CLINICAL INVESTIGATION
Bacteremia in Hemodialysis Patients With Hepatitis C Puja Chebrolu, BS, Rhonda E. Colombo, MD, Stephanie Baer, MD, T. Ryan Gallaher, MD, Sara Atwater, MD, Mufaddal Kheda, MD, N. Stanley Nahman, Jr, MD and Kristina W. Kintziger, PhD
Abstract: Background: Hepatitis C virus (HCV) infection and bacteremia are common comorbidities in hemodialysis patients. A specific relationship between HCV infection and bacteremia has not been defined; however, there is evidence of immune compromise in both HCV-infected and uremic patients, suggesting that this group may be at higher risk for infection. Methods: We investigated risk factors and mortality associated with bacteremia in HCV-infected hemodialysis patients from the United States Renal Data System. Results: During the 4-year study period, HCV was present in 2.1% of 355,084 patients initiating hemodialysis. When compared with the total population, the rate of bacteremia was significantly higher in patients with HCV (38.3% versus 21.8%). The adjusted relative risk (RR) for bacteremia was higher in HCV versus all patients (relative risk, 95% confidence interval [CI]) in the presence of methicillin-resistant Staphylococcus aureus infection (2.64, CI: 2.58–2.70 versus 2.32, CI: 2.27–2.38), HIV (1.93, CI: 1.85–2.02 versus 1.86, CI: 1.77–1.95) urinary tract infection (1.79, CI: 1.77, 1.82 versus 1.64, CI: 1.61–1.67) and cirrhosis (1.49, CI: 1.45– 1.54 versus 1.29, CI: 1.25–1.34). The hazard ratio (95% CI) for death was higher in HCV versus all patients at 1.69 (CI: 1.58–1.81) versus 1.54 (CI: 1.53–1.56). Conclusions: These data indicate that several clinical covariates increase the risk of bacteremia in hemodialysis patients, with the magnitude of that risk being further increased by HCV infection. Improving outcomes in HCV-infected hemodialysis patients will likely be dependent on aggressive diagnosis and treatment of both HCV and bacteremia.
From the Department of Medicine (PC, RC, SB, TRG, SA, MK, NSN, KWK), Georgia Regents University, Augusta, Georgia; Charlie Norwood VAMC (SB, NSN), Augusta, Georgia; and Southwest Georgia Nephrology Clinic, PC (MK), Albany, Georgia. Submitted April 23, 2014; accepted in revised form October 21, 2014. The authors have no conflicts of interest to disclose. This work was supported by an Alpha Omega Alpha student research fellowship (P.C.), research fellowship from the Immunotherapy Center, Georgia Regents University (R.C.), a grant from Dialysis Clinic, Inc (M.K., N.S.N. and K.W.K.), the Augusta Biomedical Research Corporation (N.S.N.) and the Translational Research Program of the Department of Medicine, Georgia Regents University. Parts of this work have been previously presented and are cited below: Chebrolu P, Baer S, Colombo R, et al. Mortality from known risk factors for bacteremia in HCV-positive hemodialysis patients. Presented at the International Conference on Multidisciplinary Healthcare, All India Institute of Medical Sciences, 2014 (E-poster)—Chebrolu P, Colombo R, Baer S, et al. Effect of bacteremia and associated risk factors on survival in hemodialysis patients. Presented at the American Society of Nephrology, 2013 (poster)—Colombo R, Chebrolu P, Baer S, et al. Risk for mortality in hemodialysis patients with blood stream infection: a comparison of bacteremia and candidemia. Presented at the S AFMR (poster). J Invest Med 2013;61:442—Chebrolu P, Colombo R, Baer S, et al. Risk factors for bacteremia in hepatitis C virus-infected hemodialysis patients. Presented at the S AFMR (poster). J Invest Med 2013;61:441—Chebrolu P, Colombo RE, Baer SL, et al. Access-independent risk factors for bacteremia in hemodialysis patients. Infectious Disease Society of America ID Week, 2012 (poster). 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 Web site (www.amjmedsci.com). The data for this study were supplied by the USRDS but the opinions expressed are those of the authors and do not represent those of the NIDDK or the USRDS. Correspondence: N. Stanley Nahman, MD, Section of Nephrology, Department of Medicine, Georgia Regents University, 1120 15th Street, BA9413, Augusta, GA 30912 (E-mail: [email protected]).
The American Journal of the Medical Sciences
Key Indexing Terms: Hemodialysis; Hepatitis C; Bacteremia. [Am J Med Sci 2015;349(3):217–221.]
C
hronic hepatitis C virus (HCV) infection affects at least 170 to 200 million people worldwide.1,2 HCV is also a common comorbidity in hemodialysis patients, with a prevalence of 5% to 25%.3,4 HCV infection may lead to cirrhosis and hepatocellular carcinoma,1,2 worsens the complications associated with HIV infection,1 and in dialysis patients, increases the relative risk (RR) of infection-related mortality.5 Bloodstream infection is also an important cause of morbidity and mortality in dialysis patients,6,7 accounting for up to 20% of causes of death.6,8 A direct relationship between HCV infection and bacteremia has not been defined; however, the reported increase in RR in infection-related mortality in patients with HCV5 would suggest a potential association. To more closely address this question, we used the United States Renal Data System (USRDS) to investigate the risk factors and mortality associated with bacteremia in HCV-infected hemodialysis patients.
METHODS Data Source The USRDS is a deidentified database that includes demographic characteristics, dialysis claims, transplant history, payer source, hospitalization, physician/supplier claims (including ICD-9 and CPT diagnosis codes) and vital statistics on all patients with ESRD in the United States.9 Demographic information was available in the USRDS patient information data set, and comorbidities were defined by inpatient ICD-9 billing codes submitted to Medicare. The latest version of the form CMS-2728, implemented in 2005, was used to obtain information about vascular access type on first dialysis. Selection of Cohorts A retrospective analysis of the USRDS cohort was conducted. All patients with ESRD, older than 18 years, whose incident date of hemodialysis was between 2005 and 2008, were included. Patients were excluded if the CMS-2728 form had incomplete information about vascular access type on first dialysis or if their record contained only the older version of the form. A subcohort including people with an ICD-9 diagnosis of HCV (070.41, 070.44, 070.51, 070.54, 070.7 or V02.62) was also examined. Covariates Clinical covariates were selected based on presumed risk factors for infection and included 139 diagnostic or procedural codes (see Figure, Supplemental Digital Content 1, http://links.lww.com/MAJ/A64). The administrative nature of the data set precluded the use of specific laboratory results and thus relied on these codes for the diagnosis. The covariates included (1) potentially immunosuppressed states, that is, diabetes mellitus, leukopenia, autoimmune diseases or history of
Volume 349, Number 3, March 2015
217
Chebrolu et al
or (3) other infectious comorbidities, that is, decubitus ulcer, Clostridium difficile colitis, diverticulitis, osteomyelitis or urinary tract infection (UTI). Total parenteral nutrition (TPN) and vascular access type were also selected because of the necessity for direct access to the bloodstream. Cirrhosis was included as an indicator of advanced liver disease and peripheral vascular disease as a marker for atherosclerosis. Each covariate was considered in the analysis if the diagnosis code was present at or before the date of the first diagnosis of bacteremia.
TABLE 1. Characteristics of the study populations (%) Variable All patients HCV positive Median age in years (IQ range) Male sex Black race White Dialysis catheter AV fistula AV graft Bacteremia Most common comorbidities for both groups Diabetes mellitus PVD UTI Decubitus ulcer C. difficile colitis Higher prevalence in HCVpositive patients HIV Hepatitis B
63.5 198,600 104,945 231,307 293,173 47,732 14,179 77,288
(21.0) (56) (29.6) (65.1) (82.6) (13.4) (3.9) (21.8)
55 4,980 3,920 3,259 6,377 726 307 2,834
(14.0) (67.2) (52.9) (44.0) (86.1) (9.8) (4.1) (38.3)
137,767 41,610 28,883 15,393 9,367
(38.8) (11.7) (8.1) (4.3) (2.6)
4,277 923 905 408 349
(57.7) (12.5) (12.2) (5.5) (4.7)
2,902 (0.82) 1,726 (0.49)
Outcome Variable The primary outcome was bacteremia, which included an ICD-9 diagnosis of bacteremia (790.7) or septicemia and its associated subcodes (038.0, 038.1, 038.2, 038.3, 038.4, 038.8 and 038.9). Only the first episode of bacteremia was included for patients who had more than 1 episode. The outcome used in survival analyses was time to death from the initiation of dialysis. Statistical Analyses Descriptive analyses were performed for each cohort, with determination of median and interquartile range for continuous variables and frequencies for categorical variables. Bivariate analyses using log-binomial regression models were performed for each covariate to determine its crude risk of bacteremia. Multivariate analyses were conducted using backward elimination to determine adjusted risks. Covariates that were not significant on bivariate analysis were excluded from multivariable analyses. Because of sample size and the large number of covariates, analysis for the total population was performed using 3 separate models: (1) demographics (gender, race, ethnicity, age, year of diagnosis and access type at first dialysis), (2) potential immunosuppressed conditions and (3) infectious comorbidities and conditions. Survival analyses were
688 (9.3) 591 (8.0)
AV, arteriovenous; HCV, hepatitis C virus; HIV, human immunodeficiency virus; PVD, peripheral vascular disease; UTI, urinary tract infection.
a kidney transplant, (2) coinfections with HIV, hepatitis B (HBV, as determined by specific ICD-9 diagnosis codes), Candida sp. (including invasive infection [candidemia, Candida endocarditis or Candida endopthalmitis], colonization and candiduria), methicillin-resistant Staphylococcus aureus (MRSA, including MRSA colonization or other infections, excluding bacteremia)
TABLE 2. Comorbidities associated with bacteremia All patients Adjusted association
Bacteremia Absent N MRSA infection 1,881 HIV 1,573 UTI 15,409 Decubitus ulcer 7,472 Candidemia 257 C. difficile colitis 4,695 TPN 1,321 Cirrhosis 4,287 Leukocytopenia 680 Candida 3,471 colonization Pancytopenia 680 Hepatitis B 1,067
%
95% CIa
Present N
HCV-infected patients
%
Absent
RR Lower Upper
0.68 4,410 5.71 2.32 0.57 1,329 1.72 1.86 5.55 13,474 17.43 1.64 2.69 7,921 10.25 1.48 0.09 323 0.42 1.52 1.69 4,672 6.04 1.46 0.48 1,347 1.74 1.51 1.54 2,725 3.53 1.29 0.24 466 0.60 1.38 1.25 2,911 3.77 1.27
2.27 1.77 1.61 1.45 1.39 1.42 1.44 1.25 1.27 1.23
2.38 1.95 1.67 1.51 1.66 1.50 1.58 1.34 1.50 1.31
0.24 0.38
1.26 1.05
1.45 1.21
589 659
0.76 1.35 0.85 1.13
Adjusted association
Bacteremia
N
%
Present N
%
95% CI RR Lower Upper
74 1.62 138 4.87 2.64 382 8.35 306 10.80 1.93 483 10.56 422 4.89 1.79 187 4.09 221 7.80 1.67 14 0.31 21 0.74 1.61 190 4.15 159 5.61 1.53 32 0.70 64 2.26 1.51 1,088 23.78 760 26.82 1.49 24 0.52 33 1.16 1.36 148 3.23 119 4.20 1.34 40 341
0.87 40 7.45 250
1.41 1.31 8.82 1.27
Change in RRb
2.58 1.85 1.77 1.63 1.48 1.49 1.44 1.45 1.25 1.30
2.70 2.02 1.82 1.70 1.76 1.57 1.58 1.54 1.47 1.39
0.31 0.07 0.16 0.18 0.09 0.07 0.00c 0.20 20.03c 0.07
1.22 1.18
1.40 1.36
20.04c 0.14
a
Confidence interval. The difference in RR between HCV-infected and all patients. The change in RR was either not different or was lower in HCV infection. HCV, hepatitis C virus; HIV, human immunodeficiency virus; MRSA, methicillin-resistant Staphylococcus aureus; RR, relative risk; TPN, total parenteral nutrition; UTI, urinary tract infection. b c
218
Volume 349, Number 3, March 2015
Bacteremia and HCV in Dialysis Patients
TABLE 3. Vascular access type and the risk of bacteremia All patients Bacteremia Absent N AV fistula AV graft Dialysis catheter b
41,892 11,438 224,466
Adjusted association
Present %
15.08 4.12 80.80
HCV-infected patients
N
95% CI
%
5,840 2,741 68,707
7.56 3.55 88.90
RR
Lower c
1.00 1.34 1.74
NA 1.29 1.70
a
Bacteremia Absent
Adjusted association
Present
95% CI
Upper
N
%
N
%
RR
Lower
Upper
NA 1.39 1.79
552 208 3,816
12.06 4.55 83.39
174 99 2,561
6.14 3.49 90.37
1.00 1.39 1.78
NA 1.34 1.73
NA 1.45 1.82
a
Confidence interval. Reference group. c Not applicable. AV, arteriovenous; HCV, hepatitis C virus; RR, relative risk. b
conducted using the Cox proportional hazards regression model to determine time to death and time to bacteremia for each cohort. All analyses were conducted at an alpha level of 0.05 using SAS version 9.3 software (SAS Inst., Cary, NC). This work was approved by the Institutional Review Board of Georgia Regents University and was performed in adherence to the principles of the Declaration of Helsinki.
RESULTS Characteristics of the Study Populations The population under study consisted of all incident dialysis patients from 2005 to 2008 and was based on the most recent data sets available at the time the queries were initiated. During the 4-year study period, 355,084 hemodialysis patients had a completed CMS-2728 form and at least 1 hospitalization for which ICD-9 diagnosis codes were recorded. From the total population, the HCV subset was determined, and diagnosis of HCV was present in 7410 (2.1%).
The characteristics of the 2 groups are shown in Table 1. HCV infection was more common in African American men. When compared with the entire population, the rates of bacteremia and catheter use were higher in patients with HCV (38.3% versus 21.8% and 86.1% versus 82.6% for HCV versus all patients, respectively), as was coinfection with HIV and hepatitis B (9.3% versus 0.82% and 7.9% versus 0.49% for HCV versus all patients, respectively). Multivariable Analysis for Significant Risk Factors for Bacteremia For all patients, the adjusted RR of bacteremia from 12 comorbidities ranged from 1.13 (hepatitis B) to 2.32 (MRSA infection) and is presented in Table 2. Other comorbidities at high risk for bacteremia included HIV (RR 5 1.86) and UTI (RR 5 1.64). For HCV-positive patients, there was a similar increased risk of bacteremia for the above comorbidities. In this regard, the adjusted RR of bacteremia ranged from 1.27 (hepatitis B) to
TABLE 4. Adjusted HR of death in patients with bacteremia and other covariates All patients HCV-infected patients 95% CIa
Bacteremia Age .65 yr Catheter Cirrhosis Decubitus ulcer TPN Candidemia Pancytopenia AV graft C difficile Candida colonization HCV Hepatitis B PVD
95% CI
HR
Lower
Upper
HR
Lower
Upper
Change in HR
1.54 2.26 1.85 1.77 1.68 1.45 1.44 1.42 1.35 1.19 1.18 1.13 1.11 1.08
1.53 2.23 1.82 1.72 1.64 1.39 1.31 1.33 1.31 1.16 1.14 1.09 1.04 1.06
1.56 2.28 1.89 1.82 1.71 1.52 1.58 1.52 1.40 1.22 1.22 1.18 1.19 1.09
1.69 1.90 1.32 1.85 1.61 1.48 1.70 NSb 1.00 1.25 1.21 NAc 1.18 1.16
1.58 1.77 1.16 1.72 1.43 1.16 1.15 — 0.81 1.09 1.03 — 1.05 1.06
1.81 2.04 1.50 1.99 1.82 1.87 2.50 — 1.24 1.44 1.42 — 1.33 1.28
0.15 20.36 20.53 0.08 20.07 0.03 0.26 — 0.07 0.08 0.07 — 0.07 0.08
a
Confidence interval. Not significant for HCV. c Not applicable. AV, arteriovenous; HCV, hepatitis C virus; HR, hazard ratio; PVD, peripheral vascular disease; TPN, total parenteral nutrition. b
Copyright © 2015 by the Southern Society for Clinical Investigation.
219
Chebrolu et al
2.64 (MRSA infection), and for HIV and UTI, the RR was 1.93 and 1.79, respectively. For 9 comorbidities, the increased adjusted RR of bacteremia was greater in HCV-infected patients than the entire population. This difference was greatest for MRSA infection (2.64 versus 2.32, 31% increase in RR) and cirrhosis (1.49 versus 1.29, 20% increase in RR). Regardless of whether dialysis was initiated with a graft or catheter, the adjusted risk of bacteremia relative to a fistula was elevated in both groups, with catheters exhibiting a higher risk than grafts (Table 3).
In HCV-infected patients, the HR for death for patients with bacteremia was 1.69 and exhibited a stronger association than in the total population. The greatest risk for death also occurred in patients older than 65 years. High risk of death was also demonstrated in patients with a diagnosis of cirrhosis, candidemia, decubitus ulcer, TPN or a dialysis catheter. When compared with all patients, the greatest increase in HR for death for patients with HCV occurred in association with candidemia. The median survival for patients with bacteremia in this group was 802 days. Figure 1B indicates the survival curve for patients with HCV with and without bacteremia.
Multivariable Analysis for Survival and Cox Regression The hazard ratio (HR) for death in all patients with bacteremia, when compared with those without bacteremia, was 1.54 (Table 4). The greatest risk for death occurred in patients older than 65 years. High risk of death was also seen in association with dialysis catheters, cirrhosis, decubitus ulcer, TPN and candidemia. The median survival for patients with bacteremia in this group was 716 days. Figure 1A indicates the survival curve for all patients with and without bacteremia.
The results of this study demonstrate that HCV-infected hemodialysis patients are at greater risk for the development of bacteremia and death than the general dialysis population. Thirty-eight percent of HCV-infected patients developed bacteremia compared with 22% of the total population, and the RR of bacteremia was 1.69, compared with 1.54. Five of the 6 greatest risk factors for bacteremia were infectious comorbidities; in each case, the risk of bacteremia was higher in the HCV population. The increase in risk of bacteremia in patients with
DISCUSSION
FIGURE 1. (A) Life-table survival curves demonstrating the effect of bacteremia on mortality in all patients. Dashed line 5 bacteremia. (B) Life-table survival curves demonstrating the effect of bacteremia on mortality in HCV-infected patients. Dashed line 5 bacteremia.
220
Volume 349, Number 3, March 2015
Bacteremia and HCV in Dialysis Patients
HCV was also reflected by decreased survival time. Taken together, these studies suggest that an increased susceptibility to infection in HCV-infected hemodialysis patients may contribute to an increase in mortality. When compared with the total population, HCV-infected patients tended to be younger, African American, male and initiate dialysis with a catheter. This group also had a much higher incidence of HIV and hepatitis B infection. The explanation for this discrepancy is not clear from this study; however similar trends have been reported from the general population.10,11 Using multivariate analysis, the infectious comorbidities associated with a greater risk of bacteremia in patients with HCV included MRSA infection, HIV coinfection, UTI and candidemia. These suggest a generalized increase in susceptibility to bacteremia in the presence of bacterial, viral or fungal coinfection, although whether this is due to a direct versus indirect effect cannot be determined from these data. Using survival analysis, death in patients with bacteremia occurred earlier in the HCV-infected patients. The greatest risk for mortality in the patients with HCV was associated with age greater than 65 years, cirrhosis and bacteremia. The greatest difference in risk from the total population was associated with candidemia. Although candidal infection may be a significant clinical risk factor for death, we cannot exclude the possibility that this may represent a surrogate marker for critical illness. HCV infection has been associated with increased infectious complications in other clinical settings. For instance, patients from the Italian Cohort of Antiretroviral Naive Patients study who were coinfected with HCV and HIV were at 2.6-fold increased risk for developing AIDS-defining illnesses12,13 and had 3.2- and 3.9-fold higher risks for bacterial and fungal infections, respectively.12 It was postulated that in HIV coinfection, HCV may have a direct effect on the immune system by impeding CD4 cell proliferation with accelerated apoptosis, and/or increasing CD8 cell activation, leading to defective responses to opportunistic pathogens.14 Thus, HCV may possess negative immunomodulatory properties that are additive to the immunosuppression exhibited by HIV infection. HCV-infected patients are at increased risk of cirrhosis, which may also negatively impact the immune responses. Chronic liver disease decreases in vivo clearance of bacteria,15 and cirrhosis increases the risk of sepsis.16 The findings from this study are consistent with this observation; a diagnosis of cirrhosis carried the greatest hazard ratio for death in HCVinfected patients (HR: 1.85). Of note, a diagnosis of cirrhosis was based on ICD-9 diagnosis coding and may underestimate its true incidence. Thus, we cannot exclude a role for subclinical cirrhosis as a contributing factor to the observed increases in the rate of bacteremia and risk for death in the HCV group. The clinical interpretations of these queries are limited by the necessity of making certain assumptions. We have presumed that the described clinical events are the correct interpretation of the submitted ICD-9 diagnosis codes or the information contained on form 2728, and further that the information from these sources is accurate.17 Specific clinical data are not readily available from the data set, thus, precise information such as laboratory values are not accessible. On this basis, only the broadest conclusions can be drawn from queries of this nature. However, this limitation is balanced by the large number of patients under consideration for each question, thus allowing for meaningful clinical inferences to be made.17
Copyright © 2015 by the Southern Society for Clinical Investigation.
CONCLUSIONS In summary, HCV infection in hemodialysis patients is associated with an increase in both bacteremia and risk of death. This may in part be the result of acquired immune compromise from both HCV and liver disease. Improving outcomes in HCV-infected hemodialysis patients will likely be dependent on aggressive diagnosis and treatment of HCV and infectious complications in these patients. REFERENCES 1. Lauer GM, Walker BD. Hepatitis C virus infection. N Engl J Med 2001;345:41–52. 2. Wendt A, Adhoute X, Castellani P, et al. Chronic hepatitis C: future treatment. Clin Pharmacol 2014;6:1–17. 3. Kalantar-Zadeh K, Kilpatrick RD, McAllister CJ, et al. Hepatitis C virus and death risk in hemodialysis patients. J Am Soc Nephrol 2007; 18:1584–93. 4. Goodkin DA, Bragg-Gresham JL, Koenig KG, et al. Association of comorbid conditions and mortality in hemodialysis patients in Europe, Japan, and the United States: the Dialysis Outcomes and Practice Patterns Study (DOPPS). J Am Soc Nephrol 2003;14:3270–7. 5. Fabrizi F, Takkouche B, Lunghi G, et al. The impact of hepatitis C virus infection on survival in dialysis patients: meta-analysis of observational studies. J Viral Hepat 2007;14:697–703. 6. Kato S, Chmielewski M, Honda H, et al. Aspects of immune dysfunction in end-stage renal disease. Clin J Am Soc Nephrol 2008;3:1526–33. 7. Patel PR, Yi SH, Booth S, et al. Bloodstream infection rates in outpatient hemodialysis facilities participating in a collaborative prevention effort: a quality improvement report. Am J Kidney Dis 2013;62:322–30. 8. Rojas L, Munoz P, Kestler M, et al. Bloodstream infections in patients with kidney disease: risk factors for poor outcome and mortality. J Hosp Infect 2013;85:196–205. 9. U.S. Renal Data System, USRDS. 2013 Annual data report: Atlas of chronic kidney disease and end-stage renal disease in the United States. Bethesda (MD): National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2013. 10. Johnson AS, Beer L, Sionean C, et al. HIV infection—United States, 2008 and 2010. MMWR Surveill Summ 2013;62(suppl 3):112–9. 11. Nguyen GC, Thuluvath PJ. Racial disparity in liver disease: biological, cultural, or socioeconomic factors. Hepatology 2008;47:1058–66. 12. d’Arminio Monforte A, Cozzi-Lepri A, Castagna A, et al. Risk of developing specific AIDS-defining illnesses in patients coinfected with HIV and hepatitis C virus with or without liver cirrhosis. Clin Infect Dis 2009;49:612–22. 13. Andreoni M, Giacometti A, Maida I, et al. HIV-HCV co-infection: epidemiology, pathogenesis and therapeutic implications. Eur Rev Med Pharmacol Sci 2012;16:1473–83. 14. Piroth L. Coinfection with hepatitis C virus and HIV: more than double trouble. Clin Infect Dis 2009;49:623–5. 15. Ashare A, Stanford C, Hancock P, et al. Chronic liver disease impairs bacterial clearance in a human model of induced bacteremia. Clin Transl Sci 2009;2:199–205. 16. Foreman MG, Mannino DM, Moss M. Cirrhosis as a risk factor for sepsis and death: analysis of the National Hospital Discharge Survey. Chest 2003;124:1016–20. 17. Eggers PW. CMS 2728: what good is it? Clin J Am Soc Nephrol 2010; 5:1908–9.
221
Bacteremia in Hemodialysis Patients With Hepatitis C Puja Chebrolu, BS, Rhonda E. Colombo, MD, Stephanie Baer, MD, T. Ryan Gallaher, MD, Sara Atwater, MD, Mufaddal Kheda, MD, N. Stanley Nahman, Jr, MD and Kristina W. Kintziger, PhD
Abstract: Background: Hepatitis C virus (HCV) infection and bacteremia are common comorbidities in hemodialysis patients. A specific relationship between HCV infection and bacteremia has not been defined; however, there is evidence of immune compromise in both HCV-infected and uremic patients, suggesting that this group may be at higher risk for infection. Methods: We investigated risk factors and mortality associated with bacteremia in HCV-infected hemodialysis patients from the United States Renal Data System. Results: During the 4-year study period, HCV was present in 2.1% of 355,084 patients initiating hemodialysis. When compared with the total population, the rate of bacteremia was significantly higher in patients with HCV (38.3% versus 21.8%). The adjusted relative risk (RR) for bacteremia was higher in HCV versus all patients (relative risk, 95% confidence interval [CI]) in the presence of methicillin-resistant Staphylococcus aureus infection (2.64, CI: 2.58–2.70 versus 2.32, CI: 2.27–2.38), HIV (1.93, CI: 1.85–2.02 versus 1.86, CI: 1.77–1.95) urinary tract infection (1.79, CI: 1.77, 1.82 versus 1.64, CI: 1.61–1.67) and cirrhosis (1.49, CI: 1.45– 1.54 versus 1.29, CI: 1.25–1.34). The hazard ratio (95% CI) for death was higher in HCV versus all patients at 1.69 (CI: 1.58–1.81) versus 1.54 (CI: 1.53–1.56). Conclusions: These data indicate that several clinical covariates increase the risk of bacteremia in hemodialysis patients, with the magnitude of that risk being further increased by HCV infection. Improving outcomes in HCV-infected hemodialysis patients will likely be dependent on aggressive diagnosis and treatment of both HCV and bacteremia.
From the Department of Medicine (PC, RC, SB, TRG, SA, MK, NSN, KWK), Georgia Regents University, Augusta, Georgia; Charlie Norwood VAMC (SB, NSN), Augusta, Georgia; and Southwest Georgia Nephrology Clinic, PC (MK), Albany, Georgia. Submitted April 23, 2014; accepted in revised form October 21, 2014. The authors have no conflicts of interest to disclose. This work was supported by an Alpha Omega Alpha student research fellowship (P.C.), research fellowship from the Immunotherapy Center, Georgia Regents University (R.C.), a grant from Dialysis Clinic, Inc (M.K., N.S.N. and K.W.K.), the Augusta Biomedical Research Corporation (N.S.N.) and the Translational Research Program of the Department of Medicine, Georgia Regents University. Parts of this work have been previously presented and are cited below: Chebrolu P, Baer S, Colombo R, et al. Mortality from known risk factors for bacteremia in HCV-positive hemodialysis patients. Presented at the International Conference on Multidisciplinary Healthcare, All India Institute of Medical Sciences, 2014 (E-poster)—Chebrolu P, Colombo R, Baer S, et al. Effect of bacteremia and associated risk factors on survival in hemodialysis patients. Presented at the American Society of Nephrology, 2013 (poster)—Colombo R, Chebrolu P, Baer S, et al. Risk for mortality in hemodialysis patients with blood stream infection: a comparison of bacteremia and candidemia. Presented at the S AFMR (poster). J Invest Med 2013;61:442—Chebrolu P, Colombo R, Baer S, et al. Risk factors for bacteremia in hepatitis C virus-infected hemodialysis patients. Presented at the S AFMR (poster). J Invest Med 2013;61:441—Chebrolu P, Colombo RE, Baer SL, et al. Access-independent risk factors for bacteremia in hemodialysis patients. Infectious Disease Society of America ID Week, 2012 (poster). 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 Web site (www.amjmedsci.com). The data for this study were supplied by the USRDS but the opinions expressed are those of the authors and do not represent those of the NIDDK or the USRDS. Correspondence: N. Stanley Nahman, MD, Section of Nephrology, Department of Medicine, Georgia Regents University, 1120 15th Street, BA9413, Augusta, GA 30912 (E-mail: [email protected]).
The American Journal of the Medical Sciences
Key Indexing Terms: Hemodialysis; Hepatitis C; Bacteremia. [Am J Med Sci 2015;349(3):217–221.]
C
hronic hepatitis C virus (HCV) infection affects at least 170 to 200 million people worldwide.1,2 HCV is also a common comorbidity in hemodialysis patients, with a prevalence of 5% to 25%.3,4 HCV infection may lead to cirrhosis and hepatocellular carcinoma,1,2 worsens the complications associated with HIV infection,1 and in dialysis patients, increases the relative risk (RR) of infection-related mortality.5 Bloodstream infection is also an important cause of morbidity and mortality in dialysis patients,6,7 accounting for up to 20% of causes of death.6,8 A direct relationship between HCV infection and bacteremia has not been defined; however, the reported increase in RR in infection-related mortality in patients with HCV5 would suggest a potential association. To more closely address this question, we used the United States Renal Data System (USRDS) to investigate the risk factors and mortality associated with bacteremia in HCV-infected hemodialysis patients.
METHODS Data Source The USRDS is a deidentified database that includes demographic characteristics, dialysis claims, transplant history, payer source, hospitalization, physician/supplier claims (including ICD-9 and CPT diagnosis codes) and vital statistics on all patients with ESRD in the United States.9 Demographic information was available in the USRDS patient information data set, and comorbidities were defined by inpatient ICD-9 billing codes submitted to Medicare. The latest version of the form CMS-2728, implemented in 2005, was used to obtain information about vascular access type on first dialysis. Selection of Cohorts A retrospective analysis of the USRDS cohort was conducted. All patients with ESRD, older than 18 years, whose incident date of hemodialysis was between 2005 and 2008, were included. Patients were excluded if the CMS-2728 form had incomplete information about vascular access type on first dialysis or if their record contained only the older version of the form. A subcohort including people with an ICD-9 diagnosis of HCV (070.41, 070.44, 070.51, 070.54, 070.7 or V02.62) was also examined. Covariates Clinical covariates were selected based on presumed risk factors for infection and included 139 diagnostic or procedural codes (see Figure, Supplemental Digital Content 1, http://links.lww.com/MAJ/A64). The administrative nature of the data set precluded the use of specific laboratory results and thus relied on these codes for the diagnosis. The covariates included (1) potentially immunosuppressed states, that is, diabetes mellitus, leukopenia, autoimmune diseases or history of
Volume 349, Number 3, March 2015
217
Chebrolu et al
or (3) other infectious comorbidities, that is, decubitus ulcer, Clostridium difficile colitis, diverticulitis, osteomyelitis or urinary tract infection (UTI). Total parenteral nutrition (TPN) and vascular access type were also selected because of the necessity for direct access to the bloodstream. Cirrhosis was included as an indicator of advanced liver disease and peripheral vascular disease as a marker for atherosclerosis. Each covariate was considered in the analysis if the diagnosis code was present at or before the date of the first diagnosis of bacteremia.
TABLE 1. Characteristics of the study populations (%) Variable All patients HCV positive Median age in years (IQ range) Male sex Black race White Dialysis catheter AV fistula AV graft Bacteremia Most common comorbidities for both groups Diabetes mellitus PVD UTI Decubitus ulcer C. difficile colitis Higher prevalence in HCVpositive patients HIV Hepatitis B
63.5 198,600 104,945 231,307 293,173 47,732 14,179 77,288
(21.0) (56) (29.6) (65.1) (82.6) (13.4) (3.9) (21.8)
55 4,980 3,920 3,259 6,377 726 307 2,834
(14.0) (67.2) (52.9) (44.0) (86.1) (9.8) (4.1) (38.3)
137,767 41,610 28,883 15,393 9,367
(38.8) (11.7) (8.1) (4.3) (2.6)
4,277 923 905 408 349
(57.7) (12.5) (12.2) (5.5) (4.7)
2,902 (0.82) 1,726 (0.49)
Outcome Variable The primary outcome was bacteremia, which included an ICD-9 diagnosis of bacteremia (790.7) or septicemia and its associated subcodes (038.0, 038.1, 038.2, 038.3, 038.4, 038.8 and 038.9). Only the first episode of bacteremia was included for patients who had more than 1 episode. The outcome used in survival analyses was time to death from the initiation of dialysis. Statistical Analyses Descriptive analyses were performed for each cohort, with determination of median and interquartile range for continuous variables and frequencies for categorical variables. Bivariate analyses using log-binomial regression models were performed for each covariate to determine its crude risk of bacteremia. Multivariate analyses were conducted using backward elimination to determine adjusted risks. Covariates that were not significant on bivariate analysis were excluded from multivariable analyses. Because of sample size and the large number of covariates, analysis for the total population was performed using 3 separate models: (1) demographics (gender, race, ethnicity, age, year of diagnosis and access type at first dialysis), (2) potential immunosuppressed conditions and (3) infectious comorbidities and conditions. Survival analyses were
688 (9.3) 591 (8.0)
AV, arteriovenous; HCV, hepatitis C virus; HIV, human immunodeficiency virus; PVD, peripheral vascular disease; UTI, urinary tract infection.
a kidney transplant, (2) coinfections with HIV, hepatitis B (HBV, as determined by specific ICD-9 diagnosis codes), Candida sp. (including invasive infection [candidemia, Candida endocarditis or Candida endopthalmitis], colonization and candiduria), methicillin-resistant Staphylococcus aureus (MRSA, including MRSA colonization or other infections, excluding bacteremia)
TABLE 2. Comorbidities associated with bacteremia All patients Adjusted association
Bacteremia Absent N MRSA infection 1,881 HIV 1,573 UTI 15,409 Decubitus ulcer 7,472 Candidemia 257 C. difficile colitis 4,695 TPN 1,321 Cirrhosis 4,287 Leukocytopenia 680 Candida 3,471 colonization Pancytopenia 680 Hepatitis B 1,067
%
95% CIa
Present N
HCV-infected patients
%
Absent
RR Lower Upper
0.68 4,410 5.71 2.32 0.57 1,329 1.72 1.86 5.55 13,474 17.43 1.64 2.69 7,921 10.25 1.48 0.09 323 0.42 1.52 1.69 4,672 6.04 1.46 0.48 1,347 1.74 1.51 1.54 2,725 3.53 1.29 0.24 466 0.60 1.38 1.25 2,911 3.77 1.27
2.27 1.77 1.61 1.45 1.39 1.42 1.44 1.25 1.27 1.23
2.38 1.95 1.67 1.51 1.66 1.50 1.58 1.34 1.50 1.31
0.24 0.38
1.26 1.05
1.45 1.21
589 659
0.76 1.35 0.85 1.13
Adjusted association
Bacteremia
N
%
Present N
%
95% CI RR Lower Upper
74 1.62 138 4.87 2.64 382 8.35 306 10.80 1.93 483 10.56 422 4.89 1.79 187 4.09 221 7.80 1.67 14 0.31 21 0.74 1.61 190 4.15 159 5.61 1.53 32 0.70 64 2.26 1.51 1,088 23.78 760 26.82 1.49 24 0.52 33 1.16 1.36 148 3.23 119 4.20 1.34 40 341
0.87 40 7.45 250
1.41 1.31 8.82 1.27
Change in RRb
2.58 1.85 1.77 1.63 1.48 1.49 1.44 1.45 1.25 1.30
2.70 2.02 1.82 1.70 1.76 1.57 1.58 1.54 1.47 1.39
0.31 0.07 0.16 0.18 0.09 0.07 0.00c 0.20 20.03c 0.07
1.22 1.18
1.40 1.36
20.04c 0.14
a
Confidence interval. The difference in RR between HCV-infected and all patients. The change in RR was either not different or was lower in HCV infection. HCV, hepatitis C virus; HIV, human immunodeficiency virus; MRSA, methicillin-resistant Staphylococcus aureus; RR, relative risk; TPN, total parenteral nutrition; UTI, urinary tract infection. b c
218
Volume 349, Number 3, March 2015
Bacteremia and HCV in Dialysis Patients
TABLE 3. Vascular access type and the risk of bacteremia All patients Bacteremia Absent N AV fistula AV graft Dialysis catheter b
41,892 11,438 224,466
Adjusted association
Present %
15.08 4.12 80.80
HCV-infected patients
N
95% CI
%
5,840 2,741 68,707
7.56 3.55 88.90
RR
Lower c
1.00 1.34 1.74
NA 1.29 1.70
a
Bacteremia Absent
Adjusted association
Present
95% CI
Upper
N
%
N
%
RR
Lower
Upper
NA 1.39 1.79
552 208 3,816
12.06 4.55 83.39
174 99 2,561
6.14 3.49 90.37
1.00 1.39 1.78
NA 1.34 1.73
NA 1.45 1.82
a
Confidence interval. Reference group. c Not applicable. AV, arteriovenous; HCV, hepatitis C virus; RR, relative risk. b
conducted using the Cox proportional hazards regression model to determine time to death and time to bacteremia for each cohort. All analyses were conducted at an alpha level of 0.05 using SAS version 9.3 software (SAS Inst., Cary, NC). This work was approved by the Institutional Review Board of Georgia Regents University and was performed in adherence to the principles of the Declaration of Helsinki.
RESULTS Characteristics of the Study Populations The population under study consisted of all incident dialysis patients from 2005 to 2008 and was based on the most recent data sets available at the time the queries were initiated. During the 4-year study period, 355,084 hemodialysis patients had a completed CMS-2728 form and at least 1 hospitalization for which ICD-9 diagnosis codes were recorded. From the total population, the HCV subset was determined, and diagnosis of HCV was present in 7410 (2.1%).
The characteristics of the 2 groups are shown in Table 1. HCV infection was more common in African American men. When compared with the entire population, the rates of bacteremia and catheter use were higher in patients with HCV (38.3% versus 21.8% and 86.1% versus 82.6% for HCV versus all patients, respectively), as was coinfection with HIV and hepatitis B (9.3% versus 0.82% and 7.9% versus 0.49% for HCV versus all patients, respectively). Multivariable Analysis for Significant Risk Factors for Bacteremia For all patients, the adjusted RR of bacteremia from 12 comorbidities ranged from 1.13 (hepatitis B) to 2.32 (MRSA infection) and is presented in Table 2. Other comorbidities at high risk for bacteremia included HIV (RR 5 1.86) and UTI (RR 5 1.64). For HCV-positive patients, there was a similar increased risk of bacteremia for the above comorbidities. In this regard, the adjusted RR of bacteremia ranged from 1.27 (hepatitis B) to
TABLE 4. Adjusted HR of death in patients with bacteremia and other covariates All patients HCV-infected patients 95% CIa
Bacteremia Age .65 yr Catheter Cirrhosis Decubitus ulcer TPN Candidemia Pancytopenia AV graft C difficile Candida colonization HCV Hepatitis B PVD
95% CI
HR
Lower
Upper
HR
Lower
Upper
Change in HR
1.54 2.26 1.85 1.77 1.68 1.45 1.44 1.42 1.35 1.19 1.18 1.13 1.11 1.08
1.53 2.23 1.82 1.72 1.64 1.39 1.31 1.33 1.31 1.16 1.14 1.09 1.04 1.06
1.56 2.28 1.89 1.82 1.71 1.52 1.58 1.52 1.40 1.22 1.22 1.18 1.19 1.09
1.69 1.90 1.32 1.85 1.61 1.48 1.70 NSb 1.00 1.25 1.21 NAc 1.18 1.16
1.58 1.77 1.16 1.72 1.43 1.16 1.15 — 0.81 1.09 1.03 — 1.05 1.06
1.81 2.04 1.50 1.99 1.82 1.87 2.50 — 1.24 1.44 1.42 — 1.33 1.28
0.15 20.36 20.53 0.08 20.07 0.03 0.26 — 0.07 0.08 0.07 — 0.07 0.08
a
Confidence interval. Not significant for HCV. c Not applicable. AV, arteriovenous; HCV, hepatitis C virus; HR, hazard ratio; PVD, peripheral vascular disease; TPN, total parenteral nutrition. b
Copyright © 2015 by the Southern Society for Clinical Investigation.
219
Chebrolu et al
2.64 (MRSA infection), and for HIV and UTI, the RR was 1.93 and 1.79, respectively. For 9 comorbidities, the increased adjusted RR of bacteremia was greater in HCV-infected patients than the entire population. This difference was greatest for MRSA infection (2.64 versus 2.32, 31% increase in RR) and cirrhosis (1.49 versus 1.29, 20% increase in RR). Regardless of whether dialysis was initiated with a graft or catheter, the adjusted risk of bacteremia relative to a fistula was elevated in both groups, with catheters exhibiting a higher risk than grafts (Table 3).
In HCV-infected patients, the HR for death for patients with bacteremia was 1.69 and exhibited a stronger association than in the total population. The greatest risk for death also occurred in patients older than 65 years. High risk of death was also demonstrated in patients with a diagnosis of cirrhosis, candidemia, decubitus ulcer, TPN or a dialysis catheter. When compared with all patients, the greatest increase in HR for death for patients with HCV occurred in association with candidemia. The median survival for patients with bacteremia in this group was 802 days. Figure 1B indicates the survival curve for patients with HCV with and without bacteremia.
Multivariable Analysis for Survival and Cox Regression The hazard ratio (HR) for death in all patients with bacteremia, when compared with those without bacteremia, was 1.54 (Table 4). The greatest risk for death occurred in patients older than 65 years. High risk of death was also seen in association with dialysis catheters, cirrhosis, decubitus ulcer, TPN and candidemia. The median survival for patients with bacteremia in this group was 716 days. Figure 1A indicates the survival curve for all patients with and without bacteremia.
The results of this study demonstrate that HCV-infected hemodialysis patients are at greater risk for the development of bacteremia and death than the general dialysis population. Thirty-eight percent of HCV-infected patients developed bacteremia compared with 22% of the total population, and the RR of bacteremia was 1.69, compared with 1.54. Five of the 6 greatest risk factors for bacteremia were infectious comorbidities; in each case, the risk of bacteremia was higher in the HCV population. The increase in risk of bacteremia in patients with
DISCUSSION
FIGURE 1. (A) Life-table survival curves demonstrating the effect of bacteremia on mortality in all patients. Dashed line 5 bacteremia. (B) Life-table survival curves demonstrating the effect of bacteremia on mortality in HCV-infected patients. Dashed line 5 bacteremia.
220
Volume 349, Number 3, March 2015
Bacteremia and HCV in Dialysis Patients
HCV was also reflected by decreased survival time. Taken together, these studies suggest that an increased susceptibility to infection in HCV-infected hemodialysis patients may contribute to an increase in mortality. When compared with the total population, HCV-infected patients tended to be younger, African American, male and initiate dialysis with a catheter. This group also had a much higher incidence of HIV and hepatitis B infection. The explanation for this discrepancy is not clear from this study; however similar trends have been reported from the general population.10,11 Using multivariate analysis, the infectious comorbidities associated with a greater risk of bacteremia in patients with HCV included MRSA infection, HIV coinfection, UTI and candidemia. These suggest a generalized increase in susceptibility to bacteremia in the presence of bacterial, viral or fungal coinfection, although whether this is due to a direct versus indirect effect cannot be determined from these data. Using survival analysis, death in patients with bacteremia occurred earlier in the HCV-infected patients. The greatest risk for mortality in the patients with HCV was associated with age greater than 65 years, cirrhosis and bacteremia. The greatest difference in risk from the total population was associated with candidemia. Although candidal infection may be a significant clinical risk factor for death, we cannot exclude the possibility that this may represent a surrogate marker for critical illness. HCV infection has been associated with increased infectious complications in other clinical settings. For instance, patients from the Italian Cohort of Antiretroviral Naive Patients study who were coinfected with HCV and HIV were at 2.6-fold increased risk for developing AIDS-defining illnesses12,13 and had 3.2- and 3.9-fold higher risks for bacterial and fungal infections, respectively.12 It was postulated that in HIV coinfection, HCV may have a direct effect on the immune system by impeding CD4 cell proliferation with accelerated apoptosis, and/or increasing CD8 cell activation, leading to defective responses to opportunistic pathogens.14 Thus, HCV may possess negative immunomodulatory properties that are additive to the immunosuppression exhibited by HIV infection. HCV-infected patients are at increased risk of cirrhosis, which may also negatively impact the immune responses. Chronic liver disease decreases in vivo clearance of bacteria,15 and cirrhosis increases the risk of sepsis.16 The findings from this study are consistent with this observation; a diagnosis of cirrhosis carried the greatest hazard ratio for death in HCVinfected patients (HR: 1.85). Of note, a diagnosis of cirrhosis was based on ICD-9 diagnosis coding and may underestimate its true incidence. Thus, we cannot exclude a role for subclinical cirrhosis as a contributing factor to the observed increases in the rate of bacteremia and risk for death in the HCV group. The clinical interpretations of these queries are limited by the necessity of making certain assumptions. We have presumed that the described clinical events are the correct interpretation of the submitted ICD-9 diagnosis codes or the information contained on form 2728, and further that the information from these sources is accurate.17 Specific clinical data are not readily available from the data set, thus, precise information such as laboratory values are not accessible. On this basis, only the broadest conclusions can be drawn from queries of this nature. However, this limitation is balanced by the large number of patients under consideration for each question, thus allowing for meaningful clinical inferences to be made.17
Copyright © 2015 by the Southern Society for Clinical Investigation.
CONCLUSIONS In summary, HCV infection in hemodialysis patients is associated with an increase in both bacteremia and risk of death. This may in part be the result of acquired immune compromise from both HCV and liver disease. Improving outcomes in HCV-infected hemodialysis patients will likely be dependent on aggressive diagnosis and treatment of HCV and infectious complications in these patients. REFERENCES 1. Lauer GM, Walker BD. Hepatitis C virus infection. N Engl J Med 2001;345:41–52. 2. Wendt A, Adhoute X, Castellani P, et al. Chronic hepatitis C: future treatment. Clin Pharmacol 2014;6:1–17. 3. Kalantar-Zadeh K, Kilpatrick RD, McAllister CJ, et al. Hepatitis C virus and death risk in hemodialysis patients. J Am Soc Nephrol 2007; 18:1584–93. 4. Goodkin DA, Bragg-Gresham JL, Koenig KG, et al. Association of comorbid conditions and mortality in hemodialysis patients in Europe, Japan, and the United States: the Dialysis Outcomes and Practice Patterns Study (DOPPS). J Am Soc Nephrol 2003;14:3270–7. 5. Fabrizi F, Takkouche B, Lunghi G, et al. The impact of hepatitis C virus infection on survival in dialysis patients: meta-analysis of observational studies. J Viral Hepat 2007;14:697–703. 6. Kato S, Chmielewski M, Honda H, et al. Aspects of immune dysfunction in end-stage renal disease. Clin J Am Soc Nephrol 2008;3:1526–33. 7. Patel PR, Yi SH, Booth S, et al. Bloodstream infection rates in outpatient hemodialysis facilities participating in a collaborative prevention effort: a quality improvement report. Am J Kidney Dis 2013;62:322–30. 8. Rojas L, Munoz P, Kestler M, et al. Bloodstream infections in patients with kidney disease: risk factors for poor outcome and mortality. J Hosp Infect 2013;85:196–205. 9. U.S. Renal Data System, USRDS. 2013 Annual data report: Atlas of chronic kidney disease and end-stage renal disease in the United States. Bethesda (MD): National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2013. 10. Johnson AS, Beer L, Sionean C, et al. HIV infection—United States, 2008 and 2010. MMWR Surveill Summ 2013;62(suppl 3):112–9. 11. Nguyen GC, Thuluvath PJ. Racial disparity in liver disease: biological, cultural, or socioeconomic factors. Hepatology 2008;47:1058–66. 12. d’Arminio Monforte A, Cozzi-Lepri A, Castagna A, et al. Risk of developing specific AIDS-defining illnesses in patients coinfected with HIV and hepatitis C virus with or without liver cirrhosis. Clin Infect Dis 2009;49:612–22. 13. Andreoni M, Giacometti A, Maida I, et al. HIV-HCV co-infection: epidemiology, pathogenesis and therapeutic implications. Eur Rev Med Pharmacol Sci 2012;16:1473–83. 14. Piroth L. Coinfection with hepatitis C virus and HIV: more than double trouble. Clin Infect Dis 2009;49:623–5. 15. Ashare A, Stanford C, Hancock P, et al. Chronic liver disease impairs bacterial clearance in a human model of induced bacteremia. Clin Transl Sci 2009;2:199–205. 16. Foreman MG, Mannino DM, Moss M. Cirrhosis as a risk factor for sepsis and death: analysis of the National Hospital Discharge Survey. Chest 2003;124:1016–20. 17. Eggers PW. CMS 2728: what good is it? Clin J Am Soc Nephrol 2010; 5:1908–9.
221
