Early 72-h serum creatinine kinetics predict 90-day mortality in hepatorenal syndrome-associated AKI - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Ren Fail . 2026 Apr 14;48(1):2656009. doi: 10.1080/0886022X.2026.2656009 Search in PMC Search in PubMed View in NLM Catalog Add to search Early 72-h serum creatinine kinetics predict 90-day mortality in hepatorenal syndrome-associated AKI Julian Müller-Kühnle Julian Müller-Kühnle a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany b Robert Bosch Gesellschaft für Medizinische Forschung (Robert Bosch Society for Medical Research), Stuttgart, Germany c Paracelsus Medizinische Universität (Paracelsus Medical University), Salzburg, Austria Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Writing – original draft Find articles by Julian Müller-Kühnle a, b, c, ✉ , Dominik Marschner Dominik Marschner d Department of Internal Medicine I, Universitätsklinikum Freiburg (University Medical Center Freiburg), Freiburg, Germany Conceptualization, Formal analysis, Methodology, Writing – review & editing Find articles by Dominik Marschner d , Severin Schricker Severin Schricker a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Data curation, Investigation, Writing – review & editing Find articles by Severin Schricker a , Arthur Schmidt Arthur Schmidt e Department of Gastroenterology, Hepatology and Endocrinology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Data curation, Investigation, Methodology, Writing – review & editing Find articles by Arthur Schmidt e , Mathias Becker Mathias Becker f Department of Cardiology and Angiology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Data curation, Formal analysis, Methodology, Writing – review & editing Find articles by Mathias Becker f , Leonie Kraft Leonie Kraft a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Data curation, Methodology, Writing – review & editing Find articles by Leonie Kraft a , Jörg Latus Jörg Latus a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Data curation, Project administration, Supervision, Writing – review & editing Find articles by Jörg Latus a , Moritz Schanz Moritz Schanz a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Conceptualization, Methodology, Project administration, Supervision, Validation, Writing – review & editing Find articles by Moritz Schanz a Author information Article notes Copyright and License information a Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany b Robert Bosch Gesellschaft für Medizinische Forschung (Robert Bosch Society for Medical Research), Stuttgart, Germany c Paracelsus Medizinische Universität (Paracelsus Medical University), Salzburg, Austria d Department of Internal Medicine I, Universitätsklinikum Freiburg (University Medical Center Freiburg), Freiburg, Germany e Department of Gastroenterology, Hepatology and Endocrinology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany f Department of Cardiology and Angiology, Robert Bosch Krankenhaus (Robert Bosch Hospital), Stuttgart, Germany Supplemental data for this article is available online at https://doi.org/10.1080/0886022X.2026.2656009 ✉ CONTACT Julian Müller-Kühnle [email protected] Department of General Internal Medicine and Nephrology, Robert Bosch Krankenhaus, Auerbachstraße 110, Stuttgart, 70376, Germany Roles Julian Müller-Kühnle : Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Writing – original draft Dominik Marschner : Conceptualization, Formal analysis, Methodology, Writing – review & editing Severin Schricker : Data curation, Investigation, Writing – review & editing Arthur Schmidt : Data curation, Investigation, Methodology, Writing – review & editing Mathias Becker : Data curation, Formal analysis, Methodology, Writing – review & editing Leonie Kraft : Data curation, Methodology, Writing – review & editing Jörg Latus : Data curation, Project administration, Supervision, Writing – review & editing Moritz Schanz : Conceptualization, Methodology, Project administration, Supervision, Validation, Writing – review & editing Received 2026 Feb 25; Accepted 2026 Apr 1; Collection date 2026. © 2026 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License ( http://creativecommons.org/licenses/by-nc/4.0/ ), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. PMC Copyright notice PMCID: PMC13081340 PMID: 41978512 Abstract Hepatorenal syndrome-associated acute kidney injury (HRS-AKI) carries high short-term mortality. Prognosis is usually based on creatinine-defined AKI stage, but early creatinine trajectories may add information. In this retrospective cohort study (January 2019–September 2024), we included adults with cirrhosis fulfilling contemporary ADQI/ICA-based criteria for HRS-AKI. Using a prespecified 72-h landmark approach, excluding patients who died within the first 72h, we classified serum creatinine kinetics over the first 72 h after diagnosis as increasing or decreasing and evaluated associations with 90-day all-cause mortality after diagnosis. Landmark Cox models adjusted for age, sex, baseline creatinine, Child–Pugh class, and platelet count and tested interaction with AKI stage. Seventy-five patients comprised the landmark cohort (rising n = 36; falling n = 39). Ninety-day survival differed by kinetics (log-rank p = 0.0018). Rising creatinine independently predicted mortality (adjusted hazard ratio 2.24, 95% CI 1.20–4.18; p = 0.011) alongside Child–Pugh class (hazard ratio per class 3.99, 95% CI 1.67–9.52; p = 0.002). There was no evidence of effect modification by AKI stage (interaction p = 0.27). Renal replacement therapy occurred more often with rising creatinine (36.1 vs. 10.3%; p = 0.012). Early 72-h creatinine kinetics provides prognostic information beyond conventional AKI stage in HRS-AKI; an increasing trajectory identifies patients at high risk of death and dialysis and may support early risk stratification. Keywords: Hepatorenal syndrome, acute kidney injury, serum creatinine, kinetics, mortality, cirrhosis Introduction Cirrhosis is frequently complicated by acute kidney injury (AKI), which occurs in up to half of hospitalized patients and substantially worsens prognosis [ 1 ]. Among the etiologies of renal dysfunction in cirrhosis, hepatorenal syndrome-associated acute kidney injury (HRS-AKI) represents one of the most severe syndromes, characterized by rapidly evolving circulatory dysfunction, limited treatment response, and high short-term mortality [ 2 , 3 ]. Although HRS-AKI has traditionally been considered largely functional, accumulating evidence suggests that subclinical structural injury and systemic inflammatory mechanisms may contribute, particularly in advanced disease and acute-on-chronic liver failure (ACLF) [ 4–8 ]. Risk stratification remains a central challenge in HRS-AKI. Current diagnostic and severity frameworks (ICA and the updated ADQI/ICA consensus) rely predominantly on serum creatinine changes; the updated ADQI/ICA consensus additionally incorporates urine output where available [ 9 , 10 ]. However, creatinine is an imperfect surrogate of kidney function in cirrhosis and a static stage at diagnosis may not capture the subsequent clinical trajectory, treatment response, or evolving multi-organ failure [ 9–13 ]. In critical care and sepsis-associated AKI, early trajectories of kidney function have repeatedly shown prognostic relevance beyond peak or baseline values, identifying transient versus persistent AKI phenotypes and correlating with survival and renal recovery [ 14–19 ]. Whether analogous trajectory-based information improves risk stratification in rigorously defined HRS-AKI cohorts remains insufficiently studied, despite the clinical need for early decisions regarding escalation of care, renal support strategies, and transplant evaluation [ 10 ]. Importantly, the dynamic serum creatinine criteria of ICA/ADQI were used in this study to establish the diagnosis of AKI/HRS-AKI, whereas the 72-h creatinine kinetics examined here were assessed only after diagnosis and were intended as a post-diagnostic prognostic marker rather than a diagnostic criterion. Therefore, we investigated whether early serum creatinine kinetics during the first 72 h after HRS-AKI diagnosis predict 90-day mortality, and whether this dynamic parameter adds prognostic information beyond conventional AKI stage and routinely available clinical markers. Materials and methods Study design, setting, and reporting We conducted a retrospective, single-center cohort study at Robert Bosch Hospital, Stuttgart, Germany, evaluating the prognostic value of early serum creatinine kinetics in patients with hepatorenal syndrome–associated acute kidney injury (HRS-AKI). The study period spanned January 2019 to September 2024. The primary objective was to determine whether early creatinine kinetics (first 72 h after HRS-AKI diagnosis) predict 90-day mortality and provide prognostic information beyond conventional AKI staging and established clinical covariates. Ethics approval and consent This retrospective cohort study was conducted in accordance with the principles of the Declaration of Helsinki and its later amendments. The study protocol was approved by the Ethics Committee of the Medical Faculty of the Eberhard Karls University of Tübingen, Tübingen, Germany (approval number: 587/2024BO2). Robert Bosch Hospital is an academic teaching hospital affiliated with the University of Tübingen; therefore, this ethics committee served as the responsible institutional review board. The requirement for informed consent was waived by the ethics committee due to the retrospective design and the use of routinely collected clinical data. Data were processed in a pseudonymized manner in accordance with applicable data protection regulations. Study population and case ascertainment Patients were initially identified through ICD-10 coding for hepatorenal syndrome during the study period (January 2019 to September 2024). Among 174 initially identified cases, 148 unique patient records remained after removal of duplicate entries and repeated episodes. These records underwent a detailed chart review to confirm eligibility. Adults (≥18 years) were eligible if they had cirrhosis and fulfilled contemporary consensus-based diagnostic criteria for HRS-AKI. In brief, HRS-AKI required (i) cirrhosis with ascites, (ii) AKI defined by serum creatinine increase ≥0.3 mg/dL within 48 h or ≥50% from baseline within 7 days, (iii) absence of shock, (iv) no recent exposure to nephrotoxic drugs, and (v) no evidence of structural kidney disease (proteinuria ≤500 mg/day, hematuria <50 red blood cells/high-power field, and normal renal ultrasound). Cases were adjudicated by detailed chart review in accordance with the contemporary ICA/ADQI framework, including assessment of the clinical course and response to initial management where documented. After adjudication against these criteria, 77 patients constituted the confirmed HRS-AKI cohort. Exclusion criteria were duplicate records or repeated episodes beyond the first qualifying episode, age <18 years, and failure to meet ICA/ADQI criteria for HRS-AKI after chart review. For kinetics-based landmark survival analyses, patients who died before the 72-h landmark were additionally excluded. If patients had multiple HRS-AKI episodes during the study period, only the first qualifying episode was analyzed to ensure independence of observations. Definitions Index date, baseline, and follow-up The index date was defined as the date/time of clinical HRS-AKI diagnosis (time zero for defining early kinetics). Baseline covariates were defined as values closest to the index time (typically within the first 24 h), unless otherwise specified. The primary endpoint was all-cause mortality within 90 days after the index date. For time-to-event analyses, follow-up ended at death or administrative censoring at day 90 after diagnosis. AKI stage AKI stage at diagnosis was defined by ICA criteria using baseline serum creatinine and the creatinine value at the time of diagnosis, and categorized as stage 1, 2, or 3. Early creatinine kinetics (primary exposure) Early creatinine kinetics were defined a priori using routine in-hospital serum creatinine measurements obtained during the first 72 h after HRS-AKI diagnosis. Patients were categorizedaccording to the overall direction of change between the creatinine value at diagnosis and the last available creatinine measurement within the subsequent 72 h. Patients were classified as having increasing creatinine if the 72-h value was higher than the value at diagnosis, and decreasing creatinine if the 72-h value was lower. Patients without sufficient creatinine data to assign a trajectory category were excluded from kinetics-based analyses. Patients receiving renal replacement therapy (RRT) within the first 72 h were not excluded a priori ; however, because RRT may influence short-term serum creatinine independently of the underlying disease course, the potential impact of early RRT on trajectory assignment was specifically examined in sensitivity analyses. Sensitivity definitions of early trajectory To assess the robustness of the binary kinetics definition, we prespecified two additional continuous measures within the same 72-h window. First, creatinine slope (mg/dL per day) was estimated from available measurements between 0 and 72 h, requiring at least two time points. Second, percent change from baseline to 72 h was calculated as ( creatinin e 72 − hcreatinine _ baseline ) / creatinine _ baseline × 100 . Because percent change required an observed 72-h creatinine value, this analysis was performed in a reduced sample when 72-h values were missing. Landmark analysis to avoid immortal-time bias Because the exposure (creatinine kinetics) is defined using information accrued during the first 72 h after diagnosis, patients who die before 72 h cannot be classified and would induce immortal-time bias. Therefore, we performed a 72-h landmark analysis for all kinetics-based survival models: patients with survival <72 h after diagnosis were excluded, and time-to-event was measured from the landmark time (72 h after diagnosis) forward. To preserve clinical interpretability of ‘90-day mortality after diagnosis’ while using a landmark time origin, models were administratively censored at day 90 post-diagnosis, corresponding to 87 days of follow-up after the 72-h landmark. Outcomes Primary outcome The primary outcome was all-cause mortality within 90 days after HRS-AKI diagnosis (evaluated from the 72-h landmark for kinetics-based survival analyses). Secondary outcomes Secondary outcomes within 90 days after HRS-AKI diagnosis included (1) initiation of RRT, (2) intensive care unit (ICU) admission, and (3) receipt of terlipressin therapy. These outcomes were summarized descriptively and compared between the increasing and decreasing creatinine kinetics groups. Data collection Data were extracted from electronic medical records using a standardized abstraction approach. Variables collected included demographics, anthropometrics, cirrhosis etiology, comorbidities, baseline vital signs, and laboratory values, including renal function, electrolytes, hematology, coagulation parameters, inflammatory markers, liver enzymes, and serum albumin. We also recorded major therapies/interventions, including terlipressin therapy, ICU admission, and RRT; these treatment variables were collected descriptively and were not prespecified as outcome measures unless stated otherwise. Survival status at 90 days was determined from hospital records and/or follow-up documentation. Statistical analysis General approach Continuous variables were summarized as mean ± standard deviation ( SD ) or median [interquartile range (IQR)], as appropriate; categorical variables were summarized as n (%). Baseline characteristics were compared between groups using the Mann–Whitney U test for continuous variables and χ 2 test or Fisher’s exact test for categorical variables, as appropriate. These comparisons were considered descriptive; p -values were not used for covariate selection, and no adjustment for multiple comparisons was applied. Kaplan-Meier analyses Kaplan-Meier methods were used to describe survival up to 90 days after diagnosis. For analyses involving creatinine kinetics, curves were displayed from the 72-h landmark, and groups were compared using the log-rank test. Survival by AKI stage was assessed using log-rank testing; for three-group comparisons, pairwise post-hoc comparisons were adjusted using the Holm-Bonferroni procedure. Cox proportional hazards modeling (primary model) In the 72-h landmark cohort, we fitted a prespecified multivariable Cox proportional hazards model to quantify the association between early creatinine kinetics (increasing vs. decreasing) and mortality. The adjustment set was defined a priori based on clinical relevance and to reduce collinearity and overfitting, and included age (years), sex, baseline serum creatinine (mg/dL), Child-Pugh class modeled as an ordinal variable ( A = 1, B = 2, C = 3), and platelet count scaled per 50 × 10 9 /L. Results are reported as hazard ratios (HRs) with 95% confidence intervals (CIs). To assess whether creatinine kinetics provided incremental prognostic information beyond the baseline covariates, we compared nested models (without vs. with kinetics) using a likelihood-ratio test. Interaction analysis To assess whether the prognostic effect of creatinine kinetics differed by baseline AKI stage, we tested an interaction term (kinetics × AKI stage) within the landmark Cox framework and reported the corresponding likelihood-ratio test p -value. Model assumptions and performance The proportional hazards assumption was assessed using Schoenfeld residual-based tests; results were summarized by covariate-level and global tests. Discrimination was quantified using Harrell’s C-index. Prognostic performance was additionally evaluated at 87 days post-landmark using time-dependent AUC and Brier score, and calibration was inspected graphically using standard calibration curves for survival predictions. Sensitivity analyses We performed prespecified sensitivity analyses to evaluate the robustness of the primary association between early creatinine kinetics and mortality. First, we replaced the binary kinetics exposure with two alternative continuous trajectory measures within the same 72-h window-creatinine slope (mg/dL/day) and percent change from baseline to 72 h. Second, we repeated the primary model after excluding patients receiving RRT at or before the 72-h landmark. In addition, because early RRT may influence short-term serum creatinine trajectories, we specifically reviewed patients receiving RRT within the first 72 h after diagnosis and performed an additional exclusion check of these patients. Third, we conducted subgroup analyses by Child–Pugh class (B vs. C) by fitting separate Cox models within each subgroup to estimate the association between creatinine kinetics and mortality. All tests were two-sided. A p -value <0.05 was considered statistically significant for primary analyses. Software All statistical analyses were performed using R version 4.5 (R Foundation for Statistical Computing, Vienna, Austria). Survival analyses used the survival package; prognostic performance (time-dependent AUC, Brier score, calibration) was assessed with riskRegression and related packages. Results Study population Between January 2019 and September 2024, 174 ICD-10-coded cases of hepatorenal syndrome were identified and screened. After removal of duplicate records and repeated episodes, 148 unique patient records underwent detailed chart review. Following application of strict ADQI/ICA criteria, 77 patients constituted the confirmed HRS-AKI cohort ( Figure 1 ). For kinetics-based time-to-event analyses, a prespecified 72-h landmark was applied to avoid immortal-time bias, excluding two patients who died within the first 72 h after diagnosis. The resulting landmark cohort comprised 75 patients and formed the basis for Kaplan-Meier and Cox regression analyses. Time zero for survival analyses corresponds to the 72-h landmark; follow-up therefore spans 0–87 days post-landmark (equivalent to 90 days post-diagnosis). Figure 1. Open in a new tab Study flow diagram and cohort derivation. A total of 174 ICD-10-coded cases of hepatorenal syndrome were identified during the study period (January 2019 to September 2024). After removal of duplicate entries and repeated episodes, 148 unique patient records underwent detailed chart review. Following adjudication according to strict ADQI/ICA 2023–2024 criteria, 77 patients constituted the confirmed HRS-AKI cohort. For kinetics-based survival analyses, a prespecified 72-h landmark approach was applied to avoid immortal-time bias; two patients who died within 72 h after diagnosis were excluded, yielding a final landmark cohort of 75 patients. Baseline characteristics Baseline demographic and clinical characteristics of the landmark cohort are summarized in Table 1 . Mean age was 62 ± 13 years, 49 (65%) patients were male, and alcohol-related cirrhosis was the most frequent etiology (47 [63%]). Liver disease severity was advanced, with Child–Pugh class C in 55 (73%). Baseline serum creatinine was 2.3 ± 1.6 mg/dL, baseline serum albumin was 2.7 ± 0.5 g/dL, and the median MELD score was 24 [IQR 19–29] ( Table 1 ). A comprehensive summary of baseline laboratory and comorbidity data is provided in Supplementary Table S1 . Table 1. Baseline demographic and clinical characteristics of the study cohort ( n = 75). Characteristic Value Age, years 62 ± 13 Male sex, n (%) 49 (65) BMI, kg/m² 28.6 ± 8.4 Cirrhosis etiology, n (%) Alcohol-related 47 (63) Viral hepatitis 8 (10) NASH 3 (4) Other 17 (22) Child–Pugh class, n (%) A 1 (1) B 19 (25) C 55 (73) Baseline creatinine, mg/dL 2.3 ± 1.6 Albumin at baseline, g/dL 2.7 ± 0.5 MELD score, median [IQR] 24 [19–29] Open in a new tab Values are presented as mean ± SD or median [IQR], as appropriate; categorical variables are shown as n (%). Laboratory values refer to the baseline assessment at HRS-AKI diagnosis. MELD was calculated from baseline serum creatinine, total bilirubin, INR, and sodium. Survival by AKI stage In the landmark cohort, Kaplan-Meier curves stratified by AKI stage showed limited separation over follow-up (overall log-rank p = 0.051; Figure 2 ). Overall, prognostic separation by AKI stage was modest in this cohort. Figure 2. Open in a new tab Kaplan-Meier curves for survival up to 90 days after HRS-AKI diagnosis, displayed from the 72-h landmark, stratified by AKI stage. To avoid immortal-time bias, patients who died within the first 72 h were excluded (landmark cohort). Time zero corresponds to the 72-h landmark; follow-up therefore spans 0–87 days post-landmark (equivalent to 90 days post-diagnosis). Shaded areas indicate 95% confidence intervals. Survival differences across AKI stages were assessed using the log-rank test ( p = 0.051). Survival by early creatinine kinetics In contrast, early creatinine kinetics within the first 72 h showed pronounced prognostic separation (log-rank p = 0.0018; Figure 3 ). Creatinine kinetics were categorized as increasing ( n = 36) versus decreasing ( n = 39). Patients with increasing creatinine experienced substantially worse survival across follow-up compared with those with decreasing creatinine ( Figure 3 ). Figure 3. Open in a new tab Kaplan–Meier curves for survival up to 90 days after HRS-AKI diagnosis, displayed from the 72-h landmark, stratified by early creatinine kinetics. To avoid immortal-time bias, patients who died within the first 72 h were excluded (landmark cohort, n = 75). Creatinine kinetics were defined over the first 72 h after diagnosis and categorized as increasing ( n = 36) versus decreasing ( n = 39). Time zero corresponds to the 72-h landmark; follow-up therefore spans 0–87 days post-landmark (equivalent to 90 days post-diagnosis). Shaded areas indicate 95% confidence intervals. Survival curves were compared using the log-rank test ( p = 0.0018). Patient characteristics by creatinine kinetics Patient characteristics and selected clinical course variables stratified by creatinine kinetics are shown in Table 2 . Compared with patients with decreasing creatinine, those with increasing creatinine were more frequently male and differed in anthropometrics (height and weight), coagulation parameters (INR), serum albumin, and platelet count; they also more often received renal replacement therapy. These comparisons are descriptive and should be interpreted cautiously given the number of variables examined ( Table 2 ). Table 2. Baseline demographic, clinical, and laboratory characteristics by early creatinine kinetics (first 72 h). Variable Increasing creatinine ( n = 36) Decreasing creatinine ( n = 39) p -Value Demographics/anthropometrics Age, years 63.5 [53.2–76.8] 62.0 [52.5–71.5] 0.346 Height, cm 177.5 [168.0–181.5] 166.0 [159.0–173.5] 0.003 Weight, kg 82.2 [74.2–96.2] 72.5 [62.9–87.8] 0.018 BMI, kg/m² 29.0 [25.8–32.6] 26.0 [22.4–30.9] 0.180 Observed survival time*, days 16.0 [7.5–33.0] 56.0 [22.0–234.0] <0.001 Vital signs at baseline Systolic blood pressure, mmHg 117.0 [100.0–130.0] 116.0 [103.8–137.0] 0.339 Heart rate, beats/min 90.5 [73.2–108.5] 86.5 [74.2–103.5] 0.878 Temperature, °C 36.5 [36.1–36.9] 36.4 [36.2–36.8] 0.634 Oxygen saturation, % 98.0 [95.2–99.8] 98.0 [96.5–100.0] 0.474 Coagulation/inflammation/hematology INR 1.8 [1.4–2.0] 1.4 [1.3–1.8] 0.019 Leukocytes, ×10⁹/L 11.8 [7.1–17.2] 10.4 [7.6–13.5] 0.436 Hemoglobin, g/dL 10.1 [7.6–11.6] 9.7 [8.1–11.9] 0.803 Platelet count, ×10⁹/L 82.5 [56.0–118.2] 126.0 [79.2–187.5] 0.020 Ionized calcium, mmol/L 1.1 [1.1–1.2] 1.1 [1.1–1.1] 0.745 Lactate, mg/dL 31.6 [22.7–61.0] 25.4 [18.1–41.3] 0.145 Phosphate, mmol/L 1.2 [1.1–1.6] 1.2 [0.9–1.4] 0.463 Urea at baseline, mg/dL 88.8 [45.0–111.0] 75.5 [45.5–130.2] 0.711 Sodium at baseline, mmol/L 132.0 [129.0–137.0] 130.0 [126.0–136.0] 0.232 Potassium at baseline, mmol/L 4.2 [3.8–4.9] 4.7 [3.9–5.1] 0.254 Total bilirubin at baseline, mg/dL 5.1 [2.1–9.5] 3.9 [1.3–10.2] 0.652 Albumin at baseline, g/dL 2.5 [2.3–2.8] 2.9 [2.5–3.3] 0.003 AST at baseline, U/L 84.5 [61.8–150.0] 88.0 [41.0–167.0] 0.395 ALT at baseline, U/L 40.0 [26.0–60.5] 38.0 [22.0–61.0] 0.669 Alkaline phosphatase at baseline, U/L 156.0 [112.5–289.2] 141.0 [108.0–242.0] 0.311 γ-GT at baseline, U/L 136.5 [70.2–325.2] 149.0 [75.0–431.0] 0.669 CRP at baseline, mg/L 3.9 [1.2–10.0] 3.0 [1.2–5.6] 0.196 Procalcitonin at baseline, ng/mL 0.8 [0.4–1.5] 0.6 [0.4–1.4] 0.680 Clinical characteristics/comorbidities, n (%) Sex (male)** 30 (83.3) 19 (48.7) 0.005 Arterial hypertension 13 (36.1) 13 (33.3) 1.000 Hepatic encephalopathy 19 (52.8) 19 (48.7) 1.000 Diabetes mellitus 10 (27.8) 9 (23.1) 0.948 Asthma 1 (2.8) 1 (2.6) 1.000 COPD 2 (5.6) 4 (10.3) 0.695 Coronary artery disease 6 (16.7) 8 (20.5) 0.809 Heart failure 2 (5.6) 1 (2.6) 0.982 Malignancy 5 (13.9) 9 (23.1) 0.405 Thromboembolic events 10 (27.8) 7 (17.9) 0.542 Hypothyroidism 4 (11.1) 7 (17.9) 0.545 Smoking 7 (19.4) 11 (28.2) 0.495 Statin use 9 (25.0) 4 (10.3) 0.205 TIPS 3 (8.3) 7 (17.9) 0.331 Prior stroke 4 (11.1) 2 (5.1) 0.647 Obesity (diagnosis or BMI >30 kg/m²) 9 (25.0) 9 (23.1) 1.000 Active GI bleeding 7 (19.4) 14 (35.9) 0.143 Continued alcohol use 15 (41.7) 17 (43.6) 0.893 Child–Pugh class, n (%) B: 4 (11.1); C: 32 (88.9) A: 1 (2.6); B: 13 (33.3); C: 25 (64.1) 0.039 Open in a new tab Baseline demographic, clinical, and laboratory characteristics are shown for patients with increasing ( n = 36) versus decreasing ( n = 39) serum creatinine within the first 72 h after HRS-AKI diagnosis (total n = 75 with classifiable kinetics). Continuous variables are presented as median [IQR] and categorical variables as n (%). p -Values are from the Mann–Whitney U test for continuous variables and the χ 2 test or Fisher’s exact test for categorical variables, as appropriate. p -Values are descriptive and were not adjusted for multiple comparisons. *Observed survival time is reported for descriptive purposes only and is not a baseline characteristic. **For sex, Fisher’s exact test was used. Observed survival time, renal replacement therapy, ICU admission, and terlipressin therapy are shown as clinical course variables and are not baseline characteristics. Independent predictors of mortality in the landmark Cox model In the pre-specified multivariable Cox proportional hazards model starting at the 72-h landmark ( Table 3 and Figure 4 ), increasing creatinine kinetics remained independently associated with higher mortality risk (HR 2.24, 95% CI 1.20–4.17; p = 0.011). Child–Pugh class (ordinal A→B→C) was also strongly associated with mortality (HR 3.99, 95% CI 1.67–9.52; p = 0.002). Age, sex, baseline creatinine, and platelet count did not show statistically significant independent associations in this model ( Table 3 ). Table 3. Multivariable Cox regression for mortality up to 90 days after diagnosis from the 72-h landmark in patients with HRS-AKI. Variable HR 95% CI p -Value Creatinine kinetics (increasing vs. decreasing; ref = decreasing) 2.24 1.20–4.17 0.011 Age (per year) 1.01 0.99–1.04 0.238 Sex (female vs. male; ref = male) 0.88 0.44–1.74 0.713 Child–Pugh class (per ordinal step A→B→C) 3.99 1.67–9.52 0.002 Baseline creatinine (per 1 mg/dL) 0.93 0.79–1.10 0.410 Platelet count (per 50 × 10⁹/L) 1.09 0.92–1.28 0.333 Open in a new tab Hazard ratios (HR) with 95% confidence intervals (CI) and p -values are shown for covariates included in the landmark Cox model. Follow-up starts at the 72-h landmark (day 3) and is administratively censored at 90 days after diagnosis (i.e., 87 days post-landmark). Creatinine kinetics are defined over the first 72 h and coded as increasing versus decreasing (reference). Figure 4. Open in a new tab Forest plot of adjusted hazard ratios for mortality up to 90 days after HRS-AKI diagnosis from the 72-h landmark. Hazard ratios (HR) with 95% confidence intervals (CI) from the prespecified multivariable Cox proportional hazards model are shown on a logarithmic scale (landmark cohort: n = 75; events = 56). Follow-up starts at the 72-h landmark (day 3) and is administratively censored at day 90 after diagnosis (i.e., 87 days post-landmark). The model included creatinine kinetics (increasing vs. decreasing within the first 72 h), age, sex, baseline serum creatinine, Child–Pugh class (ordinal A–C), and platelet count (scaled as modeled). The dashed vertical line indicates HR = 1 (no association). Model diagnostics supported the adequacy of the proportional hazards assumption ( Supplementary Table S3 ). An interaction term between creatinine kinetics and AKI stage did not indicate effect modification ( p = 0.27), suggesting that the prognostic relevance of early creatinine direction was broadly similar across AKI stages. Added prognostic value and model performance Adding binary creatinine kinetics to the baseline covariate set improved model fit (likelihood-ratio test p = 0.017). However, improvements in global discrimination were small: time-dependent AUC at 87 days post-landmark was 77.8% without kinetics and 78.1% with binary kinetics, and Brier scores at 87 days were similar across models (∼15%). Harrell’s C-index increased modestly from 0.662 (without kinetics) to 0.694 (with binary kinetics). Additional model-performance and calibration results are provided in Supplementary Table S2 ; global proportional-hazards checks were acceptable ( Supplementary Table S3 ). Sensitivity analyses Results were consistent in sensitivity analyses. Only two patients received renal replacement therapy (RRT) within the first 72 h after diagnosis, one in the increasing-creatinine group and one in the decreasing-creatinine group. In both cases, the first dialysis session occurred at ∼48 h after diagnosis, and the direction of creatinine change was already apparent before dialysis initiation. These patients were retained in the primary analysis; excluding them in an additional check did not materially alter the results. As shown in Supplementary Table S4 , the association between increasing creatinine kinetics and mortality also remained robust in the prespecified sensitivity analysis excluding patients receiving RRT at or before the 72-h landmark (HR 3.05, 95% CI 1.31–7.12; p = 0.009). In subgroup analyses by Child–Pugh class, the association between increasing kinetics and mortality remained significant in Child-Pugh C (HR 3.42, 95% CI 1.41–8.31; p = 0.007), whereas estimates in Child–Pugh B were directionally consistent but imprecise due to limited events (HR 2.01, 95% CI 0.42–9.54; p = 0.37; Supplementary Table S5 ). Alternative kinetics measures (creatinine slope and percent change from baseline to 72 h) were not statistically significant and did not improve AUC or Brier metrics compared with the binary kinetics definition. Percent-change analyses were limited by missing 72-h creatinine values, reducing sample size and statistical power. Secondary outcomes Secondary outcomes within 90 days after diagnosis are summarized in Table 4 . Overall, 17/75 (22.7%) patients required renal replacement therapy (RRT). RRT was more frequent in patients with increasing versus decreasing creatinine (36.1 vs. 10.3%, p = 0.012). ICU admission occurred in 63/75 (84.0%) patients and did not differ by kinetics group (86.1 vs. 82.1%, p = 0.757). Terlipressin therapy was administered in 58/75 (77.3%) patients and also did not differ significantly between groups (83.3 vs. 71.8%, p = 0.278). Among patients receiving RRT, 90-day survival was 2/17 (12%) and is reported descriptively due to small subgroup sizes ( Table 4 ). Table 4. Secondary outcomes within 90 days after HRS-AKI diagnosis in the 72-h landmark cohort, stratified by early creatinine kinetics. Outcome Overall ( n = 75) Increasing creatinine ( n = 36) Decreasing creatinine ( n = 39) p -Value Renal replacement therapy 17 (22.7%) 13 (36.1%) 4 (10.3%) 0.012 ICU admission 63 (84.0%) 31 (86.1%) 32 (82.1%) 0.757 Terlipressin therapy 58 (77.3%) 30 (83.3%) 28 (71.8%) 0.278 90-day Survival among patients receiving RRT 2/17 (12.0%) 1/13 (8.0%)* 1/4 (25.0%)* — Open in a new tab Values are shown as n (%) unless stated otherwise. The landmark cohort excludes patients who died within the first 72 h (overall n = 75). p -Values compare increasing versus decreasing creatinine kinetics using Fisher’s exact test. Survival among patients receiving renal replacement therapy (RRT) is reported descriptively only due to small subgroup sizes (*). Discussion In this retrospective cohort of patients with cirrhosis and HRS-AKI, early creatinine kinetics within the first 72 h after diagnosis emerged as a strong and independent prognostic marker of short-term survival. Patients with rising creatinine experienced substantially worse outcomes than those with early decline, with survival curves separating early and remaining distinct over follow-up. Importantly, this association persisted after adjustment for key clinical covariates and in sensitivity analyses, supporting early creatinine direction as a robust risk signal in a population with advanced liver disease and high competing mortality. Our findings align with evidence from critical illness and sepsis indicating that renal trajectories can outperform static measures for mortality risk stratification. In a prospective sepsis cohort, the early evolution of creatinine, rather than peak creatinine, independently associated with mortality [ 20 ]. Similarly, early recovery patterns after sepsis-associated AKI were strongly linked to 90-day outcomes, whereas partial recovery conferred no clear survival benefit [ 21 ]. Together with our results, these data support a central concept: the temporal course of kidney function captures clinically meaningful information about physiologic recovery potential that is not contained in a single creatinine value. From a pathophysiologic perspective, HRS-AKI in advanced decompensated cirrhosis - particularly in patients with ACLF - should be viewed within the broader systemic context of circulatory dysfunction, inflammation, and evolving organ failure rather than as an isolated renal event [ 7 , 8 , 22 , 23 ]. Within this framework, early creatinine kinetics likely reflect the balance between hemodynamic stabilization, inflammatory burden, and renal perfusion during the early critical phase of disease [ 7 ]. Accordingly, creatinine kinetics may serve as a pragmatic surrogate of ‘trajectory of multi-organ failure versus recovery’, rather than a kidney-only marker. In contrast, conventional AKI staging by ADQI/ICA criteria showed limited discriminatory capacity in our cohort. This is consistent with the recognized limitations of creatinine-based staging systems in cirrhosis and with the notion that static stage definitions may not fully capture prognosis in advanced disease [ 10 , 11 , 13 ]. More broadly, our data underscore a methodological point: while staging systems are essential for diagnostic standardization [ 10 ], they remain intrinsically static and may miss the prognostic information contained in early disease dynamics. Similar conclusions have been drawn in sepsis, where peak-creatinine–based definitions failed to capture outcome-relevant information that was reflected by early trends [ 20 ]. Supporting this trajectory-based paradigm, distinct creatinine courses have differentiated transient from sustained AKI and have been associated with mortality in broader hospital settings [ 17 ], and persistent or worsening kidney injury trajectories have repeatedly identified the highest-risk patients in critical care cohorts [ 18 , 19 ]. Extending these observations into the cirrhosis field, our study suggests that early creatinine kinetics provide additional prognostic granularity using routinely available laboratory data. This practical advantage is particularly relevant because consensus statements highlight both the limitations of creatinine and the need for improved prognostic approaches in cirrhosis-related AKI [ 10 ]. In this context, early creatinine direction offers a readily implementable tool that can be operationalized at the bedside without additional cost, specialized biomarkers, or complex modeling. The potential clinical implications are straightforward. A persistently rising creatinine within 72 h may identify patients on a high-risk trajectory who warrant intensified monitoring and earlier escalation of care, including expedited transplant evaluation and early consideration of renal replacement therapy in appropriate clinical contexts [ 22 , 23 ]. Conversely, early decline may indicate stabilization or response to therapy, supporting continuation of medical management and structured reassessment. In our cohort, meaningful distortion of trajectory classification by early renal replacement therapy appears unlikely: only two patients received RRT within the first 72 h, one in each kinetics group, and in both cases dialysis was initiated only after ∼48 h, when the direction of creatinine change was already apparent. Moreover, excluding these two patients in an additional check did not materially alter the results. These observations support that early creatinine kinetics primarily reflected the underlying clinical course rather than dialysis exposure in our cohort. By reframing creatinine as a dynamic signal rather than a static threshold, creatinine kinetics may augment clinical judgment in a population with narrow therapeutic windows and high mortality. Beyond creatinine trajectories, Child–Pugh class remained strongly associated with mortality, consistent with the central role of hepatic dysfunction and systemic inflammation in determining outcomes in advanced cirrhosis [ 7 , 8 ]. In this setting, early creatinine kinetics may provide complementary information: while Child–Pugh captures global liver disease severity, early creatinine direction captures the early renal/systemic course that may reflect treatment responsiveness and evolving organ failure dynamics. This study has several strengths, including strict application of updated ADQI/ICA diagnostic criteria [ 10 ], careful phenotyping, and survival analyses supported by sensitivity testing. Nonetheless, limitations should be considered. The retrospective, single-center design may limit generalizability and introduces the potential for residual confounding, including confounding by indication in treatment decisions (e.g., timing of renal replacement therapy or transplant referral). Sample size constrained subgroup inference, particularly in Child–Pugh B. Finally, serum creatinine remains an imperfect marker in cirrhosis, influenced by muscle mass, volume status, and hepatic dysfunction. Future studies should prospectively validate early creatinine kinetics in multicenter cohorts and evaluate how trajectory-based risk stratification can be integrated into clinical pathways for HRS-AKI [ 10 ]. Building on critical-care experience, evaluation of earlier trajectory windows (e.g., 24–48 h) may enable even earlier recognition of high-risk courses [ 17 , 20 ]. Ultimately, improving prognostic precision in HRS-AKI is clinically consequential: it can support timely transplant prioritization, guide escalation decisions, and inform resource allocation in a syndrome characterized by rapid deterioration and high short-term mortality. Supplementary Material Supplemental Material IRNF_A_2656009_SM1881.docx (28.2KB, docx) Acknowledgments We thank all members of the multidisciplinary HRS-AKI project team at Robert Bosch Hospital for their excellent collaboration, including the clinicians, and nursing staff involved in the care of the study population. We also acknowledge the institutional support that enabled this retrospective analysis. Generative AI support was used for language editing: ChatGPT (OpenAI; model GPT-5.2 Thinking; accessed January 2026). All AI-assisted text was critically reviewed, edited, and verified by the authors, who take full responsibility for the accurary, integrity, and originality of the manuscript Funding Statement JMK received funding from the Eva Mayr-Stihl Foundation (Vorgangsnummer: FP00017; funding period: 1 March 2024 to 31 December 2025). Funder website: https://eva-mayr-stihl-stiftung.de . The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Disclosure statement No potential conflict of interest was reported by the author(s). Data availability statement The data that support the findings of this study are available from the corresponding author, JMK, upon reasonable request. The data are not publicly available due to ethical and data protection restrictions because they contain information that could compromise the privacy of research participants. References 1. Appenrodt B, Lammert F.. Renal failure in patients with liver cirrhosis: novel classifications, biomarkers, treatment. Visc Med. 2018;34(4):246–252. doi: 10.1159/000492587. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Angeli P, Garcia-Tsao G, Nadim MK, et al. News in pathophysiology, definition and classification of hepatorenal syndrome: a step beyond the International Club of Ascites (ICA) consensus document. J Hepatol. 2019;71(4):811–822. doi: 10.1016/j.jhep.2019.07.002. [ DOI ] [ PubMed ] [ Google Scholar ] 3. Chaney A. A review for the practicing clinician: hepatorenal syndrome, a form of acute kidney injury, in patients with cirrhosis. Clin Exp Gastroenterol. 2021;14:385–396. doi: 10.2147/ceg.S323778. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Malakar S, Rungta S, Samanta A, et al. Understanding acute kidney injury in cirrhosis: current perspective. World J Hepatol. 2025;17(5):104724. doi: 10.4254/wjh.v17.i5.104724. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Arnold J, Avila E, Idalsoaga F, et al. Advances in the diagnosis and management of hepatorenal syndrome: insights into HRS-AKI and liver transplantation. eGastroenterology. 2023;1(2):e100009. doi: 10.1136/egastro-2023-100009. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Pyrsopoulos NT, Bernstein D, Kugelmas M, et al. Improving outcomes in hepatorenal syndrome-acute kidney injury with early diagnoses and implementation of approved treatment regimens. Gastroenterol Hepatol. 2023;19(12 Suppl 7):3–13. [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Moreau R, Jalan R, Gines P, et al. Acute-on-chronic liver failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenterology. 2013;144(7):1426–1437, 1437.e1–9. doi: 10.1053/j.gastro.2013.02.042. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Clària J, Stauber RE, Coenraad MJ, et al. Systemic inflammation in decompensated cirrhosis: characterization and role in acute-on-chronic liver failure. Hepatology. 2016;64(4):1249–1264. doi: 10.1002/hep.28740. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Angeli P, Gines P, Wong F, et al. Diagnosis and management of acute kidney injury in patients with cirrhosis: revised consensus recommendations of the International Club of Ascites. Gut. 2015;64(4):531–537. doi: 10.1136/gutjnl-2014-308874. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Nadim MK, Kellum JA, Forni L, et al. Acute kidney injury in patients with cirrhosis: acute Disease Quality Initiative (ADQI) and International Club of Ascites (ICA) joint multidisciplinary consensus meeting. J Hepatol. 2024;81(1):163–183. doi: 10.1016/j.jhep.2024.03.031. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Khatua CR, Sahu SK, Meher D, et al. Acute kidney injury in hospitalized cirrhotic patients: risk factors, type of kidney injury, and survival. JGH Open. 2021;5(2):199–206. doi: 10.1002/jgh3.12467. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Bansho ETO, Silva PES, Colombo BS, et al. Prognostic significance of the new criteria for acute kidney injury in cirrhosis. Ann Hepatol. 2018;17(3):461–469. doi: 10.5604/01.3001.0011.7390. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Chancharoenthana W, Leelahavanichkul A.. Acute kidney injury spectrum in patients with chronic liver disease: where do we stand? World J Gastroenterol. 2019;25(28):3684–3703. doi: 10.3748/wjg.v25.i28.3684. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Wang TJ, Huang CT, Wu CL, et al. Predictive approach for liberation from acute dialysis in ICU patients using interpretable machine learning. Sci Rep. 2024;14(1):13142. doi: 10.1038/s41598-024-63992-y. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Alfieri F, Ancona A, Tripepi G, et al. Continuous and early prediction of future moderate and severe acute kidney injury in critically ill patients: development and multi-centric, multi-national external validation of a machine-learning model. PLOS One. 2023;18(7):e0287398. doi: 10.1371/journal.pone.0287398. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Itenov TS, Berthelsen RE, Jensen JU, et al. Predicting recovery from acute kidney injury in critically ill patients: development and validation of a prediction model. Crit Care Resusc. 2018;20(1):54–60. [ PubMed ] [ Google Scholar ] 17. Warnock DG, Powell TC, Siew ED, et al. Serum creatinine trajectories for community- versus hospital-acquired acute kidney injury. Nephron. 2016;134(3):177–182. doi: 10.1159/000447757. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Uhel F, Peters-Sengers H, Falahi F, et al. Mortality and host response aberrations associated with transient and persistent acute kidney injury in critically ill patients with sepsis: a prospective cohort study. Intensive Care Med. 2020;46(8):1576–1589. doi: 10.1007/s00134-020-06119-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Ozrazgat-Baslanti T, Loftus TJ, Ren Y, et al. Association of persistent acute kidney injury and renal recovery with mortality in hospitalised patients. BMJ Health Care Inform. 2021;28(1):e100458. doi: 10.1136/bmjhci-2021-100458. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Vanmassenhove J, Lameire N, Dhondt A, et al. Prognostic robustness of serum creatinine based AKI definitions in patients with sepsis: a prospective cohort study. BMC Nephrol. 2015;16(1):112. doi: 10.1186/s12882-015-0107-4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Luo X, Yan P, Zhang N, et al. Early recovery status and outcomes after sepsis-associated acute kidney injury in critically ill patients. Zhong Nan Da Xue Xue Bao Yi Xue Ban. 2022;47(5):535–545. doi: 10.11817/j.issn.1672-7347.2022.210368. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Kabbani AR, Tergast TL, Manns MP, et al. [Treatment strategies for acute-on-chronic liver failure]. Med Klin Intensivmed Notfmed. 2021;116(1):3–16. doi: 10.1007/s00063-019-00613-x. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Hübener P, Braun G, Fuhrmann V.. [Acute-on-chronic liver failure: a diagnostic and therapeutic challenge for intensive care]. Med Klin Intensivmed Notfmed. 2018;113(8):649–657. doi: 10.1007/s00063-017-0263-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplemental Material IRNF_A_2656009_SM1881.docx (28.2KB, docx) Data Availability Statement The data that support the findings of this study are available from the corresponding author, JMK, upon reasonable request. The data are not publicly available due to ethical and data protection restrictions because they contain information that could compromise the privacy of research participants. Articles from Renal Failure are provided here courtesy of Taylor & Francis ACTIONS View on publisher site PDF (673.7 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top