Abstract
Purpose
Detail the trajectory and outcomes of patients with severe sepsis admitted from the emergency department to a non-ICU setting and identify risk factors associated with adverse outcomes.
Material and Methods
Single-center retrospective cohort study conducted at a tertiary, academic hospital in the United States between 2005 and 2009. The primary outcome was a composite outcome of ICU transfer within 48 hours of admission and/or 28-day mortality.
Results
Of 1853 patients admitted with severe sepsis, 841 (45%) were admitted to a non-ICU setting, the rate increased over time (p<0.001), and 12.5% of these patients were transferred to the ICU within 48 hours and/or expired within 28 days. In multivariable models, age (p<0.001), an oncology diagnosis (p<0.001), and illness severity as measured by APACHE II (p=0.04) and high (≥ 4 mmol/L) initial serum lactate levels (p=0.005) were associated with the primary outcome.
Conclusions
Patients presenting to the ED with severe sepsis were frequently admitted to a non-ICU setting and the rate increased over time. One of eight patients admitted to the hospital ward was transferred to the ICU within 48 hours and/or expired within 28 days of admission. Factors present at admission were identified that were associated with adverse outcomes.
Keywords: severe sepsis, infection, outcomes, ICU transfer, mortality
Introduction
Severe sepsis is common, costly, and frequently, life-altering (1–7). Severe sepsis afflicts as many as 3,000,000 adults in the United States annually and results in an estimated 750,000 deaths (4). It is estimated that the majority of patients with severe sepsis are admitted through the Emergency Department (ED) (2–4). As a leading cause of morbidity and mortality internationally, epidemiologic studies of severe sepsis have focused on those patients admitted to the Intensive Care Unit (ICU) (8–10).
Because there is a wide spectrum of disease severity at presentation, ranging from single organ dysfunction to multi-system organ dysfunction that results in death, many patients with severe sepsis will be admitted to a non-ICU setting (1, 7). While it is known that patients transferred from the hospital ward to the ICU are more likely to die than patients admitted directly to the ICU from the ED (11), and the sepsis population appears to be especially vulnerable (12–13), little is known about the epidemiology and outcomes of patients admitted to the wards with severe sepsis.
We conducted an observational cohort study to examine the epidemiology of this understudied patient population. We describe the trajectory of care and the frequency of adverse outcomes in severe sepsis cases admitted to the hospital ward at a single center over 5 years. We hypothesized that admission to a non-ICU setting would increase over time and that we could identify clinical factors present at presentation associated with adverse outcomes, including death.
Materials and Methods
This retrospective observational cohort study was reviewed by the Institutional Review Board at the University of Pennsylvania. The study was approved with an exemption for obtaining informed consent.
Study Population
We examined hospital admissions to the Hospital of the University of Pennsylvania who met consensus criteria for severe sepsis in the ED between 2005 and 2009. We have previously described the validated selection strategy to identify severe sepsis cases in the ED using the ED electronic medical record (EMR) (14–18). First, we examined the charts of all patients with an ED serum lactate measurement or physician documentation that suggested suspected infection (e.g., “sepsis,” “severe sepsis”) to identify sepsis cases (19–20). In 2004, we had instituted a severe sepsis protocol that recommended serum lactate measurement be drawn during initial venous sampling in patients with suspected infection to identify patients eligible for quantitative resuscitation (21). The protocol includes the suggestion that patients receiving quantitative resuscitation be admitted to an ICU (16), although the final decision was at the discretion of the ED provider team. We used antibiotic administration in the ED to satisfy the suspected infection criteria and the EMR to satisfy the systemic inflammatory response syndrome criteria (19). Next, we applied consensus conference criteria to identify cases of severe sepsis (20). We excluded trauma patients, patients not admitted to the hospital, patients not fulfilling severe sepsis criteria, and repeat patient visit(s).
Data Collection
Trained investigators used the case abstraction form to collect demographic and clinical data from the EMR and medical record to permit calculation of the Charlson Comorbidity Index with age score adjustment (22). We identified the etiology of sepsis based on the EMR and medical record, recorded therapies administered in the ED, including central venous catheterization (CVC) and protocol-directed resuscitation defined as resuscitation guided by CVC-derived end-points (15, 21), and documented care-limiting orders (e.g., do-not-resuscitate) from the EMR. We calculated an ED Acute Physiology and Chronic Health Evaluation (APACHE) II score (14–18, 23). Abstraction forms were examined for accuracy and completeness by a separate investigator at the time of data entry.
We categorized hospital admission as admission to the hospital ward or ICU. Patients were categorized as admitted to the ICU if they were admitted to any of the following ICU or intermediate (step-down) care locations: Medical ICU, Intermediate Medical Care Unit, Coronary Care Unit, Coronary Intermediate Care Unit, or a Surgical ICU. We also abstracted whether patients received care in an ICU during the hospitalization and discharge disposition. We validated the admission locale, ICU admission status, and discharge disposition through the use of Horizon Performance Manager (McKesson Information Solutions, Alpharetta, GA), a hospital administrative database. We determined mortality status using the hospital record and the Social Security Death Index.
Our primary outcome was a composite measure of 28-day mortality and/or ICU transfer within 48 hours of admission (“adverse outcome”). We a priori selected the 48 hour threshold, given our focus on clinical factors present at presentation and precedent(24), and our observed data validated this selected definition. We identified the cause of ICU transfer within 48 hours of admission, categorized as related to respiratory failure, hemodynamic instability, change in mental status, or other when not captured by one of the aforementioned categories. We focused on 28-day mortality, rather than in-hospital mortality, to capture patients transitioned to in- or out-patient hospice during the index hospitalization, a frequent occurrence at our institution, and to avoid capturing patients who died at the end of a lengthy hospitalization for reasons potentially unrelated to the severe sepsis presentation. Secondary outcomes included in-hospital mortality, 60-day mortality, ICU transfer at any time during the hospitalization, and disposition status, categorized as home, skilled care facility, and other.
We a priori hypothesized risk factors that would be associated with adverse outcomes during the hospitalization. To minimize the potential for a Type I error, we limited our hypothesized clinical risk factors to age, comorbidities, severity of illness as measured by initial ED serum lactate levels and ED APACHE II score, and etiology of sepsis (14–15, 22–25). We considered year of admission (17), ED occupancy (26), and do-not-resuscitate order at presentation as potential confounders (27–28).
Statistical Analysis
We used the chi-squared statistic or Fisher’s exact test to compare groups. We compared groups using the Student’s t-test for normally distributed continuous variables and the Wilcoxon’s rank-sum test for non-normally distributed continuous variables. We used the non-parametric test for trend to test whether ICU admission and ED process of care measures decreased over time (29).
We used multivariable logistic regression to adjust for potential confounding and specifically to determine whether candidate risk factors were independently associated with the primary outcome (adverse outcomes, defined as ICU transfer within 48 hours and/or 28-day mortality) in patients admitted to the hospital ward. We included candidate risk factors associated with the primary outcome in univariate analyses at a significance level of p<0.05. We adjusted for potential covariates associated with adverse outcomes in univariate analyses at a significance level of p<0.20 one-at-a-time into the multivariable models and maintained the covariate if its inclusion altered the odds ratio for a potential risk factor by >10% (30).
We assessed multicollinearity using variance inflation factors (31). We found collinearity between age, Charlson comorbidity index, and APACHE II. We deconstructed these scores to identify the clinical variables associated with adverse outcomes. After removing the age and oncology adjustment for the comorbidity index score, there was no association between the modified score and outcomes (p=0.58). Although attenuated, APACHE II without age adjustment remained associated with adverse outcomes. As such, we included the following candidate risk factors in multivariable risk factor models: age, initial serum lactate levels, oncology diagnosis, and non-age adjusted ED APACHE II scores. Given the non-normal distribution of initial serum lactate measures, we stratified measures as low (0 – 1.9 mmol/L), intermediate (2.0 – 3.9), and high (4.0 and greater) in multivariable models, in accord with prior studies (14, 32).
We conducted multiple secondary analyses. First, we performed two separate multivariable models where we used the individual components of the composite outcome as the dependent variable to determine if the risk factors identified in the primary analyses were independently associated with ICU transfer within 48 hours and 28-day mortality (33). In addition, we altered the definition of our primary outcome to include 28-day mortality and/or ICU transfer at any time during the hospitalization and we repeated the primary analyses after excluding patients presenting with a do-not-resuscitate order. As the risk factors associated with ICU transfer within 48 hours differed from the other models, we repeated the univariate analyses to further examine which factors were associated with ICU transfer within 48 hours. Last, to examine the ability of APACHE II and initial serum lactate levels to predict an adverse outcome, we used a logistic regression model and calculated the area under the receiver operating curve (AUC) to assess model discrimination and conducted the Hosmer-Lemeshow test to assess model calibration. We used Stata 12.0 software to perform statistical analyses (Stata Datacorp, College Station, TX) (30). We considered two-sided p-values ≤ 0.05 to be significant.
Results
Baseline characteristics for ICU and Hospital Ward Admissions
Of 1853 patients with severe sepsis admitted through the ED, 841 (45.4%, 95% confidence interval (CI): 43.1, 47.7) were admitted to the hospital ward (Figure 1). As shown in Table 1, compared to patients admitted to an ICU setting, patients admitted to a non-ICU setting were younger (p<0.001), less severely ill (p<0.001), had fewer comorbidities (p<0.001), were more likely to present with a genitourinary (p<0.001) or soft-tissue source of infection (p=0.002), were less likely to present with a respiratory source of infection (p<0.001), and were less likely to receive CVC or protocol-directed resuscitation in the ED (p<0.001). Of the 841 subjects admitted to the hospital ward, 217 (25.8%) experienced transient hypotension in the ED, 29 (3.5%) experienced sustained hypotension (i.e., shock) in the ED, and 86 (10.2%) presented with an initial serum lactate of 4 or higher.
Figure 1.
Flow diagram of triage and disposition status of the cohort of patients admitted from the emergency department with severe sepsis.
* Discharged home group included 181 subjects discharged home with home health care and 6 subjects who left against medical advice.
† Discharged to skilled care facility group included subjects discharged to acute rehabilitation facilities (N=16), skilled nursing facilities (N=100), and other facilities (N=27).
Table 1.
Patient-level factors associated with admission to a hospital ward or Intensive Care Unit.
| Hospital Ward Admissions (N=841) | ICU Admissions (N=1012) | p-value | |
|---|---|---|---|
| Clinical Factors at Presentation
| |||
| Age, years | 53 ± 18 | 59 ± 17 | < 0.001 |
| Female sex (n, %) | 399 (47.4) | 451 (44.6) | 0.22 |
| Race (n, %) | |||
| White | 280 (39.8) | 368 (42.2) | 0.40 |
| Black | 381 (54.2) | 462 (53.0) | |
| Other | 42 (6.0) | 41 (4.7) | |
| ED SIRS criteria | |||
| Maximal temperature, ° F | 100.8 ± 2.1 | 99.6 ± 3.1 | < 0.001 |
| Maximal heart rate/minute | 115 ± 18 | 120 ± 24 | < 0.001 |
| Maximal respiratory rate/minute | 23 ± 6 | 28 ± 9 | < 0.001 |
| White blood cell count | 12.1 (7.3 – 16.6) | 13.0 (8 – 18.5) | 0.001 |
| Comorbidities (n, %) | |||
| CAD | 80 (9.6) | 112 (11.2) | 0.26 |
| CHF | 71 (8.5) | 127 (12.7) | 0.004 |
| Chronic liver disease | 26 (3.1) | 69 (6.8) | < 0.001 |
| Chronic renal disease | 107 (12.7) | 153 (15.1) | 0.14 |
| COPD | 48 (5.7) | 81 (8.0) | 0.05 |
| Diabetes mellitus | 168 (20.0) | 207 (20.5) | 0.79 |
| ESRD | 67 (8.0) | 81 (8.0) | 0.97 |
| HIV | 29 (3.4) | 45 (4.4) | 0.27 |
| Hypertension | 347 (41.3) | 464 (45.8) | 0.05 |
| Oncology | 297 (35.3) | 301 (29.7) | 0.01 |
| Transplant | 85 (10.1) | 115 (11.4) | 0.38 |
| Charlson Comorbidity Index | 4 (2 – 6) | 5 (3 – 7) | < 0.001 |
| Do not resuscitate order at admission (n, %) | 30 (3.6) | 46 (4.6) | 0.29 |
| Illness Severity at Presentation | |||
| Initial serum lactate (mmol/L) | 2.4 (1.7 – 3.1) | 3.6 (2.3 – 5.6) | < 0.001 |
| Hyperlactatemia (≥ 4 mmol/L) | 86 (10.2) | 455 (45.0) | <0.001 |
| Lowest systolic blood pressure, mm Hg | 106 ± 22 | 94 ± 24 | < 0.001 |
| Hypotension (n, %) * | 217 (25.8) | 525 (51.9) | < 0.001 |
| Acute kidney injury (n, %) * | 124 (14.7) | 311 (30.7) | <0.001 |
| APACHE II | 14 ± 5 | 19 ± 7 | < 0.001 |
| ED Processes of Care (n, %) | |||
| ED central venous catheterization | 14 (1.7) | 401 (39.6) | < 0.001 |
| Protocol-directed resuscitation initiated in ED *† | 11 (1.3) | 367 (36.3) | < 0.001 |
| Site of infection (n, %) | |||
| Bacteremia | 104 (12.4) | 136 (13.4) | 0.49 |
| Respiratory | 191 (22.7) | 318 (31.4) | <0.001 |
| Genitourinary | 213 (25.3) | 209 (20.6) | <0.001 |
| Gastrointestinal | 104 (12.4) | 141 (13.9) | 0.32 |
| Soft-tissue | 105 (12.5) | 82 (8.1) | 0.002 |
Definition of abbreviation: COPD=chronic obstructive pulmonary disease; HIV=human immunodeficiency virus; ICU=Intensive Care Unit; APACHE II= Acute Physiologic and Chronic Health Evaluation II scores; SIRS=systemic inflammatory response criteria.
Categorical variables are reported as a percentage; continuous variables are reported as mean and standard deviation or as median and interquartile range. Missing data was infrequent for each variable listed, except for race (<5% for each variable).
Hypotension defined as systolic blood pressure ≤ 90 mm Hg. Acute kidney injury defined as serum creatinine at presentation of ≥ 2 mg/dL. Protocol directed resuscitation defined as resuscitation guided to central venous catheter derived end-points (15, 20).
Amongst eligible patients (serum lactate ≥ 4 mmol/L or septic shock), protocol-directed resuscitation was initiated more often in patients admitted to an ICU (347/619 (56.1%) vs. 9/113 (8.0%), p<0.001).
Admission to an ICU setting decreased over time (64.1% in 2005 to 48.6% in 2009, p-value for trend <0.001) overall and, specifically, amongst patients eligible for protocol-directed resuscitation (91.2% in 2005 to a nadir of 76.4% in 2008, p=0.001). In parallel, CVC in the ED decreased over time (28.9% in 2005 to 17.0% in 2009, p-value for trend <0.001), as did initiation of protocol-directed resuscitation amongst eligible patients (59.3% in 2005 to 37.8% in 2009, p-value for trend <0.001).
Of 841 patients admitted directly to the hospital ward from the ED, 87 (10.3%, 95% CI: 8.4, 12.6) were transferred to the ICU during their hospital stay (Table 2). The median time to ICU transfer was 2 days after admission, with interquartile range of 1 – 5 days and range from 2 hours to 43 days (Figure 2). The most common cause for ICU transfer within 48 hours was progressive hemodynamic instability, followed by respiratory failure (Table 3), and 8 of the 47 patients (17.0%) transferred within 48 hours of hospital admission were intubated prior to or upon arrival to the ICU. In-hospital, 28-day, and 60-day mortality rates were 5%, 8%, and 11%, respectively, for the 841 patients admitted to the hospital ward. In total, 130 patients (15.5%, 95% CI: 13.1, 18.1) initially admitted to the hospital ward were transferred to the ICU during their hospitalization and/or expired within 28 days of admission. In general, severe sepsis survivors admitted to the hospital ward had a favorable disposition, with 78% of survivors being discharged to home (Figure 1). However, many survivors required assistance, as 23% required home health care, 18% were discharged to a skilled care facility, and 4% were transitioned to hospice at discharge.
Table 2.
Outcomes by admission location in patients with severe sepsis.
| Hospital Ward Admissions (N=841) | ICU Admissions (N=1012) | p-value | |
|---|---|---|---|
| Outcomes | |||
|
| |||
| Mortality | |||
|
| |||
| In-hospital mortality (n, %) | 44 (5.2) | 228 (22.5) | < 0.001 |
| 28-day mortality (n, %) | 71 (8.4) | 276 (27.3) | < 0.001 |
| 60-day mortality (n, %) | 91 (10.8) | 311 (30.7) | < 0.001 |
|
| |||
| ICU Transfer | |||
|
| |||
| ICU transfer within 48 hours | 47 (5.6) | --- | --- |
| ICU transfer during hospitalization | 87 (10.3) | --- | --- |
|
| |||
| Adverse outcome (ICU transfer within 48 hours and/or 28-day mortality) | |||
|
| |||
| Composite outcome | 105 (12.5) | --- | --- |
Definition of abbreviation: COPD=chronic obstructive pulmonary disease; HIV=human immunodeficiency virus; ICU=Intensive Care Unit; APACHE II= Acute Physiologic and Chronic Health Evaluation II scores.
Categorical variables are reported as a percentage; continuous variables are reported as mean and standard deviation or as median and interquartile range.
Figure 2.
Timing of intensive care unit transfer during hospital stay in patients with severe sepsis admitted from the emergency department to the hospital ward.
Table 3.
Cause for ICU transfer within 48 hours after admission to the hospital ward.
| Cause | n (%) |
|---|---|
| Hemodynamic instability * | 28 (59.6) |
| Respiratory failure | 9 (19.2) |
| Mental status change | 4 (8.5) |
| Other † | 6 (12.8) |
| Total: N = 47 |
Hemodynamic instability cases included: 22 cases of shock, 3 cases of cardiac arrest, and 3 cases of arrhythmia.
Other cases included: management of hypothermia (n=1), acute kidney injury (n=1), and resuscitation for hyperlactatemia (n=1), gastrointestinal bleeding (n=1), and pre-operatively (n=2).
Despite being younger and less severely ill at hospital admission compared to patients directly admitted to the ICU, as measured by initial serum lactate levels and APACHE II scores, the 28-day mortality rate for patients transferred to the ICU within 48 hours was 27.7% (13/47) and for patients transferred to the ICU during the hospitalization was 32.2% (23/87), similar to that experienced by patients admitted directly to the ICU (Table 2). After adjusting for APACHE II and initial serum lactate levels, compared to those directly admitted to the ICU, 28-day mortality was higher in those transferred within 48 hours (odds ratio (OR): 1.59, 95% confidence interval (CI): 0.80, 3.14, p=0.19) or during the hospitalization (OR: 2.11, 95% 1.28, 3.49, p=0.004), achieving statistical significance in the latter.
An adverse outcome occurred in 105 patients (12.5%, 95% CI: 10.3, 14.9) admitted to the hospital ward, defined as being transferred to the ICU within 48 hours and/or expiring within 28 days of admission. Compared to ward admissions who did not experience an adverse outcomes, those who experienced an adverse outcome were significantly older (p<0.001), more likely to have a higher burden of comorbid conditions (p<0.001), an oncology diagnosis in particular (p<0.001), a do-not-resuscitate order at admission (p<0.001), and were more severely ill, as measured by initial serum lactate levels (p=0.01), eligibility for protocol-directed resuscitation (p=0.01), and baseline APACHE II levels (p<0.001). There was no association between an adverse outcome and year of admission (p=0.20), ED occupancy (p=0.23), or site of infection (Table 4). In analyses using ICU transfer within 48 hours as the outcome, we confirmed that each of the illness severity measures were associated (Table 5).
Table 4.
Patient-level factors associated with adverse outcomes in severe sepsis hospital ward admissions.
| No Adverse Outcome (N=736) | Adverse Outcome (N=105) | p-value | |
|---|---|---|---|
| Clinical Factors at Presentation | |||
|
| |||
| Age, years | 54 ± 18 | 63 ± 17 | < 0.001 |
| Female sex (n, %) | 346 (47.0) | 53 (50.5) | 0.51 |
| Race (n, %) | |||
| White | 240 (38.8) | 40 (47.1) | 0.38 |
| Black | 340 (55.0) | 41 (48.2) | |
| Other | 38 (6.1) | 4 (4.7) | |
| ED SIRS criteria | |||
| Maximal temperature, ° F | 100.9 ± 2.0 | 99.7 ± 2.7 | < 0.001 |
| Maximal heart rate/minute | 115 ± 18 | 114 ± 19 | 0.65 |
| Maximal respiratory rate/minute | 23 ± 6 | 24 ± 6 | 0.13 |
| White blood cell count | 12 (7.2 – 16.5) | 12.7 (8.2 – 19.8) | 0.08 |
| Comorbidities (n, %) | |||
| CAD | 72 (9.9) | 8 (7.7) | 0.45 |
| CHF | 63 (8.6) | 8 (7.7) | 0.75 |
| Chronic liver disease | 22 (3.0) | 4 (3.8) | 0.65 |
| Chronic renal disease | 93 (12.6) | 14 (13.3) | 0.84 |
| COPD | 41 (5.6) | 7 (6.7) | 0.65 |
| Diabetes mellitus | 145 (19.7) | 23 (21.9) | 0.60 |
| ESRD | 57 (7.7) | 10 (9.5) | 0.53 |
| HIV | 25 (3.4) | 4 (3.8) | 0.78 |
| Hypertension | 305 (41.4) | 42 (40.0) | 0.78 |
| Oncology | 239 (32.5) | 58 (55.2) | <0.001 |
| Transplant | 79 (10.7) | 6 (5.7) | 0.11 |
| Charlson Comorbidity Index | 4 (2 – 6) | 6 (4 – 9) | < 0.001 |
| Do not resuscitate order at admission (n, %) | 16 (2.2) | 14 (13.3) | <0.001 |
| Illness Severity at Presentation | |||
| Initial serum lactate (mmol/L) | 2.3 (1.6 – 3.1) | 2.5 (2.0 – 3.5) | 0.01 |
| Hyperlactatemia (≥ 4 mmol/L) * | 67 (9.1) | 19 (18.1) | 0.004 |
| Lowest systolic blood pressure, mm Hg | 106 ± 22 | 105 ± 21 | 0.64 |
| Hypotension (n, %) * | 189 (25.7) | 28 (26.7) | 0.83 |
| Eligible for protocol-directed resuscitation * | 90 (12.2) | 23 (21.9) | 0.01 |
| Acute kidney injury (n, %) * | 105 (14.3) | 19 (18.1) | 0.30 |
| APACHE II | 13 ± 5 | 16 ± 5 | < 0.001 |
| Site of infection (n, %) | |||
| Bacteremia | 86 (11.7) | 18 (17.1) | 0.11 |
| Respiratory | 165 (22.4) | 26 (24.8) | 0.59 |
| Genitourinary | 186 (25.3) | 27 (25.7) | 0.92 |
| Gastrointestinal | 89 (12.1) | 15 (14.3) | 0.52 |
| Soft-tissue | 95 (12.9) | 10 (9.5) | 0.33 |
Definition of abbreviation: COPD=chronic obstructive pulmonary disease; HIV=human immunodeficiency virus; ICU=Intensive Care Unit; APACHE II= Acute Physiologic and Chronic Health Evaluation II scores (age-adjusted).
Categorical variables are reported as a percentage; continuous variables are reported as mean and standard deviation or as median and interquartile range. Missing data was infrequent for each variable listed, except for race (<5% for each variable).
Hypotension defined as systolic blood pressure ≤ 90 mm Hg and protocol-directed resuscitation eligibility defined as serum lactate ≥ 4 mmol/L or fluid-refractory hypotension.
Table 5.
Patient-level factors associated with ICU transfer within 48 hours of hospital ward admission in patients with severe sepsis.
| No Transfer (N=794) | ICU Transfer (N=47) | p-value | |
|---|---|---|---|
| Clinical Factors at Presentation | |||
|
| |||
| Age, years | 55 ± 18 | 55 ± 19 | 0.91 |
| Female sex (%) | 368 (46.4) | 31 (66.0) | 0.01 |
| Race (%) | |||
| White | 262 (33.0) | 18 (38.3) | 0.72 |
| Black | 362 (45.6) | 19 (40.4) | |
| Other | 170 (21.4) | 10 (21.3) | |
| ED SIRS criteria | |||
| Maximal temperature, ° F | 100.8 ± 2.0 | 100.0 ± 3.1 | 0.01 |
| Maximal heart rate/minute | 115 ± 18 | 115 ± 19 | 0.98 |
| Maximal respiratory rate/minute | 23 ± 6 | 24 ± 6 | 0.22 |
| White blood cell count | 12.1 (7.2 – 16.5) | 12.9 (9.7 – 21.8) | 0.06 |
| Comorbidities (%) | |||
| CAD | 76 (9.7) | 4 (8.5) | 1.00 |
| CHF | 67 (8.5) | 4 (8.7) | 1.00 |
| Chronic liver disease | 23 (2.9) | 3 (6.4) | 0.17 |
| Chronic renal disease | 98 (12.3) | 9 (19.2) | 0.18 |
| COPD | 45 (5.7) | 3 (6.4) | 0.75 |
| Diabetes mellitus | 158 (19.9) | 10 (21.3) | 0.85 |
| ESRD | 60 (7.6) | 7 (14.9) | 0.09 |
| HIV | 26 (3.3) | 3 (6.4) | 0.22 |
| Hypertension | 327 (41.2) | 20 (42.6) | 0.88 |
| Oncology | 282 (35.5) | 15 (31.9) | 0.75 |
| Transplant | 83 (10.4) | 2 (4.3) | 0.22 |
| Charlson Comorbidity Index | 4 (2 – 6) | 5 (2 – 7) | 0.15 |
| Do not resuscitate order at admission (n, %) | 28 (3.5) | 2 (4.3) | 0.79 |
| Severity of Illness Measure | |||
| Initial serum lactate (mmol/L) | 2.4 (1.7 – 3.1) | 2.5 (1.9 – 3.7) | 0.19 |
| Hyperlactatemia (≥ 4 mmol/L) | 77 (9.7) | 9 (19.2) | 0.04 |
| Lowest systolic blood pressure, mm Hg | 106 ± 22 | 104 ± 22 | 0.53 |
| Hypotension (%) * | 203 (25.6) | 14 (29.8) | 0.52 |
| Eligible for protocol-directed resuscitation * | 102 (12.8) | 11 (23.4) | 0.04 |
| APACHE II | 14 ± 5 | 16 ± 5 | 0.01 |
| Site of infection | |||
| Bacteremia | 94 (11.8) | 10 (21.3) | 0.07 |
| Respiratory | 184 (23.2) | 7 (14.9) | 0.21 |
| Genitourinary | 200 (25.2) | 13 (27.7) | 0.73 |
| Gastrointestinal | 97 (12.2) | 7 (14.9) | 0.65 |
| Soft-tissue | 99 (12.5) | 6 (12.7) | 1.00 |
Definition of abbreviation: COPD=chronic obstructive pulmonary disease; HIV=human immunodeficiency virus; ICU=Intensive Care Unit; APACHE II= Acute Physiologic and Chronic Health Evaluation II scores.
Categorical variables are reported as a percentage; continuous variables are reported as mean and standard deviation or as median and interquartile range.
Hypotension defined as systolic blood pressure ≤ 90 mm Hg and protocol-directed resuscitation eligibility defined as serum lactate ≥ 4 mmol/L or fluid-refractory hypotension.
In multivariable models adjusted for potential confounders, age (p<0.001), an oncology diagnosis (p<0.001), and illness severity at presentation, as measured by APACHE II (p=0.04) and high (≥ 4 mmol/L) initial serum lactate levels (p=0.005) were independently associated with an adverse outcome in severe sepsis patients admitted to the hospital ward (Table 6). Illness severity, as measured by APACHE II, discriminated those who experienced an adverse outcome with an AUC of 0.62 (95% CI: 0.57, 0.67) and was well calibrated (p=0.57), whereas serum lactate did not significantly predict an adverse outcome (AUC 0.56, 95% CI: 0.47, 0.64).
Table 6.
Association between clinical risk factors and adverse outcomes in severe sepsis hospital ward admissions using multivariable logistic regression.
| Independent Variable * | Adjusted Odds Ratio (95% CI) | p-value |
|---|---|---|
| Age (years) * | 1.02 (1.01 – 1.04) | <0.001 |
| Oncology | 2.22 (1.44 – 3.42) | <0.001 |
| APACHE II (baseline) * † ‡ | 1.05 (1.00 – 1.09) | 0.04 |
| Initial Serum Lactate Strata | ||
| Low (0 – 1.9) | Reference | Reference |
| Intermediate (2.0 – 3.9) | 1.42 (0.85 – 2.37) | 0.18 |
| High (4.0 and greater) | 2.68 (1.35 – 5.32) | 0.005 |
Definition of abbreviation: APACHE =Acute physiology and chronic health evaluation score; CI=confidence interval.
Odds ratio for each 1-unit increase in age and baseline APACHE II score. An adjusted odds ratio of greater than 1 is indicative of increased odds of developing an adverse outcome. Inclusion of potential confounders (i.e., do not resuscitation order at admission) did not alter the odds ratio estimates significantly.
The APACHE II calculation included in the model was not age-adjusted given collinearity.
In multivariable models using 28-day mortality as the dependent variable, we confirmed that age (p<0.001), an oncology diagnosis (p<0.001), and high initial serum lactate levels (p<0.001) were associated with 28-day mortality after adjusting for potential confounders (Table 7). In models using ICU transfer within 48 hours as the dependent variable, we found that illness severity as measured by baseline APACHE II (p=0.005) and high initial serum lactate levels (p=0.046) were associated with ICU transfer within 48 hours of admission (Table 8). Further, patients transferred to the ICU within 48 hours who expired within 28 days were significantly more likely to be oncology patients (p=0.001) and were more severely ill at presentation, as measured by APACHE II scores (p=0.02) and higher initial serum lactate levels (p=0.03) (Table 9). Finally, when we altered the definition of the primary composite outcome to incorporate ICU transfer during the hospitalization, rather than limiting to ICU transfer within 48 hours, or excluded patients with a do-not-resuscitate order at presentation in our primary analyses, with the exception that baseline APACHE II was attenuated to the null, the results were materially unchanged.
Table 7.
Association between clinical risk factors and 28-day mortality in severe sepsis hospital ward admissions using multivariable logistic regression.
| Independent Variable | Adjusted Odds Ratio (95% CI) | p-value |
|---|---|---|
| Age (years) * | 1.05 (1.03 – 1.07) | <0.001 |
| Oncology | 6.02 (3.34 – 10.87) | <0.001 |
| APACHE II (baseline) * † ‡ | 1.04 (0.98 – 1.09) | 0.16 |
| Initial Serum Lactate Strata | ||
| Low (0 – 1.9) | Reference | Reference |
| Intermediate (2.0 – 3.9) | 1.90 (0.96 – 3.76) | 0.06 |
| High (4.0 and greater) | 4.86 (2.04 – 11.53) | <0.001 |
| Year of admission | 0.77 (0.63 – 0.94) | 0.01 |
Definition of abbreviation: APACHE =Acu te physiology and chronic health evaluation score; CI=confidence interval.
Odds ratio for each 1-unit increase in age and baseline APACHE II score. An adjusted odds ratio of greater than 1 is indicative of increased odds of 28-day mortality. Inclusion of the potential confounder variable “do not resuscitate order at admission” did not alter the odds ratio estimates significantly. Year of admission altered the odds ratio estimates significantly and remained associated with 28-day mortality. Emergency department occupancy (p=0.44) was not associated with 28-day mortality.
The APACHE II calculation included in the model was not age-adjusted.
Table 8.
Association between clinical risk factors and ICU transfer within 48 hours of hospital ward admission in patients with severe sepsis.
| Independent Variable | Adjusted Odds Ratio (95% CI) | p-value |
|---|---|---|
| Age (years) * | 1.00 (0.98 – 1.02) | 0.95 |
| Oncology | 0.72 (0.38 – 1.38) | 0.32 |
| APACHE II (baseline) * † ‡ | 1.08 (1.02 – 1.15) | 0.005 |
| Initial Serum Lactate Strata | ||
| Low (0 – 1.9) | Reference | Reference |
| Intermediate (2.0 – 3.9) | 1.16 (0.58 – 2.35) | 0.68 |
| High (4.0 and greater) | 2.53 (1.02 – 6.29) | 0.046 |
Definition of abbreviation: APACHE =Acu te physiology and chronic health evaluation score; CI=confidence interval; ICU=intensive care unit.
Odds ratio for each 1-unit increase in age and baseline APACHE II score. An adjusted odds ratio of greater than 1 is indicative of increased odds of being transferred to an ICU within 48 hours of admisison. None of the potential confounders (year of admission (p=0.55), emergency department occupancy (p=0.40), do not resuscitation order at admission (p=0.79)) were associated with ICU transfer within 48 hours.
The APACHE II calculation included in the model was not age-adjusted.
Table 9.
Patient-level factors associated with 28-day mortality in severe sepsis patients transferred to the ICU within 48 hours of hospital ward admission.
| 28-day Survivors (N=34) | Non-Survivors (N=13) | p-value | |
|---|---|---|---|
| Clinical Factors at ED Presentation | |||
|
| |||
| Age, years | 53 ± 19 | 61 ± 17 | 0.23 |
| Oncology (n, %) | 6 (17.6) | 9 (69.2) | 0.001 |
| Illness Severity at Presentation | |||
| Initial serum lactate (mmol/L) | 2.3 (1.8 – 3.0) | 3.7 (2.1 – 4.1) | 0.03 |
| Hyperlactatemia, ≥4 mmol/L (n,%) | 4 (11.8) | 5 (38.5) | 0.04 |
| Eligible for protocol-directed resuscitation (n, %) * | 6 (17.7) | 5 (38.5) | 0.25 |
| APACHE II (baseline) | 14 ± 5 | 18 ± 4 | 0.02 |
Definition of abbreviation: APACHE II= Acute Physiologic and Chronic Health Evaluation II scores.
Categorical variables are reported as a percentage; continuous variables are reported as mean and standard deviation or as median and interquartile range.
Protocol-directed resuscitation eligibility defined as serum lactate ≥ 4 mmol/L or fluid-refractory hypotension.
Discussion
In this retrospective cohort study conducted over five years, we found that patients presenting to the hospital through the ED with severe sepsis were frequently admitted to a non-ICU setting and the rate increased over time. Once admitted to a non-ICU setting, adverse outcomes, defined as transfer to the ICU within 48 hours and/or 28-day mortality, occurred in one out of eight patients. Factors associated with adverse outcomes included patient age, oncologic diagnosis, and illness severity upon presentation to the ED, with measures of illness severity being associated with ICU transfer within 48 hours and 28-day mortality.
We found that nearly half of patients with severe sepsis admitted through the ED were admitted to a non-ICU setting, consistent with prior work (1, 7, 34). Over the study period, the proportion of patients admitted to a non-ICU setting increased significantly. This admission pattern coincided with less frequent ED central venous catheterizations and less frequent initiation of protocol-directed resuscitation in eligible patients over time. In the recently completed randomized trial of protocol-based care for early septic shock, protocol-based early goal-directed therapy (EGDT) in the ED, which involved CVC in 94% of cases, did not result in improved outcomes (35). Consistent with the two non-EGDT arms of ProCESS (35), we found that 15% of patients eligible for protocol-directed resuscitation were admitted to a non-ICU setting, initiation of quantitative resuscitation was rare in these patients, and admission to a non-ICU setting increased over time in these patients. Because the results of ProCESS support a less (i.e., lactate clearance (36)) invasive approach in many patients eligible for quantitative resuscitation initially, it will be important to examine triage decisions and the effects of these decisions in future studies.
We found that 16% of severe sepsis patients admitted to the hospital ward were transferred to the ICU during the hospitalization and/or expired within 28-days. As such, 84% of severe sepsis patients admitted to a non-ICU setting were managed effectively to the degree that they were discharged alive from the hospital without requiring ICU-level care. However, 10% of ward admissions required ICU transfer at some point during their hospitalization, half of these occurring within the first 48 hours and three-quarters occurring within the first 5 days (Figure 2). For those transferred within the first 48 hours, hemodynamic instability was the most common event prompting transfer, followed by management of respiratory failure, and nearly one out of five patients required intubation proximate to the time of transfer. Whether the deterioration could have been prevented had the triage decision or care provisions been different remains unclear and warrants further investigation.
While the overall mortality rates for patients admitted to the hospital ward were significantly less than those patients admitted directly to the ICU, the observed mortality rates for ward admissions are at least two-fold higher than the average hospitalization (37). Further, despite significantly lower illness severity at hospital admission, the 28-day mortality incurred by those requiring ICU transfer was similar to the mortality rate for those directly admitted to the ICU. Because the adjusted analyses suggest that outcomes may differ by triage decision, increased attention to those at increased risk is justified. Whether the observed mortality rate would have been less had those who required ICU transfer been directly admitted to an ICU remains unclear and requires further investigation. To estimate the causal effects of these critical triage decisions on patient-centered outcomes using observational data will require robust analyses to balance observed and unobserved covariates, such as instrumental variables or matching methods (38, 39).
We found that patient age, comorbid conditions (oncology patients), and illness severity were associated with adverse outcomes. High initial serum lactate levels, a marker of illness severity and an established criterion to initiate quantitative resuscitation (14–15, 21), along with higher APACHE II scores, were associated with the composite outcome of ICU transfer within 48 hours and/or 28-day hospital mortality, as well as the individual components. These observations suggest that elevated serum lactate levels, previously demonstrated to be associated with morbidity and mortality in sepsis in general (14, 18, 32, 40), also appear to effectively risk-stratify patients with severe sepsis admitted to the hospital ward. While plausible, whether outcomes would have been better for these patients had care (21, 35, 41) or triage decisions (i.e., ICU care from time of admission) differed remains unclear and warrants further investigation. Additionally, these findings have important implications for early warning systems aimed to identify at-risk ward patients, given the potential utility of a strategy designed to augment a physiologic scoring system (42) with additional clinical data to permit improved risk-stratification and precision regarding the target population.
Survivors of severe sepsis admitted to the hospital ward, in general, were able to return home. And yet, 23% of survivors required home health care, 18% required placement in a skilled care facility, and 4% transitioned to hospice. In 2010, Iwashyna and colleagues revealed that severe sepsis was associated with new and clinically meaningful cognitive and physical impairments in elderly survivors (5). Rohde and colleagues recently demonstrated that functional disability after severe sepsis is not confined to the most severely ill, as physical impairment and increased care needs were common in patients admitted to the medical ward (43). Our work supports and extends these observations by detailing the frequent care needs of the broader population of hospital ward admissions and the urgent need to identify strategies, such as early physical therapy (44), to mitigate these impairments.
There are potential limitations to our study. First, our observational study design limits our ability to detail why patients were admitted to a non-ICU setting and whether outcomes would have differed had patients been triaged and managed differently. Future studies, designed to prospectively examine clinical decision-making related to triage and care delivery, are warranted. Second, important, unobserved covariates may have influenced our results. For example, while we examined ED occupancy, because delayed transfer of care is known to adversely affect outcomes (45), further study is required to examine how ICU occupancy impacts triage decisions, trajectory and outcomes of patients admitted with severe sepsis. Related, as a single-center study with high ICU occupancy (46) and a robust oncology service, our results may not generalize to other hospitals. Third, while we designed our study to examine patient-specific risk factors, additional study is required to consider structural (e.g., time of admission, day of admission) and process of care factors (e.g., time to antibiotics (16), management by a rapid response team on the ward) and to develop prediction scoring models to aid in triage decisions. Fourth, we did not include a penalty for multiple comparisons in our risk factor analyses. However, to account for this potential limitation, we limited our a priori hypotheses and conducted multiple sensitivity analyses. Further, had we applied a Bonferroni correction and defined significance as a p-value of 0.01 based on the number of candidate risk factors, each of the risk factors identified in the primary analysis would have remained significant, save for baseline APACHE II.
In conclusion, many patients presenting to an ED with severe sepsis were admitted to a non-ICU setting and the rate appears to be increasing. One out of eight patients admitted to the hospital ward was transferred to the ICU within 48 hours and/or expired within 28 days of admission. We identified patient-specific factors associated with adverse outcomes in these patients. Whether outcomes could be altered through improved recognition and management of the patients identified as at-risk requires further investigation.
Highlights.
Severe sepsis patients were frequently admitted to a non-ICU setting
Rate of admission to a non-ICU setting appeared to be increasing
One out of eight patients admitted to the hospital ward was transferred to the ICU within 48 hours and/or expired within 28 days
Patients admitted to a non-ICU setting with a high (≥ 4 mmol/L) initial serum lactate level were more likely to experience an adverse outcome
Acknowledgments
Funding: The study was supported in part by National Institutes of Health, National Heart, Lung and Blood Institute (NIH NHLBI) Loan Repayment Program, Bethesda, MD
Footnotes
Disclosures: For each of the above authors, no financial or other potential conflicts of interest exist related to the work. Presented as an abstract, in part, at the Society of Critical Care Medicine Congress in San Diego, CA in 2011.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.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;29:1303–1310. doi: 10.1097/00003246-200107000-00002. [DOI] [PubMed] [Google Scholar]
- 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;35:1244–1250. doi: 10.1097/01.CCM.0000261890.41311.E9. [DOI] [PubMed] [Google Scholar]
- 3.Lagu T, Rothberg MB, Shieh MS, Pekow PS, Steingrub JS, Lindenauer PK. Hospitalizations, costs, and outcomes of severe sepsis in the United States 2003 to 2007. Crit Care Med. 2012;40:754–761. doi: 10.1097/CCM.0b013e318232db65. [DOI] [PubMed] [Google Scholar]
- 4.Gaieski DF, Edwards JM, Kallan MJ, Carr BG. Benchmarking the incidence and mortality of severe sepsis in the United States. Crit Care Med. 2013;41(5):1167–1174. doi: 10.1097/CCM.0b013e31827c09f8. [DOI] [PubMed] [Google Scholar]
- 5.Iwashyna TJ, Ely EW, Smith DM, Langa KM. Long-term cognitive impairment and functional disability among survivors of severe sepsis. JAMA. 2010;304(16):1787–1794. doi: 10.1001/jama.2010.1553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Guerra C, Linde-Zwirble WT, Wunsch H. Risk factors for dementia after critical illness in elderly medicare beneficiaries. Crit Care. 2012;16(6):R233. doi: 10.1186/cc11901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang HE, Shapiro NI, Angus DC, Yealy DM. National estimates of severe sepsis in United States emergency departments. Crit Care Med. 2007;35:1928–1936. doi: 10.1097/01.CCM.0000277043.85378.C1. [DOI] [PubMed] [Google Scholar]
- 8.Brun-Buisson C, Meshaka P, Pinton P, Vallet B EPISEPSIS Study Group. EPISEPSIS : a reappraisal of the epidemiology and outcome of severe sepsis in French intensive care units. Intensive Care Med. 2004;30:580–8. doi: 10.1007/s00134-003-2121-4. [DOI] [PubMed] [Google Scholar]
- 9.Alberti C, Brun-Buisson C, Burchard H, Martin C, Goodman S, Artigas A, et al. Epidemiology of sepsis and infection in ICU patients from an international multicentre cohort study. Intensive Care Med. 2002;28:108–121. doi: 10.1007/s00134-001-1143-z. [DOI] [PubMed] [Google Scholar]
- 10.Brun-Buisson C, Doyon F, Carlet J, Dellamonica P, Gouin F, Lepoutre A, et al. Incidence, risk factors and outcome of severe sepsis and septic shock in adults: a multicenter prospective study in Intensive Care Units. JAMA. 1995;274(12):968–974. [PubMed] [Google Scholar]
- 11.Liu V, Kipnis P, Rizk NW, Escobar GJ. Adverse outcomes associated with delayed intensive care unit transfers in an integrated healthcare system. J Hosp Med. 2012;7:224–230. doi: 10.1002/jhm.964. [DOI] [PubMed] [Google Scholar]
- 12.Frost SA, Alexandrou E, Bogdanovski T, Salamonson Y, Parr MJ, Hillman KM. Unplanned admission to intensive care after emergency hospitalisation: risk factors and development of a nomogram for individualizing risk. Resusc. 2009;80:224–230. doi: 10.1016/j.resuscitation.2008.10.030. [DOI] [PubMed] [Google Scholar]
- 13.Delgado MK, Liu V, Pines JM, Kipnis P, Gardner MN, Escobar GJ. Risk factors for unplanned transfer to intensive care within 24 hours of admission from the emergency department in an integrated healthcare system. J Hosp Med. 2013;8:13–9. doi: 10.1002/jhm.1979. [DOI] [PubMed] [Google Scholar]
- 14.Mikkelsen ME, Miltiades AN, Gaieski DF, Goyal M, Fuchs BD, Shah CV, et al. Serum lactate is associated with mortality in severe sepsis independent of organ failure and shock. Crit Care Med. 2009;37:1670–5. doi: 10.1097/CCM.0b013e31819fcf68. [DOI] [PubMed] [Google Scholar]
- 15.Mikkelsen ME, Gaieski DF, Goyal M, Miltiades AN, Munson JC, Pines JM, et al. Factors associated with non-adherence with early goal-directed therapy in the emergency department. Chest. 2010;138:551–8. doi: 10.1378/chest.09-2210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gaieski DF, Mikkelsen ME, Band RA, Pines JM, Massone R, Furia FF, et al. Impact of time to antibiotics on survival in patients with severe sepsis or septic shock in whom early goal-directed therapy was initiated in the emergency department. Crit Care Med. 2010;38:1045–1053. doi: 10.1097/CCM.0b013e3181cc4824. [DOI] [PubMed] [Google Scholar]
- 17.Whittaker SA, Mikkelsen ME, Gaieski DF, Koshy S, Kean C, Fuchs BD. Severe sepsis cohorts derived from claims-based strategies appear to be biased toward a more severely ill patient population. Crit Care Med. 2013;41:945–953. doi: 10.1097/CCM.0b013e31827466f1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mikkelsen ME, Shah CV, Meyer NJ, Gaieski DF, Lyon S, Miltiades AN, et al. The epidemiology of acute respiratory distress syndrome in patients presenting to the emergency department with severe sepsis. Shock. 2013;40:375–381. doi: 10.1097/SHK.0b013e3182a64682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM, Knaus WA, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest. 1992;101:1644–1655. doi: 10.1378/chest.101.6.1644. [DOI] [PubMed] [Google Scholar]
- 20.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. doi: 10.1097/01.CCM.0000050454.01978.3B. [DOI] [PubMed] [Google Scholar]
- 21.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;345:1368–1377. doi: 10.1056/NEJMoa010307. [DOI] [PubMed] [Google Scholar]
- 22.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 23.Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: A severity of disease classification system. Crit Care Med. 1985;13:818–829. [PubMed] [Google Scholar]
- 24.Kennedy M, Joyce N, Howell MD, Lawrence Mottley J, Shapiro NI. Identifying infected emergency department patients admitted to the hospital ward at risk of clinical deterioration and intensive care unit transfer. Acad Emerg Med. 2010;17:1080–5. doi: 10.1111/j.1553-2712.2010.00872.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mato AR, Luger SM, Heitjan DF, Mikkelsen ME, Olson E, Ujjani C, et al. Elevation in serum lactate at the time of febrile neutropenia in hemodynamically-stable patients with hematologic malignancies is associated with the development of septic shock within 48 hours. Cancer Biology and Treatment. 2010;9(8):1–5. doi: 10.4161/cbt.9.8.11270. [DOI] [PubMed] [Google Scholar]
- 26.Shin TG, Jo IJ, Choi DJ, Kang MJ, Jeon K, Suh GY, et al. Adverse effect of emergency department crowding on compliance with the resuscitation bundle in the management of severe sepsis and septic shock. Crit Care. 2013;17:R224. doi: 10.1186/cc13047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Beach MC, Morrison RS. The effect of do-not-resuscitate orders on physician decision-making. J Am Geriatr Soc. 2002;50:2057–2061. doi: 10.1046/j.1532-5415.2002.50620.x. [DOI] [PubMed] [Google Scholar]
- 28.Shepardson LB, Youngner SJ, Speroff T, Rosenthal GE. Increased risk of death in patients with do-not-resuscitate orders. Med Care. 1999;37:727–737. doi: 10.1097/00005650-199908000-00003. [DOI] [PubMed] [Google Scholar]
- 29.Cuzick J. A Wilcoxon-type test for trend. Statistics in Medicine. 1985;4:87–90. doi: 10.1002/sim.4780040112. [DOI] [PubMed] [Google Scholar]
- 30.Maldonado G, Greenland S. Simulation study of confounder-selection strategies. Amer J of Epid. 1993;138:923–936. doi: 10.1093/oxfordjournals.aje.a116813. [DOI] [PubMed] [Google Scholar]
- 31.Stata Statistical Software, Release 12.0. College Station, TX: Stata Corporation; 2011. [Google Scholar]
- 32.Trzeciak S, Dellinger RP, Chansky ME, Arnold RC, Schorr C, Milcarek B, et al. Serum lactate as a predictor of mortality in patients with infection. Int Care Med. 2007;33:970–7. doi: 10.1007/s00134-007-0563-9. [DOI] [PubMed] [Google Scholar]
- 33.Freemantle N, Calvert M, Wood J, Eastaugh J, Griffin C. Composite outcomes in randomized trials: greater precision but with greater uncertainty? JAMA. 2003;289(19):2554–9. doi: 10.1001/jama.289.19.2554. [DOI] [PubMed] [Google Scholar]
- 34.Esteban A, Frutos-Vivar F, Ferguson ND, Penuelas O, Lorente JA, Gordo F, et al. Sepsis incidence and outcome: contrasting the intensive care unit with the hospital ward. Crit Care Med. 2007;35:1284–9. doi: 10.1097/01.CCM.0000260960.94300.DE. [DOI] [PubMed] [Google Scholar]
- 35.The ProCESS Investigators. Yealy DM, Kellum JA, Huang DT, Barnato AE, Weissfeld LA, Pike F, et al. A randomized trial of protocol-based care for early septic shock. New Engl J Med. 2014;370(18):1683–93. doi: 10.1056/NEJMoa1401602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.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;32:1637–1642. doi: 10.1097/01.ccm.0000132904.35713.a7. [DOI] [PubMed] [Google Scholar]
- 37.Walkey AJ, Wiener RS Ghobrial JM, Curtis LH, Benjamin EJ. Incident stroke and mortality associated with new-onset atrial fibrillation in patients hospitalized with severe sepsis. JAMA. 2011;306:2248–2254. doi: 10.1001/jama.2011.1615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Baiocchi M, Cheng J, Small DS. Instrumental variable methods for causal inference. Stat Med. 2014;33(33):2297–2340. doi: 10.1002/sim.6128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1–21. doi: 10.1214/09-STS313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Bakker J, Nijsten MW, Jansen TC. Clinical use of lactate monitoring in critically ill patients. Ann Intensive Care. 2013;3:12. doi: 10.1186/2110-5820-3-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.The ProCESS/ARISE/ProMISe Methodology Writing Committee. Harmonizing international trials of early goal-directed resuscitation for severe sepsis and septic shock: methodology of ProCESS, ARISE, and ProMISe. Intensive Care Med. 2013;39:1760–1775. doi: 10.1007/s00134-013-3024-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Subbe CP, Kruger M, Rutherford P, Gemmel L. Validation of a modified Early Warning Score in medical admissions. Qual J Med. 2001;94:521–526. doi: 10.1093/qjmed/94.10.521. [DOI] [PubMed] [Google Scholar]
- 43.Odden AJ, Rohde JM, Bonham C, Kuhn L, Malani PN, Chen LM, et al. Functional outcomes of general medical patients with severe sepsis. BMC Infectious Dis. 2013;13:588. doi: 10.1186/1471-2334-13-588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Schweickert WD, Pohlman MC, Pohlman AS, Nigos C, Pawlik AJ, Esbrook CL, et al. Early physical and occupational therapy in mechanically ventilated, critically ill patients: a randomized, controlled trial. Lancet. 2009;373:1874–82. doi: 10.1016/S0140-6736(09)60658-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Chalfin DB, Trzeciak S, Likourezos A, Baumann BM, Dellinger RP DELAYED study group. Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Crit Care Med. 2007;35:1477–1483. doi: 10.1097/01.CCM.0000266585.74905.5A. [DOI] [PubMed] [Google Scholar]
- 46.Kerlin MP, Harhay MO, Vranas KC, Cooney E, Ratcliffe SJ, Halpern SD. Objective factors associated with physicians’ and nurses’ perceptions of intensive care unit capacity strain. Ann Am Thorac Soc. 2014;11:167–172. doi: 10.1513/AnnalsATS.201306-141OC. [DOI] [PMC free article] [PubMed] [Google Scholar]