Upcoding Emergency Admissions for Non-Life-Threatening Injuries to Children

Objectives:

To assess the influence of investorowned for-profit (IOFP) status on upcoding pediatric inpatient admissions for inconsequential injuries as emergency when urgent or elective would be more suitable.

Study Design:

Using Florida inpatient discharge data for children 15 years and younger during 2001 to 2010, we examined injuries originating from the emergency departments (EDs) resultingin 1 overnight stay. Only non—life-threatening injuries were included. We assessed the probability of emergency categorization (vs urgent/elective) of admissions at IOFP hospitals compared with other types of hospitals (public, not for profit).

Methods:

Logistic regression was used to explore the probability that hospital admission following non—life-threatening injury to a child was classified as an emergency on the billing claim. The model controlled for age, race, sex, Hispanic ethnicity, trauma center status, insurance type and status, number of injuries, and market competition conditions.

Results:

For those patients satisfying the inclusion criteria (n = 8694), about 68% of the time hospitals classified the admissions as emergent. The model provides strong statistical evidence that IOFP hospitals had a higher probability (odds ratio = 1.1) of reporting emergency priorities for children admitted to the hospital from the ED, holding all other variables constant.

Conclusions:

Upcoding by IOFP hospitals may be a consequence of payer payment practices, utilization management policies, and local market dynamics. Florida Medicaid regulators and managed care organizations should examine their policies to identify inefficiencies associated with pediatric patients admitted for non—life-threatening injuries.

Am J Manag Care. 2013;19(11):917-924For-profit status was found to influence the probability of upcoding for inpatient cases involving non—life-threatening injuries with implications for Medicaid and other insurers.

• Payer payment practices and policies affect the probability of upcoding.

• Differences based on payer type (commercial vs public) point to the influence of utilization management policies that address the potential for upcoding.

• Florida Medicaid regulators and managed care organizations should examine their policies to identify inefficiencies associated with pediatric patients admitted for non—lifethreatening injuries.

Improper billing aimed at maximizing revenue, whether intentional or not, creates inefficiencies that could be identified and prevented. This study examines a specific type of upcoding related to classification of injured pediatric patients upon inpatient admission following emergency department (ED) evaluation. This study asked how the type of hospital ownership affects admission categorization (emergent, urgent, elective) for non—life-threatening injuries for children admitted to Florida hospitals from the ED from 2001 to 2010. We hypothesized the presence of a statistically significant greater probability of emergency categorization of inpatient admissions (vs urgent/elective) at investor-owned, for-profit (IOFP) facilities versus other types of hospitals (public, not for profit [NFP]).

Cost-control regulations and management practices have addressed wasteful healthcare provider billing practices to some extent, but opportunities for improvement remain.^1,2 Providers retain considerable leeway in preparing claims from medical charts.³ For example, ambiguity in payment policies may lead to inpatient admissions for inconsequential injuries being categorized as “emergency” when “urgent” or “elective” would be more suitable to the patient’s condition. This type of upcoding is similar to substituting patient Diagnosis-Related Groups that qualify for higher rates of reimbursement without any associated change in illness severity or treatment intensity, a practice that has been studied by Silverman and Skinner³ and Danfy.⁴ Categorizing minor injuries asemergency for hospital admissions from the ED has 2 potential financial impacts on the hospital in case of Medicaid-covered patients.⁵ First, concerning all age groups, the emergency classification is necessary to circumvent prior authorization requirements, therefore reducing administrative costs and consequently increasing residual revenue. The second reason applies to the adult population and concerns lifting the $1500 cap associated with procedures performed in the ED before inpatient admission. However, there are alternative relevant criteria that must be met for the cap to be lifted in association with a surgical procedure, one of which is classification as an emergency. While the latter reason does not affect pediatric patientsdirectly, it may impact general institutional procedures when classifying patients’ conditions as either urgent or emergent.

Prior analysis of the Florida Agency for Health Care Administration (AHCA) inpatient discharge data analysis revealed that injury, poisonings, toxic effects, and other external cause diagnoses (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 800-999) were the second-most common reason for hospitalization in the pediatric population. (The most common reason for admission in the inpatient pediatric population concerned diagnoses related to the respiratory system.) We selected injury for 3 reasons. First, the severity measure used in the study, the International Classification Injury Severity Score (ICISS), is well established and available for the study population for the past 20 years. Next, the injury diagnoses we used lent themselves more easily to the notion of non—life-threatening admissions. While the ICISS main criterion, mortality, could have been determined for other types of admissions (eg, respiratory diagnoses), it was necessary to account for severity for those hospitalizations thatwere not associated with mortality. The count of injuries has been documented in the literature as a valid proxy for apparent severity.⁶ Finally, it was deemed important to distinguish among hospitals’ treatment capabilities. The relevant experience and specialization in treating injury diagnoses are accounted for through the Florida trauma center (TC) certification variable in the model.

Inconsequential injuries resulting in admissions are highly prevalent.6 In other words, many of these admissions, while credible, do not result in death for children and may not require emergency prioritization upon admission. Despite their relatively mild conditions, the hospitals claimed that the patients required “immediate medical intervention as a result of severe, life-threatening or potentially disabling condition.”⁷

Investor-owned, for-profit hospitals have been found to be more likely to maximize profits8 and increase revenues through billing processes than their NFP counterparts, a practice perhaps related to IOFP hospitals’ abilities to implement more effective managerial and financial strategies.⁹ In addition, the studies that specifically addressed upcoding billing practices found that for-profit hospitals are willing to risk regulatory investigation and subsequent reputation damages in favor of profit obtained by upcoding.^3,4 These findings are consistent with the altruism model of hospital behavior theorized by Newhouse.¹⁰ Pressures on IOFP management, including investors’ expectation for return on investment and the requirement to pay taxes¹¹ ordifferences in the incentives provided to managers,¹² may be at the core of differences in billing behaviors.

STUDY DESIGN

We analyzed the inpatient discharge data set from the Florida AHCA for 2001 to 2010. The data contained patient information related to diagnosis (primary and up to 30 other diagnoses) and demographics (age, sex, race, Hispanic ethnicity). Each observation includes a hospital identifier, which was used to derive facility-specific information such as hospital ownership type (IOFP, NFP, or public) and TC certification status. Finally, the inpatient data set also indicates the payer type (eg, commercial, Medicaid, KidCare, or uninsured).

The outcome variable of interest was defined as dichotomous and indicated the type/priority assigned to the admission. Value options for this variable included emergency (the patient requires immediate medical intervention as a result of a severe, life-threatening, or potentially disabling condition) versus urgent (the patient requires attention for the care and treatment of a physical or mental disorder) or elective (the patient’s condition permits adequate time to schedule the availability of a suitable accommodation).⁷ Newborn admissions were excluded. These admission priority codes are entered by the hospitals on the billing claim (Inpatient Uniform Billing-04, form locator-14).

The study population was defined as children aged 0 to 15 years who were admitted to the hospital through the ED with a primary diagnoses in the ICD-9-CM code range of 800 to 959, excluding late effects of injuries, poisonings, and toxic effects, and effects of foreign bodies entering through an orifice. In addition, the population was limited to those children with non—life-threatening injuries. To define non–life-threatening injuries, this study used the ICISS.^6,13,14 This technique measures the proportion of patients who survive after admission with a specific combination of ICD-9-CM codes, and only those diagnoses in which all patients survived (ICISS = 1) were included. The ICISS values were calculated using survival risk ratios dating back to 1991, indicating at least a 10-year history of zero mortality associated with the injuries. The inclusion criterion of an ICISS of 1 accounts for the observed risk associated with the injuries. To eliminate bias from unobserved characteristics that might indicate higher injury severity not captured by the ICISS methodology, all inpatient episodes lasting longer than a single overnight stay were also omitted from the analysis. Based on these inclusion and exclusion criteria, the data set contained 8694 pediatric patients in Florida categorized as emergent or urgent/elective upon hospital admission from the ED.

METHODS

Given the dichotomous nature of the dependent variable, a logistic regression was used to predict the probability that a patient admitted to the hospital from the ED with non—life-threatening injuries would be classified as emergency as opposed to urgent or elective. Model variables may be conceptually divided into 5 categories: injury type and count (fractures, skull and spinal cord injuries [SSCIs], traumatic brain injury [TBI], vascular injury, thorax injury, and burns); hospital types (ownership, trauma certification); payer types and status (uninsured, commercial, Medicaid, KidCare);patient geography and demographics (distance to hospital, age, race, sex, and Hispanic ethnicity); and market structure (hospital and health maintenance organization [HMO] concentration).

The study population included pediatric patients admitted to the hospital from the ED due to injury, but with an ICISS of 1, indicating zero associated mortality. Therefore, the measure of severity used in our analysis contained no variation. Nonetheless, variation in severity not measured by the ICISS value was expected to influence the probability of classifying patients as emergent as opposed to urgent or elective. To capture the relative severity of study injuries not associated with mortality, the model included the number of individual injuries, as defined above. A positive relationship between emergency status and injury count was hypothesized. The remaining criteria, specifying the types of injury, indicate whether the patient was admitted with TBI, SSCI (other than TBI), a fracture (other than TBI or SSCI), a vascular injury, an injury to the thorax, or a burn. The International Classification of Diseases, Ninth Revision, Clinical Modification ranges for these injuries are discussed in more detail by Pracht and colleagues.⁶ Patients admitted with TBI were used as the control group to examine the impact of injury type.

Also, during the study period, the Florida Department of Health designated 22 hospitals as TCs, including 2 pediatriconly TCs. There is strong statistical evidence of improved mortality for children with severe injuries when treatment is provided in designated TCs versus nontrauma centers.⁶ However, the advantage associated with Florida TCs that treat children with non—life-threatening injuries is not well understood. We included TC status in the model to control for the greater experience, both pertaining to institutional operation and to potential medical expertise and proficiency in recognizing potential mortality risks. These qualities may allow TCs to more accurately assess severity and, by extension, reduce the probability of admitting patients for non–life-threatening injuries. Because 20 of the 22 TCs are NFP hospitals, the inclusion of this variable was important to keep its hypothesized influence from incorrectly getting attributed to NFP hospitals in general, therefore biasing the IOFP treatment variable in the model.

A study by Selassie and colleagues¹⁵ addressed the likelihood of admission to the hospital among adult patients, and their finding suggested that among mildly injured patients, African American females were significantly less likely to be admitted compared with white females and that females were less likely to be admitted than males. In addition, the uninsured were less likely to be admitted than those with private insurance.¹⁶ Also, Csikesz and colleagues¹⁷ found that patients with low socioeconomic status were 2.5 times more likely than patients with high socioeconomic status to present to the ED emergently. These studies identify the existence of an "admission practice bias.”¹⁸ To control for this systematic influence of patient socioeconomic and demographic characteristics, the confounding factors of race, sex, Hispanic ethnicity, and payer type were statistically controlled for in the analysis.

In addition to patient demographics, a variable indicating the linear distance from the patient’s residence to the hospital was included in the model. The distance was calculated from the geographic center of the patient’s residence zip code to the hospital. The distance was included in the model to reflect the time cost to the patient associated with travel to the hospital. The greater the distance to the hospital, the more reluctant a patient may be to leave the hospital in case of uncertainty associated with the severity and risk involved with the injuries. To the extent patients (or their guardians) influence the probability of inpatient admission, distance is hypothesized to influence the process. In such cases, the admission is less likely to reflect true emergency, but rather indicates caution. Therefore, increased distance is hypothesized to have a negative relationship to an admission being categorized as an emergency.

Table 1

The distance variable also incorporated the initial hospital selection decision of the patient. Distance is expected to have an important influence on a patient’s selection of hospital, as shown by Pracht and colleagues.⁶ To determine whether distance to types of hospitals could bias the results, the distribution of the distance from the patient’s residence to the hospital was examined (). There was virtually no difference in the distribution based on hospital type. Thus, there appears to be no direct evidence of preferential selection of any hospital type. Furthermore, it is noteworthy that in instances of trauma the Florida’s emergency medical services’ protocol dictates that patients be transported to the nearest ED.¹⁹

Finally, the Herfindahl-Hirschman Index (HHI) in the model accounts for the potential influence of the countylevel market dynamics. The number of insurers in a market may influence hospital behavior. In part, the findings of Zwanziger and colleagues²⁰ and Feldman and colleagues²¹ suggest that for-profit hospitals may be less willing to contract with insurers due to operational constraints, including claim payment policies. The HMO HHI was calculated from Florida HMO Enrollment Reports for 2001 to 2005—the only years available. Data for the latter half of the analysis period were based on the 2005 values to maintain cross-county variation. Any significant changes in the numberof HMOs operating in specific counties from 2006 to 2010 would bias the results; however, in the absence of any media reports of HMOs exiting or entering the state, it is reasonable to believe the HHI did not change consequentially. The hospital services HHI was calculated from the AHCA data using the market share in terms of inpatient bed days at the county level. Hospitals with a high degree of monopoly power, holding everything else constant, are hypothesized to be more capable of exploiting payment policy ambiguity to their benefit, including billing practices. For-profit hospitals and NFPs have been shown to exert market power in equal measure.²²

RESULTS

The number of pediatric patients aged 0 to 15 years admitted to Florida hospitals from EDs with non—lifethreatening injuries included in the overall analysis was 8694. Table 1 summarizes the statistics pertaining to the model and presents the proportions associated with emergency admission for all hospitals as well as IOFP hospitals. In Florida, IOFP institutions make up about 50% of the hospital market. Among the 3128 pediatric patients admitted to IOFP inpatient facilities, almost 70% were categorized as emergencies, whereas NFP/public hospitals (n = 5566 pediatric patients) categorized 67% as emergencies. Males made up a larger proportion of the emergency admissions than females—about 65% in both hospital ownership types. As expected, those racially categorized as white represented the largest group of study subjects, but IOFP hospitals had a lower proportion than NFP hospitals (54% vs 59%). In contrast, Hispanics represented a higher proportion of admissions at IOFP hospitals compared with NFP/public hospitals (22% vs 17%). The payer distribution was relatively similar between IOFP and NFP hospitals; Medicaid accounted for approximately 38% to 39%, and the uninsured made up almost 7% in both categories.

Table 2

The results of the logistic regression are shown in . The model provides evidence that IOFP hospitals had a statistically significant higher probability of billing non—life-threatening pediatric inpatient admissions from the ED as emergencies (P = .043), after controlling for the confounding influence of age, sex, race, ethnicity, TC status, injury type, severity, and market structure variables. Compared with patients with TBI, those admitted with other fractures had a lower probability of being classified as emergencies. The SSCI variable also had a negative coefficient, but was not statistically significant at the P = .05 level. Similarly, the other injury type variables were not statistically significant. The number of injuries, which provides a measure of severity, was highly significantly associated with a higher probability of classification as emergent. Trauma center status was associated with a significantly lower probability of classifying patients with non—life-threatening injuries as emergencies.

The payer variables indicate that compared with commercially insured patients, the uninsured, nonpayment, and Medicaid patients all had a statistically significant higher probability of classification as emergencies. (Nonpayment is defined as charity, professional courtesy, no charge, research/clinical trial, refusal to pay/bad debt, Hill-Burton free care, or research/donor that is known at the time of reporting.) Female patients were less likely to be classified as emergent, while age had no statistical influence on the dependent variable. Children categorized as white (P <.001) were less likely to be classified as emergencies, while children of Hispanic ethnicity (P = .007) were found to have a statistically significant higher probability of being classified as emergencies, holding all other variables fixed.

As expected, higher hospital HHI (less competition among hospitals) was associated with an increased probability of patients being classified as emergencies (P <.001), holding all other variables fixed. Conversely, higher HMO HHI (less competition among insurers) was associated with a lower probability of patients being classified as emergencies (P = .008), holding all other variables fixed.

CONCLUSIONS

Identifying and controlling inaccurate healthcare provider billing practices are important goals for payers. Differences in billing behaviors between IOFP and other types of hospitals are well established,^3,4,9 yet little research has focused on priority admission billing for non—life-threatening injuries. Given that hospital admissions from the ED are the most profitable type,²³ the practice may result in significant unnecessary expense to payers.

A logistic regression model was used to predict the probability that a pediatric patient with non—life-threatening injuries admitted to the hospital from the ED would be categorized as an emergency (vs urgent/elective). This study provides compelling evidence that IOFP hospitals have a significantly higher probability of classifying pediatric patients with non–life-threatening injuries admitted to the hospital from the ED as emergencies, after controlling for the influence of patient demographics, injury type and severity, and market structure.

The statistical evidence provided here suggests that the increased probability of billing for emergency admissions from the ED among mildly injured children by IOFP hospitals does not stem from patient mix, payer mix, or geographical variances in practice patterns. Upcoding of emergency admissions by IOFP hospitals may be a consequence of the pressures to produce return on investment to private shareholders, the requirement to pay taxes on profits, or administration incentives. In reaction to these performance demands, IOFP hospitals may take advantage of payment policy loopholes of some healthcare payers. Alternatively, it is possible that the results reflect systematic downcoding by NFP and public hospitals. While it was not possible to test whether this was the case, it is not likely for at least 2 reasons. First, the subset of hospitalizations used in the analysis, as described in the Study Design and Methods sections, was specifically limited to episodes not associated with a risk of mortality dating back at least to 1991. In contrast, AHCA specifically links an emergency designation with “life threatening or potentially disabling conditions.” Therefore, the episodes in the analysis are, holding everything else constant, presumed to be lower in priority than emergencies. Second, like their for-profit counterparts, NFP hospitals must to some extent at least be sensitive to their financial viability, again making this alternative explanation less likely.

Medicaid represents one of the most common payer types for pediatric patients categorized as emergency when admitted to the hospital through the ED for Florida hospitals. As a result, the Florida Medicaid payment policy may influence the billing behaviors of hospitals across all payer types. By coding claims for patients admitted to the hospital from the ED as emergencies, hospitals justify increased payment for ED services and are exempt from Medicaid’s prior authorization requirements, which are dictated by the Balanced Budget Act of 1997. For example, according to AHCA policy, patient admissions categorized as urgent or elective are ineligible for payment for ED screening services and must receive prior authorization for admission.⁵ It is noteworthy that uninsured and Medicaid patients may have less access to ambulatory care, making it possible that hospitals would be more likely to end up admitting them. However, such patients would likely be admitted “for observation,” which would be associated with a code of urgent or elective as opposed to emergency. If this dynamic was a significant factor, it would serve to render the findings more robust as it would logically decrease the proportion of uninsured and Medicaid patients with an emergency designation.

In addition, this study suggests that commercial insurer payment policies may reduce to some extent the reporting of emergency priorities for children admitted to the hospital from the ED, as patients with other payer types (Medicaid, KidCare, and uninsured) had statistically significant higher probabilities of emergency admission prioritization compared with commercially insured patients. The relatively lower probability of emergency admissions of children with commercial insurance may be the result of differences in their payment and prior authorization policies regarding admission priority codes. Regarding patients associated with commercial insurers, hospitals do not have the same incentives for categorizing inconsequential injuries as emergencies (eg, increased payment, guaranteed payment for the first night’s stay). The regulatory and statutory obligations that exempt prior authorization requirements for emergency cases under Medicaid are not present in the case of commercial insurance; therefore, there is increased risk of claim denial upon retrospective review by the latter (Susan Kohler, Senior Director of Ethics & Compliance, Centene Corporation, personal communication, December 29, 2012). It is also possible that other factors (eg, parents/guardians with commercial insurance who exert more control over their child’s health service utilization) lead to hospital admissions that may not otherwise occur.

The results also indicate that the uninsured are more likely to be classified as emergent after controlling for the influence of the other variables in the model. A potential explanation for this finding is that the uninsured may delay treatment for economic reasons, increasing the likelihood of increased severity when they eventually do visit the ED and are subsequently hospitalized. ²⁴ While the potential for this behavior could not be tested for the population used in this study, there is plenty of evidence for it in the overall US population.²⁵ It is noteworthy, however, that the percentages of the uninsured and commercially insured who were discharged after a single day were identical at 41%.

Limitations associated with this study may have implications for general inferences. First, the differences in hospital markets and payment policies make it difficult to generalize the results beyond Florida. Second, because the study population was limited to children aged 0 to 15 years, conclusions regarding hospital billing practice for adults should be made with caution. Finally, this study examined non—life-threatening injuries exclusively, so generalizations from this study regarding billing practices for more severe trauma should be restrained.

The objective of this study was to analyze the likelihood of an emergency classification for mildly injured pediatric patients. The conclusion was that for-profit hospitals are more likely to use emergency versus urgent or elective categorization. The most likely reason for this systematic increase is financial, as it increases third-party reimbursement. Perhaps the more difficult question, which this study was not designed to answer, is how such systematic differences manifest themselves. A recent investigative report by CBS’s 60 Minutes²⁶ provided a plausible answer to this question. Interviews with more than 100 current and former employees, including physicians, revealed how one large for-profithospital chain in Florida “relentlessly pressured its doctors to admit more and more patients—regardless of medical need&mdash;in order to increase revenues.” While the report focused on admissions themselves and not the associated emergency classification, the incentives and mechanisms are the same. The report further uncovered computer software (since removed) designed to maximize revenue; the software automated much of the admission process, including ordering a battery of tests without physician input. Again, while the focus was not specifically on emergency classification of patients, it provides a clear potential explanation of the mechanism by which that too would differ by the hospitals’ ownership status.

Additional research questions on hospital billing practices may prove valuable, such as the influence of hospital occupancy rates, ED crowding, and staffing levels on the propensity to categorize non—life-threatening injuries as emergency admissions.Author Affiliations: From University of South Florida (EP, ZP), Health Policy and Management, Tampa FL.

Funding Source: None.

Author Disclosures: The authors (EP, ZP) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (ZP); acquisition of data (ZP, EP); analysis and interpretation of data (ZP); drafting of the manuscript (ZP); critical revision of the manuscript for important intellectual content(EP); statistical analysis (EP); and provision of study materials or patients(EP).

Address correspondence to: Etienne Pracht, PhD, University of South Florida, Health Policy and Management, 13201 Bruce B. Downs Blvd, MDC56, Tampa, FL 33612. E-mail: epracht@health.usf.edu.1. Bentley TG, Effros RM, Palar K, Keeler EB. Waste in the U.S. healthcare system: a conceptual framework. Milbank Q. 2008;86(4):629-659.

2. The Commonwealth Fund Commission on a High Performance Health System. Why Not the Best? Results from the National Scorecard on U.S. Health System Performance, 2008. http://www.commonwealthfund.org/Publications/Fund-Reports/2008/Jul/Why-Notthe-Best--Results-from-the-National-Scorecard-on-U-S--Health-System-Performance--2008.aspx. Published July 17, 2008. Accessed September 25, 2013.

3. Silverman E, Skinner J. Medicare upcoding and hospital ownership. J Health Econ. 2004;23(2):369-389.

4. Danfy LS. How do hospitals respond to price changes? Am Econ Rev. 2005;95(5):1525-1547.

5. Agency for Health Care Administration. Hospital Services Coverage and Limitations Handbook. Tallahassee, FL: Florida Department of State; December 2011.

6. Pracht EE, Tepas JJ 3rd, Celso BG, Langland-Orban B, Flint L. Survival advantage associated with treatment of injury at designated trauma centers: a bivariate probit model with instrumental variables. Med Care Res Rev. 2007;64(1):83-97.

7. Agency for Health Care Administration. Provider General Handbook. Tallahassee, FL: Florida Department of State; 2008.

8. Rotarius T, Trujillo AJ, Liberman A, Ramirez B. Not-for-profit versus for-profit health care providers—part II: comparing and contrasting their records. Health Care Manag (Frederick). 2006;25(1):12-25.

9. Sear AM. Comparison of efficiency and profitability of investorowned multihospital systems with non-for-profit hospitals. Health Care Manage Rev. 1991;16(2):31-37.

10. Newhouse J. Towards a theory of non-profit institutions: an economic model of a hospital. Am Econ Rev. 1970;60(1):64-74.

11. Rotarius T, Trujillo AJ, Liberman A, Ramirez B. Not-for-profit versus for-profit health care providers—part I: comparing and contrasting their records. Health Care Manag (Frederick). 2005;24(4):296-310.

12. Dafny L, Dranove D. Regulatory exploitation and management changes: upcoding in the hospital industry. J Law Econ. 2009;52(2): 223-250.

13. Meredith JW, Evans G, Kilgo PD, et al. A comparison of the abilitiesof nine scoring algorithms in predicting mortality. J Trauma. 2002;53(4): 621-628.

14. Osler TM, Rutledge RR, Deis J, Bedrick E. ICISS: an international classification of disease-based injury severity score. J Trauma. 1996;41(3): 380-388.

15. Selassie AW, McCarthy ML, Pickelsimer EE. The influence of insurance, race, and gender on emergency department disposition. Acad Emerg Med. 2003;10(11):1260-1270.

16. Newton MF, Keirns CC, Cunningham R, Hayward RA, Stanley R. Uninsured adults presenting to US emergency departments: assumptions vs data. JAMA. 2008;300(16):1914 -1924. socioeconomic status on presentation and treatment of diverticular disease. J Gastrointest Surg. 2009;13(11):1993-2002.

18. Miller MG, Miller LS, Fireman B, Black SB. Variation in practice for discretionary admissions: impact on estimates of quality of hospital care. JAMA. 1994;271(19):1493-1498.

19. Florida Department of Health. Department of Health Trauma Transport Protocols Manual. http://www.doh.state.fl.us/demo/Trauma/PDFs/TTP_Manual-December_2004-Updated_64J_10-1-08.pdf. Published December 2004. Accessed October 24, 2012.

20. Zwanziger J, Meirowitz A. Managed care and changing health care markets. In: Morrisey MA, ed. Strategic Factors in Hospital Selection for HMO And PPO Networks. Washington, DC: AEI Press; 1998: 1-23.

21. Feldman R, Chan HC, Kralewski J, Dowd B, Shapiro J. Effects of HMOs on the creation of competitive markets for hospital services. J Health Econ. 1990;9(2):207-222.

22. Gaynor M, Vogt WB. Competition Among Hospitals. National Bureau of Economic Research Working Paper No. 9471. http://www.nber.org/papers/w9471. Published February 2003. Accessed September 25, 2013.

23. Handel DA, John McConnell K. The financial impact of ambulance diversion on inpatient hospital revenues and profits. Acad Emerg Med. 2009;16(1):29-33.

24. Irvin CB, Fox JM, Smude B. Are there disparities in emergency care for uninsured, Medicaid, and privately insured patients? Acad Emerg Med. 2003;10(11):1271-1277.

25. Wilper AP, Woolhandler S, Lasser KE, McCormick D, Bor DH, Himmelstein DU. Health insurance and mortality in US adults. Am J Public Health. 2009;99(12):2289-2295.

26. Kroft S. Hospitals: the cost of admission. 60 Minutes [television news program script]. CBS. http://www.cbsnews.com/8301-18560_162- 57556670/hospitals-the-cost-of-admission/. Broadcast December 2, 2012. Accessed December 25, 2012.