Decomposition of trauma care prices between 2019 and 2022 in the US commercially insured population

Hospital-based trauma centers (TCs) provide critical care services for trauma patients on a 24/7 basis. But maintaining trauma care readiness imposes financial burden on hospitals.1 To help cover these additional costs, TCs are allowed to charge revenue code 068x, also known as a trauma activation fee, which represents an additional billed charge on top of the usual facility fee. Hospitals are only allowed to add a trauma activation fee for patients for whom there has been a prehospital arrival notification, meaning that a patient arrived at the hospital by some form of emergency medical services, but not by self, family, or friend drop-off. While previous work has examined the trauma activation charges published in hospital transparency databases, the evidence on the actual payments for trauma care is limited.2 We assessed whether the presence of a trauma activation fee in hospital claims is associated with higher negotiated prices (allowed amounts for facility claims) for commercially insured patients with injuries, compared with similar patients with injuries, but without an additional charge for trauma activation (see Appendix 1 for details). We also decomposed the premium on prices for trauma patients to determine which clinical, patient, or hospital characteristics explain most of the payment differential. We used commercial claims between 2019 and 2022 from a large national insurer. We classified hospitals as designated TCs if they recorded at least 1 admission with a “trauma activation fee” revenue code (068x) in each year between 2011 and 2022. As a robustness test, we identified TCs using American Hospital Association Annual Surveys (Table A1). We focused our analysis on patients aged 18–64 years who were admitted to TCs with at least 1 injury-related International Classification of Diseases, Tenth Revision (ICD-10), diagnosis code on their claims because 90% of admissions with trauma activation fees involve patients with injury diagnoses. As a robustness test, we identified injury diagnoses among trauma patients and limited the injury sample (trauma and non-trauma) to admissions with these diagnoses to exclude any injuries uncommon among trauma patients (Table A2). We limited the non–trauma injury admissions to those that match trauma injury admissions on patient's age, gender, and lagged Elixhauser index as well as length of stay and the injury severity score using the nonparametric coarsened exact matching algorithm,3 with evidence on balance presented in Table A3. Our independent variable is a binary indicator signifying whether an admission included trauma activation. To account for confounding factors, we incorporated patient (age, gender, lagged Elixhauser index), clinical (hospital length of stay, the New Injury Severity Score [NISS], and Medicare Diagnosis-Related Group fixed effects), and hospital (nonprofit ownership, number of beds, operational volume, and Hospital Referral Region [HRR]–level Herfindahl-Hirschman index [HHI] based on relative bed capacity distribution, lagged by 1 year) characteristics and hospital and year fixed effects in our regression model. This study includes 587 TCs with 25 694 injury-related hospital admissions, comprising 16 437 non–trauma injury admissions and 9257 trauma injury admissions. In the unadjusted model, the total price for trauma injury admission was $12 641 higher than for non–trauma injury admission (P < .001) (Table 1). After controlling for year and hospital fixed effects, along with patient demographics and the Elixhauser score, the differential decreased to $11 966 or by approximately 5% (P < .001). Upon further adjusting for clinical complexity including NISS, length of hospital stay, and diagnosis-related group (DRG) fixed effects, the payment differential decreased by 5%, 13%, and 13%, respectively, to $8594 (P < .001) (Table 1). Estimated differences in prices: trauma injury vs non–trauma injury patients, 2019 −2022. Data from commercial claims data and American Hospital Association annual surveys. Linear regression models with Coarsened Exact Matching (CEM) were used. The first row is a bivariate regression of indicator for the presence of trauma activation fee in claims on the allowed amounts. The second row adds hospital and year fixed effects. The third row adds patient (age, gender, Elixhauser index) and hospital (binary indicator that equals 1 if hospital is nonprofit, bed size, the number of total annual inpatient commercial admissions, and lagged HRR-level HHI) characteristics. The fourth row adds the new patient injury score (NISS). The fifth row adds length of inpatient stay (days). The sixth row adds Medicare DRG fixed effects and represents our preferred specification. Abbreviations: DRG, diagnosis-related group; HHI, Herfindahl-Hirschman index; HRR, Hospital Referral Region; NISS, New Injury Severity Score. Overall, the total adjustment across all models accounted for $4047 or a 47% reduction in the payment differential relative to the baseline model, with DRG fixed effects and length of stay contributing $2734, or approximately 48% of the total reduction. In the fully specified model with all patient, clinical, and hospital characteristics and hospital fixed effects, the total price for trauma injury admissions was still 25% higher than for comparable non–trauma injury admissions. The data also revealed substantial variation in the additional payments for trauma injury admissions. The Pacific region exhibited the highest mean difference in prices, at $25 694 (P < .001) (Table 1). For-profit hospitals also exhibited the largest price differences, with trauma injury admissions incurring an additional $12 091 (P < .001) (Table 1). To our knowledge, this study provides the first descriptive evidence that commercial insurers pay a premium for care of trauma injury admissions, beyond what is paid for non–trauma injury admissions, based on observable hospital, clinical, and patient characteristics. This direct payment differential for trauma admissions represents only 1 component of the financial flows to TCs. Previous research found that hospitals designated as TCs charge higher prices for non-trauma services, suggesting a pricing “spillover effect” of their trauma designation.4 This dual financial revenue stream—direct higher prices for trauma injury admissions and indirect higher prices for non-trauma services—likely provides important financial support to TCs. Further research is needed to explore whether patients with trauma injury treated on an outpatient emergency basis also generate higher allowed amounts compared with similar non–trauma injury emergency department patients. Our study focuses on commercially insured inpatients aged 18–64 years. Future research should quantify financing from government payors such as Medicare and Medicaid and other public sources that may provide grants to TCs. Medicare regulations allow for additional payments for trauma activation—up to a maximum of $500 per patient stay.5 However, the effects of these Medicare regulations have yet to be studied. Should further research confirm limits to differential payments for trauma patients by Medicare and other government payors, it would raise questions about equitable financing across payor types, particularly given that TCs often serve large Medicare and Medicaid populations. Multipayor payment information is needed to construct a more comprehensive view of the financing of TCs and the role that commercial insurers play in supporting the trauma system infrastructure in the United States. Our study has several limitations. First, the differential in payments likely represents a combination of trauma-specific payments plus unobserved factors rather than a pure effect of trauma activation alone. Second, we do not have data on the negotiated prices specifically for the trauma activation, which are often outlined in the hospital contracts but are not recorded in the claims, which makes our findings an upper bound on the true payment differential. Supplementary material is available at Health Affairs Scholar online. Funded in part by Arnold Ventures. The data that support this study’s findings are confidential and proprietary to UnitedHealthcare and are not publicly available due to legal restrictions. The research authors accessed the disaggregated data, stored in a cloud-based environment, under a research agreement that stipulates terms and conditions for the use of that data and information. Only aggregated/summarized data is available in the publication. Data access requests can be submitted to the authors and will be reviewed for permission by the relevant parties.