SaO007IMPLEMENTATION OF AN AUTOMATED PRIMARY CARE ACUTE KIDNEY INJURY (AKI) WARNING SYSTEM: A REVIEW OF 2 YEARS EXPERIENCE

Abstract

Introduction and Aims: Acute kidney injury (AKI) is associated with increased mortality and the development of CKD. Early detection and effective intervention are required to improve outcomes. AKI clinician-alert systems following detection of a rise in serum creatinine (sCr) have seen increased use in secondary care and are now being implemented nationally in England, with plans for subsequent primary care adoption. In 2012 we pioneered an automated alerting system for primary care clinicians (PCCs). A sCr value >= 1.5X the previous value generates an automated message that reports in the primary care electronic results system. This is linked to clinical guidance and the PCC receives an advisory phone-call from laboratory staff. We have retrospectively analysed the patient demographics and outcomes of community generated AKI alerts and qualitatively evaluated the system through a survey of all the PCCs using our laboratory services. Method(s): We analysed data on all patients who triggered an AKI clinician-alert from community blood tests between June 2012 and June 2014.We retrieved their sCr, urea and eGFR results between August 2011 and November 2014 to establish actual baseline sCr, comorbidity, alert validity, KDIGO stage, AKI incidence, AKI on CKD and renal recovery. The cause of AKI, action taken as a result of the AKI-alert, length of stay (if admitted), 3-month post-alert mortality were obtained from our hospital records or the patient's PCC. A qualitative survey was sent to all PCCs to evaluate awareness of the alerting system, satisfaction, utilisation of services and actions taken when an AKI alert is received. Result(s): 142 AKI alerts were generated in the 2-year period of study. The mean age was 69 years (20-96) with 54% being female. 64% were KDIGO stage 1, 27% stage 2 and 9% stage 3. 20% previously had AKI alerts with 25% going on to have future AKIs. 3-month mortality was 12%, with the majority of these patients having a severity of >=KDIGO 2. 78% of AKIs were due to pre-renal causes with renal and post-renal disease causative in 11%. 43% of patients with AKI alerts had not returned to baseline renal function (within 20% of baseline sCr value). Of the data analysed 56% of the AKI alerts resulted in admission with a 10 day average length of stay. 15% were referred for hospital follow-up and 29% received community AKI care or were monitored. 86% of PCCs surveyed were aware of the AKI alerting system with 84% finding it a "useful service that has altered practice". 56% of PCCs referred their patient into hospital services via the acute take or outpatient services whilst a further 44% saw their patients themselves. 55% of PCCs administered AKI care in the community as a result of the alerting system. Conclusion(s): As far as we are aware, this is the world's first implementation of an automated primary care AKI clinician-alert and this system has now been alerting community clinicians for nearly 3 years. The community AKI alerting system is a novel, valued and useful service which has triggered community interventions such as rapid PCC review, community care or hospital admission. This system has been sequentially scaled in with linked advice to 4 other acute hospitals in our sector. This experience should inform further quality improvement developments in community AKI management and strategies for national implementation.