Presented at the American College of Emergency Physicians Scientific Assembly Research Forum, Las Vegas, NV, September 2010.
National Study of Emergency Department Observation Services
Article first published online: 30 AUG 2011
© 2011 by the Society for Academic Emergency Medicine
Academic Emergency Medicine
Volume 18, Issue 9, pages 959–965, September 2011
How to Cite
Wiler, J. L., Ross, M. A. and Ginde, A. A. (2011), National Study of Emergency Department Observation Services. Academic Emergency Medicine, 18: 959–965. doi: 10.1111/j.1553-2712.2011.01151.x
The authors have no relevant financial information or potential conflicts of interest to disclose.
Supervising Editor: Jacob Ufberg, MD.
- Issue published online: 11 SEP 2011
- Article first published online: 30 AUG 2011
- Received December 20, 2010; revision received February 27, 2011; accepted March 21, 2011.
ACADEMIC EMERGENCY MEDICINE 2011; 18:959–965 © 2011 by the Society for Academic Emergency Medicine
Objectives: The objective was to describe patient and facility characteristics of emergency department (ED) observation services in the United States.
Methods: The authors analyzed the 2007 National Hospital Ambulatory Medical Care Survey (NHAMCS). Characteristics of EDs with observation units (OUs) were compared to those without, and patients with a disposition of ED observation were compared to those with a “short-stay” (<48 hour) hospital admission. Results are descriptive and without formal statistical comparisons for this observational analysis.
Results: An estimated 1,746 U.S. EDs (36%) reported having OUs, of which 56% are administratively managed by ED staff. Fifty-two percent of hospitals with ED-managed OUs are in an urban location, and 89% report ED boarding, compared to 29 and 65% of those that do not have an OU. The admission rate is 38% at those with ED-managed OUs and 15% at those without OUs. Of the 15.1% of all ED patients who are kept in the hospital following an ED visit, one-quarter are kept for either a short-stay admission (1.8%) or an ED observation admission (2.1%). Most (82%) ED observation patients were discharged from the ED. ED observation patients were similar to short-stay admission patients in terms of age (median = 52 years for both, interquartile range = 36 to 70 years), self-pay (12% vs. 10%), ambulance arrival (37% vs. 36%), urgent/emergent triage acuity (77% vs. 74%), use of ≥1 ED medication (64% vs.76%), and the most common primary chief complaints and primary diagnoses.
Conclusions: Over one-third of U.S. EDs have an OU. Short-stay admission patients have similar characteristics as ED observation patients and may represent an opportunity for the growth of OUs.
Curbing escalating health care costs was noted to be a major catalyst in recent passage of health care reform legislation.1,2 The medical necessity for select health services, including inpatient admissions, is increasingly scrutinized by insurers, including the nation’s largest health care insurer, the Centers for Medicare and Medicaid Services (CMS).3,4 Emergency department (ED) observation care offers the ability to avoid costly inpatient admissions, specifically “short-stay” admissions (under 48 hours), while mitigating the risk of immediate discharge from the ED for those patients who require ongoing care or treatment beyond ED stabilization.5–7 Single-center studies of ED observation unit (OU) care have suggested that this setting may be used as an alternative to inpatient admission, with improved clinical and economic outcomes.5–11 A 2003 survey estimated that 19% of all U.S. hospitals had an OU, with 12% planning one.12 However, little is known about the current national prevalence and characteristics of such units, if the anticipated growth occurred, or how many ED patients might be eligible for observation care. While ED utilization and case mix has been tracked and reported for years, the 2007 National Hospital Ambulatory Medical Care Survey (NHAMCS, data released in September 2010) was the first to include questions regarding admissions to an OU.13
We analyzed these new data to describe the prevalence and characteristics of ED OUs in the United States, compared facility characteristics of hospitals with and without OUs, compared facility characteristics of hospitals with OUs administratively managed by ED or inpatient staff, compared patient-level characteristics of ED observation and short-stay inpatient admissions, and assessed whether an unrealized potential exists for ED-based observation services.
This was a retrospective, cross-sectional analysis of the 2007 NHAMCS.13 This study was given institutional review board approval as an exempt protocol.
Study Setting and Population
A detailed description of the NHAMCS survey methods is given elsewhere.14 Briefly, the NHAMCS is a four-stage probability sample conducted annually by the National Center for Health Statistics and covers geographic primary sampling units, hospitals within primary sampling units, EDs within hospitals, and patients within EDs. A sample of 482 U.S. noninstitutional general and short-stay hospitals was selected for the 2007 NHAMCS, of which 405 were eligible and 362 participated.
Trained staff collected demographic and clinical data during a randomly assigned 4-week data period for each of the sampled hospitals, approximately once every 15 months. Completed data collection forms were sent to the Constella Group (Durham, NC), where they were coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).15
Before participation in this survey, hospitals complete a standardized 20-page hospital induction interview form.14 In the 2007 induction form, two questions were added relevant to OUs, “Does your ED have an observation or clinical decision unit?” and if yes, “Is your observation or clinical decision unit administratively a part of the ED or the inpatient side of the hospital?” Additionally, in the disposition section of the individual ED patient case report form, an additional disposition category was added for “Admit to observation unit,” indicating an admission to ED observation. We grouped ED visits into three disposition categories: discharge from the ED, admission to ED observation, and short-stay hospital admission to a floor or telemetry bed. Patients with short-stay admissions were classified as those whose disposition was admission to an inpatient floor, mental health/detoxification, or telemetry bed and whose hospital length of stay was ≤48 hours. Patients admitted to the intensive care unit, stepdown unit, cardiac catheterization laboratory, or operating rooms were excluded, because they were not considered as potentially comparable to ED observation patients.
We analyzed the data by hospital facility characteristics and included the hospital metropolitan statistical area status (urban and nonurban); U.S. region (Northeast, Midwest, South, and West); hospital ownership (nonprofit, government [nonfederal], and private/for profit); and median income in region of the ED by zip code. Region and metropolitan statistical area categories represent standardized geographical divisions defined by the U.S. Bureau of the Census. ED facility characteristics also included presence of fast track, ambulance diversion during the past year, and ED inpatient boarding during the past year.
We also analyzed the data in terms of patient-level characteristics such as age, sex, race/ethnicity, and source of payment. Additionally, we analyzed visit characteristics including mode of arrival; triage acuity; arrival time; day of arrival (weekend or weekday); month of arrival; imaging, tests, or medications ordered; ED length of stay; primary chief complaint; and primary ICD-9-CM diagnosis.
We performed all analyses using Stata 10.1 software (StataCorp, College Station, TX) to determine nationally representative estimates with 95% confidence intervals (CIs) using assigned weights, which account for the complex weighted survey design. Because this descriptive study did not have a specific hypothesis, we compared estimated proportions between groups without formal statistical testing.
In 2007, of the estimated 4,891 U.S. EDs, 1,746 (36%) reported having an OU in the ED (Tables 1A and 1B). ED characteristics by OU status are reported in Table 2. There was variation in the proportion of EDs with OUs by several patient and ED characteristics.
|ED OU in Hospital||No ED OU in Hospital||Unknown|
|Total ED visits (117 M)||47 M (40%)||66 M (56%)||3.7 M (3%)|
|Total admitted to ED observation status (2.5 M)||1.2 M (49%)*||1.1 M (44%)||0.2 M (7%)|
|Total U.S. hospitals (4,891)||1,746 (36%)†||3,065 (63%)||80 (2%)|
|ED Managed OU||Inpatient Managed OU||Unknown|
|Total ED visits admitted to ED OU||0.8 M (69%)||0.3 M (23%)||0.1 M (8%)|
|Total U.S. hospitals with ED OU||902 (52%)||707 (40%)||137 (8%)|
|Hospital Characteristics||ED Managed OU,*% (95% CI)||Inpatient Managed OU,*% (95% CI)||No ED OU, % (95% CI)|
|Nonwhite patients (%)|
|0%–19%||34 (20–50)||46 (24–70)||46 (33–59)|
|20%–39%||26 (14–41)||32 (17–53)||25 (17–35)|
|40%–59%||13 (7–22)||12 (3–34)||18 (12–27)|
|≥60%||28 (17–43)||10 (4–22)||12 (7–20)|
|0%–4%||19 (11–30)||21 (10–39)||23 (15–34)|
|5%–14%||29 (19–42)||13 (6–27)||39 (32–48)|
|15%–24%||33 (21–46)||29 (11–56)||19 (13–27)|
|≥25%||19 (9–35)||37 (20–60)||18 (11–28)|
|Admit rate (observation + hospital)|
|0%–9%||35 (22–50)||53 (30–75)||39 (28–52)|
|10%–14%||17 (8–31)||3 (1–12)||27 (17–40)|
|15%–19%||11 (6–20)||19 (7–43)||19 (11–29)|
|≥20%||38 (25–53)||25 (10–48)||15 (10–23)|
|Large urban||52 (38–66)||20 (10–37)||29 (21–38)|
|Medium–small urban||33 (21–49)||39 (21–61)||25 (17–35)|
|Rural||15 (6–33)||41 (22–63)||47 (33–60)|
|Northeast||19 (11–32)||9 (3–20)||13 (9–19)|
|Midwest||24 (15–37)||32 (14–56)||32 (23–43)|
|South||30 (20–42)||51 (30–73)||40 (31–50)|
|West||27 (15–42)||8 (3–23)||14 (7–26)|
|Nonprofit||78 (63–88)||67 (48–82)||68 (60–76)|
|Government, nonfederal||14 (7–25)||28 (14–49)||15 (9–25)|
|For-profit||8 (4–18)||5 (2–14)||16 (10–26)|
|Admitted patients ever boarded in ED|
|Yes||89 (76–95)||30 (15–50)||65 (52–76)|
|No||11 (5–24)||70 (50–85)||35 (24–48)|
|Any ambulance diversion in 2006|
|Yes||33 (22–47)||10 (4–22)||23 (16–31)|
|No||44 (29–59)||69 (45–85)||55 (45–65)|
|Data not available||24 (13–38)||22 (9–44)||22 (16–31)|
|Yes||52 (37–66)||29 (13–51)||29 (22–38)|
|No||48 (34–63)||71 (49–87)||71 (62–78)|
|Median income for ED zip code|
|$0–$32,793||25 (15–40)||47 (28–66)||39 (28–51)|
|$32,794–$40,626||28 (16–45)||28 (13–51)||30 (21–41)|
|$40,627–$52,387||27 (18–40)||20 (7–46)||18 (11–30)|
|≥$52,388||19 (11–31)||5 (2–17)||13 (8–20)|
Of all U.S. ED visits in 2007, 2.1% were admitted to ED observation, and 13% were admitted to a hospital bed. The short-stay inpatient admission patients accounted for 1.8% of total ED visits and 14% of all hospital admissions. In total, 15% of all ED patients were admitted either to ED observation or to a hospital bed, with 27% of these patients being either short-stay or ED observation patients. Of patients admitted to ED observation, 36% were at a hospital with an ED-managed OU, 12% at an ED with an inpatient-managed OU, and 47% at an ED without an OU. Patient and facility characteristics by disposition status are presented in Table 3. Patient visits admitted to ED observation were similar to those admitted to short-stay hospital admissions for all demographic and clinical characteristics measured, except that more ED observation patients arrived during the night (11:00 pm to 6:59 am). Of ED observation patients, 18% were ultimately converted to a hospital inpatient admission.
|Characteristics||ED Observation, % (95% CI)||Short Stay (<2 day) Hospital Admission, % (95% CI)|
|Median age (yr)|
|<18||6 (4–10)||9 (7–13)|
|18–44||30 (25–35)||27 (22–33)|
|45–64||32 (27–37)||33 (29–37)|
|65+||31 (27–37)||30 (25–36)|
|Female sex||56 (51–61)||49 (43–54)|
|NH white||67 (59–74)||68 (61–74)|
|NH black or African American||18 (13–26)||18 (14–22)|
|Hispanic||11 (7–16)||12 (8–17)|
|Other||4 (3–7)||3 (2–5)|
|Source of payment|
|Private||32 (26–39)||36 (31–41)|
|Medicare||30 (24–36)||31 (26–36)|
|Medicaid||24 (19–29)||22 (17–28)|
|Self-pay||12 (7–18)||10 (7–13)|
|Other||3 (nc)||2 (nc)|
|Northeast||20 (13–31)||18 (12–25)|
|Midwest||30 (19–43)||23 (15–32)|
|South||28 (17–44)||39 (29–51)|
|West||22 (9–43)||20 (14–28)|
|ED visit features|
|EMS arrival||37 (32–42)||36 (31–41)|
|<15 minutes||29 (23–35)||34 (28–40)|
|15–60 minutes||46 (36–56)||40 (36–45)|
|61–120 minutes||9 (6–14)||7 (5–10)|
|>121 minutes||3 (1–5)||2 (nc)|
|None/unknown||13 (8–22)||16 (12–22)|
|Time of arrival|
|7:00am–2:59pm||44 (38–50)||39 (33–45)|
|3:00pm–10:59pm||39 (33–44)||38 (34–43)|
|11:00pm–6:59am||17 (14–22)||23 (19–28)|
|Weekend arrival||30 (26–35)||25 (21–29)|
|Number diagnostic tests|
|0||7 (4–11)||4 (3–7)|
|1–3||22 (17–27)||19 (15–23)|
|4–6||27 (23–33)||29 (25–34)|
|≥7||44 (36–53)||48 (41–54)|
|Cardiac enzymes||41 (30–54)||46 (41–52)|
|Any imaging||65 (59–71)||71 (63–78)|
|Medication in ED|
|0||36 (28–44)||24 (19–29)|
|1||17 (13–22)||20 (17–24)|
|>1||47 (41–54)||56 (51–61)|
|ED LOS, hours|
|0–6||63 (55–70)||73 (67–78)|
|7–12||21 (16–27)||20 16–25)|
|13–24||11 (7–15%)||3 (nc)|
|>24||5 (2–10)||4 (nc)|
The primary chief complaints by category were also similar between patient visits admitted to ED observation and those admitted as short-stay inpatient admissions, with the top four chief complaints being circulatory, digestive, respiratory, and injury/poisoning for ED observation patients and circulatory, digestive, neurologic, and respiratory for short-stay admission patients (Table 4). The most common circulatory system chief complaint was chest pain (84% for ED observation and 89% for short-stay admission patients), and the most common digestive system chief complaint was abdominal pain (55% for ED observation and 43% for short-stay admission patients).
|Characteristics||ED Observation, % (95% CI)||Short Stay (<2-day) Hospital Admission, % (95% CI)|
|Primary chief complaints|
|Circulatory system||25 (21–29)||27 (23–32)|
|Digestive system||15 (12–19)||19 (16–23)|
|Respiratory system||14 (10–18)||10 (8–13)|
|Injury and poisoning||8 (6–11)||8 (6–11)|
|Mental disorders||7 (4–10)||2 (1–4)|
|Nervous system||6 (4–9)||13 (10–18)|
|Musculoskeletal system||6 (4–9)||6 (4–8)|
|General/ill-defined symptoms||6 (4–9)||5 (4–8)|
|Genitourinary system||5 (3–8)||2 (nc)|
|General/screening examinations||4 (2–6)||3 (nc)|
|General/ill-defined conditions||33 (28–39)||44 (40–49)|
|Circulatory system||9 (6–14)||14 (10–18)|
|Injury and poisoning||13 (9–19)||10 (7–14)|
|Respiratory system||7 (5–11)||5 (3–7)|
|Mental disorders||6 (4–10)||3 (2–5)|
|Digestive system||6 (4–9)||6 (4–9)|
|Genitourinary system||6 (4–9)||2 (nc)|
|Nervous system||5 (3–8)||3 (2–6)|
|Skin/subcutaneous tissue||3 (nc)||2 (nc)|
|Musculoskeletal system||2 (nc)||2 (nc)|
The most common primary diagnoses of patients who received ED observation care compared to short-stay admission patients were also similar, with general/ill-defined conditions, circulatory system, and injury/poisoning being the most common. Of those categorized as “general/ill-defined,” the diagnosis was most often described as “other symptoms involving abdomen/pelvis” (ICD-9 789.x; 20% overall), chest pain (ICD-9 786.5×; 18% overall), fever (ICD-9 780.6×; 7% overall), headache (ICD-9 784.0×; 7% overall), or nausea/vomiting (ICD-9 787.0×; 7% overall).
Previous studies from 1989 and 2003 have estimated the number of EDs with OUs nationwide to be 19% to 27%, with 12% to 16% of EDs reporting plans to open an OU.12,16 We found that over the past decade, the percentage of U.S. hospitals reporting the use of an OU has increased to 36%, with over half of ED OUs administratively managed by ED staff. This is slightly lower than previous reports of 59% to 93% of ED OUs being administratively managed by ED staff in 1989.16 Currently, regional variation of ED OU densities does exist; a greater percentage of Western hospitals surveyed have an ED OU when compared to those surveyed in the southern United States.
There are several possible reasons why the prevalence of ED OUs may be increasing. Over the past decade, EDs have become increasingly crowded (often as a result of hospital inpatient crowding), which leads to ambulance diversion, patients who leave before being seen, and suboptimal clinical conditions.17 Observation services have become an alternative to inpatient hospital admission for several conditions, serving as a way to improve efficiency, clinical care quality, and patient satisfaction, while minimizing the costs associated with inpatient treatment.6,8–10,18–21 Previous studies have reported that opening an OU decreased total inpatient admissions by 12%, ambulance diversion by 40%, and patients who leave without being seen by 53%.22,23 Based on these and other studies, the Institute of Medicine report on the state of emergency medicine suggested that establishment of OUs is part of the solution to the crowding crisis that emergency medicine faces.24 Consistent with these findings, our analysis found that EDs having an OU seemed more likely to have higher admission rates, ED patient boarding, and EMS diversion rates. It is likely that the OU is an attempt to address these issues.
Not surprisingly, the most common ED chief complaints of patients who received observation care were symptoms related to the circulatory system (e.g., chest pain) and digestive system (e.g., abdominal pain). Although the ED chief complaint may be similar to the admission diagnosis for observation care, information about the reason for admission to ED observation services was not available. Future versions of the NHAMCS might consider addressing this issue. Ultimately, the most common final primary diagnosis of patients who received observation care was classified as “general/ill-defined conditions,” including chest pain and abdominal pain. This condition case mix correlates well with prior studies.12,15,22
Of the approximately 117 million U.S. ED visits in 2007, nearly equal numbers of patients were admitted to either ED observation or for an inpatient short-stay hospital admission. Notably, more than one out of four patients admitted from the ED was either admitted to ED observation or a short-stay admission. Our analysis demonstrates that admissions to ED observation and inpatient short stay were remarkably similar with regard to patient demographics, ED presentation acuity, chief complaint, resource utilization, and final diagnosis. Previous studies have also reported that an estimated 15% to 26% of ED OU patients are eventually admitted as an inpatient.12,22,25 Our data support the findings of these previous studies with an admission rate of 18%.
The CMS currently defines observation services as “a well-defined set of specific, clinically appropriate services, which include ongoing short term treatment, assessment, and reassessment before a decision can be made regarding whether patients will require further treatment as hospital inpatients or if they are able to be discharged from the hospital [ED].”2 They go on to clarify that observation care can be delivered in any treatment space within the ED or hospital. This does not require an observation or clinical decision unit. Since the current NHAMCS survey did not identify the status (outpatient observation or inpatient) of patients placed in inpatient beds, we cannot comment on how many of these short stay patients were admitted under observation status. However, our findings suggest that short-stay admission patients may be candidates for ED-based observation care and may represent a potential cost savings strategy for hospitals and payers and be a revenue opportunity for EDs,11 as ED-based observation care has been shown to decrease the costs associated with inpatient admission.5–7
Medicare is currently employing recovery audit contractors (RAC) to recover perceived “overpayments” made to hospitals for services rendered.3,4,26 Short-stay inpatient hospital admissions deemed retrospectively to be more appropriately categorized as observation care have been identified as a priority of the RAC program.4 As of March 27, 2008, after a 3-year demonstration project, the RAC program successfully collected more than $1.03 billion in previously distributed Medicare payments.26 ED observation services have been shown to be safe and effective and are a “best practice” when managed appropriately.27 As such, they offer a potential alternative to short-stay inpatient hospital admission for appropriate patients.
Accordingly, ED-based observation care may become a more important service line as hospitals attempt to decrease patient readmissions. Financial penalties for “avoidable” hospital readmissions are considered an important part of the new health care reform legislation and are expected to decrease health care costs by approximately $25 billion over 10 years.1 Thus, increasing pressure will be placed on EDs to decrease inpatient readmissions with observation care, which is considered an outpatient service by payers.
Given the manner in which the data are collected, the NHAMCS is subject to the limitations of general survey research, with possible errors in data collection and coding. In particular, data abstractors may have recorded incorrect or incomplete data on type of ED services provided and patient disposition. The effect of this potential error is unknown. In addition, limited questions pertinent to observation care were collected on this survey. However, NHAMCS data have been used widely by others to report epidemiology of a variety of characteristics and conditions, using rigorous methodology and data known to be collected with a high level of accuracy.14
This was the first year that data about observation services was collected. The available information about this patient population is limited. For instance, data concerning the observation admission chief complaint was not collected (only initial triage chief complaint). Additionally, 47% of ED observation services were performed at hospitals without an ED OU (presumably observation services were performed in the ED treatment space). For patients who received ED observation care, the ED length of stay data included ED service time only if the patient was moved to an OU, but combined ED and observation service times if the patient remained in the ED treatment space (i.e., same bed). Finally, if a patient was placed in an inpatient hospital bed, rather than an ED OU, it is not known if the patient was assigned inpatient status or outpatient observation status. As such, we cannot conclude how many of the short-stay admissions were inpatient versus outpatient observation status in an inpatient bed. Regardless, our analysis suggests that this population may be appropriate for ED-based OU care. Future NHAMCS surveys are expected to clarify these issues with new observation care questions.
The prevalence of observation units has increased to over one-third of all U.S. hospitals, with ED staff managing more than half of these units. A small, but similar percentage of ED visits are admitted either to ED observation or to the hospital for short-stay admissions. Patients admitted to the hospital for short-stay admissions have very similar characteristics to those admitted to ED observation, suggesting an opportunity for growth of ED observation care and units. As hospitals and insurers look for ways to decrease costs associated with inpatient admissions, we find that short-stay inpatient admissions may be appropriate for ED-based observation services and may represent a revenue opportunity for EDs. The effect of this disposition location on the cost and quality of comprehensive observation services merits further study.
- 1The White House. Coming Together to Bring Down the Cost of Healthcare Fact Sheet. Available at: http://healthreform.gov/factsheethealthreformstake.pdf. Accessed Jun 23, 2011.
- 2Department of Health & Human Services (DHHS), Centers for Medicare & Medicaid Services (CMS). July 2009 Update of the Hospital Outpatient Prospective Payment System (OPPS), CMS Manual System, Pub 100-02 Medicare Benefit Policy, Transmittal 10, Change Request 6492. Available at: https://www.cms.gov/transmittals/downloads/R107BP.pdf. Accessed Jun 23, 2011.
- 13National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Rep. 2010; (26):1–31., , .
- 14Plan and operation of the National Hospital Ambulatory Medical Survey. Series 1: programs and collection procedures. Vital Health Stat. 1994; (34):1–78., .
- 15The International Classification of Diseases. 9th rev. Clinical Modification, 5th ed. Salt Lake City, UT: Medicode Publications, 1997.
- 24Institute of Medicine. Hospital-Based Emergency Care: At the Breaking Point. Committee on the Future of Emergency Care in the United States Health System. Washington DC: The National Academies Press, 2006.
- 26Jackson Davis HealthCare. The Medicare Recovery Audit Contractor (RAC) Program: An Evaluation of the 3-year Demonstration. Available at: http://www.racaudits.com/uploads/RAC_Demonstration_Evaluation_Report.pdf. Accessed Jun 23, 2011.