| Summary: | Background
Acute Kidney Injury AKI occurs in up to 15% of hospital admissions , and is associated with an increased risk of death1: some 40,000 excess deaths/year in England are attributed to AKI1. It is also associated with an increased length of stay2, and the development of chronic kidney disease3. This is expensive- one recent economic analysis estimated the associated healthcare costs for England to be over £1bn/year2. Whilst poor outcomes relate in part to patients’ underlying disease states, much of this impact appears causally attributable to AKI itself4. It may be possible to halt its progression and prevent its clinical sequelae with timely and appropriate intervention5. Recent drives to reduce ‘avoidable harm’ in hospitals are underpinned by both early identification of deteriorating patients, and rapid and appropriate intervention. Despite this, the 2009 NCEPOD enquiry reported delays in the recognition of AKI to be commonplace, and care of patients with established AKI to be “good” less than half of the time6.
The potential for reduction of patient morbidity, mortality and healthcare costs are thus clear. AKI is a priority workstream in the “reduction of avoidable harm” and “patient safety” domains of the NHS Outcomes Framework, and has been identified as an area of focus for reducing avoidable mortality in the NHS “Five Year Forward View”7. NHS England have produced an algorithm for the identification and stratification of AKI, mandating that it be incorporated into all hospital Laboratory Information Management Systems by Spring 20158. However, this is likely to have minimal impact in its current form; a recent RCT testing the efficacy of simple e-alerting for AKI resulted in no significant differences in patient’s peak creatinine, requirement for dialysis or risk of death9. Delivering improvements in outcome will depend on a more complex intervention that accounts for the diverse nature of the patient group and ensures that clinicians adhere to “best practice” guidelines.
Aims
We have used the NHS England algorithm to create a cloud-based software platform that will detect cases of AKI within milliseconds of creatinine being assayed. This novel platform produces a patient-specific, data-rich report in the same timeframe. Relevant blood tests are displayed graphically, and available biochemistry and imaging results are used to produce decision support information to help guide further investigation and management. This technology represents a substantial advance on existing systems of automated AKI alerting. An example report using dummy data is appended.
The report is automatically sent via secure mobile device to a designated response team, comprised of critical care outreach nurses. Cases in which specialist input is required (e.g. severe AKI or where blood tests suggest parenchymal kidney disease) are identified and flagged to the on-call nephrology team, automating the referral process for much of the caseload. The report will be used to drive a timely protocol-driven intervention, the individual strands of which map to the principle causes of AKI. As occurred with the “Sepsis Six” education programme and care protocol, we aim to make “best practice” routine10.
We believe that this combination of enhanced detection, analytics, automated referral and early therapy will improve outcomes for patients developing AKI in hospital. Feasibility and stakeholder engagement work is now underway; a full pilot study at The Royal Free London NHS trust will then run for one year, data from which will be used in the design of a larger randomised trial.
Conclusions
The ability to provide rich, patient-specific information to clinicians is one of the primary incentives to developing expensive clinical information systems. To date, efforts to use automated alerting for AKI have largely failed. We have developed a software platform and allied care bundle that directly addresses existing deficiencies in care and the difficulties encountered in previous trials. Specific understanding of organizational management and behaviour change will be required to maximize usability and facilitate integration into clinician workflow11.
Disclosures
VN is supported by the Medical Research Council in the form of a clinical training fellowship. No conflict of interest is declared.
|
|---|