Severe Sepsis in Hematopoietic Stem Cell Transplant Recipients* Gagan Kumar, MD1; Shahryar Ahmad, MD2; Amit Taneja, MD1; Jayshil Patel, MD1; Achuta Kumar Guddati, MD, PhD3; Rahul Nanchal, MD1; the Milwaukee Initiative in Critical Care Outcomes Research Group of Investigators
Objective: Severe sepsis requires timely management and has high mortality if care is delayed. Hematopoietic stem cell transplant recipients are more likely to be immunocompromised and are predisposed to serious infections. Reports of outcomes of severe sepsis in this population are limited to data from single, tertiary care centers, and national outcomes data are missing. Design: Retrospective analysis of an administrative database. Setting: Twenty percent of community hospitals in United States, excluding federal hospitals. Subject: Patients with severe sepsis. Intervention: None. Measurements and Main Results: We used International Classification of Diseases, 9th Edition, Clinical Modification codes indicating the presence of sepsis and organ system failure to identify hospitalizations for severe sepsis between 2000 and 2008. We also used International Classification of Diseases, 9th Edition, Clinical Modification codes to identify hematopoietic stem cell transplant recipients. We compared outcomes of hematopoietic stem cell transplant recipients with severe sepsis during engraftment and subsequent admissions with a non–hematopoietic stem cell transplant cohort and excluded solid-organ transplantation from this cohort. We used *See also p. 501. 1 Department of Medicine, Division of Pulmonary and Critical Care Medicine, Medical College of Wisconsin, Milwaukee, WI. 2 Department of Medicine, Division of Hospital Medicine, Medical College of Wisconsin, Milwaukee, WI. 3 Department of Medicine, Division of Hospital Medicine, Massachusetts General Hospital, Boston, MA. Dr. Kumar designed the study, did statistical analysis, and wrote the article. Dr. Ahmad wrote the article. Drs. Taneja, Patel, and Guddati did critical review and revised the article. Dr. Nanchal designed the study, did statistical analysis, and wrote the article. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ccmjournal). The authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2015 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000714
Critical Care Medicine
mixed effect, multivariate logistic regression modeling with propensity score adjustment to examine factors associated with mortality of severe sepsis in hematopoietic stem cell transplant recipients. A total of 21,898 hematopoietic stem cell transplant recipients with severe sepsis were identified. The frequency of severe sepsis in hematopoietic stem cell transplant recipients was five times higher when compared with the non–hematopoietic stem cell transplant cohort. The unadjusted mortality was 32.9% in non–hematopoietic stem cell transplant cohort, which was similar to autologous hematopoietic stem cell transplant recipients (30.1%) and those who did not develop graft-versus-host disease (35%). Mortality was significantly higher in allogeneic transplants (55.1%, p < 0.001) and in those who developed graft-versus-host disease (47.9%, p < 0.001). After adjustment, during engraftment admission, the odds of in-hospital mortality in allogeneic hematopoietic stem cell transplant (odds ratio, 3.81; 95% CI, 2.39–6.07) and autologous hematopoietic stem cell transplant (odds ratio, 1.28; 95% CI, 1.06–1.53) recipients was significantly higher than non–hematopoietic stem cell transplant patients. Similarly, in subsequent admissions, hematopoietic stem cell transplant recipients with graft-versus-host disease (odds ratio, 2.14; 95% CI, 1.88–2.45) and without graft-versus-host disease (odds ratio, 1.35; 95% CI, 1.19–1.54) had significantly higher odds of mortality than non–hematopoietic stem cell transplant patients. Among patients with hematopoietic stem cell transplant, persons with autologous hematopoietic stem cell transplant and those without graft-versus-host disease fared better as compared with their allogeneic and graft-versus-host disease counterparts. Conclusions: Hematopoietic stem cell transplant recipients are more likely to develop severe sepsis and die following a severe sepsis episode than nontransplant patients. Autologous hematopoietic stem cell transplant recipients and those who do not develop graft-versus-host disease have significantly better outcomes than allogeneic and graft-versus-host disease patients. (Crit Care Med 2015; 43:411–421) Key Words: administrative data; allogeneic hematopoietic stem cell transplant; autologous hematopoietic stem cell transplants; graft-versus-host disease; hematopoietic stem cell transplant; mortality; outcomes; severe sepsis
www.ccmjournal.org
411
Kumar et al
S
evere sepsis has emerged as a global health problem. The incidence continues to rise, and the latest estimates report that nearly one in 40 hospitalizations in the United States alone is complicated by severe sepsis. Patients with hematopoietic stem cell transplantation (HSCT) represent a unique set of individuals who are at particular risk of developing severe sepsis secondary to their immunocompromised state, arising as part of the underlying disorder, the conditioning regimen, or treatment for complications such as graft-versus-host disease (GVHD). Between 3.3% and 55% of such patients require ICU admission (1–13), and severe sepsis is a common reason for such admission (5, 9, 14, 15). Most randomized controlled trials for severe sepsis have excluded HSCT recipients (16–18). A few epidemiological studies have reported outcomes of persons with HSCT admitted to the ICU, but none have specifically examined the consequences of developing severe sepsis (2–4, 6, 7). We, therefore, carried out the present study to characterize the frequency and outcomes of severe sepsis in adult patients with HSCT and compare them to persons without HSCT. To better understand this phenomenon, we analyzed two distinct cohorts: persons admitted for engraftment and persons admitted after engraftment. We further classified persons admitted for engraftment into autologous and allogeneic and those admitted after engraftment into people with GVHD and those without GVHD. We hypothesized that the rates of severe sepsis and associated mortality would be higher in the cohort with HSCT. We used a large administrative dataset maintained by the Agency of Health Care Quality and Research (AHRQ) to enhance the generalizability of our results.
METHODS Data Source We used the Healthcare Cost and Utilization Project–Nationwide Inpatient Sample (NIS) database from the year 2000– 2008 for the purposes of our study. NIS was created by the AHRQ as part of the Healthcare Cost and Utilization Project and contains data on hospital stays from about 1,000 hospitals sampled to approximate a 20% stratified sample of U.S. community hospitals (19). Each hospitalization is treated as an individual entry in the database and is coded with one principal diagnosis, up to 14 secondary diagnoses, and 15 procedural diagnoses associated with that stay. NIS includes information on all hospitalizations, regardless of the payer. To facilitate the production of national estimates, both hospital and discharge weights are provided. The details of NIS structure are available online (19). Since we used a publicly available dataset with no patient identifiers, this study was deemed exempt from review by the Institutional Review Board of the Medical College of Wisconsin, Milwaukee, WI. Study Population We used the International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM) codes to identify patients 412
www.ccmjournal.org
with HSCT who are 18 years old or older. We identified classified persons into the engraftment admission cohort if they had an ICD-9-CM code indicating the receipt of the HSCT procedure. Persons with an ICD-9-CM diagnosis code indicating HSCT but without a procedure code for HSCT were classified as subsequent admission. Further, using standard ICD-9-CM codes, we separated engraftment admissions into autologous and allogeneic and subsequent admissions into persons with GVHD and those without GVHD. The schema of our classification is shown in Figure 1. We excluded kidney, pancreas, heart, liver, and lung transplant from our comparator group and termed this group “non-HSCT.” Based on the coding system previously used and validated by Martin et al (20), we defined severe sepsis as either 1) ICD9-CM code 995.92 (systemic inflammatory response syndrome due to infectious process with organ failure) or 2) ICD-9-CM code for septicemia, bacteremia, or fungemia and at least one organ failure code. We identified comorbidities using ICD-9-CM codes as shown in Appendix 1 (Supplemental Digital Content 1, http:// links.lww.com/CCM/B128). We similarly identified the underlying etiology for undergoing HSCT—multiple myeloma (ICD-9-CM code 203), acute lymphoid leukemia (ICD-9-CM code 204), acute myeloid leukemia (AML) (ICD-9-CM code 205), non-Hodgkin lymphoma (NHL) (ICD-9-CM code 200, 202), and Hodgkin lymphoma (ICD-9-CM code 201). We used ICD-9-CM codes to identify patients who underwent mechanical ventilation (ICD-9-CM code 96.70– 96.72), tracheostomy (ICD-9-CM code 31.1, 31.21, and 31.29) or had new dialysis, blood transfusion (ICD-9-CM code 99.03 and 99.04), or total parenteral nutrition (ICD9-CM code 99.15). We considered a patient having received new dialysis if they had an ICD-9-CM code indicating acute renal failure (ICD-9-CM code 584.5–584.9) and a procedure code for hemodialysis (ICD-9-CM code 39.95). We used number of organs failing as a surrogate marker for severity of acute illness (20). We used NIS variables to identify patient age, gender, race, and primary payer. We divided age into four categories: 18– 34 years, 35–49 years, 50–64 years, and 65 years and above. Information of race is missing in 20–25% of records, and we categorized these as “unknown.” As per the NIS, we classified the primary payer into five groups: 1) private including Health Maintenance Organization, 2) Medicare, 3) Medicaid, 4) selfpay, and 5) others. We also used NIS classifications for a hospital’s teaching status and size. Outcomes Our primary outcomes of interest were the frequency of hospitalizations for severe sepsis and all-cause in-hospital mortality. Secondary outcomes included length of hospital stay and discharge dispositions. We used the NIS variable “DISPUNIFORM” to classify discharge disposition as deaths, discharges to home, discharges to skilled nursing facilities, discharges with home care, and others (includes discharges against medical advice and others). February 2015 • Volume 43 • Number 2
Clinical Investigations
Figure 1. Identification and grouping of hematopoietic stem cell transplant (HSCT) patients into four groups. GVHD = graft-versus-host disease, ICD-9-CM = International Classification of Diseases, 9th Edition, Clinical Modification.
Statistical Analysis We performed all statistical analysis in STATA IC 11.0 (STATA Corp, College Station, TX), using the strata weights and survey commands to generate national estimates of patients with HSCT admitted to the hospital with severe sepsis. We used Pearson chi-square test to compare categorical and Wilcoxon rank test to compare nonnormally distributed variables. We made pairwise comparisons between each of the four individual HSCT groups and the non-HSCT group. We used Bonferroni correction for multiple comparisons. Since we made four comparisons, p value was kept at 0.0125. We assessed predictors for in-hospital mortality in severe sepsis using multivariable logistic regression modeling. The model accounted for the survey design of the NIS. First, we used single predictor logistic regression to identify associations between putative risk factors for mortality. Variables that were significant in the univariate analysis at p value of less than 0.10 were candidates for inclusion in the final multivariate model. These variables were then checked for multicollinearity using tolerance and a variable Critical Care Medicine
inflation factor. We examined two-way interaction between the variables and HSCT; all interactions with p value of less than 0.10 were kept in the model. We then developed a propensity score approach to further adjust for differences by constructing a logistic regression model to calculate the likelihood that a person would have HSCT. This model included factors regardless of their individual statistical significance. Thus, we included study year, patient age, gender, race, hospital characteristics and severity of sepsis (number of organ failures), and individual comorbidities. We then divided the propensity score into quintiles and used it as a variable in the previously developed model. To exclude that the differences in mortality reflected a differential hospital effect, that is, more transplanted patients being treated in tertiary care and academic centers, we constructed a multilevel model that included a random effect for each individual hospital along with the fixed effects for all the variables in the previous model. The covariance structure used a distinct variance for each random effect within a random-effects equation and assumed that all covariances were zero. We thus constructed four models www.ccmjournal.org
413
Kumar et al
for each HSCT group (autologous and allogeneic in the engraftment admission and GVHD and no GVHD in the subsequent admission) and compared each to the cohort without HSCT. We used similar techniques and constructed multivariable models to examine risk factors associated with higher mortality within each HSCT cohort (engraftment admission and subsequent admission) with severe sepsis.
RESULTS Of the 291,179 discharges with HSCT from 2000 to 2008, severe sepsis developed in 21,898 patients (7.5%). The frequency of severe sepsis in HSCT recipients was five times higher when compared with the non-HSCT cohort (Table 1). Of all HSCT recipients developing severe sepsis, 6,168 developed severe sepsis during their engraftment admission. Of these, 2,750 received autologous and 3,418 received allogeneic HSCT. Severe sepsis developed in 15,730 patients with HSCT during their subsequent hospitalizations. Of these, 6,713 had GVHD. Rates of severe sepsis in HSCT recipients during engraftment admission through years 2000 to 2008 is shown in Figure 1. Demographical Characteristics HSCT recipients were younger than non-HSCT cohort. Male patients constituted a larger proportion of patients in each HSCT category. Teaching hospitals accounted for the majority of admissions for persons with HSCT compared with persons without HSCT. Significantly higher proportion of HSCT recipients was privately insured. Demographic details of these patients are shown in Table 1. Clinical Characteristics Engraftment Admission. All studied comorbid conditions except pulmonary embolism were lower in the HSCT cohort as compared with non-HSCT. Patients with allogeneic transplant had higher rates of severe sepsis as compared with autologous (13.2% vs 5.2%; p < 0.001) (Table 1). During the engraftment admission, the proportion of patients with three or more organ failures was lower in autologous transplants (18%) in comparison with allogeneic transplant (24.5%) and non-HSCT patients (20.3%). Allogeneic HSCT recipients had higher rates of respiratory failure and renal failure as compared with autologous and non-HSCT recipients. Rates of new dialysis were significantly higher in allogeneic transplants when compared with autologous and non-HSCT cohorts (19.3% vs 11.4% vs 5.7%, p < 0.001) (Table 2). Rates of cardiac failure were lowest in allogeneic recipients in comparison with autologous and non-HSCT cohorts. Subsequent Admission. Patients with GVHD developed severe sepsis more often than those who did not develop GVHD (10.4% vs 6.1%, p < 0.001) (Table 1). The proportion of patients with three or more organ failure was highest in HSCT recipients with GVHD followed by HSCT recipients without GVHD and non-HSCT patients (Table 2). Rates of respiratory, cardiac, and renal failure, though similar among persons with GVHD and non-HSCT cohort, remained higher than persons without GVHD. Occurrence of new dialysis was more frequent in GVHD cohort (Table 2). 414
www.ccmjournal.org
Outcomes Engraftment Admission. Although the unadjusted in-hospital mortality of allogeneic HSCT recipients with severe sepsis was significantly higher than non-HSCT patients (55.1% vs 32.9%, p < 0.001), the mortality rates for autologous HSCT recipients were no different than the non-HSCT cohort (Fig. 3A). On adjusted analysis, the odds of mortality were 3.81 times (95% CI, 2.39–6.07) higher in allogeneic HSCT recipients and 1.28 times (95% CI, 1.06–1.53) higher in autologous HSCT recipients when compared with non-HSCT patients (Fig. 4). The length of stay (LOS) in survivors was significantly longer in allogeneic transplant recipients (median, 39 d) as compared with autologous (median, 26 d) and non-HSCT patients (median, 11 d) (Table 3). Subsequent Admission. The in-hospital mortality was significantly higher in HSCT recipients with GVHD than nonHSCT patients (47.9% vs 32.9%, p < 0.001). However, in HSCT recipients without GVHD, mortality was no different (Fig. 3B). After adjustment, the odds of mortality were 2.14 times (95% CI, 1.88–2.45) higher in HSCT recipients with GVHD and 1.35 times (95% CI, 1.19–1.54) higher in HSCT recipients without GVHD when compared with non-HSCT patients (Fig. 4). The LOS in survivors was similar in GVHD, non-GVHD, and nonHSCT cohorts (Table 3). Discharge Destinations Discharges to home and home with healthcare were significantly higher in all HSCT cohorts (Table 3). Factors Associated With Mortality in HSCT Patients With Severe Sepsis Engraftment Admission. Allogeneic transplant recipients had 2.12 times higher odds of mortality than autologous recipients. Multiple myeloma as the reason for receiving HSCT was associated with lower odds of death with the occurrence of severe sepsis. Other factors associated with worse odds of mortality were advanced age and cirrhosis. Organ failures, particularly respiratory (odds ratio [OR], 12.1; 95% CI, 8.64–16.8) and hepatic (OR, 5.22; 95% CI, 2.29–11.8), were the strongest predictors of mortality. We did not find association of methicillin-resistant Staphylococcus aureus, pseudomonas, Clostridium difficile, aspergillus, or candida infections with increased mortality (Table 4). Subsequent Admission. During subsequent admissions, presence of GVHD was associated with 1.38 times higher odds of mortality when compared with those without GVHD. AML and NHL as reasons for HSCT were associated with higher odds of mortality. Other factors included advancing age, female gender, presence of stroke, nonvariceal upper gastrointestinal hemorrhage, and aspergillus infection. Teaching hospitals and larger hospitals had higher mortality. Similar to engraftment admission, respiratory (OR, 10.1; 95% CI, 8.47– 11.9) and hepatic failure (OR, 4.39; 95% CI, 2.54–7.59) had the strongest association with mortality. Sensitivity analyses were performed using 1) individual organ failures in place of total number of organ failure, 2) individual comorbidities in place of Charlson comorbidity index, and 3) February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 1. Demographic Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis From 2000 to 2008 HSCT Engraftment Admission Variable
Non-HSCT
Total admissions
286,285,073
Severe sepsis, n (%)
4,654,341 (1.6)
Autologous
53,337 2,750 (5.2)
Allogeneic
25,950 3,418 (13.2)
Subsequent Admission GVHD
64,878 6,713 (10.4)
No GVHD
146,909 9,017 (6.1)
Age category,a yr 18–34
3.9
10.8
26
22.4
13.5
35–49
10.6
20.4
33.4
32
24.9
50–64
22.9
47.2
36.3
39.3
45.8
≥ 65
62.7
21.6
4.3
6.3
15.7
49.7
43.4
36.6
39.4
42.6
White
52.6
57.9
55
56.3
60.1
African American
12.7
6.2
4.8
5.4
7.0
Hispanic
7.3
7.4
10.3
8.5
7.0
Asian
2.2
1.4
3.4
2.5
2.5
Others
2.5
3.4
2.9
2.1
2.4
22.8
23.7
23.5
25.2
21
Private
17.8
59.2
69.6
64.8
58.3
Medicare
66.5
24.9
11.7
18.7
28.5
Medicaid
10.3
10.8
13.6
11.7
8.3
Self
2.9
1.8
1.4
2.1
1.8
Others
2.5
3.2
3.8
2.7
3.1
46.2
94.5
97.1
89.3
77.1
Small
10.7
10.8
12.1
8.7
9.8
Medium
24.7
13.9
8.5
7.5
16.1
Large
64.7
75.3
79.3
83.8
74.2
Sex
a
Female Race
Unknown Primary payer
a
Hospital characteristics Teaching statusa Bed size
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125.
excluding the unknown category of race. The OR and their CIs varied by less than 5% for each sensitivity analysis performed.
DISCUSSION Using nationally representative data, we show that during the engraftment admission, severe sepsis occurs more frequently in HSCT recipients than non-HSCT patients. Furthermore, the associated odds of mortality, after adjustment for confounding Critical Care Medicine
factors, are higher for both allogeneic and autologous HSCT. Similarly, during subsequent admissions, HSCT recipients have higher rates of severe sepsis and higher odds of death in both GVHD and non-GVHD cohorts. Prior studies mostly performed at single centers have reported outcomes of HSCT patients admitted to the ICU for a variety of diagnoses (1, 4, 7, 9). The strengths of our study are its large size, national representation, and restricted attention to severe sepsis which is a www.ccmjournal.org
415
Kumar et al
Table 2. Clinical Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Non-HSCT
Subsequent Admission
Autologous
Allogeneic
GVHD
No GVHD
Comorbidities (%) Congestive heart failurea
28.4
10.2
10.5
8.3
8.9
Diabetes mellitusa
22.4
6.8
4.1
6.3
8.1
Cirrhosisa
7.0
2.4
7.9
7.5
4.2
End-stage kidney diseasea
5.9
3.7
0.9
2.9
4.3
Chronic pulmonary diseasea
34.3
11.8
8.2
9.9
13.1
3.8
2.8
1.9
2.5
1.6
10.1
8.1
1.5
1.6
4.9
7.4
4.7
9.8
14.7
9.6
6.9
8.8
1.7
1.8
1.0
2.2
0.8
0.03
1.6
0
0
0.9
0.23
7.8
3.2
2.2
2.8
0.15
10.4
1.8
3.8
5.8
0.8
37
4.5
7.4
25.2
Strokea Solid organ cancera Nonvariceal upper gastrointestinal bleed
a
Atrial fibrillation
20
a
Pulmonary embolism
10
4.2
Etiology (%) Amyloidosisa Non-Hodgkin lymphoma Hodgkin lymphoma Multiple myeloma
a
a
a
Acute lymphocytic leukemia Acute myeloid leukemia
a
a
0.8
1.0
16.7
14.6
7.5
0.9
6.9
45.4
40.1
23.6
Number of organ system failures (%) One
52.5
60
48
48.1
52.4
Two
27.3
21.9
27.5
27
24.6
Three
13.5
12.4
15.9
16.3
13.5
Four
5.2
4.7
6.7
6.7
6.2
Five
1.4
0.9
1.7
1.6
2.8
Six or seven
0.2
0
0
0.3
0.5
20.3
18.0a
24.5a
24.9a
23.0a
Respiratorya
52.5
41.7
57.1
50
42.1
Cardiaca
30.3
24.6
20.2
28.7
33.8
46.8
44
59.3
47
40.8
15.7
32.5
22.5
33.7
40.4
16.5
14.2
12.6
16
16.5
7.4
4.0
5.4
4.0
3.6
5.0
2.4
9.9
7.2
5.3
Three or more organ failures Type of organ failure (%)
Renal
a
Hematological
a
Metabolic Neurological Hepatic
a
a
(Continued)
416
www.ccmjournal.org
February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 2. (Continued) Clinical Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Non-HSCT
Autologous
Subsequent Admission
Allogeneic
GVHD
No GVHD
Infection (%) 5.9
12
13.4
9.2
5.8
Gram positivea
27.8
33
38.3
36.1
32.9
Gram negativea
20.9
16
15.8
25.3
22
Clostridium difficilea
Aspergillusa
0.21
2.2
5.8
5.8
2.8
Candidaa
1.5
3.3
5.4
2.7
1.6
39.4
30.9
40.4
37.6
31.8
6.8
3.8
5.8
3.4
2.8
23.6
39.4
35.8
35
37.5
5.7
11.4
19.3
11
8.2
6.0
13.7
21.5
17.7
6.8
Procedures (%) Invasive mechanical ventilationa Tracheostomy Blood
a
a
New dialysis
a
Total parenteral nutrition
a
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125. All comparisons are with non–hematopoietic stem cell transplant group.
with autologous (319/588, 58.3%) recipients. We report 55% in-hospital mortality in allogeneic and 30% in autologous HSCT recipients. The numbers reported in literature may be higher due to the fact that acute lung injury is another common cause for admission of these patients to an ICU and is associated with even higher mortality up to 80–85% (3–5, 7, 9, 21). Other studies examining the outcomes of ICU stay during transplant admissions report ICU mortality ranging from 38% to 72% (3, 5, 7–9, 13). Limitations of these studies include being single-center studies, smaller sample size and reporting the overall ICU mortality, and not specifically describing mortality in severe sepsis. In these studies, sepsis and respiratory failure were reported to be commonest cause of admission. These studies do not report more recent outcomes for severe sepsis after initiation of early goal–directed therapies and other advances in taking care of patients with sepsis. Thus, the higher mortality may represent the pre–early goal–directed therapy era (16). A recent single-center study from Canada in autologous HSCT by Trinkaus et al (5) reported 55% ICU mortality in patients with sepsis, although the numbers are very small (only 11 patients with sepsis). Lungs and kidneys were the most common organs to fail in both HSCT and nontransplant patients (9, 22). However, this was significantly higher for allogeneic transplant patients. The Figure 2. Rate of hematopoietic stem cell transplant recipients developing severe sepsis during engraftment admission. rates of tracheostomy were lower in the HSCT frequent occurrence and perhaps the primary reason for ICU admission in HSCT recipients. Further, our analysis of four distinct HSCT cohorts adds to the previously reported literature, as risk factors for the development and outcomes from severe sepsis are probably very different in each cohort. Among HSCT recipients developing severe sepsis, we report two-fold higher odds for mortality for allogeneic HSCT recipients than autologous recipients. In a review, Afessa and Azoulay (9) examined literature reporting outcomes of HSCT patients admitted to the ICU and found that mortality was higher in allogeneic (604/867; 70%) when compared
Critical Care Medicine
www.ccmjournal.org
417
Kumar et al
Figure 3. Trends in mortality with severe sepsis from 2000 to 2008. Error bars show sem. A, Comparison of the mortality trends of severe sepsis during engraftment admission—autologous and allogeneic hematopoietic stem cell transplant with nontransplant group. B, Comparison of the mortality trends of severe sepsis during subsequent admissions—graft-versus-host disease (GVHD) status with nontransplant group.
(25) demonstrated that mortality of HSCT recipients receiving mechanical ventilation was very high and had not changed in 20 years. In parallel, the mortality of HSCT recipients receiving invasive mechanical ventilation remained unchanged between 70% and 75% during our study period from 2000 to 2008 (Appendix 3, Supplemental Digital Content 1, http://links.lww. com/CCM/B128). The differences in numbers may reflect a shift of ventilation strategies toward lung protective ventilation (26) and our analysis of the specific diagnosis of severe sepsis. Although we used a well-characterized national database, our study has important limitations. Our choice of codes to characterize organ failure, although identical to those validated by Martin et al (20), may not reliably identify sepsis-related organ failure. Specifically, we cannot exclude the possibility of variations in coding practices between hospitals as there is lack of standardization in the administrative literature. Further temporal association between the organ failures and occurrence of infection cannot be elicited from the database. Because the NIS is a discharge-level database, we were unable to identify that multiple hospitalizations may have been by the same patient. This also precluded us from using readmissions as an outcome measure. However, this limitation would apply to “subsequent admissions” only since engraftment admissions were selected using the procedure codes. During engraftment admissions, we also could not differentiate whether the severe sepsis occurred during the preengraftment, early postengraftment phase, or late posttransplant phase. Since differences in immunity and predilection to infections occur during each of these stages, inclusion of such information would have enhanced the robustness of our multivariable models. Secondary to the administrative nature of our data source were also unable to determine the conditioning regimen, associated neutropenia and its duration, the receipt of total body irradiation, and the status and duration of prophylactic antibiotics. The limited clinical detail of Figure 4. Forrest plot showing odds ratio of mortality in severe sepsis—hematopoietic our ICD-9 codes may not detect important severstem cell transplant admissions compared with nonhematopoietic stem cell transplant. ity differences. For example, different degrees of GVHD = graft-versus-host disease. patients; however, this may simply reflect significantly higher mortality in those developing respiratory failure in HSCT recipients. We report that renal replacement therapy was used in 11.4% for autologous and 19.3% for allogeneic HSCT recipients. These numbers are remarkably similar to other reports in the literature (2, 4, 7). Although observed rates of renal failure were similar or lower in autologous HSCT and non-GVHD cohorts as compared with the non-HSCT cohort, the utilization of new dialysis was 2–3 times higher. This may suggest either greater degrees of renal injury not described by ICD9-CM codes or provision of more aggressive care in HSCT recipients. Similar to other studies, we found that the number of failing organs were the strongest predictors of mortality in HSCT recipients in whom severe sepsis develops (10, 23). The odds of mortality increased with each additional organ failing (Appendix 2, Supplemental Digital Content 1, http://links. lww.com/CCM/B128). Individually, respiratory (OR, 10.3) and hepatic (OR, 5.1) failure had the strongest association with mortality. A few small studies have previously reported worse outcomes of HSCT recipients developing liver dysfunction (5, 7). Conversely, mechanical ventilation is a well-established risk factor for incremental mortality in HSCT recipients (2–6, 8, 21, 24). In a systemic review published in 2001, Bach et al
418
www.ccmjournal.org
February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 3. Outcomes of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Subsequent Admission
Non-HSCT
Autologous
Allogeneic
GVHD
32.9
30.1
55.1a
47.9a
35
Home
27.3
58.7a
58.4a
45.7a
50.3a
Home care
16
26.9a
33a
33.5a
29a
Facility
49.4
12.7a
6.6a
14.6a
13.2a
6.3
1.5a
2.0a
6.1
7.3
Mortality (%)
No GVHD
Disposition in survivors (%)
Transfers Time to death (d)b Length of stay in survivors (d)b
7 (2–16)
34 (21–54)a
44 (27–61)a
21 (10–37)a
11 (6–20)
26 (21–37)a
39 (30–58)a
12 (6–29)
10 (4–23) 9 (5–17)
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125. b Median (interquartile range). All comparisons are made with non–hematopoietic stem cell transplant group.
Table 4. Predictor Variables Associated With Mortality in Hematopoietic Stem Cell Transplant Patients With Severe Sepsis
Variable
Engraftment Admission
Subsequent Admission
OR
OR
Hematopoietic stem cell transplant status Autologous
Reference
Allogeneic
2.12 (1.55–2.90)
No GVHD
Reference
GVHD
1.38 (1.14–1.67)
Age category (yr) 18–34
Reference
Reference
35–49
1.68 (1.08–2.60)
1.38 (1.03–1.85)
50–64
1.31 (0.81–2.08)
1.68 (1.28–2.21)
≥ 65
2.08 (1.13–3.83)
2.74 (1.85–4.05)
Reference
Reference
NS
1.25 (1.02–1.53)
Stroke
2.03 (0.77–5.37)
2.26 (1.17–4.37)
Cirrhosis
4.49 (1.81–11.1)
NS
Nonvariceal upper gastrointestinal bleed
1.85 (0.98–3.49)
3.63 (2.43–5.43)
0.59 (0.40–0.89)
NS
Sex Male Female Individual comorbidities
Etiology Multiple myeloma
(Continued)
Critical Care Medicine
www.ccmjournal.org
419
Kumar et al
Table 4. (Continued) Predictor Variables Associated With Mortality in Hematopoietic Stem Cell Transplant Patients With Severe Sepsis Engraftment Admissiona
Subsequent Admissionb
Variable
OR
OR
Acute myeloid leukemia
NS
1.67 (1.37–2.02)
Acute lymphocytic leukemia
NS
NS
Non-Hodgkin lymphoma
NS
1.76 (1.07–2.91)
Hodgkin lymphoma
NS
NS
NS
1.78 (1.12–2.83)
Respiratory
12.1 (8.64–16.8)
10.1 (8.47–11.9)
Cardiac
2.42 (1.59–3.66)
1.64 (1.33–2.02)
Renal
2.64 (1.95–3.56)
1.83 (1.53–2.19)
NS
NS
1.63 (1.07–2.49)
1.54 (1.16–2.04)
NS
NS
5.22 (2.29–11.8)
4.39 (2.54–7.59)
Reference
Reference
NS
1.51 (1.15–1.99)
Reference
Reference
Medium
NS
1.67 (0.96–2.91)
Large
NS
1.93 (1.18–3.17)
2000
Reference
Reference
2001
0.96 (0.49–1.88)
0.76 (0.43–1.33)
2002
0.44 (0.23–0.83)
0.61 (0.36–1.03)
2003
0.91 (0.51–1.60)
0.59 (0.33–1.06)
2004
0.52 (0.25–1.06)
0.45 (0.28–0.71)
2005
0.40 (0.23–0.72)
0.52 (0.31–0.86)
2006
0.36 (0.18–0.72)
0.47 (0.29–0.75)
2007
0.37 (0.21–0.60)
0.33 (0.21–0.53)
2008
0.34 (0.19–0.60)
0.34 (0.21–0.54)
Infection Aspergillus Type of organ failure
Hematological Metabolic Neurological Hepatic Hospital type Nonteaching Teaching Small
Year
OR = odds ratio, GVHD = graft-versus-host disease, NS = not significant.
renal failure cannot be distinguished using these codes. Thus, it is possible that the improving survival we observed simply reflects decreasing severity that is not detected by the available codes. Finally, we were unable to identify those persons who may have refused escalation of care and preferred a palliative or hospice approach. This would falsely lower our estimates of mortality in each group. 420
www.ccmjournal.org
Despite its limitations, our study provides important information about severe sepsis in an understudied population. HSCT recipients are much more likely to be admitted with or develop severe sepsis once hospitalized than their non-HSCT counterparts. Once sepsis occurs, they are also more likely to suffer worse mortality. Clinicians should be vigilant about the occurrence of severe sepsis in this population and pay particular February 2015 • Volume 43 • Number 2
Clinical Investigations
attention to therapies that are known to improve outcomes such as early administration of antibiotics. Furthermore, we also report novel risk factors of mortality within the HSCT cohorts developing severe sepsis, and these should be incorporated into clinical decision making and prognostication.
REFERENCES
1. Horak DA, Forman SJ: Critical care of the hematopoietic stem cell patient. Crit Care Clin 2001; 17:671–695 2. Jackson SR, Tweeddale MG, Barnett MJ, et al: Admission of bone marrow transplant recipients to the intensive care unit: Outcome, survival and prognostic factors. Bone Marrow Transplant 1998; 21:697–704 3. Kew AK, Couban S, Patrick W, et al: Outcome of hematopoietic stem cell transplant recipients admitted to the intensive care unit. Biol Blood Marrow Transplant 2006; 12:301–305 4. Scales DC, Thiruchelvam D, Kiss A, et al: Intensive care outcomes in bone marrow transplant recipients: A population-based cohort analysis. Crit Care 2008; 12:R77 5. Trinkaus MA, Lapinsky SE, Crump M, et al: Predictors of mortality in patients undergoing autologous hematopoietic cell transplantation admitted to the intensive care unit. Bone Marrow Transplant 2009; 43:411–415 6. Paz HL, Garland A, Weinar M, et al: Effect of clinical outcomes data on intensive care unit utilization by bone marrow transplant patients. Crit Care Med 1998; 26:66–70 7. Pène F, Aubron C, Azoulay E, et al: Outcome of critically ill allogeneic hematopoietic stem-cell transplantation recipients: A reappraisal of indications for organ failure supports. J Clin Oncol 2006; 24:643–649 8. Soubani AO, Kseibi E, Bander JJ, et al: Outcome and prognostic factors of hematopoietic stem cell transplantation recipients admitted to a medical ICU. Chest 2004; 126:1604–1611 9. Afessa B, Azoulay E: Critical care of the hematopoietic stem cell transplant recipient. Crit Care Clin 2010; 26:133–150 10. Afessa B, Tefferi A, Dunn WF, et al: Intensive care unit support and Acute Physiology and Chronic Health Evaluation III performance in hematopoietic stem cell transplant recipients. Crit Care Med 2003; 31:1715–1721 11. Cargo CA, Yates E, Marley C, et al: Allogeneic bone marrow transplant in Belfast—An outcome overview of the first 25 years. Ulster Med J 2008; 77:185–190 12. Gil L, Styczynski J, Komarnicki M: Infectious complication in 314 patients after high-dose therapy and autologous hematopoietic stem
Critical Care Medicine
cell transplantation: Risk factors analysis and outcome. Infection 2007; 35:421–427 13. Almyroudis NG, Fuller A, Jakubowski A, et al: Pre- and post-engraftment bloodstream infection rates and associated mortality in allogeneic hematopoietic stem cell transplant recipients. Transpl Infect Dis 2005; 7:11–17 14. Mitsui H, Karasuno T, Santo T, et al: Analysis of sepsis in allogeneic bone marrow transplant recipients: A single-center study. J Infect Chemother 2003; 9:238–242 15. Lalitha MK, Pai R, Manoharan A, et al: Systemic bacterial infections in bone marrow transplant patients. Indian J Cancer 2000; 37:10–14 16. Rivers E, Nguyen B, Havstad S, et al; Early Goal-Directed Therapy Collaborative Group: Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001; 345:1368–1377 17. Sprung CL, Annane D, Keh D, et al; CORTICUS Study Group: Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008; 358:111–124 18. Bernard GR, Vincent JL, Laterre PF, et al; Recombinant human protein C Worldwide Evaluation in Severe Sepsis (PROWESS) study group: Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344:699–709 19. Introduction to the HCUP Nationwide Inpatient Sample (NIS) 2008. Available at: http://www.hcup-us.ahrq.gov/db/nation/nis/NIS_ Introduction_2008.jsp. Accessed January 21, 2014 20. Martin GS, Mannino DM, Eaton S, et al: The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med 2003; 348:1546–1554 21. Naeem N, Eyzaguirre A, Kern JA, et al: Outcome of adult umbilical cord blood transplant patients admitted to a medical intensive care unit. Bone Marrow Transplant 2006; 38:733–738 22. Létourneau I, Dorval M, Bélanger R, et al: Acute renal failure in bone marrow transplant patients admitted to the intensive care unit. Nephron 2002; 90:408–412 23. Haire WD, Ruby EI, Gordon BG, et al: Multiple organ dysfunction syndrome in bone marrow transplantation. JAMA 1995; 274:1289–1295 24. Huaringa AJ, Leyva FJ, Giralt SA, et al: Outcome of bone marrow transplantation patients requiring mechanical ventilation. Crit Care Med 2000; 28:1014–1017 25. Bach PB, Schrag D, Nierman DM, et al: Identification of poor prognostic features among patients requiring mechanical ventilation after hematopoietic stem cell transplantation. Blood 2001; 98:3234–3240 26. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308
www.ccmjournal.org
421
Objective: Severe sepsis requires timely management and has high mortality if care is delayed. Hematopoietic stem cell transplant recipients are more likely to be immunocompromised and are predisposed to serious infections. Reports of outcomes of severe sepsis in this population are limited to data from single, tertiary care centers, and national outcomes data are missing. Design: Retrospective analysis of an administrative database. Setting: Twenty percent of community hospitals in United States, excluding federal hospitals. Subject: Patients with severe sepsis. Intervention: None. Measurements and Main Results: We used International Classification of Diseases, 9th Edition, Clinical Modification codes indicating the presence of sepsis and organ system failure to identify hospitalizations for severe sepsis between 2000 and 2008. We also used International Classification of Diseases, 9th Edition, Clinical Modification codes to identify hematopoietic stem cell transplant recipients. We compared outcomes of hematopoietic stem cell transplant recipients with severe sepsis during engraftment and subsequent admissions with a non–hematopoietic stem cell transplant cohort and excluded solid-organ transplantation from this cohort. We used *See also p. 501. 1 Department of Medicine, Division of Pulmonary and Critical Care Medicine, Medical College of Wisconsin, Milwaukee, WI. 2 Department of Medicine, Division of Hospital Medicine, Medical College of Wisconsin, Milwaukee, WI. 3 Department of Medicine, Division of Hospital Medicine, Massachusetts General Hospital, Boston, MA. Dr. Kumar designed the study, did statistical analysis, and wrote the article. Dr. Ahmad wrote the article. Drs. Taneja, Patel, and Guddati did critical review and revised the article. Dr. Nanchal designed the study, did statistical analysis, and wrote the article. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (http://journals.lww.com/ccmjournal). The authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2015 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0000000000000714
Critical Care Medicine
mixed effect, multivariate logistic regression modeling with propensity score adjustment to examine factors associated with mortality of severe sepsis in hematopoietic stem cell transplant recipients. A total of 21,898 hematopoietic stem cell transplant recipients with severe sepsis were identified. The frequency of severe sepsis in hematopoietic stem cell transplant recipients was five times higher when compared with the non–hematopoietic stem cell transplant cohort. The unadjusted mortality was 32.9% in non–hematopoietic stem cell transplant cohort, which was similar to autologous hematopoietic stem cell transplant recipients (30.1%) and those who did not develop graft-versus-host disease (35%). Mortality was significantly higher in allogeneic transplants (55.1%, p < 0.001) and in those who developed graft-versus-host disease (47.9%, p < 0.001). After adjustment, during engraftment admission, the odds of in-hospital mortality in allogeneic hematopoietic stem cell transplant (odds ratio, 3.81; 95% CI, 2.39–6.07) and autologous hematopoietic stem cell transplant (odds ratio, 1.28; 95% CI, 1.06–1.53) recipients was significantly higher than non–hematopoietic stem cell transplant patients. Similarly, in subsequent admissions, hematopoietic stem cell transplant recipients with graft-versus-host disease (odds ratio, 2.14; 95% CI, 1.88–2.45) and without graft-versus-host disease (odds ratio, 1.35; 95% CI, 1.19–1.54) had significantly higher odds of mortality than non–hematopoietic stem cell transplant patients. Among patients with hematopoietic stem cell transplant, persons with autologous hematopoietic stem cell transplant and those without graft-versus-host disease fared better as compared with their allogeneic and graft-versus-host disease counterparts. Conclusions: Hematopoietic stem cell transplant recipients are more likely to develop severe sepsis and die following a severe sepsis episode than nontransplant patients. Autologous hematopoietic stem cell transplant recipients and those who do not develop graft-versus-host disease have significantly better outcomes than allogeneic and graft-versus-host disease patients. (Crit Care Med 2015; 43:411–421) Key Words: administrative data; allogeneic hematopoietic stem cell transplant; autologous hematopoietic stem cell transplants; graft-versus-host disease; hematopoietic stem cell transplant; mortality; outcomes; severe sepsis
www.ccmjournal.org
411
Kumar et al
S
evere sepsis has emerged as a global health problem. The incidence continues to rise, and the latest estimates report that nearly one in 40 hospitalizations in the United States alone is complicated by severe sepsis. Patients with hematopoietic stem cell transplantation (HSCT) represent a unique set of individuals who are at particular risk of developing severe sepsis secondary to their immunocompromised state, arising as part of the underlying disorder, the conditioning regimen, or treatment for complications such as graft-versus-host disease (GVHD). Between 3.3% and 55% of such patients require ICU admission (1–13), and severe sepsis is a common reason for such admission (5, 9, 14, 15). Most randomized controlled trials for severe sepsis have excluded HSCT recipients (16–18). A few epidemiological studies have reported outcomes of persons with HSCT admitted to the ICU, but none have specifically examined the consequences of developing severe sepsis (2–4, 6, 7). We, therefore, carried out the present study to characterize the frequency and outcomes of severe sepsis in adult patients with HSCT and compare them to persons without HSCT. To better understand this phenomenon, we analyzed two distinct cohorts: persons admitted for engraftment and persons admitted after engraftment. We further classified persons admitted for engraftment into autologous and allogeneic and those admitted after engraftment into people with GVHD and those without GVHD. We hypothesized that the rates of severe sepsis and associated mortality would be higher in the cohort with HSCT. We used a large administrative dataset maintained by the Agency of Health Care Quality and Research (AHRQ) to enhance the generalizability of our results.
METHODS Data Source We used the Healthcare Cost and Utilization Project–Nationwide Inpatient Sample (NIS) database from the year 2000– 2008 for the purposes of our study. NIS was created by the AHRQ as part of the Healthcare Cost and Utilization Project and contains data on hospital stays from about 1,000 hospitals sampled to approximate a 20% stratified sample of U.S. community hospitals (19). Each hospitalization is treated as an individual entry in the database and is coded with one principal diagnosis, up to 14 secondary diagnoses, and 15 procedural diagnoses associated with that stay. NIS includes information on all hospitalizations, regardless of the payer. To facilitate the production of national estimates, both hospital and discharge weights are provided. The details of NIS structure are available online (19). Since we used a publicly available dataset with no patient identifiers, this study was deemed exempt from review by the Institutional Review Board of the Medical College of Wisconsin, Milwaukee, WI. Study Population We used the International Classification of Diseases, 9th Edition, Clinical Modification (ICD-9-CM) codes to identify patients 412
www.ccmjournal.org
with HSCT who are 18 years old or older. We identified classified persons into the engraftment admission cohort if they had an ICD-9-CM code indicating the receipt of the HSCT procedure. Persons with an ICD-9-CM diagnosis code indicating HSCT but without a procedure code for HSCT were classified as subsequent admission. Further, using standard ICD-9-CM codes, we separated engraftment admissions into autologous and allogeneic and subsequent admissions into persons with GVHD and those without GVHD. The schema of our classification is shown in Figure 1. We excluded kidney, pancreas, heart, liver, and lung transplant from our comparator group and termed this group “non-HSCT.” Based on the coding system previously used and validated by Martin et al (20), we defined severe sepsis as either 1) ICD9-CM code 995.92 (systemic inflammatory response syndrome due to infectious process with organ failure) or 2) ICD-9-CM code for septicemia, bacteremia, or fungemia and at least one organ failure code. We identified comorbidities using ICD-9-CM codes as shown in Appendix 1 (Supplemental Digital Content 1, http:// links.lww.com/CCM/B128). We similarly identified the underlying etiology for undergoing HSCT—multiple myeloma (ICD-9-CM code 203), acute lymphoid leukemia (ICD-9-CM code 204), acute myeloid leukemia (AML) (ICD-9-CM code 205), non-Hodgkin lymphoma (NHL) (ICD-9-CM code 200, 202), and Hodgkin lymphoma (ICD-9-CM code 201). We used ICD-9-CM codes to identify patients who underwent mechanical ventilation (ICD-9-CM code 96.70– 96.72), tracheostomy (ICD-9-CM code 31.1, 31.21, and 31.29) or had new dialysis, blood transfusion (ICD-9-CM code 99.03 and 99.04), or total parenteral nutrition (ICD9-CM code 99.15). We considered a patient having received new dialysis if they had an ICD-9-CM code indicating acute renal failure (ICD-9-CM code 584.5–584.9) and a procedure code for hemodialysis (ICD-9-CM code 39.95). We used number of organs failing as a surrogate marker for severity of acute illness (20). We used NIS variables to identify patient age, gender, race, and primary payer. We divided age into four categories: 18– 34 years, 35–49 years, 50–64 years, and 65 years and above. Information of race is missing in 20–25% of records, and we categorized these as “unknown.” As per the NIS, we classified the primary payer into five groups: 1) private including Health Maintenance Organization, 2) Medicare, 3) Medicaid, 4) selfpay, and 5) others. We also used NIS classifications for a hospital’s teaching status and size. Outcomes Our primary outcomes of interest were the frequency of hospitalizations for severe sepsis and all-cause in-hospital mortality. Secondary outcomes included length of hospital stay and discharge dispositions. We used the NIS variable “DISPUNIFORM” to classify discharge disposition as deaths, discharges to home, discharges to skilled nursing facilities, discharges with home care, and others (includes discharges against medical advice and others). February 2015 • Volume 43 • Number 2
Clinical Investigations
Figure 1. Identification and grouping of hematopoietic stem cell transplant (HSCT) patients into four groups. GVHD = graft-versus-host disease, ICD-9-CM = International Classification of Diseases, 9th Edition, Clinical Modification.
Statistical Analysis We performed all statistical analysis in STATA IC 11.0 (STATA Corp, College Station, TX), using the strata weights and survey commands to generate national estimates of patients with HSCT admitted to the hospital with severe sepsis. We used Pearson chi-square test to compare categorical and Wilcoxon rank test to compare nonnormally distributed variables. We made pairwise comparisons between each of the four individual HSCT groups and the non-HSCT group. We used Bonferroni correction for multiple comparisons. Since we made four comparisons, p value was kept at 0.0125. We assessed predictors for in-hospital mortality in severe sepsis using multivariable logistic regression modeling. The model accounted for the survey design of the NIS. First, we used single predictor logistic regression to identify associations between putative risk factors for mortality. Variables that were significant in the univariate analysis at p value of less than 0.10 were candidates for inclusion in the final multivariate model. These variables were then checked for multicollinearity using tolerance and a variable Critical Care Medicine
inflation factor. We examined two-way interaction between the variables and HSCT; all interactions with p value of less than 0.10 were kept in the model. We then developed a propensity score approach to further adjust for differences by constructing a logistic regression model to calculate the likelihood that a person would have HSCT. This model included factors regardless of their individual statistical significance. Thus, we included study year, patient age, gender, race, hospital characteristics and severity of sepsis (number of organ failures), and individual comorbidities. We then divided the propensity score into quintiles and used it as a variable in the previously developed model. To exclude that the differences in mortality reflected a differential hospital effect, that is, more transplanted patients being treated in tertiary care and academic centers, we constructed a multilevel model that included a random effect for each individual hospital along with the fixed effects for all the variables in the previous model. The covariance structure used a distinct variance for each random effect within a random-effects equation and assumed that all covariances were zero. We thus constructed four models www.ccmjournal.org
413
Kumar et al
for each HSCT group (autologous and allogeneic in the engraftment admission and GVHD and no GVHD in the subsequent admission) and compared each to the cohort without HSCT. We used similar techniques and constructed multivariable models to examine risk factors associated with higher mortality within each HSCT cohort (engraftment admission and subsequent admission) with severe sepsis.
RESULTS Of the 291,179 discharges with HSCT from 2000 to 2008, severe sepsis developed in 21,898 patients (7.5%). The frequency of severe sepsis in HSCT recipients was five times higher when compared with the non-HSCT cohort (Table 1). Of all HSCT recipients developing severe sepsis, 6,168 developed severe sepsis during their engraftment admission. Of these, 2,750 received autologous and 3,418 received allogeneic HSCT. Severe sepsis developed in 15,730 patients with HSCT during their subsequent hospitalizations. Of these, 6,713 had GVHD. Rates of severe sepsis in HSCT recipients during engraftment admission through years 2000 to 2008 is shown in Figure 1. Demographical Characteristics HSCT recipients were younger than non-HSCT cohort. Male patients constituted a larger proportion of patients in each HSCT category. Teaching hospitals accounted for the majority of admissions for persons with HSCT compared with persons without HSCT. Significantly higher proportion of HSCT recipients was privately insured. Demographic details of these patients are shown in Table 1. Clinical Characteristics Engraftment Admission. All studied comorbid conditions except pulmonary embolism were lower in the HSCT cohort as compared with non-HSCT. Patients with allogeneic transplant had higher rates of severe sepsis as compared with autologous (13.2% vs 5.2%; p < 0.001) (Table 1). During the engraftment admission, the proportion of patients with three or more organ failures was lower in autologous transplants (18%) in comparison with allogeneic transplant (24.5%) and non-HSCT patients (20.3%). Allogeneic HSCT recipients had higher rates of respiratory failure and renal failure as compared with autologous and non-HSCT recipients. Rates of new dialysis were significantly higher in allogeneic transplants when compared with autologous and non-HSCT cohorts (19.3% vs 11.4% vs 5.7%, p < 0.001) (Table 2). Rates of cardiac failure were lowest in allogeneic recipients in comparison with autologous and non-HSCT cohorts. Subsequent Admission. Patients with GVHD developed severe sepsis more often than those who did not develop GVHD (10.4% vs 6.1%, p < 0.001) (Table 1). The proportion of patients with three or more organ failure was highest in HSCT recipients with GVHD followed by HSCT recipients without GVHD and non-HSCT patients (Table 2). Rates of respiratory, cardiac, and renal failure, though similar among persons with GVHD and non-HSCT cohort, remained higher than persons without GVHD. Occurrence of new dialysis was more frequent in GVHD cohort (Table 2). 414
www.ccmjournal.org
Outcomes Engraftment Admission. Although the unadjusted in-hospital mortality of allogeneic HSCT recipients with severe sepsis was significantly higher than non-HSCT patients (55.1% vs 32.9%, p < 0.001), the mortality rates for autologous HSCT recipients were no different than the non-HSCT cohort (Fig. 3A). On adjusted analysis, the odds of mortality were 3.81 times (95% CI, 2.39–6.07) higher in allogeneic HSCT recipients and 1.28 times (95% CI, 1.06–1.53) higher in autologous HSCT recipients when compared with non-HSCT patients (Fig. 4). The length of stay (LOS) in survivors was significantly longer in allogeneic transplant recipients (median, 39 d) as compared with autologous (median, 26 d) and non-HSCT patients (median, 11 d) (Table 3). Subsequent Admission. The in-hospital mortality was significantly higher in HSCT recipients with GVHD than nonHSCT patients (47.9% vs 32.9%, p < 0.001). However, in HSCT recipients without GVHD, mortality was no different (Fig. 3B). After adjustment, the odds of mortality were 2.14 times (95% CI, 1.88–2.45) higher in HSCT recipients with GVHD and 1.35 times (95% CI, 1.19–1.54) higher in HSCT recipients without GVHD when compared with non-HSCT patients (Fig. 4). The LOS in survivors was similar in GVHD, non-GVHD, and nonHSCT cohorts (Table 3). Discharge Destinations Discharges to home and home with healthcare were significantly higher in all HSCT cohorts (Table 3). Factors Associated With Mortality in HSCT Patients With Severe Sepsis Engraftment Admission. Allogeneic transplant recipients had 2.12 times higher odds of mortality than autologous recipients. Multiple myeloma as the reason for receiving HSCT was associated with lower odds of death with the occurrence of severe sepsis. Other factors associated with worse odds of mortality were advanced age and cirrhosis. Organ failures, particularly respiratory (odds ratio [OR], 12.1; 95% CI, 8.64–16.8) and hepatic (OR, 5.22; 95% CI, 2.29–11.8), were the strongest predictors of mortality. We did not find association of methicillin-resistant Staphylococcus aureus, pseudomonas, Clostridium difficile, aspergillus, or candida infections with increased mortality (Table 4). Subsequent Admission. During subsequent admissions, presence of GVHD was associated with 1.38 times higher odds of mortality when compared with those without GVHD. AML and NHL as reasons for HSCT were associated with higher odds of mortality. Other factors included advancing age, female gender, presence of stroke, nonvariceal upper gastrointestinal hemorrhage, and aspergillus infection. Teaching hospitals and larger hospitals had higher mortality. Similar to engraftment admission, respiratory (OR, 10.1; 95% CI, 8.47– 11.9) and hepatic failure (OR, 4.39; 95% CI, 2.54–7.59) had the strongest association with mortality. Sensitivity analyses were performed using 1) individual organ failures in place of total number of organ failure, 2) individual comorbidities in place of Charlson comorbidity index, and 3) February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 1. Demographic Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis From 2000 to 2008 HSCT Engraftment Admission Variable
Non-HSCT
Total admissions
286,285,073
Severe sepsis, n (%)
4,654,341 (1.6)
Autologous
53,337 2,750 (5.2)
Allogeneic
25,950 3,418 (13.2)
Subsequent Admission GVHD
64,878 6,713 (10.4)
No GVHD
146,909 9,017 (6.1)
Age category,a yr 18–34
3.9
10.8
26
22.4
13.5
35–49
10.6
20.4
33.4
32
24.9
50–64
22.9
47.2
36.3
39.3
45.8
≥ 65
62.7
21.6
4.3
6.3
15.7
49.7
43.4
36.6
39.4
42.6
White
52.6
57.9
55
56.3
60.1
African American
12.7
6.2
4.8
5.4
7.0
Hispanic
7.3
7.4
10.3
8.5
7.0
Asian
2.2
1.4
3.4
2.5
2.5
Others
2.5
3.4
2.9
2.1
2.4
22.8
23.7
23.5
25.2
21
Private
17.8
59.2
69.6
64.8
58.3
Medicare
66.5
24.9
11.7
18.7
28.5
Medicaid
10.3
10.8
13.6
11.7
8.3
Self
2.9
1.8
1.4
2.1
1.8
Others
2.5
3.2
3.8
2.7
3.1
46.2
94.5
97.1
89.3
77.1
Small
10.7
10.8
12.1
8.7
9.8
Medium
24.7
13.9
8.5
7.5
16.1
Large
64.7
75.3
79.3
83.8
74.2
Sex
a
Female Race
Unknown Primary payer
a
Hospital characteristics Teaching statusa Bed size
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125.
excluding the unknown category of race. The OR and their CIs varied by less than 5% for each sensitivity analysis performed.
DISCUSSION Using nationally representative data, we show that during the engraftment admission, severe sepsis occurs more frequently in HSCT recipients than non-HSCT patients. Furthermore, the associated odds of mortality, after adjustment for confounding Critical Care Medicine
factors, are higher for both allogeneic and autologous HSCT. Similarly, during subsequent admissions, HSCT recipients have higher rates of severe sepsis and higher odds of death in both GVHD and non-GVHD cohorts. Prior studies mostly performed at single centers have reported outcomes of HSCT patients admitted to the ICU for a variety of diagnoses (1, 4, 7, 9). The strengths of our study are its large size, national representation, and restricted attention to severe sepsis which is a www.ccmjournal.org
415
Kumar et al
Table 2. Clinical Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Non-HSCT
Subsequent Admission
Autologous
Allogeneic
GVHD
No GVHD
Comorbidities (%) Congestive heart failurea
28.4
10.2
10.5
8.3
8.9
Diabetes mellitusa
22.4
6.8
4.1
6.3
8.1
Cirrhosisa
7.0
2.4
7.9
7.5
4.2
End-stage kidney diseasea
5.9
3.7
0.9
2.9
4.3
Chronic pulmonary diseasea
34.3
11.8
8.2
9.9
13.1
3.8
2.8
1.9
2.5
1.6
10.1
8.1
1.5
1.6
4.9
7.4
4.7
9.8
14.7
9.6
6.9
8.8
1.7
1.8
1.0
2.2
0.8
0.03
1.6
0
0
0.9
0.23
7.8
3.2
2.2
2.8
0.15
10.4
1.8
3.8
5.8
0.8
37
4.5
7.4
25.2
Strokea Solid organ cancera Nonvariceal upper gastrointestinal bleed
a
Atrial fibrillation
20
a
Pulmonary embolism
10
4.2
Etiology (%) Amyloidosisa Non-Hodgkin lymphoma Hodgkin lymphoma Multiple myeloma
a
a
a
Acute lymphocytic leukemia Acute myeloid leukemia
a
a
0.8
1.0
16.7
14.6
7.5
0.9
6.9
45.4
40.1
23.6
Number of organ system failures (%) One
52.5
60
48
48.1
52.4
Two
27.3
21.9
27.5
27
24.6
Three
13.5
12.4
15.9
16.3
13.5
Four
5.2
4.7
6.7
6.7
6.2
Five
1.4
0.9
1.7
1.6
2.8
Six or seven
0.2
0
0
0.3
0.5
20.3
18.0a
24.5a
24.9a
23.0a
Respiratorya
52.5
41.7
57.1
50
42.1
Cardiaca
30.3
24.6
20.2
28.7
33.8
46.8
44
59.3
47
40.8
15.7
32.5
22.5
33.7
40.4
16.5
14.2
12.6
16
16.5
7.4
4.0
5.4
4.0
3.6
5.0
2.4
9.9
7.2
5.3
Three or more organ failures Type of organ failure (%)
Renal
a
Hematological
a
Metabolic Neurological Hepatic
a
a
(Continued)
416
www.ccmjournal.org
February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 2. (Continued) Clinical Characteristics of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Non-HSCT
Autologous
Subsequent Admission
Allogeneic
GVHD
No GVHD
Infection (%) 5.9
12
13.4
9.2
5.8
Gram positivea
27.8
33
38.3
36.1
32.9
Gram negativea
20.9
16
15.8
25.3
22
Clostridium difficilea
Aspergillusa
0.21
2.2
5.8
5.8
2.8
Candidaa
1.5
3.3
5.4
2.7
1.6
39.4
30.9
40.4
37.6
31.8
6.8
3.8
5.8
3.4
2.8
23.6
39.4
35.8
35
37.5
5.7
11.4
19.3
11
8.2
6.0
13.7
21.5
17.7
6.8
Procedures (%) Invasive mechanical ventilationa Tracheostomy Blood
a
a
New dialysis
a
Total parenteral nutrition
a
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125. All comparisons are with non–hematopoietic stem cell transplant group.
with autologous (319/588, 58.3%) recipients. We report 55% in-hospital mortality in allogeneic and 30% in autologous HSCT recipients. The numbers reported in literature may be higher due to the fact that acute lung injury is another common cause for admission of these patients to an ICU and is associated with even higher mortality up to 80–85% (3–5, 7, 9, 21). Other studies examining the outcomes of ICU stay during transplant admissions report ICU mortality ranging from 38% to 72% (3, 5, 7–9, 13). Limitations of these studies include being single-center studies, smaller sample size and reporting the overall ICU mortality, and not specifically describing mortality in severe sepsis. In these studies, sepsis and respiratory failure were reported to be commonest cause of admission. These studies do not report more recent outcomes for severe sepsis after initiation of early goal–directed therapies and other advances in taking care of patients with sepsis. Thus, the higher mortality may represent the pre–early goal–directed therapy era (16). A recent single-center study from Canada in autologous HSCT by Trinkaus et al (5) reported 55% ICU mortality in patients with sepsis, although the numbers are very small (only 11 patients with sepsis). Lungs and kidneys were the most common organs to fail in both HSCT and nontransplant patients (9, 22). However, this was significantly higher for allogeneic transplant patients. The Figure 2. Rate of hematopoietic stem cell transplant recipients developing severe sepsis during engraftment admission. rates of tracheostomy were lower in the HSCT frequent occurrence and perhaps the primary reason for ICU admission in HSCT recipients. Further, our analysis of four distinct HSCT cohorts adds to the previously reported literature, as risk factors for the development and outcomes from severe sepsis are probably very different in each cohort. Among HSCT recipients developing severe sepsis, we report two-fold higher odds for mortality for allogeneic HSCT recipients than autologous recipients. In a review, Afessa and Azoulay (9) examined literature reporting outcomes of HSCT patients admitted to the ICU and found that mortality was higher in allogeneic (604/867; 70%) when compared
Critical Care Medicine
www.ccmjournal.org
417
Kumar et al
Figure 3. Trends in mortality with severe sepsis from 2000 to 2008. Error bars show sem. A, Comparison of the mortality trends of severe sepsis during engraftment admission—autologous and allogeneic hematopoietic stem cell transplant with nontransplant group. B, Comparison of the mortality trends of severe sepsis during subsequent admissions—graft-versus-host disease (GVHD) status with nontransplant group.
(25) demonstrated that mortality of HSCT recipients receiving mechanical ventilation was very high and had not changed in 20 years. In parallel, the mortality of HSCT recipients receiving invasive mechanical ventilation remained unchanged between 70% and 75% during our study period from 2000 to 2008 (Appendix 3, Supplemental Digital Content 1, http://links.lww. com/CCM/B128). The differences in numbers may reflect a shift of ventilation strategies toward lung protective ventilation (26) and our analysis of the specific diagnosis of severe sepsis. Although we used a well-characterized national database, our study has important limitations. Our choice of codes to characterize organ failure, although identical to those validated by Martin et al (20), may not reliably identify sepsis-related organ failure. Specifically, we cannot exclude the possibility of variations in coding practices between hospitals as there is lack of standardization in the administrative literature. Further temporal association between the organ failures and occurrence of infection cannot be elicited from the database. Because the NIS is a discharge-level database, we were unable to identify that multiple hospitalizations may have been by the same patient. This also precluded us from using readmissions as an outcome measure. However, this limitation would apply to “subsequent admissions” only since engraftment admissions were selected using the procedure codes. During engraftment admissions, we also could not differentiate whether the severe sepsis occurred during the preengraftment, early postengraftment phase, or late posttransplant phase. Since differences in immunity and predilection to infections occur during each of these stages, inclusion of such information would have enhanced the robustness of our multivariable models. Secondary to the administrative nature of our data source were also unable to determine the conditioning regimen, associated neutropenia and its duration, the receipt of total body irradiation, and the status and duration of prophylactic antibiotics. The limited clinical detail of Figure 4. Forrest plot showing odds ratio of mortality in severe sepsis—hematopoietic our ICD-9 codes may not detect important severstem cell transplant admissions compared with nonhematopoietic stem cell transplant. ity differences. For example, different degrees of GVHD = graft-versus-host disease. patients; however, this may simply reflect significantly higher mortality in those developing respiratory failure in HSCT recipients. We report that renal replacement therapy was used in 11.4% for autologous and 19.3% for allogeneic HSCT recipients. These numbers are remarkably similar to other reports in the literature (2, 4, 7). Although observed rates of renal failure were similar or lower in autologous HSCT and non-GVHD cohorts as compared with the non-HSCT cohort, the utilization of new dialysis was 2–3 times higher. This may suggest either greater degrees of renal injury not described by ICD9-CM codes or provision of more aggressive care in HSCT recipients. Similar to other studies, we found that the number of failing organs were the strongest predictors of mortality in HSCT recipients in whom severe sepsis develops (10, 23). The odds of mortality increased with each additional organ failing (Appendix 2, Supplemental Digital Content 1, http://links. lww.com/CCM/B128). Individually, respiratory (OR, 10.3) and hepatic (OR, 5.1) failure had the strongest association with mortality. A few small studies have previously reported worse outcomes of HSCT recipients developing liver dysfunction (5, 7). Conversely, mechanical ventilation is a well-established risk factor for incremental mortality in HSCT recipients (2–6, 8, 21, 24). In a systemic review published in 2001, Bach et al
418
www.ccmjournal.org
February 2015 • Volume 43 • Number 2
Clinical Investigations
Table 3. Outcomes of Hematopoietic Stem Cell Transplant Patients Admitted With Severe Sepsis Over 2000–2008 HSCT Engraftment Admission Variable
Subsequent Admission
Non-HSCT
Autologous
Allogeneic
GVHD
32.9
30.1
55.1a
47.9a
35
Home
27.3
58.7a
58.4a
45.7a
50.3a
Home care
16
26.9a
33a
33.5a
29a
Facility
49.4
12.7a
6.6a
14.6a
13.2a
6.3
1.5a
2.0a
6.1
7.3
Mortality (%)
No GVHD
Disposition in survivors (%)
Transfers Time to death (d)b Length of stay in survivors (d)b
7 (2–16)
34 (21–54)a
44 (27–61)a
21 (10–37)a
11 (6–20)
26 (21–37)a
39 (30–58)a
12 (6–29)
10 (4–23) 9 (5–17)
HSCT = hematopoietic stem cell transplant, GVHD = graft-versus-host disease. a p < 0.0125. b Median (interquartile range). All comparisons are made with non–hematopoietic stem cell transplant group.
Table 4. Predictor Variables Associated With Mortality in Hematopoietic Stem Cell Transplant Patients With Severe Sepsis
Variable
Engraftment Admission
Subsequent Admission
OR
OR
Hematopoietic stem cell transplant status Autologous
Reference
Allogeneic
2.12 (1.55–2.90)
No GVHD
Reference
GVHD
1.38 (1.14–1.67)
Age category (yr) 18–34
Reference
Reference
35–49
1.68 (1.08–2.60)
1.38 (1.03–1.85)
50–64
1.31 (0.81–2.08)
1.68 (1.28–2.21)
≥ 65
2.08 (1.13–3.83)
2.74 (1.85–4.05)
Reference
Reference
NS
1.25 (1.02–1.53)
Stroke
2.03 (0.77–5.37)
2.26 (1.17–4.37)
Cirrhosis
4.49 (1.81–11.1)
NS
Nonvariceal upper gastrointestinal bleed
1.85 (0.98–3.49)
3.63 (2.43–5.43)
0.59 (0.40–0.89)
NS
Sex Male Female Individual comorbidities
Etiology Multiple myeloma
(Continued)
Critical Care Medicine
www.ccmjournal.org
419
Kumar et al
Table 4. (Continued) Predictor Variables Associated With Mortality in Hematopoietic Stem Cell Transplant Patients With Severe Sepsis Engraftment Admissiona
Subsequent Admissionb
Variable
OR
OR
Acute myeloid leukemia
NS
1.67 (1.37–2.02)
Acute lymphocytic leukemia
NS
NS
Non-Hodgkin lymphoma
NS
1.76 (1.07–2.91)
Hodgkin lymphoma
NS
NS
NS
1.78 (1.12–2.83)
Respiratory
12.1 (8.64–16.8)
10.1 (8.47–11.9)
Cardiac
2.42 (1.59–3.66)
1.64 (1.33–2.02)
Renal
2.64 (1.95–3.56)
1.83 (1.53–2.19)
NS
NS
1.63 (1.07–2.49)
1.54 (1.16–2.04)
NS
NS
5.22 (2.29–11.8)
4.39 (2.54–7.59)
Reference
Reference
NS
1.51 (1.15–1.99)
Reference
Reference
Medium
NS
1.67 (0.96–2.91)
Large
NS
1.93 (1.18–3.17)
2000
Reference
Reference
2001
0.96 (0.49–1.88)
0.76 (0.43–1.33)
2002
0.44 (0.23–0.83)
0.61 (0.36–1.03)
2003
0.91 (0.51–1.60)
0.59 (0.33–1.06)
2004
0.52 (0.25–1.06)
0.45 (0.28–0.71)
2005
0.40 (0.23–0.72)
0.52 (0.31–0.86)
2006
0.36 (0.18–0.72)
0.47 (0.29–0.75)
2007
0.37 (0.21–0.60)
0.33 (0.21–0.53)
2008
0.34 (0.19–0.60)
0.34 (0.21–0.54)
Infection Aspergillus Type of organ failure
Hematological Metabolic Neurological Hepatic Hospital type Nonteaching Teaching Small
Year
OR = odds ratio, GVHD = graft-versus-host disease, NS = not significant.
renal failure cannot be distinguished using these codes. Thus, it is possible that the improving survival we observed simply reflects decreasing severity that is not detected by the available codes. Finally, we were unable to identify those persons who may have refused escalation of care and preferred a palliative or hospice approach. This would falsely lower our estimates of mortality in each group. 420
www.ccmjournal.org
Despite its limitations, our study provides important information about severe sepsis in an understudied population. HSCT recipients are much more likely to be admitted with or develop severe sepsis once hospitalized than their non-HSCT counterparts. Once sepsis occurs, they are also more likely to suffer worse mortality. Clinicians should be vigilant about the occurrence of severe sepsis in this population and pay particular February 2015 • Volume 43 • Number 2
Clinical Investigations
attention to therapies that are known to improve outcomes such as early administration of antibiotics. Furthermore, we also report novel risk factors of mortality within the HSCT cohorts developing severe sepsis, and these should be incorporated into clinical decision making and prognostication.
REFERENCES
1. Horak DA, Forman SJ: Critical care of the hematopoietic stem cell patient. Crit Care Clin 2001; 17:671–695 2. Jackson SR, Tweeddale MG, Barnett MJ, et al: Admission of bone marrow transplant recipients to the intensive care unit: Outcome, survival and prognostic factors. Bone Marrow Transplant 1998; 21:697–704 3. Kew AK, Couban S, Patrick W, et al: Outcome of hematopoietic stem cell transplant recipients admitted to the intensive care unit. Biol Blood Marrow Transplant 2006; 12:301–305 4. Scales DC, Thiruchelvam D, Kiss A, et al: Intensive care outcomes in bone marrow transplant recipients: A population-based cohort analysis. Crit Care 2008; 12:R77 5. Trinkaus MA, Lapinsky SE, Crump M, et al: Predictors of mortality in patients undergoing autologous hematopoietic cell transplantation admitted to the intensive care unit. Bone Marrow Transplant 2009; 43:411–415 6. Paz HL, Garland A, Weinar M, et al: Effect of clinical outcomes data on intensive care unit utilization by bone marrow transplant patients. Crit Care Med 1998; 26:66–70 7. Pène F, Aubron C, Azoulay E, et al: Outcome of critically ill allogeneic hematopoietic stem-cell transplantation recipients: A reappraisal of indications for organ failure supports. J Clin Oncol 2006; 24:643–649 8. Soubani AO, Kseibi E, Bander JJ, et al: Outcome and prognostic factors of hematopoietic stem cell transplantation recipients admitted to a medical ICU. Chest 2004; 126:1604–1611 9. Afessa B, Azoulay E: Critical care of the hematopoietic stem cell transplant recipient. Crit Care Clin 2010; 26:133–150 10. Afessa B, Tefferi A, Dunn WF, et al: Intensive care unit support and Acute Physiology and Chronic Health Evaluation III performance in hematopoietic stem cell transplant recipients. Crit Care Med 2003; 31:1715–1721 11. Cargo CA, Yates E, Marley C, et al: Allogeneic bone marrow transplant in Belfast—An outcome overview of the first 25 years. Ulster Med J 2008; 77:185–190 12. Gil L, Styczynski J, Komarnicki M: Infectious complication in 314 patients after high-dose therapy and autologous hematopoietic stem
Critical Care Medicine
cell transplantation: Risk factors analysis and outcome. Infection 2007; 35:421–427 13. Almyroudis NG, Fuller A, Jakubowski A, et al: Pre- and post-engraftment bloodstream infection rates and associated mortality in allogeneic hematopoietic stem cell transplant recipients. Transpl Infect Dis 2005; 7:11–17 14. Mitsui H, Karasuno T, Santo T, et al: Analysis of sepsis in allogeneic bone marrow transplant recipients: A single-center study. J Infect Chemother 2003; 9:238–242 15. Lalitha MK, Pai R, Manoharan A, et al: Systemic bacterial infections in bone marrow transplant patients. Indian J Cancer 2000; 37:10–14 16. Rivers E, Nguyen B, Havstad S, et al; Early Goal-Directed Therapy Collaborative Group: Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001; 345:1368–1377 17. Sprung CL, Annane D, Keh D, et al; CORTICUS Study Group: Hydrocortisone therapy for patients with septic shock. N Engl J Med 2008; 358:111–124 18. Bernard GR, Vincent JL, Laterre PF, et al; Recombinant human protein C Worldwide Evaluation in Severe Sepsis (PROWESS) study group: Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344:699–709 19. Introduction to the HCUP Nationwide Inpatient Sample (NIS) 2008. Available at: http://www.hcup-us.ahrq.gov/db/nation/nis/NIS_ Introduction_2008.jsp. Accessed January 21, 2014 20. Martin GS, Mannino DM, Eaton S, et al: The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med 2003; 348:1546–1554 21. Naeem N, Eyzaguirre A, Kern JA, et al: Outcome of adult umbilical cord blood transplant patients admitted to a medical intensive care unit. Bone Marrow Transplant 2006; 38:733–738 22. Létourneau I, Dorval M, Bélanger R, et al: Acute renal failure in bone marrow transplant patients admitted to the intensive care unit. Nephron 2002; 90:408–412 23. Haire WD, Ruby EI, Gordon BG, et al: Multiple organ dysfunction syndrome in bone marrow transplantation. JAMA 1995; 274:1289–1295 24. Huaringa AJ, Leyva FJ, Giralt SA, et al: Outcome of bone marrow transplantation patients requiring mechanical ventilation. Crit Care Med 2000; 28:1014–1017 25. Bach PB, Schrag D, Nierman DM, et al: Identification of poor prognostic features among patients requiring mechanical ventilation after hematopoietic stem cell transplantation. Blood 2001; 98:3234–3240 26. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308
www.ccmjournal.org
421
