Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Kidney Int. Author manuscript; available in PMC 2010 June 1.
Published in final edited form as:
Published online 2009 December 23. doi:  10.1038/ki.2009.502
PMCID: PMC2871010

Acute kidney injury in non-severe pneumonia is associated with an increased immune response and lower survival

Raghavan Murugan,1 Vijay Karajala-Subramanyam,1 Minjae Lee,1,2 Sachin Yende,1 Lan Kong,1,2 Melinda Carter,1 Derek C. Angus,1 and John A. Kellum1, on behalf of the Genetic and Inflammatory Markers of Sepsis (GenIMS) Investigators


While sepsis is a leading cause of acute kidney injury in critically ill patients, the relationship between immune response and acute kidney injury in less severely ill patients with infection is not known. Here we studied the epidemiology, 1-year mortality, and immune response associated with acute kidney injury in 1836 hospitalized patients with community-acquired severe and non-severe pneumonia. Acute kidney injury developed in 631 patients of whom 329 had severe and 302 had non-severe sepsis. Depending on the subgroup classification, 16–25% of the patients with non-severe pneumonia also developed acute kidney injury. In general, patients with acute kidney injury were older, had more comorbidity, and had higher biomarker concentrations (interleukin-6, tumor necrosis factor, D-dimer) even among patients without severe sepsis. The risk of death associated with acute kidney injury varied when assessed by Gray's survival model and after adjusting for differences in age, gender, ethnicity, and comorbidity. This risk was significantly higher immediately after hospitalization but gradually fell over time in the overall cohort and in those with non-severe pneumonia. A significantly higher risk of death (hazard ratio 1.29) was also present in those never admitted to an intensive care unit. Hence acute kidney injury is common even among patients with non-severe pneumonia and is associated with higher immune response and an increased risk of death.

Keywords: acute kidney injury, inflammation, mortality, pneumonia, severe sepsis, severity

Sepsis, an immune response to infection, is a leading cause of acute kidney injury (AKI) in severely ill patients,13 and development of AKI is associated with increased risk of death.2,4 Although several studies reported outcome of sepsis-induced AKI in critically ill patients, 1,2,48 less attention has been paid to infected patients with AKI who are less severely ill. For instance, community-acquired pneumonia (CAP) is a common, infectious cause of hospitalization in developed countries.9 Most patients with CAP are not severely ill and are often treated in non-intensive care settings. However, the epidemiology and outcome of AKI in patients with non-severe CAP is unknown. Furthermore, patients with CAP who develop milder forms of AKI will not meet existing criteria for severe sepsis (i.e., sepsis with coexisting severe acute organ dysfunction),10 and are therefore unlikely to be enrolled in therapeutic trials of sepsis. 1113 If AKI is associated with increased risk of death in patients with non-severe CAP and in those without severe sepsis, then early recognition and treatment could improve outcome from AKI.

Inflammation is thought to have a causal role in sepsis-induced AKI in animals, 1416 but data in humans are inconclusive. Several authors found increased concentrations of inflammatory, coagulation, and fibrinolysis markers in critically ill patients with AKI.1719 However, an important limitation of these studies is that it is unclear whether immune markers were associated with AKI itself, or coexisting severe organ dysfunction (severe sepsis) that may have contributed to increased immune response.20,21 Understanding immune response associated with AKI in patients without severe sepsis is essential for development of immune modulating therapies for AKI.

Using Risk, Injury, and Failure (RIFLE) criteria to classify AKI in a large multicenter cohort of patients hospitalized with CAP, the goals of this prospective observational study were threefold. First, to describe differences in baseline characteristics and outcomes between hospitalized CAP patients with and without AKI. Second, to examine the risk of AKI in patients with milder forms of CAP and whether development of AKI in an otherwise non-severe pneumonia is associated with adverse outcomes. We used the Pneumonia Severity Index classes I–III,22 confusion, uremia, respiratory rate, low blood pressure, age 65 years or older (CURB-65) group 1,23 those who never developed severe sepsis,24 and those who were never admitted to intensive care unit, as proxies for non-severe CAP. Third, to examine immune, coagulation, and fibrinolysis pathways among patients with and without AKI, in those who do and do not develop severe sepsis.


Baseline characteristics of study participants

Of the 2320 patients enrolled in the study, 291 were excluded because they were discharged from the emergency department, and another 134 patients were excluded because the clinical team ruled out CAP during the first 3 days of hospitalization. Of the remaining 1895 patients with CAP, we excluded patients who were either receiving dialysis or had history of end-stage renal disease (n=48), and in whom serum creatinine was never obtained (n=11). Of the 1836 patients who formed the study cohort (Figure 1), premorbid baseline creatinine was estimated in 1745 patients using the recommended Modification of Diet in Renal Disease equation.

Figure 1
Subject disposition for the Genetic and Inflammatory Markers of Sepsis study cohort

Of the 631 patients (34%) who met criteria for AKI, 307 (49%), 135 (21%), and 189 (30%) reached maximum RIFLE stages, respectively, during hospitalization. Of the 91 patients with known premorbid baseline creatinine, we found moderate agreement between premorbid creatinine and the estimated creatinine to classify patients with AKI (Cohen's κ coefficient=0.70, 95% CI 0.52–0.87, P<0.0001). Nearly two-thirds of patients developing AKI had already done so at hospital admission (n=399, 63.2% of patients with AKI). Table 1 shows the demographic characteristics and severity of illness on day 1 of hospital admission stratified by AKI, and by severity of AKI. Patients with AKI were predominantly older, of white race, had more preexisting comorbidity, and had more severe CAP.

Table 1
Baseline characteristics of patients with AKI following CAP

Hospital course and outcome for patients with and without AKI

Table 2 shows hospital course and outcomes for patients with and without AKI. One-third (31.1%, n=572) of CAP patients developed severe sepsis (i.e., had a Sequential Organ Failure Assessment (SOFA) score10 of [gt-or-equal, slanted]3 in at least one organ system during the entire hospitalization), of which more than one-half (57.5%) developed AKI (n=329). Although overall intensive care use was only 16%, more than a third of patients with AKI (39%) were admitted to the intensive care unit at some point during hospitalization, of which 18% received mechanical ventilation. Patients with AKI, on an average, incurred higher median hospital length of stay compared to patients without AKI (8 vs 5 days, P<0.001).

Table 2
Hospital course and outcomes in patients with and without AKI in the overall cohort

Patients with AKI had a higher risk of death at each time point, such as by hospital discharge (11 vs 1.3%, P<0.001), by 90 days (24 vs. 9.8%, P<0.001), and by 1 year (36.3 vs 20.1%, P<0.001; Table 2). Mortality increased with increased severity of AKI and was 29.6, 40.7, and 43.9% for maximum stages RIFLE, respectively, at 1 year (P<0.001; Figure 2a).

Figure 2
One-year mortality in patients with and without AKI in the overall cohort and within non-severe CAP subgroups

Using Gray's survival model, we found that the hazard ratios for association between AKI and risk of death varied significantly over 365 days after hospitalization for CAP. The unadjusted hazard ratios were highest during the first 100 days after presentation, subsequently decreased, but nevertheless persisted over the entire 1 year of follow-up (hazard ratio range 1.45–2.79, P<0.001; Table 4). When adjusted for differences in age, gender, race, and the Charlson Comorbidity Index the hazard ratios were attenuated (adjusted hazard ratio range 1.10–2.10, P<0.001; Table 4), but nevertheless remained significant up to 100 days after CAP (Figure 3a).

Figure 3
Varying risk of death associated with AKI over one year after CAP hospitalization
Table 4
Hazard ratios for association between AKI and 1-year mortality

Similar risk of death was present when absolute (adjusted hazard ratio 1.12, 95% CI 1.04–1.20, P=0.001), relative (adjusted hazard ratio 1.12, 95% CI 1.04–1.19, P=0.002), as well as ratio (adjusted hazard ratio 1.10, 95% CI 1.04–1.17, P=0.001) of change in glomerular filtration rate were used in the model instead of AKI. We also found that the mortality rates were not different between patients who presented with AKI at hospital admission and those who developed AKI during hospitalization. Similarly, there was no difference in mortality associated with AKI among those with and without chronic kidney disease (data not shown).

Risk of AKI, hospital course, and outcome in non-severe CAP

Not surprisingly, the risk of AKI was lower in patients with non-severe CAP. Of the 1030 patients within Pneumonia Severity Index classes I–III, 20.3% of patients (n=209) developed AKI. Of the 806 patients within CURB-65 group 1, 16.4% of patients (n=132) developed AKI. Of patients who never developed severe sepsis (n=1264) and those never admitted to the intensive care unit (n=1544), the risk of AKI was 23.8% (n=302) and 25% (n=386). Patients with AKI within Pneumonia Severity Index classes I–III and CURB-65 group 1 were more likely to develop severe sepsis, be admitted to intensive care unit, and be mechanically ventilated (P<0.001; Table 3).

Table 3
Hospital course and outcomes in patients with and without AKI in non-severe CAP subgroups

Non-severe CAP patients with AKI incurred higher length of hospital stay, in-hospital and 90-day mortality (Table 3). At 1 year, mortality varied from 17.2 to 34.2% for patients with AKI within the four non-severe CAP subgroups (Figure 2b). Patients admitted to medical wards with AKI had higher mortality compared to those without AKI (34.2 vs 19.7%, P<0.001). This increased risk of death was not only seen with more severe AKI. Of patients with stage Risk (61%), the mortality was higher with AKI (30.1 vs 19.6%, P<0.001) (hazard ratio 1.67, 95% CI 1.28–2.18, P<0.001).

The unadjusted hazard ratios for 1-year mortality ranged from 1.03 to 3.65 within the four non-severe CAP subgroups (Table 4). The hazard ratios remained unchanged when adjusted for age, gender, race, and the Charlson Comorbidity Index for those admitted to medical wards with AKI (adjusted hazard ratio 1.29, 95% CI 1.03–1.60, P<0.023). Using Gray's model, we found that the adjusted hazard ratio for death in patients within CURB-65 group 1 varied over the 1-year follow-up (adjusted hazard ratio range 0.85–2.97, P=0.001). The risk of death associated with AKI was high immediately after CAP in patients admitted to hospital (hazard ratio 2.73, 95% CI 1.52–4.92). Subsequently, although the hazard ratios declined, the risk of death associated with AKI nevertheless remained significant up to 50 days (Figure 3b).

Biomarker concentration in patients with and without AKI

Figure 4 shows the plasma biomarker concentrations during the first week of hospitalization among patients with and without AKI stratified by severe sepsis. Patients with AKI who developed severe sepsis had higher concentrations of interleukin-6, 10, and tumor necrosis factor at presentation to the emergency department and during the first week of CAP (Figure 4a). The geometric mean concentration on day 1 for AKI vs no AKI was—interleukin-6: 111.8 vs 42 pg/ml, P<0.001; interleukin-10: 8.9 vs 6.0 pg/ml, P=0.026; and tumor necrosis factor: 8.4 vs 4.5 pg/ml, P<0.001. Importantly, of patients without severe sepsis, similar patterns in interleukin-6 and tumor necrosis factor concentrations were noted in patients with and without AKI during the first week of CAP (P<0.001), although the magnitude of difference was only modest. The geometric mean concentration on day 1 in those without severe sepsis for AKI vs no AKI was—interleukin-6: 48.6 vs 35 pg/ml, P<0.03; and tumor necrosis factor: 6.5 vs 4.9 pg/ml, P<0.001.

Figure 4
Circulating inflammatory, coagulation, and fibrinolysis marker concentrations in 1380, 712, and 710 patients with and without acute kidney injury, stratified by severe sepsis during first week of CAP

Significant derangement in coagulation occurred in patients with AKI (Figure 4b). Of patients who developed severe sepsis, higher concentration of thrombin-antithrombin complexes (P<0.001) and lower concentration of antithrombin persisted during the first week of CAP (P<0.001) in patients with AKI. Day 1 geometric mean concentration for AKI vs no AKI was 6.3 vs 4.9 μg/ml for thrombin-antithrombin complexes (P=0.08) and 79.8 vs 88.1 μg/ml, for antithrombin (P<0.001). In contrast, of patients without severe sepsis, circulating concentrations of thrombin-antithrombin complexes, factor IX, and antithrombin did not differ between those with developing and not developing AKI (Figure 4b).

Patients with severe sepsis who developed AKI had marked elevation in D-dimer and plasminogen activator inhibitor-1 concentrations during the first week of pneumonia (P=0.001; Figure 4c). The geometric mean concentration on day 1 in AKI vs no AKI was—D-dimer: 901 vs 643 ng/ml, P=0.019; and plasminogen activator inhibitor–1: 7.9 vs 4.0 ng/ml, P=0.001. In contrast, of patients without severe sepsis, only circulating D-dimer concentration was higher on day 1 (689.3 vs 494.9 ng/ml, P=0.001) and during the first week of CAP (P<0.001).


We found, perhaps not surprisingly, that AKI was very common in CAP. Importantly, however, even in those who appeared to have an uncomplicated course, as defined by different criteria, AKI was still very common, developing in around one-quarter of patients. In this latter group, development of AKI was associated with considerable mortality. Within those developing severe sepsis, AKI occurred in patients who also had evidence of greater activation of inflammatory, coagulation, and fibrinolysis pathways. Among the patients with less severe CAP, AKI was associated with modest but significant differences in immune and fibrinolysis activation. This study, to the best of our knowledge, is the first large-scale investigation of the epidemiology and outcome of AKI, and immune response to CAP in patients with AKI. The result of our study suggests that AKI is common in patients with CAP even among those without severe sepsis or even severe CAP. Furthermore, the immune response to infection is different in patients with and without AKI.

An overall 34% incidence of AKI in patients with CAP is very similar to that reported for critically ill patients.5 Our estimate is likely to rise further with increasing prevalence of chronic kidney disease as the populations in the developed world and even in parts of the developing world age. Development of AKI was associated with high risk of death early on, declining subsequently, yet persisting up to 100 days after hospitalization for CAP. Our findings are surprising given that most patients with AKI were discharged by 8 days, suggesting that the risk of death persists even though the clinical signs have resolved and subjects appeared to be stable for discharge after CAP.25

We also found that AKI was associated with a significant increase in the risk of death even in patients with non-severe CAP. We chose four different definitions for ‘non-severe’ CAP. Pneumonia Severity Index and CURB-65 are widely validated severity indices for CAP. We also used absence of severe sepsis and lack of intensive care use as two additional methods to identify patients who were less severely ill throughout hospitalization with pneumonia. Surprisingly, we found that many patients with non-severe CAP admitted to medical wards developed AKI and although most had mild AKI (Risk), there was still an increased risk of death. This finding is consistent with other studies of AKI showing that small changes in kidney function are associated with higher mortality.2628 Similarly, of the lower severity patients defined by CURB-65, development of AKI was still associated with higher mortality. Importantly, these non-severe CAP patients with AKI would not have met inclusion criteria for any of the major sepsis clinical trials and yet the presence of AKI was associated with a significant increase in mortality. We speculate that this group of patients may be better suited to the study of novel therapies for AKI because their comorbidities are fewer and the risk of death attributable to AKI may be more readily modifiable than that for AKI in severely ill patients.

As expected, we found higher concentrations of inflammatory, coagulation, and fibrinolysis markers in patients with AKI who developed severe sepsis similar to other studies.17,29 However, a key finding of our study is that patients with AKI but without severe sepsis also had higher inflammatory and fibrinolysis markers. These observations suggest that the immune response in CAP is different in patients with AKI and may have a bidirectional relationship. Higher immune response might have caused AKI or might have occurred as a consequence of AKI (e.g., decreased elimination, increased cytokine release) or both. Because many of our patients had AKI on day 1, we were unable to delineate a temporal association to support any cause-effect relationship between AKI and immune response.

There are important limitations to our study. First, as is the case with most studies of AKI, we did not know premorbid creatinine in many patients and relied on screening for chronic kidney disease by medical history and estimated creatinine according to recommended guidelines.30 This might have resulted in an overestimate in the incidence of AKI if there was a higher than reported underlying prevalence of chronic kidney disease among our patients. Nevertheless, we found that there was a moderate concordance between known premorbid creatinine and estimated baseline creatinine in classifying patients with AKI, suggesting that our results are less likely to be affected by misclassification bias.31

Second, we were unable to control for fluid, hemodynamic, and other concurrent interventions that could have influenced course of AKI, biomarker patterns, and outcomes. Third, there is no gold standard for the term ‘non-severe CAP’. We therefore presented results defining non-severe CAP in four ways. We chose these definitions on the basis of clinical face validity, but recognize that the definitions are arbitrary. Fourth, we did not account for the competing risk between mortality associated with sepsis-induced AKI and subsequent decline in kidney function, ultimately reflected as either chronic kidney disease or end-stage renal disease, due to smaller cohort and shorter duration of follow-up.

Our study also has several strengths. First, being a multicenter study, our finding of high risk of AKI and its association with adverse outcomes independent of severity of CAP is highly generalizable. Second, we showed that the risk of death associated with AKI persists long after discharge following CAP hospitalization, though the mechanism responsible for late death is unclear in patients with AKI. Third, by considering subjects with CAP alone, we enrolled a more homogenous group of patients with early sepsis (CAP), unlike earlier studies that enrolled patients with different infections and at varying time points. Fourth, we were able to quantify that most sepsis-induced AKI occurs before hospitalization and therefore were able to examine the timing of AKI. Fifth, we carefully measured biomarkers from the time of admission when biomarker activation is generally highest and when concentrations are least likely to be modified by therapy. We stratified biomarker analysis by severe sepsis to account for the independent relationship between AKI and biomarkers in the absence of coexisting organ dysfunction; unlike many other studies that measured cytokines late and after coexisting organ dysfunctions were already evident.

In summary, the results of our study show that AKI occurs in approximately one-third of patients with CAP, and in a quarter of patients with non-severe CAP. Occurrence of AKI is associated with higher concentrations of immune and fibrinolysis markers even in those without severe sepsis, and an increased risk of death. Our study underscores the importance of increased awareness of this high incidence of AKI and mortality after pneumonia, and the need for development of prevention/treatment strategies especially for patients with non-severe CAP who have not typically been included in sepsis trials. In addition, further research is needed to understand the mechanism of death after AKI, particularly as the risk appears to persist long after AKI and CAP have resolved.

Materials and Methods

Study design and selection of participants

The analyses were based on all hospitalized patients enrolled in the Genetic and Inflammatory Markers of Sepsis (GenIMS) study.32 GenIMS is a large, multicenter, observational cohort study of subjects with CAP presenting to the emergency departments of 28 teaching and nonteaching hospitals in the United States between November 2001 and November 2003. Eligible patients were [gt-or-equal, slanted]18 years old and had a clinical and radiological diagnosis of CAP using criteria by Fine et al.22 Exclusion criteria included transfer from another hospital, discharge from an acute care hospital within the previous 10 days, pneumonia within the previous 30 days, chronic dependency on mechanical ventilation, cystic fibrosis, active pulmonary tuberculosis, admission for palliative care, prior enrollment in the study, incarceration, and pregnancy. The institutional review boards at all participating sites approved the study, and we obtained written informed consent from all participants or their proxies.

Methods of data collection

We gathered baseline and sequential clinical information by structured patient or proxy interviews. We prospectively ascertained comorbid conditions using the Charlson Comorbidity Index,33 and severity of illness using the Acute Physiology and Chronic Health Evaluation III score,34 and Pneumonia Severity Index.22 We calculated CURB-65 score23 retrospectively with altered mental status or a new change in Glasgow Coma Scale score as proxy measures for confusion.35 We defined severe sepsis as infection and acute organ dysfunction, following international consensus criteria.24 We defined acute organ dysfunction as a new SOFA score of three or greater in any of six organ systems.10

Definition of acute kidney injury

We classified AKI using the maximum RIFLE stages, as proposed by the Acute Dialysis Quality Initiative30 at any time during hospitalization. The RIFLE stage was determined based on the worse of either serial serum creatinine or urine output. For patients with no known premorbid creatinine and no known medical history of chronic kidney disease, we estimated premorbid creatinine using the Modification of Diet in Renal Disease equation,36 as recommended by the Acute Dialysis Quality Initiative. We then selected the lower creatinine value from either the hospital admission creatinine or the Modification of Diet in Renal Disease creatinine as the baseline value.37 Patients were classified as stage Risk, if serum creatinine was 1.5 times the baseline creatinine, or urine output <0.5 ml/kg/h for 6h; stage Injury, if serum creatinine was twice the baseline, or urine output <0.5 ml/kg/h for 12 h; stage Failure, if serum creatinine was thrice the baseline, or creatinine [gt-or-equal, slanted]4 mg/dl with an acute rise >0.5 mg/dl, or urine output <0.3 ml/kg/h for 24 h, or anuria for 12 h.30

Criteria for non-severe CAP

We used four different sets of criteria to classify the severity of CAP.2224,38 We defined non-severe CAP as classes I–III for Pneumonia Severity Index calculated at admission and group 1 for CURB-65 to identify patients who had non-severe CAP at hospital admission. We also chose patients who did not develop severe sepsis according to the SOFA score of less than three in any of six organ systems, and those never admitted to the intensive care unit to identify patients who were less severely ill with CAP throughout index hospitalization.

Outcome variables

Our primary outcome was 1-year all-cause mortality. Secondary outcomes included length of stay, in-hospital and 90-day mortality. Study coordinators ascertained deaths in hospital and post-discharge mortality was ascertained by telephone and National Death Index search.

Laboratory procedures

Blood was drawn for cytokines (tumor necrosis factor, interleukin-6, -10), coagulation (thrombin-antithrombin complexes, antithrombin, factor IX), and fibrinolysis (plasminogen activator inhibitor-1, and D-dimer) markers in the emergency department and daily for the first week, while a subject remained in hospital. Detailed sample handling and laboratory procedures have been described elsewhere.32 We rechecked all outlier values and masked laboratory personnel to clinical data. Although we analyzed inflammatory markers in 1380 patients, for logistical and cost reasons, coagulation and fibrinolysis markers were assessed only in a random subset of 712 and 710 patients.

Statistical analysis

We conducted univariate comparisons using χ2_tests, Student's t-tests, or their nonparametric equivalents, as appropriate. We assessed 1-year mortality by comparing Kaplan–Meier failure plots using log-rank test. We conducted survival analyses using Cox proportional hazard model to estimate hazard ratios for association between AKI and 1-year mortality. We used Gray's model to estimate hazard ratios for nine time nodes and eight intervals, if hazards failed Cox's proportionality assumption.39 We adjusted for potential confounders by expanding the models to include age, sex, race/ethnicity, and the Charlson Comorbidity Index.

We present sequential biomarker data as plots of daily means stratified by development of AKI and severe sepsis. We assumed a log-normal distribution of biomarkers and analyzed data in natural log scale. We constructed Tobit models to account for truncated data that fell below detection thresholds,40 and generated daily geometric means to compare the concentration between groups for a single time point. For sequential data, we conducted regression analysis with mixed models that accounted for correlation of repeated measures over time,41 incorporating Tobit models as necessary.42 Models included linear and quadratic terms to allow evaluation of trends. All analyses were performed using SAS 9.0 (SAS Institute, Cary, NC, USA) assuming statistical significance at P<0.05.


The GenIMS Study was funded by the National Institute of General Medical Sciences (R01 GM61992) with additional support from GlaxoSmithKline (Philadelphia, PA, USA) for enrollment and clinical data collection, and from Diagnostic Products Corporation (Los Angeles, CA, USA) for the cytokine and coagulation assays. The agencies had no role in the design and conduct of the study; the collection, analysis, and interpretation of the data; or the preparation, or approval of the paper. We thank the nurses, respiratory therapists, phlebotomists, physicians, and other health care professionals who participated in GenIMS, as well as the patients and their families who supported this study. Source of support: GenIMS was funded by NIGMS R01 GM61992, with additional support from GlaxoSmithKline for cytokine assays. This publication was made possible by grant KL2 RR024154 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and the NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Information on NCRR is available at Information on Re-engineering the Clinical Research Enterprise can be obtained from This publication was also supported in part by career development grant K23GM83215 from NIGMS.


This study was presented in part at the Society of Critical Care Medicine's 37th Critical Care Congress, Honolulu, Hawaii, 2-6 February 2008 and published in part as an abstract in Crit Care Med 2007 Vol 35, No.12 (Suppl).

Disclosure: All the authors declared no competing interests.

Author contributions: Murugan, Lee, Kong and Kellum had had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Murugan, Kellum, and Angus. Acquisition of data: Kellum and Angus. Cytokine analysis: Carter. Analysis and interpretation of data: Murugan, Kellum, Yende, Karajala-Subramanyam, Lee, Kong, and Angus. Drafting of the paper: Murugan, Karajala-Subramanyam, Kellum, Yende, and Angus. Critical revision of the paper for important intellectual content: Kellum, Yende, Angus, Kong, Lee, and Carter. Statistical analysis: Lee and Kong. Obtainment of funding: Kellum and Angus. Administrative, technical, or material support: Kong, Kellum, and Angus. Study supervision: Kellum and Angus.


1. Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294:813–818. [PubMed]
2. Bagshaw SM, Uchino S, Bellomo R, et al. Septic acute kidney injury in critically ill patients: clinical characteristics and outcomes. Clin J Am Soc Nephrol. 2007;2:431–439. [PubMed]
3. Schrier RW, Wang W. Acute renal failure and sepsis. N Engl J Med. 2004;351:159–169. [PubMed]
4. Chen YC, Jenq CC, Tian YC, et al. RIFLE classification for predicting in-hospital mortality in critically ill sepsis patients. Shock. 2009;31:139–145. [PubMed]
5. Bagshaw SM, George C, Bellomo R. Early acute kidney injury and sepsis: a multicentre evaluation. Crit Care. 2008;12:R47. [PMC free article] [PubMed]
6. Yegenaga I, Hoste E, Van Biesen W, et al. Clinical characteristics of patients developing ARF due to sepsis/systemic inflammatory response syndrome: results of a prospective study. Am J Kidney Dis. 2004;43:817–824. [PubMed]
7. Janssen vD, Spapen H, Geers C, et al. Sepsis-related acute kidney injury: a protective effect of drotrecogin alpha (activated) treatment? Acta Anaesthesiol Scand. 2008;52:1259–1264. [PubMed]
8. Lopes JA, Jorge S, Resina C, et al. Acute kidney injury in patients with sepsis: a contemporary analysis. Int J Infect Dis. 2009;13:176–181. [PubMed]
9. Angus DC, Linde-Zwirble WT, Lidicker J, et al. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Crit Care Med. 2001;29:1303–1310. [PubMed]
10. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med. 1996;22:707–710. [PubMed]
11. Abraham E, Reinhart K, Opal S, et al. Efficacy and safety of tifacogin (recombinant tissue factor pathway inhibitor) in severe sepsis: a randomized controlled trial. JAMA. 2003;290:238–247. [PubMed]
12. Bernard GR, Vincent JL, Laterre PF, et al. Efficacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med. 2001;344:699–709. [PubMed]
13. Warren BL, Eid A, Singer P, et al. Caring for the critically ill patient. High-dose antithrombin III in severe sepsis: a randomized controlled trial. JAMA. 2001;286:1869–1878. [PubMed]
14. Fiedler VB, Loof I, Sander E, et al. Monoclonal antibody to tumor necrosis factor-alpha prevents lethal endotoxin sepsis in adult rhesus monkeys. J Lab Clin Med. 1992;120:574–588. [PubMed]
15. Knotek M, Rogachev B, Wang W, et al. Endotoxemic renal failure in mice: role of tumor necrosis factor independent of inducible nitric oxide synthase. Kidney Int. 2001;59:2243–2249. [PubMed]
16. Cunningham PN, Dyanov HM, Park P, et al. Acute renal failure in endotoxemia is caused by TNF acting directly on TNF receptor-1 in kidney. J Immunol. 2002;168:5817–5823. [PubMed]
17. Chawla LS, Seneff MG, Nelson DR, et al. Elevated plasma concentrations of IL-6 and elevated APACHE II score predict acute kidney injury in patients with severe sepsis. Clin J Am Soc Nephrol. 2007;2:22–30. [PubMed]
18. Iglesias J, Marik PE, Levine JS. Elevated serum levels of the type I and type II receptors for tumor necrosis factor-alpha as predictive factors for ARF in patients with septic shock. Am J Kidney Dis. 2003;41:62–75. [PubMed]
19. Garcia-Fernandez N, Montes R, Purroy A, et al. Hemostatic disturbances in patients with systemic inflammatory response syndrome (SIRS) and associated acute renal failure (ARF) Thromb Res. 2000;100:19–25. [PubMed]
20. Nathan C. Points of control in inflammation. Nature. 2002;420:846–852. [PubMed]
21. Scaffidi P, Misteli T, Bianchi ME. Release of chromatin protein HMGB1 by necrotic cells triggers inflammation. Nature. 2002;418:191–195. [PubMed]
22. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997;336:243–250. [PubMed]
23. Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax. 2003;58:377–382. [PMC free article] [PubMed]
24. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med. 2003;31:1250–1256. [PubMed]
25. Halm EA, Teirstein AS. Clinical practice. Management of community-acquired pneumonia. N Engl J Med. 2002;347:2039–2045. [PubMed]
26. Bihorac A, Yavas S, Subbiah S, et al. Long-term risk of mortality and acute kidney injury during hospitalization after major surgery. Ann Surg. 2009;249:851–858. [PubMed]
27. Coca SG, Peixoto AJ, Garg AX, et al. The prognostic importance of a small acute decrement in kidney function in hospitalized patients: a systematic review and meta-analysis. Am J Kidney Dis. 2007;50:712–720. [PubMed]
28. Newsome BB, Warnock DG, McClellan WM, et al. Long-term risk of mortality and end-stage renal disease among the elderly after small increases in serum creatinine level during hospitalization for acute myocardial infarction. Arch Intern Med. 2008;168:609–616. [PubMed]
29. Simmons EM, Himmelfarb J, Sezer MT, et al. Plasma cytokine levels predict mortality in patients with acute renal failure. Kidney Int. 2004;65:1357–1365. [PubMed]
30. Bellomo R, Ronco C, Kellum JA, et al. Acute renal failure—definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care. 2004;8:R204–R212. [PMC free article] [PubMed]
31. Bagshaw SM, Uchino S, Cruz D, et al. A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant. 2009;24:2739–2744. [PubMed]
32. Kellum JA, Kong L, Fink MP, et al. Understanding the inflammatory cytokine response in pneumonia and sepsis: results of the Genetic and Inflammatory Markers of Sepsis (GenIMS) study. Arch Intern Med. 2007;167:1655–1663. [PMC free article] [PubMed]
33. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383. [PubMed]
34. Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system. Risk prediction of hospital mortality for critically ill hospitalized adults. Chest. 1991;100:1619–1636. [PubMed]
35. Aujesky D, Auble TE, Yealy DM, et al. Prospective comparison of three validated prediction rules for prognosis in community-acquired pneumonia. Am J Med. 2005;118:384–392. [PubMed]
36. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002;39:S1–266. [PubMed]
37. Hoste EA, Clermont G, Kersten A, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care. 2006;10:R73. [PMC free article] [PubMed]
38. Yealy DM, Auble TE, Stone RA, et al. Effect of increasing the intensity of implementing pneumonia guidelines: a randomized, controlled trial. Ann Intern Med. 2005;143:881–894. [PubMed]
39. Kasal J, Jovanovic Z, Clermont G, et al. Comparison of Cox and Gray's survival models in severe sepsis. Crit Care Med. 2004;32:700–707. [PubMed]
40. Tobin J. Estimation of relationships for limited dependent variables. Econometrica. 1958;26:24–36.
41. Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics. 1982;38:963–974. [PubMed]
42. Epstein MP, Lin X, Boehnke M. A tobit variance-component method for linkage analysis of censored trait data. Am J Hum Genet. 2003;72:611–620. [PubMed]