MySurgeryRisk: Development and Validation of a Machine-learning Risk Algorithm for Major Complications and Death After Surgery.

OBJECTIVE: To accurately calculate the risk for postoperative complications and death after surgery in the preoperative period using machine-learning modeling of clinical data. BACKGROUND: Postoperative complications cause a 2-fold increase in the 30-day mortality and cost, and are associated with long-term consequences. The ability to precisely forecast the risk for major complications before surgery is limited. METHODS: In a single-center cohort of 51,457 surgical patients undergoing major inpatient surgery, we have developed and validated an automated analytics framework for a preoperative risk algorithm (MySurgeryRisk) that uses existing clinical data in electronic health records to forecast patient-level probabilistic risk scores for 8 major postoperative complications (acute kidney injury, sepsis, venous thromboembolism, intensive care unit admission >48 hours, mechanical ventilation >48 hours, wound, neurologic, and cardiovascular complications) and death up to 24 months after surgery. We used the area under the receiver characteristic curve (AUC) and predictiveness curves to evaluate model performance. RESULTS: MySurgeryRisk calculates probabilistic risk scores for 8 postoperative complications with AUC values ranging between 0.82 and 0.94 [99% confidence intervals (CIs) 0.81-0.94]. The model predicts the risk for death at 1, 3, 6, 12, and 24 months with AUC values ranging between 0.77 and 0.83 (99% CI 0.76-0.85). CONCLUSIONS: We constructed an automated predictive analytics framework for machine-learning algorithm with high discriminatory ability for assessing the risk of surgical complications and death using readily available preoperative electronic health records data. The feasibility of this novel algorithm implemented in real time clinical workflow requires further testing.

Percutaneous cholecystostomy: prognostic factors and comparison to cholecystectomy.

BACKGROUND: Data regarding long-term outcomes following percutaneous cholecystostomy (PC) are limited, and comparisons to cholecystectomy (CCY) are lacking. We hypothesized that chronic disease burden would predict 1-year mortality following PC, and that outcomes following PC and CCY would be similar when controlling for preprocedural risk factors. METHODS: We performed a 10-year retrospective cohort analysis of patients with acute cholecystitis managed by PC (n = 114) or CCY (n = 234). Treatment response was assessed by systemic inflammatory response syndrome (SIRS) criteria at PC/CCY and 72 h later. Logistic regression identified predictors of 30-day and 1-year mortality following PC. PC and CCY patients were matched by age, Tokyo Guidelines (TG13) cholecystitis severity grade, and VASQIP calculator predicted mortality (n = 42/group). RESULTS: The presence of SIRS at 72 h following PC was associated with 30-day mortality [OR 8.9 (95% CI 2.6-30)]. SIRS at 72 h was present in and 21.4% of all PC patients, significantly higher than unmatched CCY patients (4.7%, p = 0.048). Independent predictors of 1-year mortality following PC were DNR status [19.7 (2.1-186)], disseminated cancer [7.5 (2.1-26)], and congestive heart failure [3.9 (1.4-11)]. PC patients with none of these risk factors had 17.9% 90-day mortality and no deaths after 90 days; late deaths continued to occur among patients with DNR, CHF, or disseminated cancer. At baseline, PC patients had greater acute and chronic disease burden than CCY patients. After matching, PC and CCY patients had similar age (69 vs. 70 years), TG13 grade (2.4 vs. 2.4), and predicted 30-day mortality (5.5 vs. 6.8%). Matched PC patients had higher 30-day mortality (14.3 vs. 2.4%, p = 0.109) and 180-day mortality (28.6 vs. 7.1%, p = 0.048). CONCLUSIONS: Treatment response to PC predicted 30-day mortality; DNR status, and chronic diseases predicted 1-year mortality. Although the matching procedure did not eliminate selection bias, PC was associated with persistent systemic inflammation and higher long-term mortality than CCY.

The Pattern of Longitudinal Change in Serum Creatinine and 90-Day Mortality After Major Surgery.

OBJECTIVE: Calculate mortality risk that accounts for both severity and recovery of postoperative kidney dysfunction using the pattern of longitudinal change in creatinine. BACKGROUND: Although the importance of renal recovery after acute kidney injury (AKI) is increasingly recognized, the complex association that accounts for longitudinal creatinine changes and mortality is not fully described. METHODS: We used routinely collected clinical information for 46,299 adult patients undergoing major surgery to develop a multivariable probabilistic model optimized for nonlinearity of serum creatinine time series that calculates the risk function for 90-day mortality. We performed a 70/30 cross validation analysis to assess the accuracy of the model. RESULTS: All creatinine time series exhibited nonlinear risk function in relation to 90-day mortality and their addition to other clinical factors improved the model discrimination. For any given severity of AKI, patients with complete renal recovery, as manifested by the return of the discharge creatinine to the baseline value, experienced a significant decrease in the odds of dying within 90 days of admission compared with patients with partial recovery. Yet, for any severity of AKI, even complete renal recovery did not entirely mitigate the increased odds of dying, as patients with mild AKI and complete renal recovery still had significantly increased odds for dying compared with patients without AKI [odds ratio: 1.48 (95% confidence interval: 1.30-1.68)]. CONCLUSIONS: We demonstrate the nonlinear relationship between both severity and recovery of renal dysfunction and 90-day mortality after major surgery. We have developed an easily applicable computer algorithm that calculates this complex relationship.

National surgical quality improvement program underestimates the risk associated with mild and moderate postoperative acute kidney injury.

OBJECTIVES: In a single-center cohort of surgical patients, we assessed the association between postoperative change in serum creatinine and adverse outcomes and compared the American College of Surgeons National Surgical Quality Improvement Program's definition for acute kidney injury with consensus risk, injury, failure, loss, and end-stage kidney and Kidney Disease: Improving Global Outcomes definitions. DESIGN: Retrospective single-center cohort. SETTING: Academic tertiary medical center. PATIENTS: Twenty-seven thousand eight hundred forty-one adult patients with no previous history of chronic kidney disease undergoing major surgery. INTERVENTIONS: Risk, injury, failure, loss, and end-stage kidney defines acute kidney injury as change in serum creatinine greater than or equal to 50% while Kidney Disease: Improving Global Outcomes uses 0.3 mg/dL change from the reference serum creatinine. Since National Surgical Quality Improvement Program defines acute kidney injury as serum creatinine change greater than 2 mg/dL, it may underestimate the risk associated with less severe acute kidney injury. MEASUREMENTS AND MAIN RESULTS: The optimal discrimination limits for both percent and absolute serum creatinine changes were calculated by maximizing sensitivity and specificity along the receiver operating characteristic curves for postoperative complications and mortality. Although prevalence of risk, injury, failure, loss, and end-stage kidney-acute kidney injury was 37%, only 7% of risk, injury, failure, loss, and end-stage kidney-acute kidney injury patients would be diagnosed with acute kidney injury using the National Surgical Quality Improvement Program definition. In multivariable logistic models, patients with risk, injury, failure, loss, and end-stage kidney or Kidney Disease: Improving Global Outcomes-acute kidney injury had a 10 times higher odds of dying compared to patients without acute kidney injury. The optimal discrimination limits for change in serum creatinine associated with adverse postoperative outcomes were as low as 0.2 mg/dL while the National Surgical Quality Improvement Program discrimination limit of 2.0 mg/dL had low sensitivity (0.05-0.28). CONCLUSIONS: Current American College of Surgeons National Surgical Quality Improvement Program definition underestimates the risk associated with mild and moderate acute kidney injury otherwise captured by the consensus risk, injury, failure, loss, and end-stage kidney and Kidney Disease: Improving Global Outcomes criteria.

Postoperative acute kidney injury in adult non-cardiac surgery: joint consensus report of the Acute Disease Quality Initiative and PeriOperative Quality Initiative.

Postoperative acute kidney injury (PO-AKI) is a common complication of major surgery that is strongly associated with short-term surgical complications and long-term adverse outcomes, including increased risk of chronic kidney disease, cardiovascular events and death. Risk factors for PO-AKI include older age and comorbid diseases such as chronic kidney disease and diabetes mellitus. PO-AKI is best defined as AKI occurring within 7 days of an operative intervention using the Kidney Disease Improving Global Outcomes (KDIGO) definition of AKI; however, additional prognostic information may be gained from detailed clinical assessment and other diagnostic investigations in the form of a focused kidney health assessment (KHA). Prevention of PO-AKI is largely based on identification of high baseline risk, monitoring and reduction of nephrotoxic insults, whereas treatment involves the application of a bundle of interventions to avoid secondary kidney injury and mitigate the severity of AKI. As PO-AKI is strongly associated with long-term adverse outcomes, some form of follow-up KHA is essential; however, the form and location of this will be dictated by the nature and severity of the AKI. In this Consensus Statement, we provide graded recommendations for AKI after non-cardiac surgery and highlight priorities for future research.

Acute kidney injury is associated with increased long-term mortality after cardiothoracic surgery.

BACKGROUND: Long-term survival after acute kidney injury (AKI) is poorly studied. We report the relationship between long-term mortality and AKI with small changes in serum creatinine during hospitalization after various cardiothoracic surgery procedures. METHODS AND RESULTS: This was a retrospective study of 2973 patients with no history of chronic kidney disease who were discharged from the hospital after cardiothoracic surgery between 1992 and 2002. AKI was defined by the RIFLE classification (Risk, Injury, Failure, Loss, and End stage), which requires at least a 50% increase in serum creatinine and stratifies patients into 3 grades of AKI: Risk, injury, and failure. Patient survival was determined through the National Social Security Death Index. Long-term survival was analyzed with a risk-adjusted Cox proportional hazards regression model. Survival was worse among patients with AKI and was proportional to its severity, with an adjusted hazard ratio of 1.23 (95% CI 1.06 to 1.42) for the least severe RIFLE risk class and 2.14 (95% CI 1.73 to 2.66) for the RIFLE failure class compared with patients without AKI. Survival was worse among all subgroups of cardiothoracic surgery with AKI except for valve surgery. Patients with complete renal recovery after AKI still had an increased adjusted hazard ratio for death of 1.28 (95% CI 1.11 to 1.48) compared with patients without AKI. CONCLUSIONS: The risk of death associated with AKI after cardiothoracic surgery remains high for 10 years regardless of other risk factors, even for those patients with complete renal recovery. Improved renal protection and closer postdischarge follow-up of renal function may be warranted.

Brain natriuretic peptide is not reno-protective during renal ischemia-reperfusion injury in the rat.

BACKGROUND: Acute kidney injury (AKI) occurs in 30% of patients undergoing complex cardiovascular surgery, and renal ischemia-reperfusion (I/R) injury is often a contributing factor. A recent meta-analysis observed that perioperative natriuretic peptide administration was associated with a reduction in AKI requiring dialysis in cardiovascular surgery patients. This study was designed to further clarify the potential reno-protective effect of brain natriuretic peptide (BNP) using an established rat model of renal I/R injury. METHODS: The study comprised three groups (n = 10 kidneys each): (1) control (no injury); (2) I/R injury (45 min of bilateral renal ischemia followed by 3 h of reperfusion); and (3) BNP (I/R injury plus rat-BNP pretreatment at 0.01 µg/kg/min). Glomerular filtration rate (GFR) and a biomarker of AKI, urinary neutrophil gelatinase-associated lipocalin (uNGAL), were measured at baseline and at 30 minute intervals post-ischemia. Groups were compared using two-way repeated measures analysis of variance (mean ± SD, significance P < 0.05). RESULTS: Baseline GFR measurements for control, I/R, and BNP groups were 1.07 ± 0.55, 0.88 ± 0.51, and 1.03 ± 0.59 mL/min (P = 0.90), respectively. Post-ischemia, GFR was significantly lower in I/R and BNP compared with controls at 30 min, 1.29 ± 0.97, 0.08 ± 0.04, and 0.06 ± 0.05 mL/min (P < 0.01), and remained lower through 3 h, 1.79 ± 0.44, 0.30 ± 0.17, and 0.32 ± 0.12 mL/min (P < 0.01). Comparing I/R to BNP groups, GFR did not differ significantly at any time point. There was no significant difference in uNGAL levels at 1 h (552 ± 358 versus 516 ± 259 ng/mL, P = 0.87) or 2 h (1073 ± 589 versus 989 ± 218 ng/mL, P = 0.79) between I/R and BNP. CONCLUSIONS: BNP does not reduce the renal injury biomarker, urinary NGAL, or preserve GFR in acute renal ischemia-reperfusion injury.

Long-term risk of mortality and acute kidney injury during hospitalization after major surgery.

OBJECTIVE: To determine the relationship between long-term mortality and acute kidney injury (AKI) during hospitalization after major surgery. SUMMARY BACKGROUND DATA: AKI is associated with a risk of short-term mortality that is proportional to its severity; however the long-term survival of patients with AKI is poorly studied. METHODS: This is a retrospective cohort study of 10,518 patients with no history of chronic kidney disease who were discharged after a major surgery between 1992 and 2002. AKI was defined by the RIFLE (Risk, Injury, Failure, Loss, and End-stage Kidney) classification, which requires at least a 50% increase in serum creatinine (sCr) and stratifies patients into 3 severity stages: risk, injury, and failure. Patient survival was determined through the National Social Security Death Index. Long-term survival was analyzed using a risk-adjusted Cox proportional hazards regression model. RESULTS: In the risk-adjusted model, survival was worse among patients with AKI and was proportional to its severity with an adjusted hazard ratio of 1.18 (95% confidence interval [CI], 1.08-1.29) for the RIFLE-Risk class and 1.57 (95% CI, 1.40-1.75) for the RIFLE-Failure class, compared with patients without AKI (P < 0.001). Patients with complete renal recovery after AKI still had an increased adjusted hazard ratio for death of 1.20 (95% CI, 1.10-1.31) compared with patients without AKI (P < 0.001). CONCLUSIONS: In a large single-center cohort of patients discharged after major surgery, AKI with even small changes in sCr level during hospitalization was associated with an independent long-term risk of death.

The rise and fall of natriuretic peptides in acute kidney injury: a misunderstood relationship?

The vasodilatory, natriuretic, and diuretic properties of natriuretic peptides (NPs) make them attractive agents in the treatment of acute kidney injury (AKI). However, there is conflicting evidence of their beneficial effects. This article examines the reasons for the differences, and provides insight that the reported outcomes may be related to the unique physiologic effects and mechanisms of action of NPs, the designs and cohorts of the trials, and the characteristic renal hemodynamics associated with AKI. NPs are effective in the prevention of AKI when applied prophylactically, in lower doses, for prolonged duration, in patients with mild to moderate impairment in renal function, and in predictable clinical settings with clearly defined outcome measurements.

The Impact of Changes in Medicare's Physician Payment System on Critical Care.

In 2015 President Obama signed the Medicare Access and CHIP Reauthorization Act (MACRA) which repealed the Sustainable Growth Rate (SGR) mechanism for Medicare physician reimbursement and mandated that CMS develop alternative payment methodologies to "reward health care providers for giving better care not more just more care." MACRA makes 3 major changes to Medicare reimbursements: (1) it ends the SGR formula; (2) it establishes a new framework to reward physicians based on performance and health outcomes rather than volume; and (3) it aims to combine existing quality reporting programs into one streamlined system. Beginning in 2019, physicians must enter one of two new tracks for payment: the Merit-based Incentive Payment System (MIPS) or Alternative Payment Models (APMs). SCCM has a unique opportunity as the largest multidisciplinary critical care organization to comment upon and, ideally, to help develop the new physician payment models specifically for critical care services. The time is now for SCCM and its individual members to become involved in the process.

Applications for detection of acute kidney injury using electronic medical records and clinical information systems: workgroup statements from the 15(th) ADQI Consensus Conference.

Electronic medical records and clinical information systems are increasingly used in hospitals and can be leveraged to improve recognition and care for acute kidney injury. This Acute Dialysis Quality Initiative (ADQI) workgroup was convened to develop consensus around principles for the design of automated AKI detection systems to produce real-time AKI alerts using electronic systems. AKI alerts were recognized by the workgroup as an opportunity to prompt earlier clinical evaluation, further testing and ultimately intervention, rather than as a diagnostic label. Workgroup members agreed with designing AKI alert systems to align with the existing KDIGO classification system, but recommended future work to further refine the appropriateness of AKI alerts and to link these alerts to actionable recommendations for AKI care. The consensus statements developed in this review can be used as a roadmap for development of future electronic applications for automated detection and reporting of AKI.

Les dossiers médicaux électroniques et les systèmes de renseignements cliniques sont de plus en plus utilisés dans les hôpitaux. Ces éléments pourraient être mis à profit pour faciliter le dépistage de l'insuffisance rénale aigüe (IRA) et améliorer les soins offerts aux patients qui en souffrent. Lors de la dernière réunion du Acute Dialysis Quality Initiative (ADQI), un groupe de travail s'est réuni pour établir un consensus autour de principes régissant la constitution d'un système automatisé de détection de l'IRA. Un système qui permettrait de produire des alertes en temps réel pour dépister les cas d'IRA (alertes IRA). Le groupe de travail a reconnu que de telles alertes représenteraient des opportunités de procéder à une évaluation clinique ou un dépistage précoce de la maladie et donc, à des interventions plus rapides, plutôt que de ne constituer qu'un indicateur diagnostique. Les membres du groupe de travail se sont entendus pour que le système d'alertes IRA soit développé en se basant sur la classification établie par le KIDGO. Ils ont toutefois recommandé que des travaux ultérieurs soient effectués pour raffiner les alertes et pour que celles-ci soient suivies de recommandations applicables et assorties d'un plan concret de soins à offrir aux patients. Les déclarations consensuelles présentées dans ce compte-rendu pourraient constituer le plan de développement pour la mise au point d'applications électroniques permettant la détection et le signalement de cas d'IRA de façon automatisée.

Acute kidney injury is associated with early cytokine changes after trauma.

BACKGROUND: Acute kidney injury (AKI) occurs in 26% of trauma patients and is associated with increased mortality and risk for nosocomial infections (NCIs). We compared serial plasma cytokine levels in patients with posttraumatic AKI to determine whether the early cytokine changes are associated with the occurrence of AKI and NCI. METHODS: We performed a secondary analysis of the Inflammation and the Host Response to Injury database to include adult blunt trauma patients who had available plasma proteomic analyses. AKI was defined by the RIFLE (Risk, Injury, Failure, Loss, and End-stage Kidney) classification, which requires a 50% increase in serum creatinine. The association among AKI, NCI, and plasma cytokines was analyzed using a mixed model analyses and logistic regression. RESULTS: Among 147 patients in the cohort, prevalence of NCI was 73% and 52% for patients with and without AKI, respectively. In mixed model analyses adjusted for clinical factors, AKI patients developed significant early increase in IL-1ra, IL-8, MCP1, and IL-6; early decrease in sTNFR2; and late decrease in IL-1ra, IL-4, and IL-6 concentrations, compared with patients without AKI and regardless of NCI. The change in cytokine pattern differed for sIL1R2, CXCL1, and MIP1ß, depending on the occurrence of NCI. Patients with AKI and NCI had lower early and late sIL1R2 and higher early and late CXCL1 and MIP1ß levels. Within the first 24 hours of injury, adding plasma levels of IL-1ra, IL-8, MCP1, IL-6, and sTNFR2 to clinical parameters of injury severity provided a predictive model for AKI superior to clinical model only (p < 0.001). CONCLUSION: AKI trauma patients exhibit simultaneous changes in proinflammatory and anti-inflammatory serial plasma cytokine levels. The predictive model for AKI that combines plasma cytokine levels with clinical data within 24 hours of injury requires further prospective validation in larger studies. LEVEL OF EVIDENCE: Prognostic study, level III.