1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References


The authors describe national trends in use, reasons for visit, most common diagnoses, and resource utilization in patients admitted to intensive care units (ICUs) from hospital-based emergency departments (EDs) in the United States.


This was an observational study using data from the National Hospital Ambulatory Care Survey, a nationally representative, weighted sample of U.S. hospital-based EDs from 2002 through 2009. The sample comprised a total of 4,267 patients aged 18 years or older admitted to the ICU from the ED, which represent over 14.5 million ED encounters from 2002 through 2009.


Over the study period, ICU admissions from EDs increased from 2.79 million in 2002/2003, to 4.14 million in 2008/2009, an absolute increase of 48.8% and a mean biennial increase of 14.2%. By comparison, overall ED visits increased a mean of 5.8% per biennial period. The three most common diagnoses for ICU admissions were unspecified chest pain, congestive heart failure, and pneumonia. Utilization rates of most tests and services delivered to patients admitted to the ICU from the ED increased, with the largest increase occurring in computed tomography (CT) and magnetic resonance imaging (MRI), which increased from 16.8% in 2002/2003 to 37.4% in 2008/2009, a 6.9% mean biennial increase. Across all years, mean ED length of stay (LOS) for ICU admissions was 304 minutes (95% confidence interval [CI] = 286 to 323 minutes), and mean hospital LOS was 6.6 days (95% CI = 6.2 to 7.0 days). There was no significant change in either mean ED or hospital LOS over the study period.


Intensive care unit admissions from EDs are increasing at a greater rate than both population growth and overall ED visits. ED resource use, specifically advanced diagnostic imaging, has increased markedly among ICU admissions. While mean ED and hospital LOS have not changed significantly, the mean ICU admission spends over 5 hours in the ED prior to transfer to an ICU bed. A greater emphasis on the ED–ICU interface and critical care delivered in the ED may be warranted.


Aumento de los Ingresos en Unidades de Cuidados Intensivos desde los Servicios de Urgencias en Estados Unidos desde 2002 a 2009


Se describen las tendencias nacionales en el uso, las razones para la visita, los diagnósticos más comunes y el consumo de recursos en los pacientes ingresados en las unidades de cuidados intensivos (UCI) desde los servicios de urgencias (SU) hospitalarios en Estados Unidos.


Estudio observacional que usó datos de la National Hospital Ambulatory Care Survey, con una muestra representativa nacional de los SU hospitalarios desde 2002 hasta 2009. La muestra consistió en 4.267 pacientes de 18 años o más ingresados en la UCI desde SU, que representa más de 14,5 millones de visitas a lel SU desde 2002 hasta 2009.


Durante el periodo del estudio, los ingresos en las UCI desde los SU se incrementaron de 2,79 millones en 2002/2003 a 4,14 millones en 2008/2009, un incremento absoluto del 48,8% y bienal del 14,2%. Por comparación, el total de visitas al SU se incrementó un 5,8% por cada bienio. Los tres diagnósticos más frecuentes de ingreso en la UCI fueron el dolor torácico no especificado, la insuficiencia cardiaca congestiva y la neumonía. Los porcentajes de utilización de la mayoría de las pruebas diagnósticas y servicios por parte de los pacientes ingresados en la UCI desde el SU se incrementaron y el mayor aumento ocurrió en tomografías computarizadas y resonancias magnéticas, que aumentaron del 16,8% en 2002/2003 al 37,4% en 2008/2009, un promedio de incremento bienal del 6,9%. A lo largo de todos los años, la estancia media en el SU para los ingresos en UCI fue de 304 minutos (IC 95% = 286 a 323 minutos) y la media de estancia en el hospital fue de 6,6 días (IC95% = 6,2 a 7,0 días). No hubo cambio significativo ni en la media de estancia en SU ni el hospital durante el periodo del estudio.


Los ingresos en la UCI desde el SU están aumentado más que el crecimiento de la población y que las visitas a urgencias. El consumo de recursos del SU, específicamente del diagnóstico por imagen avanzado, se ha incrementado marcadamente en los ingresos enla UCI. Mientras que la estancia media en el SU y en el hospital no ha cambiado significativamente, el promedio de tiempo desde el SU hasta el ingreso en UCI es superior a 5 horas. Se debe garantizar un mayor énfasis en la interfase SU-UCI y en la atención de cuidados críticos en el SU.

The costs of providing care in intensive care units (ICUs) in the United States were estimated at $81.7 billion in 2005, making up about 5% of health care costs.[1] Both the demand for and supply of ICU resources have been increasing: the number of ICU beds grew by 6.5% overall between 2000 and 2005, while mean number of days spent in the ICU grew by 10.6%.[1] ICU admissions are increasing with an aging population that requires more intensive services when they are critically ill and at the end of life.[1-4] Medicare reimbursements for sepsis, which is commonly treated in ICUs, illustrate the increased demand. From 2008 to 2010, hospital days for Medicare beneficiaries with sepsis increased 15.5%.[5, 6]

Several areas of ICU use have been studied, focusing on condition-specific topics such as sepsis,[7-9] international comparisons across health systems,[10-12] and ICU readmission trends.[13-15] One area that has not been well explored is the characteristics of patients admitted to ICUs from hospital-based emergency departments (EDs) in the United States. EDs represented the single largest source (up to 58%) of ICU admissions between 2002 and 2004.[10] The initial care for patients in the ED is vitally important in initial stabilization, rapid diagnosis, and placement in the appropriate level of care. Intensive care initiated in the ED for ICU-bound patients has been associated with large enhancements in survival for patients with sepsis, acute myocardial infarction, trauma, stroke, and post–cardiac arrest syndrome.[16-18]

The increasing intensity of critical care services in the ED is superimposed on a wider emergency care system that is already strained and overcrowded.[19] Between 1993 and 2006, overall hospital admissions increased by 15%, whereas admissions from the ED increased by 50.4%.[20] Demand for ED-based services is increasing at a rapid rate that is outpacing population growth, while, at the same time, the number of hospitals is declining.[21, 22] To our knowledge, no studies have explored the broad demographic changes in ICU patients admitted from U.S. EDs. In this study, we used data from a large, nationally representative sample of EDs in the United States to explore demographic changes, along with use and treatment trends, in patients admitted to ICUs from 2002 through 2009.


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Study Design

We used data from the 2002 through 2009 National Hospital Ambulatory Care Survey (NHAMCS). NHAMCS data have been collected by the Centers for Disease Control and Prevention National Center for Health Statistics (NCHS) annually since 1992. NHAMCS data use a multistate estimation procedure to form a nationally representative sample of ED visits in the United States, from which unbiased national estimates can be derived. A detailed description of NHAMCS is provided elsewhere.[23]

The NHAMCS contains publically available data that contain no patient-identifiable information. Therefore, this study was deemed to be not human subjects' research by the institutional review board at George Washington University.

Study Setting and Population

Samples from NHAMCS 2002 through 2009 comprised a total of 286,422 ED encounters. Taken together, these encounters represent over 943.7 million ED visits nationally over this time period. Due to sample sizes of patients admitted to ICUs and minimum cell requirements of the complex survey design of the NHAMCS for subgroup reporting, we conducted analyses by combining data from two consecutive yearly surveys. Data were combined into four time blocks for 2002/2003, 2004/2005, 2006/2007, and 2008/2009. Through the study period, we computed mean changes between sets of years (e.g., 2002/2003 vs. 2004/2005) and report the mean biennial percent change.

Study Protocol

First, we examined the data by separating encounters by age group and excluding patients younger than 18 years of age. Data were separated into the following age groups: 18–34, 35–44, 45–54, 55–64, 65–74, 75–84, and 85 years or older. We then tabulated the number of patients of all ages by demographics, payment source, geographic region, hospital location, and hospital type. We also examined the numbers of patients who presented with a variety of physiologic markers of illness to assess if the severity of the physiology has changed over time. This included pulse oximetry levels < 93%, heart rates > 89 beats/min, and temperatures > 100.4 or < 96.8°F. These cutoff values were derived from the systemic inflammatory response syndrome (SIRS) criteria.[24]

Next, we compiled estimates of resource utilization in the ED by patients admitted to ICUs. We tabulated the numbers of visits in which patients received any imaging study: x-rays; ultrasound; computed tomography (CT), magnetic resonance imaging (MRI), or both, which were combined due to NHAMCS data restrictions (so abbreviated to CT/MRI); and electrocardiograms (ECGs). In addition, we tabulated the numbers of visits in which patients received any laboratory blood test or urinalysis. We also calculated estimates of the numbers of visits in which patients received specific services in the ED, such as medications, intravenous (IV) fluids, any ED procedure, and intubation. To assess trends in ED crowding, we also calculated the mean length of ED encounter, mean waiting time, and mean length of hospital stay for visits leading to ICU admission in each biennial period. This was done by calculating the mean value with the weighted survey data with 95% confidence intervals (CIs). Median values, which are more commonly used with time data, could not be estimated with weighted survey data. To assess the frequency of visit by complaint and diagnosis, we combined data across all years in our sample and tabulated the top ten most frequent first-listed reasons for visit and first-listed diagnoses. We were unable to assess changes in reasons for visit and diagnoses over the years because of the minimum cell sizes required by NCHS to make national estimates.

Data Analysis

All data were tabulated using Stata, version 12 (StataCorp, College Park, TX) and used complex survey analytic methods. To assess for trends over time, we used weighted linear regression models, with the year as the independent variable and each study variable as the dependent variable. Due to nonnormal distribution of the length of visit, waiting time, and hospital length of stay (LOS) variables, we used a natural log transformation of these variables when assessing trends in the regression analysis. A p-value ≤ 0.05 was considered significant.


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Main Results

From 2002 through 2009, NHAMCS included data on 4,267 separate ED encounters that led to ICU admissions, representing over 13.9 million U.S. ED ICU admissions using the population weighting in the data set. Over the study period, ICU admissions from EDs increased from 2.79 million in 2002/2003 to 4.14 million in 2008/2009, an absolute increase of 48.8% and a mean biennial increase of 14.2% (p = 0.005 for trend). By comparison, overall ED visits increased a mean of 5.8% per biennial period (p = 0.001 for trend). The proportion of ED visits that resulted in ICU admissions increased from 1.7% in 2002/2003 to 2.1% in 2008/2009 (p = 0.005 for trend). The proportion in the age groups studied was mostly stable; however, the largest growth was among visits by patients 85 years or older, which grew at a mean biennial rate of 25.2% (p = 0.04 for trend; Table 1).

Table 1. Patient and Hospital Characteristics of Patients Admitted to ICUs from EDs, 2002–2009, with 95% CIs
Variable2002/20032004/20052006/20072008/20092002–2009: Mean Biennial Increase2002–2009; Mean Biennial Growthp-value for Test for Trend
  1. ICUs = intensive care units.

Patient characteristics
Total estimated ICU admissions
Overall2,785,827 (2,351,286–3,220,368)3,160,833 (2,494,229–3,827,437)3,789,524 (3,041,511–4,537,537)4,144,785 (3,289,273–5,000,297)452,98614.2%0.005
Total estimated ED visits by patients >17 years
Overall167,919,499 (154,434,077–181,404,921)168,768,295 (145,026,930–192,509,660)182,971,923 (155,522,534–210,421,312)198,586,819 (167,891,373–229,282,265)10,222,4405.8%0.001
Proportion of ED visits resulting in ICU admissions (95% CI)
Overall1.7% (1.4–1.9)1.9% (1.6–2.2)2.1% (1.8–2.4)2.1% (1.8–2.4)0.1%7.4%0.005
Total estimated ICU admissions, by age, yr
85+303,287 (219,413–387,161)319,823 218,539–421,107)521,766 (389,060–654,472)558,503 (418,540–698,466)85,07225.2%0.04
75–84550,373 (429,727–671,019)779,324 (598,595–960,053)688,162 (522,164–854,160)754,993 (554,146–955,840)68,20713.2%0.15
65–74526,349 (402,775–649,923)472,872 (359,718–586,026)576,659 (434,865–718,543)783,390 (604,384–962,396)85,68015.9%0.83
55–64451,901 (346,888–556,914)594,960 (424,033–765,887)658,300 (503,763–866,837)605,272 (406,326–804,218)51,12411.4%0.28
45–54501,434 (384,097–618,771)430,771 (313,581–547,961)666,536 (492,222–840,850)720,991 (527,859–914,123)73,18616.3%0.83
35–44240,513 (167,753–312,273)277,007 (172,177–381,837)305,315 (211,591–399,039)353,037 (227,730–478,344)37,50813.7%0.38
18–34211,970 (139,816–284,124)286,076 (184,981–387,171)345,786 (216,225–475,347)368,599 (262,190–475,008)52,21020.8%0.39
Mean age, yr62.6 (60.9–64.4)62.7 (61.0–64.5)62.3 (60.4–64.1)63.0 (61.4–64.7)0.10.2% 
Female1,338,945 (1,094,462–1,583,428)1,440,239 (1,135,489–1,744,989)1,851,125 (1,492,801–2,209,449)2,043,324 (1,536,860–2,549,788)234,79315.5%0.14
Male1,446,882 (1,198,657–1,695,107)1,720,594 (1,341,889–2,099,299)1,938,399 (1,529,104–2,347,694)2,101,461 (1,695,551–2,507,371)218,19313.3%0.62
 Race and ethnicity
White2,314,721 (1,952,000–2,677,442)2,488,907 (1,970,318–3,007,496)2,857,541 (2,264,124–3,450,958)3,100,451 (2,474,838–3,726,064)261,91010.3%0.005
Black386,719 (286,913–486,525)558,739 (396,509–720,969)829,389 (577,972–1,080,806)843,013 (582,504–1,103,522)152,09831.5%0.02
Other84,387 (47,517–121,257)113,187 (40,644–185,730)102,594 (49,061–156,127)201,321 (87,797–314,845)38,97840.3%0.14
Source of payment
Private701,163 (555,465–846,861)1,105,383 (837,981–1,372,785)1,512,292 (1,133,606–1,890,978)1,740,051 (1,320,948–2,159,154)346,29636.5%0.001
Medicare1,328,179 (1,108,704–1,547,654)1,546,439 (1,195,719–1,897,159)1,934,200 (1,559,997–2,308,403)2,280,866 (1,807,668–2,754,064)317,56219.8%0.007
Medicaid285,283 (195,640–374,926)485,293 (346,158–624,428)722,851 (565,070–880,632)788,342 (610,236–966,448)167,68642.7%0.001
Hospital characteristics
 Geographic region
Northeast422,191 (305,403–538,979)859,268 (573,190–1,145,346)839,575 (562,448–1,116,702)869,051 (439,764–1,298,338)148,95334.9%0.50
Midwest822,664 (630,556–1,014,772)662,264 (433,309–891,219)958,358 (630,152–1,286,564)903,150 (580,752–1,225,548)26,8296.5%0.30
South976,284 (673,702–1,278,866)1,067,063 (597,457–1,536,669)1,345,667 (8,551,492–1,839,842)1,378,771 (972,128–1,785,414)134,16212.6%0.75
West564,888 (438,836–690,540)572,238 (328,619–815,857)645,924 (343,407–948,441)993,813 (505,905–1,481,721)142,97522.7%0.52
Urban2,167,643 (1,810,632–2,524,654)2,855,953 (2,206,076–3,505,830)3,408,665 (2,672,044–4,145,286)3,649,385 (2,829,733–4,469,037)493,91419.4%0.14
Nonurban618,184 (289,856–946,512)304,880 (114,348–495,412)380,859 (153,508–608,210)495,400 (154,531–836,269)-40,9281.4%0.14
 Hospital type
Nonprofit2,105,028 (1,764,816–2,445,240)2,513,439 (1,943,977–3,082,901)2,876,488 (2,270,845–3,482,131)3,127,245 (2,436,990–3,817,500)340,73914.2%0.86
Government444,683 (274,417–614,949)451,051 (269,761–632,341)548,717 (266,538–830,896)459,610 (248,927–670,293)4,9762.3%0.21
For-profit236,116 (102,103–370,129)196,343 (63,205–329,481)364,319 (137,577–591,061)557,930 (178,287–937,573)107,27140.6%0.19

Demographic and Hospital Trends

The mean age of patients admitted to ICUs ranged from 62.6 years (95% CI = 60.9 to 64.4 years) in 2002/2003, to 63.0 years (95% CI = 61.4 to 64.7 years) in 2008/2009, with no significant trend. With regard to race, the number of visits by white patients admitted to ICUs from EDs increased significantly at a mean of 10.3% biennially (p = 0.005 for trend), while there were larger increases in the numbers of nonwhite patient visits. Specifically, the numbers of African American patient visits increased 31.5% every 2 years over the study period (p = 0.02 for trend), while visits by patients from other races increased 40.1%, although there was no significant trend. Medicare was the payment source associated with the smallest increase in visits of 19.8% every 2 years (p = 0.007 for trend), while private insurance increased 36.5% (p = 0.001 for trend), and admissions among Medicaid-insured patient visits increased 42.7% (p = 0.001) on average. There were no significant trends in regional distribution for ICU admissions over the study period. There were also no significant trends among hospital types (Table 1).

Resource Use and ED Crowding

There was a consistent increase in the number of tests and services ordered for ICU admissions over the study period. The number of visits that received any imaging increased a mean of 16.8% per biennial period, while the proportion of visits that received any imaging increased 1.8% (p = 0.01 for trend). The largest increase was observed in CT/MRI, in which the mean biennial increase was 50.3%, while the proportion of patient visits where a CT or MRI was performed increased 6.9% per 2-year period (p = 0.001 for trend). For comparison, we calculated the rates of growth in CT/MRI utilization for overall ED encounters. The number of visits in which CT/MRIs were ordered increased 32.1% on average for each biennial period (p = 0.001 for trend). The number of patient visits in which patients received urinalysis increased by a mean of 25.8% per biennial period (p = 0.001 for trend; Table 2).

Table 2. ED Resource Utilization in Patients Admitted to ICUs, 2002–2009
 2002–20032004–20052006–20072008–20092002–2009: Mean Percent Increasep-value for Test for Trend
  1. Data are reported as N or% (95% CI).

  2. ECG = electrocardiogram; ICUs = intensive care units; MRI = magnetic resonance imaging.

Physiologic variables
 Pulse oximetry < 93%
Number of patient visits820,213 (626,684–1,013,742)1,077,277 (771,458–1,383,096)31.30.001
Proportion of patient visits21.6% (18.0–25.9)26.0% (21.0–31.7)4.4 
 Heart rate > 89 beats/min.
Number of patient visits1,484,747 (1,234,411–1,735,083)1,681,960 (1,377,395–1,986,525)1,911,993 (1,562,294–2,261,692)2,136,592 (1,772,998–2,500,196)12.90.47
Proportion of patient visits53.3% (48.4–58.2)53.2% (49.2–57.2)50.5% (46.6–54.3)51.6% (46.8–56.3)–0.6 
 Temperature > 100.4 or < 96.8 F
Number of patient visits881,372 (704,343–1,058,401)932,549 (723,026–1,142,072)1,096,605 (885,194–1,308,016)1,072,016 (846,106–1,297,926)7.10.04
Proportion of patient visits31.6% (26.8–36.9)29.5% (25.3–34.1)28.9% (25.7–32.4)25.9% (22.3–29.8)–1.9 
Tests ordered
 Any imaging
Number of patient visits2,139,274 (1,800,310–2,478,238)2,616,566 (2,108,206–3,124,926)2,948,495 (2,478,318–3,418,672)3,402,157 (2,784,465–4,019,849)16.80.01
Proportion of patient visits76.7% (71.9–81.1)82.8% (79.3–85.8)77.8% (74.4–80.9)82.1% (78.5–85.2)1.8 
Number of patient visits1,966,348 (1,641,267–2,291,429)2,358,947 (1,894,428–2,823,466)2,559,565 (2,139,442–2,979,688)2,867,400 (2,340,547–3,394,253)13.50.29
Proportion of patient visits70.6% (65.4–75.3)74.6% (70.3–78.6)67.5% (63.7–71.1)69.2% (65.1–73.0)–0.5 
Number of patient visits467,557 (353,781–581,333)835,777 (648,192–1,023,362)1,143,835 (871,807–1,415,863)1,548,089 (1,206,546–1,889,632)50.30.001
Proportion of patient visits16.8% (13.7–20.3)26.4% (23.2–30.0)30.2% (26.5–34.1)37.4% (32.9–42.1)6.9 
Number of patient visits128,037 (70,425–185-649)107,704 (47,140–168,268)127,058 (59,372–194-744)149,691 (84,053–215,329)6.60.66
Proportion of patient visits4.6% (3.0–7.1)3.4% (2.0–5.9)3.4% (2.1–5.4)3.6% (2.5–5.3)-0.3 
Number of patient visits2,142,509 (1,797,330–2,487,688)2,468,327 (2,014,716–2,921,938)2,699,698 (2,241,902–3,157,494)3,020,229 (2,476,051–3,564,407)12.20.05
Proportion of patient visits76.9% (72.3–80.9)78.1% (74.1–81.6)71.2% (67.1–75.1)72.9% (68.9–76.5)–1.3 
Laboratory tests
 Blood labs
Number of patient visits2,494,345 (2,091,652–2,897,038)2,852,244 (2,313,853–3,394,635)3,389,918 (2,814,159–3,965,677)3,785,271 (3,098,425–4,472,117)15.00.41
Proportion of patient visits89.5% (85.0–92.8)90.3% (87.4–92.6)89.5% (86.6–91.8)91.3% (88.5–93.5)0.6 
Number of patient visits902,326 (703,993–1,100,659)1,034,695 (794,147–1,275,243)1,457,372 (1,180,951–1,733,793)1,777,225 (1,431,322–2,123,128)25.80.001
Proportion of patient visits32.4% (28.2–36.9)32.7% (28.9–36.8)38.5% (34.2–42.9)42.9% (39.2–46.7)3.5 
Services performed
 Received medicines in ED
Number of patient visits2,386,688 (1,999,845–2,773,531)2,721,148 (2,192,347–3,249,949)3,172,363 (2,609,394–3,735,332)3,605,962 (2,953,619–4,258,305)14.80.001
Proportion of patient visits85.7% (80.8–89.4)86.1% (82.4–89.1)83.7% (80.3–86.7)87.0% (83.9–89.6)0.4 
Mean number of medicines received3.39 (3.16–3.62)3.64 (3.33–3.92)3.35 (3.12–3.58)3.58 (3.31–3.85)0.40
 Received IV fluids in ED
Number of patient visits1,986,438 (1,674,607–2,298,269)2,398,197 (1,893,099–2,903,295)2,849,613 (2,301,431–3,397,795)3,436,413 (2,777,036–4,095,790)20.00.002
Proportion of patient visits71.3% (65.6–76.4)75.9% (70.6–80.5)75.2% (70.4–79.4)82.9% (78.1–86.8)3.9 
 Received procedure in ED
Number of patient visits2,230,978 (1,896,227–2,565,729)2,630,074 (2,101,513–3,158,635)3,157,933 (2,572,804–3,743,062)3,646,120 (2,980,396–4,311,844)17.80.04
Proportion of patient visits80.1% (74.9–84.4)83.2% (78.6–87.0)83.3% (79.4–86.7)88.0% (83.8–91.2)2.6 
 Intubated in ED
Number of patient visits238,040 (170,415–305,665)161,506 (108,423–214,589)221,955 (148,505–295,405)225,361 (143,449–307,273)2.30.13
Proportion of patient visits8.5% (6.5–11.2)5.1% (3.7–7.0)5.9% (4.3–7.9)5.4% (3.8–7.7)–1.0 
 ED crowding measures
Mean ED LOS, min285.1 (253.5–316.6)319.0 (285.6–352.3)302.8 (277.7–327.8)307.1 (279.1–335.2)2.80.18
Mean waiting time, min16.2 (13.1–19.3)15.6 (14.0–17.2)16.6 (15.0–18.1)17.6 (16.1–19.2)2.90.12
Mean hospital LOS, days6.5 (5.9–7.1)6.5 (5.9–7.1)6.8 (6.2–7.4)2.20.26

The proportion of patients who received medications increased 0.4% (p = 0.001 for trend), while the mean number of medicines administered per patient visit did not change significantly. IV fluid utilization increased at a mean rate of 3.9% per biennial period (p = 0.002 for trend), while the number of patient visits receiving IV fluids increased 20.0%. The number of patient visits receiving a procedure increased 17.8% per biennial period (p = 0.04 for trend).

Measures of crowding and hospital LOS were not significantly different over the study period. The mean LOS in the ED for patients admitted to ICUs was 286 minutes in 2002/2003 and 307 minutes in 2008/2009, and there was no significant trend across study years. Mean waiting time to be seen by a provider also remained stable, with estimates ranging from 16 minutes in 2002/2003 to 15 minutes in 2008/2009. Mean hospital LOS for patients admitted to ICUs from the ED did not increase significantly from 2004/2005 (the first years in which these data were available) to 2008/2009.

Physiologic Characteristics

The proportion of visits in which patients presented with heart rates greater than 89 beats/min did not change significantly over the study period. The number of visits in which patients met SIRS temperature criteria increased at a mean rate of 7.1% per biennial period (p = 0.04 for trend). The number of visits by patients admitted to ICUs from EDs whose pulse oximetry levels were below 93% increased 31.3% in 2008/2009, relative to 2006/2007 levels (the first years for which these data were available), while the proportion of patient visits meeting this criterion increased 4.4% (p = 0.001 for trend).

Most Common Reasons for Visit and Diagnoses Leading to ICU Admission

The top 10 reasons for visit composed 48.8% of all ICU admissions from EDs over the study period, while the top 10 diagnoses composed only 34.6% of total ICU admissions. The three most common reasons for admission were chest pain/soreness, shortness of breath/dyspnea, and abdominal pain (Table 3). The three most common diagnoses were unspecified chest pain, congestive heart failure, and pneumonia of unspecified origin (Table 4).

Table 3. Top 10 Reasons for Visit to EDs Leading to ICU Admissions, 2002–2009
Reason for VisitEstimated NumberPercentage of Total ICU Admissions From EDs
  1. ICU = intensive care unit.

Chest pain, soreness2,315,92715.9
Shortness of breath1,719,46511.8
Abdominal pain467,9163.2
General weakness458,0003.2
Psychological symptoms345,3952.4
Unintentional overdose238,5761.6
Table 4. Top 10 Diagnoses in EDs Leading to ICU Admissions, 2002–2009
DiagnosisEstimated NumberPercentage of Total ICU Admissions
  1. ICU = intensive care unit.

Chest pain, unspecified1,341,6779.2
Congestive heart failure779,1825.4
Pneumonia, organism unspecified554,1393.8
Hemorrhage of gastrointestinal tract, unspecified496,5033.4
Intermediate coronary syndrome399,3842.8
Cerebral artery occlusion, unspecified with cerebral infarction364,4882.5
Acute myocardial infarction of unspecified site, episode of care unspecified282,6371.9
Other respiratory abnormalities271,8351.9
Syncope and collapse271,0581.9
Acute respiratory failure264,3371.9


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Increases in ED use have outpaced population growth over the past 10 years. As a subset of ED visits, ICU admissions are increasing at a rate nearly three times faster than general ED visit rates. Our data are consistent with those of other studies documenting increased demand for ICU services in hospitals across the world.[1, 10, 12] Some of the increases may be accounted for by rising rates of ICU admissions among the 85 years and older age group; however, increasing ICU admissions among younger patients also account for much of the increase. Two of the major contributors to higher ICU use are admissions by nonwhite patients and patients with Medicaid insurance. These trends by insurance status are similar to wider trends in overall ED use, with Medicaid increasing at a greater rate than Medicare.[1] Higher rates of general ED and ICU use in these groups may be a symptom of less access to primary care and preventive services, contributing to both increased use of the ED and increasingly higher severity presentations requiring ICU-level care.

Consistent with wider trends in ED service use, services and tests ordered during visits by patients admitted to ICUs also increased substantially over the study period, indicating higher service intensity for ICU admissions in the ED.[22] Several studies have demonstrated that high-resource care in the ED, such as early goal-directed therapy in septic shock and targeted temperature management in the post–cardiac arrest syndrome, can produce dramatic improvements in survival.[16-18] In many hospitals, these treatments are delivered in the ED because of often-constrained ICU resources, provider desires, or ICU provider availability. Our study demonstrated a clear trend toward a large increase in the intensity of the services provided in the ED, such as IV fluids and medications. In addition, the increased use of diagnostic testing such as CT/MRI and laboratory tests in EDs has been documented in other studies, a trend that appears to be similar among ICU admissions.[22, 25, 26]

Emergency department LOS over the study period remained stable, with the mean stay in the ED for ICU patients being more than 5 hours. ED-based ICU care can consume a tremendous amount of ED staffing resources. This can result in further crowding, as ED staff are required to care for critically ill patients, sometimes for prolonged periods of time, while ICU beds and staff become available. ED crowding is associated with poor outcomes in several studies.[27-29] Furthermore, the relationship between prolonged ED LOS and mortality has been documented in patients admitted to ICU settings.[30] This suggests that if the trend toward more and more ED-based critical care continues, the emergency care system may require broad changes in staffing, training, and space. Some of these trends are already occurring, with more ED providers becoming fellowship-trained in critical care medicine.[31] However, the broader system capacity issues still need to be addressed at the local, regional, and national levels.

The most common reasons for ED visits leading to ICU admissions were symptoms representing potentially serious cardiac, respiratory, and gastrointestinal emergencies, with chest pain, breathing difficulty, and abdominal pain being the most common, which also reflect the most common diagnoses. Despite this, the top 10 complaints and diagnoses only made up 50 and 35%, respectively, of all ICU admissions. This demonstrates the heterogeneity of critically ill patients cared for in ED settings, further underscoring the need for critical care training in emergency medicine residency and beyond.

It is notable that, while use of time-intensive diagnostics and procedures increased over the study period, the mean LOS in the ED did not significantly change. This finding may suggest that delays in transfer from ED to ICU may be more dependent upon the availability of ICU beds than the level of resource utilization in the ED. Future research could examine this potentially important relation as more recent data regarding ICU capacity become available.


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

The NHAMCS is a weighted survey design that can be used to generate national-level estimates. Given that ICU admission is, overall, a relatively rare outcome of ED encounters, sample size was a limitation in some subgroup analyses. While each reported estimate met published NCHS criteria for generating reliable and valid estimates, the small numbers in each analytic cell may reduce the precision of the estimates. Information bias may be present due to changes in data collection techniques; in particular, between the 2004 and 2005 versions of the survey, the survey instruments changed in the way that ICU disposition was written on the form. In 2004 and before, it was written as “Admit to ICU/CCU” and in 2005, this changed to admission to “critical care unit.” It is possible this could have changed the local abstractor decisions with regard to what was classified as an ICU admission. However, the clear increase in visits after 2005 makes the survey change unlikely to account for the observed increases in ICU admissions.


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References

Our results indicate that intensive care unit admissions from EDs are increasing at an even greater rate than overall ED visits. In addition, the proportion of ED visits that result in intensive care unit admissions is growing. As the U.S. population ages, these growth trends should be considered when planning for future ICU capacity. In addition, an increased emphasis on the ED–intensive care unit interface and critical care training of emergency providers may be warranted, as many patients stay in EDs for prolonged periods of time.


  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. References