Early Nephrology Consultation Alone Based on AI Risk Score Fails to Reduce AKI Severity: JAMA
USA: A randomized clinical trial has found that early nephrology consultation (ENC), triggered by a machine-learning risk score for severe acute kidney injury (AKI), did not reduce the rise in serum creatinine (SCr) among hospitalized at-risk patients compared with usual care.
A major limitation was that most nephrology recommendations were not implemented, which may have reduced the intervention's effectiveness. The findings, published in JAMA Network Open, suggest that early consultation alone is insufficient to improve AKI outcomes, and future studies should focus on improving adherence to specialist recommendations.
Acute kidney injury (AKI) is a frequent and serious complication in hospitalized patients, but it remains uncertain whether early nephrology involvement can improve outcomes in those identified as high risk by artificial intelligence.
To address this, researchers led by Matthew M. Churpek from the Department of Medicine, University of Wisconsin–Madison, conducted a randomized clinical trial at the University of Chicago involving 180 hospitalized adults identified by the ESTOP-AKI machine-learning model as being at high risk for severe AKI but without evidence of AKI at enrollment.
Participants were randomly assigned to receive either a structured early nephrology consultation or usual care. The primary outcome was the peak change in serum creatinine over seven days, while secondary outcomes included AKI incidence, need for kidney replacement therapy, hospital readmission, and mortality.
Key findings of the study include:
- Early nephrology consultation did not significantly reduce the peak rise in serum creatinine compared with usual care.
- Rates of stage 1 or higher AKI and stage 2 or higher AKI were similar between the two groups.
- There were no significant differences in the need for kidney replacement therapy, 90-day readmission, or 90-day mortality.
- Nephrologists made substantially more treatment recommendations in the early consultation group than in the usual care group.
- However, recommendations involving medication adjustments, fluid or diuretic management, and vasopressor use were implemented less frequently in the early consultation group than in the usual care group, potentially limiting the intervention's impact.
The researchers found that the machine-learning model accurately identified patients at high risk of severe AKI, but early nephrology consultation did not improve kidney outcomes, largely because specialist recommendations were often not implemented.
They noted that the single-center design may limit the generalizability of the findings and that although the COVID-19 pandemic delayed recruitment, it did not appear to affect the results.
The authors concluded that while machine learning can effectively identify high-risk patients, early nephrology consultation alone is unlikely to improve outcomes without better adherence to specialist recommendations. They called for future studies to evaluate strategies that improve the implementation of evidence-based nephrology care.
Reference:
Churpek MM, Fatima A, Anjorin O, et al. Early Nephrology Consultation and Acute Kidney Injury in Hospitalized Patients: A Randomized Clinical Trial. JAMA Netw Open. 2026;9(7):e2622554. doi:10.1001/jamanetworkopen.2026.22554
Next Story