Effects of nationwide alerting for acute kidney injury on healthcare and patient outcomes: population based, regression discontinuity analysis
Bibliographic record
Abstract
Abstract Objective To determine whether the implementation of electronic alerts for acute kidney injury improves care and outcomes in real-world clinical practice. Design Population-based, regression discontinuity analysis Setting Hospital and community-based systems across all 8 health boards in Wales, from 2016 to 2021 following implementation of AKI alerts. Participants 3.1 million adults aged ≥18 years resident and registered at GP practices in Wales between the years of 2016-2021, following implementation of AKI e-alerts across all 8 Wales health boards. Exposure Electronic alerts for acute kidney injury provided in either a passive (7 boards) or interruptive (1 board) manner to physicians through laboratory reporting systems. Main outcome measures Mortality, hospital admission or readmission, severity and recovery of AKI, coding of acute kidney injury, proteinuria and blood pressure measurements. Results Among 861,494 and 354,505 eligible patient encounters, 5.8% and 2.0% AKI alerts were generated from hospital and community settings, respectively; mean age 64 years, 54% female. In both hospital and community settings, respectively, AKI alerts led to no significant differences in mortality [complier average treatment effect +1.31% (95% CI −3.07, 4.74); +2.07% (95% CI −3.44, 6.65)] or admissions/readmissions [+0.13% (95% CI −3.82, 4.21); +4.07% (95% CI −1.84, 8.27)]. There was a modest increase in hospital coding of AKI with alerts [+5.88% (95% CI 2.22, 7.58)], but no difference in primary care coding of AKI after discharge [+0.72% (95% CI −0.67, 1.30)]. Alerts exerted small increases on subsequent checks for proteinuria and of blood pressure. Findings were consistent for passive and interruptive alerts. There were no meaningful differences by rurality, deprivation, sex, history of surgery, diabetes, or vascular disease. Conclusions The nationwide implementation of AKI alerts in Wales produced small effects on documentation of AKI and some processes of care but exerted no effects on survival or hospital admissions in acute care or community settings. The consistently poor outcomes, and the deficiencies in documentation and care after AKI highlight the ongoing need for an improved clinical response. Registration A prespecified analysis protocol is available at Open Science Framework (OSF) repository ( https://osf.io/f4wz9/?view_only=d20e49817bc74f1fa66afc7aa8dda0ab ) Summary What is known Previous randomised trials and real-world observational studies of electronic alerts for acute kidney injury (AKI) have produced mixed results. The applicability of previous trials for real-world clinical practice remains uncertain and contested. AKI e-alerts are still widely used. What we did We applied Regression Discontinuity Design (RDD), a method gaining traction in clinical research. RDD can work like an ideal target trial to enable causal inference in real-world settings. RDD is particularly effective when clinical actions are triggered by a specific measurement threshold. This approach allowed us we to evaluate a nationwide AKI e-alert initiative across Wales. What we found We found no evidence that AKI e-alerts improved or worsened outcomes across any clinical setting, patient subgroup, or alert delivery method. Nonetheless, the consistently poor outcomes, and the deficiencies in documentation and care after AKI highlight the ongoing need for an improved clinical response.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.062 | 0.123 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.008 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.004 | 0.004 |
| Research integrity | 0.002 | 0.005 |
| Insufficient payload (model declined to judge) | 0.005 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".