Research Forum Abstracts Results: Case study hospitals had 207,775 ED visits from which there were 7,699 EDOU visits having average stays of 15.3 hours, with 9.4% staying beyond 24 hours and 0.1% beyond 48 hours. 83.9% were discharged from the EDOU. HCUP regional data found 6.45 million ED visits, and 507,239 OS visits whose mean LOS was 20 hours, with 29% having OS stays beyond 24 hours and 7.2% having OS stays beyond 48 hours. The NHAMCS survey estimated 129.5 million annual ED visits, with 2.54 million annual OS visits, of which 38% were managed in an EDOU. Twenty-seven percent EDOU stays were beyond 24 hours and 4% were beyond 48 hours. In NHAMCS data, units that were managed by the ED had shorter stays (24.2 vs. 17.1 hr) and lower admit rates (25% vs. 51%). The case mix of conditions being treated in observation was similar across all comparison groups. If case study hospital length of stays were applied to regional and national groups, this would reduce OS days by 18% annually. Conclusions: Redesigned protocol-driven EDOUs may provide a potential solution to prolonged inpatient observation stays.
157
The Persistent Emergency Department Superuser: Defining a Population to Target Limited Resources
Peabody CR, Gruber PF, Menchine MD, McCollough M/LAC þ USC Department of Emergency Medicine, Los Angeles, CA
Study Objective: Frequent users of the emergency department (ED) may contribute to department crowding, increased waiting times, and inappropriate resource utilization. Currently, no consensus definition of what constitutes an ED frequent user exists. Previous literature defines a frequent user as an individual with between 3-20 visits per year, with the subset of “superuser” describing individuals with 10-20 visits per year. These inconsistent definitions may explain heterogeneity in responses to interventions aimed at reducing frequent ED use. We seek to determine if there is a distinct number of ED visits per patient that identifies a distinct frequent user/superuser group and if this group remains persistent over multiple time periods. Methods: We studied a single-center, retrospective cohort at a large (>180,000 patient encounters per year), urban, safety-net hospital in Los Angeles. Using all ED patient registrations from 06/2009-12/2012, we graphically analyzed the distribution of ED visits per patient to determine if an observable discontinuity could identify a distinct (or multiple distinct) subpopulation of frequent users/superusers. We then compared traditional superusers (>20 visits per year) from the year 2011 to 2012 to identify whether this group is stable over time and may be a viable target for interventions, or if their use simply regresses to the mean. We hypothesized, that there might exist a distinct frequent user population, and that after a few visits, there may be a rise in the amount of patient registrations by a large and distinct subset population, resulting in a bimodal distribution of visits per patient. Results: Since June 2009 to the end of 2012, there have been 607,181 unique ED patient registrations by 324,894 patients. The distribution of visits ranged from 1-461; with 66.8% of patients registering for only one ED visit (Figure). In fact, 99% of patients had ten visits or fewer. There was no discernable bimodal distribution in the number of ED visits per patient. Therefore, instead of a unique patient population, frequent users may be the tail end of a Poisson (expected frequency over time) distribution of the overall users of the ED. This population may simply regress to the population mean over time. To assess stability of the tail end of this distribution, we took the arbitrarily defined “superuser” population (>20 visits per year) from 2011 and compared their ED use in 2011 and 2012. In 2011, 139 patients had 20 visits or more (0.001% of total population), accounting for 4,782 total visits. In 2012, these same superusers had 2624 visits; however, only 47 of the original 139 superusers (33%) had over 20 visits or more. These persistent superusers may benefit most from resources aimed at reducing ED use. These patients may also have unique characteristics that may help predict future superusers. Conclusions: Frequent users of the ED are rare, and a large, unique population does not exist. Most frequent use is transient; however, the persistent superuser may have unique characteristics that should be the focus of further research and the target for limited resources aimed at reducing ED use.
158
Variation in Charges for Emergency Department Visits in California
Ravikumar D, Antwi YA, Hsia RY/University of California, San Francisco, San Francisco, CA; Indiana University-Purdue University Indianapolis, Indianapolis, IN
Background: Previous studies show that charges for common conditions and price indices vary substantially, however, there is scant data on variation of charges for emergency department (ED) visits. Study Objectives: To determine the variation in charges for levels 1-5 ED visits and the hospital and market-level factors associated with higher charges at hospitals across the state of California. Methods: We conducted a descriptive analysis of charges for ED visits level 1-5 in 2011 using chargemaster data for each hospital’s 25 most common outpatient procedures as reported to the California Office of Statewide Health Planning and Development (OSHPD). We then regressed hospital- and market-level characteristics on reported charges for ED visits to determine which factors explained some of the observed variation in charges between hospitals. Results: There was a great degree of variability for charges across all visit levels. For example, charges for a level four visit ranged from $185-$6,662 (median: $1,411). We found that both hospital and market level factors were associated with charges for ED visits. Using level four visits as an example, we found that visiting a government hospital was associated with lower charges, while visiting a hospital with a high proportion of Medicaid patients, a high case mix severity, or in an area with a high cost of living or high proportion uninsured was associated with higher charges. Level two and three visits showed similar trends. For level 4 visits, our multivariate regression analysis revealed 39.5% lower charges for (95% CI: 18.1% to 61%) for government-owned hospitals compared to not-for-profit hospitals. For each percentage point increase in the percent of Medicaid patients, there was an increase in charges of .67% (95% CI: .4% to 1.3%). Hospitals with the highest mix of severe cases were associated with a 27.1% increase in charges (95% CI: 5.9% to 48.5%). In terms of market-level characteristics, regression analysis revealed that areas with a higher wage index were associated with 33.2% higher charges (95% CI: 9.5% to 56.8%). For each percentage point decrease in the percent of uninsured patients, charges decreased by 5.1% (95% CI: 1.8% to 8.3%). Conclusions: The wide variation in charges for ED visits supports past literature suggesting patients are unable to act as rational consumers of ED care. Given the lack of available data regarding charges for ED visits and treatments, patients cannot make informed decisions about where to receive their care and could receive any of the wide range of charges we document. Further studies should determine if medical charge transparency could help narrow existing variation. Table. Data for ED visits for hospitals in California (Charges in dollars). CPT 99281 99282 99283 99284 99285
Average Std Dev # Data Charge Charge Median Points Min
Description ED ED ED ED ED
Visit Visit Visit Visit Visit
159
Level Level Level Level Level
1 2 3 4 5
192 572 991 1680 1233
60 237 469 965 401
196 550 900 1411 1274
4 240 244 238 6
119 105 155 185 683
Max 257 1422 3130 6662 1823
Emergency Physician Variation in Admissions for Common Chief Complaints
Levine MB, Moore AB, Kuehl D, Franck C, Li J/Virginia Tech Carilion School of Medicine, Roanoke, VA; Carilion Clinic Department of Emergency Medicine, Roanoke, VA; Virginia Tech Department of Statistics, Blacksburg, VA
Study Objectives: The emergency department (ED) represents an increasing source of hospital admissions, placing costly decisions in the hands of emergency physicians. Emergency physicians have been shown to vary dramatically in their admission rates, suggesting some providers may admit patients unnecessarily. It is unclear whether this variation in provider admission rates exists across all types of patient presentations. Determining the degree of admission rate variation for common chief complaints may help delineate clinical presentations that increase risk of inappropriate and costly hospital admissions. Methods: This retrospective study examined board certified/eligible emergency physicians practicing at an 850-bed tertiary level 1 trauma center with 85,000 annual
S60 Annals of Emergency Medicine
Volume 62, no. 4s : October 2013