Acute Kidney Injury, Anesthesia, Surgery Complications
Conditions
Keywords
Acute Kidney Injury, Surgical Outcomes, Machine Learning, Clinical Decision Support, Electronic Health Records
Brief summary
This investigator-initiated, pragmatic trial evaluates whether displaying a machine learning (ML)- derived perioperative AKI risk score-alone or paired with an interruptive Best/Our Practice Advisory (BPA/OPA)-improves kidney-protective care and reduces kidney injury after non-obstetric surgery at UCSF. Approximately 75-100 attending anesthesiologists (clusters) are randomized 1:1:1 to: (a) Control (risk score hidden), (b) Score Only (visible preoperative AKI risk probability with passive KDIGO bundle recommendation), or (c) Score + BPA (visible risk plus interruptive KDIGO prompt for high-risk patients). CRNAs/residents follow their attending' s assignment. Adult inpatients (age ≥18) with expected overnight stay and eGFR ≥15 mL/min/1.73 m² are included; obstetrics, chronic dialysis, and kidney transplant patients are excluded. The underlying preoperative model was prospectively validated at UCSF and outperforms anesthesiologist risk estimation reported in the literature. The model was reviewed and approved by the AI Oversight Committee at UCSF. Primary endpoint is the continuous change in serum creatinine (mg/dL) from baseline to POD 1-2. Secondary outcomes include KDIGO-defined AKI, adherence to bundle elements (hemodynamics, balanced fluids, nephrotoxin avoidance, glycemic control), intraoperative hypotension time, fluid volumes, nephrotoxin exposure, perioperative hyperglycemia, length of stay, unplanned ICU transfer, readmission, dialysis, and in-hospital mortality. Data are obtained from the EHR; analysts are blinded. No direct subject interaction is planned; the investigators will request a waiver of patient consent. The study aims to demonstrate that ML-enabled, workflow-embedded decision support can safely and feasibly improve guideline concordant care and decrease early postoperative kidney injury.
Interventions
DEVICEEHR-Embedded AKI Risk Score
A non-adaptive, machine learning-based clinical decision support tool integrated into the electronic health record that generates a preoperative probability of acute kidney injury (AKI) using routinely collected patient data. For patients identified as high risk, the tool displays the risk estimate to anesthesia providers without an accompanying Best Practice Advisory (BPA) recommending consideration of a KDIGO-based kidney-protective bundle. The intervention is advisory only, does not mandate clinical actions, and is designed to support provider decision-making within the existing clinical workflow.
DEVICEEHR-Embedded AKI Risk Score with Best Practice Advisory
A non-adaptive, machine learning-based clinical decision support tool integrated into the electronic health record that generates a preoperative probability of acute kidney injury (AKI) using routinely collected patient data. For patients identified as high risk, the tool displays the risk estimate to anesthesia providers with an accompanying Best Practice Advisory (BPA) recommending consideration of a KDIGO-based kidney-protective bundle. The intervention is advisory only, does not mandate clinical actions, and is designed to support provider decision-making within the existing clinical workflow.
Sponsors
University of California, San Francisco
National Institute of General Medical Sciences (NIGMS)
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
SCREENING
Masking
NONE
Intervention model description
This is a pragmatic, single-center, three-arm, parallel-group, cluster-randomized controlled trial. Attending anesthesiologists are the unit of randomization and are assigned in a 1:1:1 ratio to one of three groups: (1) control (AKI risk score not displayed), (2) score only (visible preoperative machine learning-derived AKI risk score with passive KDIGO bundle recommendation), or (3) score plus Best Practice Advisory (visible risk score with an interruptive KDIGO-based alert for high-risk patients). All eligible surgical cases managed by a given attending anesthesiologist inherit that provider's assigned study arm. Trainees and nurse anesthetists follow the assignment of the supervising attending. The intervention is delivered within the electronic health record at the point of care. The clinical decision support tools are advisory only and do not mandate any clinical actions. There is no crossover between groups, and allocation remains fixed for the duration of the study.
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
* Adults ≥18 years undergoing non-obstetric surgery at UCSF. * Inpatient cases with expected overnight stay. * Baseline eGFR ≥15 mL/min/1.73 m². * Managed by an attending anesthesiologist randomized to one of three arms (CRNAs/residents follow attending). * Data available in the UCSF EHR for risk scoring and outcomes.
Exclusion criteria
* Obstetric procedures. * Chronic dialysis patients. * Kidney transplant recipients. * Cases without baseline creatinine/eGFR or missing essential EHR elements needed for scoring/outcomes (operational exclusions). * Outpatient procedures without expected overnight stay.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Post-operative Change in Creatinine | From pre-operative baseline to 1-2 days post-operative level | Maximum continuous change in serum creatinine (mg/dL) from baseline to post-operative day 1-2 |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Acute Kidney Injury | Operation to Post-operative Day 7 | Acute Kidney Injury as defined by KDIGO |
| KDIGO Bundle Adherence | Intra-operative | Measurement of provider adherence to KDIGO components |
| Intra-Operative Time and Severity of Hypotension | Intra-operative | Intra-Operative Time and Severity (meaning how far below the threshold) where patient is in hypotension, defined as systolic blood pressure \<90 mmHg and mean arterial pressure \<65 mmHg during surgery |
| Total intra-operative intravenous fluid volume administered (mL) | Intra-operative | Provider administration of intravenous fluids during the intra-operative period, measured in milliliters (mL). Intravenous fluids include normal saline, lactated Ringer's, Plasma-Lyte, other balanced crystalloids, and colloid solutions such as albumin. |
| Length of Stay | Operation to Post-operative Day 180 | Duration of patient admission in hospital in days |
| Intra-operative Hyperglycemic Events | Intra-operative | Number of intra-operative hyperglycemic events, defined as the number of recorded blood glucose measurements exceeding 180 mg/dL. |
| Intra-operative Nephrotoxin Exposure | Intra-operative | Number of nephrotoxic medications administered intra-operatively and duration of intra-operative exposure |
| In-Hospital Mortality | Operation to Post-operative Day 180 | Patient death while admitted in the hospital |
| ICU Transfer and total time in the ICU | Postoperative | Any transfers to the ICU while admitted and the total time the patient spends in the ICU |
| Hospital Readmission | Operation to Post-operative Day 180 | Readmission back to a UCSF hospital following operation |
| Dialysis Requirement | Operation to Post-operative Day 180 | Patients requiring dialysis following surgery |
| Dilution Corrected KDIGO AKI measurement (Stage 1 or higher) | AKI is defined per KDIGO as corrected creatinine increase ≥0.3 mg/dL within 48 hours or ≥1.5× baseline within 7 days. This measure captures "hidden AKI" - kidney injury masked by fluid dilution that would be missed using standard uncorrected creatinine. | Acute kidney injury (AKI) assessed using KDIGO creatinine criteria applied to dilution-corrected postoperative serum creatinine. Creatinine is corrected for hemodilution from perioperative fluid retention using the formula: Corrected Creatinine (mg/dL) = Measured Creatinine × (1 + Net Fluid Balance / Total Body Water) Where: * Net Fluid Balance (L) = Fluid inputs - urine output - blood loss - other outputs * Total Body Water (L) = 0.6 × weight (kg) for males; 0.5 × weight (kg) for females |
| Total intra-operative packed red blood cells administered (units transfused) | intraoperative | Provider administration of packed red blood cells during the intra-operative period, measured as total units transfused. |
| Total intra-operative fresh frozen plasma administered (units transfused) | intraoperative | Provider administration of fresh frozen plasma during the intra-operative period, measured as total units transfused. |
| Total intra-operative platelets administered (units transfused) | intraoperative | Provider administration of platelets during the intra-operative period, measured as total units transfused. |
| Total intra-operative cryoprecipitate administered (units transfused) | intraoperative | Provider administration of cryoprecipitate during the intra-operative period, measured as total units transfused. |
Countries
United States
Contacts
CONTACTAndrew Bishara, MD
PRINCIPAL_INVESTIGATORAndrew Bishara, MD
University of California, San Francisco
Outcome results
None listed