An Algorithmic Approach for Causal Health Equity: A Look at Race Differentials in Intensive Care Unit (ICU) Outcomes

TL;DR

This paper addresses racial/ethnic disparities in ICU outcomes by applying a causal fairness framework grounded in a Standard Fairness Model (SFM) with $X$ as the protected attribute, $Z$ as confounders, $W$ as mediators, and $Y$ as the outcome. It decomposes the marginal mortality difference $\mathbb{E}[Y|X=x_1]-\mathbb{E}[Y|X=x_0]$ into direct ($x$-DE), indirect ($x$-IE), and confounded ($x$-CE) components using $TV$-based reasoning and efficient influence-function–based debiasing, validated across Australian (ANZICS APD) and US (MIMIC-IV) ICU cohorts. The analysis reveals three consistent causal mechanisms: confounded effects through age/SES that reduce mortality for minorities, indirect effects via worse chronic health and non-elective/medical admissions that increase mortality, and a counterintuitive direct effect that is protective when covariates are equal, linked to higher ICU admission prevalence (tipping-over concept). The Indigenous Intensive Care Equity (IICE) Radar provides a geography-aware monitor of ICU-admission risk for Indigenous populations, offering a practical tool for tracking primary-care access gaps and guiding policy to address health inequities.

Abstract

The new era of large-scale data collection and analysis presents an opportunity for diagnosing and understanding the causes of health inequities. In this study, we describe a framework for systematically analyzing health disparities using causal inference. The framework is illustrated by investigating racial and ethnic disparities in intensive care unit (ICU) outcome between majority and minority groups in Australia (Indigenous vs. Non-Indigenous) and the United States (African-American vs. White). We demonstrate that commonly used statistical measures for quantifying inequity are insufficient, and focus on attributing the observed disparity to the causal mechanisms that generate it. We find that minority patients are younger at admission, have worse chronic health, are more likely to be admitted for urgent and non-elective reasons, and have higher illness severity. At the same time, however, we find a protective direct effect of belonging to a minority group, with minority patients showing improved survival compared to their majority counterparts, with all other variables kept equal. We demonstrate that this protective effect is related to the increased probability of being admitted to ICU, with minority patients having an increased risk of ICU admission. We also find that minority patients, while showing improved survival, are more likely to be readmitted to ICU. Thus, due to worse access to primary health care, minority patients are more likely to end up in ICU for preventable conditions, causing a reduction in the mortality rates and creating an effect that appears to be protective. Since the baseline risk of ICU admission may serve as proxy for lack of access to primary care, we developed the Indigenous Intensive Care Equity (IICE) Radar, a monitoring system for tracking the over-utilization of ICU resources by the Indigenous population of Australia across geographical areas.

Figures (9)

• Figure 1: Diagram of the performed analysis steps.
• Figure 2: (a) Standard Fairness Model constructed for the data (b) Decomposition of the marginal disparity in outcome into confounded, indirect, and direct effects, on ANZICS APD and MIMIC-IV datasets.
• Figure 3: (a) Age in Australian data; (b) Age in US data; (c) Deciles of socioeconomic status in Australian data; (d) SOFA score in Australian data; (e) SOFA score in US data; (f) chronic health status in US data (Charlson comorbidity index).
• Figure 4: Heterogeneity according to age group and admission type of (a) direct effect of minority status on death on ANZICS APD; (b) baseline risk ratio of ICU admission in Australia; (c) direct effect of minority status on readmission on ANZICS APD; (d) direct effect of minority status on death on MIMIC-IV; (e) direct effect of minority status on readmission on MIMIC-IV.
• Figure 5: Indigenous Intensive Care Equity (IICE) Radar.