Automated AKI Alert System Improves Patient Outcomes

Acute kidney injury (AKI) is associated with loss of kidney function and increased risk of mortality, yet there is no established universal treatment for AKI. Patients with AKI are treated with an individualized approach and avoidance of nephrotoxic events. It is clear that early detection of AKI events is key to improving outcomes; the role of the nephrologist is also crucial in treating patients experiencing AKI.

AKI is largely defined by criteria based on serum creatinine. Serum creatinine is widely tested and can be reported quickly using comprehensible numeric values. There have been attempts to utilize serum creatinine measurements to create an efficient surveillance system for AKI, known as an AKI alert. However, results of the only published randomized controlled trial of AKI alerts were negative; in addition, despite promising preliminary outcomes other studies had several limitations. A change in clinicians’ behavior was the most common assumption needed for the success of an alert system.

Clinicians in Seoul, Korea, implemented an AKI alert system on June 1, 2014, at a tertiary referral hospital with >1000 general admission beds. In that system, the attending clinicians could easily generate automatic direct consultation with the nephrology division. Sehoon Park, MD, and colleagues reported results of an assessment of the system using a comparison of outcomes of patients with AKI events prior to and following launch of the system in the American Journal of Kidney Diseases [2018;71(1):9-19].

Primary outcomes of interest were overlooked AKI events, defined as not measuring the follow-up creatinine value, and the consultation pattern of clinicians. Secondary outcomes were severe AKI events; AKI recovery, defined based on the creatinine-based criterion; and patient mortality.

The system defined the baseline serum creatinine concentration level as the minimum level within 2 weeks prior to admission date; when there was no earlier laboratory value available, the baseline level was the first serum creatinine concentration measured during the hospitalization. The system screened AKI events, defined as serum creatinine concentration elevation of at least 1.5-fold or 0.3 mg/dL from baseline, every midnight. When the physician opened the electronic medical record (EMR) the following morning, a pop-up window displayed the following message: “(Mild/Moderate/Severe) acute kidney injury (stage X), Do you want to send a consultation request to nephrology division?”

There were 21,554 index admission cases in the historical cohort and 25,487 in the study cohort. Following exclusion of cases of death on the day of AKI, there were 1739 patients with AKI events following the launch of the alert system. Following the same criteria, there were 2189 patients identified as having AKI events in the historical cohort. After applying additional exclusion criteria, the final cohort included 1884 patients in the usual-care group and 1309 patients in the alert group.

The two groups had significant differences in baseline characteristics. Compared with patients in the usual-care group, patients in the alert group were older, more likely to be male, and more likely to have had surgery during the hospitalization (P<.001 for all), have poorer kidney function at baseline (P<.001), and lower hemoglobin (P=.002) and albumin (P<.001) concentrations. The medical histories of patients in the alert group also included higher rates of hypertension (P<.001), diabetes mellitus (P<.001), ischemic heart disease (P=.001), and heart failure (P<.001). Patients in the alert group were more likely to use diuretics (P<.001), while patients in the usual-care group were more likely to use nonsteroidal anti-inflammatory drugs (P<.001). Patients in the alert group developed AKI earlier and were hospitalized longer following the AKI event (P<.001 for both).

Following implementation of the alert system, the odds of an AKI event being overlooked decreased significantly (adjusted odds ratio [OR], 0.40; 95% confidence interval [CI], 0.30-0.52; P<.001). There was also a marked reduction in the odds of an AKI case not receiving a nephrologist consultation (adjusted OR, 0.23; 95% CI, 0.19-0.29; P<.001); the odds of a late consultation, defined as >3 days from AKI, were similar in both groups. The rate of early consultation increased more than 4-fold (adjusted OR for early consultation, 6.13; 95% CI, 4.80-7.82; P<.001).

In analyses of characteristics associated with AKI, patients in the alert group had lesser odds of a severe AKI event (adjusted OR, 0.75; 95% CI, 0.64-0.89; P<.001) compared with patients in the usual-care group. In addition, patients in the alert group demonstrated an improved recovery rate compared with those in the usual-care group, and the calculated number to treat was 4.03. There were significant interactions between the effect of the alert system and the presence of decreased kidney function (estimated glomerular filtration rate <60 mL/min/1.73 m2; P<.001), sex (P=.007), and medical history of cancer (P<.001). The alert system was associated with facilitation of AKI recovery in every subgroup tested; the benefit was greater in patients with relatively reserved kidney function, admission to nonsurgical departments, and no history of cancer; the benefit was also greater in women.

There were no statistical differences between the two groups in survival rate. The effect of time period also did not show significant interaction with the effects of implementation of the alert and AKI recovery rate was not significantly different according to quarter-years in both the usual-care and alert groups.

Study limitations cited by the authors included the trial design not being randomized; the inability to record AKI events in real time; not establishing criteria for baseline serum creatinine concentrations; possible selection bias because the alert group was admitted for longer periods and had worse baseline characteristics; and finally, the possibility that the results are not generalizable to other hospitals depending on size or location.

“In conclusion, our EMR-based AKI alert system altered the behavior of clinicians, increased the involvement of specialists, and improved AKI outcomes. Therefore, adoption of an AKI alert system linked to early nephrology intervention could be considered in hospitals to improve patient prognosis,” the authors said.

Takeaway Points

  1. Researchers in South Korea conducted an assessment of an acute kidney injury (AKI) alert system with a before-and-after quality improvement study.
  2. The system utilized automatic messages sent to clinicians each morning following measurement of patients’ serum creatinine concentrations, allowing the physician to generate automated consultations to the nephrology division for all hospitalized patients.
  3. The system significantly lowered the odds of overlooked AKI events and increased the odds of an early consultation with a nephrologist. The likelihood of AKI recovery was also improved with implementation of the alert system.