Alerting Clinicians to 1-Year Mortality Risk in Patients Hospitalized With Heart Failure: The REVEAL-HF Randomized Clinical Trial.

Importance: Heart failure is a major cause of morbidity and mortality worldwide. The use of risk scores has the potential to improve targeted use of interventions by clinicians that improve patient outcomes, but this hypothesis has not been tested in a randomized trial. Objective: To evaluate whether prognostic information in heart failure translates into improved decisions about initiation and intensity of treatment, more appropriate end-of-life care, and a subsequent reduction in rates of hospitalization or death. Design, Setting, and Participants: This was a pragmatic, multicenter, electronic health record-based, randomized clinical trial across the Yale New Haven Health System, comprising small community hospitals and large tertiary care centers. Patients hospitalized for heart failure who had N-terminal pro-brain natriuretic peptide (NT-proBNP) levels of greater than 500 pg/mL and received intravenous diuretics within 24 hours of admission were automatically randomly assigned to the alert (intervention) or usual-care groups. Interventions: The alert group had their risk of 1-year mortality calculated using an algorithm that was derived and validated using similar historic patients in the electronic health record. This estimate, including a categorical risk assessment, was presented to clinicians while they were interacting with a patient's electronic health record. Main Outcomes and Measures: The primary outcome was a composite of 30-day hospital readmissions and all-cause mortality at 1 year. Results: Between November 27, 2019, through March 7, 2021, 3124 patients were randomly assigned to the alert (1590 [50.9%]) or usual-care (1534 [49.1%]) group. The alert group had a median (IQR) age of 76.5 (65-86) years, and 796 were female patients (50.1%). Patients from the following race and ethnicity groups were included: 13 Asian (0.8%), 324 Black (20.4%), 136 Hispanic (8.6%), 1448 non-Hispanic (91.1%), 1126 White (70.8%), 6 other ethnicity (0.4%), and 127 other race (8.0%). The usual-care group had a median (IQR) age of 77 (65-86) years, and 788 were female patients (51.4%). Patients from the following race and ethnicity groups were included: 11 Asian (1.4%), 298 Black (19.4%), 162 Hispanic (10.6%), 1359 non-Hispanic (88.6%), 1077 White (70.2%), 13 other ethnicity (0.9%), and 137 other race (8.9%). Median (IQR) NT-proBNP levels were 3826 (1692-8241) pg/mL in the alert group and 3867 (1663-8917) pg/mL in the usual-care group. A total of 284 patients (17.9%) and 270 patients (17.6%) were admitted to the intensive care unit in the alert and usual-care groups, respectively. A total of 367 patients (23.1%) and 359 patients (23.4%) had a left ventricular ejection fraction of 40% or less in the alert and usual-care groups, respectively. The model achieved an area under the curve of 0.74 in the trial population. The primary outcome occurred in 619 patients (38.9%) in the alert group and 603 patients (39.3%) in the usual-care group (P = .89). There were no significant differences between study groups in the prescription of heart failure medications at discharge, the placement of an implantable cardioverter-defibrillator, or referral to palliative care. Conclusions and Relevance: Provision of 1-year mortality estimates during heart failure hospitalization did not affect hospitalization or mortality, nor did it affect clinical decision-making. Trial Registration: ClinicalTrials.gov Identifier NCT03845660.

Blood pressure response to commonly administered antihypertensives for severe inpatient hypertension.

BACKGROUND: Blood pressure (BP) elevations are commonly treated in hospitalized patients; however, treatment is not guideline directed. Our objective was to assess BP response to commonly prescribed antihypertensives after the development of severe inpatient hypertension (HTN). METHODS: This is a cohort study of adults, excluding intensive care unit patients, within a single healthcare system admitted for reasons other than HTN who developed severe HTN (systolic BP>180 or diastolic BP >110 mmHg at least 1 hour after admission). We identified the most commonly administered antihypertensives given within 6 hours of severe HTN (given to >10% of treated patients). We studied the association of treatment with each antihypertensive vs. no treatment on BP change in the 6 hours following severe HTN development using mixed-effects model after adjusting for demographics and clinical characteristics. RESULTS: Among 23,147 patients who developed severe HTN, 9,166 received antihypertensive treatment. The most common antihypertensives given were oral metoprolol (n = 1991), oral amlodipine (n = 1812), oral carvedilol (n = 1116), IV hydralazine (n = 1069) and oral hydralazine (n = 953). In the fully adjusted model, treatment with IV hydralazine led to 13 [-15.9, -10.1], 18 [-22.2, -14] and 11 [-14.1, -8.3] mmHg lower MAP, SBP, and DBP in the 6 hours following severe HTN development compared to no treatment. Treatment with oral hydralazine and oral carvedilol also resulted in significantly lower BPs in the 6 hours following severe HTN development (6 [-9.1, -2.1 and -7 [-9.1, -4.2] lower MAP, respectively) compared to no treatment. Receiving metoprolol and amlodipine did not result in a drop in BP compared to no treatment. CONCLUSION: Among commonly used antihypertensives, IV hydralazine resulted in the most significant drop in BP following severe HTN, while metoprolol and amlodipine did not lower BP. Further research to assess the effect of treatment on clinical outcomes and if needed which antihypertensives to administer are necessary.

Severe inpatient hypertension prevalence and blood pressure response to antihypertensive treatment.

Severe hypertension (HTN) that develops during hospitalization is more common than admission for HTN; however, it is poorly studied, and treatment guidelines are lacking. Our goal is to characterize hospitalized patients who develop severe HTN and assess blood pressure (BP) response to treatment. This is a multi-hospital retrospective cohort study of adults admitted for reasons other than HTN who developed severe HTN. The authors defined severe inpatient HTN as the first documented BP elevation (systolic BP > 180 or diastolic BP > 110) at least 1 hour after admission. Treatment was defined as receiving antihypertensives (intravenous [IV] or oral) within 6h of BP elevation. As a measure of possible overtreatment, the authors studied the association between treatment and time to mean arterial pressure (MAP) drop ≥ 30% using the Cox proportional hazards model. Among 224 265 hospitalized adults, 10% developed severe HTN of which 40% were treated. Compared to patients who did not develop severe HTN, those who did were older, more commonly women and black, and had more comorbidities. Incident MAP drop ≥ 30% among treated and untreated patients with severe HTN was 2.2 versus 5.7/1000 person-hours. After adjustment, treated versus. untreated patients had lower rates of MAP drop ≥ 30% (hazard rate [HR]: 0.9 [0.8, 0.99]). However, those receiving only IV treatment versus untreated had greater rates of MAP drop ≥ 30% (1.4 [1.2, 1.7]). Overall, the authors found that clinically significant MAP drop is observed among inpatients with severe HTN irrespective of treatment, with greater rates observed among patients treated only with IV antihypertensives. Further research is needed to phenotype inpatients with severe HTN.

Predicting patients with false negative SARS-CoV-2 testing at hospital admission: A retrospective multi-center study.

IMPORTANCE: False negative SARS-CoV-2 tests can lead to spread of infection in the inpatient setting to other patients and healthcare workers. However, the population of patients with COVID who are admitted with false negative testing is unstudied. OBJECTIVE: To characterize and develop a model to predict true SARS-CoV-2 infection among patients who initially test negative for COVID by PCR. DESIGN: Retrospective cohort study. SETTING: Five hospitals within the Yale New Haven Health System between 3/10/2020 and 9/1/2020. PARTICIPANTS: Adult patients who received diagnostic testing for SARS-CoV-2 virus within the first 96 hours of hospitalization. EXPOSURE: We developed a logistic regression model from readily available electronic health record data to predict SARS-CoV-2 positivity in patients who were positive for COVID and those who were negative and never retested. MAIN OUTCOMES AND MEASURES: This model was applied to patients testing negative for SARS-CoV-2 who were retested within the first 96 hours of hospitalization. We evaluated the ability of the model to discriminate between patients who would subsequently retest negative and those who would subsequently retest positive. RESULTS: We included 31,459 hospitalized adult patients; 2,666 of these patients tested positive for COVID and 3,511 initially tested negative for COVID and were retested. Of the patients who were retested, 61 (1.7%) had a subsequent positive COVID test. The model showed that higher age, vital sign abnormalities, and lower white blood cell count served as strong predictors for COVID positivity in these patients. The model had moderate performance to predict which patients would retest positive with a test set area under the receiver-operator characteristic (ROC) of 0.76 (95% CI 0.70-0.83). Using a cutpoint for our risk prediction model at the 90th percentile for probability, we were able to capture 35/61 (57%) of the patients who would retest positive. This cutpoint amounts to a number-needed-to-retest range between 15 and 77 patients. CONCLUSION AND RELEVANCE: We show that a pragmatic model can predict which patients should be retested for COVID. Further research is required to determine if this risk model can be applied prospectively in hospitalized patients to prevent the spread of SARS-CoV-2 infections.

REVeAL-HF: Design and Rationale of a Pragmatic Randomized Controlled Trial Embedded Within Routine Clinical Practice.

Heart failure (HF) is one of the most common causes of hospitalization in the United States and carries a significant risk of morbidity and mortality. Use of evidence-based interventions may improve outcomes, but their use is encumbered in part by limitations in accurate prognostication. The REVeAL-HF (Risk EValuation And its Impact on ClinicAL Decision Making and Outcomes in Heart Failure) trial is the first to definitively evaluate the impact of knowledge about prognosis on clinical decision making and patient outcomes. The REVeAL-HF trial is a pragmatic, completely electronic, randomized controlled trial that has completed enrollment of 3,124 adults hospitalized for HF, defined as having an N-terminal pro-B-type natriuretic peptide level of >500 pg/ml and receiving intravenous diuretic agents within 24 h of admission. Patients randomized to the intervention had their risk of 1-year mortality generated with information in the electronic health record and presented to their providers, who had the option to give feedback on their impression of this risk assessment. The authors are examining the impact of this information on clinical decision-making (use of HF pharmacotherapies, referral to electrophysiology, palliative care referral, and referral for advanced therapies like heart transplantation or mechanical circulatory support) and patient outcomes (length of stay, post-discharge 30-day rehospitalizations, and 1-year mortality). The REVeAL-HF trial will definitively examine whether knowledge about prognosis in HF has an impact on clinical decision making and patient outcomes. It will also examine the relationship between calculated, perceived, and real risk of mortality in this patient population. (Risk EValuation And Its Impact on ClinicAL Decision Making and Outcomes in Heart Failure [REVeAL-HF]; NCT03845660).

Assessment of Acute Kidney Injury and Longitudinal Kidney Function After Hospital Discharge Among Patients With and Without COVID-19.

Importance: Acute kidney injury (AKI) occurs in up to half of patients hospitalized with coronavirus disease 2019 (COVID-19). The longitudinal effects of COVID-19-associated AKI on kidney function remain unknown. Objective: To compare the rate of change in estimated glomerular filtration rate (eGFR) after hospital discharge between patients with and without COVID-19 who experienced in-hospital AKI. Design, Setting, and Participants: A retrospective cohort study was conducted at 5 hospitals in Connecticut and Rhode Island from March 10 to August 31, 2020. Patients who were tested for COVID-19 and developed AKI were screened, and those who survived past discharge, did not require dialysis within 3 days of discharge, and had at least 1 outpatient creatinine level measurement following discharge were included. Exposures: Diagnosis of COVID-19. Main Outcomes and Measures: Mixed-effects models were used to assess the association between COVID-19-associated AKI and eGFR slope after discharge. The secondary outcome was the time to AKI recovery for the subgroup of patients whose kidney function had not returned to the baseline level by discharge. Results: A total of 182 patients with COVID-19-associated AKI and 1430 patients with AKI not associated with COVID-19 were included. The population included 813 women (50.4%); median age was 69.7 years (interquartile range, 58.9-78.9 years). Patients with COVID-19-associated AKI were more likely to be Black (73 [40.1%] vs 225 [15.7%]) or Hispanic (40 [22%] vs 126 [8.8%]) and had fewer comorbidities than those without COVID-19 but similar rates of preexisting chronic kidney disease and hypertension. Patients with COVID-19-associated AKI had a greater decrease in eGFR in the unadjusted model (-11.3; 95% CI, -22.1 to -0.4 mL/min/1.73 m2/y; P = .04) and after adjusting for baseline comorbidities (-12.4; 95% CI, -23.7 to -1.2 mL/min/1.73 m2/y; P = .03). In the fully adjusted model controlling for comorbidities, peak creatinine level, and in-hospital dialysis requirement, the eGFR slope difference persisted (-14.0; 95% CI, -25.1 to -2.9 mL/min/1.73 m2/y; P = .01). In the subgroup of patients who had not achieved AKI recovery by discharge (n = 319), COVID-19-associated AKI was associated with decreased kidney recovery during outpatient follow-up (adjusted hazard ratio, 0.57; 95% CI, 0.35-0.92). Conclusions and Relevance: In this cohort study of US patients who experienced in-hospital AKI, COVID-19-associated AKI was associated with a greater rate of eGFR decrease after discharge compared with AKI in patients without COVID-19, independent of underlying comorbidities or AKI severity. This eGFR trajectory may reinforce the importance of monitoring kidney function after AKI and studying interventions to limit kidney disease after COVID-19-associated AKI.

Modeling expression ranks for noise-tolerant differential expression analysis of scRNA-seq data.

Systematic delineation of complex biological systems is an ever-challenging and resource-intensive process. Single-cell transcriptomics allows us to study cell-to-cell variability in complex tissues at an unprecedented resolution. Accurate modeling of gene expression plays a critical role in the statistical determination of tissue-specific gene expression patterns. In the past few years, considerable efforts have been made to identify appropriate parametric models for single-cell expression data. The zero-inflated version of Poisson/negative binomial and log-normal distributions have emerged as the most popular alternatives owing to their ability to accommodate high dropout rates, as commonly observed in single-cell data. Although the majority of the parametric approaches directly model expression estimates, we explore the potential of modeling expression ranks, as robust surrogates for transcript abundance. Here we examined the performance of the discrete generalized beta distribution (DGBD) on real data and devised a Wald-type test for comparing gene expression across two phenotypically divergent groups of single cells. We performed a comprehensive assessment of the proposed method to understand its advantages compared with some of the existing best-practice approaches. We concluded that besides striking a reasonable balance between Type I and Type II errors, ROSeq, the proposed differential expression test, is exceptionally robust to expression noise and scales rapidly with increasing sample size. For wider dissemination and adoption of the method, we created an R package called ROSeq and made it available on the Bioconductor platform.

Electronic health record alerts for acute kidney injury: multicenter, randomized clinical trial.

OBJECTIVE: To determine whether electronic health record alerts for acute kidney injury would improve patient outcomes of mortality, dialysis, and progression of acute kidney injury. DESIGN: Double blinded, multicenter, parallel, randomized controlled trial. SETTING: Six hospitals (four teaching and two non-teaching) in the Yale New Haven Health System in Connecticut and Rhode Island, US, ranging from small community hospitals to large tertiary care centers. PARTICIPANTS: 6030 adult inpatients with acute kidney injury, as defined by the Kidney Disease: Improving Global Outcomes (KDIGO) creatinine criteria. INTERVENTIONS: An electronic health record based "pop-up" alert for acute kidney injury with an associated acute kidney injury order set upon provider opening of the patient's medical record. MAIN OUTCOME MEASURES: A composite of progression of acute kidney injury, receipt of dialysis, or death within 14 days of randomization. Prespecified secondary outcomes included outcomes at each hospital and frequency of various care practices for acute kidney injury. RESULTS: 6030 patients were randomized over 22 months. The primary outcome occurred in 653 (21.3%) of 3059 patients with an alert and in 622 (20.9%) of 2971 patients receiving usual care (relative risk 1.02, 95% confidence interval 0.93 to 1.13, P=0.67). Analysis by each hospital showed worse outcomes in the two non-teaching hospitals (n=765, 13%), where alerts were associated with a higher risk of the primary outcome (relative risk 1.49, 95% confidence interval 1.12 to 1.98, P=0.006). More deaths occurred at these centers (15.6% in the alert group v 8.6% in the usual care group, P=0.003). Certain acute kidney injury care practices were increased in the alert group but did not appear to mediate these outcomes. CONCLUSIONS: Alerts did not reduce the risk of our primary outcome among patients in hospital with acute kidney injury. The heterogeneity of effect across clinical centers should lead to a re-evaluation of existing alerting systems for acute kidney injury. TRIAL REGISTRATION: ClinicalTrials.gov NCT02753751.

The Association of COVID-19 With Acute Kidney Injury Independent of Severity of Illness: A Multicenter Cohort Study.

RATIONALE & OBJECTIVE: Although coronavirus disease 2019 (COVID-19) has been associated with acute kidney injury (AKI), it is unclear whether this association is independent of traditional risk factors such as hypotension, nephrotoxin exposure, and inflammation. We tested the independent association of COVID-19 with AKI. STUDY DESIGN: Multicenter, observational, cohort study. SETTING & PARTICIPANTS: Patients admitted to 1 of 6 hospitals within the Yale New Haven Health System between March 10, 2020, and August 31, 2020, with results for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) testing via polymerase chain reaction of a nasopharyngeal sample. EXPOSURE: Positive test for SARS-CoV-2. OUTCOME: AKI by KDIGO (Kidney Disease: Improving Global Outcomes) criteria. ANALYTICAL APPROACH: Evaluated the association of COVID-19 with AKI after controlling for time-invariant factors at admission (eg, demographic characteristics, comorbidities) and time-varying factors updated continuously during hospitalization (eg, vital signs, medications, laboratory results, respiratory failure) using time-updated Cox proportional hazard models. RESULTS: Of the 22,122 patients hospitalized, 2,600 tested positive and 19,522 tested negative for SARS-CoV-2. Compared with patients who tested negative, patients with COVID-19 had more AKI (30.6% vs 18.2%; absolute risk difference, 12.5% [95% CI, 10.6%-14.3%]) and dialysis-requiring AKI (8.5% vs 3.6%) and lower rates of recovery from AKI (58% vs 69.8%). Compared with patients without COVID-19, patients with COVID-19 had higher inflammatory marker levels (C-reactive protein, ferritin) and greater use of vasopressors and diuretic agents. Compared with patients without COVID-19, patients with COVID-19 had a higher rate of AKI in univariable analysis (hazard ratio, 1.84 [95% CI, 1.73-1.95]). In a fully adjusted model controlling for demographic variables, comorbidities, vital signs, medications, and laboratory results, COVID-19 remained associated with a high rate of AKI (adjusted hazard ratio, 1.40 [95% CI, 1.29-1.53]). LIMITATIONS: Possibility of residual confounding. CONCLUSIONS: COVID-19 is associated with high rates of AKI not fully explained by adjustment for known risk factors. This suggests the presence of mechanisms of AKI not accounted for in this analysis, which may include a direct effect of COVID-19 on the kidney or other unmeasured mediators. Future studies should evaluate the possible unique pathways by which COVID-19 may cause AKI.

Predicting patients with false negative SARS-CoV-2 testing at hospital admission: A retrospective multi-center study.

Importance: False negative SARS-CoV-2 tests can lead to spread of infection in the inpatient setting to other patients and healthcare workers. However, the population of patients with COVID who are admitted with false negative testing is unstudied. Objective: To characterize and develop a model to predict true SARS-CoV-2 infection among patients who initially test negative for COVID by PCR. Design: Retrospective cohort study. Setting: Five hospitals within the Yale New Haven Health System between 3/10/2020 and 9/1/2020. Participants: Adult patients who received diagnostic testing for SARS-CoV-2 virus within the first 96 hours of hospitalization. Exposure: We developed a logistic regression model from readily available electronic health record data to predict SARS-CoV-2 positivity in patients who were positive for COVID and those who were negative and never retested. Main Outcomes and Measures: This model was applied to patients testing negative for SARS-CoV-2 who were retested within the first 96 hours of hospitalization. We evaluated the ability of the model to discriminate between patients who would subsequently retest negative and those who would subsequently retest positive. Results: We included 31,459 hospitalized adult patients; 2,666 of these patients tested positive for COVID and 3,511 initially tested negative for COVID and were retested. Of the patients who were retested, 61 (1.7%) had a subsequent positive COVID test. The model showed that higher age, vital sign abnormalities, and lower white blood cell count served as strong predictors for COVID positivity in these patients. The model had moderate performance to predict which patients would retest positive with a test set area under the receiver-operator characteristic (ROC) of 0.76 (95% CI 0.70 - 0.83). Using a cutpoint for our risk prediction model at the 90th percentile for probability, we were able to capture 35/61 (57%) of the patients who would retest positive. This cutpoint amounts to a number-needed-to-retest range between 15 and 77 patients. Conclusion and Relevance: We show that a pragmatic model can predict which patients should be retested for COVID. Further research is required to determine if this risk model can be applied prospectively in hospitalized patients to prevent the spread of SARS-CoV-2 infections.

Optical configuration of modified Fredkin gate using lithium-niobate-based Mach-Zehnder interferometer.

The continuous quest for reversible computation that could be extensively used in applications such as digital signal processing, quantum computing, quantum-dot cellular automata, and nanotechnology has recently discovered its optical implementation as light tenders high-speed computing with the slightest information loss. The electro-optic effect of a lithium-niobate-based Mach-Zehnder interferometer is explored to configure a 4×4 modified Fredkin gate, capable of furnishing as many as 16 logical combinations, and thus showing potential of curbing the area overhead. The optical design is carried out using the beam propagation method. We have also performed the mathematical modeling and analyzed the results in MATLAB.

Real-Time Prediction of Acute Kidney Injury in Hospitalized Adults: Implementation and Proof of Concept.

RATIONALE & OBJECTIVE: Acute kidney injury (AKI) is diagnosed based on changes in serum creatinine concentration, a late marker of this syndrome. Algorithms that predict elevated risk for AKI are of great interest, but no studies have incorporated such an algorithm into the electronic health record to assist with clinical care. We describe the experience of implementing such an algorithm. STUDY DESIGN: Prospective observational cohort study. SETTING & PARTICIPANTS: 2,856 hospitalized adults in a single urban tertiary-care hospital with an algorithm-predicted risk for AKI in the next 24 hours>15%. Alerts were also used to target a convenience sample of 100 patients for measurement of 16 urine and 6 blood biomarkers. EXPOSURE: Clinical characteristics at the time of pre-AKI alert. OUTCOME: AKI within 24 hours of pre-AKI alert (AKI24). ANALYTICAL APPROACH: Descriptive statistics and univariable associations. RESULTS: At enrollment, mean predicted probability of AKI24 was 19.1%; 18.9% of patients went on to develop AKI24. Outcomes were generally poor among this population, with 29% inpatient mortality among those who developed AKI24 and 14% among those who did not (P<0.001). Systolic blood pressure<100mm Hg (28% of patients with AKI24 vs 18% without), heart rate>100 beats/min (32% of patients with AKI24 vs 24% without), and oxygen saturation<92% (15% of patients with AKI24 vs 6% without) were all more common among those who developed AKI24. Of all biomarkers measured, only hyaline casts on urine microscopy (72% of patients with AKI24 vs 25% without) and fractional excretion of urea nitrogen (20% [IQR, 12%-36%] among patients with AKI24 vs 34% [IQR, 25%-44%] without) differed between those who did and did not develop AKI24. LIMITATIONS: Single-center study, reliance on serum creatinine level for AKI diagnosis, small number of patients undergoing biomarker evaluation. CONCLUSIONS: A real-time AKI risk model was successfully integrated into the EHR.

A Time-Updated, Parsimonious Model to Predict AKI in Hospitalized Children.

BACKGROUND: Timely prediction of AKI in children can allow for targeted interventions, but the wealth of data in the electronic health record poses unique modeling challenges. METHODS: We retrospectively reviewed the electronic medical records of all children younger than 18 years old who had at least two creatinine values measured during a hospital admission from January 2014 through January 2018. We divided the study population into derivation, and internal and external validation cohorts, and used five feature selection techniques to select 10 of 720 potentially predictive variables from the electronic health records. Model performance was assessed by the area under the receiver operating characteristic curve in the validation cohorts. The primary outcome was development of AKI (per the Kidney Disease Improving Global Outcomes creatinine definition) within a moving 48-hour window. Secondary outcomes included severe AKI (stage 2 or 3), inpatient mortality, and length of stay. RESULTS: Among 8473 encounters studied, AKI occurred in 516 (10.2%), 207 (9%), and 27 (2.5%) encounters in the derivation, and internal and external validation cohorts, respectively. The highest-performing model used a machine learning-based genetic algorithm, with an overall receiver operating characteristic curve in the internal validation cohort of 0.76 [95% confidence interval (CI), 0.72 to 0.79] for AKI, 0.79 (95% CI, 0.74 to 0.83) for severe AKI, and 0.81 (95% CI, 0.77 to 0.86) for neonatal AKI. To translate this prediction model into a clinical risk-stratification tool, we identified high- and low-risk threshold points. CONCLUSIONS: Using various machine learning algorithms, we identified and validated a time-updated prediction model of ten readily available electronic health record variables to accurately predict imminent AKI in hospitalized children.

A simple real-time model for predicting acute kidney injury in hospitalized patients in the US: A descriptive modeling study.

BACKGROUND: Acute kidney injury (AKI) is an adverse event that carries significant morbidity. Given that interventions after AKI occurrence have poor performance, there is substantial interest in prediction of AKI prior to its diagnosis. However, integration of real-time prognostic modeling into the electronic health record (EHR) has been challenging, as complex models increase the risk of error and complicate deployment. Our goal in this study was to create an implementable predictive model to accurately predict AKI in hospitalized patients and could be easily integrated within an existing EHR system. METHODS AND FINDINGS: We performed a retrospective analysis looking at data of 169,859 hospitalized adults admitted to one of three study hospitals in the United States (in New Haven and Bridgeport, Connecticut) from December 2012 to February 2016. Demographics, medical comorbidities, hospital procedures, medications, and laboratory data were used to develop a model to predict AKI within 24 hours of a given observation. Outcomes of AKI severity, requirement for renal replacement therapy, and mortality were also measured and predicted. Models were trained using discrete-time logistic regression in a subset of Hospital 1, internally validated in the remainder of Hospital 1, and externally validated in Hospital 2 and Hospital 3. Model performance was assessed via the area under the receiver-operator characteristic (ROC) curve (AUC). The training set cohort contained 60,701 patients, and the internal validation set contained 30,599 patients. External validation data sets contained 43,534 and 35,025 patients. Patients in the overall cohort were generally older (median age ranging from 61 to 68 across hospitals); 44%-49% were male, 16%-20% were black, and 23%-29% were admitted to surgical wards. In the training set and external validation set, 19.1% and 18.9% of patients, respectively, developed AKI. The full model, including all covariates, had good ability to predict imminent AKI for the validation set, sustained AKI, dialysis, and death with AUCs of 0.74 (95% CI 0.73-0.74), 0.77 (95% CI 0.76-0.78), 0.79 (95% CI 0.73-0.85), and 0.69 (95% CI 0.67-0.72), respectively. A simple model using only readily available, time-updated laboratory values had very similar predictive performance to the complete model. The main limitation of this study is that it is observational in nature; thus, we are unable to conclude a causal relationship between covariates and AKI and do not provide an optimal treatment strategy for those predicted to develop AKI. CONCLUSIONS: In this study, we observed that a simple model using readily available laboratory data could be developed to predict imminent AKI with good discrimination. This model may lend itself well to integration into the EHR without sacrificing the performance seen in more complex models.

Electronic Alerts for Acute Kidney Injury Amelioration (ELAIA-1): a completely electronic, multicentre, randomised controlled trial: design and rationale.

INTRODUCTION: Acute kidney injury (AKI) is common among hospitalised patients and under-recognised by providers and yet carries a significant risk of morbidity and mortality. Electronic alerts for AKI have become more common despite a lack of strong evidence of their benefits. We designed a multicentre, randomised, controlled trial to evaluate the effectiveness of AKI alerts. Our aim is to highlight several challenges faced in the design of this trial, which uses electronic screening, enrolment, randomisation, intervention and data collection. METHODS AND ANALYSIS: The design and implementation of an electronic alert system for AKI was a reiterative process involving several challenges and limitations set by the confines of the electronic medical record system. The trial will electronically identify and randomise 6030 adults with AKI at six hospitals over a 1.5-2 year period to usual care versus an electronic alert containing an AKI-specific order set. Our primary outcome will be a composite of AKI progression, inpatient dialysis and inpatient death within 14 days of randomisation. During a 1-month pilot in the medical intensive care unit of Yale New Haven Hospital, we have demonstrated feasibility of automating enrolment and data collection. Feedback from providers exposed to the alerts was used to continually improve alert clarity, user friendliness and alert specificity through refined inclusion and exclusion criteria. ETHICS AND DISSEMINATION: This study has been approved by the appropriate ethics committees for each of our study sites. Our study qualified for a waiver of informed consent as it presents no more than minimal risk and cannot be feasibly conducted in the absence of a waiver. We are committed to open dissemination of our data through clinicaltrials.gov and submission of results to the NIH data sharing repository. Results of our trial will be submitted for publication in a peer-reviewed journal. TRIAL REGISTRATION NUMBER: NCT02753751; Pre-results.

Identification of Patients Expected to Benefit from Electronic Alerts for Acute Kidney Injury.

BACKGROUND AND OBJECTIVES: Electronic alerts for heterogenous conditions such as AKI may not provide benefit for all eligible patients and can lead to alert fatigue, suggesting that personalized alert targeting may be useful. Uplift-based alert targeting may be superior to purely prognostic-targeting of interventions because uplift models assess marginal treatment effect rather than likelihood of outcome. DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: This is a secondary analysis of a clinical trial of 2278 adult patients with AKI randomized to an automated, electronic alert system versus usual care. We used three uplift algorithms and one purely prognostic algorithm, trained in 70% of the data, and evaluated the effect of targeting alerts to patients with higher scores in the held-out 30% of the data. The performance of the targeting strategy was assessed as the interaction between the model prediction of likelihood to benefit from alerts and randomization status. The outcome of interest was maximum relative change in creatinine from the time of randomization to 3 days after randomization. RESULTS: The three uplift score algorithms all gave rise to a significant interaction term, suggesting that a strategy of targeting individuals with higher uplift scores would lead to a beneficial effect of AKI alerting, in contrast to the null effect seen in the overall study. The prognostic model did not successfully stratify patients with regards to benefit of the intervention. Among individuals in the high uplift group, alerting was associated with a median reduction in change in creatinine of -5.3% (P=0.03). In the low uplift group, alerting was associated with a median increase in change in creatinine of +5.3% (P=0.005). Older individuals, women, and those with a lower randomization creatinine were more likely to receive high uplift scores, suggesting that alerts may benefit those with more slowly developing AKI. CONCLUSIONS: Uplift modeling, which accounts for treatment effect, can successfully target electronic alerts for AKI to those most likely to benefit, whereas purely prognostic targeting cannot.

Approaches to Predicting Outcomes in Patients with Acute Kidney Injury.

Despite recognition that Acute Kidney Injury (AKI) leads to substantial increases in morbidity, mortality, and length of stay, accurate prognostication of these clinical events remains difficult. It remains unclear which approaches to variable selection and model building are most robust. We used data from a randomized trial of AKI alerting to develop time-updated prognostic models using stepwise regression compared to more advanced variable selection techniques. We randomly split data into training and validation cohorts. Outcomes of interest were death within 7 days, dialysis within 7 days, and length of stay. Data elements eligible for model-building included lab values, medications and dosages, procedures, and demographics. We assessed model discrimination using the area under the receiver operator characteristic curve and r-squared values. 2241 individuals were available for analysis. Both modeling techniques created viable models with very good discrimination ability, with AUCs exceeding 0.85 for dialysis and 0.8 for death prediction. Model performance was similar across model building strategies, though the strategy employing more advanced variable selection was more parsimonious. Very good to excellent prediction of outcome events is feasible in patients with AKI. More advanced techniques may lead to more parsimonious models, which may facilitate adoption in other settings.