March 29, 2018Clinical Journal of the American Society of Nephrology
Identification of Patients Expected to Benefit from Electronic Alerts for Acute Kidney Injury
Key result
Among patients identified by uplift modeling as likely to benefit, electronic alerts for acute kidney injury significantly reduced the maximum relative change in creatinine by a median of 25.3% compared to usual care.
Why the study?
Does an automated, electronic alert system improve creatinine trajectories in adult patients with AKI?
Population
2278 adult patients with AKI
Comparison
Automated, electronic alert system vs Usual care
Design
Cohort, randomized
Follow-up
3 days after randomization
Authors
Discussion
Loading...
Member takes
Implication
Supports targeted AKI alerts to improve creatinine trajectories; hypothesis-generating pending randomized validation.
Study Design
Type
RCT (n=2,278)
Randomization
1:1
Multicenter
No
Structured PICO
Does an automated, electronic alert system improve creatinine trajectories in adult patients with AKI?
P
Population
2,278 hospitalized adults with KDIGO stage 1 acute kidney injury, followed for 3 days post-randomization to evaluate the effect of targeted electronic alerts.
I
Intervention
Automated, electronic alert system
C
Comparator
Usual care
O
Outcome
Maximum relative change in creatinine from the time of randomization to 3 days after randomizationsurrogate
Main Result
Effect estimate: Median difference -25.3% (95% CI -29.9 to -20.06)
Absolute Event Rate: -2.6% vs 7.9%
p-value: p=0.03
Uplift modeling can successfully target electronic alerts for AKI to those most likely to benefit, improving creatinine trajectories and potentially reducing alert fatigue.
Limitations
- Retrospective nature of the secondary analysis
- Model may be specific to the alert and healthcare system in which it was developed
- Did not model provider characteristics
- Primary outcome focused on change in creatinine is not as clinically meaningful as rate of dialysis or death
- Single imputation used to account for missing variables
Cite This Study
Biswas et al. (2018) conducted an RCT in Acute Kidney Injury (n=2,278). Electronic alerts for AKI targeted via uplift modeling vs. Usual care was evaluated on Maximum relative change in creatinine from time of randomization to 3 days post-randomization (Median difference -25.3%, 95% CI -29.9 to -20.06, p=0.03). Among patients identified by uplift modeling as likely to benefit, electronic alerts for acute kidney injury significantly reduced the maximum relative change in creatinine by a median of 25.3% compared to usual care.