You are on page 1of 8

Injury 53 (2022) 2915–2922

Contents lists available at ScienceDirect

Injury
journal homepage: www.elsevier.com/locate/injury

Change in outcomes for trauma patients over time: Two decades of a


state trauma system
Elinore J. Kaufman a,∗, Patrick M. Reilly a, Justin S. Hatchimonji b, Ruiying Aria Xiong c,
Wei Yang c, C. William Schwab a, Jay A. Yelon a,d, Daniel N. Holena a,#
a
Division of Traumatology, Surgical Critical Care, and Emergency Surgery, University of Pennsylvania Perelman School of Medicine
b
Department of Surgery, University of Pennsylvania Perelman School of Medicine
c
Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania
d
Naval Strategic Health Alliance for Readiness and Performance, Navy Medicine Operational Training Command

a r t i c l e i n f o a b s t r a c t

Article history: Background: Trauma center mortality rates are benchmarked to expected rates of death based on patient
Accepted 9 June 2022 and injury characteristics. The expected mortality rate is recalculated from pooled outcomes across a
trauma system each year, obscuring system-level change across years. We hypothesized that risk-adjusted
Keywords: mortality would decrease over time within a state-wide trauma system.
Trauma systems Methods: We identified adult trauma patients presenting to Level I and II Pennsylvania trauma centers,
Shock 1999–2018, using the Pennsylvania Trauma Outcomes Study. Multivariable logistic regression generated
risk-adjusted models for mortality in all patients, and in key subgroups: penetrating torso injury, blunt
multisystem trauma, and patients presenting in shock.
Results: Of 162,646 included patients, 123,518 (76.1%) were white and 108,936 (67.0%) were male. The
median age was 49 (interquartile range [IQR] 29–70), median injury severity score was 16 (IQR 10–24),
and 87.5% of injuries were blunt. Overall, 9.9% of patients died, and compared to 1999, no year had sig-
nificantly higher adjusted odds of mortality. Overall mortality was significantly lower in 20 07–20 09 and
2011–2018. Of patients with blunt, multisystem injuries, 17.7% died, and adjusted mortality improved over
time. Mortality rates were 24.9% for penetrating torso injury, and 56.9% for shock, with no significant
change. Mortality improved for patients with ISS < 25, but not for the most severely injured.
Conclusions: Over 20 years, Pennsylvania trauma centers demonstrated improved risk-adjusted mortality
rates overall, but improvement remains lacking in high-risk groups despite numerous innovations and
practice changes in this time period. Identifying change over time can help guide focus to these critical
gaps.
© 2022 Elsevier Ltd. All rights reserved.

Introduction of improving center performance over time. Center reports typi-


cally compare each center’s performance to that of their peers, us-
Trauma systems have long been leaders in quality and per- ing an risk-adjusted outcome metric such as observed-to-expected
formance improvement. Quality improvement programs, such as (O:E) mortality ratios [1]. These collaborative systems allow cen-
the American College of Surgeons’ Trauma Quality Improvement ters to evaluate their own performance against that of their peers,
Program (TQIP) provide risk-adjusted benchmarking with the aim and can spark performance improvement and the sharing of best
practices [1]. However, they do not allow the system or an in-
dividual center to assess its performance over time in absolute

Corresponding author. terms. As the expected mortality rate is calculated from the risk-
E-mail addresses: elinore.kaufman@pennmedicine.upenn.edu (E.J. Kaufman), adjusted outcomes of patients within the system, if system-wide
patrick.reilly2@pennmedicine.upenn.edu (P.M. Reilly), performance changes from year to year, this change will not ap-
jhat@pennmedicine.upenn.edu (J.S. Hatchimonji), aria.xiong@pennmedicine.upenn.edu pear in quality reports [1]. This is particularly true as patient fac-
(R.A. Xiong), weiyang@pennmedicine.upenn.edu (W. Yang),
tors included in risk adjustment have changed over time, and in-
charles.schwab@pennmedicine.upenn.edu (C.W. Schwab),
daniel.holena@pennmedicine.upenn.edu, dholena@mcw.edu (D.N. Holena). jured patients are older and have more comorbid conditions than
#
Current affiliation: Department of Surgery, Medical College of Wisconsin previously [2].

https://doi.org/10.1016/j.injury.2022.06.011
0020-1383/© 2022 Elsevier Ltd. All rights reserved.

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Implementation and coordination of trauma systems decreases Analysis


mortality after severe injury [3–5], but less is known about change
over time within an established trauma system. In the first decade Demographics, injury characteristics and severity variables were
of the 21st century, the proportion of deaths due to trauma in- described using chi-squared test for categorical variables. Trends
creased at the population level, with the largest increase in older over time in patient characteristics were assessed using the non-
adults [6]. During this time period, risk-adjusted trauma mortal- parametric trend test [14]. We constructed multivariable logistic
ity was stable nationally [2], but declined by 29% in Pennsylvania regression models designed to approximate the TQIP approach [1].
trauma centers [7]. However, these benefits were not equally dis- In the model, we adjusted for Glasgow coma scale (GCS) motor
tributed. Moderate to severely injured patients showed the great- component, systolic blood pressure, and heart rate (all measured
est improvement. There was no change in outcomes for patients on presentation), patient age, mechanism of injury, maximum ab-
presenting with hypotension or for the mildly or very severely breviated injury score (AIS) for head and for lower extremity, over-
injured. Despite these encouraging results, progress for the most all maximum AIS, and patient comorbidities: cancer, liver disease,
severely injured and ill remains elusive, particularly for patients heart disease, hypertension, dialysis, functional dependence, bleed-
who present with hemodynamic compromise. ing disorder, peripheral vascular disease, and dementia as defined
In this study, we assessed performance over time in a statewide by PTOS [15]. Data were evaluated for missingness. As < 5% of pa-
trauma system. Our goal was to determine how outcomes would tients had missing values for the variables included in the regres-
have differed if an identical patient with identical injuries pre- sion model, we excluded patients with missing exposure variables
sented 5 or 10 years earlier or later. We aimed to extend the find- and conducted a complete case analysis. We conducted a sensi-
ings of the prior Pennsylvania study over a longer time period, and tivity analysis using multiple imputation for missing values of ex-
to focus on key, high-risk groups of injured patients. We hypothe- posure variables. This analysis yielded nearly identical results. For
sized that risk-adjusted mortality would decrease over time. simplicity, we present here only the complete case analysis. To as-
sess for change over time, we included a categorical variable for
Patients and methods year in the model.
Beyond overall performance, we identified three key subgroups
Data source and population for performance analysis: 1. Blunt multisystem injury, defined as
blunt mechanism of injury with AIS ≥ 3 in at least 2 body regions;
We performed a retrospective cohort study of injured patients 2. Penetrating truncal injury, defined as penetrating mechanism of
presenting to all level I and II trauma centers in Pennsylvania from injury with AIS ≥ 3 in the chest, abdomen, or pelvis; 3. Shock, de-
1999 to 2018. The Pennsylvania Trauma Systems Foundation (PTSF) fined as patients presenting with presenting systolic blood pres-
accredits all trauma centers in Pennsylvania [8]. PTSF also main- sure < 90 mmHg. We also stratified our analysis by injury sever-
tains a statewide trauma registry, the Pennsylvania Trauma Out- ity and assessed adjusted odds of mortality in mild-moderately in-
comes Study (PTOS) [9]. PTOS includes demographic, physiologic, jured patients (ISS < 16), severely[16–24], and very severely in-
injury and outcomes data for patients presenting to accredited jured (ISS ≥ 25). Analysis was performed using Stata (Version 16,
trauma centers in the state of Pennsylvania. This registry includes StataCorp) and R (Version 4.2, R Foundation).
all patients admitted for injury, as well as all trauma deaths, and
all admitted transfers in or out. We excluded patients <16 years
Results
old, those with a mechanism of burn, and those transferred out
(as final outcomes could not be determined). We included only
We included 162,646 patients at 23 Level I and II trauma cen-
patients treated at centers that participated throughout the study
ters. The median age was 49 (interquartile range 29–70), 68.4%
period. Injury diagnoses were classified according to International
were male, and 87.4% were injured by blunt mechanism. Median
Classification of Disease, version 9 (ICD-9) codes for data through
ISS was 16 (interquartile range 10–24). Patient characteristics are
September 2015 and with ICD-10 codes thereafter. Data for this
shown in Table 1.
work were provided by the Pennsylvania Trauma Systems Foun-
Key patient characteristics changed significantly over the 20-
dation (Mechanicsburg, PA), which specifically disclaims respon-
year study period, as shown in Fig. 1. Median age increased from
sibility for any analyses, interpretations, or conclusions presented
40 in 1999 to 62 in 2018 (p for trend < 0.001), with the propor-
herein. The Institutional Review Board of the University of Penn-
tion age 65 or older increasing from 24.9 to 45.8. The proportion
sylvania approved this study.
of women increased from 31.4% to 38.2%, p<0.001. The proportion
injured by fall increased from 21.5% to 53.5% (p < 0.001), while
Variables
the proportion injured by firearm decreased from 11.3 to 6.7% (p
< 0.001) and the proportion injured in a motor vehicle crash de-
The outcome of interest was in-hospital mortality. The exposure
creased from 37.7% to 15.5% (p < 0.001). Median injury severity
of interest was year of injury. Patient demographics, comorbidities,
was 16 throughout the study period.
injury mechanism, diagnoses, and severity, and presenting physiol-
Overall, 116,406 (71.6%) of patients had at least one comorbid-
ogy were collected and compared. The Center for Disease Control-
ity recorded, and this proportion increased from 53.2% in 1999 to
recommended framework for external-cause-of-injury-codes was
85.3% in 2018 (p for trend < 0.001). The proportion of patients
used to determine the major mechanism and intention of injury
with heart disease increased from 9.4 to 16.3% (p < 0.001), while
[10].
hypertension increased from 18.6% to 45.0%. The proportions of pa-
The database includes registrar-recorded AIS and ISS scores.
tients with liver disease, cancer, and those on dialysis increased
However, the AIS version was updated in 2011 from AIS 1998 to
significantly but remained < 2%.
AIS 2005 [11]. This yielded AIS values that were not directly com-
parable across years. That is, a given injury in 2010 would have
received a different (and often lower) AIS score if it had occurred Overall mortality
in 2012. There is no crosswalk between AIS versions. Therefore,
to achieve consistency across all years we calculated AIS and ISS A total of 16,118 patients died, for an unadjusted mortality rate
scores from ICD-9 and 10 codes using a validated algorithm, ICD- of 9.9%. In multivariable analysis, adjusted odds of mortality de-
PIC [12,13]. creased during the study period, with an odds ratio for 2018 of

2916

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Table 1
Characteristics of patients treated at Pennsylvania trauma centers, 1999–2018.

All patients Blunt multisystem Penetrating truncal injury Patients in shock

N = 162,646 N = 20,303 N = 17,058 N = 11,186


Age# 49 (29–70) 43 (26–61) 27 (21–38) 37 (24–57)
Male 108,936 (67.0) 13,879 (68.4) 15,338 (89.9) 8584 (76.7)
GCS Motor score
1–2 21,175 (13.0) 5694 (28.0) 4666 (27.4) 6650 (59.4)
3–4 3690 (2.3) 794 (3.9) 363 (2.1) 330 (3.0)
5–6 137,781 (84.7) 13,815 (68.0) 12,029 (70.5) 4206 (37.6)
Systolic blood pressure# 137 (118- 154) 127 (105–146) 120 (89–141) 63 (0–80)
Heart rate# 88 (75–102) 95 (80–112) 92 (72–110) 76 (0–107)
Injury severity score
< 16 80,006 (49.2) 66 (0.3) 8369 (49.1) 3367 (30.1)
16–24 59,413 (36.5) 7342 (36.2) 4103 (24.1) 3160 (28.2)
≥25 23,227 (14.3) 12,895 (63.5) 4586 (26.9) 4659 (41.7)
Mechanism of injury
Fall 49,581 (30.5) 3325 (16.4) 42 (0.2) 1069 (9.6)
Pedestrian/Pedal cyclist 9099 (5.6) 1982 (9.8) 10 (0.1) 860 (7.7)
Motor vehicle crash 43,569 (26.8) 10,237 (50.4) 7 (0.0) 2975 (26.6)
Motorcycle crash 10,222 (6.3) 2448 (12.1) 1 (0.0) 762 (6.8)
Firearm injury 14,935 (9.2) 17 (0.1) 11,574 (67.9) 3863 (34.5)
Struck 7593 (4.7) 447 (2.2) 55 (0.3) 183 (1.6)
Cut or stabbed 5521 (3.4) 28 (0.1) 4922 (28.9) 667 (6.0)
Other 22,126 (13.6) 1819 (9.0) 447 (2.6) 807 (7.2)
Comorbidities
Heart disease 20,888 (12.8) 1782 (8.8) 247 (1.4) 761 (6.8)
Liver disease 1666 (1.0) 191 (0.9) 43 (0.3) 124 (1.1)
Cancer 2209 (1.4) 167 (0.8) 46 (0.3) 95 (0.8)
Hypertension 52,095 (32.0) 4666 (23.0) 1205 (7.1) 1540 (13.8)
Dialysis 1076 (0.7) 87 (0.4) 14 (0.1) 47 (0.4)
Dementia 6820 (4.2) 418 (2.1) 19 (0.1) 117 (1.0)
Functional dependence∗ 4237 (2.6) 242 (1.2) 26 (0.2) 59 (0.5)
Peripheral vascular disease∗ 687 (0.4) 61 (0.3) 19 (0.1) 19 (0.2)

Variables available 2012–2018.
#
Median (interquartile range). All others N (%).

Fig. 1. Change in Pennsylvania trauma patient characteristics over time, 1999–2018.

2917

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Fig. 2. Adjusted odds ratio of mortality for Pennsylvania trauma patients, 1999–2018
A: Overall mortality and blunt multisystem injury
B: Penetrating torso injury and patients presenting in shock.

2918

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Fig. 3. Adjusted odds ratio of mortality for Pennsylvania trauma patients by injury severity, 1999–2018.

0.75 (95% CI 0.66, 0.86). As shown in Fig. 2A, in multivariable anal- tile range 21–38) and 89.9% were male. Firearm injuries accounted
ysis, odds of mortality were significantly lower than baseline in for 67.9% and cut or stab injuries for 28.9%. Median ISS was 16
20 07–20 09 and 2011–2018 compared to 1999. Additional factors (interquartile range 9–25). Unadjusted mortality was 24.9%, and
that were significantly associated with mortality in multivariable there was no significant change in adjusted mortality over time
analysis included patient age (OR 1.04, 95% CI 1.04–1.05), low GCS as shown in Fig. 2B. Factors that were independently associated
motor score, abnormal vital signs, and higher ISS. Patients trans- with odds of mortality in multivariable analysis were similar to the
ferred in had lower odds of death (OR 0.62, 95% CI 0.58–0.65), overall group. Compared to firearm injury, cut/stab injuries were
as did those with a mechanism of firearm injury or pedestrian associated with lower odds of mortality (OR 0.18, 95% CI 0.15–
struck. Comorbidities associated with higher odds of mortality in- 0.23).
cluded heart disease (OR 1.68, 95% CI 1.58–1.79), cancer (OR 2.16,
95% CI 1.89–2.46), and liver disease (OR 3.85, 95% CI 3.31–4.47).
Full regression results are included in the online supplementary Shock
material.
Overall, 11,186 patients (6.9%) presented in shock. The median
age was 37 (IQR 24–57) and 76.7% were male. The most common
Blunt multisystem injury
mechanism of injury was firearm (34.5%) followed by motor vehi-
cle crash (26.3) and median ISS was 21 (interquartile range 13–29).
We identified 20,303 patients with blunt multisystem injury
Of the 6362 (56.9) patients in shock who died, 3664 (57.1%) died
(12.5% of trauma patients). The median age was 43 (IQR 26–61)
within an hour of presentation, and another 1246 died between 1
and 68.4% were male. The most common mechanism of injury was
and 6 h. There was no consistent change in risk-adjusted mortality
motor vehicle crash (50.4) followed by fall (16.4%) and pedestrian
rates over time, but rates were higher in 2005 and 2016 as shown
struck (9.8%). Median ISS was 27 (interquartile range 22–34). Unad-
in Fig. 2B.
justed mortality was 17.7%. In multivariable analysis, adjusted odds
of mortality decreased during the study period, with an odds ra-
tio for 2018 of 0.68 (95% CI 0.50–0.91). Odds of mortality were Analysis stratified by injury severity
significantly below baseline in 2003 and in 2006–2018, as shown
in Fig. 2A. Factors independently associated with mortality in this Of the 80,006 (49.2%) of patients with mild to moderate in-
group were similar to those in the overall population, though the juries (ISS < 16), unadjusted mortality was 4.7%. For the 59,413
only mechanism independently associated with higher mortality (36.5%) with severe injuries (ISS 16–24), 10.0% died. Among the
was pedestrian struck (OR 1.69, 95% CI 1.43, 2.02). 23,227 (14.3%) with very severe injuries (ISS ≥ 25), 27.7% died.
In multivariable analysis, mortality declined significantly for mild-
Penetrating truncal injury moderate and severe injuries. For the very severe injuries, adjusted
odds of mortality were lower during the middle of the study pe-
We identified 17,058 patients (10.5%) with penetrating injury to riod, but returned to baseline between 2014 and 2016, as shown in
the neck, chest, or abdomen. The median age was 27 (interquar- Fig. 3.

2919

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Discussion cused resuscitation and on reducing time to definitive surgical con-


trol. Rapid identification of patients at risk is necessary along with
Over 20 years, risk-adjusted mortality decreased by 25% for rapid intervention, but at least two key innovations in Pennsylva-
the injured patients treated at Pennsylvania trauma centers. When nia have not shown clear benefit. First, in Philadelphia, an increas-
expected rates of mortality are recalculated annually, the bench- ing majority of patients with penetrating injury are transported
mark changes over time. By calculating a single expected rate of by police with no prehospital intervention and exceedingly short
mortality for the 20-year period, we were able to compare out- prehospital times. Although many have hypothesized that rapid
comes over time to assess absolute, as well as relative, improve- transit to definitive care is life-saving for these patients, studies
ments. These results provide important context for performance have shown equivalent outcomes to ambulance transport, without
improvement efforts not revealed by annual performance reports demonstrating a clear advantage [25,26]. A recent analysis demon-
based on observed-to-expected ratios. Mortality improved signif- strated that patients transported by police were less likely to be
icantly for key groups including patients with blunt multisystem dead upon arrival[26] and Maher et al. found that patients trans-
injury and those with mild to severe injuries. While this progress ported by police had more severe injuries but similar survival to
is encouraging, it is limited, and vexing challenges remain in the those transported by EMS [27]. While this may indicate an inter-
care for the most severely injured patients, and the overlapping venable subgroup, it is also possible that rapid transport of pa-
groups of patients with penetrating injury and those presenting tients with unsurvivable injuries falsely elevates in-hospital mor-
in shock. Moreover, the bulk of the improvement we see appears tality rates in a way not fully accounted for by our risk adjustment
to have occurred in the first 10 years of our study, as previously models. Second, the University of Pittsburgh has pioneered the use
documented by Glance et al. [7]. without additional improvement of plasma for prehospital resuscitation. These results are relatively
in more recent years. Numerous practice changes and innovations recent, and may not be reflected in survival rates reported in this
have focused on these high-risk patients, but with no discernible study [28]. Likewise, optimizing use of pre- and in-hospital tranex-
change in risk-adjusted outcomes across Pennsylvania. While all amic acid administration holds some promise for patients in hem-
patients stand to benefit from ongoing innovation and quality im- orrhagic shock [29]. Efforts to shorten transport times through al-
provement, those who have been refractory to our efforts thus far location of transport resources; strategic location of trauma centers
deserve our focused attention moving forward. [30]; and reducing time spent on field interventions may still prove
The retrospective nature of this study prevents us from iden- useful. Rigorous implementation of transfusion protocols is cru-
tifying any particular clinical innovation or performance improve- cial, including thromboelastography[31], balanced transfusion ra-
ment effort responsible for the improvements we saw, but several tios[32,33], whole blood[34], and medical adjuncts such as tranex-
major changes in trauma care during this period are likely to have amic acid[35], calcium [36], and arginine vasopressin[37]. Emerg-
benefited patients with blunt, multi-system injury and the overall ing hemorrhage control technologies for field and hospital use may
trauma population. Cross sectional imaging has become widely and also help move us toward zero preventable deaths [38].
readily available, as have interventional radiologic techniques for Our study has inherent limitations, including the potential for
controlling hemorrhage and infection. The concept of damage con- variation in data entry and coding inherent in a retrospective, reg-
trol has become central to the management of seriously injured istry study. We saw substantial changes in the population and in-
patients during this time period and can decrease mortality sub- juries treated at Pennsylvania trauma centers over 20 years. Con-
stantially in high-risk patients [16]. Damage control resuscitation sistent with national evidence, the proportion of patients who
and balanced blood product resuscitation have also been widely were 65 or older nearly doubled in this time frame [39]. It is
adopted over the last two decades, and have been shown to reduce also possible that the nature of the injured population of the state
mortality [17]. Moreover, these 20 years have seen increasing focus changed less or differently than our results would make it ap-
on performance improvement within trauma systems, as evidenced pear. If subgroups were more or less likely to be admitted to the
by widespread adoption of the Trauma Quality Improvement Pro- hospital or to present to or be transferred into a trauma cen-
gram, initiated in 2008 [18]. While these initiatives in themselves ter, their representation in this data would be altered. Without
are not enough to improve quality of care [19], they have brought true population-level surveillance that includes non-trauma cen-
a focus on preventing and managing complications, which can be ters and non-admitted patients, our ability to assess injury out-
key to reducing later mortality after injury [20]. For example, since comes accurately remains limited [40]. We were not able to assess
2016, the Pennsylvania Trauma System Foundation quality collab- preventability of death. If we could reliably exclude the subset of
orative has used TQIP data to identify statewide priorities for im- deaths that were non-preventable, an analysis focusing on the re-
provement and to share best practices and strategies for change mainder might be more accurate. To compensate for the change
[21]. in AIS system, we used diagnosis codes to estimate AIS and ISS
Patients in shock represent the highest risk group of injured scores, which may have introduced bias. Our study spanned the
patients. Death rates remain high, and the bulk of deaths occur transition from ICD-9 to 10, and changes in diagnosis coding for
soon after injury. Our findings are consistent with prior evidence— injuries and comorbidities may have influenced our results. Injury
for example, in the Resuscitation Outcomes Consortium multicen- severity score remains an imperfect metric for assessing risk, par-
ter randomized trial, Tisherman et al. found that median time to ticularly for those with penetrating injury [41]. We did not assess
death for patients in shock was 2 h [22]. This short time frame for within-center correlation in this analysis. Finally, both the cur-
may limit our opportunity for intervention, but we must do better. rent study and the American College of Surgeons’ Trauma Quality
We also saw no improvement in survival for patients with pen- Improvement Project focus on outcomes at the level of the trauma
etrating injury, with deaths again occurring primarily early and center, but factors that contribute to outcomes after injury span
among patients who also presented in shock. These results par- a continuum which includes the prehospital setting and presenta-
allel findings that patients with penetrating trauma and hypoten- tions to non-trauma centers followed by transfer to trauma centers.
sion have very short times to death [23]. Our findings reinforce It is therefore likely that some of the factors which influence pa-
those reported by Glance et al. for the same trauma system in a tient outcomes are beyond the control of trauma centers and may
time period overlapping the first portion of the current study [7]. have changed in ways that we cannot account for in the current
A recent study of patients undergoing massive transfusion showed study. We chose to focus on in-hospital mortality as this outcome
similar lack of improvement [24]. Since the advent of hemostatic is unambiguous and universally assessed. It is not, however, the
resuscitation, efforts to offer better care to these patients have fo- only or necessarily the most important outcome. Complications,

2920

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

length of stay, long-term functional status, and costs all must play Dep Econ 2010. [cited 2016 May 15]. (Statistical Software Components). Avail-
a role in assessing the quality of trauma care but have not been able from https://ideas.repec.org/c/boc/bocode/s457028.html .
[13] Greene NH, Kernic MA, Vavilala MS, Rivara FP. Validation of ICDPIC soft-
recorded in a consistent fashion at all centers over this 20-year ware injury severity scores using a large regional trauma registry. Inj Prev
time frame. Tracking these outcomes would allow us to assess care 2015;21(5):325–30 Oct.
more accurately, particularly in groups of patients with low rates [14] Cuzick J. A Wilcoxon-type test for trend. Stat Med 1985;4(1):87–90 Mar.
[15] Pennsylvania Trauma Systems Foundation. Pennsylvania Trauma Systems Foun-
of inpatient mortality. Ground level falls, for example, make up a dation Operational Manual for the Data Base Collection System [Internet].
rising segment of trauma care, and may have long-term implica- [cited 2021 Oct 26]. Available from: https://www.ptsf.org/wp-content/uploads/
tions for health and well-being that are not captured in mortality 2021/01/2021- PTOS- Manual- March.pdf
[16] Rotondo MF, Schwab CW, McGonigal MD, Phillips GR, Fruchterman TM,
statistics [42,43].
Kauder DR, et al. Damage control”: an approach for improved survival in
exsanguinating penetrating abdominal injury. J Trauma 1993;35(3):375–82 Sep
Conclusion discussion 382-383.
[17] Damage control resuscitation in patients with severe traumatic hemorrhage
- Practice Management Guideline [Internet]. [cited 2020 Nov 18]. Avail-
Over 20 years, risk-adjusted mortality decreased by 253% for able from: https://www.east.org/education/practice-management-guidelines/
the injured patients treated at Pennsylvania trauma centers, but damage-control-resuscitation-in-patients-with-severe-traumatic-hemorrhage
gains did not extend to patients with penetrating injury or those [18] Trauma Quality Improvement Program (TQIP) [Internet]. American College
of Surgeons. [cited 2020 Nov 18]. Available from: https://www.facs.org/
presenting in shock. While trauma center performance reporting quality-programs/trauma/tqp/center-programs/tqip
typically compares centers to one another in a cross-sectional [19] Etzioni DA, Wasif N, Dueck AC, Cima RR, Hohmann SF, Naessens JM, et al. As-
fashion, this practice may obscure system-level change over time. sociation of Hospital Participation in a Surgical Outcomes Monitoring Program
With Inpatient Complications and Mortality. JAMA 2015 Feb 3;313(5):505.
Center- and system-level performance improvement should con- [20] Holena DN, Kaufman EJ, Delgado MK, Wiebe DJ, Carr BG, Christie JD, et al. A
sider absolute as well as relative progress in patient outcomes. metric of our own: Failure to rescue after trauma. J Trauma Acute Care Surg
Longitudinal, system-level outcomes are crucial to identifying key 2017;83(4):698–704.
[21] PA-TQIP Collaboration | PA Trauma Systems Foundation [Internet]. PA
areas for performance improvement, particularly for challenging Trauma. [cited 2021 Nov 18]. Available from: https://www.ptsf.org/
injury patterns and patient groups. performance-improvement/pa-tqip-collaboration/
[22] Tisherman SA, Schmicker RH, Brasel KJ, Bulger EM, Kerby JD, Minei JP, et al.
Detailed Description of all Deaths in Both the Shock and Traumatic Brain In-
Funding/Financial support jury Hypertonic Saline Trials of the Resuscitation Outcomes Consortium. Ann
Surg 2015;261(3):586–90 Mar.
No specific funding was obtained for this research. Dr. Kaufman [23] Remick KN, Schwab CW, Smith BP, Monshizadeh A, Kim PK, Reilly PM. Defining
the optimal time to the operating room may salvage early trauma deaths. J
is supported by AHRQ (grant 5K12 HS026372).
Trauma Acute Care Surg 2014;76(5):1251–8 May.
[24] Duchesne J, Taghavi S, Ninokawa S, Harris C, Schroll R, McGrew P, et al. After
Declaration of Competing Interest 800 MTP Events, Mortality due To Hemorrhagic Shock Remains High And Un-
changed Despite Several In-Hospital Hemorrhage Control Advancements. Shock
2021 May 27.
The authors have no conflicts of interest to report. [25] Band RA, Salhi RA, Holena DN, Powell E, Branas CC, Carr BG. Severity-Ad-
justed Mortality in Trauma Patients Transported by Police. Ann Emerg Med
2014;63(5):608–14 May e3.
Supplementary materials
[26] Winter E., Hynes A.M., Shultz K., Holena D.N., Malhotra N.R., Cannon J.W. As-
sociation of Police Transport With Survival Among Patients With Penetrating
Supplementary material associated with this article can be Trauma in Philadelphia, Pennsylvania. Public Health.:11.
found, in the online version, at doi:10.1016/j.injury.2022.06.011. [27] Maher Z, Beard JH, Dauer E, Carroll M, Forman S, Topper GV, et al. Police trans-
port of firearm-injured patients-more often and more injured. J Trauma Acute
Care Surg 2021 Jul 1;91(1):164–70.
References [28] Pusateri AE, Moore EE, Moore HB, Le TD, Guyette FX, Chapman MP, et al. As-
sociation of Prehospital Plasma Transfusion With Survival in Trauma Patients
[1] Newgard CD, Fildes JJ, Wu L, Hemmila MR, Burd RS, Neal M, et al. Methodology With Hemorrhagic Shock When Transport Times Are Longer Than 20 Minutes:
and Analytic Rationale for the American College of Surgeons Trauma Quality A Post Hoc Analysis of the PAMPer and COMBAT Clinical Trials. JAMA Surg
Improvement Program. J Am Coll Surg 2013;216(1):147–57 Jan. 2020;155(2):e195085 01.
[2] DiMaggio CJ, Avraham JB, Lee DC, Frangos SG, Wall SP. The Epidemiology of [29] Guyette FX, Brown JB, Zenati MS, Early-Young BJ, Adams PW, Eastridge BJ, et al.
Emergency Department Trauma Discharges in the United States. Acad Emerg Tranexamic Acid During Prehospital Transport in Patients at Risk for Hemor-
Med 2017;24(10):1244–56. rhage After Injury: A Double-blind, Placebo-Controlled, Randomized Clinical
[3] Nathens AB, Jurkovich GJ, Cummings P, Rivara FP, Maier RV. The effect of or- Trial. JAMA Surg 2021 Jan 1;156(1):11–20.
ganized systems of trauma care on motor vehicle crash mortality. JAMA 20 0 0 [30] Winchell RJ, Broecker J, Kerwin AJ, Eastridge B, Crandall M. Comparing
Apr 19;283(15):1990–4. GIS-based estimates with trauma center registry data to assess the effects of
[4] MacKenzie Ellen J, Rivara Frederick P, Jurkovich Gergory J, Nathens Avery B, additional trauma centers on system access. J Trauma Acute Care Surg 2020
Frey Katherine P, Egleston Brian L, et al. A national evaluation of the effect of Sep 10.
trauma-center care on mortality. New Engl J Med 2006;354:366–78. [31] Tapia NM, Chang A, Norman M, Welsh F, Scott B, Wall MJ, et al.
[5] MacKenzie Ellen J, Steinwachs Donald M, Ramzy Ameen I. Evaluating Per- TEG-guided resuscitation is superior to standardized MTP resuscitation in
formance of Statewide Regionalied Systems of Trauma Care. J Trauma massively transfused penetrating trauma patients. J Trauma Acute Care Surg
1990;30(6):681–8. 2013;74(2):378–85 Feb discussion 385-386.
[6] Rhee P, Joseph B, Pandit V, Aziz H, Vercruysse G, Kulvatunyou N, et al. Increas- [32] Holcomb JB, del Junco DJ, Fox EE, Wade CE, Cohen MJ, Schreiber MA, et al. The
ing Trauma Deaths in the United States. Ann Surg 2014;260(1):13–21 Jul. Prospective, Observational, Multicenter, Major Trauma Transfusion (PROMMTT)
[7] Glance LG, Osler TM, Mukamel DB, Dick AW. Outcomes of Adult Trauma Pa- Study: Comparative Effectiveness of a Time-Varying Treatment With Compet-
tients Admitted to Trauma Centers in Pennsylvania, 20 0 0-20 09. Arch Surg ing Risks. JAMA Surgery 2013 Feb 1;148(2):127.
2012 Aug 1;147(8):732–7. [33] Holcomb JB, Tilley BC, Baraniuk S, Fox EE, Wade CE, Podbielski JM, et al. Trans-
[8] Pennsylvania Trauma Centers | PA Trauma Systems Foundation [Internet]. PA fusion of Plasma, Platelets, and Red Blood Cells in a 1:1:1 vs a 1:1:2 Ratio and
Trauma. [cited 2021 Oct 26]. Available from: https://www.ptsf.org/ Mortality in Patients With Severe Trauma: The PROPPR Randomized Clinical
[9] Wiebe DJ, Holena DN, Delgado MK, McWilliams N, Altenburg J, Carr BG. The Trial. JAMA 2015 Feb 3;313(5):471.
Pennsylvania Trauma Outcomes Study Risk-Adjusted Mortality Model: Results [34] Repine TB, Perkins JG, Kauvar DS, Blackborne L. The use of fresh whole blood
of a Statewide Benchmarking Program. Am Surg 2017 May 1;83(5):445–52. in massive transfusion. J Trauma 2006;60(6 Suppl):S59–S69 Jun.
[10] U.S. Centers for Disease Control and Prevention. Proposed Matrix of E-code [35] Cole E, Davenport R, Willett K, Brohi K. Tranexamic acid use in severely injured
Groupings [Internet]. [cited 2016 Nov 23]. Available from: http://www.cdc.gov/ civilian patients and the effects on outcomes: a prospective cohort study. Ann
injury/wisqars/ecode_matrix.html Surg 2015;261(2):390–4 Feb.
[11] Abbreviated Injury Scale (AIS) [Internet]. Association for the Advancement of [36] MacKay EJ, Stubna MD, Holena DN, Reilly PM, Seamon MJ, Smith BP, et al. Ab-
Automotive Medicine. [cited 2020 Jun 18]. Available from: https://www.aaam. normal Calcium Levels During Trauma Resuscitation Are Associated With In-
org/abbreviated-injury-scale-ais/ creased Mortality, Increased Blood Product Use, and Greater Hospital Resource
[12] Clark DE, Osler TM, Hahn DR. ICDPIC: Stata module to provide methods for Consumption: A Pilot Investigation. Anesth Analg 2017;125(3):895–901 Sep.
translating International Classification of Diseases (Ninth Revision) diagnosis [37] Sims CA, Holena D, Kim P, Pascual J, Smith B, Martin N, et al. Effect of Low-
codes into standard injury categories and/or scores [Internet]. Boston College Dose Supplementation of Arginine Vasopressin on Need for Blood Product

2921

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
E.J. Kaufman, P.M. Reilly, J.S. Hatchimonji et al. Injury 53 (2022) 2915–2922

Transfusions in Patients With Trauma and Hemorrhagic Shock: A Randomized Surg 2021. Jul 6 [cited 2021 Jul 13]; Available from: http://journals.lww.com/
Clinical Trial. JAMA Surg [Internet] 2019. Aug 28 [cited 2019 Aug 29]; Available jtrauma/Abstract/90 0 0/Developing_a_measure_of_overall_intensity_of.97369.
from: https://jamanetwork.com/journals/jamasurgery/fullarticle/2749069 . aspx.
[38] National Academies of Sciences, Engineering, and Medicine A national [41] Smith BP, Goldberg AJ, Gaughan JP, Seamon MJ. A comparison of Injury Sever-
trauma care system: integrating military and civilian trauma systems ity Score and New Injury Severity Score after penetrating trauma: A prospec-
to achieve zero preventable deaths after injury [Internet], Washington, tive analysis. J Traum Acute Care Surg 2015;79(2):269–74 Aug.
DC: The National Academies Press; 2016. [cited 2019 Sep 19]. Avail- [42] Ayoung-Chee P, McIntyre L, Ebel BE, Mack CD, McCormick W, Maier RV.
able from: http://www.nationalacademies.org/hmd/Reports/2016/A-National- Long-term outcomes of ground-level falls in the elderly. J Traum Acute Care
Trauma- Care- System- Integrating- Military- and- Civilian- Trauma- Systems.aspx. Surg 2014;76(2):498–503 Feb.
[39] Khurrum M, Chehab M, Ditillo M, Richards J, Douglas M, Bible L, et al. Trends [43] Kaufman EJ, Zebrowski AM, Holena DN, Loher P, Wiebe DJ, Carr BG. The Short
in Geriatric Ground-Level Falls: Report from the National Trauma Data Bank. J and the Long of it: Timing of Mortality for Older Adults in a State Trauma
Surg Res 2021;266:261–8 Oct. System. J Surg Res 2021;268:17–24 Dec.
[40] Zebrowski AM, Hsu JY, Holena DN, Wiebe DJ, Carr BG. Developing a measure
of overall intensity of injury care: A latent class analysis. J Traum Acute Care

2922

Downloaded for Ankur Verma (ankur.verma@maxhealthcare.com) at Max Smart Super Speciality Hospital from ClinicalKey.com by Elsevier
on October 12, 2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.

You might also like