Article, Emergency Medicine

Predictors of acute decompensation after admission in ED patients with sepsis

Brief Report

Predictors of Acute decompensation after admission in ED patients with sepsis?

Jeffrey M. Caterino MD a,?, Tracy Jalbuena MD a, Benjamin Bogucki MD b

aDepartment of Emergency Medicine, The Ohio State University, Columbus, OH 43210, USA

bDepartment of Medicine, Riverside Methodist Hospital, Columbus, OH 43214, USA

Received 9 March 2009; accepted 8 April 2009

Abstract

Purposes: The aim of the study was to identify predictors of acute decompensation within 48 hours of admission among infected emergency department (ED) patients admitted to a regular nursing floor. Procedures: This used a case control study of infected ED patients admitted to a regular nursing floor and who received a discharge diagnosis of sepsis. A multivariate logistic regression model was constructed with the dependent variable as transfer to an intensive care unit (ICU) within 48 hours of admission.

Findings: Seventy-eight patients were enrolled–34 in the ICU group and 44 in the floor group. Only low bicarbonate (b20 mmol/L) (odds ratio [OR], 7.40; 95% confidence interval [CI], 2.35-23.30) and absence of fever (OR, 3.66; 95% CI, 1.11-12.60) were predictive of ICU transfer.

Conclusions: Among infected ED patients admitted to a regular floor, absence of fever and low bicarbonate were independently associated with ICU transfer within 48 hours. Particular attention should be paid to similar patients to ensure appropriate identification of severe infection and appropriate risk stratification.

(C) 2010

Introduction

Emergency departments (EDs) in the United States treat an estimated 280 000 to 500 000 cases of sepsis each year [1,2]. Rapid aggressive resuscitation of the patient who presents with or subsequently develops severe sepsis or septic shock has been recognized as key to improving

? This study was funded, in part, by a 2005 Samuel J. Roessler Memorial Medical Scholarship from the Ohio State University College of Medicine (Columbus, Ohio). The scholarship provided support for BB. The funding body had no role in study design; collection, analysis, or interpretation of data; manuscript writing; or the decision to submit for publication.

* Corresponding author.

E-mail address: [email protected] (J.M. Caterino).

outcomes and decreasing mortality rates that range from 35% to 65% [3,4]. ED care and Disposition decisions can therefore have significant effects on outcome in patients with severe infection. Most attempts to develop instruments for assessing infection severity in ED patients have used in-hospital or 30- day mortality as the primary outcome measure [5-9]. However, this outcome may not be the optimal measure for assessing infection severity in the ED itself. Mortality may be affected by multiple factors occurring after the ED visit, particularly for those patients with lengthy hospital stays. This complicates the attempt to understand the relationship between outcome and ED Care decisions.

It therefore seems reasonable to identify ED variables predicting outcome measures more proximate to the ED visit. One such measure would be decompensation on the Inpatient floor soon after admission. It has been noted that patients

0735-6757/$ – see front matter (C) 2010 doi:10.1016/j.ajem.2009.04.020

who have worsening sepsis or develop septic shock on regular nursing floors have delays in Processes of care as compared to those who develop septic shock in the intensive care unit (ICU) [10]. Therefore, we determined to identify factors predictive of decompensation on the inpatient floor within 48 hours after admission among those ED patients initially admitted to a regular nursing floor. We focused on data available to the emergency physician.

The primary objective of this study was to identify independent predictors of acute decompensation of infected patients admitted through the ED to a regular nursing floor. We chose transfer to an ICU within 48 hours of admission as the marker of acute decompensation. We identified and compared ED characteristics of patients initially admitted to the floor but transferred to an ICU within 48 hours with those remaining on a regular nursing (non- ICU) floor during that time.

Methods

We conducted a case-control study of ED patients admitted to an urban tertiary care hospital. The emergency department has 55 000 annual patient visits, is staffed by board certified emergency physicians, and has an established emergency medicine residency program. Inclusion criteria included treatment in the ED and admission to a regular nursing (non-ICU) floor, proven or suspected infection in the ED, and receipt of a hospital discharge diagnosis of sepsis as defined by an International Classification of Diseases-9- Clinical Modification code of 038.49, 038.9, or 995.92. Patients initially admitted to an ICU and those without documentation of a proven or suspected infection in the ED were excluded. The study included patients presenting to the ED from January 1, 2003, to June 30, 2005. Institutional review board approval was obtained.

Potentially eligible patients were identified using the health system’s electronic medical record repository. We identified all patients admitted from the ED to a non-ICU floor who had an eligible International Classification of Diseases-9-Clinical Modification discharge diagnosis. Patients were then stratified into 2 groups. The first group, the “ICU group,” included all patients who were transferred to an ICU within 48 hours of admission. The second group, the “non-ICU” or “floor” group, consisted of all patients who remained on the regular nursing floor for at least 48 hours.

A random sample of charts from each group was chosen for chart abstraction. A combination of electronic and paper medical records were used to confirm study admission criteria and to obtain study data. Data abstraction was performed by trained study personnel (TJ and BB) using a standardized abstraction form. A code book was created, and standard value checking was used. Abstractors were not blinded to study hypotheses.

First, patients without presence or suspicion of infection in the ED were excluded by abstractors. An infection was

considered present or suspected if there was a documented ED diagnosis of an infectious condition, if there was an ED diagnosis to “rule out” an infectious condition, or if blood cultures were ordered in the ED. Abstracted data included demographics, nursing home residence, comorbidities, ED vital signs, ED diagnostics, previous do-not-resuscitate status, altered mental status in the ED, and administration of vasopressors and antibiotics in the ED. To assess the burden of comorbidities, the Charlson Comorbidity Index was calculated [11]. Altered mental status in the ED was defined as any ED documentation of altered mental status, confusion, delirium, or dementia. Immunosuppression was defined as the presence of HIV, AIDS, cancer, multiple myeloma, chemotherapy, recent systemic steroid use (within 30 days), organ transplant, or use of any immunosuppressive medica- tion. All laboratory results considered were the initial ED values. As per the laboratory’s reference range, low serum bicarbonate was defined as values less than 20 mmol/L.

Vital sign measurements were examined in several different ways. First, we considered the initial measurements in the ED. Second, to identify patients whose illness severity may have changed throughout their ED stay, we identified the highest ED value for heart rate and respiratory rate as well as the lowest ED value for systolic blood pressure and oxygen saturation. Finally, we examined dichotomized vital signs by defining the presence of fever as temperature of 38.0?C or higher, tachycardia as heart rate of 100 beats per minute or higher, tachypnea as respiratory rate of 20 breaths per minute or higher, and hypotension as systolic blood pressure of less than 90 mm Hg.

The primary study outcome variable was transfer to an ICU within 48 hours of admission. To identify differences in patient outcome between the 2 study groups, we also obtained data on in-hospital mortality, hospital length of stay, and intubation during the first 48 hours of hospitalization.

Descriptive statistics (means, medians, proportions) with 95% confidence intervals (CIs) were obtained for study population characteristics. Differences between groups were analyzed using t tests, ?2 tests, and Fisher exact test as appropriate. The P value for significance of all calculations was set at .05. All calculations were performed using STATA

v.10 (STATACorp, College Station, Tex).

To identify independent predictors of ICU transfer within 48 hours of admission, we constructed a forward-selection multivariate logistic regression model with transfer to an ICU as the dependent variable. Potential independent variables were first tested in a series of univariate models. The independent variable with the most significant P value was chosen as the first variable to be added to the multivariate model. For subsequent iterations, all remaining variables were individually added to the model, and the variable with the most significant results was retained. Additional variables were added as long as the P value of the Wald test was less than .05. For Wald test values of less than 0.1, a likelihood ratio test was performed to confirm that variables were not significant. After identification of

these variables, model assumptions were tested, and diagnostics were performed. This included testing for linearity in the logit of any continuous variables retained in the model, testing for the presence of all biologically plausible interactions, and assessing for goodness of fit by use of the Hosmer-Lemeshow ?2 goodness-of-Fit test. Variables in the model found to not be linear in the logit using locally weighted smoothed scatter plots (LOWESS) smoothing or fractional polynomial analysis were categor- ized, and the model was recreated using the categorical variable. To assess model discrimination, area under the curve was calculated. By the “rule of 10s,” for 40 patients

with the outcome event, we would expect to be able to retain up to 4 variables in the model.

Results

The data search for the study period identified 243 potential subjects admitted from the ED who received a discharge diagnosis of sepsis. Of these, 60 were transferred to the ICU within 48 hours of admission and 183 remained on the floor for at least 48 hours. The ICU group consisted of 40 and the non-ICU group of 50 randomly selected charts

Table 1 Characteristics of study patients

Entire population (n = 78)

Transferred to ICU (n = 34)

Remained on regular floor (n = 44)

Means of continuous variables Age

Charlson comorbidity score Temperature–initial (?F) Heart rate–initial Respiratory rate–initial

Systolic blood pressure–initial (mm Hg) Temperature–highest in ED (?F)

Heart rate–highest in ED Respiratory rate–highest in ED Systolic blood pressure–lowest in ED (mm Hg)

Oxygen saturation–lowest in ED White blood cell count (1000/uL) Hemoglobin level (g/dL)

Platelets (1000/uL) band forms (%) Sodium (mmol/L) Potassium (mmol/L)

Blood urea nitrogen (mg/dL) Creatinine (mg/dL)

Glucose (mg/dL)

Serum bicarbonate (mmol/L)

Proportions of dichotomous variables (n%) Male sex

Previously existing do-not-resuscitate order

Nursing home resident Temperature, >=38.0?C Tachycardia (N100 beats/min) Tachypnea (N20 breaths/min)

Hypotension (systolic, b90 mm Hg) Low serum bicarbonate (b20 mmol/L) Immunosuppression present a

Altered mental status in ED b Vasopressors used in the ED Antibiotics administered in the ED Blood cultures ordered in the ED Urine culture ordered in the ED

55.6

53.5

57.2

2.4

2.6

2.1

98.6

97.7

99.4

102

104

102

21

21

21

112

106

115

99.2

98.3

99.9

107

107

107

23

24

22

100

99

101

94

95

93

13.8

13.4

14.2

11.5

12

11.1

285

262

302

10

14

8.5

137

137

136

4.1

4.2

4.1

37

38

36.7

3.4

2.8

3.9

143

145

143

22

19

25

52.5

64.7

43.2

18.2

16.1

23.1

27.6

11.7

38.6

32.0

17.6

43.2

64.1

64.7

63.6

11.5

17.6

6.8

39.7

41.1

38.6

32.9

55.9

14.3

26.9

35.3

20.4

43.4

46.9

40.9

2.5

2.9

2.2

79.5

55.9

97.7

48.7

37.5

56.8

26.3

28.0

25.0

a Immunosuppression defined as the presence of any of the following: HIV infection, AIDS, any cancer, multiple myeloma, current chemotherapy, recent systemic steroid use (last 30 days), history of organ transplant, or use of any immunosuppressive medication.

b Altered mental status defined as any documentation of altered mental status, confusion, dementia, or delirium in the ED chart.

among patients with both ED and inpatient data available. Of these, 6 in each group were found not to have a suspected infection in the ED and were excluded from further analysis.

Seventy-eight patients were entered into the study–34 in the ICU group and 44 in the floor group. The mean age of the entire study population was 70 years. Additional characteristics of the population and of each group are noted in Table 1. Among the differences in the 2 populations, those in the ICU group were more likely to be males and have low serum bicarbonate. They were less likely to have a fever, be from a nursing home, or have blood cultures drawn. Vital signs and laboratory values were quite similar between groups.

Age, hospital length of stay, and hospital mortality were available for all patients. Race was not uniformly documen- ted in the ED record and so was not included as a potential predictor. Vital signs were present for all patients except temperature that was missing in 1 patient. Complete blood counts were available in 73 and electrolytes in 75 patients. For the univariate analyses, each analysis was performed only on those patients who had documentation of the potential predictor variable.

Table 2 Results of univariate testing for prediction of ICU transfer within 48 hours of admission. P values b .05 are bolded.

Results of the univariate logistic regression models are shown in Table 2. Only nursing home residence, initial temperature in the ED, the highest measured temperature in the ED, the absence of a fever (temperature, b38.0?C), serum bicarbonate, and low serum bicarbonate (b20 mmol/L) were significant predictors of ICU transfer in the univariate models. The continuous variable serum bicarbonate was not linear in the logit in either the univariate or multivariate models.

In the multivariate analysis, all potential univariate predictors were tested, starting with the most significant. Only absence of fever and decreased serum bicarbonate were significant contributors to the model. Low serum bicarbonate (b20 mmol/L) had an odds ratio (OR) of 7.40 (95% CI, 2.35- 23.30) and absence of fever had an OR of 3.66 (95% CI, 1.11-12.60) for prediction of ICU transfer. Other ED demographics, vital signs, comorbidities, physical examina- tion findings, and laboratory data were not independently associated with floor to ICU transfer within 48 hours of admission. Three patients in the floor group were excluded form the model–1 for absence of a temperature reading and

2 for absence of serum bicarbonate. Test for interaction

Variables

OR (95% CI)

P

Age

0.99 (0.96-1.01)

.354

Male sex

2.40 (0.96-6.07)

.061

Previously existing do-not-resuscitate order

0.64 (0.13-3.20)

.587

Nursing home resident

0.21 (0.06-0.71)

.012

Charlson comorbidity score

1.15 (0.91-1.47)

.242

Immunosuppression

2.12 (0.77-5.86)

.147

Altered mental status in ED

1.27 (0.51-3.19)

.605

Temperature–initial (?F)

0.56 (0.39-0.78)

.001

Heart rate–initial

1.00 (0.98-1.02)

.719

Respiratory rate–initial

1.01 (0.95-1.07)

.833

Systolic blood pressure–initial (mm Hg)

0.99 (0.97-1.00)

.191

Temperature–highest in ED (?F)

0.68 (0.53-0.88)

.003

Heart rate–highest in ED

1.00 (0.98-1.02)

.936

Respiratory rate–highest in ED

1.03 (0.96-1.11)

.318

Systolic blood pressure–lowest in ED (mm Hg)

1.00 (0.98-1.01)

.704

Oxygen saturation–lowest in ED

1.12 (0.96-1.29)

.141

White blood cell count (K/uL)

0.98 (0.92-1.050

.624

Hemoglobin level (g/dL)

1.15 (0.95-1.39)

.140

Platelets (K/uL)

1.00 (1.00-1.00)

.303

Band forms (%)

1.02 (0.94-1.12)

.601

Sodium (mmol/L)

1.04 (0.96-1.13)

.362

Potassium (mmol/L)

1.08 (0.66-1.78)

.761

Blood urea nitrogen (mg/dL)

1.00 (0.99-1.02)

.848

Creatinine (mg/dL)

0.91 (0.78-1.05)

.187

Glucose (mg/dL)

1.00 (0.99-1.01)

.900

Serum bicarbonate (mmol/L)

0.88 (0.81-0.96)

.004

Absence of fever (temperature, b38.0?C)

3.65 (1.24-10.71)

.018

Tachycardia (N100 beats/min)

1.05 (0.41-2.66)

.922

Tachypnea (N20 breaths/min)

2.92 (0.68-12.70)

.151

Hypotension (systolic, b90 mm Hg)

1.11 (0.44-2.77)

.820

Low serum bicarbonate (b20 mmol/L)

7.60 (2.54-22.8)

b.001

Table 3 Patient outcomes stratified by group as a percentage or mean with 95% CIs

Outcome

Entire population (n = 78)

ICU group (n = 34)

Non-ICU group (n = 44)

P

In-hospital mortality

20.5 (12.2-31.2)

32 .3 (17.4-50.5)

11.4 (3.8-24.6)

.045

Intubation within 48 h

24.4 (15.3-35.4)

61.3 (42.2-78.2)

0 (0-8.0)

b.001

Mean hospital length of stay (d)

14.1 (11.4-16.7)

16.8 (11.9-21.6)

12.0 (9.2-14.8)

.07

between the 2 variables retained in the model revealed no statistically significant interaction (P = .65). By the Hosmer- Lemeshow goodness-of-fit test, there was no evidence of lack of fit (P = .9011). The area under the curve for the model was 0.76, indicating good discrimination.

Several variables were near to inclusion in the final model. Those with P values less than .10 included male sex (OR, 2.6; 95% CI, 0.88-7.8) (P = .084), nursing home

residence (OR, 0.30; 95% CI, 0.08-1.10) (P = .069), and

hemoglobin level (OR, 1.22; 95% CI, 0.96-1.55) (P = .093). Table 3 shows the differences in outcomes between the

2 groups. In-hospital mortality was 11.4% in the non-ICU group and 32.4% in the ICU group (P = .045 by Fisher exact test). The ICU group had increased mortality and intubations as compared to the non-ICU group. There was also a trend toward increased length of stay in the ICU group.

Discussion

For the ED physician, it is relatively easy to determine that patients in severe sepsis or septic shock require ICU level care. Many other septic patients can be safely cared for on and are admitted to non-ICU floors. However, patients may have severe infection in the absence of traditional markers such as hypotension [6,12]. As a result, a certain proportion of patients with infection will decompensate soon after admission from the ED to a regular floor. We examined this group to determine if this acute decompensation could be predicted from data available to the ED physician. As a surrogate for acute decompensation, we chose floor-to-ICU transfer within 48 hours of admission because need for ICU care represents a good representation of illness severity and requirement for higher Levels of care such as central venous access, vasopressor administration, intubation, and others [3]. In this study of variables available to the ED physician, only serum bicarbonate and absence of fever were independently predictive of patient decompensation.

On our initial screening of charts, 24.7% of patients ultimately diagnosed with sepsis who were admitted to the floor (60/243) were transferred to the ICU within 48 hours. Although a certain percentage of these patients were not suspected of infection in the ED, this still points to a large subset of patients at risk for inadequate or delayed care. Early identification of this subgroup has the potential to impact their Care pathways and thus their clinical trajectory. It has been shown that patients who develop sepsis on regular nursing floors get worse or delayed care including

increased times to ICU transfer, intravenous fluid boluses, and inotropic support [10]. This small study also demon- strated a trend toward increased mortality in the floor patients. Given recent data and guideline statements on the importance of rapid, early, aggressive identification and treatment of severe sepsis/septic shock, correct ED admission decisions are critical [3].

Literature studying sepsis in the ED has tended to concentrate on Mortality prediction, particularly in-hospital mortality. Several mortality prediction scores specific to the ED have been developed for use in patients with infection [5,7]. Others have studied end points such as worsening of sepsis syndrome or complicated clinical course [13]. From a practical ED standpoint, 30-day or in-hospital mortality may not be the best outcome measure as it does not account for the variety of factors and events that may affect patient outcome during long hospital stays. Such factors might include development of additional infections; complications of procedures, surgeries, or other disease processes; worsen- ing of comorbid conditions; or development of new disease processes. In addition, inpatient therapeutic choices will play a large role in outcome.

Both Abnormal vital signs and certain patient character- istics have been important predictors in previously published mortality prediction models for ED and ICU patients [5,7,8,14]. For example, Shapiro et al [5] reported that, among other factors, age, nursing home residence, altered mental status, low platelets, and high bands were significant predictors of mortality. These patient factors were not significant for the outcome studied in our population. To confirm this, we extensively analyzed vital sign abnormal- ities in various ways to ensure that we did not miss patients undergoing decompensation while still in the ED. Failure to find significant relationships for these factors is likely due to the spectrum bias inherent in our study as patients with the most abnormal vital signs, and most concerning demo- graphic characteristics were likely identified and admitted to the ICU initially rather than to a regular floor.

Both variables identified as independent predictors in our model are likely associated with failure of the treating physician to recognize the presence of a severe infection. For example, it may be that patients without fever were clinically felt to be less ill than those with fever. As a result, initial triage decisions were more likely to be inappropriate in these patients.

The presence of low serum bicarbonate was the strongest predictor of decompensation in our population. It has been shown that organ hypoperfusion may occur even in the

absence of hypotension and that organ hypoperfusion is associated with poor outcomes [6,15]. The increasing use of serum lactate as a marker of infection severity and prognostic indicator irrespective of blood pressure is consistent with these observations [6]. Study patients were enrolled in a period before widespread use of lactate, and so, this was not available as a potential variable. In our study, the low bicarbonate likely represents occult hypoperfusion. Failure to identify this hypoperfusion may have resulted in patient admission to the floor rather than the ICU.

These findings reinforce the need for particular caution in admitting patients without obvious signs of severe infection to the floor. Emergency and admitting physicians should consider the possibility of worsening infection severity in patients with low serum bicarbonate. In addition, the absence of fever should not be used as a sign that a patient does not have severe infection. Such considerations might impact both admitting decisions and degree of watchfulness exercised over patients once they are admitted.

We recognize several limitations to this study. The inclusion of only patients admitted to a regular floor introduces a spectrum bias as noted above. Although our chosen population was designed to answer the specific study hypothesis, the results should not be applied to all ED patients with infection in the ED. The inherent limitations of chart review methods may also inject bias both in the initial recording of data and in the abstraction process itself. Three additional variables had P values less than .10 in our multivariate model. Given our limited sample size, attention should be paid to these variables in future studies.

One factor not considered as a predictor in our model was administration of ED antibiotics. Because of the study’s chart review methods, we could not reliably determine ED antibiotic timing or administration. We also could not reliably determine administration of antibiotics before arrival in our ED. Rapid administration of antibiotics has been shown to have a significant effect on mortality in patients with severe infection [3,16-18], but their effect on acute decompensation has not been well studied. In performing analyses using the imperfect antibiotic administration information we did possess, low serum bicarbonate remained as the most significant variable in the model, followed by ED antibiotic administration. Future prospective studies should consider ED antibiotic administration for inclusion in the model.

In conclusion, among infected ED patients initially admitted to a regular nursing floor, absence of fever and low serum bicarbonate were independently associated with transfer to an ICU within 48 hours of admission. Particular attention should be paid to patients with these characteristics to ensure appropriate identification of severe infection and appropriate risk stratification.

References

  1. Strehlow MC, Emond SD, Shapiro NI, Pelletier AJ, Camargo CA Jr. National study of emergency department visits for sepsis, 1992 to 2001. Ann Emerg Med 2006;48(3):326-31, 331.
  2. Wang HE, Shapiro NI, Angus DC, Yealy DM. National estimates of severe sepsis in United States emergency departments. Crit Care Med 2007;35(8):1928-36.
  3. 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. Crit Care Med 2008;36(1):296-327.
  4. 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.
  5. 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.
  6. Shapiro NI, Howell MD, Talmor D, Nathanson LA, Lisbon A, Wolfe RE, et al. Serum lactate as a predictor of mortality in emergency department patients with infection. Ann Emerg Med 2005;45(5):524-8.
  7. Olsson T, Terent A, Lind L. Rapid emergency medicine score: a new Prognostic tool for in-hospital mortality in nonsurgical emergency department patients. J Intern Med 2004;255(5):579-87.
  8. Olsson T, Lind L. Comparison of the rapid emergency medicine score and APACHE II in nonsurgical emergency department patients. Acad Emerg Med 2003;10(10):1040-8.
  9. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, et al. A prediction rule to identify low-risk patients with community- acquired pneumonia. N Engl J Med 1997;336(4):243-50.
  10. Lundberg JS, Perl TM, Wiblin T, Costigan MD, Dawson J, Nettleman MD, et al. Septic shock: an analysis of outcomes for patients with onset on hospital wards versus intensive care units. Crit Care Med 1998;26(6):1020-4.
  11. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: develop- ment and validation. J Chronic Dis 1987;40(5):373-83.
  12. Shapiro N, Howell MD. Sick? Or, not sick? Crit Care Med 2005;33(5):

1151-3.

  1. Smith SW, Pheley A, Collier R, Rahmatullah A, Johnson L, Peterson PK. Severe sepsis in the emergency department and its association with a complicated clinical course. Acad Emerg Med 1998;5(12): 1169-76.
  2. Jones AE, Fitch MT, Kline JA. Operational performance of validated physiologic scoring systems for predicting in-hospital mortality among critically ill emergency department patients. Crit Care Med 2005;33(5):974-8.
  3. Nguyen HB, Rivers EP, Knoblich BP, Jacobsen G, Muzzin A, Ressler JA, et al. Early Lactate clearance is associated with improved outcome in severe sepsis and septic shock. Crit Care Med 2004;32(8):1637-42.
  4. Bochud PY, Bonten M, Marchetti O, Calandra T. Antimicrobial therapy for patients with severe sepsis and septic shock: an evidence- based review. Crit Care Med 2004;32(11 Suppl):S495-S512.
  5. Garnacho-Montero J, Garcia-Garmendia JL, Barrero-Almodovar A, Jimenez-Jimenez FJ, Perez-Paredes C, Ortiz-Leyba C. Impact of adequate empirical antibiotic therapy on the outcome of patients admitted to the intensive care unit with sepsis. Crit Care Med 2003; 31(12):2742-51.
  6. Kollef MH, Sherman G, Ward S, Fraser VJ. Inadequate Antimicrobial treatment of infections: a risk factor for hospital mortality among critically ill patients. Chest 1999;115(2):462-74.

Leave a Reply

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