American Journal of Emergency Medicine 32 (2014) 1215–1219
Contents lists available at ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Resource utilization and health care charges associated with the most frequent ED users Cory Ondler, DO a, G.G. Hegde, PhD b, Jestin N. Carlson, MD a, c,⁎ a b c
Department of Emergency Medicine, Saint Vincent Hospital, Erie, PA University of Pittsburgh Katz School of Business, Pittsburgh, PA Department of Emergency Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA
a r t i c l e
i n f o
Article history: Received 21 April 2014 Accepted 21 July 2014
a b s t r a c t Study objective: Emergency department (ED) visits have continued to rise, and frequent ED users account for up to 8% of all ED visits. Reducing visits by frequent ED users may be one way to help reduce health care costs. We hypothesize that frequent users have unique ED utilization patterns resulting in differences in health care charges. Methods: We conducted a retrospective review of electronic medical records from an urban community teaching hospital for the year 2012 comparing the top 108 frequent ED users (N12 visits/year) to a randomly selected group of 108 nonfrequent users (b4 visits/year). We compared demographic characteristics, distance lived from the hospital, medical and psychiatric history, substance abuse history, diagnostic testing, disposition, and amount charged to the patient for each visit. We compared data using χ2 for proportions and t test or Wilcoxon rank sum based on normality of the data. Results: The top 108 frequent ED users accounted for 1922 visits (2.9%), whereas the 108 nonfrequent users accounted for 150 visits (0.2%), in 2012 (all ED visits n = 65,398). Frequent users were more often unemployed, have public insurance, have mental health conditions, use tobacco, have a greater number of allergies to medications, and live closer to the hospital (P b .01). Disposition and median charge per visit did not differ between frequent and nonfrequent users ($1220 vs $1280). The total charges of the frequent ED users’ visits were $10,465,216.07 versus $1,012,610.21 for nonfrequent users. Conclusions: Frequent users have unique medical and social characteristics; however, disposition and visit charges did not differ from nonfrequent users. © 2014 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Background In recent years, utilization of emergency departments (EDs) in the United States has continued to increase. From 1991 to 2011, the number of visits to US EDs jumped from 88.5 million to 129.5 million— an increase of almost 46% [1]. As visits have risen, so has Emergency Medicine’s share of overall health care spending. In 2011, Americans spent approximately US $2.4 trillion on health care overall, with ED costs accounting for between 2% and 10% of the total expenditure—a range of US $47 to $240 billion [2]. Although roughly 20% of US adults visit an ED annually [3], there is a subgroup of patients deemed “frequent users” (N 4 ED visits/year) that account for up to 8% of all ED patients but contribute to 21% to 28% of all ED visits [4]. These users are suspected of costing hospitals millions of dollars while decreasing the efficiency of the ED [5]. A recent study in South Carolina concluded that frequent visitors to EDs ⁎ Corresponding author. Department of Medicine, Division of Emergency Medicine, Saint Vincent Health Center, 232 West 25th St, Erie, PA 16544. E-mail address: [email protected] (J.N. Carlson). http://dx.doi.org/10.1016/j.ajem.2014.07.013 0735-6757/© 2014 Elsevier Inc. All rights reserved.
account for 10% of the total costs [6]. Previous studies have helped to define frequent users, with many studies investigating the impact these users place on the resources of the facilities they visit [3–5,7,8]. However, there are few objective data comparing the resources utilized by frequent users and cost associated with their ED visits. 1.2. Objective We sought to determine the demographic characteristics, ED utilization, and differences in charges between frequent ED users and nonfrequent users. We hypothesize that frequent users utilize unique ED resources resulting in differences in charges. 2. Methods 2.1. Setting and selection of participants The ED at our urban, teaching hospital averages approximately 65,000 visits a year. We defined frequent users as those with at least 4 visits/year [7,9]. We identified those patients that visited the ED most frequently during 2012 (frequent users) through visit numbers associated with patient medical record numbers. We identified all
1216
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219
patients who had at least 4 ED visits to our facility in 2012. We then focused our efforts on those patients who had more than 12 ED visits per year, defined previously as heavy users of the ED (n = 108) [4]. We then randomly selected medical record numbers using randomly generated numbers from a list of all nonfrequent users (b4 visits/ year) that visited our ED in 2012 (n = 108). There were no limitations on age. 2.2. Study design We performed a retrospective review of electronic medical records from a community teaching hospital ED for the year 2012. The study was approved by the hospital’s Internal Review Board. The desired charts were initially pulled by a senior informatics analyst. Two research assistants (RAs) were trained on data abstraction from the medical records through several sessions with the investigators (CO and JNC), and each assistant analyzed half of the frequent and nonfrequent users’ charts. The RAs used a standardized statistical collection tool with standard definitions of the predetermined variables but were not blinded to the primary hypothesis. The RAs had data elements randomly selected and evaluated on a monthly basis by one of the investigators for consistency, accuracy, and reliability [10]. 2.3. Data elements and definitions There are variations on the term frequent users in the literature including heavy users, superusers, and highly frequent users based on various thresholds of ED visits/year [3,4,7,9]. A priori, we defined patients with more than 12 ED visits to our institution in 2012 as frequent users for the purposes of our study. We defined nonfrequent users as patients with less than 4 ED visits to our institution in 2012. We abstracted total number of visits, age, sex, race, marital status, distance lived from the hospital, insurance status, whether the patient identified a primary care provider (PCP), diagnostic testing, prescriptions received on discharge, disposition, method of arrival to the ED, and total amount charged to the patient for each visit from the facility. We also determined the distance the patient lived from the hospital by mapping the distance from the address provided upon registration to the hospital using a publicly available mapping program (www. google.com/maps). We abstracted medical and mental health conditions as coded in the medical record. Frequent users and highly frequent users (N 17 visits/year) often have a history of substance abuse [3,8,9]. We defined a history of substance abuse if alcohol abuse, tobacco use, or illegal drug use was coded in the chart. Alcohol abuse was defined as more than 2 alcoholic drinks daily or if the chart documented “alcohol abuse.” We also collected data on chief concern. Previous work has suggested that frequent users present to EDs for pain-related concerns with a history of substance abuse [3,8,9]. As a result, concerns were divided into 3 exclusive categories: pain, psychiatric, and other. We defined concerns as Pain as any chief concern that included the term pain (eg, chest pain, abdominal pain). We defined concerns as Psychiatric if the chief concern included any of the following terms: anxiety, suicide or suicidal, homicidal, psychiatric or psychiatric evaluation, medical clearance, panic attack, thoughts or racing thoughts, depressed or depression, hallucinations. We also included concerns that included terms such as overdose and ingestion if the documentation in the chart mentioned that the overdose was the result of a suicidal ideation or suicide attempt. Cases where the overdose or ingestion was not the result of a suicide attempt (eg, accidental ingestion by a pediatric patient) were not coded as Psychiatric. All other chief concerns were defined as Other. All other variables (age, sex, insurance, etc) were also predefined and abstracted by the RAs using standard definitions and a standardized data collection form [10].
2.4. Statistical analysis We present characteristics using descriptive statistics. We compared characteristics between frequent and nonfrequent users using χ 2 for proportions and t test or Wilcoxon rank sum for continuous variables based on normality of the data. We compared patient characteristics at the patient level. We compared visit-level data to account for the multiple visits of frequent users. We also compared the rate of resource utilization per number of visits between frequent users and nonfrequent users to account for clustering around individual patients and assess the impact of increasing number of visits on individual variables (ie, is resource utilization dependent on the number of visits/year?). We initially defined our most frequent users as those with more than 12 ED visits/year [4]. Since the inception of our study, Doupe et al [3] has suggested a standard definition of highly frequent users as those with more than 17 ED visits/year. We therefore ran a sensitivity analysis using the same parameters and variable definitions as above but with patients who had more than 17 visits/year. Data were compared using Stata v 12 (Stata Corp, College Station, TX).
3. Results During calendar year 2012, there were 65,398 visits to our ED. We identified 2960 frequent visitors (N 4 visits/year), accounting for 17,216 total ED visits (26.3% of all ED visits). The top 108 frequent users (N12 visits/year) included in this study accounted for 1922 visits (2.9% of all ED visits), whereas the 108 nonfrequent visitors accounted for 150 visits (0.2% of all ED visits). Frequent and nonfrequent ED users did not differ by age, sex, race, or access to a PCP (Table 1). Frequent users did live in closer proximity to the hospital (median miles [interquartile range {IQR}], 1.8 [0.75, 3.5] vs 3.35 [1.6, 6.5]; P b .001). Frequent users were more often insured with either Medicaid or Medicare (86% vs 55%), more often unemployed (15% vs 60%), and had psychiatric conditions more often than the nonfrequent users (62% vs 10%) (P b .01). Frequent users more often had significant medical history such as hypertension (46% vs 24 %, P = .001) and diabetes mellitus (27% vs 6%, P b .01) than nonfrequent users. Frequent users also had a greater substance abuse history compared to nonfrequent users including tobacco use (67% vs 22%, P b .01), alcohol abuse (8%, vs 2%, P = .03), and illicit drug use (16% vs 0%, P b .01). Frequent users had a total of 1922 visits with a median of 17 (IQR 15, 22) per patient, whereas the nonfrequent group had a total of 150 visits with a median of 2 (IQR 1, 2) (Table 2). Both groups had similar rates of chief concerns of pain and had laboratory studies performed at similar rates (41% vs 48%, P = .13). Frequent users had radiographs (42% vs 30%, P b .01) less frequently than nonfrequent users. Disposition was similar between frequent and nonfrequent users (admission, 16% vs 18%; P = .36), although frequent user visits were transported via emergency medical services (EMS) more often than those of nonfrequent users (19% vs 14%, P b .01). The median charge per visit (IQR) for a frequent visitor was $1220.10 ($490, $4179.04), with a total charge for all visits of $10,465,216.07 (Table 2). The distribution of charges for nonfrequent users (median charge [IQR] per visit, $1280.00 [$573, $4569]) did not differ (P = .97) but totaled $1,012,610.00. In our sensitivity analysis including only those highly frequent users of more than 17 ED visits/year, highly frequent ED users had similar demographics and patterns of ED utilization to nonfrequent ED users. This subset of frequent users had lower rates of radiographs than nonfrequent users but higher rates of EMS service. The ED admission rates did not differ between highly frequent users and nonfrequent users, nor did charges.
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219 Table 1 Patient demographics
Mean age, y (SD) Sex Male Female Race/ethnicity White Black Asian Other Hispanic Married (if age N17 y) PCP Insured None Medicare/Medicaid Private insurance Employed Distance Median number of allergies (IQR) NSAID allergies Substance abuse Tobacco Alcohol use Illicit drugs Health status CAD HTN DM COPD/asthma CHF Afib CRF CA Other Mental health status Any psychiatric history Anxiety Bipolar Depression Schizophrenia Fibromyalgia Other
1217
Table 2 Visit information Nonfrequent users (n = 108)
Frequent users (n = 108)
P value
37.6 (25.2)
42.6 (13.7)
.073 1
45% (49) 55% (59)
45% (49) 55% (59)
75% (81) 18% (19) b1% (1) 2% (2) 5% (5) 51% (40/78) 95% (103)
69% 26% 0% 0% 5% 58% 95%
(75) (28) (0) (0) (5) (63/108) (103)
5% (5) 55% (60) 40% (43) 60% (34/57) 3.35 (1.6, 6.5) 0 (0, 1) 6% (6)
5% 86% 9% 15% 1.8 2 41%
(5) (93) (10) (15/101) (0.75, 3.5) (1,4) (44)
22% (24) 2% (2/100) 0% (0)
67% (72) 8% (9) 16% (17)
b.01 .03 b.01
6% (7) 24% (26) 6% (7) 8% (9) 2% (2) 3% (3) 2% (2) 4% (4) 17% (18)
12% 46% 27% 21% 3% 3% 4% 9% 41%
(13) (50) (29) (23) (3) (3) (4) (10) (44)
.16 .01 b.01 .01 .75 1 .41 .1 b.01
10% (11) 3% (3) 3% (3) 6% (6) 0% (0) 2% (2) 2% (2)
62% 19% 44% 25% 10% 4% 5%
(67) (21) (48) (27) (11) (4) (5)
b.01 b.01 b.01 b.01 b.01 .41 .24
.29
.34 1 b.01
b.01 b.01 b.01 b.01
NSAID: nonsteroidal anti-inflammatory drug, CAD: coronary artery disease, HTN: hypertension, DM: diabetes mellitus, COPD: chronic obstructive pulmonary disease, CHF: congestive heart failure, Afib: atrial fibrillation, CRF: chronic renal failure, CA: cancer.
4. Limitations Our study has several limitations. First, we only evaluated 1 hospital ED in an urban environment. Locally, we have 4 hospitals within a 2.5-mile radius. Frequent users may utilize multiple health care resources; and without access to medical records at these nearby facilities, it is difficult to obtain the total number of ED visits annually for any 1 individual. However, the characteristics of our studied frequent ED users are similar to previously published work with regards to age, sex, and other demographics [4,5]. We feel that this gives our data face validity and that our data are representative of the broader population of frequent ED users and not unique to our institution. Second, there were more than 2900 patients that that had at least 4 ED visits to our ED in 2012. Multiple definitions of frequent ED users exist in the literature including more than 4, more than 12, and more than 17 visits/year [3–5]. Because of limitations in resources, we focused on those with the highest ED usage: more than 12 and more than 17 visits/year. The remainder of the frequent users may have demographics that differentiate them from the most frequent users. Those patients with more than 12 visits/year represent nearly 3% of our ED’s annual volume but have charges similar to nonfrequent users. We believe that this group represents the ideal population
Total number of visits (n = 2072) Median number of visits/patient (IQR) Charges Total charges in US dollars Median charges in US dollars/visit (IQR) Chief concern Pain Psychiatric Other (nonpsych or nonpain concern) Diagnostic testing Laboratories Radiograph ECG CT Ultrasound Venous Doppler V/Q MRI Disposition Discharged Admitted/transferred LWBS/eloped Left AMA Died in ED Narcotic Rx if discharged Arrival by EMS
Nonfrequent users
Frequent users
150
1922
2 (1, 2)
1,012,610.21 1280 (573, 4569)
P value
17 (15, 22)
10,465,216.07 1220.17 (490, 4179.04)
.97
57% (85) 4% (6) 47% (71)
59% (1130) 11% (205) 30% (582)
.6 .01 b.01
41% (62) 42% (63) 26% (39) 17% (25) 1% (2) 3% (4) b 1% (1) 0% (0)
48% (917) 30% (584) 19% (371) 11% (220) 1% (26) b 1% (12) 0% (0) b1% (6)
.13 b.01 .05 .06 .98 .01 b.01 .49
81% (122) 18% (27) b 1% (1) 0% (0) 0% (0) 22% (27/123) 14% (20/147)
80% (1529) 16% (317) 3% (66) b1% (9) b1% (1) 33% (535/1605) 19% (362/1916)
.36
.01 b.01
CC: chief concern, ECG: electrocardiogram, CT: computed tomography, V/Q: venous perfusion scan, MRI: magnetic resonance imaging, LWBS: left without being seen, AMA: left against medical advice, Rx: prescription.
where interventions to find alternative care pathways may have the greatest impact by reducing ED visits and reducing cost. We performed a retrospective chart review and, as a result, are unable to determine “why” patients chose to present to the ED as opposed to another health care provider/facility. As a result in limitations in support staff, we did not evaluate interrater reliability. The RAs had data elements randomly selected and evaluated on a monthly basis by one of the investigators for consistency and accuracy [10]. Given the relative paucity of data regarding charges associated with frequent ED users, we did not perform a sample size calculation. We performed a review of all patients with more than 12 visits to our ED in 2012, representing nearly 3% of our ED’s annual volume. As a result of the retrospective nature of our work, we were unable to control sample size. Although it would able to expand the definition of frequent users to include more patients, we chose to use definitions of frequent users in the existing literature. We chose to look at diagnostic testing as a marker of resource utilization. Although there may be other markers of resource utilization (medications administered in the ED, intravenous placement, etc), we chose to focus our efforts on those resources that we felt a priori help emergency providers evaluate for acute lifethreatening conditions. We used the amount charged to the patient as a representation of the cost of the visit. The actual cost of each visit is often challenging to determine. Although charges may vary by institution, we felt that this was the most consistent way to estimate the dollar amount associated with the care of our study population. Lastly, we performed a retrospective chart review. Both abstractors were trained to abstract these data similarly, used a standard data collection form, and had their work regularly reviewed by one of the investigators to ensure correct abstraction. Although we have attempted to standardize the data collection, much of these data are self-reported by the patients—such as substance abuse or psychiatric
1218
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219
and medical histories. As nonfrequent users presented less than 4 times in the year, it could be that this information was not asked or simply assumed to be negative by the providers who saw the patient. Frequent users have more opportunities to be asked about this information and, as such, may have a disproportionately higher instance of these conditions being documented (eg, tobacco use, medical and psychiatric history). 5. Discussion Physicians, health care workers, and policy makers may have a preconceived notion of ED frequent visitors [5]. Previous work has detailed providers’ misconception including that they are more often uninsured minorities that utilize the ED for basic care or drug-seeking behavior [5,9]. However, others have shown that frequent ED users more often are white, are female, have a mean age of 40 years, and are covered by insurance [5]. The uninsured represent less than 15% of frequent users and are no more likely to be frequent users than those that use the ED occasionally [5]. There are, however, a disproportionately higher number of publicly insured (Medicaid/Medicare) patients that are frequent users (60%) compared to nonfrequent users (36%). In line with national data, the frequent users of our ED consisted predominantly of women (55%), had a mean age of 42.5 years, and were predominantly white (69%) [5]. The demographics of our frequent users parallel other studies; however, prior to our work, little was known regarding the charges related to these frequent ED visits. Although the median charge per visit was similar between frequent and nonfrequent users, the overall charges for frequent users were nearly 10 times those of nonfrequent users because of the sheer number of visits by frequent users. Finding alternative plans of care for these patients could help to reduce health care costs, although equally high number of patients in each group reported having a PCP (95%). Although we did not evaluate the specific reasons frequent users presented to the ED instead of their PCP, frequent users did live in closer proximity to the ED and their use may be related to the ED’s ease of access. Like their nonfrequent counterparts, the majority of frequent users reported that they had PCPs. Prior studies have reported that most frequent users prefer ED care for a variety of reasons [7,9]. Although some frequent users present to the ED because they feel that their PCP may be unable to adequately care for their issues, many utilized the ED due to its accessibility (provides care 24 hours a day and no need for an appointment) [5,7,9]. Frequent users are also believed to utilize the ED for treatment of chronic pain and desire narcotic pain medication—with estimates of drug-seeking patients accounting for nearly 25% of all ED visits [7]. Although variation in presenting concerns depends on the site of the study, with urban EDs having a higher percentage of psychiatric and substance abuse concerns, pain-related concerns may be seen in a variety of settings [5]. We did not note a difference in pain-related concerns between frequent and nonfrequent users. However, prescriptions for narcotic pain medications were provided at ED discharge more often to frequent users (33%) than nonfrequent users (22%). Although we are unable to determine why ED providers prescribed narcotic pain medications more frequently, it may be related to the allergies reported by frequent users. Frequent users reported allergies to nonsteroidal anti-inflammatory medications more often than nonfrequent users (44% vs 6%). Previous efforts aimed at reducing the prescription of narcotics to frequent users have been shown to markedly decrease ED usage in this population [7]. Mental health as well as substance abuse issues play important roles in frequent use of the ED [7–9]. One prior study showed that frequent ED use is defined by substance abuse more strongly than any other factor [3]. We found that frequent users abused tobacco, illicit drugs, and alcohol at higher rates then nonfrequent users. A recent study stated that frequent users had significant burden of disease and that behavioral health issues were only a small contributing factor
[11]. Our study supports the fact that many frequent users have many chronic medical issues but also psychiatric illness. We found that many frequent ED users often report mental health conditions including anxiety (19%), bipolar disorder (44%), depression (25%), and schizophrenia (10%). Nonfrequent users rarely reported such histories. Previous work has detailed geographic variation in psychiatric disease, which may explain the differences noted between our study population and that of others [12,13]. The relatively high percentage of frequent users reporting mental health conditions suggests that alternative care pathways targeting frequent users may benefit from input from mental health specialists. We did not notice a difference in mean charges between frequent and nonfrequent users. Both groups had similar diagnostic workups including blood work and usage of computed tomography. In addition, both groups were admitted to the hospital equally. Radiographs were performed less often in frequent users; however, these have a relatively lower cost associated with them than other diagnostics tests, such as computed tomography, which may be the larger drivers of the visit charges. Prior research has suggested that frequent users tend to have a higher utilization of other resources including higher rates of admission and are transported by ambulance more frequently [5,8]. Although we did note increased ambulance transport rate, we did not note any difference in admission rates between the 2 groups. Of the 1922 visits by frequent users, more than 1500 were not admitted to the hospital. One study reported that approximately 29% of frequent user visits were more appropriate for primary care where the cost of care is greatly reduced [14]. We identified that 2.9% of our ED visits were from our 108 frequent users with a median charge of $1220/ visit. Developing alternative care pathways for frequent users including directing them to a PCP in place of the ED could result in significant reductions in health care dollars. Using a median of $1220/ visit, reducing the number of frequent ED visits by the same 29% nationally could reduce health care expenditures by up to $1.35 billion annually [15]. Although our data detail the financial magnitude of frequent ED users, before fully implementing alternative care pathways, future work should further detail why frequent users present to the ED as opposed to other care sites. 6. Conclusion In this single-center study, frequent users have unique demographics and live closer to the hospital than nonfrequent users; however, their dispositions are similar. Although charges per visit did not differ between frequent and nonfrequent users, the sheer number of frequent user visits amounted to approximately 10 times the total amount. Interventions aimed at lowering frequent ED use must be tailored to certain demographics and needs but could significantly reduce health care costs. Acknowledgments The authors would like to thank Dr. Arvind Venkat for his assistance in proof-reading this manuscript. References [1] Avalere. US Census Bureau: national and state population estimates for community hospitals. http://www.census.gov/popest/data/state/totals/2011/ index.html; 2011. [Accessed July 2011, 2011]. [2] Lee MH, Schuur JD, Zink BJ. Owning t0068cbee cost of emergency medicine: beyond 2%. Ann Emerg Med 2013;62(5):498–505.e493. [3] Doupe MB, Palatnick W, Day S, et al. Frequent users of emergency departments: developing standard definitions and defining prominent risk factors. Ann Emerg Med 2012;60(1):24–32. [4] Blank FS, Li H, Henneman PL, et al. A descriptive study of heavy emergency department users at an academic emergency department reveals heavy ED users have better access to care than average users. J Emerg Nurs 2005;31(2):139–44.
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219 [5] LaCalle E, Rabin E. Frequent users of emergency departments: the myths, the data, and the policy implications. Ann Emerg Med 2010;56(1):42–8. [6] A Report on Frequent Users of Hospital Emergency Departments in South Carolina. http://imph.org/a-report-on-frequent-users-of-hospital-emergency-departmentsin-south-carolina/; 2011. [Accessed February 1, 2014]. [7] Grover CA, Close RJ, Villarreal K, Goldman LM. Emergency department frequent user: pilot study of intensive case management to reduce visits and computed tomography. West J Emerg Med 2010;11(4):336–43. [8] DiPietro BY, Kindermann D, Schenkel SM. Ill, itinerant, and insured: the top 20 users of emergency departments in Baltimore city. ScientificWorldJournal 2012;2012:726568. [9] Althaus F, Paroz S, Hugli O, et al. Effectiveness of interventions targeting frequent users of emergency departments: a systematic review. Ann Emerg Med 2011;58(1): 41–52.e42.
1219
[10] Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996; 27(3):305–8. [11] Billings J, Raven MC. Dispelling an urban legend: frequent emergency department users have substantial burden of disease. Health Aff (Millwood) 2013;32(12):2099–108. [12] Lewis G, Booth M. Regional differences in mental health in Great Britain. J Epidemiol Community Health 1992;46(6):608–11. [13] Bijl RV, de Graaf R, Hiripi E, et al. The prevalence of treated and untreated mental disorders in five countries. Health Aff (Millwood) 2003;22(3):122–33. [14] Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52(2):126–36. [15] Carlson JN, Menegazzi JJ, Callaway CW. Magnitude of national ED visits and resource utilization by the uninsured. Am J Emerg Med 2013;31(4):722–6.
Contents lists available at ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Resource utilization and health care charges associated with the most frequent ED users Cory Ondler, DO a, G.G. Hegde, PhD b, Jestin N. Carlson, MD a, c,⁎ a b c
Department of Emergency Medicine, Saint Vincent Hospital, Erie, PA University of Pittsburgh Katz School of Business, Pittsburgh, PA Department of Emergency Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA
a r t i c l e
i n f o
Article history: Received 21 April 2014 Accepted 21 July 2014
a b s t r a c t Study objective: Emergency department (ED) visits have continued to rise, and frequent ED users account for up to 8% of all ED visits. Reducing visits by frequent ED users may be one way to help reduce health care costs. We hypothesize that frequent users have unique ED utilization patterns resulting in differences in health care charges. Methods: We conducted a retrospective review of electronic medical records from an urban community teaching hospital for the year 2012 comparing the top 108 frequent ED users (N12 visits/year) to a randomly selected group of 108 nonfrequent users (b4 visits/year). We compared demographic characteristics, distance lived from the hospital, medical and psychiatric history, substance abuse history, diagnostic testing, disposition, and amount charged to the patient for each visit. We compared data using χ2 for proportions and t test or Wilcoxon rank sum based on normality of the data. Results: The top 108 frequent ED users accounted for 1922 visits (2.9%), whereas the 108 nonfrequent users accounted for 150 visits (0.2%), in 2012 (all ED visits n = 65,398). Frequent users were more often unemployed, have public insurance, have mental health conditions, use tobacco, have a greater number of allergies to medications, and live closer to the hospital (P b .01). Disposition and median charge per visit did not differ between frequent and nonfrequent users ($1220 vs $1280). The total charges of the frequent ED users’ visits were $10,465,216.07 versus $1,012,610.21 for nonfrequent users. Conclusions: Frequent users have unique medical and social characteristics; however, disposition and visit charges did not differ from nonfrequent users. © 2014 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Background In recent years, utilization of emergency departments (EDs) in the United States has continued to increase. From 1991 to 2011, the number of visits to US EDs jumped from 88.5 million to 129.5 million— an increase of almost 46% [1]. As visits have risen, so has Emergency Medicine’s share of overall health care spending. In 2011, Americans spent approximately US $2.4 trillion on health care overall, with ED costs accounting for between 2% and 10% of the total expenditure—a range of US $47 to $240 billion [2]. Although roughly 20% of US adults visit an ED annually [3], there is a subgroup of patients deemed “frequent users” (N 4 ED visits/year) that account for up to 8% of all ED patients but contribute to 21% to 28% of all ED visits [4]. These users are suspected of costing hospitals millions of dollars while decreasing the efficiency of the ED [5]. A recent study in South Carolina concluded that frequent visitors to EDs ⁎ Corresponding author. Department of Medicine, Division of Emergency Medicine, Saint Vincent Health Center, 232 West 25th St, Erie, PA 16544. E-mail address: [email protected] (J.N. Carlson). http://dx.doi.org/10.1016/j.ajem.2014.07.013 0735-6757/© 2014 Elsevier Inc. All rights reserved.
account for 10% of the total costs [6]. Previous studies have helped to define frequent users, with many studies investigating the impact these users place on the resources of the facilities they visit [3–5,7,8]. However, there are few objective data comparing the resources utilized by frequent users and cost associated with their ED visits. 1.2. Objective We sought to determine the demographic characteristics, ED utilization, and differences in charges between frequent ED users and nonfrequent users. We hypothesize that frequent users utilize unique ED resources resulting in differences in charges. 2. Methods 2.1. Setting and selection of participants The ED at our urban, teaching hospital averages approximately 65,000 visits a year. We defined frequent users as those with at least 4 visits/year [7,9]. We identified those patients that visited the ED most frequently during 2012 (frequent users) through visit numbers associated with patient medical record numbers. We identified all
1216
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219
patients who had at least 4 ED visits to our facility in 2012. We then focused our efforts on those patients who had more than 12 ED visits per year, defined previously as heavy users of the ED (n = 108) [4]. We then randomly selected medical record numbers using randomly generated numbers from a list of all nonfrequent users (b4 visits/ year) that visited our ED in 2012 (n = 108). There were no limitations on age. 2.2. Study design We performed a retrospective review of electronic medical records from a community teaching hospital ED for the year 2012. The study was approved by the hospital’s Internal Review Board. The desired charts were initially pulled by a senior informatics analyst. Two research assistants (RAs) were trained on data abstraction from the medical records through several sessions with the investigators (CO and JNC), and each assistant analyzed half of the frequent and nonfrequent users’ charts. The RAs used a standardized statistical collection tool with standard definitions of the predetermined variables but were not blinded to the primary hypothesis. The RAs had data elements randomly selected and evaluated on a monthly basis by one of the investigators for consistency, accuracy, and reliability [10]. 2.3. Data elements and definitions There are variations on the term frequent users in the literature including heavy users, superusers, and highly frequent users based on various thresholds of ED visits/year [3,4,7,9]. A priori, we defined patients with more than 12 ED visits to our institution in 2012 as frequent users for the purposes of our study. We defined nonfrequent users as patients with less than 4 ED visits to our institution in 2012. We abstracted total number of visits, age, sex, race, marital status, distance lived from the hospital, insurance status, whether the patient identified a primary care provider (PCP), diagnostic testing, prescriptions received on discharge, disposition, method of arrival to the ED, and total amount charged to the patient for each visit from the facility. We also determined the distance the patient lived from the hospital by mapping the distance from the address provided upon registration to the hospital using a publicly available mapping program (www. google.com/maps). We abstracted medical and mental health conditions as coded in the medical record. Frequent users and highly frequent users (N 17 visits/year) often have a history of substance abuse [3,8,9]. We defined a history of substance abuse if alcohol abuse, tobacco use, or illegal drug use was coded in the chart. Alcohol abuse was defined as more than 2 alcoholic drinks daily or if the chart documented “alcohol abuse.” We also collected data on chief concern. Previous work has suggested that frequent users present to EDs for pain-related concerns with a history of substance abuse [3,8,9]. As a result, concerns were divided into 3 exclusive categories: pain, psychiatric, and other. We defined concerns as Pain as any chief concern that included the term pain (eg, chest pain, abdominal pain). We defined concerns as Psychiatric if the chief concern included any of the following terms: anxiety, suicide or suicidal, homicidal, psychiatric or psychiatric evaluation, medical clearance, panic attack, thoughts or racing thoughts, depressed or depression, hallucinations. We also included concerns that included terms such as overdose and ingestion if the documentation in the chart mentioned that the overdose was the result of a suicidal ideation or suicide attempt. Cases where the overdose or ingestion was not the result of a suicide attempt (eg, accidental ingestion by a pediatric patient) were not coded as Psychiatric. All other chief concerns were defined as Other. All other variables (age, sex, insurance, etc) were also predefined and abstracted by the RAs using standard definitions and a standardized data collection form [10].
2.4. Statistical analysis We present characteristics using descriptive statistics. We compared characteristics between frequent and nonfrequent users using χ 2 for proportions and t test or Wilcoxon rank sum for continuous variables based on normality of the data. We compared patient characteristics at the patient level. We compared visit-level data to account for the multiple visits of frequent users. We also compared the rate of resource utilization per number of visits between frequent users and nonfrequent users to account for clustering around individual patients and assess the impact of increasing number of visits on individual variables (ie, is resource utilization dependent on the number of visits/year?). We initially defined our most frequent users as those with more than 12 ED visits/year [4]. Since the inception of our study, Doupe et al [3] has suggested a standard definition of highly frequent users as those with more than 17 ED visits/year. We therefore ran a sensitivity analysis using the same parameters and variable definitions as above but with patients who had more than 17 visits/year. Data were compared using Stata v 12 (Stata Corp, College Station, TX).
3. Results During calendar year 2012, there were 65,398 visits to our ED. We identified 2960 frequent visitors (N 4 visits/year), accounting for 17,216 total ED visits (26.3% of all ED visits). The top 108 frequent users (N12 visits/year) included in this study accounted for 1922 visits (2.9% of all ED visits), whereas the 108 nonfrequent visitors accounted for 150 visits (0.2% of all ED visits). Frequent and nonfrequent ED users did not differ by age, sex, race, or access to a PCP (Table 1). Frequent users did live in closer proximity to the hospital (median miles [interquartile range {IQR}], 1.8 [0.75, 3.5] vs 3.35 [1.6, 6.5]; P b .001). Frequent users were more often insured with either Medicaid or Medicare (86% vs 55%), more often unemployed (15% vs 60%), and had psychiatric conditions more often than the nonfrequent users (62% vs 10%) (P b .01). Frequent users more often had significant medical history such as hypertension (46% vs 24 %, P = .001) and diabetes mellitus (27% vs 6%, P b .01) than nonfrequent users. Frequent users also had a greater substance abuse history compared to nonfrequent users including tobacco use (67% vs 22%, P b .01), alcohol abuse (8%, vs 2%, P = .03), and illicit drug use (16% vs 0%, P b .01). Frequent users had a total of 1922 visits with a median of 17 (IQR 15, 22) per patient, whereas the nonfrequent group had a total of 150 visits with a median of 2 (IQR 1, 2) (Table 2). Both groups had similar rates of chief concerns of pain and had laboratory studies performed at similar rates (41% vs 48%, P = .13). Frequent users had radiographs (42% vs 30%, P b .01) less frequently than nonfrequent users. Disposition was similar between frequent and nonfrequent users (admission, 16% vs 18%; P = .36), although frequent user visits were transported via emergency medical services (EMS) more often than those of nonfrequent users (19% vs 14%, P b .01). The median charge per visit (IQR) for a frequent visitor was $1220.10 ($490, $4179.04), with a total charge for all visits of $10,465,216.07 (Table 2). The distribution of charges for nonfrequent users (median charge [IQR] per visit, $1280.00 [$573, $4569]) did not differ (P = .97) but totaled $1,012,610.00. In our sensitivity analysis including only those highly frequent users of more than 17 ED visits/year, highly frequent ED users had similar demographics and patterns of ED utilization to nonfrequent ED users. This subset of frequent users had lower rates of radiographs than nonfrequent users but higher rates of EMS service. The ED admission rates did not differ between highly frequent users and nonfrequent users, nor did charges.
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219 Table 1 Patient demographics
Mean age, y (SD) Sex Male Female Race/ethnicity White Black Asian Other Hispanic Married (if age N17 y) PCP Insured None Medicare/Medicaid Private insurance Employed Distance Median number of allergies (IQR) NSAID allergies Substance abuse Tobacco Alcohol use Illicit drugs Health status CAD HTN DM COPD/asthma CHF Afib CRF CA Other Mental health status Any psychiatric history Anxiety Bipolar Depression Schizophrenia Fibromyalgia Other
1217
Table 2 Visit information Nonfrequent users (n = 108)
Frequent users (n = 108)
P value
37.6 (25.2)
42.6 (13.7)
.073 1
45% (49) 55% (59)
45% (49) 55% (59)
75% (81) 18% (19) b1% (1) 2% (2) 5% (5) 51% (40/78) 95% (103)
69% 26% 0% 0% 5% 58% 95%
(75) (28) (0) (0) (5) (63/108) (103)
5% (5) 55% (60) 40% (43) 60% (34/57) 3.35 (1.6, 6.5) 0 (0, 1) 6% (6)
5% 86% 9% 15% 1.8 2 41%
(5) (93) (10) (15/101) (0.75, 3.5) (1,4) (44)
22% (24) 2% (2/100) 0% (0)
67% (72) 8% (9) 16% (17)
b.01 .03 b.01
6% (7) 24% (26) 6% (7) 8% (9) 2% (2) 3% (3) 2% (2) 4% (4) 17% (18)
12% 46% 27% 21% 3% 3% 4% 9% 41%
(13) (50) (29) (23) (3) (3) (4) (10) (44)
.16 .01 b.01 .01 .75 1 .41 .1 b.01
10% (11) 3% (3) 3% (3) 6% (6) 0% (0) 2% (2) 2% (2)
62% 19% 44% 25% 10% 4% 5%
(67) (21) (48) (27) (11) (4) (5)
b.01 b.01 b.01 b.01 b.01 .41 .24
.29
.34 1 b.01
b.01 b.01 b.01 b.01
NSAID: nonsteroidal anti-inflammatory drug, CAD: coronary artery disease, HTN: hypertension, DM: diabetes mellitus, COPD: chronic obstructive pulmonary disease, CHF: congestive heart failure, Afib: atrial fibrillation, CRF: chronic renal failure, CA: cancer.
4. Limitations Our study has several limitations. First, we only evaluated 1 hospital ED in an urban environment. Locally, we have 4 hospitals within a 2.5-mile radius. Frequent users may utilize multiple health care resources; and without access to medical records at these nearby facilities, it is difficult to obtain the total number of ED visits annually for any 1 individual. However, the characteristics of our studied frequent ED users are similar to previously published work with regards to age, sex, and other demographics [4,5]. We feel that this gives our data face validity and that our data are representative of the broader population of frequent ED users and not unique to our institution. Second, there were more than 2900 patients that that had at least 4 ED visits to our ED in 2012. Multiple definitions of frequent ED users exist in the literature including more than 4, more than 12, and more than 17 visits/year [3–5]. Because of limitations in resources, we focused on those with the highest ED usage: more than 12 and more than 17 visits/year. The remainder of the frequent users may have demographics that differentiate them from the most frequent users. Those patients with more than 12 visits/year represent nearly 3% of our ED’s annual volume but have charges similar to nonfrequent users. We believe that this group represents the ideal population
Total number of visits (n = 2072) Median number of visits/patient (IQR) Charges Total charges in US dollars Median charges in US dollars/visit (IQR) Chief concern Pain Psychiatric Other (nonpsych or nonpain concern) Diagnostic testing Laboratories Radiograph ECG CT Ultrasound Venous Doppler V/Q MRI Disposition Discharged Admitted/transferred LWBS/eloped Left AMA Died in ED Narcotic Rx if discharged Arrival by EMS
Nonfrequent users
Frequent users
150
1922
2 (1, 2)
1,012,610.21 1280 (573, 4569)
P value
17 (15, 22)
10,465,216.07 1220.17 (490, 4179.04)
.97
57% (85) 4% (6) 47% (71)
59% (1130) 11% (205) 30% (582)
.6 .01 b.01
41% (62) 42% (63) 26% (39) 17% (25) 1% (2) 3% (4) b 1% (1) 0% (0)
48% (917) 30% (584) 19% (371) 11% (220) 1% (26) b 1% (12) 0% (0) b1% (6)
.13 b.01 .05 .06 .98 .01 b.01 .49
81% (122) 18% (27) b 1% (1) 0% (0) 0% (0) 22% (27/123) 14% (20/147)
80% (1529) 16% (317) 3% (66) b1% (9) b1% (1) 33% (535/1605) 19% (362/1916)
.36
.01 b.01
CC: chief concern, ECG: electrocardiogram, CT: computed tomography, V/Q: venous perfusion scan, MRI: magnetic resonance imaging, LWBS: left without being seen, AMA: left against medical advice, Rx: prescription.
where interventions to find alternative care pathways may have the greatest impact by reducing ED visits and reducing cost. We performed a retrospective chart review and, as a result, are unable to determine “why” patients chose to present to the ED as opposed to another health care provider/facility. As a result in limitations in support staff, we did not evaluate interrater reliability. The RAs had data elements randomly selected and evaluated on a monthly basis by one of the investigators for consistency and accuracy [10]. Given the relative paucity of data regarding charges associated with frequent ED users, we did not perform a sample size calculation. We performed a review of all patients with more than 12 visits to our ED in 2012, representing nearly 3% of our ED’s annual volume. As a result of the retrospective nature of our work, we were unable to control sample size. Although it would able to expand the definition of frequent users to include more patients, we chose to use definitions of frequent users in the existing literature. We chose to look at diagnostic testing as a marker of resource utilization. Although there may be other markers of resource utilization (medications administered in the ED, intravenous placement, etc), we chose to focus our efforts on those resources that we felt a priori help emergency providers evaluate for acute lifethreatening conditions. We used the amount charged to the patient as a representation of the cost of the visit. The actual cost of each visit is often challenging to determine. Although charges may vary by institution, we felt that this was the most consistent way to estimate the dollar amount associated with the care of our study population. Lastly, we performed a retrospective chart review. Both abstractors were trained to abstract these data similarly, used a standard data collection form, and had their work regularly reviewed by one of the investigators to ensure correct abstraction. Although we have attempted to standardize the data collection, much of these data are self-reported by the patients—such as substance abuse or psychiatric
1218
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219
and medical histories. As nonfrequent users presented less than 4 times in the year, it could be that this information was not asked or simply assumed to be negative by the providers who saw the patient. Frequent users have more opportunities to be asked about this information and, as such, may have a disproportionately higher instance of these conditions being documented (eg, tobacco use, medical and psychiatric history). 5. Discussion Physicians, health care workers, and policy makers may have a preconceived notion of ED frequent visitors [5]. Previous work has detailed providers’ misconception including that they are more often uninsured minorities that utilize the ED for basic care or drug-seeking behavior [5,9]. However, others have shown that frequent ED users more often are white, are female, have a mean age of 40 years, and are covered by insurance [5]. The uninsured represent less than 15% of frequent users and are no more likely to be frequent users than those that use the ED occasionally [5]. There are, however, a disproportionately higher number of publicly insured (Medicaid/Medicare) patients that are frequent users (60%) compared to nonfrequent users (36%). In line with national data, the frequent users of our ED consisted predominantly of women (55%), had a mean age of 42.5 years, and were predominantly white (69%) [5]. The demographics of our frequent users parallel other studies; however, prior to our work, little was known regarding the charges related to these frequent ED visits. Although the median charge per visit was similar between frequent and nonfrequent users, the overall charges for frequent users were nearly 10 times those of nonfrequent users because of the sheer number of visits by frequent users. Finding alternative plans of care for these patients could help to reduce health care costs, although equally high number of patients in each group reported having a PCP (95%). Although we did not evaluate the specific reasons frequent users presented to the ED instead of their PCP, frequent users did live in closer proximity to the ED and their use may be related to the ED’s ease of access. Like their nonfrequent counterparts, the majority of frequent users reported that they had PCPs. Prior studies have reported that most frequent users prefer ED care for a variety of reasons [7,9]. Although some frequent users present to the ED because they feel that their PCP may be unable to adequately care for their issues, many utilized the ED due to its accessibility (provides care 24 hours a day and no need for an appointment) [5,7,9]. Frequent users are also believed to utilize the ED for treatment of chronic pain and desire narcotic pain medication—with estimates of drug-seeking patients accounting for nearly 25% of all ED visits [7]. Although variation in presenting concerns depends on the site of the study, with urban EDs having a higher percentage of psychiatric and substance abuse concerns, pain-related concerns may be seen in a variety of settings [5]. We did not note a difference in pain-related concerns between frequent and nonfrequent users. However, prescriptions for narcotic pain medications were provided at ED discharge more often to frequent users (33%) than nonfrequent users (22%). Although we are unable to determine why ED providers prescribed narcotic pain medications more frequently, it may be related to the allergies reported by frequent users. Frequent users reported allergies to nonsteroidal anti-inflammatory medications more often than nonfrequent users (44% vs 6%). Previous efforts aimed at reducing the prescription of narcotics to frequent users have been shown to markedly decrease ED usage in this population [7]. Mental health as well as substance abuse issues play important roles in frequent use of the ED [7–9]. One prior study showed that frequent ED use is defined by substance abuse more strongly than any other factor [3]. We found that frequent users abused tobacco, illicit drugs, and alcohol at higher rates then nonfrequent users. A recent study stated that frequent users had significant burden of disease and that behavioral health issues were only a small contributing factor
[11]. Our study supports the fact that many frequent users have many chronic medical issues but also psychiatric illness. We found that many frequent ED users often report mental health conditions including anxiety (19%), bipolar disorder (44%), depression (25%), and schizophrenia (10%). Nonfrequent users rarely reported such histories. Previous work has detailed geographic variation in psychiatric disease, which may explain the differences noted between our study population and that of others [12,13]. The relatively high percentage of frequent users reporting mental health conditions suggests that alternative care pathways targeting frequent users may benefit from input from mental health specialists. We did not notice a difference in mean charges between frequent and nonfrequent users. Both groups had similar diagnostic workups including blood work and usage of computed tomography. In addition, both groups were admitted to the hospital equally. Radiographs were performed less often in frequent users; however, these have a relatively lower cost associated with them than other diagnostics tests, such as computed tomography, which may be the larger drivers of the visit charges. Prior research has suggested that frequent users tend to have a higher utilization of other resources including higher rates of admission and are transported by ambulance more frequently [5,8]. Although we did note increased ambulance transport rate, we did not note any difference in admission rates between the 2 groups. Of the 1922 visits by frequent users, more than 1500 were not admitted to the hospital. One study reported that approximately 29% of frequent user visits were more appropriate for primary care where the cost of care is greatly reduced [14]. We identified that 2.9% of our ED visits were from our 108 frequent users with a median charge of $1220/ visit. Developing alternative care pathways for frequent users including directing them to a PCP in place of the ED could result in significant reductions in health care dollars. Using a median of $1220/ visit, reducing the number of frequent ED visits by the same 29% nationally could reduce health care expenditures by up to $1.35 billion annually [15]. Although our data detail the financial magnitude of frequent ED users, before fully implementing alternative care pathways, future work should further detail why frequent users present to the ED as opposed to other care sites. 6. Conclusion In this single-center study, frequent users have unique demographics and live closer to the hospital than nonfrequent users; however, their dispositions are similar. Although charges per visit did not differ between frequent and nonfrequent users, the sheer number of frequent user visits amounted to approximately 10 times the total amount. Interventions aimed at lowering frequent ED use must be tailored to certain demographics and needs but could significantly reduce health care costs. Acknowledgments The authors would like to thank Dr. Arvind Venkat for his assistance in proof-reading this manuscript. References [1] Avalere. US Census Bureau: national and state population estimates for community hospitals. http://www.census.gov/popest/data/state/totals/2011/ index.html; 2011. [Accessed July 2011, 2011]. [2] Lee MH, Schuur JD, Zink BJ. Owning t0068cbee cost of emergency medicine: beyond 2%. Ann Emerg Med 2013;62(5):498–505.e493. [3] Doupe MB, Palatnick W, Day S, et al. Frequent users of emergency departments: developing standard definitions and defining prominent risk factors. Ann Emerg Med 2012;60(1):24–32. [4] Blank FS, Li H, Henneman PL, et al. A descriptive study of heavy emergency department users at an academic emergency department reveals heavy ED users have better access to care than average users. J Emerg Nurs 2005;31(2):139–44.
C. Ondler et al. / American Journal of Emergency Medicine 32 (2014) 1215–1219 [5] LaCalle E, Rabin E. Frequent users of emergency departments: the myths, the data, and the policy implications. Ann Emerg Med 2010;56(1):42–8. [6] A Report on Frequent Users of Hospital Emergency Departments in South Carolina. http://imph.org/a-report-on-frequent-users-of-hospital-emergency-departmentsin-south-carolina/; 2011. [Accessed February 1, 2014]. [7] Grover CA, Close RJ, Villarreal K, Goldman LM. Emergency department frequent user: pilot study of intensive case management to reduce visits and computed tomography. West J Emerg Med 2010;11(4):336–43. [8] DiPietro BY, Kindermann D, Schenkel SM. Ill, itinerant, and insured: the top 20 users of emergency departments in Baltimore city. ScientificWorldJournal 2012;2012:726568. [9] Althaus F, Paroz S, Hugli O, et al. Effectiveness of interventions targeting frequent users of emergency departments: a systematic review. Ann Emerg Med 2011;58(1): 41–52.e42.
1219
[10] Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996; 27(3):305–8. [11] Billings J, Raven MC. Dispelling an urban legend: frequent emergency department users have substantial burden of disease. Health Aff (Millwood) 2013;32(12):2099–108. [12] Lewis G, Booth M. Regional differences in mental health in Great Britain. J Epidemiol Community Health 1992;46(6):608–11. [13] Bijl RV, de Graaf R, Hiripi E, et al. The prevalence of treated and untreated mental disorders in five countries. Health Aff (Millwood) 2003;22(3):122–33. [14] Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52(2):126–36. [15] Carlson JN, Menegazzi JJ, Callaway CW. Magnitude of national ED visits and resource utilization by the uninsured. Am J Emerg Med 2013;31(4):722–6.
