Reckoning with Redlining and Other Structural Barriers to Health of Critically Ill Children: Addressing systemic racism will require shifting the focus from micro- to macro- level analysis of social risks

In “Geospatial Analysis of Social Determinants of Health Identifies Neighborhood Hot Spots Associated with Pediatric Intensive Care Use for Acute Respiratory Failure Requiring Mechanical Ventilation,” Najjar et al. used geospatial analysis to identify neighborhood hot spots, which, when compared to non-hot spots, had higher rates of admission to the pediatric intensive care unit (PICU) and were not associated with readmission but were associated with longer inpatient bed day rates for children with acute respiratory failure in two large hospital systems in Georgia (1). The authors found that these hot spots were associated with higher social vulnerability index (SVI) and lower child opportunity index (COI) composite scores. In addition, the authors examined the individual components of each neighborhood level marker and found differences between hot spots across all domains of both the SVI and COI. Finally, the authors noted that Black Americans and those with public health insurance were more likely to reside in neighborhood hot spots. Perhaps the most significant limitation of this study is the that the retrospective design used unvalidated electronic health record (EHR) data to identify the addresses which were subsequently utilized to establish associations between admission and inpatient bed day rates and the neighborhood level SVI and COI measures upon which the authors draw their conclusions. The findings from this study, however, add substantially to the growing literature on disparities in pediatric critical care, and, in particular, to increasing knowledge about the effect of macro-level factors—the built environment, neighborhood risk and resilience—on the health of the most vulnerable children.

Multiple other studies have shown disparities in many aspects of pediatric critical illness including children living in impoverished neighborhoods having higher rates of PICU admission and inpatient bed day rates (2), higher severity of illness for children residing in predominantly Latino neighborhoods (3), differential treatment and outcomes for children with traumatic brain injury based on insurance (4), lower odds of admission to receive critical care services for children with traumatic injury (5), differences in sepsis outcomes based on race and insurance (6) and hot-spots associated with critical asthma (7). Additionally, one’s neighborhood has also been linked to geographic access to pediatric critical care services (8).

Screening for individual level social determinants of health has long been recommended by the American Academy of Pediatrics at routine primary care visits, recognizing the impact of poverty and social determinants of health on a child’s overall health (9). More recently, some have advocated for screening in acute settings such as the emergency department and inpatient units. Screening in these settings is designed to discover addressable social risks, providing opportunities to provide resources to mitigate these risks and their associated disparities. Screening tools for social determinants of health generally address individual level social risk factors including race, ethnicity, age, gender, food insecurity, housing stability, employment, education, assistance with public benefits including insurance, and may include questions related to adverse experiences such as abuse, neglect, or family separation. These screening tools, however, do not routinely note neighborhood level factors that may also impact one’s level of social risk. Yet, best practices for screening for social determinants of health would include screening for these macro-level risks.

Several metrics have been developed that can identify degree of social risk at a neighborhood level. Common measures utilized to describe neighborhoods include the SVI and COI used by Najjar et al., as well as the Area Deprivation Index (ADI). The Centers for Disease Control Social Vulnerability Index measures fifteen social factors, grouped into four domains that include socioeconomic status, household composition, race/ethnicity/language, and housing/transportation to estimate the level of community vulnerability in a neighborhood in order to identify areas that may need support before, during, or after a disaster, indexed to the level of census tract within neighborhoods (10). The Area Deprivation Index ranks neighborhoods by level of socioeconomic disadvantage at the census block level, organized into four domains including income, education, employment and housing quality (11). Finally, the Childhood Opportunity Index is a 29-indicator measure of neighborhood factors specifically relevant to a child’s healthy development, indexed to census tract or zip code, and grouped into three domains, including education, health and environment and social and economic (12). While each measure may include overlapping factors with other measures, in general, there is variability in how they ascertain risk factors versus protective factors, making it useful to become familiar with each of the different measures.

Rather than identifying individual level social risks, differences that are identified by using neighborhood level metrics describe structural barriers to attaining health equity. Although differences may be attributable to multiple domains, they often reflect differential care associated with systemic racism and other systems or structural level problems. The National Institute for Minority Health and Disparities convened a two-year series of workshops in 2015 envisioning a future for research directions on social determinants of health, which encouraged identification of such barriers. Palmer et al. summarized best practices from this convening, including understanding social determinants of health in the setting of their upstream contributing factors and downstream consequences, positing the mechanisms through which social risks can have a serious deleterious impacts on health, specific acknowledgment that the place and context in which one lives (i.e., neighborhood factors) should be identified, recognizing the importance of protective factors and resilience and accepting that intersectionality of multiple factors and domains makes identifying and redressing social determinants of health a complex but necessary problem to confront (13). Reporting neighborhood level associations with critical illness fits within this framework. To further move this work forward, these identified neighborhood level factors must be discussed in the context of historical and current contributors that lead to conditions of higher social needs. Where protective factors have been identified, these too are critical to explore as, if they are effective, they may provide opportunities for implementation in areas where similar social needs exist in order to mitigate the impact of these needs.

In order to utilize neighborhood-level, and individual-level, social determinants of health data in the most effective way, pediatric intensivists need to be educated about social determinants of health and their various levels and complexities throughout medical education. As existing evidence supports that the majority of one’s health is framed outside of individual health care interactions and instead shaped by their social and built environments, those looking to make a meaningful impact on overall health will need to incorporate knowledge of social risks, needs, available resources, and systems level barriers into their practice. As a critical care community, this may mean using individual level social risks data to provide resources to individuals during their critical care admission, requiring implementation of screening for such needs during admission, a shift from current practice in many locations. As more is learned about both individual level social risk data and neighborhood level data, tools may be developed to help risk stratify those individuals who may present as sicker on admission, have a more complicated course, or may be expected to have complicated post-ICU courses. Prospectively tracking neighborhood level metrics, based on zip code or census tract data, upon admission to the PICU along with implementation of individual level screening will be critical to developing such tools. If EHR data will be used to identify geographic characteristics of patients, this information will also need to prospectively validated.

Najjar et al. acknowledge the multiple ways in which pediatric intensivists can become involved to address the individual and structural barriers to health faced by children with critical illness, recommending—as described above—screening for individual level risk, offering resources where available to address modifiable risks, as well as systemic advocacy on behalf of patients. A first step towards implementing this approach is a shift in mindset to thinking about both micro-level individual and macro-level neighborhood social risks as relevant to child health and critical illness in particular.

Finally, addressing factors ascertained by studying neighborhood level data will require more involvement of critical care providers in learning how to describe the systems level factors that are identified as well as intensivists advocating for systems level solutions to problems that may not otherwise be seen as linked to contributing to critical illness. This requires knowledge of and comfort with the different tools to assess social risk at individual and neighborhood levels as well as the learning the appropriate ways in which structural barriers to health should be discussed and studied. In particular, differences attributed to race and ethnicity must be contextualized as social constructs associated with racism and racially motivated policy decisions (14). For example, growing research supports the concept that historic redlining, or making differential choices about neighborhood housing investment and lending based on race, has disparate impacts on health (15). Although such concepts may not have immediately apparent relevance to critical illness, investigations such as the one completed by Najjar et al. teach the critical care community that these linkages exist, and, if we are to truly have an impact on the health of children, are concepts we will need to reckon with as a critical care community.