Examining the Association Between MIS-C and the Child Opportunity Index at a Single Center

This retrospective case control study included children ≤21 years hospitalized and evaluated for MIS-C at a tertiary care children’s hospital. Our first study group included children diagnosed with confirmed MIS-C between March 2020 and June 2021 using the Centers for Disease Control and Prevention’s definition.⁸  Our control group included children evaluated for MIS-C between March 2020 and September 2020 after presenting with fever, laboratory evidence of inflammation, and 2 organ system involvement, but ultimately ruled out. Our MIS-C task force met weekly to review every hospitalized patient evaluated for MIS-C and created this control group from those children determined not to have MIS-C per our institutional evaluation algorithm. This control group was used previously, and we did not apply additional exclusion criteria for the current study.⁹  Our institutional review board approved this study.

Overall COI and its 3 subtypes were determined for each patient by using their home address mapped to the census tract.⁵  Categorical census tract COI (very low, low, moderate, high, very high), each containing 20% of the United States’ child population, was used.⁶ 

Importantly, we did not evaluate disparities in MIS-C diagnosis by race or ethnicity, although previous work has established racial and ethnic disparities among children with MIS-C. Race and ethnicity often serve as proxies for other social and structural determinants of health, which we directly studied here using the COI.¹⁰ 

MIS-C diagnosis was our primary outcome. As an exploratory cross-sectional subanalysis, we evaluated associations between census tract COI and MIS-C severity among the study group of children diagnosed with MIS-C using 2 approaches: requiring intensive-level care or echocardiogram abnormalities. The first categorized children who required advanced respiratory or pressor support at any time during the patient’s hospitalization (yes/no). The second categorized children by abnormal echocardiogram findings (yes/no) using previously published criteria focused on coronary artery abnormalities (z-score >2), decreased ventricular systolic function (ejection fraction <55% or shortening fraction <28%), greater than trivial pericardial effusion, and atrioventricular regurgitation (greater than trivial mitral valve regurgitation and/or greater than mild tricuspid valve regurgitation).⁹ 

Demographic and clinical characteristics were abstracted from chart review. COI was obtained from diversitydatakids.org. Variables were described with frequencies. Kruskall-Wallis compared the COI distributions between children with MIS-C versus non-MIS-C and between children with severe versus nonsevere MIS-C. Logistic regression further determined associations between COI and MIS-C diagnosis. Significant odds ratios were adjusted for significant demographic covariates.

Our study population included 172 hospitalized children: 111 children were diagnosed with MIS-C and 61 children had an alternate diagnosis (Supplemental Table 3). Of children with an alternate diagnosis, ∼68% had an infection (n = 42), most commonly a viral syndrome followed by urinary tract infection/pyelonephritis.⁹ 

Children with MIS-C compared with those with an alternate diagnosis had a significantly different distribution of census tract overall COI (Table 1; P = .03). More than half of children diagnosed with MIS-C (55%) lived in census tracts with very low to low overall COI (Fig 1) compared with only 36% of children diagnosed with an alternate diagnosis. When evaluating distributions of COI subtype, 65% of children diagnosed with MIS-C lived in census tracts with very low to low health/environment COI compared with 36% of children with an alternate diagnosis (P <.001).

When evaluating associations between MIS-C diagnosis and census tract COI, children experiencing very low to low COI had significantly higher odds (2.99, 95% CI: 1.34–6.70, P = .008) of being diagnosed with MIS-C compared with children experiencing moderate COI but not compared with children experiencing high to very high COI (1.78, 95% CI: 0.81–3.92, P = .15; Table 2). Specifically, children experiencing very low to low health/environment COI had significantly increased odds (4.28, 95% CI: 1.99–9.22, P <.001) of MIS-C diagnosis compared with children experiencing high to very high health/environment COI. There were no significant associations between social/economic or education COI and MIS-C diagnosis.

The distribution of census tract COI did not vary by MIS-C severity (Supplemental Table 4).

Our study applied the COI to uniquely frame the evaluation of place-based disparities in MIS-C hospitalizations among children. In our single-center cohort, children hospitalized with MIS-C more often resided in very low to low overall COI census tracts, specifically with very low to low health/environment opportunity, compared with children hospitalized for an alternative diagnosis. Moreover, we identified hospitalized children experiencing very low to low overall and health/environment COI were significantly more likely to be diagnosed with MIS-C compared with children living in census tracts with more opportunity. We did not detect disparities in the severity of MIS-C illness by the COI, which aligns with previous work that did not find associations between MIS-C severity and socioeconomic status or SVI.³ 

Drivers for our identified associations between the COI and MIS-C are likely complex and multifactorial.¹¹  Previous literature suggests disparities among children with MIS-C may be related to experienced cultural or language barriers, increased SVI or systemic racism, decreased socioeconomic status, and/or increased hesitancy to seek care.^2,3,11,12  These social inequities may compound existing prepandemic health disparities in communities with very low to low overall COI, leading to increased COVID-19 exposure and subsequent development of MIS-C.^2,3  The health/environment COI was the only subtype significantly associated with MIS-C diagnosis. This subtype includes measures of health insurance coverage, access to healthy food and green space, area walkability, housing vacancy, and environmental pollutants.⁶  It is possible that decreased health insurance leading to decreased health care access, increased exposure to toxic air pollution, or decreased or unsafe outdoor space may collectively create environments that make children who live in them more susceptible to developing COVID-19 and, subsequently, MIS-C.

Using comprehensive neighborhood measures like the COI can facilitate the creation of place-based interventions founded on health care system and community partnerships to reduce health disparities with a child-centered lens. Existing multifaceted interventions have focused on underresourced neighborhoods to improve health outcomes. In Ohio, a hospital–community partnership that focused on improving preventative care for chronic pediatric conditions, transitions of care, and social risks reduced hospitalization bed day rates among children.¹³  During the pandemic, a health care system–community collaboration reduced disparities in COVID-19 incidence by increasing case management and COVID-19-related education and testing.¹⁴  Similar partnerships can be developed in neighborhoods with very low to low overall COI identified in this study to reduce COVID-19 incidence and later development of MIS-C among children. Addressing elements of the health/environment COI subtype such as improving health insurance coverage or increasing safe outdoor spaces for children, could complement these existing model partnerships to optimize program success.

Our study’s main limitation is our findings may be biased by ecological fallacy when population-level measures do not reflect specific individuals’ lived experiences.¹⁵  Also, our study focused on a single center and preceded the δ and ο variants. Our definition of severity only included the need for respiratory support, pressors, or cardiac abnormalities; future work could include other markers of MIS-C severity like neurologic or renal complications.

Finally, we could not separate if COI associations are reflective of COVID-19 exposure or MIS-C development.

Our findings build on previous work reporting racial and ethnic disparities among children with MIS-C by highlighting the contribution of place-based social determinants of health. Residing in a very low to low overall COI neighborhood, particularly with very low to low health/environment opportunity, was associated with MIS-C diagnosis among hospitalized children. As the pandemic progresses, we recommend efforts be directed to areas with lower COI to minimize COVID exposure and subsequent MIS-C development. In addition, it is important to consider additional supports in care guidelines for children with MIS-C and their families that address these place-based factors both during hospitalization and follow-up care.

We acknowledge the Children’s National Hospital’s MIS-C Taskforce for clinical care guidance and support for this article.