(148) Disparities in AMI Treatment in the ICU: Insights From Multiple Datasets

Introduction: Health equity is crucial in critical care because it is a matter of life and death. ICU datasets are invaluable to critical care research yet much is unknown about how the conclusions made from these datasets compare. This work examines how differences in treatment of patients with Acute Myocardial Infarction(AMI) from two ICU datasets reflect when the datasets are analyzed separately or as a single dataset.

Methods: A retrospective analysis of MIMIC-III, a single-center database of patients hospitalized in 2008-2014, and eICU, a multi-center database of patients hospitalized in 2014-2015. Patients with a primary diagnosis of AMI and no comorbidities were extracted, and, statistical hypothesis tests used to compute differences in orders of analgesia and AMI-related drugs (ACE-inhibitors, beta-blockers, antiplatelets and statins) placed during their hospitalization. Differences were assessed by race, sex, insurance and geographical region.

Results: 2521 MIMIC and 644 eICU patients fit the inclusion criteria. In eICU, more patients from MidWest and NorthEast regions received analgesia and all AMI-related drugs except beta-blockers(p < 0.001 for all) and more female patients received multimodal analgesia(p=0.04). In MIMIC, more male patients received non-opioid analgesia(p=0.04), ace-inhibitors(p=0.002), aspirin(p=0.017), beta-blockers(p=0.03) and statins(p=0.006), more patients with government insurance received ace-inhibitors(p=0.002) and non-aspirin antiplatelets(p=0.049), and more White Americans received analgesia, aspirin, statins(p < 0.001 for all) and ace-inhibitors(p=0.01) while more Hispanic Americans received beta-blockers and non-aspirin antiplatelets(p < 0.001 for both). When the datasets are merged, there were no significant differences by sex but more White Americans received analgesia(p < 0.001), aspirin(p < 0.001) and statins(p=0.002) and more Hispanic Americans received beta-blockers(p < 001).

Conclusions: Analyzed separately, the differences in eICU are by region, while in MIMIC they are by sex and race. When combined, differences are by race. This shows that 1) health inequities manifest differently at various institutions, which is only part of the whole picture, 2) additional research on creating tools to generalize findings across various cohorts and datasets is required.