residency program), or nonteaching (no residency or fellowship program). Hospital bed number was categorized as less than 200 beds, 200-499 beds, or 500 or more beds. Geographic region was characterized as North Mid Atlantic, South Atlantic and Puerto Rico, North Central, South Central, or Mountain and Pacific. Hospital racial composition was defined as a hospital with a patient population that is less than 10% Black, 10%-30% Black, or more than 30% Black adopted from a previous study analyzing hospital racial composition and cardiac arrest outcomes.⁶ Hospital racial composition has been shown to influence cardiac arrest outcomes with hospitals with a higher proportion of Black patients having worse outcomes after cardiac arrest.⁶

Statistical analysis

To assess the association of race with MET evaluation, propensity weighted regression was utilized to reduce potential confounding due to a priori selected variables including age, sex, illness category, pre-existing conditions, cardiac arrest location, and hospital characteristics.¹⁹ While propensity weighted regression is traditionally used to balance confounders between treatment groups, it is equally well-suited to balance confounders between Black and White patients. Three steps were utilized to construct a cohort of Black and White patients that are balanced with respect to potentially confounding factors.

First, a multivariable logistic regression model of race was created using the variables described above as predictors. For patients with any missing pre-existing condition, “missing” was coded as a separate category to indicate the patient was missing this information and to equally balance missing data between the propensity-matched cohorts. Variables included in the propensity model included age, sex, illness category, pre-existing conditions, cardiac arrest location, and hospital characteristics. A butterfly plot of propensities was constructed for Black and White patients with acute physiologic decline within 24 hours of cardiac arrest (Supplemental Figure 1). Patients with propensity scores outside of the region of overlap between the histograms were excluded from the final multivariable model assessing the association between race and MET evaluation.

Second, patients were weighted utilizing stabilized inverse probability of treatment weights. This weighting increases the balance of a priori characteristics between Black and White patients, creating a weighted cohort in which patients differ with respect to race but are balanced with respect to potentially confounding variables. This is the mechanism by which confounding from characteristics is reduced. Differences in means and proportions were compared between Black and White patients. The cohorts were balanced when the absolute standardized difference for each variable was <0.10 (Supplemental Table 2).

Third, a logistic regression model was constructed using the weighted cohort to assess the association of race and MET evaluation prior to IHCA for all patients with acute physiologic decline. The model utilized a robust variance estimator to account for the correlation due to clustering of patients within hospitals and further adjusted for hospital teaching status to account for variability of MET implementation across hospitals.¹²

All analyses were performed using SAS 9.4 (SAS Institute; Cary, NC) and the AHA Precision Medicine Platform (https://precision.heart.org/) with reported p-values based on a two-sided alternative and considered significant if less than 0.05.