“One Piece at a Time”: The Cache of Acute Kidney Injury Data in the Electronic Medical Record*

    loading  Checking for direct PDF access through Ovid

Excerpt

Acute kidney injury (AKI) is associated with poor ICU outcomes in all patient groups studied and has been identified as an independent risk factor for PICU mortality, longer length of stay, and longer length of mechanical ventilation (1–3). Yet despite these risks, early interventions to ameliorate the burden of AKI have been hampered by difficulties in early identification.
Until recent years, there was a lack of a consensus definition of AKI, which was previously a variable definition based on threshold creatinine levels. The first consensus definition was the Risk of renal dysfunction, Injury to the kidney, Failure of kidney function, Loss of kidney function, and End-stage renal disease (RIFLE) criteria model published in 2004 by the Acute Dialysis Quality Initiative Group and featured a change in serum creatinine (SCr) relative to the patient’s own baseline (4). Akcan-Arikan et al (2) created a modified pediatric version (pRIFLE) that both identified and classified the severity of AKI in critically ill children. Most recently, the Kidney Disease Improving Global Outcomes classification system has been devised and validated in subsets of patients (1, 5). These consensus definitions use a change in creatinine and urine output (UOP) as diagnostic variables with categorizations of disease based on threshold change and/or the need for renal support therapy. Although creatinine is a universally available test and changes in SCr are measurable in an electronic medical record (EMR), it has issues as a sole measurement of renal function. A rise in SCr may not be detected until 48 hours after the glomerular filtration rate (GFR) has fallen, and because creatinine is both filtered and secreted the measure may overestimate GFR (6). SCr can also be affected by nutrition and muscle mass, which makes it a difficult biomarker to use given the heterogeneity of the pediatric population (7). Changes in SCr require a baseline creatinine level for comparison, which may not be available in up to 83% of noncritically ill patients (8). Although UOP measures have been included in AKI consensus definitions, they have not added substantially to the SCr definitions. More importantly, although fluid overload (FO) has been associated with AKI, it has not yet been incorporated into AKI classification systems which include UOP (8).
The widespread use of EMRs was touted as a way to provide real-time patient safety alerts and has allowed the development of multiple predictive models using patient data. Given the difficulty with early AKI recognition, several groups have sought to create EMR-derived alert systems with the goal of encouraging earlier intervention or prevention for AKI. Several alert systems for adults and children have been devised using SCr changes as a marker of AKI and as the basis for provider alert, but with variable success (9–11). Alerts based on nephrotoxic medication exposure have also been reported with some success (12, 13). One group has even created an AKI alert that is independent of SCr measures, but none have included FO in the alert system (14). To date, no alert has been studied that combines changes in SCr, FO measurement, and exposure to nephrotoxic medications.
In this issue of Pediatric Critical Care Medicine, Akcan-Arikan et al (15) report their experience in the creation of a novel EMR-derived decision support tool called the “Fluid Overload Kidney Injury Score” (FOKIS). FOKIS is a compilation of variables that are associated with AKI, including changes in SCr, FO status, UOP, and the use of nephrotoxic medications. SCr changes and UOP changes are based on the pRIFLE criteria with three categories of AKI assessment totalling six maximum points.

Related Topics

    loading  Loading Related Articles