a b s t r a c t

Background: The Patient Protection and Affordable Care Act supports the establishment of Accountable Care Organizations (ACOs) as care delivery models designed to save costs. The potential for these cost savings has been demonstrated in the primary care and inpatient populations, but not for patients with emergency condi- tions or traumatic injuries.

Methods: Our study evaluated adult trauma patients transferred to the Tertiary care hospitals of a pioneer ACO, com- paring those who were transferred from within the ACO to those from outside the ACO in terms of overall cost of hospitalization. Hospital length of stay and number of imaging studies were predetermined secondary outcomes. Results: The study population included 7696 hospitalizations for traumatic injuries over a 5-year period, 85.1% of which were for patients transferred from outside the ACO. Patients transferred from within the ACO had a 7.2% lower overall cost of hospitalization (P = .032). Mean injury severity scores were not significantly different between groups. Differences in mortality, Intensive care unit length of stay, and overall hospital length of stay were not significant. However, analysis of radiology studies performed during the hospitalization revealed that patients transferred from within the ACO had, on average, 0.47 fewer advanced imaging studies per hospitalization than did those transferred from outside the ACO (3.55 vs 4.02 studies per hospitalization, P = .003).

Conclusions: Adult trauma patients transferred from within an ACO have significantly lower Total costs of hospi- talization than do those transferred from outside the system, without significant differences in Disease burden, hospital length of stay, or mortality.

(C) 2015

Introduction

The Affordable Care Act of 2010 authorizes the creation of a Medicare Shared Savings Program and the accountable care organizations (ACOs) with which it will contract. The potential benefits of ACOs, organizations accountable for the health care of a defined patient population and charged with improving quality while decreasing costs, have been demonstrated in a number of populations; cost savings have been

? The authors have no conflicts of interest to report and there was no internal or external funding for this work.

?? Presented at the American College of Emergency Medicine Scientific Assembly, 2013.

* Corresponding author. Department of Emergency Medicine, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114. Tel.: +1 617 726 3104; fax: +1 617 500 9951.

E-mail addresses: [email protected] (B.C. Geyer), [email protected] (D.A. Peak), [email protected] (G.C. Velmahos), [email protected] (J.D. Gates), [email protected] (Y. Michaud), [email protected] (L. Petrovick), [email protected] (J. Lee), [email protected] (B.J. Yun), [email protected] (B.A. White), [email protected] (A.S. Raja).

achieved in primary care group practices, multispecialty groups, and integrated delivery systems [1-3], and a recent review of 2 years of data from 32 ACOs demonstrated improvements in expenditures and resource use for patients in ACOs [4]. However, there is no evidence that similar savings can be achieved in patients presenting to the hospital with emergency conditions, including acutely ill patients with trauma.

trauma outcomes in the United States have been greatly improved by the regionalization of trauma care [5]; patients injured at outlying hospitals are transferred to American College of Surgeons-verified trauma centers, where they receive care from teams of trauma specialists. However, the acute treatment cost of patients with trauma is higher than that of patients with other Acute conditions [6]. Some of these costs have been attributed to the use of high-cost imaging in trauma because its use has risen significantly more than the proportion of severely injured trauma patients requiring such imaging would seem to warrant [7]. Recent studies have found that a portion of this increase in imaging may be due to inefficiencies caring for Transferred patients, as repeat imaging is common for patients with trauma transferred between

http://dx.doi.org/10.1016/j.ajem.2015.11.067

0735-6757/(C) 2015

456 B.C. Geyer et al. / American Journal of Emergency Medicine 34 (2016) 455-458

hospitals [8,9]. One solution for decreasing these costs may be improved electronic image sharing, a potential benefit of the Transfer of patients within an ACO.

Our objective was to compare trauma patients transferred from within a pioneer ACO to those transferred from outside the ACO in terms of total cost of hospitalization, overall length of stay (LOS), and overall imaging received during their hospitalizations. We hypothesized that trauma transfers occurring within an ACO would have lower costs and less diagnostic imaging.

Methods

Study design and setting

This was a retrospective review conducted on all trauma patients transferred into 2 associated level 1 trauma centers between January 1, 2005, and December 31, 2009. These 2 academic medical centers anchor an integrated health care system that is a pioneer ACO, which also includes 5 community acute care hospitals. Each of the level 1 trauma centers receives approximately 1500 to 2000 trauma patients per year as transfers from lower level and undesignated trauma centers in the neighboring region.

Selection of participants

We compared adult patients transferred from acute care hospitals within the pioneer ACO to those transferred from outside the system. All patients were considered eligible for inclusion in the study if they were included in either of the participating level 1 trauma center’s insti- tutional trauma registries. The only exclusion criterion was age b 18 years. Patients were grouped in the ACO cohort if they were transferred from 1 of the 5 community hospitals that comprise the remainder of the ACO network. All other patients were grouped in the non-ACO cohort.

Data collection

Data were obtained for this Health Insurance Portability and Accountability Act-compliant, institutional review board-approved study from institutional trauma registries, cost databases, and an inte- grated research data repository containing Demographic and clinical data. All data were combined using Microsoft Access 2013 (Microsoft, Redmond, WA) according to standard query methodology. Any data field with greater than 20% data missing was excluded from the analysis. The trauma registry data were abstracted from the individual patient medical records and the trauma registry (Lancet Technology, Trauma One) of 2 American College of Surgeons-verified level 1 trauma centers. The data were then entered into a secure, Web-based application developed by Vanderbilt University, REDCap-Version 5.5.10. The data were reviewed and prepared for analysis using Microsoft Excel; analyses were performed to aggregate the data, identify trends, and

describe the medical care.

Costs were obtained from institutional cost databases that track the cost of providing patient care and are stored in a secure database attached to patient-specific identifiers for analysis within the finance departments of the individual institutions. This information is stored separately from patient charges, which were not used in this analysis.

The research data repository is a centralized clinical data registry, maintained by our health care system, which gathers clinical data from within the hospital system and stores the data in one central Health Insurance Portability and Accountability Act-compliant data warehouse. Researchers are able to query these data by using an online query tool to obtain an aggregate total number of patients that meet the specified criteria as well as patients’ detailed medical records. The detailed medical records are returned to researchers in an encrypted Microsoft Access file and text (.txt) files. Detailed medical records may include the following types of data: demographics, diagnoses, discharge

notes, medications, microbiology, operative notes, pathology reports, procedures, providers, and Radiology reports.

Outcome measures

Demographic, clinical, and outcome variables were collected for all patients through the databases above. Demographic variables included age (in years) and sex. Clinical variables included heart rate (abnormal was defined as b 60 or N 100 beats/min) and blood pressure (abnormal was defined as a systolic blood pressure of b 90 mm Hg), both measured on arrival to the level 1 trauma center. Additional clinical variables recorded included injury severity score, diagnosis (categorized by body area injured), and the time from injury to arrival at the level 1 trauma center. The outcome variables collected included overall LOS (from time of hospital admission to discharge, in days), intensive care unit (ICU) LOS (from time of ICU admission to ICU discharge, in days), and total advanced imaging obtained at the level 1 trauma center (categorized by modality: ultrasound, magnetic resonance imaging [MRI], and computed tomography [CT]). The individual patient cost of care was calculated as follows: cost of care was the sum of direct and indirect costs, which were maintained in a secure institutional database of all patient encounters as part of the ongoing accounting and other business activities of our institutional finance department. Indirect costs were calculated as a sum of all overhead costs (ie, building and nondedicated personnel costs, etc) as a function of the length and type of hospitalization. Direct costs were calculated by allocating actual cost of care according to the charge codes attributed to the patient over their hospitalization.

Data management and statistical analysis

We compared patients who were transferred from facilities within the ACO with patients who were transferred from outside the ACO. Data were analyzed in STATA 12 (StataCorp, College Station, TX) using t test or ?² tests for continuous and dichotomous variables, respectively. For multivariate analysis, both overall hospital LOS and total cost of hospitalization were analyzed by analysis of variance. Crude data were log transformed with sequential adjustment for differences in principal trauma diagnosis category and overall mortality between the groups. A P value less than 0.05 was considered statistically significant for all sta- tistical tests. After statistical analysis, cost was expressed as a percent- age, per institutional policy that restricts publication of cost data in specific monetary values.

Results

There were 7696 patients transferred during the study period, in- cluding 1143 (14.9%) from within the integrated health care system. Missing data were exceedingly rare in this analysis. Virtually all fields had more than 98% complete data, including all data relating to the primary outcome. There were no records that lacked the predefined greater than 20% of missing fields. Patients transferred from within the system were older and more likely to be female (Table 1). They were also more likely to have abnormal heart rates upon arrival to the level 1 trauma center and to have shorter intervals between their estimated times of injury and their presentations at a level 1 trauma center. However, injury severity scores, rates of abnormal blood pres- sure readings, LOS in the ICUs, and in-hospital mortality were similar between groups. In addition, there were slight differences in the specific Injury profiles between the groups (Table 2).

Patients transferred from outside the ACO had a 7.1% higher overall cost of hospitalization when compared with those patients transferred from within the system, after adjusting for differences in diagnoses between the groups (P = .047; Table 3). This was despite a similar hospital LOS (3.9 vs 3.7 days, P = .095). To account for the possibility that in-hospital mortality might create artificial differences in overall

B.C. Geyer et al. / American Journal of Emergency Medicine 34 (2016) 455-458 457

Table 1

Demographic, physiologic, and outcomes for transfer trauma patients

Table 3

Cost and LOS for transfer trauma patients

ACO transfer (n = 1143)

Demographics					Adjusted for diagnosis	100.0%	107.1%	.047
Age (y)	57.1	53.6	b.001		Adjusted for diagnosis and mortality	100.0%	107.2%	.032
Female (%) Vital signs	44.6	39.0	b.001		Hospital LOS Adjusted for diagnosis	3.7	3.9	.086
HR b60 or N 100 beats/min (%)	25.9	22.9	.012		Adjusted for diagnosis and mortality	3.5	3.6	.095
SBP b90 mm Hg (%)	2.1	1.8	.458
Injury to level 1 ED interval (h)	5.0	5.4	.001

Non-ACO transfer P

(n = 6553)

Cost of hospitalization

ACO Non-ACO P

Injury severity score Outcomes	12.3	12.0	.129
ICU LOS (d)	2.0	1.8	.22
In-hospital mortality (%)	5.6	4.6	.077

Abbreviations: HR, heart rate; SBP, systolic blood pressure.

cost and LOS, we also conducted analysis of variance adjusted for in- hospital mortality as well as diagnosis, with similar results in both cost and LOS (Table 3).

A total of 30 401 CT, MRI, and Ultrasound studies were ordered for the study population during their hospitalizations. Overall, there was signifi- cantly less advanced imaging performed on patients transferred from within the ACO (3.55 vs 4.02 advanced imaging studies per patient hospi- talization, P = .003). This trend was true across all imaging modalities; there were fewer CT studies (3.00 vs 3.37, P b .001), MRI studies (0.45 vs 0.52, P = .025), and ultrasound studies (0.11 vs 0.13, P = .02) per- formed on ACO patients compared with non-ACO patients (Table 4).

This study has a number of limitations, primarily its retrospective nature. In addition, although we were able to control for in-hospital mortality and diagnosis, we did not control for age and sex. In addition, the integrated health care system studied is currently a pioneer ACO; it was only designated as one during the end of the study period. However, the cost-saving strategies seen as benefits of ACOs-including a shared electronic longitudinal medical record, image and laboratory transfer between institutions, facilitated secure communication between providers, and reliable and sophisticated systemwide performance measurement-were present in the system during the entire time of the study. Finally, we were unable to gather costs accumulated in the ACO and non-ACO hospitals prior to transfer, and so our conclusions can only be considered valid for the costs incurred at the destination, rather than transferring, hospitals.

These results support those of prior studies that have demonstrated potential savings of integrated Health care delivery systems. Evaluation of an ACO prototype in 2010 demonstrated that 5 physician group prac- tices were able to save more than $38 million by adopting ACO-like organizational structures and care management programs [2]. A study of inpatient surgical procedures in an integrated system found that payments for hip replacements were lower because of lower costs for postdischarge care [3]. A recent review of emergency department imaging use after implementation of a health information exchange demonstrated a reduction in redundant imaging [10]. To our knowl- edge, our findings are the first to demonstrate that these benefits are achievable by an ACO while managing patients with acute trauma.

There are likely a number of potential areas of cost savings for these patients within an ACO-like system, with these results likely being a direct result of information and image sharing between hospitals in our pioneer ACO. Patients with trauma routinely require imaging, and these imaging studies are often repeated after transfer [9,11]. However, prior research has demonstrated that, if the imaging studies themselves are available for viewing, they are significantly less likely to be repeated [12]. Similarly, the study noted above demonstrated that shared health information across a health information exchange could reduce repeat imaging [10]. This availability of imaging, via a shared electronic longi- tudinal medical record, may have led to the decreased use of imaging in our study, as well as a portion of the decreased costs.

It should be noted that the potential cost savings associated with the formation of ACOs might be even greater than demonstrated herein. The formation of ACOs may lead to standardization of practice and decreased variation in the use of resources [13]; there is a significant

Table 4

Use of advanced imaging among transfer trauma patients

Discussion

CT

This study demonstrates that patients with traumatic injuries who were transferred to a level 1 trauma center from a referring institution within the same ACO-like system had significantly lower total costs of hospitalization and overall imaging use than did patients transferred from outside the system.

Head CT	1.43	1.27	b.001
Max/face CT	0.09	0.14	b.001
C-spine CT	0.35	0.38	.03
Neck CT	0.09	0.11	.005
Chest CT	0.27	0.40	b.001
Abdominal/Pelvis CT	0.15	0.22	b.001
Abdominal CT	0.14	0.17	.0017
Pelvis CT	0.19	0.24	b.001
T/L/S-spine CT	0.13	0.23	b.001
Extremity CT	0.16	0.21	b.001
CT total MRI	3.00	3.37	b 0.001

ACO

(tests per patient)

Non-ACO P

(tests per patient)

Table 2

Injury profile for transfer trauma patients

Diagnosis ACO Non-ACO P

Fracture of skull	10.1%	9.3%	.361	Head MRI	0.19	0.19	.661
Fracture of neck and trunk	10.8%	14.0%	.003	C-spine MRI	0.11	0.12	.34
Fracture of limb	27.7%	25.5%	.116	Neck MRI	0.04	0.04	.971
Dislocation/sprain/strain	0.9%	1.5%	.114	Abdominal MRI	0.01	0.01	.294
Intracranial injury	27.6%	18.3%	b.001	Pelvis MRI	0.01	0.01	.769
Internal Injury of the thorax, abdomen, and pelvis	3.1%	4.6%	.026	T/L/S-spine MRI	0.06	0.11	b.001
Open wound of the head, neck, and trunk and limb	2.9%	6.3%	b.001	Extremity MRI	0.02	0.04	.018
Injury to blood vessels	0.4%	0.6%	.551	MRI total	0.45	0.52	.025
Superficial injury/contusion	2.5%	2.3%	.631 Ultrasound
Burns	0.5%	2.2%	b.001	Abdominal US	0.05	0.06	.14
Injury to nerves and spinal cord	0.6%	1.4%	.033	Lower Extremity US	0.06	0.08	.056
Certain traumatic complications and unspecified injuries	3.1%	2.8%	.655	Ultrasound total	0.11	0.13	.019
Unclassified	9.6%	11.3%	.103	Overall total	3.55	4.02	0.003

458 B.C. Geyer et al. / American Journal of Emergency Medicine 34 (2016) 455-458

variation in the use of blood transfusion during surgery and head CT in pa- tients with trauma [14,15], and ACO-level protocols may decrease this. In addition, the formation of ACOs may allow health care systems to focus on more than just the acute care of patients with trauma. By focusing on population-level care, the systems may allow for an increased focus on so- cial services and injury prevention [16]. Prior studies have demonstrated the benefit of injury prevention programs-interventions promoting hel- met use and discouraging alcohol consumption have already been proven to decrease head injuries and trauma recidivism, respectively [17,18].

In summary, the transfer of patients with trauma within an ACO-like health care system was associated with significantly lower cost of hospitalization when compared with contemporaneous patients trans- ferred from other hospitals. This may have been partially due to the sig- nificantly fewer imaging studies performed on patients transferred from within the system. Further research is needed to determine other factors that may have led to these differences in cost.

References

1. McClellan M, McKethan AN, Lewis JL, Roski J, Fisher ES. A national strategy to put accountable care into practice. Health Aff 2010;29:982-90. http://dx.doi.org/10. 1377/hlthaff.2010.0194.
2. Iglehart JK. Assessing an ACO prototype-Medicare’s physician group practice demon- stration. N Engl J Med 2011;364:198-200. http://dx.doi.org/10.1056/NEJMp1013896.
3. Miller DC, Ye Z, Gust C, Birkmeyer JD. Anticipating the effects of accountable care orga- nizations for inpatient surgery. JAMA Surg 2013;148:549-54. http://dx.doi.org/10. 1001/jamasurg.2013.1699.
4. Nyweide DJ, Lee W, Cuerdon TT, Pham HH, Cox M, Rajkumar R, et al. Association of Pioneer Accountable Care Organizations vs traditional Medicare fee for service with spending, utilization, and patient experience. JAMA 2015;313:2152-61. http://dx. doi.org/10.1001/jama.2015.4930.
5. Nathens AB, Jurkovich GJ, Cummings P, Rivara FP, Maier RV. The effect of organized sys- tems of trauma care on motor vehicle crash mortality. JAMA 2000;283:1990-4. http:// dx.doi.org/10.1001/jama.283.15.1990.
6. Willenberg L, Curtis K, Taylor C, Jan S, Glass P, Myburgh J. The variation of acute treatment costs of trauma in high-income countries. BMC Health Services Research 2012;12:267. http://dx.doi.org/10.1186/1472-6963-12-267.
7. Korley FK, Pham JC, Kirsch TD. Use of advanced radiology during visits to US emer- gency departments for injury-related conditions, 1998-2007. JAMA 2010;304: 1465-71. http://dx.doi.org/10.1001/jama.2010.1408.
8. Emick DM, Carey TS, Charles AG, Shapiro ML. Repeat imaging in trauma transfers: a retrospective analysis of computed tomography scans repeated upon arrival to a level I trauma center. J Trauma Acute Care Surg 2012;72:1255-62. http://dx.doi. org/10.1097/TA.0b013e3182452b6f.
9. Sung JC, Sodickson A, Ledbetter S. Outside CT imaging among emergency depart- ment transfer patients. J Am Coll Radiol 2009;6:626-32. http://dx.doi.org/10.1016/ j.jacr.2009.04.010.
10. Lammers EJ, Adler-Milstein J, Kocher KE. Does health information exchange reduce redundant imaging? Evidence from emergency departments. Med Care 2014;52: 227-34. http://dx.doi.org/10.1097/MLR.0000000000000067.
11. Moore HB, Loomis SB, Destigter KK, Mann-Gow T, Dorf L, Streeter MH, et al. Airway, breathing, computed tomographic scanning: duplicate computed tomographic imaging after transfer to trauma center. J Trauma Acute Care Surg 2013;74:813-7. http://dx.doi.org/10.1097/TA.0b013e3182789399.
12. Sodickson A, Opraseuth J, Ledbetter S. Outside imaging in emergency department transfer patients: CD import reduces rates of subsequent imaging utilization. Radiology 2011;260:408-13. http://dx.doi.org/10.1148/radiol.11101956.
13. Schuur JD, Carney DP, Lyn ET, Raja AS, Michael JA, Ross NG, et al. A top-five list for emergency medicine: a pilot project to improve the value of emergency care. JAMA Intern Med 2014. http://dx.doi.org/10.1001/jamainternmed.2013.12688.
14. Bennett-Guerrero E, Zhao Y, O’Brien SM, Ferguson Jr TB, Peterson ED, Gammie JS, et al. Variation in use of blood transfusion in Coronary artery bypass graft surgery. JAMA 2010;304:1568-75. http://dx.doi.org/10.1001/jama.2010.1406.
15. Andruchow JE, Raja AS, Prevedello LM, Zane RD, Khorasani R. Variation in head com- puted tomography use for emergency department trauma patients and physician risk tolerance. Arch Intern Med 2012;172:660-1. http://dx.doi.org/10.1001/ archinternmed.2011.2243.
16. Noble DJ, Casalino LP. Can accountable care organizations improve population health?: Should they try? JAMA 2013;309:1119-20. http://dx.doi.org/10.1001/jama.2013.592.
17. Mock CN, Maier RV, Boyle E, Pilcher S, Rivara FP. Injury Prevention strategies to pro- mote helmet use decrease Severe head injuries at a level I trauma center. J Trauma 1995;39:29-33 [discussion 34-5].
18. Gentilello LM, Rivara FP, Donovan DM, Jurkovich GJ, Daranciang E, Dunn CW, et al. Alcohol interventions in a trauma center as a means of reducing the risk of injury re- currence. Ann Surg 1999;230:473.