Electronic Alerting Tool to Help Prevent Acute Kidney Injury

'Preventable Acute Kidney Injury (AKI) Should Never Occur.' The Use of a Novel Electronic Prediction Alerting Tool to Deliver an Individualised Care Package for Hospital In-patients at Risk of AKI

Around a third of patients who develop acute kidney injury (AKI) do so after a hospital admission (hospital-acquired - HA-AKI). The primary aim of the study is to prospectively test whether introducing a complex intervention (a 'care package' - comprising a clinical prediction rule incorporating an electronic alert which generates a checklist for patient management to relevant health professionals) can identify patients on admission to hospital who are at risk of developing HA-AKI, highlight the need for closer monitoring and allow putative preventative measures to be put in place. The investigators will introduce the care package in one acute hospital and evaluate its effectiveness in reducing HA-AKI and its associated morbidity, over ten months, compared to a sister hospital within the same Trust (which will act as a control site). The investigators will extend evaluation for a further ten months to assess sustainability on the first site and introduce the package at the control hospital to assess generalisability. The primary aim is reducing HA-AKI, but secondary aims will include improved outcomes associated with HA-AKI, management of patients already with AKI on admission to hospital (whose care may also benefit from the checklist) and a cost-effectiveness analysis.

Acute Kidney Injury (AKI) is common in hospital (incidence of 10-20% - up to 70% in the critically ill), with high associated morbidity and mortality. Even small changes in renal function are associated with increased mortality. The 2009 National Confidential Enquiry into Patient Outcome and Death examined the care of patients who had died in hospital with a primary diagnosis of AKI. Over 40% of cases had an unacceptable delay in diagnosis and in 20% of cases, AKI was thought to be predictable and avoidable. Electronic alerts have been studied for patients with established AKI, however, they have highlighted rises in creatinine after insult rather than identifying patients at risk of AKI - a third of hospital AKI cases occur after admission (HA-AKI). Risk factors have been reported in surgical and burns patients. However, strategies to identify patients admitted as medical emergencies at risk of developing AKI are lacking - the group accounting for most Intensive Care admissions with AKI. The investigators multidisciplinary team, with significant experience utilising technology in healthcare, have developed a novel prediction score - Acute Kidney injury Prediction Score (APS). Utilising physiological measurements, biochemical parameters and known co-morbidities, the APS identifies patients at risk of developing HA-AKI following admission (1/3 of all AKI cases). A 'care package' has been devised incorporating the APS into an automated electronic algorithm to send realtime alerts to staff on the Observation chart, e-mail the patient's Consultant and advise on a checklist. Alongside this, an E-learning AKI module for ward staff has been developed building on NICE Guidance with additional information regarding the APS. Aims Primary: investigate whether introducing a 'care package' can reduce HA-AKI in patients admitted to hospital as an emergency. Secondary include determining whether the intervention: reduces associated complications; improves outcomes in patients with AKI on admission; and is cost-effective. Research Questions Primary: can a 'care package' by systematically recognising the 'at risk' patient, alerting and prompting management to staff educated in the problem, reduce HA-AKI? Secondary: Can harms associated with AKI be reduced? Is the intervention acceptable to staff and are there barriers to implementation? Are improvements sustainable and can the intervention be successfully applied to a second hospital? Does the intervention improve outcomes of those with AKI on admission? Is the intervention cost-effective? Design A prospective, non-randomised, parallel cohorts study, with before-after trial periods, at intervention and control hospital sites, will be performed. A run-in phase (10 months) for baseline data collection and prospective external validation is followed by the intervention on one site (Worthing) with the Chichester site acting as control (Phase 1) for 10 months. The intervention will then be introduced at Chichester whilst continuing at Worthing (Phase 2) for 10 months. The additional period will allow analysis of the interventions' impact on readmissions. Potential benefits: Preventing morbidity and mortality including secondary complications such as chronic kidney disease (immediate and longer term). Reducing length of stay, thus reducing potential exposure to harm. Prevent requirement for renal replacement therapies and escalation to Intensive Care with associated morbidity, mortality and psychological harm. A Cost-effectiveness analysis will aim to demonstrate whether such a strategy could provide savings locally and to the wider health economy (short and long-term). Inform the healthcare community on applicability of information technology to benefit patients and improve staff engagement, with potential utilisation in a number of other conditions.

Patients identified as high risk of AKI by the electronic clinical prediction model will be managed with a care bundle of best practice.

HA-AKI will be defined as per KDIGO change in serum creatinine i.e. a ≥26.4μmol/L increase within a 48 hour period (during the first 7 days of admission to hospital) or a 1.5 times increase vs the admission result within the first 7 days of admission to hospital.

At any time point during the admission under analysis i.e. from admission to either discharge from the hospital or death in-hospital, participants will be followed for the duration of hospital stay, expected average of 7 days.

During the index hospital admission. Each participant will be followed for the duration of hospital stay, an expected average of 7 days.

AKI on admission defined as at least 1.5 times baseline serum creatinine or ≥354μmol/L without a baseline. Baseline defined using NHS algorithm.

During the index hospital admission. Each participant will be followed for the duration of hospital stay, an expected average of 7 days.

This will be measured in both KDIGO stage: increase from Stage 1 to Stage 2 (200% increase in serum creatinine) or Stage 3 (300% increase in serum creatinine) and also peak mean increase in serum creatinine.

During the hospital admission. During the index hospital admission. Each participant will be followed for the duration of hospital stay, an expected average of 7 days.

Inclusion Criteria: Admission as an emergency Spending at least one night as an in-patient Exclusion Criteria: Patients under 18 Patients not admitted as emergencies or staying less than one night in hospital.

Hodgson LE, Roderick PJ, Venn RM, Yao GL, Dimitrov BD, Forni LG. The ICE-AKI study: Impact analysis of a Clinical prediction rule and Electronic AKI alert in general medical patients. PLoS One. 2018 Aug 8;13(8):e0200584. doi: 10.1371/journal.pone.0200584. eCollection 2018. Erratum In: PLoS One. 2018 Aug 23;13(8):e0203183.