Good neighbors? The effect of a level 1 trauma center on the performance of nearby level 2 trauma centers.

OBJECTIVE: In this study, we sought to determine whether the proximity of a level 1 trauma center (TC) might affect the performance of a nearby level 2 TC. BACKGROUND: With the exception of research and teaching programs, level 2 TC must function at a level similar to that of level 1 TC, and provide high quality, definitive care to severely injured patients. However, the role of a level 2 TC within a region might vary significantly depending on the local trauma care environment. We postulated that the case mix, regional role and outcomes of level 2 TC are greatly influenced by the regional presence of a level 1 TC. METHODS: Data were derived from the National Trauma Databank (9.0), limiting to adults with Injury Severity Score ≥9. Level 2 TC were classified as either isolated trauma centers (ITC, >30 miles from the closest level 1 TC) or neighbored trauma centers (NTC, ≤30 miles from the closest level 1 TC). Regression was used to calculate risk-adjusted mortality at each center type. RESULTS: Fifty-five thousand six hundred and fifty-five patients were identified at 161 centers; 55% of patients were cared for at ITC (n = 84 centers). Case mix varied significantly across center type; in particular, ITC received significantly more transfer patients than NTC. After adjusting for differences in case mix, patients at ITC had a 12% lower risk of death than patients treated at NTC (0.88, 95% CI 0.78-0.98). CONCLUSIONS: Level 2 TC assume different roles depending on the local trauma system configuration. Ideally, a level 2 TC should benefit from the presence of a nearby level 1 TC through collaborations in care protocols and shared case reviews. However, these data suggest the opposite: level 2 centers in proximity to level 1 centers might perform at a lower than expected level.

Duplex ultrasound screening for deep vein thrombosis in asymptomatic trauma patients: a survey of individual trauma surgeon opinions and current trauma center practices.

BACKGROUND: Many national agencies have suggested that deep vein thrombosis (DVT) rates measure quality of hospital care. However, none provide recommendations for standardized screening. If screening practices vary among clinicians or hospitals, DVT rates could be biased-centers which perform more duplex ultrasounds report more DVTs. We hypothesized that trauma surgeons have varying opinions regarding duplex ultrasound screening for DVT in asymptomatic trauma patients, which result in varying practice patterns. METHODS: We conducted two web-based surveys regarding the use of duplex ultrasound screening for DVT in asymptomatic trauma patients. The first (individual provider level) surveyed members of two national trauma surgery organizations (American Association for the Surgery of Trauma and Eastern Association for the Surgery of Trauma). The second (trauma center level) surveyed practice patterns of National Trauma Data Bank hospitals. RESULTS: Three hundred seventeen individual surgeons completed surveys. There was wide variation in individual opinions regarding DVT screening in asymptomatic trauma patients (53% agree, 36% disagree, and 11% neither agree nor disagree). Two hundred thirteen National Trauma Data Bank hospitals completed surveys of which 28% (n=60) have a written guideline regarding DVT screening in asymptomatic trauma patients. The proportion of centers with a written protocol varied significantly by trauma center level (p<0.001) but not by teaching status. Opinions and practice patterns suggest that screening should start early and be performed weekly. The main risk factors used to suggest DVT screening are spinal cord injury and pelvic fracture. CONCLUSIONS: There are wide variations in trauma surgeons' opinions and trauma centers' practices regarding duplex ultrasound screening for DVT in asymptomatic trauma patients. This variability combined with the fact that performing more duplex ultrasounds finds more DVTs may influence reported DVT rates. DVT rates alone are biased and not reflective of true quality of trauma care.

The attributable mortality and length of stay of trauma-related complications: a matched cohort study.

OBJECTIVE: To determine the attributable mortality (AM) and excess length of stay because of complications or complication groupings in the National Trauma Data Bank. SUMMARY BACKGROUND DATA: Resources devoted to performance improvement activities should focus on complications that significantly impact mortality and length of stay. To determine which post-traumatic complications impact these outcomes, we conducted a matched cohort study. AM is the proportion of all deaths that can be prevented if the complication did not occur. METHODS: We identified severely injured patients (Injury Severity Score, > or =9) at centers that contribute complications to the National Trauma Data Bank. To estimate the AM, a patient with a specific complication was matched to 5 patients without the complication. Matching was based on demographics and injury characteristics. Residual confounding was addressed through a logistic regression model. To estimate excess length of stay, matching covariates were identified through a Poisson regression model. Each case was required to match the control on all variables, and one control was selected per case. RESULTS: Of the 94,795 patients who met the inclusion criteria, 3153 died. The overall mortality rate was 3.33%, and 10,478 (11.1%) patients developed at least 1 complication. Four complication groupings (cardiovascular, acute respiratory distress syndrome, renal failure, and sepsis) were associated with significant AM. Infectious complications (surgical infections, sepsis, and pneumonia) were associated with the greatest excess length of stay. CONCLUSIONS: This study used AM and excess length of stay to identify trauma-related complications for external benchmarking. Guideline development and performance improvement activities need to be focused on these complications to significantly reduce the probability of poor outcomes following injury.

Trauma and Injury Severity Score (TRISS) coefficients 2009 revision.

BACKGROUND: Currently used Trauma and Injury Severity Score (TRISS) coefficients, which measure probability of survival (PS), were derived from the Major Trauma Outcome Study (MTOS) in 1995 and are now unlikely to be optimal. This study aims to estimate new TRISS coefficients using a contemporary database of injured patients presenting to emergency departments in the United States; and to compare these against the MTOS coefficients. METHODS: Data were obtained from the National Trauma Data Bank (NTDB) and the NTDB National Sample Project (NSP). TRISS coefficients were estimated using logistic regression. Separate coefficients were derived from complete case and multistage multiple imputation analyses for each NTDB and NSP dataset. Associated PS over Injury Severity Score values were graphed and compared by age (adult >or=15 years; pediatric <15 years) and injury mechanism (blunt; penetrating) groups. Area under the Receiver Operating Characteristic curves was used to assess coefficients' predictive performance. RESULTS: Overall 1,072,033 NTDB and 1,278,563 weighted NSP injury events were included, compared with 23,177 used in the original MTOS analyses. Large differences were seen between results from complete case and imputed analyses. For blunt mechanism and adult penetrating mechanism injuries, there were similarities between coefficients estimated on imputed samples, and marked divergences between associated PS estimates and those from the MTOS. However, negligible differences existed between area under the receiver operating characteristic curves estimates because the overwhelming majority of patients had minor trauma and survived. For pediatric penetrating mechanism injuries, variability in coefficients was large and PS estimates unreliable. CONCLUSIONS: Imputed NTDB coefficients are recommended as the TRISS coefficients 2009 revision for blunt mechanism and adult penetrating mechanism injuries. Coefficients for pediatric penetrating mechanism injuries could not be reliably estimated.

Hips can lie: impact of excluding isolated hip fractures on external benchmarking of trauma center performance.

BACKGROUND: Trauma centers (TCs) vary in the inclusion of patients with isolated hip fractures (IHFs) in their registries. This inconsistent case ascertainment may have significant implications on the assessment of TC performance and external benchmarking efforts. METHODS: Data were derived from the National Trauma Data Bank (2007-8.1). We included patients (aged 16 years or older) with Injury Severity Score value ≥ 9 who were admitted to Level I and II TCs. To ensure data quality, we limited the study to TC that routinely reported comorbidities and Abbreviated Injury Scale codes. IHF were defined as patients, aged 65 years or older, injured as a result of falls, with Abbreviated Injury Scale codes for hip fracture and without other significant injuries. TCs were stratified according to their reported inclusion of IHF in their registry. Observed-to-expected mortality ratios were used to rank TC performance first with and then, without the inclusion of patients with IHF. RESULTS: In total, 91,152 patients in 132 TCs were identified; 5% (n = 4,448) were IHF. The proportion of IHF per TC varied significantly, ranging from 0% to 31%. When risk-adjusted mortality was evaluated, excluding patients with IHF had significant effects: 37% (n = 49) of TCs changed their performance rank by ≥ 3 (range, 1-25) and 12% of centers changed their performance quintile. The greatest change in rank performance was evident in centers that routinely include IHF in their registries. CONCLUSIONS: Given the fact that IHFs in the elderly significantly influence risk-adjusted outcomes and are variably reported by TCs, these patients should be excluded from subsequent benchmarking efforts.

Health care reform at trauma centers--mortality, complications, and length of stay.

OBJECTIVE: The Trauma Quality Improvement Program has demonstrated existence of significant variations in risk-adjusted mortality across trauma centers. However, it is unknown whether centers with lower mortality rates also have reduced length of stay (LOS), with associated cost savings. We hypothesized that LOS is not primarily determined by unmodifiable factors, such as age and injury severity, but is primarily dependent on the development of potentially preventable complications. METHODS: The National Trauma Data Bank (2002-2006) was used to include patients (older than 16 years) with at least one severe injury (Abbreviated Injury Scale score ≥ 3) from Level I and II trauma centers (217,610 patients, 151 centers). A previously validated risk-adjustment algorithm was used to calculate observed-to-expected mortality ratios for each center. Poisson regression was used to determine the relationship between LOS, observed-to-expected mortality ratios, and complications while controlling for confounding factors, such as age, gender, mechanism, insurance status, comorbidities, and injuries and their severity. RESULTS: Large variations in LOS (median, 4-8 days) were observed across trauma centers. There was no relationship between mortality and LOS. The most important predictor of LOS was complications, which were associated with a 62% increase. Injury severity score, shock, gunshot wounds, brain injuries, intensive care unit admission, and comorbidities were less important predictors of LOS. CONCLUSION: Quality improvement programs focusing on mortality alone may not be associated with reduced LOS. Hence, the Trauma Quality Improvement Program should also focus on processes of care that reduce complications, thereby shortening LOS, which may lead to significant cost savings at trauma centers.

The Trauma Quality Improvement Program: pilot study and initial demonstration of feasibility.

OBJECTIVE: The American College of Surgeons Committee on Trauma has created a "Trauma Quality Improvement Program" (TQIP) that uses the existing infrastructure of Committee on Trauma programs. As the first step toward full implementation of TQIP, a pilot study was conducted in 23 American College of Surgeons verified or state designated Level I and II trauma centers. This study details the feasibility and acceptance of TQIP among the participating centers. METHODS: Data from the National Trauma Data Bank for patients admitted to pilot study hospitals during 2007 were used (15,801 patients). A multivariable logistic regression model was developed to estimate risk-adjusted mortality in aggregate and on three prespecified subgroups (1: blunt multisystem, 2: penetrating truncal, and 3: blunt single-system injury). Benchmark reports were developed with each center's risk adjusted mortality (expressed as an observed-to-expected [O/E] mortality ratio and 90% confidence interval [CI]) and crude complication rates available for comparison. Reports were deidentified with only the recipient having access to their performance relative to their peers. Feedback from individual centers regarding the utility of the reports was collected by survey. RESULTS: Overall crude mortality was 7.7% and in cohorts 1 to 3 was 16.4%, 12.4%, and 5.1%, respectively. In the aggregate risk-adjusted analysis, three trauma centers were low outliers (O/E and 90% CI <1) and two centers were high outliers (O/E and 90% CI >1) with the remaining 18 centers demonstrating average mortality. Challenges identified were in benchmarking mortality after penetrating injury due to small sample size and in the limited capture of complications. Ninety-two percent of survey respondents found the report clear and understandable, and 90% thought that the report was useful. Sixty-three percent of respondents will be taking action based on the report. CONCLUSIONS: Using the National Trauma Data Bank infrastructure to provide risk-adjusted benchmarking of trauma center mortality is feasible and perceived as useful. There are differences in O/E ratios across similarly verified or designated centers. Substantial work is required to allow for morbidity benchmarking.

Creating a nationally representative sample of patients from trauma centers.

BACKGROUND: The National Trauma Data Bank (NTDB) was developed as a convenience sample of registry data from contributing trauma centers (TCs), thus, inferences about trauma patients may not be valid at the national level. The NTDB National Sample was created to obtain nationally representative estimates of trauma patients treated in the US level I and II TCs. METHODS: Level I and II TCs in the Trauma Information Exchange Program were identified and a random stratified sample of 100 TCs was selected. The probability-proportional-to-size method was used to select TCs and sample weights were calculated. National Sample Program estimates from 2003 to 2006 were compared with raw NTDB data, and to a subset of TCs in the Healthcare Cost and Utilization Project Nationwide Inpatient Sample, a population-based dataset drawn from community hospitals. RESULTS: Weighted estimates from the NTDB National Sample range from 484,000 (2004) to 608,000 (2006) trauma incidents. Crude NTDB data over-represented the proportion of younger patients (0 years-14 years) compared with the NTDB National Sample, which does not include children's hospitals. Few TCs in Trauma Information Exchange Program are included in Healthcare Cost and Utilization Project Nationwide Inpatient Sample, but estimates based on this subset indicate a higher percentage of older patients (age 65 year or older, 23.98% versus 17.85%), lower percentage male patients, and a lower percentage of motor vehicle accidents compared with NTDB National Sample. CONCLUSION: Although nationally representative data regarding trauma patients are available in other population-based samples, they do not represent TCs patients and lack the specificity of National Sample Program data, which contains detailed information on injury mechanisms, diagnoses, and hospital treatment.

Use of performance reports among trauma medical directors and programme managers in the American College of Surgeons' Trauma Quality Improvement Program: a qualitative analysis.

BACKGROUND: The American College of Surgeons' Trauma Quality Improvement Program (TQIP) provides trauma centres with performance reports on their processes and outcomes of care relative to their peers. This study explored how performance reports are used by trauma centre leaders to engage in performance improvement and perceived barriers to use. STUDY DESIGN: Qualitative focus group study with trauma medical directors (TMDs) and trauma programme managers (TPMs) in US trauma centres. Consistent with qualitative descriptive analysis, data collection and interpretation were inductively and iteratively completed. Major themes were derived using a constant comparative technique. RESULTS: Six focus groups were conducted involving 22 TMDs and 22 TPMs. Three major themes were captured: (1) technical uses of performance reports; (2) cultural uses of performance reports; (3) opportunities to enhance the role and value of TQIP. First, technical uses included using reports to assess data collection procedures, data quality and areas of poor performance relative to peers. In this domain, barriers to report use included not trusting others' data quality and challenges with report interpretation. Second, reports were used to influence practice change by fostering inter-specialty discussions, leveraging resources for quality improvement, community engagement and regional collaboratives. Perceived lack of specialist engagement was viewed as an impediment in this domain. Lastly, identified opportunities for TQIP to support report use involved clarifying the relationship between verification and performance reports, and increasing partnerships with nursing associations. CONCLUSION: Trauma centre improvement leaders indicated practical and social uses of performance reports that can affect intention and ability to change. Recommendations to optimise programme participation include a focus on data quality, adequate resource provision and enhanced support for regional collaboratives.

Effectiveness of low-molecular-weight heparin versus unfractionated heparin to prevent pulmonary embolism following major trauma: A propensity-matched analysis.

BACKGROUND: Pulmonary embolism (PE) is a leading cause of delayed mortality in patients with severe injury. While low-molecular-weight heparin (LMWH) is often favored over unfractionated heparin (UH) for thromboprophylaxis, evidence is lacking to demonstrate an effect on the occurrence of PE. This study compared the effectiveness of LMWH versus UH to prevent PE in patients following major trauma. METHODS: Data for adults with severe injury who received thromboprophylaxis with LMWH or UH were derived from the American College of Surgeons Trauma Quality Improvement Program (2012-2015). Patients who died or were discharged within 5 days were excluded. Rates of PE were compared between propensity-matched LMWH and UH groups. Subgroup analyses included patients with blunt multisystem injury, penetrating truncal injury, shock, severe traumatic brain injury, and isolated orthopedic injury. A center-level analysis was performed to determine if practices with respect to choice of prophylaxis type influence hospital PE rates. RESULTS: We identified 153,474 patients at 217 trauma centers who received thromboprophylaxis with LMWH or UH. Low-molecular-weight heparin was given in 74% of patients. Pulmonary embolism occurred in 1.8%. Propensity score matching yielded a well-balanced cohort of 75,920 patients. After matching, LMWH was associated with a significantly lower rate of PE compared with UH (1.4% vs. 2.4%; odds ratio, 0.56; 95% confidence interval, 0.50-0.63). This finding was consistent across injury subgroups. Trauma centers in the highest quartile of LMWH utilization (median LMWH use, 95%) reported significantly fewer PE compared with centers in the lowest quartile (median LMWH use, 39%; 1.2% vs. 2.0%; odds ratio, 0.59; 95% confidence interval, 0.48-0.74). CONCLUSIONS: Thromboprophylaxis with LMWH (vs. UH) was associated with significantly lower risk of PE. Trauma centers favoring LMWH-based prophylaxis strategies reported lower rates of PE. Low-molecular-weight heparin should be the anticoagulant agent of choice for prevention of PE in patients with major trauma. LEVEL OF EVIDENCE: Therapeutic study, level III.

Inclusion of Highest Glasgow Coma Scale Motor Component Score in Mortality Risk Adjustment for Benchmarking of Trauma Center Performance.

BACKGROUND: The Glasgow Coma Scale (GCS) is the most widely used measure of traumatic brain injury (TBI) severity. Currently, the arrival GCS motor component (mGCS) score is used in risk-adjustment models for external benchmarking of mortality. However, there is evidence that the highest mGCS score in the first 24 hours after injury might be a better predictor of death. Our objective was to evaluate the impact of including the highest mGCS score on the performance of risk-adjustment models and subsequent external benchmarking results. STUDY DESIGN: Data were derived from the Trauma Quality Improvement Program analytic dataset (January 2014 through March 2015) and were limited to the severe TBI cohort (16 years or older, isolated head injury, GCS ≤8). Risk-adjustment models were created that varied in the mGCS covariates only (initial score, highest score, or both initial and highest mGCS scores). Model performance and fit, as well as external benchmarking results, were compared. RESULTS: There were 6,553 patients with severe TBI across 231 trauma centers included. Initial and highest mGCS scores were different in 47% of patients (n = 3,097). Model performance and fit improved when both initial and highest mGCS scores were included, as evidenced by improved C-statistic, Akaike Information Criterion, and adjusted R-squared values. Three-quarters of centers changed their adjusted odds ratio decile, 2.6% of centers changed outlier status, and 45% of centers exhibited a ≥0.5-SD change in the odds ratio of death after including highest mGCS score in the model. CONCLUSIONS: This study supports the concept that additional clinical information has the potential to not only improve the performance of current risk-adjustment models, but can also have a meaningful impact on external benchmarking strategies. Highest mGCS score is a good potential candidate for inclusion in additional models.

Characteristics of ACS-verified Level I and Level II trauma centers: A study linking trauma center verification review data and the National Trauma Data Bank of the American College of Surgeons Committee on Trauma.

BACKGROUND: The Trauma Quality Improvement Project of the American College of Surgeons (ACS) has demonstrated variations in trauma center outcomes despite similar verification status. The purpose of this study was to identify structural characteristics of trauma centers that affect patient outcomes. METHODS: Trauma registry data on 361,187 patients treated at 222 ACS-verified Level I and Level II trauma centers were obtained from the National Trauma Data Bank of ACS. These data were used to estimate each center's observed-to-expected (O-E) mortality ratio with 95% confidence intervals using multivariate logistic regression analysis. De-identified data on structural characteristics of these trauma centers were obtained from the ACS Verification Review Committee. Centers in the lowest quartile of mortality based on O-E ratio (n = 56) were compared to the rest (n = 166) using Classification and Regression Tree (CART) analysis to identify institutional characteristics independently associated with high-performing centers. RESULTS: Of the 72 structural characteristics explored, only 3 were independently associated with high-performing centers: annual patient visits to the emergency department of fewer than 61,000; proportion of patients on Medicare greater than 20%; and continuing medical education for emergency department physician liaison to the trauma program ranging from 55 and 113 hours annually. Each 5% increase in O-E mortality ratio was associated with an increase in total length of stay of one day (r = 0.25; p < 0.001). CONCLUSIONS: Very few structural characteristics of ACS-verified trauma centers are associated with risk-adjusted mortality. Thus, variations in patient outcomes across trauma centers are likely related to variations in clinical practices. LEVEL OF EVIDENCE: Therapeutic study, level III.

Timing of Pharmacologic Venous Thromboembolism Prophylaxis in Severe Traumatic Brain Injury: A Propensity-Matched Cohort Study.

BACKGROUND: Patients with severe traumatic brain injury (sTBI) are at high risk for developing venous thromboembolism (VTE). Nonetheless, pharmacologic VTE prophylaxis is often delayed out of concern for precipitating extension of intracranial hemorrhage (ICH). The purpose of this study was to compare the effectiveness of early vs late VTE prophylaxis in patients with sTBI, and to characterize the risk of subsequent ICH-related complication. STUDY DESIGN: Adults with isolated sTBI (head Abbreviated Injury Scale score ≥3 and total Glasgow Coma Scale score ≤8) who received VTE prophylaxis with low-molecular-weight or unfractionated heparin were derived from the American College of Surgeons Trauma Quality Improvement Program (2012 to 2014). Patients were divided into EP (<72 hours) or LP (≥72 hours) groups. Propensity score matching was used to minimize selection bias. The primary end point was VTE (pulmonary embolism or deep vein thrombosis). Secondary outcomes were defined as late neurosurgical intervention (≥72 hours) or death. RESULTS: We identified 3,634 patients with sTBI. Early prophylaxis was given in 43% of patients. Higher head injury severity, presence of ICH, and early neurosurgery were associated with late prophylaxis. Propensity score matching yielded a well-balanced cohort of 2,468 patients. Early prophylaxis was associated with lower rates of both pulmonary embolism (odds ratio = 0.48; 95% CI, 0.25-0.91) and deep vein thrombosis (odds ratio = 0.51; 95% CI, 0.36-0.72), but no increase in risk of late neurosurgical intervention or death. CONCLUSIONS: In this observational study of patients with sTBI, early initiation of VTE prophylaxis was associated with decreased risk of pulmonary embolism and deep vein thrombosis, but no increase in risk of late neurosurgical intervention or death. Early prophylaxis may be safe and should be the goal for each patient in the context of appropriate risk stratification.

Penetrating oesophageal injury: a contemporary analysis of the National Trauma Data Bank.

BACKGROUND: Oesophageal trauma is uncommon. The aim of this study was to conduct a descriptive analysis of penetrating oesophageal trauma and determine risk factors for oesophageal related complications and mortality in the National Trauma Data Bank (NTDB). METHODS: Patients with penetrating oesophageal trauma from Levels 1 and 2 trauma centres in the NTDB (2007 and 2008) that specified how complication and comorbidity data were recorded were selected. Data collected included age, injury severity score (ISS), abbreviated injury scores (AIS), lengths of stay (LOS) and ventilation days, systolic blood pressure (SBP) in the emergency department (ED), comorbidities, oesophageal related procedures, and oesophageal related complications. Univariate and multivariable analyses were conducted to identify significant predictors of oesophageal-related complications and mortality in patients with LOS>24 h. RESULTS: 227 patients from 107 centres were studied. The mean number of patients per centre was 2 (range 1-15). Overall mortality was found to be 44% with 92% of these deaths in less than 24 h. In patients with LOS>24 h, 62% had primary repair, 13% drainage, 4% resection, 1% diversion, and 20% unspecified. No significant difference in mortality was found in patients with oesophageal related complications. The time to first oesophageal related procedure was not significantly different in those with oesophageal related complications or those who died. Significant predictors of oesophageal related complications were age and AIS of the abdomen or pelvic contents ≥3 and the only significant predictor of mortality was ISS. CONCLUSIONS: Most deaths in penetrating oesophageal trauma occur in the first 24 h due to severe associated injuries. Primary repair was the most common intervention, followed by drainage and resection. Oesophageal related complications were not found to significantly increase mortality and time to first oesophageal related procedure did not affect outcomes in this subset of patients from the NTDB.

Frequency of adoption of practice management guidelines at trauma centers.

Evidence-based management guidelines have been shown to improve patient outcomes, yet their utilization by trauma centers remains unknown. This study measured adoption of practice management guidelines or protocols by trauma centers. A survey of 228 trauma centers was conducted over 1 year; 55 completed the survey. Centers were classified into three groups: noncompliant, partially compliant, and compliant with adoption of management protocols. Characteristics of compliant centers were compared with those of the other two groups. Most centers were Level I (58%) not-for-profit (67%) teaching hospitals (84%) with a surgical residency (74%). One-third of centers had an accredited fellowship in surgical critical care (37%). Only one center was compliant with all 32 management protocols. Half of the centers were compliant with 14 of 32 protocols studied (range, 4 to 32). Of the 21 trauma center characteristics studied, only two were independently associated with compliant centers: use of physician extenders and daily attending rounds (both P < .0001). Adoption of management guidelines by trauma centers is inconsistent, with wide variations in practices across centers.

The effect of dead-on-arrival and emergency department death classification on risk-adjusted performance in the American College of Surgeons Trauma Quality Improvement Program.

BACKGROUND: The American College of Surgeons' Trauma Quality Improvement Program is focused on identifying variations in outcomes across trauma centers for the purposes of performance improvement. In previous analyses, patients who died in the emergency department were excluded. We investigated the effect of inclusion and exclusion of emergency department (ED) deaths (dead on arrival [DOA] and died in ED [DIE]) on analyses of overall risk-adjusted trauma center performance. METHODS: Data for patients admitted to 65 Trauma Quality Improvement Program hospitals during the 2009 calendar year was used. A logistic regression model was developed to estimate risk-adjusted mortality. Trauma centers were then ranked based on their observed-to-expected (O/E) mortality ratio with 90% confidence intervals (CIs) and classified by outlier status: low outliers/high performers had a 90% CI for O/E mortality ratio of less than 1, and high outliers/low performers had a 90% CI for O/E mortality ratio of greater than 1. Changes in outlier status, rank, and quartile were examined with and without DOA and DIE patients included in the analyses to discern the impact of such exclusions on overall risk-adjusted center-specific performance. RESULTS: Thirty-one trauma centers (48%) reported no DOA patients in 2009, while 6 centers (9%) reported more than 10. Of 224 patients, 14 (6.2%) had a documented time of death of more than 30 minutes after ED arrival despite being recorded as DOA. Forty-one trauma centers (63%) changed rank by three positions or less. Ten trauma centers changed their quartile ranking by a single quartile, but no centers were found to change quartile rank more than one quartile. Changes in outlier status occurred for 6 trauma centers (9%). CONCLUSION: The relative frequency of patients classified as DOA varies greatly between trauma centers. Misclassification of patients as DOA occurs. Inclusion of ED deaths in risk-adjusted analysis of mortality results in a small but insignificant change in predicting the outcome results of a trauma center. This change is less than the rate of finding a center to be a high or low outlier by chance alone using the 90% CI. Inclusion of DOA and DIE patients in risk-adjusted analysis of mortality is appropriate and eliminates the bias introduced by exclusion of ED deaths owing to misuse of the DOA classification. LEVEL OF EVIDENCE: Prognostic/epidemiologic study, level III.