Trauma Quality Improvement Program

A pilot study was initiated in June 2008 to refine the methodology and assess the feasibility of applying TQIP for quality improvement at different trauma centers. Twenty-three Level I and II trauma centers volunteered and were selected to participate in the study with ACS verification. Most Level I centers are university-based trauma centers with comprehensive services. Level II centers were included to increase geographic and patient diversity, as well as the statistical power of any analyses. Each participating center received a registrar training course that included information about TQIP objectives and infrastructures, critical data collection fields, and NTDS data definitions. Webinars with conference calls and test case data abstraction were used for follow-up training.[4]

Using NTDB data from patients admitted to trauma centers between January 1 and December 1, 2007, three cohorts were created. The first cohort included patients with blunt multisystem or blunt mechanism traumas with an Abbreviated Injury Scale (AIS) score ≥ 3 in two or more regions including the head, face, neck, thorax, abdomen, spine, and extremities. The second cohort was composed of trauma patients with penetrating truncal injuries with an AIS score ≥ 3 in at least one region including the neck, thorax, and abdomen. Patients in the third cohort had a blunt single-system injury with an AIS score ≥ 3 in only one AIS body region, with the remaining regions having a maximum AIS score of 2. The outcomes of interest were death during hospitalization as evidenced by an emergency department (ED) discharge disposition of “death” or hospital discharge disposition of “expired,” as well as the prevalence of inpatient complications.[4]

Data reports were created and distributed to each participating trauma center in June 2009. Results were consistent with the previous year's baseline findings. Cohorts of similar verified trauma centers had differences in risk-adjusted mortality rates with a large amount of variance between low-outlier and high-outlier trauma centers. The aggregate group had a relative risk-mortality of 3.3, while the single-system trauma cohort had a mortality 5.9 times higher for high-outlier facilities.[4]

The TQIP pilot results gave each trauma center feedback regarding their trauma outcomes relative to other hospitals. The results also shed light on appropriate actions to undertake in order to improve quality, such as by illuminating local or regional collaborative efforts that could implemented. Overall, the pilot study demonstrated that TQIP's anonymous measures of relative performance could successfully allow trauma centers to identify shortcomings and facilitate quality improvement using existing resources and systems at local, regional, and national levels.[4]

To participate in the program, patients must meet the following inclusion criteria: be an adult greater than sixteen years of age with at least one valid ICD 9 CM diagnosis code, history of blunt or penetrating mechanisms of injury, or have an AIS score ≥ 3. Eligible patients also must have emergency department or hospital dispositions available.[2]

Patients are excluded from the program if they have a pre-existing advance directive to withhold life-sustaining measures or are older than 65 years of age and have an isolated hip fracture.[2]

The program categorizes patients into different cohorts in order to evaluate different aspects of trauma care.[2] The cohorts are as follows:

1. Blunt multisystem injury with an AIS > 3 in a least two regions of the body
2. Penetrating truncal injury with an AIS > 3 in the regions of the neck, chest, or abdomen
3. Shock patients with a systolic blood pressure < 90 mmHg
4. Patients with isolated traumatic brain injury (TBI)
5. Elderly patients (over 65 years of age) without isolated hip fractures

This program focuses on mortality, complications, and resource use. Mortality is further subdivided into on arrival, in the emergency department, and in-hospital. Complications are further subdivided into urinary tract infections, deep venous thrombosis, ventilator-associated pneumonia, central line-related bacteremia, renal failure, and surgical site infections. Resource use is measured via length of stay in the hospital, length of stay in the intensive care unit, and number of days that a patient is on a ventilator.[2]

TQIP addresses quality issues by explicitly addressing certain care metrics, including monitoring intracranial pressure in patients with traumatic brain injuries, measuring time to operations, measuring the placement and timing of tracheostomies, measuring the time to hemorrhage control, and documenting the use of venous thromboembolism prophylaxis.[2]

TQIP helps ensure the quality of data by providing training for trauma registrars and data abstractors. TQIP administrators hold monthly activities, such as webinars and quizzes, to educate registrars on the importance of data quality. Administrators assess outlier values, perform internal and external validation, and perform data logic checks.[2]

TQIP measures multiple variables in its risk-adjustment models. These variables include factors such as age, race, gender, initial pulse rate in the emergency department, the mechanism of injury, etc. An 18 variable multivariable logistic regression model is used to estimate risk-adjusted mortality for trauma patients. Results provide observed-to-expected ratios and a 90% confidence interval of a trauma center's data compared to other de-identified trauma centers in order to gauge relative variability.[2]

TQIP's outcome estimates are heavily dependent on data quality and sample size for each participating trauma center. Despite the development of sophisticated statistical models which attempt to mitigate errors in precision, the small sample sizes of highly specific trauma patient populations will inevitably skew results. Similarly, centers reporting data need to be hyper-aware of their reporting for data fields for certain patient populations susceptible to biased or missing data (e.g., high prevalence of low Injury Severity Scores for patients with early death who did not receive complete diagnostic testing or autopsies).[2]

As of 2013, TQIP has not incorporated measures related to racial disparities in trauma outcomes. However, there is evidence to suggest that differences in trauma outcomes are due to the overall quality of hospitals serving higher proportions of minority patients, rather than being due to discriminatory delivery of care; that is, these hospitals are less likely to provide recommended care to patients.[7] In fact, almost 80% of trauma centers serving primarily minority patients are classified as high mortality trauma centers due to their ratios of observed-to-expected survival rates. Patients from all racial and ethnic backgrounds appear to be 40% more likely to survive when treated at low mortality trauma centers when compared with patients of the same race and ethnicity with the same injuries being treated at high mortality trauma centers.[8] Differences between groups are relatively small compared to overall discrepancies between recommended and observed care.[9] Still, high mortality status is not universal among institutions treating predominantly minority patients, and TQIP has not addressed this inequity.[8]

TQIP excludes from statistical analyses all dead on arrival (DOA) patients and those who expire in the emergency department. This is done out of concern regarding varying patient transit times among trauma centers located in more urban vs. rural environments as well as the severity of injuries more likely to be present at certain trauma centers. However, this practice has the potential to remove a sizable number of patients from a trauma center's mortality index (for example, if a trauma center that is exceptionally skillful at the resuscitation phase of care may receive no credit), thus impeding analysis of a vital component of trauma care. Similarly, there can be great variation in how trauma centers classify DOA patients, leading to differences in treatment (e.g., resuscitation attempts or other invasive procedures). While independent researchers have found that inclusion of ED deaths in statistical analyses yields only small, insignificant changes in TQIP outcomes, doing so eliminates bias that might otherwise be introduced.[10]