Article, Emergency Medicine

Lower mortality in sepsis patients admitted through the ED vs direct admission

Unlabelled imageAmerican Journal of Emergency Medicine (2012) 30, 432-439

Original Contribution

Lower mortality in sepsis patients admitted through the ED vs direct admission?,??

Emilie S. Powell MD, MBA a,b,?, Rahul K. Khare MD a,b,

D. Mark Courtney MD a, Joe Feinglass PhD b

aNorthwestern University, Department of Emergency Medicine, Feinberg School of Medicine, Chicago, IL 602611, USA

bInstitute for Healthcare Studies and Division of General Internal Medicine, Northwestern University, Feinberg School of Medicine, Chicago, IL 602611, USA

Received 21 October 2010; revised 4 January 2011; accepted 9 January 2011

Abstract

Purpose: Early aggressive resuscitation in patients with severe sepsis decreases mortality but requires extensive time and resources. This study analyzes if patients with sepsis admitted through the emergency department (ED) have lower inpatient mortality than do patients admitted directly to the hospital.

Procedures: We performed a cross-sectional analysis of hospitalizations with a principal diagnosis of sepsis in institutions with an annual minimum of 25 ED and 25 direct admissions for sepsis, using data from the 2008 Nationwide Inpatient Sample. Analyses were controlled for patient and hospital characteristics and examined the likelihood of either early (2-day postadmission) or overall inpatient mortality.

Findings: Of 98 896 hospitalizations with a principal diagnosis of sepsis, from 290 hospitals, 80,301 were admitted through the ED and 18 595 directly to the hospital. Overall sepsis inpatient mortality was 17.1% for ED admissions and 19.7% for direct admissions (P b .001). Overall early sepsis mortality was 6.9%: 6.8% for ED admissions and 7.4% for direct admissions (P = .005). Emergency department patients had a greater proportion of comorbid conditions, were more likely to have Medicaid or be uninsured (12.5% vs 8.4%; P b .001), and were more likely to be admitted to urban, large bed-size, or teaching hospitals (P b .001). The risk-adjusted odds ratio for overall mortality for ED admissions was 0.83 (95% confidence interval, 0.80-0.87) and 0.92 for early mortality (95% confidence interval, 0.86-0.98), as compared with direct admissions to the hospital. Conclusion: Admission for sepsis through the ED was associated with lower early and overall inpatient mortality in this large national sample.

(C) 2012

? This work has been presented at the Academy Health Annual Meeting in Boston, MA, in June 2010 and at the ACEP Scientific Assembly in Las Vegas, ND, in September 2010.

?? Dr Powell and Dr Khare were supported by National Research Service Award postdoctoral fellowship grants through the Institute for Healthcare

Studies at Northwestern University under institutional awards from the Agency for Healthcare Research and Quality (T-32 HS 000078 and F-32 HS 17876-01).

* Corresponding author. Tel.: +1 312 695 7521 (pager); fax: +1 312 926 6274.

E-mail address: [email protected] (E.S. Powell).

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2011.01.011

Introduction

Sepsis is a broad clinical condition defined by the presence of infection and a systemic inflammatory response and exists along a continuum with severe sepsis and septic shock [1]. It affects more than 750 000 US residents each year and is associated with a mortality rate of up to 50% to 60% once septic shock is present [2,3].

Specific interventions, including early antibiotics [4,5] and early goal directed therapy [6], have been shown to reduce mortality in severe sepsis and septic shock. The Institute for Healthcare Improvement and the Surviving Sepsis Campaign advocate an aggressive Resuscitation protocol within the first 6 hours of patient presentation [7]. These interventions may be both time and resource intensive, and significant challenges exist to their imple- mentation, including difficulties in diagnosing sepsis and high levels of resources and staff required for appropriate management [8-12].

Patients suspected of having sepsis present to the hospital and are initially resuscitated in the emergency department (ED) or are admitted directly to the inpatient hospital. It is unclear if the location of initial presentation and sepsis resuscitation impacts mortality. If the location of initial presentation impacts mortality, patients suspected of being critically ill with sepsis should be directed toward this location. We hypothesize that admission through the ED may be associated with lower inpatient mortality rates due to emergency physicians’ experience in the diagnosis and treatment of undifferentiated patients and immediate avail- ability of physicians, nurses, and resources to resuscitate septic patients.

The goal of this investigation was to use a large national hospital discharge data set to determine if there was an association between inpatient mortality and route of admission, through the ED or direct admission to the hospital, for patients with a principal diagnosis of sepsis.

Patient variables“>Materials and methods

Study design

We performed a cross-sectional analysis of the 2008 Healthcare Cost and Utilization Project Nationwide Inpa- tient Sample (NIS). The 2008 NIS includes 42 states, representative of approximately 20% of US community hospitals, and is the largest publicly available all-payer inpatient database in the United States. It is provided by the US Agency for Healthcare Research and Quality. Addi- tional detail on the NIS can be found on the Healthcare Cost and Utilization Project Web site (http://www.hcup- us.ahrq.gov/). The Northwestern University Institutional Review Board found this study of de-identified data exempt from full review.

Study setting and population

Admissions for sepsis were identified based on the principal diagnosis International Classification of Diseases, Ninth Revision (ICD-9) codes previously defined by Martin et al [13] to detect patients with sepsis from hospital discharge data: 038 (septicemia), 020.2 (septicemic plague),

790.7 (bacteremia), 117.9 (disseminated fungal infection),

112.5 (disseminated candidal infection), or 112.81 (dissem- inated fungal endocarditis. In addition, 2 ICD-9 codes for sepsis were added in 2006 and included in the analysis:

995.95 (severe sepsis) and 785.52 (septic shock).

For the primary analysis, admissions were excluded if the patient was transferred into or out of the hospital, ensuring that outcomes could be attributed to a single institution. To ensure that we analyzed only hospitals with a significant number of sepsis hospitalizations through both admission routes, we then excluded admissions from hospitals with less than 25 ED or less than 25 direct sepsis admissions in 2008. In doing so, we analyzed hospitals with a volume threshold of at least 1 sepsis hospitalization every other week per route of admission.

Of 159 370 hospitalizations in 2008 with a principal diagnosis of sepsis, 8.1% were excluded because of transferring into or out of the institution, and 32.7% of the remaining hospitalizations were excluded because of admis- sion to hospitals with less than 25 ED or direct sepsis admissions. Of the 657 excluded hospitals, 324 were rural institutions, 354 were small bed-size hospitals, 176 were medium bed-size hospitals, and 127 were large bed-size hospitals. Of the remaining 98 896 cases included in the final analysis, from 290 hospitals, 80 301 were admitted through the ED and 18 595 were admitted directly to the hospital (Fig. 1). The median number of direct sepsis admissions was 44 per year (25%-75% interquartile range, 30-75), and the median number of ED sepsis admissions per hospital was 233 per year (25%-75% interquartile range, 130-368).

Study protocol

Patient variables

Nationwide Inpatient Sample data included admission source (ED vs direct admission), discharge status (alive or dead), sex, age (divided into 5 age groups), and race and ethnicity (black, Hispanic, white/other, or missing). Race and ethnicity data were missing for 22.8% of the study population from states where this information was not mandated. Nationwide Inpatient Sample data also included length of stay, primary payer source, and patients’ median zip code Household income (divided into quartiles). As an additional proxy for Lower socioeconomic status, we coded patients younger than 65 years with a primary payer status of Medicaid or self-pay (uninsured). We used the Elixhauser ICD-9 secondary Diagnosis codes to identify 30 conditions likely to have been present on admission [14]. Elixhauser

Fig. 1 Study population selection. a399 patients met both exclusion criteria: they were both transferred into the hospital from another acute care hospital and transferred to another acute care hospital. b3004 patients met both exclusion criteria, admission to hospital with ED admission sepsis case volume less than 25 and admission to hospital with direct admission sepsis case volume less than 25, and therefore a total of 47 980 patients were excluded for low sepsis case volume (30.1% of original sample).

comorbid conditions have often been used in mortality models to adjust for severity of preexisting illness [14-16]. We first tested the univariate association of each Elixhauser comorbid condition with inpatient mortality and included the conditions found to have significant associations with greater likelihood of inpatient death in the analysis as binary variables (P b .05). We found that 12 of the 30 Elixhauser comorbid conditions were significantly associated with higher mortality (P b .05). These conditions included alcohol abuse, lymphoma, metastatic cancer, cancer without metas- tasis, chronic heart failure, coagulopathy, fluid and electro- lyte disorders, liver disease, peripheral vascular disease, pulmonary circulation disorders, renal failure, and weight loss (Appendix A, online).

Hospital variables

Hospital level data available in the NIS included hospital bed-size (categorized by NIS as small, medium, or large based on number of beds and hospital location, teaching status, and region), teaching status (member of the Council of Teaching Hospitals or not), and location (rural vs urban). Hospitals were empirically divided equally into

4 approximately equal annual hospital sepsis case volume quartiles (1: 25-171, 2: 172-287, 3: 288-438, 4: N438

sepsis cases per year).

Key outcome measures

The outcome measures in this study were overall inpatient mortality (death at any time during the hospital stay) and early inpatient mortality (death within the first 2 days of admission). Early mortality may better reflect the impact of Initial resuscitation by minimizing confounding events that could have occurred later in the hospital stay. Evaluating early inpatient mortality at a fixed interval from admission also avoided the potential pitfall of inpatient mortality related to longer lengths of stay and factors remote from the initial presentation.

Data analysis

Analyses were conducted with SPSS version 18.0 (SPSS Inc, Chicago, IL) and STATA version 10.0 (STATA Corp, College Station, TX). Univariate patient differences by route of admission were evaluated by c2 test. A population- averaged logistic regression model with an exchangeable working correlation matrix (generalized estimating equation) was used to test the effect of route of admission on the likelihood of early or overall inpatient mortality after controlling for the simultaneous effects of age, sex, significant comorbid conditions, median zip code income, payer status, hospital location, hospital bed-size, teaching status, and hospital sepsis case volume. This approach uses admissions as the unit of analysis and adjusts significance tests for clustering of patients within hospitals [17]. Race and ethnicity was not included in the adjusted analysis because the data were missing for 22.8% of hospitalizations. Results are reported as odds ratios (ORs) and 95% confidence intervals (95% CI).

A sensitivity analysis was conducted to determine the extent to which results were sensitive to the exclusions described above by running the model with all admissions, inclusive of hospitals with less than 25 ED sepsis admissions or direct sepsis admissions.

Results

Differences in patient and hospital characteristics

Table 1 displays patient and hospital characteristic differences between patients admitted through the ED vs those admitted directly to the hospital. Fig. 2 demonstrates the range of inpatient overall mortality across all institutions in this sample. There was a significant difference in overall inpatient sepsis mortality by route of admission: 17.1% for patients admitted through the ED and 19.7% for those

Table 1 Sepsis admissions in the 2008 NIS (n = 98 896)

Characteristic

Total sample (%), n = 98 896

ED

admissions (%), n = 80 301

Direct admissions n = 18 595

(%),

P

Unadjusted mortality rate Overall inpatient mortality

Early inpatient (day 2) mortality Patient characteristics

Age (y) 18-54

55-66

67-76

77-83

N83

Female sex Race a White/Other Black Hispanic

Selected comorbid conditions associated with inpatient mortality Alcohol abuse

Cancer: lymphoma Cancer: metastatic cancer Cancer: without metastasis Chronic heart failure Coagulopathy

Fluid and Electrolyte disorder Liver failure

Peripheral vascular disease Pulmonary circulation disorders Renal failure

Weight loss

Zip code median-income quartile 1 (lowest)

2

3

4 (highest) Medicaid/Self-pay Hospital characteristics Rural location

Bed size Small Medium Large

Teaching hospital

Hospital sepsis volume quartiles 1: 25-171 cases per year

2: 172-287 cases per year

3: 288-438 cases per year 4: N438 cases per year

17.6

17.1

19.7

b.001

6.9

6.8

7.4

.005

b.001

19.3

19.9

16.9

19.1

19.0

19.5

20.7

20.8

20.3

19.3

19.3

19.5

21.6

21.1

23.9

53.1

53.0

53.6

.13

b.001

77.0

76.7

79.5

13.8

13.9

13.7

9.2

9.4

6.8

3.0

3.3

1.9

b.001

2.2

2.0

2.8

b.001

5.1

5.0

5.4

.05

4.0

4.0

4.2

.18

21.6

21.8

20.5

b.001

10.7

11.0

9.4

b.001

54.2

55.3

49.3

b.001

4.4

4.5

3.8

b.001

8.3

8.3

8.2

.58

3.9

3.9

3.7

.21

22.4

22.6

21.6

.003

14.0

13.7

14.9

.35

b.001

29.9

29.8

30.3

29.7

29.8

29.1

21.9

22.2

20.4

18.6

18.2

20.1

11.7

12.5

8.4

b.001

10.1

9.4

13.1

b.001

b.001

7.1

6.8

8.4

20.2

19.5

23.5

72.7

73.7

68.1

49.6

50.1

47.5

b.001

b.001

8.6

7.2

14.7

16.8

16.5

18.1

26.7

27.4

23.8

47.9

48.9

43.4

Baseline study population patient and hospital characteristics by route of admission.

a All variables had less than 1% missing data, with the exception of race and ethnicity (22.8% missing) and Median household income by zip code quartile (2.4% missing). Race and ethnicity percentages are reported only for hospitalizations with data available (77.2% of records).

admitted directly to the hospital (P b .001). A similar unadjusted difference was demonstrated for early mortality: 6.8% for patients admitted through the ED and 7.4% for patients admitted directly to the hospital (P = .005). In this

large sample, relatively small differences by route of admission in age, race and ethnicity, income level, and proportion with Medicaid or without insurance reached statistical significance (P b .001). Patients admitted directly

Fig. 2 Unadjusted inpatient mortality rate by hospital in 290 hospitals and 98 896 hospitalizations for sepsis.

to the hospital were slightly older, with a greater proportion of patients in the highest age quintile (23.9% vs 21.1%, P b

.001). For admissions with available data, the group admitted through the ED had a higher proportion of patients with black or Hispanic race and ethnicity (13.9% and 9.4% for ED admissions vs 13.7% and 6.8% for direct admissions, respectively; P b .001). Patients admitted through the ED were more likely to be covered by Medicaid or to be without insurance (12.5% vs 8.4%, P b .001).

Patients admitted through the ED had relatively greater proportion of 8 of the 12 significant coded comorbid conditions: alcohol abuse, chronic heart failure, coagulo- pathy, fluid and electrolyte disorder, liver failure, renal failure, peripheral vascular disease, and pulmonary circulation disorder (the latter 2 were not statistically significant). Of the remaining 4 significant comorbid conditions, the cohort admitted directly to the hospital had statistically similar proportions of admissions with cancer with and without metastasis and weight loss (P >= .05), and 0.8% more admissions were coded to have lymphoma (P b .001).

There were more admissions from Rural hospitals in the direct admission population (13.1% vs 9.4%, P b .001) and more ED admissions from large bed-size hospitals (73.7% vs 68.1% for direct admissions, P b .001). Patients admitted through the ED were more likely to be admitted to teaching hospitals than patients not admitted through the ED (50.1% vs 47.5%, P b .001) and were more likely to be admitted to institutions with higher hospital sepsis case volume (P b .001).

Adjusted analysis: patient and hospital factors and mortality

Many hospital and patient factors were significantly associated with mortality when simultaneously controlling for other risk factors (Table 2). The risk of inpatient mortality and early inpatient mortality increased with age. All of the

significant comorbid conditions included in the analysis were associated with an increased risk of overall mortality, and 8 of the comorbid conditions were associated with an increased risk of early mortality. After adjustment, chronic heart failure, pulmonary circulation disorders, and renal failure were no longer associated with early mortality; weight loss was associated with improved early survival (OR, 0.45; 95% CI, 0.41-0.50).

Patients admitted to teaching hospitals had an increased likelihood of both early and overall inpatient sepsis mortality (OR, 1.13; 95% CI, 1.04-1.24; OR, 1.25; 95%

CI, 1.15-1.36, respectively). Patients admitted to large bed- size hospitals also had an increased risk of overall mortality (OR, 1.32; 95% CI, 1.13-1.54). However, higher hospital sepsis case volume was associated with a decreased risk of mortality: the odds of both early and overall inpatient mortality decreased with each subsequent hospital sepsis case volume quartile (P b .001).

Patients admitted through the ED had an 8% lower likelihood of early mortality (OR, 0.92; 95% CI, 0.86-0.98) and a 17% lower likelihood of overall inpatient mortality (OR, 0.83; 95% CI, 0.80-0.87) when compared with patients admitted directly to the hospital. These ORs were virtually identical in the sensitivity analysis including admissions to hospitals with fewer than 25 ED or direct admissions (data not shown).

Discussion

In this large nationally representative study, we found a 17% lower likelihood of inpatient overall mortality and an 8% lower likelihood of early inpatient mortality for those patients who were admitted through the ED as compared with direct admission to the hospital. Emergency depart- ments face unique challenges in delivering timely resusci- tation of critically ill patients, primarily due to rising ED volumes and overcrowded conditions [18]. This study provides support for the valuable role the ED plays in the delivery of sepsis care and early resuscitation. despite these challenges. If EDs do indeed provide superior critical care and resuscitation to patients with sepsis, patients suspected of having sepsis should be directed to the ED vs direct admission to the hospital floor.

Route of admission mortality differences could be due to unmeasured severity of illness that favors ED patients. Although NIS does not include specific severity of illness scoring, our findings demonstrate that the ED sepsis population had a greater proportion of patients with significant comorbid conditions, which may indicate similar, if not higher, severity of illness. In particular, there were greater proportions of ED patients coded with fluid and electrolyte disorder and coagulopathy, which may reflect a higher severity of acute sepsis presentation instead of chronic comorbid conditions (Table 1). Furthermore, there were similar proportions of admissions

Table 2 Adjusted association of early inpatient and overall mortality with admission through the ED (reference category: direct admission to the hospital) for 98 896 sepsis admissions in the 2008 NIS

Early inpatient Overall inpatient

mortality, mortality,

OR (95% CI) OR (95% CI)

Patient characteristics Age (y)

18-54 Reference Reference

55-66 1.42 (1.29-1.57) a 1.45 (1.37-1.55) a

67-76 1.76 (1.60-1.94) a 1.74 (1.63-1.85) a

77-83 2.19 (1.98-2.41) a 2.16 (2.03-2.30) a

N83 3.01 (2.74-3.32) a 2.78 (2.61-2.96) a

Female 1.05 (0.99-1.10) 0.97 (0.94-1.00)

Selected comorbid conditions

Alcohol abuse 1.53 (1.33-1.76) a 1.42 (1.29-1.56) a

Cancer: lymphoma 1.33 (1.14-1.56) a 1.54 (1.39-1.71) a

Cancer: metastatic 2.10 (1.91-2.30) a 2.54 (2.38-2.71) a

Cancer: without metastases 1.53 (1.37-1.71) a 1.39 (1.28-1.50) a

Chronic heart failure 0.97 (0.92-1.03) 1.31 (1.26-1.36) a

Coagulopathy 1.28 (1.19-1.38) a 2.02 (1.93-2.12) a

Fluid an electrolyte disorder 1.24 (1.18-1.31) a 1.41 (1.37-1.47) a

Liver failure 1.71 (1.53-1.92) a 1.75 (1.63-1.89) a

Peripheral vascular disease 1.29 (1.19-1.41) a 1.27 (1.20-1.34) a

Pulmonary circulation disorders 0.98 (0.86-1.11) 1.29 (1.19-1.39) a

Renal failure 0.96 (0.91-1.02) 1.15 (1.10-1.19) a

Weight loss 0.45 (0.41-0.50) a 1.25 (1.19-1.31) a

Zip code median-income quartile

Quartile 1 (lowest) Reference Reference

Quartile 2 0.95 (0.89-1.02) 0.95 (0.90-0.99) b

Quartile 3 0.92 (0.86-0.99) b 0.94 (0.89-0.99) b

Quartile 4 (highest) 0.91 (0.84-0.99) b 0.94 (0.89-1.00)

Medicaid/Self-pay 1.05 (0.96-1.16) b 1.07 (1.01-1.14) b

Hospital characteristics

Rural location 0.80 (0.70-0.92) a 0.73 (0.64-0.83) a

Bed size

Small Reference Reference

Medium 0.92 (0.79-1.08) 1.14 (0.97-1.35)

Large 1.08 (0.93-1.25) 1.32 (1.13-1.54) a

Teaching hospital 1.13 (1.04-1.24) a 1.25 (1.15-1.36) a

Hospital sepsis case volume

1: 25-171 cases per year Reference Reference

2: 172-287 cases per year 0.96 (0.85-1.07) 0.95 (0.84-1.06)

3: 288-438 cases per year 0.82 (0.73-0.93) a 0.84 (0.74-0.95) a

4: N438 cases per year 0.73 (0.64-0.84) a 0.75 (0.66-0.86) a

Admission through the ED 0.92 (0.86-0.98) b 0.83 (0.80-0.87) a

a Significant to P b .001.

b Significant to P b .05.

with metastatic cancer in the ED and direct admission populations (P = .05), the comorbid condition with the strongest association with mortality.

Our results highlight several other hospital factors that had an association with mortality. We demonstrate a relationship between greater hospital sepsis case volume and improved mortality outcome. The relationship between volume and outcome has been demonstrated in other disciplines, particularly elective surgery [19] and in the ED care of sepsis patients [20]. In this analysis, lower mortality

outcomes at high-volume institutions were likely driven by a higher proportion of ED admissions at these institutions: hospitals in the highest sepsis case volume quartile had 83% ED admissions as compared with 66% in the lowest volume quartile (P b .001). The mortality rate for patients admitted through the ED in the highest sepsis volume hospitals was 16.2% as compared with 18.3% for ED admissions in other volume quartiles and 19.7% for all direct admissions to the hospital. This would support that both route of admission and hospital sepsis case volume play independent roles and that

the lower mortality rate for ED admissions was largely driven by ED admissions to the highest volume hospitals.

We further found that hospital bed-size and teaching status had a higher likelihood of mortality, which seemingly contradicts the finding that high sepsis case volume had a protective effect. This is reflected in the multivariate model, and similarly, the unadjusted mortality for large bed-size hospitals was 17.7% and 18.6% for teaching hospitals as compared with 16.8% for patients in the highest sepsis volume quartile. However, this is explained by large hospitals and teaching institutions with low annual sepsis case volume. For example, 71.7% of large bed-size hospitals were not in the highest sepsis case volume quartile. Admissions to these low-volume hospitals make up 48.0% of total admissions to large bed-size hospitals and had an unadjusted mortality rate of 18.3%, as compared with 16.8% for admissions to large bed-size hospitals in the highest sepsis volume quartile (P b .001). A similar relationship exists for teaching institutions. However, the associations of large bed-size and teaching institutions with higher mortality may also reflect higher, but unmeasured, severity of sepsis. We hypothesize that the ED Mortality benefit could reflect more aggressive early sepsis resuscitation in the ED. The ED could have a greater availability of technology and infrastructure for sepsis resuscitation, such as central venous pressure monitors or immediate antibiotics for infection. The ED could have a greater concentration of available health care staff, including on-site physicians at all times. Emergency physicians routinely care for patients with other time- sensitive conditions such as myocardial infarction and stroke and are specifically trained to care for undifferentiated, acutely ill patients. Furthermore, the ED also has the unique capability to monitor and reassess patients during their acute presentation and then admit patients to the most appropriate

care unit, such as the intensive care unit.

It remains unclear why some patients are admitted directly to the hospital vs being admitted and initially resuscitated in the ED. This decision may relate to patient preference, outpatient physician practice patterns, hospital census at the time of admission, or institutional culture. There has been attention recently in the lay-press toward diverting patients away from the ED because of over- crowded conditions: patients with nonurgent conditions and patients who may be admitted to the hospital through other means. Although some patients may be treated in other locations, given the lower mortality rates for patients admitted through the ED with sepsis in this study, patients suspected of being septic in the outpatient setting may be preferentially directed toward initial resuscitation in the ED when possible.

There are several limitations to this study, many of which are inherent to secondary data analysis of hospital discharge data sets. We were limited by the information available in the data set and therefore could only adjust for coded comorbid conditions present on admission in predicting inpatient mortality as opposed to other validated schemes of sepsis

severity that rely on information including vital signs and laboratory data [21,22]. The Elixhauser comorbidity coding scheme has been well validated and used extensively in other studies of discharge data to predict mortality in multiple different conditions [15,16]. We were also only able to compare the 2 study populations on the information available in the NIS, and there could be other characteristics on which the populations may differ, such as sepsis secondary to community-acquired vs hospital-acquired infections. How- ever, we believe that the strengths of such a large data set and the ability to adjust for preexisting comorbid conditions in predicting mortality, our primary outcome, outweighed the limitation of lack of laboratory and vital sign data in this study. We were also only able to compare hospitals based on characteristics available, whereas additional information such as availability of intensivists may have been helpful.

There may be other coding and reporting errors; however, coding errors were unlikely to vary significantly by route of admission. The principal diagnosis is widely accepted as the reason for admission [23], and because of this, we assumed that patients with a principal diagnosis of sepsis were resuscitated on admission, in the ED, or in the inpatient hospital for direct admissions, but we were unable to confirm this for every hospitalization. We were limited to only analyzing inpatient mortality in this study because out-of-hospital deaths were not available in the NIS. Future work could expand upon the results of this analysis with the use of improved severity of illness data and methodology to confirm presence of severe sepsis and resuscitation upon admission and additional outcomes and to define which septic patients would receive the greatest benefit from early ED resuscitation. Finally, we hypothe- size that the mortality difference is secondary to superior critical care and sepsis resuscitation in the ED. Further research is necessary to confirm this hypothesis through prospectively observing and comparing the quality of sepsis care in the ED and Inpatient units.

In conclusion, after adjustment for significant comorbid conditions and other patient and hospital factors, we found lower mortality rates for patients admitted to the hospital through the ED as compared with direct admission to the hospital. This result may support a valuable role of the ED in lowering sepsis mortality by providing early resuscitation.

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ajem. 2011.01.011.

References

  1. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003;31(4):1250-6.
  2. Dombrovskiy VY, Martin AA, Sunderram J, Paz HL. Rapid increase in hospitalization and mortality rates for severe sepsis in the United States: a trend analysis from 1993 to 2003. Crit Care Med 2007;35(5): 1244-50.
  3. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and Associated costs of care. Crit Care Med 2001;29(7):1303-10.
  4. Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006;34(6):1589-96.
  5. MacArthur RD, Miller M, Albertson T, Panacek E, Johnson D, Teoh L, et al. Adequacy of early empiric antibiotic treatment and survival in severe sepsis: experience from the MONARCS trial. Clin Infect Dis 2004;38(2):284-8.
  6. Rivers E, Nguyen B, Havstad S, Ressler J, Muzzin A, Knoblich B, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001;345(19):1368-77.
  7. Dellinger RP, Levy MM, Carlet JM, Bion J, Parker MM, Jaeschke R, et al. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Intensive Care Med 2008;34(1):17-60.
  8. Reade MC, Huang DT, Bell D, Coats TJ, Cross AM, Moran JL, et al. Variability in management of early severe sepsis. Emerg Med J 27(2): 110-5.
  9. Brunkhorst FM, Engel C, Ragaller M, Welte T, Rossaint R, Gerlach H, et al. Practice and perception–a nationwide survey of therapy habits in sepsis. Crit Care Med 2008;36(10):2719-25.
  10. Carlbom DJ, Rubenfeld GD. Barriers to implementing protocol-based sepsis resuscitation in the emergency department–results of a national survey. Crit Care Med 2007;35(11):2525-32.
  11. Jones AE, Shapiro NI, Roshon M. Implementing early goal-directed therapy in the emergency setting: the challenges and experiences of translating research innovations into clinical reality in academic and Community settings. Acad Emerg Med 2007;14(11):1072-8.
  12. Poeze M, Ramsay G, Gerlach H, Rubulotta F, Levy M. An international sepsis survey: a study of doctors’ knowledge and perception about sepsis. Crit Care 2004;8(6):R409-13.
  13. Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med 2003;348(16):1546-54.
  14. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Medical Care 1998;36(1):8-27.
  15. Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administra- tive data. Medical Care 2004;42(4):355-60.
  16. Zhu H, Hill MD. Stroke: the Elixhauser Index for comorbidity adjustment of in-hospital case fatality. Neurology 2008;71(4): 283-7.
  17. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42(1):121-30.
  18. Cowan RM, Trzeciak S. Clinical review: emergency department overcrowding and the potential impact on the critically ill. Crit Care 2005;9(3):291-5.
  19. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, et al. hospital volume and surgical mortality in the United States. N Engl J Med 2002;346(15):1128-37.
  20. Powell ES, Khare RK, Courtney DM, Feinglass J. Volume of emergency department admissions for sepsis is related to inpatient mortality: results of a nationwide cross sectional analysis. Critical Care Medicine 2010;38(11):2161-8.
  21. Shapiro NI, Wolfe RE, Moore RB, Smith E, Burdick E, Bates DW. Mortality in Emergency Department Sepsis (MEDS) score: a prospectively derived and validated clinical prediction rule. Crit Care Med 2003;31(3):670-5.
  22. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, et al. The SOFA (Sepsis-Related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996;22(7):707-10.
  23. ICD-9-CM Official Guidelines for Coding and Reporting. http://www.cdc.gov/nchs/data/icd9/icdguide09.pdf. 2009. Accessed

August, 2010.

Leave a Reply

Your email address will not be published. Required fields are marked *