Trauma Center Ratings in Hospital Compare: Poor Performance or Biased Modeling?

Author(s):
Brian Young; Joseph Golob Jr M.D.; Alexis Harvey; Samuel Zolin; Vanessa Ho; Jeffrey Claridge

Background:

Surgical site infection (SSI) after colon surgery is a key component of the Centers for Medicare & Medicaid Service’s (CMS) star rating as well as incentive and penalty programs. All colon cases, including those for trauma, are evaluated; however, trauma status is not included in the logistic regression model used to calculate standardized infection ratios (SIRs) for hospital comparison.

Hypothesis:

We hypothesized that trauma centers would have worse colon SIRs and star ratings when compared to non-trauma hospitals.

Methods:

Current hospital star ratings and 2014 and 2016 colon SIRs were obtained from Hospital Compare and matched to state trauma designations from the Trauma Information Exchange Program. Trauma centers were defined as level 1-3. Hospitals without 2016 colon SIRs were excluded. SIRs and star ratings were compared for trauma and non-trauma centers and by trauma level. Student’s t-test, analysis of variance (ANOVA), and Tukey post hoc analysis were used for comparison.

Results:

1951 hospitals were included. 849 (44%) were trauma centers: 206 level 1s, 307 level 2s, and 336 level 3s, respectively. 2016 colon SIRs were higher for trauma centers than non-trauma centers (0.98±0.69 vs 0.88±0.72, p<0.01). ANOVA yielded significant variation among trauma levels for colon SIRs in 2016, p<0.01. A post hoc Tukey test showed that level 1 centers had higher SIRs than each remaining hospital type, p<0.05. These results were similar for the 2014 baseline (1.03±0.78 vs 0.90±0.76 p<0.01; ANOVA, p<0.01), Tukey p<0.05. Despite an apparent dose effect for higher trauma designation, no significant differences were noted between level 2, 3, and non-trauma hospitals.  A review of hospital star ratings noted no difference based on trauma center status alone (2.97±1.13 vs 3.01±1.21, p=0.47), but ANOVA was significant (p<0.01), with level 1 trauma centers having lower star ratings than each remaining hospital type by Tukey analysis, p<0.05.

Conclusions:

Level 1 trauma centers perform worse on CMS’s SIR calculation for colon SSIs, contributing to lower overall star ratings. Our data suggests a biased model which may be penalizing hospitals for providing care to trauma patients. Despite a new model being introduced in 2019, trauma status remains excluded.