Surgery-specific patterns of perioperative amino acid administration and associated acute kidney injury risk: a large-scale retrospective cohort study
Authors' conclusion
Perioperative amino acid administration demonstrates surgery-specific patterns with differential AKI associations.
Study context
- Population
- Adults undergoing surgery with ≥ 24-h stays, INSPIRE database, Seoul National University Hospital, 2011–2020
- Participants
- 22,972 adults
- Exposure
- Perioperative amino acid administration vs none
- Amino acid usersn = 899
- Outcome
- AKI incidence (KDIGO criteria) and amino acid utilization patterns across surgical departments
Methodological detailssingle-centre database (INSPIRE), propensity score matching, machine learning analysis
- Single-centre database (INSPIRE)
- Propensity score matching
- Machine learning analysis
Main finding
As reported in the abstract
Amino acid use
3.9%
Overall acute kidney injury (AKI) incidence
3.7%
Utilization varied 60-fold across departments
0.2-11.5%
Observational data: associations, not effects.
Further findings
| Outcome | Amino acid users | Non-users | RR (95% CI) | P value | |
|---|---|---|---|---|---|
| AKI, non-cardiac surgery (subgroup) | 7.4% | 3.4% | 2.16 (1.65–2.85) | P < 0.001 | |
| 0.20.5125Relative risk (log scale) |
| Outcome | Amino acid users | Non-users | P value |
|---|---|---|---|
| AKI, cardiac surgery (subgroup) | 0/50 | 4.2% | P = 0.267 |
Amino acid use as independent AKI predictor (multivariable analysis)
- Multivariable OR for AKI with amino acid use
- 2.01 (95% CI 1.52-2.60)
Machine learning prediction of AKI
- Amino acids the strongest AKI predictor; Random Forest AUC-ROC
- 0.782
- Significant non-linear interactions
AKI with amino acid use, propensity score matched
- PSM OR for AKI with amino acid use
- 1.63 (95% CI 1.05-2.52, p = 0.029)
- 799 matched pairs
Original publication
Surgery-specific patterns of perioperative amino acid administration and associated acute kidney injury risk: a large-scale retrospective cohort study. Perioper Med (Lond). Published 12 August 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →