Predictors of Mortality in Acute Kidney Injury Patients Admitted to Medicine Intensive Care Unit in a Rural Tertiary Care Hospital

An overview of circulating and urinary biomarkers capable of predicting the transition of acute kidney injury to chronic kidney disease

INTRODUCTION

Acute kidney injury (AKI) defined by a substantial decrease in kidney function within hours to days and is often irreversible with higher risk to chronic kidney disease (CKD) transition.

AREAS COVERED

The authors discuss the diagnostic and predictive utilities of serum and urinary biomarkers on AKI and on the risk of AKI-to-CKD progression. The authors focus on the relevant literature covering evidence of circulating and urinary biomarkers' capability to predict the transition of AKI to CKD.

EXPERT OPINION

Based on the different modalities of serum and urinary biomarkers, multiple biomarker panel seems to be potentially useful to distinguish between various types of AKI, to detect the severity and the risk of AKI progression, to predict the clinical outcome and evaluate response to the therapy. Serum/urinary neutrophil gelatinase-associated lipocalin (NGAL), serum/urinary uromodulin, serum extracellular high mobility group box-1 (HMGB-1), serum cystatin C and urinary liver-type fatty acid-binding protein (L-FABP) were the most effective in the prediction of AKI-to-CKD transition regardless of etiology and the presence of critical state in patients. The current clinical evidence on the risk assessments of AKI progression is mainly based on the utility of combination of functional, injury and stress biomarkers, mainly NGAL, L-FABP, HMGB-1 and cystatin C.

Characteristics, Risk Factors, and Outcomes in Acute Kidney Injury Patients: A Retrospective Cross-Sectional Study, Palestine

Background

Acute kidney injury (AKI) is a major medical problem affecting patients' quality of life and healthcare costs.

Objectives

This study evaluated the severity, risk factors, and outcomes of patients diagnosed with acute kidney injury (AKI), including community-acquired AKI (CA-AKI) and hospital-acquired AKI (HA-AKI), who were admitted to tertiary institutions in Palestine.

Methods

This retrospective cross-sectional study was conducted at multiple tertiary care hospitals in Palestine by reviewing patient charts from January 2020 to March 2023. The study included all patients aged ≥18 years who were admitted to the hospital and diagnosed with AKI at admission (CA-AKI) or who developed AKI 48 hours after admission (HA-AKI). Patients with incomplete medical records and those with no reported creatinine levels during their stay, pregnant women, kidney transplant patients, and end-stage renal disease patients were excluded. Data were analyzed using SPSS v22.0. The incidence of AKI in each group was compared using the chi-squared test.

Results

This study included 259 participants. HA-AKI was present in 27.3% of the patients, while CA-AKI was 72.7%. The most common stage among patients was stage 3 (55.7%, HA-AKI) (42.9%, CA-AKI), and the most common comorbidity contributing to AKI was CKD. NSAIDs, ACE-I/ARBs, and DIURETICs were the most nephrotoxic drugs contributing to AKI. Patients with hyperphosphatemia, hyperkalemia, severe metabolic acidosis, or stage 3 AKI require renal replacement therapy. In addition, our findings revealed a significant association among AKI mortality, age, and heart disease.

Conclusion

CA-AKI was more prevalent than HA-AKI in Palestinian patients admitted for AKI. Risk factors for AKI included diabetes, CKD, and medications (antibiotics, NSAID, diuretics, and ACE-I/ARB). Preventive measures, medication management, and disease state management are necessary to minimize AKI during hospital admission or in the community.

Incidence, Outcomes, and Predictors of Subphenotypes of Acute Kidney Injury among Acute Respiratory Distress Syndrome Patients: A Prospective Observational Study

Background

Acute kidney injury (AKI) is a heterogeneous syndrome with subphenotypes. Acute kidney injury is one of the most common complications in acute respiratory distress syndrome (ARDS) patients, which influences mortality.

Material and methods

It was a single-center observational study on 266 ARDS patients on invasive mechanical ventilation (IMV) to determine the subphenotypes of AKI associated with ARDS. Subphenotyping was done based on the serum creatinine (SCr) trajectories from day 1 to day 5 of IMV into resolving (subphenotype 1) or non-resolving (subphenotype 2) AKI.

Results

Out of 266 ARDS patients, 222 patients were included for data analysis. 141 patients (63.51%) had AKI. The incidence of subphenotype 2 AKI among the ARDS cohort was 78/222 (35.13%). Subphenotype 2 AKI was significantly more among the non-survivors (87.7% vs 36.2 %, p < 0.001). Subphenotype 2 AKI was an independent predictor of mortality among ARDS patients (p < 0.001, adjusted odds ratio 8.978, 95% CI [2.790-28.89]. AKI subphenotype 1 had higher median day 1 SCr than subphenotype 2 but lower levels by day 3 and day 5 of IMV. The median time of survival was 8 days in AKI subphenotype 2 vs 45 days in AKI with subphenotype 1 [Log-Rank (Mantel-Cox) p < 0.001]. The novel DRONE score (Driving pressure, Oxygenation, and Nutritional Evaluation) ≥ 4 predicted subphenotype 2 AKI.

Conclusion

The incidence of subphenotype 2 (non-resolving) AKI among ARDS patients on IMV was about 35% (vs 20% subphenotype 1 AKI), and it was an independent predictor of mortality. The DRONE score ≥4 can predict the AKI subphenotype 2.

Highlights

The serum creatinine trajectory-based subphenotype of AKI (resolving vs non-resolving) determines survival in ARDS patients. Non-resolving AKI subphenotype 2 is an independent predictor of mortality in ARDS. The novel DRONE score (driving pressure, oxygenation, and nutritional evaluation) ≥ 4 within 48 hours of IMV predicted the AKI subphenotype 2 among ventilated ARDS patients.

How to cite this article

Todur P, Nileshwar A, Chaudhuri S, Srinivas T. Incidence, Outcomes, and Predictors of Subphenotypes of Acute Kidney Injury among Acute Respiratory Distress Syndrome Patients: A Prospective Observational Study. Indian J Crit Care Med 2023;27(10):724-731.

Spectrum, Outcomes, and Mortality Predictors of Acute Kidney Injury among Non-COVID-19 Patients during COVID-19 Pandemic: Data from Four Intensive Care Units

Introduction

The data of acute kidney injury (AKI), that is, community-acquired AKI (CA-AKI) and hospital-acquired AKI (HA-AKI) among non-COVID patients from intensive care units (ICU) during the coronavirus disease-2019 (COVID-19) pandemic are scarce. We planned to study the change in the profile of such patients compared to the pre-pandemic era.

Materials and methods

This prospective observational study was conducted at four ICUs dealing with non-COVID patients at a government hospital in North India, and was aimed at assessing outcomes, and mortality predictors of AKI among non-COVID patients during the COVID-19 pandemic. Renal and patient survival at ICU transfer-out and hospital discharge, ICU and hospital stay duration, mortality predictors, and dialysis requirement at discharge were evaluated. The current or previous COVID-19 infection, previous AKI or chronic kidney disease (CKD), organ donors, and organ transplant patients were excluded.

Results

Among the 200 non-COVID-19 AKI patients, diabetes mellitus (DM), primary hypertension, and cardiovascular diseases were the predominant comorbidities in descending order. The commonest cause of AKI was severe sepsis, followed by systemic infections and post-surgery patients. Dialysis requirements at ICU admission during ICU stay and above 30 days were seen in 20.5, 47.5, and 6.5% of patients, respectively. Incidence of CA-AKI and HA-AKI was 1.24:1, whereas dialysis requirement above 30 days was 0.85:1, respectively. The 30-day mortality was 42%. Hepatic dysfunction [hazard ratio (HR): 3.471], septicemia (HR: 3.342), age above 60 years (HR: 4.000), higher sequential organ failure assessment (SOFA) score (HR: 1.107; p = 0.001), anemia (p = 0.003), and low serum iron (p = 0.001) were important mortality predictors in AKI.

Conclusion

Compared to the pre-COVID era, CA-AKI was more common than HA-AKI due to restricted elective surgeries during the COVID-19 pandemic. Acute kidney injury with multiorgan involvement and hepatic dysfunction, elderly age with higher SOFA score and sepsis were predictors of adverse renal and patient outcomes.

How to cite this article

Singh B, Dogra PM, Sood V, Singh V, Katyal A, Dhawan M, et al. Spectrum, Outcomes, and Mortality Predictors of Acute Kidney Injury among Non-COVID-19 Patients during COVID-19 Pandemic: Data from Four Intensive Care Units. Indian J Crit Care Med 2023;27(2):119-126.

Comparison of Outcomes of Mild and Severe Community- and Hospital-Acquired Acute Kidney Injury

PURPOSE

Acute kidney injury (AKI) has shown an increasingly common occurrence among hospitalized patients worldwide. We determined the incidence and compared the short- and long-term outcomes of all stages of community-acquired AKI (CA-AKI) and hospital-acquired AKI (HA-AKI), and identified predictors for such outcomes.

MATERIALS AND METHODS

This observational, single-center, retrospective study identified patients admitted between January 2013 and December 2013 who developed CA-AKI or HA-AKI. Short- and long-term patient and renal outcomes were analyzed.

RESULTS

AKI incidence was 14.3% (1882, CA-AKI 4.8% and HA-AKI 9.5%). The highest 30-day and 1-year mortality were recorded in the CA-AKI group. Thirty-day mortality rate was 11.4% in CA-AKI group and 5.7% in HA-AKI group (p<0.001). One-year mortality rates were 20.1% and 13.3%, respectively (p<0.001). More CA-AKI patients developed kidney failure with replacement therapy within 1 year (27, 4.3% vs. 18, 1.4% respectively, p<0.001).

CONCLUSION

In conclusion, patients with CA-AKI had worse short- and long-term outcomes compared to HA-AKI patients. AKI severity and discharge serum creatinine were significant independent predictors of 30-day and 1-year mortality.

Burn-Induced Acute Kidney Injury-Two-Lane Road: From Molecular to Clinical Aspects

Severe burn injuries lead to acute kidney injury (AKI) development, increasing the mortality risk up to 28-100%. In addition, there is an increase in hospitalization days and complications appearance. Various factors are responsible for acute or late AKI debut, like hypovolemia, important inflammatory response, excessive load of denatured proteins, sepsis, and severe organic dysfunction. The main measure to improve the prognosis of these patients is rapidly recognizing this condition and reversing the underlying events. For this reason, different renal biomarkers have been studied over the years for early identification of burn-induced AKI, like neutrophil gelatinase-associated lipocalin (NGAL), cystatin C, kidney injury molecule-1 (KIM-1), tissue inhibitor of metalloproteinase-2 (TIMP-2), interleukin-18 (IL-18), and insulin-like growth factor-binding protein 7 (IGFBP7). The fundamental purpose of these studies is to find a way to recognize and prevent acute renal injury progression early in order to decrease the risk of mortality and chronic kidney disease (CKD) onset.

Assessment of APACHE II Score to Predict ICU Outcomes of Patients with AKI: A Single Center Experience from Haryana, North India

Background

The acute physiology and chronic health evaluation (APACHE) II severity score has shown a good calibration and discriminatory value across a range of disease processes and remains the most widely used source of prognostic information on the risk for death in patients admitted to intensive care units (ICUs).

Objectives

To study APACHE II scores in patients of acute kidney injury (AKI) admitted in the ICU and to find its association with outcome.

Materials and methods

One hundred patients with AKI aged 18 years or above were admitted in the ICU, department of general medicine, of a tertiary care institute in Haryana, North India, from October 2019 to September 2020, were studied. Patients who had known causes of chronic kidney disease (CKD), on maintenance hemodialysis, and those who underwent renal replacement therapy (RRT) or nephrectomy were excluded. All required investigations were performed, and data were collected. The patients were followed till discharge or in-hospital mortality.

Results

The mean age of the patients was 55.92 ± 18.18 years. Male–female ratio was 1.5:1. Thirty-five percent of the admitted patients had an in-hospital mortality. Sepsis (47%) was the most common cause of AKI, and 83% of the patients had underlying comorbid conditions. The mean APACHE II score of the expired patients on admission, i.e., 24.80 ± 13.65, was found to be significantly higher compared to the mean APACHE II score (17.25 ± 10.12) of the discharged patients (p-value <0.001). APACHE II score was found to have 57.14% sensitivity, 86.15% specificity, 69% PPV, 78.9% NPV, and 76% diagnostic accuracy to predict mortality among the AKI patients.

Conclusion

APACHE II scoring system has a good discrimination and calibration when applied to ICU-admitted AKI patients and is a good predictor of prognosis in them.

How to cite this article

Patel P, Gupta S, Patel H, Bashar MDA. Assessment of APACHE II Score to Predict ICU Outcomes of Patients with AKI: A Single-center Experience from Haryana, North India. Indian J Crit Care Med 2022;26(3):276–281.

Conceptual model of adding antibiotics to dialysate fluid during renal replacement therapy

Studies have shown significant variability in antibiotic trough concentrations in critically ill patients receiving renal replacement therapy (RRT). The purpose of this study was to assess whether adding beta-lactam antibiotics to dialysate solution can maintain stable antibiotic concentrations during RRT in experimental conditions. A single compartment model reflecting the patient was constructed and connected to the RRT machine. Dialysate fluid was prepared in three different concentrations of meropenem (0 mg/L; 16 mg/L; 64 mg/L). For each dialysate concentration various combinations of dialysate and blood flow rates were tested by taking different samples. Meropenem concentration in all samples was calculated using spectrophotometry method. Constructed experimental model results suggest that decrease in blood meropenem concentration can be up to 35.6%. Moreover, experimental data showed that antibiotic loss during RRT can be minimized and stable plasma antibiotic concentration can be achieved with the use of a 16 mg/L Meropenem dialysate solution. Furthermore, increasing meropenem concentration up to 64 mg/L is associated with an increase antibiotic concentration up to 18.7–78.8%. Administration of antibiotics to dialysate solutions may be an effective method of ensuring a constant concentration of antibiotics in the blood of critically ill patients receiving RRT.

Mortality and predictors of acute kidney injury in adults: a hospital-based prospective observational study

Acute kidney injury (AKI) is a major global public health problem. It is expensive to manage and associated with a high rate of prolonged hospitalization and in-hospital mortality. Little is known about the burden of acute kidney injury in moderate to low-income countries. We aim to assess predictors of in-hospital mortality among AKI patients admitted to the medical ward. We prospectively identified patients meeting kidney disease improving global outcomes (KIDGO) AKI definitions from April to August 2019. Patients with underlying CKD and patients hospitalized for less than 48 h were excluded. The Cox regression model was fitted to identify predictors of mortality and statistical significance was considered at the p-value of less than 0.05. A total of 203 patients were enrolled over 5 months. Out of this, 121(59.6%) were males, 58(28.6%) were aged greater than 60 years, and 141(69.5%) had community-acquired acute kidney injury. The most common causes of AKI were Hypovolemia 99(48.77%), Glomerulonephritis 51(25.11%), and sepsis 32(15.79%). The overall in-hospital mortality rate was 12.8%. Stage 3 AKI (AHR = 9.61, 95% CI 1.17-28.52, p = 0.035), duration of AKI (AHR = 7.04, 95% CI 1.37-36.08, p = 0.019), length of hospital stay (AHR = 0.19, 95% CI 0.05-0.73, p = 0.012), and hyperkalemia (AHR = 3.61, 95% CI 1.12-11.71, p = 0.032) were significantly associated with in-hospital mortality. There is a high rate of acute kidney injury-related in-hospital mortality in adult patients admitted to the medical ward. The severity of AKI, hyperkalemia duration of AKI, and a short length of hospital stay were predictors of 30-days in-hospital mortality. Most of the causes of AKI are preventable and patients may benefit from early identification and treatment of these reversible causes.

Reduced alternative insulin dosing in hyperkalemia: A meta-analysis of effects on hypoglycemia and potassium reduction

STUDY OBJECTIVE

Recent studies have identified that reduced alternative intravenous insulin doses, such as 5 units or 0.1 units/kg, may reduce the risk of hypoglycemia compared to standard doses of 10 units in patients treated for hyperkalemia. However, some studies suggest that these alternative doses may reduce the ability to lower serum potassium. This study was performed to determine the impact of alternative insulin dosing on hypoglycemia and potassium reduction in patients with hyperkalemia.

DESIGN

Meta-analysis.

DATA SOURCE

PubMed/MEDLINE, CENTRAL, Ovid, and ClinicalTrials.gov were searched from inception through November 2020.

PATIENTS

Patients treated with standard (10 units) or alternative (<10 units) insulin dosing strategies for hyperkalemia. Only studies that evaluated hypoglycemia (serum glucose <70 mg/dl), severe hypoglycemia (serum glucose <50 mg/dl), and potassium reduction post-treatment were included in the meta-analysis. All articles were assessed for bias using the Cochrane Risk of Bias Assessment Tool and Newcastle-Ottawa scales for randomized prospective trials and retrospective trials, respectively.

INTERVENTION

None.

MEASUREMENTS AND MAIN RESULTS

Ten retrospective cohort studies (n = 3437) were included and had low- or moderate-risk of bias. Alternative insulin dosing strategies included 5 units, 0.1 units/kg, and <10 units. Alternative dosing had lower pooled odds of hypoglycemia (odds ratio [OR] 0.55, 95% confidence interval [CI] 0.43-0.69, I²  = 8%) and severe hypoglycemia (OR 0.41, 95% CI 0.27-0.64, I²  = 0%). No difference in potassium reduction was detected (mean difference -0.02 mmol/L, 95% CI -0.11-0.07, I²  = 53%).

CONCLUSIONS

Alternative insulin dosing strategies for hyperkalemia management resulted in less hypoglycemia and severe hypoglycemia without compromising potassium reduction compared to standard dose. Prospective studies are needed to confirm these findings.

Acute kidney injury is an independent predictor of in-hospital mortality in a general medical ward: A retrospective study from a tertiary care centre in south India

Background & objectives:

Acute kidney injury (AKI) has been identified as an independent risk factor for mortality in intensive care units. This retrospective study was conducted to determine the effect of AKI on in-hospital mortality in a general medical ward of a tertiary care hospital and risk factors for mortality in patients with AKI.

Methods:

Demographic and clinical details, and outcome data were collected from case records of patients. In all hospitalized patients, factors associated with increased in-hospital mortality, such as AKI, inotrope requirement, mechanical ventilation and primary disease were compared between patients who died during their hospital stay and those who were discharged alive. Among the hospitalized patients, who had AKI, likely predictors of in-hospital mortality were compared between dead and alive patients at discharge. Factors that were significant in univariate analysis were tested by multivariate regression analysis to identify those that independently predicted poor outcomes.

Results:

Of the 1150 patients admitted in a general medical ward in a year, 220 patients were identified to have AKI. In-hospital mortality rate among patients with AKI was 19.09 per cent as compared to 1.8 per cent without AKI [adjusted odds ratio (OR) 5.7 (95% CI: 1.56-20.74)]. The presence of AKI was an independent risk factor for death, with an adjusted OR of 6.0 [95% CI: 1.67-21.6]. Among patients with AKI, the presence of haematological malignancy, adjusted OR 25.86 (95% CI: 1.81-369.58), requirement of inotrope, adjusted OR 126.5 (95% CI: 19.39-825.93) and serum creatinine at admission (P<0.001) were found to be independent predictors of death. The presence of underlying chronic kidney disease and hospital-acquired AKI were not found to have an association with mortality.

Interpretation & conclusions:

The study showed the in-hospital mortality rate among patients with AKI in a general medical ward was 19.09 per cent. The occurrence of AKI was an independent risk factor for death, with haematological malignancy, use of vasopressors and higher serum creatinine at admission, significantly associated with death among patients with AKI. Large prospective studies need to be done to better understand the outcomes in AKI and the ways to present and manage AKI.

Performance of Severity Indices to Estimate Postoperative Complications of Myocardial Revascularization

Fundamento

Os pacientes em pós-operatório (PO) de cirurgia de revascularização miocárdica (CRM) internados em unidade de terapia intensiva (UTI) apresentam risco de complicações que aumentam o tempo de permanência e a morbimortalidade. Portanto, é fundamental o reconhecimento precoce desses riscos para otimizar estratégias de prevenção e desfecho clínico satisfatório.

Objetivo

Analisar o desempenho de índices de gravidade na predição de complicações em pacientes no PO de CRM durante a permanência na UTI.

Métodos

Estudo transversal, com análise retrospectiva de prontuários eletrônicos de pacientes com idade ≥ 18 anos submetidos à CRM isolada e admitidos na UTI de um hospital cardiológico, em São Paulo, Brasil. As áreas sob as curvas receiver operating characteristic (AUC) com intervalo de confiança de 95% foram analisadas para verificar a acurácia dos índices European System for Cardiac Operative Risk Evaluation (EuroScore), Acute Physiology and Chronic Health Evaluation (APACHE II), Simplified Acute Physiology Score (SAPS II) e Sequential Organ Failure Assessment (SOFA) na predição de complicações.

Resultados

A casuística foi composta por 366 pacientes (64,58±9,42 anos; 75,96% sexo masculino). As complicações identificadas foram respiratórias (24,32%), cardiológicas (19,95%), neurológicas (10,38%), hematológicas (10,38%), infecciosas (6,56%) e renais (3,55%). O APACHE II apresentou satisfatório desempenho para a predição de complicações neurológicas (AUC 0,72) e renais (AUC 0,78).

Conclusão

O APACHE II se destacou na previsão das complicações neurológicas e renais. Nenhum dos índices teve bom desempenho na predição das outras complicações analisadas. Portanto, os índices de gravidade não devem ser utilizados indiscriminadamente com o objetivo de predizer todas as complicações frequentemente apresentadas por pacientes após CRM. (Arq Bras Cardiol. 2020; 115(3):452-459)

Pregnancy-related acute kidney injury: mortality and survival of patients treated at a maternal intensive care unit

INTRODUCTION

Pregnancy-related acute kidney injury (AKI) can be defined as the abrupt decline in renal function during pregnancy or the postpartum period. It remains a relevant cause of fatal complications in obstetric patients. This study aimed to determine the incidence of pregnancy-related AKI in a maternal intensive care unit (ICU) as well as the associated risk factors for dialysis therapy and maternal mortality according to the KDIGO classification system.

METHODS

Retrospective analysis of observational data prospectively collected from January/2014 to April/2016 in a maternal ICU in a public tertiary maternal hospital in Brasília, Federal District, Brazil. All consecutive patients diagnosed with pregnancy-related AKI were included. Cases of renal failure before pregnancy or kidney transplantation were excluded.

RESULTS

Of the 619 admitted patients, pregnancy-related AKI was present in 172 cases (27.8%). One hundred and ten patients were classified as KDIGO 1 (64.0%), 43 as KDIGO 2 (20.9%) and 22 as KDIGO 3 (15.1%). KDIGO stages 2 and 3 were less frequent than KDIGO stage 1 in patients with gestational hypertension (p = 0.0010). Thirteen patients required hemodialysis (7.6%). Higher APACHE II (p = 0.0399) and SOFA (p = 0.0297) scores, hypovolemic shock (p = 0.0189) and septic shock (p = 0.0204) were independently associated with dialysis therapy (hemodialysis in all cases), 15 patients died (8.7%). Time to death was shorter in patients with a higher KDIGO stage (p = 0.002). Norepinephrine (p = 0.0384) and hemodialysis therapy (p = 0.0128) were independently associated with maternal mortality.

CONCLUSIONS

The incidence of pregnancy-related AKI remains high in the maternal ICU setting. Septic shock, hypovolemic shock, and higher APACHE II and SOFA scores were independently associated dialysis therapy (hemodialysis in all cases). KDIGO stages 2 and 3 were less frequent than KDIGO stage 1 in patients with gestational hypertension. Norepinephrine and hemodialysis therapy were independently associated with maternal mortality in patients with pregnancy-related AKI. KDIGO stage 3 was associated with higher maternal mortality.

Epidemiology and Pathogenesis of Acute Kidney Injury in the Critically Ill Patients

How to cite this article: Kulkarni AP, Bhosale SJ. Epidemiology and Pathogenesis of Acute Kidney Injury in the Critically Ill Patients. Indian J Crit Care Med 2020;24(Suppl 3):S84–S89.

Machine learning for the prediction of volume responsiveness in patients with oliguric acute kidney injury in critical care

BACKGROUND AND OBJECTIVES

Excess fluid balance in acute kidney injury (AKI) may be harmful, and conversely, some patients may respond to fluid challenges. This study aimed to develop a prediction model that can be used to differentiate between volume-responsive (VR) and volume-unresponsive (VU) AKI.

METHODS

AKI patients with urine output < 0.5 ml/kg/h for the first 6 h after ICU admission and fluid intake > 5 l in the following 6 h in the US-based critical care database (Medical Information Mart for Intensive Care (MIMIC-III)) were considered. Patients who received diuretics and renal replacement on day 1 were excluded. Two predictive models, using either machine learning extreme gradient boosting (XGBoost) or logistic regression, were developed to predict urine output > 0.65 ml/kg/h during 18 h succeeding the initial 6 h for assessing oliguria. Established models were assessed by using out-of-sample validation. The whole sample was split into training and testing samples by the ratio of 3:1.

MAIN RESULTS

Of the 6682 patients included in the analysis, 2456 (36.8%) patients were volume responsive with an increase in urine output after receiving > 5 l fluid. Urinary creatinine, blood urea nitrogen (BUN), age, and albumin were the important predictors of VR. The machine learning XGBoost model outperformed the traditional logistic regression model in differentiating between the VR and VU groups (AU-ROC, 0.860; 95% CI, 0.842 to 0.878 vs. 0.728; 95% CI 0.703 to 0.753, respectively).

CONCLUSIONS

The XGBoost model was able to differentiate between patients who would and would not respond to fluid intake in urine output better than a traditional logistic regression model. This result suggests that machine learning techniques have the potential to improve the development and validation of predictive modeling in critical care research.

Predictors of Mortality in Acute Kidney Injury Patients in an Intensive Care Unit: Is Hemodialysis No Good?

How to cite this article: Karim HMR, Panda CK. Predictors of Mortality in Acute Kidney Injury Patients in an Intensive Care Unit: Is Hemodialysis No Good? Indian J Crit Care Med 2019;23(9):439.