Long-term Outcomes and Prognostic Factors for Patients Requiring Renal Replacement Therapy After Cardiac Surgery

Recovery after AKI: Effects on outcomes over 15 years

PURPOSE

To examine the effect of kidney recovery on mortality, dialysis and kidney transplantation up to 15 years after AKI.

MATERIALS AND METHODS

We studied 29,726 survivors of critical illness and compared these outcomes stratified by AKI and recovery status at hospital discharge. Kidney recovery was defined as a return of serum creatinine to ≤150% of baseline without dialysis prior to hospital discharge.

RESULTS

Overall AKI occurred in 59.2% in which two thirds developed stage 2-3 AKI. Recovery rate of AKI at hospital discharge was 80.8%. Patients who did not recover experienced the worst 15-year mortality compared to those who recovered and those without AKI (57.8% vs 45.2% vs 30.3%, p < 0.001). This pattern was also found in subgroups of patients with suspected sepsis-associated (57.1% vs 47.9% vs 36.5%, p < 0.001) and cardiac surgery-associated AKI (60.1% vs 41.8% vs 25.9%, p < 0.001). The rates of dialysis and transplantation at 15 years were low and not associated with recovery status.

CONCLUSIONS

Recovery of AKI in critically ill patients at hospital discharge had an effect on long-term mortality for up to 15 years. These results have implications for acute care, follow-up and choice of endpoints for clinical trials.

Explainable Preoperative Automated Machine Learning Prediction Model for Cardiac Surgery-Associated Acute Kidney Injury

BACKGROUND

We aimed to develop and validate an automated machine learning (autoML) prediction model for cardiac surgery-associated acute kidney injury (CSA-AKI).

METHODS

Using 69 preoperative variables, we developed several models to predict post-operative AKI in adult patients undergoing cardiac surgery. Models included autoML and non-autoML types, including decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost), and artificial neural network (ANN), as well as a logistic regression prediction model. We then compared model performance using area under the receiver operating characteristic curve (AUROC) and assessed model calibration using Brier score on the independent testing dataset.

RESULTS

The incidence of CSA-AKI was 36%. Stacked ensemble autoML had the highest predictive performance among autoML models, and was chosen for comparison with other non-autoML and multivariable logistic regression models. The autoML had the highest AUROC (0.79), followed by RF (0.78), XGBoost (0.77), multivariable logistic regression (0.77), ANN (0.75), and DT (0.64). The autoML had comparable AUROC with RF and outperformed the other models. The autoML was well-calibrated. The Brier score for autoML, RF, DT, XGBoost, ANN, and multivariable logistic regression was 0.18, 0.18, 0.21, 0.19, 0.19, and 0.18, respectively. We applied SHAP and LIME algorithms to our autoML prediction model to extract an explanation of the variables that drive patient-specific predictions of CSA-AKI.

CONCLUSION

We were able to present a preoperative autoML prediction model for CSA-AKI that provided high predictive performance that was comparable to RF and superior to other ML and multivariable logistic regression models. The novel approaches of the proposed explainable preoperative autoML prediction model for CSA-AKI may guide clinicians in advancing individualized medicine plans for patients under cardiac surgery.

Development and Validation of a Web-Based Prediction Model for AKI after Surgery

BACKGROUND

AKI after surgery is associated with high mortality and morbidity. The purpose of this study is to develop and validate a risk prediction tool for the occurrence of postoperative AKI requiring RRT (AKI-dialysis).

METHODS

This retrospective cohort study had 2,299,502 surgical patients over 2015-2017 from the American College of Surgeons National Surgical Quality Improvement Program Database (ACS NSQIP). Eleven predictors were selected for the predictive model: age, history of congestive heart failure, diabetes, ascites, emergency surgery, hypertension requiring medication, preoperative serum creatinine, hematocrit, sodium, preoperative sepsis, and surgery type. The predictive model was trained using 2015-2016 data (n=1,487,724) and further tested using 2017 data (n=811,778). A risk model was developed using multivariable logistic regression.

RESULTS

AKI-dialysis occurred in 0.3% (n=6853) of patients. The unadjusted 30-day postoperative mortality rate associated with AKI-dialysis was 37.5%. The AKI risk prediction model had high area under the receiver operating characteristic curve (AUC; training cohort: 0.89, test cohort: 0.90) for postoperative AKI-dialysis.

CONCLUSIONS

This model provides a clinically useful bedside predictive tool for postoperative AKI requiring dialysis.

Copeptin concentration following cardiac surgery as a prognostic marker of postoperative acute kidney injury: a prospective cohort study

Background

Copeptin, the C-terminal portion of the arginine vasopressin precursor, is a novel candidate biomarker. This study investigated the prognostic value of copeptin levels following cardiac surgery for the occurrence of postoperative acute kidney injury.

Methods

We studied 23 patients who underwent cardiac surgery between January 2018 and December 2019. The primary endpoint was postoperative acute kidney injury onset. Copeptin levels were measured before, right after, and daily for 7 days. The patients were divided into two groups according to the copeptin levels: low (values <43.7 pmol/L) and high (values ≥43.7 pmol/L). Correlations between copeptin levels and variables, such as central venous pressure, were assessed by bivariate analysis.

Results

The high copeptin group exhibited significantly higher levels of arginine vasopressin and cortisol following surgery, compared to those of the low copeptin group. The copeptin concentration following surgery was correlated to central venous pressure (P=0.03) and norepinephrine administered dose (P=0.008). Also, the copeptin levels right after surgery robustly predicted the onset of postoperative acute kidney injury (area under the receiver operating characteristic curve of 0.83, P=0.004).

Conclusions

Elevated copeptin levels in patients following cardiac surgery predicted postoperative acute kidney injury development. Therefore, the copeptin concentration after surgery could represent a promising clinical biomarker of the postoperative cardiac outcome.

Predicting Acute Kidney Injury after Cardiac Surgery by Machine Learning Approaches

Cardiac surgery-associated AKI (CSA-AKI) is common after cardiac surgery and has an adverse impact on short- and long-term mortality. Early identification of patients at high risk of CSA-AKI by applying risk prediction models allows clinicians to closely monitor these patients and initiate effective preventive and therapeutic approaches to lessen the incidence of AKI. Several risk prediction models and risk assessment scores have been developed for CSA-AKI. However, the definition of AKI and the variables utilized in these risk scores differ, making general utility complex. Recently, the utility of artificial intelligence coupled with machine learning, has generated much interest and many studies in clinical medicine, including CSA-AKI. In this article, we discussed the evolution of models established by machine learning approaches to predict CSA-AKI.

The Clinical Course of Acute Kidney Disease after Cardiac Surgery: A Retrospective Observational Study

Acute kidney disease (AKD), or renal dysfunction persisting >7 days after an initiating event of acute kidney injury, is a rising concern. This study aimed to elucidate the clinical course of AKD after cardiac surgery with data on post-cardiac surgery patients admitted to intensive care units (ICU) at 18 Japanese hospitals during 2012–2014. Using multivariable logistic models, we evaluated the association of AKD with 90-day mortality and the 50% eGFR decline during 2-year follow-up compared to eGFR at 90 days. AKD was defined as an elevation in serum creatinine to at least 1.5-fold from baseline in >7 days after ICU admission. Of the 3,605 eligible patients undergoing cardiac surgery, 403 patients (11.2%) had AKD. Multivariable analysis revealed that the adjusted odds ratio (OR) of AKD for 90-day mortality was 63.0 (95% confidence interval [CI], 27.9–180.6). In addition, the adjusted OR of AKD for 50% eGFR decline was 3.56 (95% CI, 2.24–5.57) among hospital survivors. In conclusion, AKD after cardiac surgery was associated with higher 90-day mortality and renal function decline after hospital discharge.

Diagnostics, Risk Factors, Treatment and Outcomes of Acute Kidney Injury in a New Paradigm

Acute kidney injury (AKI) is a common clinical condition among patients admitted in the hospitals. The condition is associated with both increased short-term and long-term mortality. With the development of a standardized definition for AKI and the acknowledgment of the impact of AKI on patient outcomes, there has been increased recognition of AKI. Two advances from past decades, the usage of computer decision support and the discovery of AKI biomarkers, have the ability to advance the diagnostic method to and further management of AKI. The increasingly widespread use of electronic health records across hospitals has substantially increased the amount of data available to investigators and has shown promise in advancing AKI research. In addition, progress in the finding and validation of different forms of biomarkers of AKI within diversified clinical environments and has provided information and insight on testing, etiology and further prognosis of AKI, leading to future of precision and personalized approach to AKI management. In this this article, we discussed the changing paradigms in AKI: from mechanisms to diagnostics, risk factors, and management of AKI.

Stage 1 acute kidney injury is independently associated with infection following cardiac surgery

OBJECTIVES

Severe acute kidney injury (AKI) is a known risk factor for infection and mortality. However, whether stage 1 AKI is a risk factor for infection has not been evaluated in adults. We hypothesized that stage 1 AKI following cardiac surgery would independently associate with infection and mortality.

METHODS

In this retrospective propensity score-matched study, we evaluated 1620 adult patients who underwent nonemergent cardiac surgery at the University of Colorado Hospital from 2011 to 2017. Patients who developed stage 1 AKI by Kidney Disease Improving Global Outcomes creatinine criteria within 72 hours of surgery were matched to patients who did not develop AKI. The primary outcome was an infection, defined as a new surgical-site infection, positive blood or urine culture, or development of pneumonia. Secondary outcomes included in-hospital mortality, stroke, and intensive care unit (ICU) and hospital length of stay (LOS).

RESULTS

Stage 1 AKI occurred in 293 patients (18.3%). Infection occurred in 20.9% of patients with stage 1 AKI compared with 8.1% in the no-AKI group (P < .001). In propensity-score matched analysis, stage 1 AKI independently associated with increased infection (odds ratio [OR]; 2.24, 95% confidence interval [CI], 1.37-3.17), ICU LOS (OR, 2.38; 95% CI, 1.71-3.31), and hospital LOS (OR, 1.30; 95% CI, 1.17-1.45).

CONCLUSIONS

Stage 1 AKI is independently associated with postoperative infection, ICU LOS, and hospital LOS. Treatment strategies focused on prevention, early recognition, and optimal medical management of AKI may decrease significant postoperative morbidity.

Impact of Early versus Late Initiation of Renal Replacement Therapy in Patients with Cardiac Surgery-Associated Acute Kidney Injury: Meta-Analysis with Trial Sequential Analysis of Randomized Controlled Trials

Background

Previous studies have examined the effect of the initiation time of renal replacement therapy (RRT) in patients with cardiac surgery-associated acute kidney injury (CSA-AKI), but the findings remain controversial. The aim of this meta-analysis was to systematically and quantitatively compare the impact of early versus late initiation of RRT on the outcome of patients with CSA-AKI.

Methods

Four databases (PubMed, the Cochrane Library, ISI Web of Knowledge, and Embase) were systematically searched from inception to June 2018 for randomized clinical trials (RCTs). Two investigators independently performed the literature search, study selection, data extraction, and quality evaluation. Meta-analysis and trial sequential analysis (TSA) were used to examine the impact of RRT initiation time on all-cause mortality (primary outcome). The Grading of Recommendations Assessment Development and Evaluation (GRADE) was used to evaluate the level of evidence.

Results

We identified 4 RCTs with 355 patients that were eligible for inclusion. Pooled analyses indicated no difference in mortality for patients receiving early and late initiation of RRT (relative risk [RR] = 0.61, 95% confidence interval [CI] = 0.33 to 1.12). However, the results were not confirmed by TSA. Similarly, early RRT did not reduce the length of stay (LOS) in the intensive care unit (ICU) (mean difference [MD] = -1.04; 95% CI = -3.34 to 1.27) or the LOS in the hospital (MD = -1.57; 95% CI = -4.62 to 1.48). Analysis using GRADE indicated the certainty of the body of evidence was very low for a benefit from early initiation of RRT.

Conclusion

Early initiation of RRT had no beneficial impacts on outcomes in patients with CSA-AKI. Future larger and more adequately powered prospective RCTs are needed to verify the benefit of reduced mortality associated with early initiation of RRT.

Trial Registration

This trial is registered with PROSPERO registration number CRD42018084465, registered on 11 February 2018.

Perioperative body weight change is associated with in-hospital mortality in cardiac surgical patients with postoperative acute kidney injury

Background

Postoperative acute kidney injury (AKI) is common following cardiac surgery (CS). Body weight (BW) may be an amenable variable by representing the summation of the nutritional and the fluid status. However, the predictive role of perioperative BW changes in CS patients with severe postoperative AKI is never explored. This study aimed to evaluate this association.

Methods

This study was conducted using a prospectively collected multicenter cohort, NSARF (National Taiwan University Hospital Study Group on Acute Renal Failure) database. The adult CS patients with postoperative AKI requiring renal replacement therapy (RRT), who had clear initial consciousness, received CS within 14 days of hospitalization, and underwent RRT within seven days after CS in intensive care units from January 2001 to January 2014 were enrolled. With the endpoint of 30-day postoperative mortality, we evaluated the association between the clinical factors denoting fluid status and patients outcomes.

Results

A total of 188 patients (70 female, mean age 63.7 ± 15.2 years) were enrolled. Comparing with the survivors (n = 124), the non-survivors (n = 64) had a significantly higher perioperative BW change [3.6 ± 6.1% versus 0.1 ± 8.3%, p = 0.003] but not the postoperative and pre-RRT BW changes. By using multivariate Cox proportional hazards model, the independent indicators of 30-day postoperative mortality included perioperative BW change (p = 0.026) and packed red blood cells transfusion (p = 0.007), postoperative intra-aortic balloon pump (p = 0.001) and central venous pressure level (p = 0.005), as well as heart rate (p = 0.022), sequential organ failure assessment score (p < 0.001), logistic organ dysfunction score (p = 0.001), and blood total bilirubin level (p = 0.044) at RRT initiation. The generalized additive models further demonstrated, in a multivariate manner, that the mortality risk rose significantly during a perioperative BW change of 2% to 15%.

Conclusions

Perioperative BW change was independently associated with an increased risk for 30-day postoperative mortality in CS patients with RRT-requiring AKI.

Effects of continuous renal replacement therapy on serum cytokines, neutrophil gelatinase-associated lipocalin, and prognosis in patients with severe acute kidney injury after cardiac surgery

The aim of our study was to evaluate the effect of continuous renal replacement therapy (CRRT) on serum cytokines, neutrophil gelatinase-associated lipocalin (NGAL), and prognosis in patients with severe acute kidney injury (AKI) following cardiac surgery. A total number of 153 patients with severe AKI following cardiac surgery were treated with CRRT. They were divided into the survival and non-survival groups. Clinical data from these two groups before and after CRRT were recorded and analyzed. It was found that the number of impaired organs, MODS and APACHE II scores were significantly higher in the non-survival group than those in the survival group before CRRT. After CRRT, MODS and APACHE II scores decreased significantly. The post-CRRT levels of serum TNF-a and IL-6 were significantly decreased. After CRRT, serum NGAL decreased in the two groups, but the levels were higher in the non-survival group than those in the survival group. MODS and APACHE II scores could be used to evaluate the severity of AKI in patients after cardiac surgery. CRRT is an effective treatment for these patients and high levels of TNF-a, IL-6, and NGAL are associated with a poor prognosis in these patients.

Outcomes of renal function in elderly patients with acute kidney injury

OBJECTIVES

The aim of this study was to explore the prognostic impact of clinical factors on the short-term outcomes of renal function (RF) in very elderly patients with acute kidney injury (AKI).

PATIENTS AND METHODS

We carried out a retrospective cohort study of only very elderly patients who developed AKI at the geriatric department of a tertiary medical center during the period 2007-2015. All patients with AKI were followed up for 90 days after AKI diagnosis or until death. Survivors were divided into recovery and nonrecovery groups according to their RF 90 days post-AKI. RF recovery was defined as an estimated glomerular filtration rate (eGFR) of ≥60 mL/min/1.73 m².

RESULTS

In total, 668 patients (39.0%) developed AKI, and 652 patients were included in the final analysis. The median age of this population was 87 years, with 95.6% being male. The 90-day mortality rate was 33.6%. Of the 433 survivors, 316 (73.0%) recovered to their baseline eGFR. Body mass index (BMI), baseline eGFR, low mean aortic pressure (MAP), low prealbumin level, hypoalbuminemia, oliguria, blood urea nitrogen (BUN) level, and more severe AKI stage were independent risk factors associated with nonrenal recovery or death. AKI etiology, evaluated by peak serum creatinine (SCr) level and the requirement for dialysis, was not associated with nonrenal recovery.

CONCLUSION

Risk factors for the poor outcomes of RF in very elderly patients with AKI were BMI, baseline eGFR, low MAP, low prealbumin level, hypoalbuminemia, oliguria, BUN level, and more severe AKI stage. Identifying risk factors may help to improve patient outcomes.

Incidence, care quality and outcomes of patients with acute kidney injury in admitted hospital care

BACKGROUND/INTRODUCTION

Acute kidney injury (AKI) is common in acute hospital admission and associated with worse patient outcomes.

AIM

To measure incidence, care quality and outcome of AKI in admitted hospital care.

DESIGN

Forty-six of 168 acute NHS healthcare trusts in UK caring for 2 million acute hospital admissions per annum collected information on adults identified with AKI stage 3 (3-fold rise in serum creatinine or creatinine >354 µmol/l) through routine biochemical testing over a 5-month period in 2012.

METHODS

Information was collected on patient and care characteristics. Primary outcomes were survival and recovery of kidney function at 1 month.

RESULTS

A total of 15 647 patients were identified with biochemical AKI stage 3. Case note reviews were available for 7726 patients. In 80%, biochemical AKI stage 3 was confirmed clinically. Among this group, median age was 75 years, median length of stay was 12 days and the overall mortality within 1 month was 38%. Significant factors in a multivariable model predicting survival included age and some causes of AKI. Dipstick urinalysis, medication review, discussion with a nephrologist and acceptance for transfer to a renal unit were also associated with higher survival, but not early review by a senior doctor, acceptance for transfer to critical care or requirement for renal replacement therapy. Eighteen percent of people did not have their kidney function checked 1 month after the episode had resolved.

DISCUSSION/CONCLUSIONS

This large study of in-hospital AKI supports the efficacy of biochemical detection of AKI in common usage. AKI mortality remains substantial, length of stay comparable with single-centre studies, and much of the variation is poorly explained (model Cox and Snell R²=0.131) from current predictors.

The comparison of the commonly used surrogates for baseline renal function in acute kidney injury diagnosis and staging

BACKGROUND

Baseline serum creatinine (SCr) level is frequently not measured in clinical practice. The aim of this study was to investigate the effect of various methods of baseline SCr determination measurement on accuracy of acute kidney injury (AKI) diagnosis in critically ill patients.

METHODS

This was a retrospective cohort study. All adult intensive care unit (ICU) patients admitted at a tertiary referral hospital from January 1, 2011 through December 31, 2011, with at least one measured SCr value during ICU stay, were included in this study. The baseline SCr was considered either an admission SCr (SCrADM) or an estimated SCr, using MDRD formula, based on an assumed glomerular filtration rate (GFR) of 75 ml/min/1.73 m(2) (SCrGFR-75). Determination of AKI was based on the KDIGO SCr criterion. Propensity score to predict the likelihood of missing SCr was used to generate a simulated cohort of 3566 patients with baseline outpatient SCr, who had similar characteristics with patients whose outpatient SCr was not available.

RESULTS

Of 7772 patients, 3504 (45.1 %) did not have baseline outpatient SCr. Among patients without baseline outpatient SCr, AKI was detected in 571 (16.3 %) using the SCrADM and 997 (28.4 %) using SCrGFR-75 (p < .001). Compared with non-AKI patients, patients who met AKI only by SCrADM, but not SCrGFR-75, were significantly associated with 60-day mortality (OR 2.90; 95 % CI 1.66-4.87), whereas patients who met AKI only by SCrGFR-75, but not SCrADM, had a non-significant increase in 60-day mortality risk (OR 1.33; 95 % CI 0.94-1.88). In a simulated cohort of patients with baseline outpatient SCr, SCrGFR-75 yielded a higher sensitivity (77.2 vs. 50.5 %) and lower specificity (87.8 vs. 94.8 %) for the AKI diagnosis in comparison with SCrADM.

CONCLUSIONS

When baseline outpatient SCr was not available, using SCrGFR-75 as surrogate for baseline SCr was found to be more sensitive but less specific for AKI diagnosis compared with using SCrADM. This resulted in higher incidence of AKI with larger likelihood of false-positive cases.