Electronic Acute Kidney Injury Alert at the Brandenburg Medical School: Implementation and Follow-Up

Abstract

Introduction: Acute kidney injury (AKI) substantially worsens the prognosis of hospitalized patients worldwide. In order to optimize early AKI recognition and therapeutic intervention, AKI alert systems have been implemented and evaluated in the past. Herein, we aimed to analyze outcome variables of AKI patients under the conditions of a de novo-established AKI alert system at the Brandenburg Hospital of the Brandenburg Medical School. Methods: Automated e-mail messages were generated and sent to the nephrologist with responsibility based on an electronic algorithm. The message was exclusively generated if one of the two first KDIGO criteria was fulfilled. During period 1, all alerts were ignored. During the second period, every alert was followed up, coupled with therapeutic management of respective individuals according to an AKI care bundle. Endpoints were in-hospital death, need for dialysis, and renal recovery. Results: In periods 1 and 2, 200 and 112 patients were included. In period 1, 150 out of 200 AKI alerts were identified as correct (75%); in the second period, 93 out of 112 AKI alerts were accepted as correct (83%) (p = 0.16). Kidney replacement therapy was initiated in 21 (14%) of all period 1 patients and in 32 (34.4%) of the period 2 patients (p = 0.017). In-hospital mortality of affected patients was 24 (16%) in period 1 and 21 (22.5%) in period 2 (p = 0.4). Restoration of kidney function was 69 (46%) in period 1 and 45 (48.3%) in period 2 (p = 0.71). Conclusions: We finally conclude that an AKI alert system, as implemented and followed-up in our study, did not significantly improve clinically relevant endpoints in AKI patients. Potential weaknesses were the lack of documentation of the time between receiving the alert and patient contact, and physicians in responsibility were not particularly informed about the alert system.