Social disparities in healthcare outcomes are almost ubiquitous, and trauma care is no exception. Because social factors cannot cause a trauma outcome directly, there must exist mediating causal factors related to the nature and severity of the injury, the robustness of the victim, access to care, or processes of care. Identification of these causal factors is the first step in the movement toward health equity.
A noninferiority analysis was undertaken to compare mortality rates between Black children and White children after traumatic brain injury (TBI). Data were derived from the Trauma Quality Improvement Program (TQIP) registries for the years 2014 through 2017. Inclusion criteria were age younger than 19 years and head Abbreviated Injury Scale scores of 4, 5, or 6. A noninferiority margin of 10% was preselected. A logistic regression propensity score model was developed to distinguish Black and White children based on all available covariates associated with race at p < 0.10. Stabilized inverse probability weighting and a one-tailed 95% CI were used to test the noninferiority hypothesis.
There were 7273 observations of White children and 2320 observations of Black children. The raw mortality rates were 15.6% and 22.8% for White and Black children, respectively. The final propensity score model included 31 covariates. It had good fit (Hosmer-Lemeshow χ2 = 7.1604, df = 8; p = 0.5194) and good discrimination (c-statistic = 0.752). The adjusted mortality rates were 17.82% and 17.79% for White and Black children, respectively. The relative risk was 0.9986, with a confidence interval upper limit of 1.0865. The relative risk corresponding to the noninferiority margin was 1.1. The hypothesis of noninferiority was supported.
Data captured in the TQIP registries are sufficient to explain the observed racial disparities in mortality after TBI in childhood. Speculations about genetic or epigenetic factors are not supported by this analysis. Discriminatory care may still be a factor in TBI mortality disparities, but it is not occult. If it exists, evidence for it can be sought among the data included in the TQIP registries.