AKI Detection at Barking, Havering and Redbridge NHS Trustresults to primary care. AKI alerts went live in our Trust in November 2013 for all creatinine requests. Details of AKI detection

AKI Detection at Barking, Havering and Redbridge NHS Trust

Funmi Akinlade ([email protected]), Ajith James, Peter Ayling, Sarah de Freitas

Hospital settingBHRUHT serves a population of about 700,000 and has two hospital sites, King George and Queen’s hospitals. Each hospital has an A&E department and a full range of local hospital services including pathology laboratories on both sites. The laboratory software is Clinisys Winpathand the results lookup system is Clinisys Cyberlab. The Trust does not have an order communication system.

Details of AKI detection algorithmA creatinine baseline test code was created (non reportable) alongside all creatinine requests. Figure 1: Rule flow diagram

We designed algorithms to alert the possibility of AKI. 1. A “baseline “ is createdThe “baseline creatinine” is chosen as the previous “baseline creatinine” (in the last 120 days) or the previous creatinine if taken within the last 365 days. If there were no previous blood tests then the creatinine result on the current blood test is taken as the “baseline creatinine”. Note: Baseline creatinine cannot be below 40 umol/L. 15 – 16 % of creatininesdid not have a baseline creatinine.

2. Comparison of current and “baseline creatinine” The current creatinine is compared to the “baseline creatinine”. The percentage difference is calculated and if > 50% a comment is reported with the result. The comment is “Could this patient have AKI – stage 1/2/3?” * For stage 3 referral to a Nephrologist prompt is included in the messageDialysis patients are excluded from the alert system.

3. Resetting the baseline In some instances the creatinine result chosen to be the patient’s “baseline” is inaccurate as the patient may have been acutely unwell at presentation. Our algorithm resets the “baseline creatinine” when the creatinine decreases on subsequent samples. This therefore ensures the value chosen to be the “baseline creatinine” is as useful as possible.

4. Patients presenting with an elevated creatinine In patients presenting acutely with no previous results and a creatinine > 120 µmol/L in females or > 150 µmol/L in males (corresponding an eGFRof ~45 ml/min/1.73 m2) a) this comment is reported “This patient has an elevated creatinine and may have CKD or AKI”

Algorithm cont.4. Patients presenting with an elevated creatinine (Cont)b) If the patient’s creatinine decreases over subsequent days with appropriate treatment (i.e. it was an AKI as opposed to CKD) they get the retrospective diagnosis message “the patient may have had AKI” and the baseline creatinine resets (i.e. a drop in creatinine of > 33%).

Details of AKI alertAKI alerts are reported alongside the creatinine result. All stage 3 comments are phoned to the ward/clinicians. These comments do not require acknowledgement by end users.

Figure 2: Screen shot showing AKI comment alongside creatinine result.

The laboratory performs 5-6000 creatinine analysis per week. We report ~ 170 alerts per week on approximately 110 individual patient, 20-25 patients have more serious AKI.

Successes Referrals to Nephrology have increased by 15% and are earlier. Nephrologists are also able to use the portal to pre-emptively review results for patients who have not been referred. A portal was also created for pharmacy to enable them to act on results prior to referral. Initially this worked well however these have not been used very effectively in recent months due to workload.

The AKI detection alerts were introduced with engagement with the nephrology team. Education of clinicians and laboratory staff were essential to implementation. Educational sessions included talks at grand rounds and departmental teaching sessions.

The Trust will be employing an AKI nurse for an initial 3 month period to target patients identified with AKI stages 2 and 3.

Challenges Alerts were implemented at BHRUHT, however without a system to ensure that these are acknowledged or that there is some ownership for the follow up of alerts by a particular team it can be difficult to assess their impact.

Peculiarities of the IT system mean that the alert system is not perfect, hence the comments reported are suggestive of AKI, requiring the clinician to take responsibility for the diagnosis.

Other outputsWe have been able to easily undertake regular ‘real time’ snapshot audits for all stages. This information helps us engage colleagues.

We have been able to look at patient outcomes for AKI, confirming that the odds ratio for mortality was 10 x greater and on average length of stay longer compared with patients without AKI.

Sample received

Check for prev. baseline in last 120 days

Use prev. baseline No BL

& CR > 40

Check for prev. creat. in last 365 days

Use prev. Creat.

If Pt is inpat.or A&E and > 17 yearsold then add CICR (change in creat.)

If creat > 40 & CRBL > 40 & CICR exists then calculate CICR

Staging done and comment added

No BL , female Pt & CR > 120

No BL , male Pt & CR > 150

Current CR < BL

If recent CICR in 90 days

not inpat or A&ECR > ????

BL < CR

Raise BL to CR(flag as br)

Flag stage 3's for phoning

B/Lfound

Creat.found

No B/L found

No creat. found

Sample analysis

( Check for prev. rejected creat. results )

CICR present

No CICR – Ignore request

No comment

CICR % of 50 / 100 / 200 / -32

Comment added

Comment added

Comment addedReport issued

Update BL With current CR

No

Yes

Lower BL(and flag as bl)

No

( Basic housekeeping )

No

Yes

Yes

AKI Detection at Barnet and Chase Farm Hospitals NHS TrustAuthor: Ayrton Puja,(consultant chemical pathologist),email:[email protected]

Contributors:Imtiaz Gilani (Pathology Systems Manager), Victoria Owoyemi, (Pre- reg Hospital Pharmacist) , Katy Heaney, (Principal Biochemist),

Dr Peter Dupont (Consultant Nephrologist), Dr Faieza Qasim(Consultant Nephrologist) Dr Phillip Jacobs (Consultant Acute Physician)

Hospital settingBarnet and Chase Farm Hospitals NHS Trust is a district general hospital.The clinical biochemistry laboratory received in 2013:Total number of test requests = 2,787,437Total Number of tests = 6,986,642Approximately 40% of total test requests were from secondary care and 60% from primary care.The total number of serum creatinine requests in 2013 was: 451283. Out of these 207590(46%) were from primary care.The name of laboratory software (LIMS) system is PathNet Cerner Millennium

supplied by Cerner Corporation.We have an integrated EPR system, our results are available in PathNet for Pathology staff, PowerChart for clinical staff, and through Indigo4 which posts results to primary care.AKI alerts went live in our Trust in November 2013 for all creatinine requests.

Details of AKI detection algorithmThe AKI flag is generated via a rule in Cerner Millennium based on results generated in the PathNet Solution.The system is set up to detect a ≥ 50% rise in serum creatinine from the most recent creatinine within the previous 3 months. If this is detected than an automated comment is added onto the serum creatinineresult and a test code called AKI flag is generated which is populated with the same comment as on the serum creatinine.The comment used is:“?AKI: ACUTE KIDNEY INJURY- Creatinine increase ≥ 50% compared to previous results.Please see guideline at www.londonaki.net/clinical for further information”

Details of AKI alertThe AKI result is visible to clinicians electronically on both the hospital/GP computer systems. The AKI alert is now part of the laboratory results to be phoned policy. The policy states the following:All AKI results for inpatients to be phoned 24/7 by the laboratory BMS staff.If the GP surgery is closed or out of hours for outpatients the AKI result is to be phoned the next day. All serum creatinine results which are > than 300 umol/L are phoned 24/7.The results require end users to acknowledge the result. Below is a screen shot of the alert on the hospital computer system powerchart:

Successes Prior to introducing AKI detection alerts clinical biochemistry rallied the support of the nephrologists in our Trust to give presentations alongside the clinical biochemistry consultant at the hospital grand round. Clinical biochemistry also sent electronic communications to the GP’s and a short communication to pop up on the hospital computer system when clinicians logged on. This communication strategy just prior to going live with the alerts ensured that people were aware of what the alerts were and what to do when they saw them.We also have a rolling educational program on AKI: delivered by an acute physician and nephrology consultants to FY1 trainees, final year medical students and core medical trainees as part of their formal lecture programmes. The GP’s had an update on AKI and the new alerts as part of their CPD programme of lectures.

Challenges The total number of alerts from 1/1/2014 till 30/4/2014 were 673- on average 168 alerts per month.From 1/1/2014 to 18/5/2014: 76% of alerts were from inpatients, 6% from outpatients, 14% from primary care, 2% from other hospitals (such as community hospitals), 1% from the local mental health trust. There were a total number of 11 alerts from dialysis patients during this period.The workload implications for nephrology, acute medicine and critical care outreach teams have not been formally quantified. However, even aiming to see all AKI stage 3 would be difficult for nephrology to deliver on a service that runs on half a day per week.Barnet is currently under-resourced from a renal point of view ( bearing in mind the NCEPOD finding about poorer outcomes being associated with lack of access to specialist input). There is currently no direct renal input to the Chase Farm site.It is anticipated that these issues will be addressed following the takeover by Royal Free Hospital in July.The IT changes required IT resources and time.A clinical audit is planned to determine the impact of the detection and management of AKI post introduction of alerts compared to a prior audit, which demonstrated that less than half of patients with AKI had a documented urinalysis result. Although review of nephrotoxic medication was good, we hope to demonstrate further improvement.

Other outputsAudit: “Appropriate prescribing of medicines in patients with AKI”.

Victoria Owoyemi

The objectives of this audit were:

To determine the percentage of patients with AKI in which drugs were

appropriately prescribed in respect to their renal function from 1st December to

14th December 2013.

To identify the groups of drugs that are commonly prescribed inappropriately if

any.

Results :

52 AKI patients were identified as being suitable for the audit. This included

patients who were admitted with AKI and those that developed it while inpatient.

AKI patients receiving palliative treatment were excluded from the audit.

Drugs are sometimes prescribed inappropriately to patients with AKI renal

impairment. Some AKI patients were prescribed at least a drug with inappropriate

dose or unsuitable for their clinical condition. In AKI, patients should not be

prescribed Metformin, NSAIDs, ACE Inhibitors, Angiotensin-II receptor antagonists

or Potassium sparing diuretics however, 7.7% of patients were prescribed one of

these drugs while 1.9% were prescribed two of the drugs.

It is planned that the alert will be visible in the ICE, Cerner under a separate test code ‘AKI Alert’. To improve visibility, the AKI alert result will be the first result in the biochemistry results table.

No results will be telephoned from the laboratory and it is not possible to prompt user acknowledgement of the result. Details of AKI2 and 3 alerts will be emailed daily to a group email address that will be accessed by renal, critical care outreach and ICU teams

AKI Detection at Barts Health NHS Trust

Lou Oliver, Emily Leach, John Prowle, Mark Blunden, Chris Kirwan

Contact: [email protected]

Hospital setting

Barts Health NHS Trust is the biggest NHS Trust in the country and comprises Barts and The London NHS Trust (BLT) legacy hospitals (The Royal London, St Bartholomew's, London Chest and Mile End hospitals), Newham University Hospital and Whipps Cross University Hospital.

Although all of the hospitals use the same LIMS (Clinisys Winpath) and ICE (Cerner Millenium), different versions are used across the sites. Currently, only the BLT legacy hospitals use the same versions (Winpath 5.32; BT Cerner Millenium) so the algorithm is being implemented in BLT legacy hospitals first, expected go-live date in Aug 14. It will be rolled out to Newham and Whipps University Hospitals at a later date.

Details of AKI detection algorithm

The detection algorithm is located in the LIMS, Clinisys Winpath (Fig. 1)

Serum creatinine result ≥ 90 umol/L

Search for last creatinine result within 6 months

Calculate ratio

[current creatinine] / [last creatinine]

Is the ratio ≥ 1.5?

Y

Is serum creatinine >354 umol/l?Alert!

?AKI 3

N

Is ratio ≥ 3.0?

N

Is ratio ≥ 2.0 and < 3.0?

N


Y

Y

Alert!

?AKI 1Y

Alert!

?AKI 2Y

Serum creatinine result ≥ 90 umol/L

Search for last creatinine result within 6 months

Calculate ratio

[current creatinine] / [last creatinine]

Is the ratio ≥ 1.5?

Y

Is serum creatinine >354 umol/l?Alert!

?AKI 3

N

Is ratio ≥ 3.0?

N


N


Y

Y

Alert!

?AKI 1Y

Alert!

?AKI 2Y

• The calculation of the baseline creatinine is limited by Winpath’s capabilities. The baseline creatinine is the last creatinine result within a 6 month period.

• Patient groups excluded from the alerting system are those registered under renal consultants on the EPR, inpatients on critical care wards or those < 16 years old.

• This is a temporary solution working within the constraints of BT Cerner Millenium. Once the Trust moves to a non-BT version of Cerner, another more advanced algorithm that is fully integrated in Cerner will be implemented.

0

5

10

15

20

25

30

35

40

45

Week 1 Week 2 Week 3

# A

lert

s

True AKI

Not AKI

22

17

20

13

17

22

The use of a ≥ 90 umol/L serum creatinine cut-off

Without the use of a cut-off, over a 3 week period (Fig. 2) of 110 total inpatient alerts 59 were false positive (46%)

Fig. 2. Proportion of true vs. false AKI

Cr > 90 Cr < 90

3 true AKI

Cr > 90 Cr < 90Cr > 90 Cr < 90

3 true AKI

70 (64%)

37

(33%)

(3%)

Cr > 90 Cr < 90

14 (27%)

37 (73%)

Total alerts False positive alerts

Fig 3. Proportion of total AKI alerts above and below the cut-off

Fig 4. Proportion of false positive AKI alerts above and below the cut-off

Fig.1. AKI alert algorithm

AKI Alert AKI

Fig. 5. Screen shot of planned alert on Cerner

The use of a 90 umol/L cut-off would reduce the false positive rate to 12% (Figure 4) but means that 3% of true AKI cases would be missed (Figure 3).

Details of AKI alert

Successes Multi-disciplinary collaboration between clinicians, clinical scientists and pathology IT has been key to the implementation of the alerting system.

Challenges Major challenges include implementing the system across all hospital sites due to different software versions of the LIMS and ICE and different clinical teams, and the complexity of educating medical staff across all 6 sites.It is expected that there will be an increase in workload for renal clinicians and this will be measured by audit.

Other outputsWe are currently analysing shadow data to assess patient management and compliance with local AKI treatment algorithms.

Acute Kidney Injury (AKI)Programme Board

AKI Detection in BoltonA C J Hutchesson

Consultant Chemical Pathologist, Royal Bolton Hospital. [email protected]

Hospital settingAn integrated Foundation Trust providing District General Hospital and community services in to a population of roughly 240,000.Busy Emergency Department (3rd busiest in Greater Manchester), with an effective catchment area of roughly 333,000.Out-patient Nephrology service only, provided by regional centre.No Electronic Patient Record; using Lorenzo Enterprise 2.2 as patient information management system.

Biochemistry laboratory serving hospital and local Primary Care; using Clinisys Labcentre 1.11. Results available electronically through web browser (based on Microsoft Internet Explorer) provided by Clinisys.

Details of AKI detection algorithmAKI detection algorithm developed as program (“logic rule”) within LIMS, adding automated comments to laboratory reports.

Last available creatinine result (if measured within 12 months) used as baseline, in keeping with National Acute Kidney Injuring Audit, 2012-13.

Comments added according to following criteria:• Consistent with AKI stage 3:

• Creatinine increased by >3-fold since last measurement, or• Creatinine >354 μmol/l and increased by >45 μmol/l since

last measurement; or• Creatinine >354 μmol/l and no previous measurement

within the last year (ED and in-patients only; out-patient and GP requests received a comment about Chronic Kidney Disease).

• Consistent with AKI stage 2:• Creatinine increased by >2-fold since last measurement.

• Consistent with AKI stage 1:• Creatinine increased by >1.5-fold since last measurement,

or• Creatinine increased by >25 μmol/l since last

measurement, within 48 hours.All comments include a reminder to consider common causes of AKI (dehydration, sepsis, drugs, renal disease).No comments added to requests from Renal Out-patient clinic.

Comments available to clinicians immediately; viewed using web browser.Results with comments copied to “ghost” queue in LIMS for post-hoc validation by Duty Biochemist (requests from GPs telephoned) and audit.

Introduction of automated comments coincided with revision of local guidelines for management of AKI (performed by Manchester-wide network of nephrologists and intensivists).Feedback indicates comments have increased awareness of AKI and are appreciated by medical staff

Challenges • Unable to hyperlink guidelines in comments electronically at present.• Previous non-numeric creatinine results (e.g. “no specimen

received”) The logic rule’s default position is to interpret non-standard comments as “Zero”, and to attach an “AKI 3” comment (>3-fold rise in creatinine). This problem has decreased over time, with recognition and restriction of the range of comments used. Paediatrics. The definition of AKI in children and young adults (<18 years) is based on change in eGFR. The MDRF equation is invalid in children; the Schwartz formula requires the child’s height (not normally available to laboratories). Following discussion with a member of the Guideline Development Group for NICE CG169, it was decided that children should have AKI comments attached. Paediatricians were informed this was an approximation.

• Post-natal specimens. Fluid retention during pregnancy leads to a fall in serum creatinine concentration. Return of creatinine to pre-pregnancy values following delivery proved enough to trigger an “AKI 1” comment in a few patients.

• Primary Care. Roughly 16% of patients (see below) have been identified from Primary Care requests. As yet, guidelines for AKI management have not been targeted at this sector.

• Frequent requests. Insufficient time for creatinine to change significantly may lead to under-estimation of the severity of AKI.

Other outputsPerformance of algorithm was audited between 2/9/13 – 3/10/13.Patients identified, and source of request:

Outcome after 1 month audited and compared to experience in August 2013 (data only available for AKI 3):

Source AKI 1 AKI 2 AKI 3 Total

GP 27 2 6 35 (15.8%)

Out-patient 8 2 3 13 (5.9%)

ED/In-patient 123 21 29 173 (78.3%)

Total 158 25 38 221

0%

20%

40%

60%

80%

100%

AKI 3 (Aug 2012) AKI 1 (Sept 2013) AKI 2 AKI 3

Unknown

Died

On dialysis

Above baseline

Returned to baseline

AKI Detection at the Royal Derby HospitalJ Monaghan1, RJ Fluck2, NV Kolhe2, N Lawson1, T Reilly3, D Elliott3, R Packington2, K Horne2, NM Selby2,4

1 Clinical Chemical Pathology department, 2 Department of Renal medicine, 3 Department of Informatics, Royal Derby Hospital, Uttoxeter Road, Derby, DE22 3NE

4 Division of Medical Sciences and Graduate Entry Medicine, School of Medicine, University of Nottingham

Hospital settingRoyal Derby Hospital is 1100 bedded teaching hospital providing servicesfor a population of over 600,000 people. Each year the hospital has 127kinpatient stays, of which 58k are emergency admissions. There is atertiary referral renal unit and all major medical and surgical specialitiesexcept cardiothoracic and neuro- surgeries.The pathology system is CSC iLab (Computer Sciences Corporation). Thehospital uses another CSC product for results reporting, including the AKIresults (iCM, integrated clinical management).

AKI detection algorithmThe detection algorithm (figure 1) is a rule based system within iLabcombined with a human validation step to apply current diagnosticcriteria (from April 2010-Dec 2012 AKIN criteria, from Jan 2013-presentKDIGO criteria). This laboratory assessment of AKI is carried out severaltimes a day 365 days a year so results are rapidly available followingmeasurement of serum creatinine.Baseline is defined as the most recent stable creatinine, selected from upto 12 months previously. In our population, this means that only 8.8% ofAKI episodes do not have a previous measurement to use as baseline.Patients without previous creatinine values are given an estimatedbaseline using reverse MDRD calculation from an eGFR of 75ml/min;these results have a different alert message.The algorithm produces an AKI test result within iLab; this also includesthe baseline creatinine value (as a separate test field to allow dataextraction) and date of baseline.We have assessed and reported the diagnostic accuracy of this system:false negative rate of 0.2%, false positive rate of 1.7%, a further 3.2%received an incorrect AKIN stage. We have improved the latter twoelements by setting up a daily exceptions report that automaticallychecks all positive AKI results for accuracy.

Figure 1. Flow diagram of detection Figure 2. Example screenshots

algorithm

AKI alert• AKI results are posted in iCM (figure 2)• Each result has additional information to assistthe clinician, including reminder of diagnostic criteriaand a link to intranet AKI guidelines • AKI stage 3 results are telephoned to the relevantclinical area by clinical biochemists• In August 2013, we introduced an interruptive alertthat appears if a patient with AKI has not had an AKI care bundle completed (figure 3)

Successes • A strong team working approach was key to success with enthusiastic representation from nephrology, pathology and information analysts. • Sustainability - system has been running in clinical practice since 2010 • Ability to collect and analyse hospital wide data on AKI incidence and outcomes. • Facilitated collaborations with other centres that has included transfer of the AKI reporting system• The AKI detection system was not introduced in isolation; additional improvement measures introduced simultaneously increased its impact:

1. Comprehensive teaching & training program2. Streamlined nephrology referral and advice3. Intranet guidelines4. Electronic AKI care bundle in iCM that is now proactively

linked to AKI detection systemOver time, we have demonstrated improvements in hospital widestandards of care for AKI patients and a reduction in mortality rates(figure 4)

Challenges 1. Lack of funding. We had no funding to deliver any of this work, time

and costs were contributed by individual goodwill and within departmental budgets.

2. Limitations of LIMS software; unable to programme algorithms consistent with current diagnostic criteria. This means the clinical biochemists contribute a significant amount of time to this process

3. Dealing with patients without previous creatinine results to use as baseline – there is no perfect solution to this. Either these patients are excluded and some patients with AKI will be missed, or estimated baseline is used and patients with CKD will be wrongly classified. Having transparency about methods used and differentiating results in this situation is important.

4. The diagnostic criteria appear to perform less well in patients with chronic kidney disease (CKD). This issue remains unresolved.

5. Jaffe method is used to determine serum creatinine. We are currently moving to an enzymatic method

Other outputs• Recipient of BUPA Foundation Technology for Healthy outcomes prize 2012• Publications enabled by this work:1. Selby NM, Crowley L, Fluck RJ, et al. Clin J Am Soc Nephrol. 2012 Apr;7(4):533-40. Selby NM. Curr Opin Nephrol Hypertens. 2013 Nov;22(6):637-42.

Selby NM, Kolhe NV, McIntyre CW, et al. PLoS One. 2012;7(11):e48580. Scott RA, Austin AS, Kolhe NV, et al. Frontline Gastroenterol. 2013 Jul;4(3):191-197. Caddeo G, Williams ST, McIntyre CW, Selby NM. Nephrourol Mon. 2013 Nov;5(5):955-61.

Figure 3. Interruptive alert

Figure 4. Kaplan-Meier analysis

of hospital-wide survival to 30 days in patients with AKI, stratified by six month period (n=8411) showing significant improvement following combined intervention of e-alerts, hospital wide guidelines, educational programme and AKI care bundle.


E-alerts for AKI Detection at the Royal Devon and Exeter Hospital

Mulgrew CJ, Sully M, Reid S, Hewitt C, Wilson A (Contact: [email protected])

Hospital settingThe Royal Devon and Exeter Hospital is a large teaching hospital, serving a

local population of around 400,000. It also provides tertiary services in a number of fields, including renal medicine which serves three referring hospitals (North Devon, Taunton and Torbay) covering a population of 1million across Devon and Somerset.

Our AKI e-alert system was first introduced in August 2012 and the algorithm was refined to the one described below in November 2012The hospital laboratory system used is the SwiftLab Integrated Pathology System which is an in-house system linked to the hospital PAS, but with programming support provided by Hewlett Packard.

At present there is not an integrated EPR in Exeter, with the results reporting system being through the historically used IPS system, or the more-recently installed Medway OrderComms product from System C.

Details of AKI detection algorithmThe AKI detection algorithm is located within the IPS LIMS pathology results

system, following the algorithm as below:

AKIStage1: represents a rise in serum creatinine of more than 26 umol/L in 48 hours OR 1.5-1.9x increase in reference creatinine over the last 90** days.AKIStage2: represents 2.0-2.9x increase in reference creatinine over the last 90** days.

AKIStage3: represents 3.0x or more increase in reference creatinine over the last 90** days or current serum creatinine of 354 umol/L or more and increased by 44umol/L or more over reference creatinine over the last 90** days.

** the reference creatinine is the lowest (not latest) creatinine level recorded within the last 90 days (if available) or, only if no creatinine result in the last 90 days, look back further 91-365 days from the current specimen for the lowest (not latest) creatinine level recorded.

The baseline creatinine selected is the lowest creatinine over the past 90 days. Spurious, likely lab error results are discounted (e.g. when sCr reported as <18umol/l).

Patients receiving renal replacement therapy are not included in the alert system. A monthly update of patients joining and leaving the RRT programme is sent to the laboratory manager who adds tags to those patients within IPS, preventing AKI flags being generated.

Details of AKI alertThe AKI e-alert appears in the pathology system alongside any creatinine result

that met the criteria detailed above. At present, this does not require acknowledgement by the user. However, the new upgrade of our OrderCommssystem will require end-users to acknowledge that results have been seen and will log all results that have been viewed, even if they have not yet been acknowledged.

AKI stage 3 results are telephoned from the laboratory directly to the source clinical area. A message is then left with a member of staff.

A list of all new cases of AKI are reviewed is generated by our AKI Outreach Education team each morning (Mon-Fri). A proposal to generate lists of AKI stage 1 cases for pharmacists on each ward is now underway.

The AKI e-alerts are linked to the hospital ‘Ward Whiteboard’, allowing anyone to search for AKI cases by stage or locations in real-time across the hospital.

In recognition of the importance of AKI across the Trust, we have recruited an AKI Outreach Education team to run a pilot project on key wards over a period of 6-9months. This involves a fulltime Band 7 nurse and 1.4 Band 6 nurses. Part of their role has been to work with medical and nursing staff to raise awareness of AKI and to set up a network of link nurses for AKI across the hospital. Care plans have been written to support our AKI management care bundle.

SuccessesThe introduction of e-Alerts has helped identify patients across a large

inpatient trust. As a result of working with the initial design for a couple of months, the Outreach team suggested adaptations to how the flags worked (Length of time viewable reduced to 10 days, Removing MOST <18 Cr etc.) and this had been actioned, allowing less time to be spent reviewing spurious results and more time in clinical areas providing advice and education.Design of the T.H.I.N.K poster – provided a suitable acronym for AKI risk factors.Multi-disciplinary team working across the trust,with positive reception and support from most clinical areas. IT and teaching resources have been relatively cheap or ‘free’ to develop butprovide valuable information on a 24 hour basis.

Lessons learned / challengesGood Multi-Disciplinary communication is paramount.

We all have irreplaceable skills in patient care and need to work together and respect those skills. Intervention DOES provide favourable results! In light of NCEPOD report, AKI outreach is a no-brainer.Any algorithm can be improved, and none will be perfect.

The costs of introducing this system have been associated with recruitment of the outreach team and project support officer, needed to provide data analysis. Other costs have been relatively small one-off outlays at the beginning of the project (e.g. to install the algorithm)

Other outputsThe data collection, downloaded on a monthly basis for analysis, but daily for

clinical use, has allowed us to study patterns of incidence and then target these areas with support and education. We have also been able to look at outcomes for patients with AKI. These data were presented as an oral abstract at the American Society of Nephrology in Atlanta, 20131.

Sample screenshot of hospital AKI e-alert, with link to hospital intranet AKI Care Bundle guidance

0

100

200

300

400

500

600

700

800

900

1000

AKI incidence - Apr 2013 to Mar 2014

AKI 1

AKI 2

AKI 3

All cases of AKI Post-admission AKI

AKI stage

30-day mortality (median)

Median length of stay (if survived to discharge)

30-day mortality (median)

Median length of stay (if survived to discharge)

1 14.8% 11 days 13.7% 14 days

2 22.3% 12 days 21.4% 15 days

3 28.3% 11 days 33.3% 18 days

1. Mulgrew CJ, Wilson A, Flower B. Using e-alerts for AKI - recording incidence, measuring outcomes and guiding education. American Society of Nephrology oral presentation, November 2013

AKI Detection at Doncaster & Bassetlaw Hospitals NHS Foundation trust.R. Stott1, I. Stott2, M. Slokan1, A. Anderson3

1 Department of Pathology, DBH. 2 Department of Renal Medicine , DBH. 3 Integrated Software Solutions Limited, Winchester

Author email - [email protected]

Hospital settingDoncaster & Bassetlaw Hospitals is a large acute Trust serving a population of around 410000 with inpatients on three sites and laboratory facilities on the two larger sites. Doncaster Royal Infirmary is a large district general hospital with 515 beds and a broad range of specialties including a main renal unit. Bassetlaw Hospital is an acute hospital with approximately 170 beds, a 24-hour Emergency Department (ED) and the full range of district general hospital services including a satellite dialysis unit.

The Doncaster laboratory operates a tracked analytical system including four Abbott Architect C16000 analysers and processes the majority of primary care requests in addition to hospital work. The Bassetlaw hospital laboratory operates two C4000 series analysers. Creatinine is analysed on both sites using the same enzymatic method.

Results from analysers on both sites are interfaced to a single ISS Omnilab LIS via Modulab Gold middleware. AKI identification is provided via the LIS and free text alert comments are added to the results via rules based logic.

Results are available to clinicians via multiple routes depending on the location –• All primary care results are reported via PMIP compliant messaging. Most GP practices no-

longer routinely receive printed reports although these are sent to practices which have not opted out and to “out of area” practices which refer patients to the trust.

• All results are available for accession via LIS search routines embedded within the trust’s PAS system.

• All inpatient locations and some GP practices can access results reported since it’s implementation via the Sunquest ICE system. Rollout of request entry and phlebotomy modules continues to A&E and the remaining inpatient locations.

• Outpatient results and requests are mainly handled via paper.• Results are not yet shared electronically with other hospitals outside the trust.

Details of AKI detection algorithmThe AKI detection algorithm is incorporated as a custom calculation within the LIS. Results are calculated when analyser results are received . The alert level is then subject to analytical validation and clinical authorisation processes identical to any other result and also to the “critical result” reporting processes if the results fit the prescribed criteria. Laboratory staff are free to decide to telephone additional results if they are concerned about a trend.

The calculation is currently based on criteria established by the local renal team. This is based on a cascade process -

• Standard KDIGO “lowest value” criteria apply if one or more baseline result is available in the previous 72 hours. This provides output as AKI1, AKI2, AKI3 or “AKI not identified”. This baseline applies to around 0.8 % of our patient population.

• In the absence of immediate baseline results, standard KDIGO ratios are applied using the lowest value in the 90 days prior to the current request. This baseline applies to around 19 % of requests.

• The algorithm also calculates a recommendation to take another sample within 24 hours if there is no suitable baseline available. With the 90 days look-back this applies to around 80% of requests and is not used to generate a comment which is available to the clinical staff.

• Other work suggests that our population would still have 18% of results (over 50% of patients) without a baseline even if a 2 year look-back period was used.

Logic at the technical validation stage adds appropriate free text comments for AKI1, AKI2 and AKI3 categories. These direct the clinical staff to local management guidelines.

Details of AKI alertThe AKI result is calculated as a “hidden” test result within the Urea & Electrolytes test set which also includes a calculated GFR result. The raw result is visible to laboratory staff only. This value is used to trigger the addition of automated free text comments as appropriate to the results. This logic handles the distinction between probable AKI and CKD identified via eGFR calculations. The AKI comments take priority therefore CKD related comments only appear if there is no acute change in renal function identified.

The first occurrence of an AKI3 alert is handled as a “critical result” and telephoned to the requesting location (Not applicable to requests from the renal specialty). The telephone process uses a scripted SBAR communications process and includes recording the details of the result recipient in a dedicated “phone module” within the LIS. There is a 2 level escalation process in place in the event of failure to achieve an appropriate clinical handover of the results and the need for that process is also recorded via the phone module.

There is the option for receiving clinicians to “file” results on ICE to confirm they are taking action however that is not generally used. All accessions of results are auditable by laboratory staff with user ID being available for results accessed via ICE

At present the Trust does not have any specific AKI outreach service.

Successes Local laboratory involvement in identification of AKI patients commenced with the case identification process prior to the NCEPOD report and continued with the wider scale audit following that report. This retrospective process identified that there was work to be done on implementing the local guidelines. Identification processes were improved when Doncaster PCT required us to provide monthly reports on the quality and outcomes of AKI care although that demonstrated the need for real time reporting as the care of patients across the hospital proved to be slow to respond to improvement processes.

AKI reporting algorithm was piloted in a hidden state on the LIMS for 12 months to fault find, assess the likely workload and improve the algorithm prior to commencing reporting. This enabled us to identify potential problems and establish reporting processes which were compatible with the workload.

Alongside the introduction of AKI alerts the Trust guidance on AKI has been reviewed to make it more relevant and easier for junior staff to access. Regular education sessions on AKI have been run for junior doctors and for senior staff via the hospital grand rounds. This work has formed part of a wider strategy for improving the Trust’s emergency care pathway.

Challenges • From feedback at the introductory presentations, there seems to be a moderate degree of resistance

to stopping medications even for a short time in response to an AKI event.• Still working to improve the availability of access to guidelines. These may need to be hosted within

CCG websites for primary care to be able to access easily.• Cost to laboratory was met via developing the algorithm in partnership with the LIMS supplier.

Probably would be an implementation cost to other users.• Very significant cost implications for the laboratory could accompany any initiative to improve the

availability of baseline results. • We have identified AKI events which are missed by other algorithms but may benefit from

intervention as well as false alerts from the “lowest value” algorithm. • Also working on alternative baseline methods to ensure clinical sensitivity and specificity without

causing “alert fatigue” due to inappropriate repeat alerts and falsely positive triggers.• Clinical challenges remain around improving the quality of care. AKI cannot be seen in isolation and

needs to be tackled as part of a wider strategy for improving emergency care and reducing mortality –recruitment of staff in acute and emergency medicine is a key limiting factor here.

Other outputsA system for monitoring the frequency of AKI alerts to track changes in the incidence and outcome of AKI is in development.An audit is currently being undertaken to compare AKI care and outcomes before and after the implementation of the alerts.

Comparison of different AKI baseline algorithms using data from Doncaster & Bassetlaw Hospitals NHS Foundation trust.

R. Stott1, I. Stott2, M. Slokan1

1 Department of Pathology, DBH. 2 Department of Renal Medicine , DBH. Author email - [email protected]

Initial assessmentWe initially calculated the AKI level as a hidden test on our laboratory system to audit the performance prior to adding the logic to provide free text alerts. Our algorithm was based on the KDOKI criteria and used the lowest figure in the previous 72 hours as the baseline against which to assess any subsequent increase. A significant proportion of creatinine requests were on patients without a creatinine result in the previous 72 hours. Several baseline methods are suggested in the literature but we were unable to identify any studies which compared the performance of the calculation processes. We therefore designed our algorithm using the “lowest value” pattern and selected an arbitrary look-back period of 90 days. The system was completed on 10th April 2013 via the addition of logic to add free test comments to hospital results which triggered AKI1, AKI2 or AKI3. AKI3 alerts were telephoned from that date. Comments and telephoned alerts for AKI3 in community patients were commenced on 4/11/13.

Subsequently the ACB criteria for AKI calculation were agreed. While preparing to present our experience of AKI score calculation at a local meeting, we identified patients in whom AKI events were detected by our “lowest value” baseline method but not by a system based on a median value baseline. These patients did not have baseline results within the 3 days prior to their AKI event and therefore failed the strict KDIGO criteria but had a significant AKI event in the previous year which caused a markedly elevated median value.

There were also patients where a potentially false alert would have been produced by our “lowest value” algorithm as a consequence of a single low creatinine result which was not consistent with the patient’s baseline creatinine as assessed graphically.

We have more than 24 months of creatinine results obtained with the same enzymatic method and have therefore evaluated the performance of baselines based on the median, lowest value, age and gender and also a new method intended to avoid the potential problems with both the lowest value and median baselines.

The new method uses the same principle as the 10X Westgard rule for identifying internal QC values which may have made a stepwise change from the mean. A stable baseline should have half of the data points either side of the running median and they should be randomly distributed therefore the presence of successive data values on the same side of the median is a potential indication of a shift and becomes more likely to be a shift as more points appear on the same side of the median. The baseline is calculated starting from the oldest data point to be included and each new baseline value increments a count of the consecutive data points above and below the current median. When the count exceeds a pre-set number, the running median is re-calculated from the first point that side of the median. Therefore the baseline takes a stepwise move in the direction of any significant trend but is minimally disturbed by individual random values.

MethodAll creatinine values for 24 months were extracted from the laboratory information system using a SQL query and organised into comma delimited patient specific files with results in consecutive order. These files were identified using the lab system internal ID number which is not traceable as it is only available to limited numbers of laboratory staff with access via SQL.Programs were written in Microsoft Q Basic to create AKI alerts using each baseline process. Each program created a copy of the patient file identifying whether each data point had triggered an AKI alert and the grade. A method summary file was also created containing a list of scores for each patient file.

ResultsCharacteristics of the data available for AKI analysis were as follows -• Total data points 924602 creating files for 220123 patients. • 433474 data points were from males aged > 18 and 490784 from females aged > 18. • Less than 2.5 % of data points were for patients aged < 18 at the time the samples were taken. Mean age

of subjects was between 63 and 64 years for both genders (Range 0 to 114 years). • 44.5% of the requests originated from primary care and 2.85% were requested by the renal team as

identified via either the consultant or an associated location.• For all except the age & gender determined baseline method, there is a requirement for at least 3 data

points to generate an alert and therefore 121391 patient files (containing a total of 172918 data points) were too small to be useful and were not opened by the other evaluation programs (18.7% of the potentially available creatinine results, 55.1% of the patients).

Age & gender method. Cutoff values from Second International Consensus Conference of the Acute

Dialysis Quality Group (Bellomo et al Crit care 2004; 8: R204 – R212 – web http://ccforun.com/content/8/4/r204; As reprinted in Kidney International Supplements 2012;2, 19 – 36).

Minimum value method.

We anticipated that this method would also miss AKI events because the “lowest point” baseline method had previously identified patients who routinely ran with values below or within the creatinine reference range and who experienced well defined AKI events when their results moved into or within the reference range.

Median & “Median with baseline reset” methods. All for 365 days look back.

The effect of the reset process is to allow the baseline to follow changes in creatinine values more closely than the median and therefore detects low grade AKI events following a more significant event. Allowing the “reset” in both directions causes the algorithm to alert for fewer high points within a cluster and alert them at a lower grade because the baseline rises more rapidly than would be the case with a simple median.

Limiting the reset process to results which are below the baseline allows the algorithm to retain alerts for significant increases within an event as well as improving the ability to alert for low grade acute increases after a series of elevated results. When a reset was triggered by 2 consecutive data points below the median, this method detected all of the AKI 1 events known to be present in the patient shown in figure 1.

Conclusions

At present we do not know the total number of acute creatinine increases in our patient data and therefore cannot formally assess sensitivity and specificity values for any of the baseline calculation algorithms. However, the increased number of AKI alerts produced and a limited graphical assessment of the algorithm performance suggest that this method may be valuable in detecting clinically significant AKI events in situations where a conventional median baseline fails to trigger an alert. The current optimum number of data point to trigger the baseline reset may not be appropriate for other populations if requesting patterns vary and will need to be re-checked in future if AKI alerting causes a change in our local requesting frequency.

This method is the only one which would be available for over 50% of our patient population and was applied without access to ethnicity data. It alerts all values above a limit calculated from the lower limit for e-GFR and therefore falsely alerts for patients with elevated but stable results due to CKD.


AKI Detection at East Lancashire Hospitals NHS TrustDr Kathryn Brownbill, Consultant Biochemist ([email protected]) and Dr Ian Stanley, Consultant Anaesthetist

Hospital setting

East Lancashire Hospitals NHS Trust encompasses the Royal Blackburn and Burnley General Hospital s and covers a population of approximately 550,000. Much of the acute hospital work is centred on the Royal Blackburn site and as such Clinical Laboratory Medicine has centralised services for Blood Sciences, Microbiology and Cellular Pathology on that site. A satellite blood sciences laboratory exists on the Burnley General site, serving clinical services on that site, including a level 3 NICU, Womens and Newborn centre and Urgent Care Centre. GP Direct Access work is performed at the centralised laboratory.

The laboratory information software (LIMS) system is Telepath with the Sunquest ICE order comms and results reporting system.

Details of AKI detection algorithm• AKI detection algorithm is located within the LIMS system• Description of the algorithm methodology :

Details of AKI alert• AKI alert is sent to clinician via ICE desktop (see example screenshot

below).• A daily list of all AKI alerts is sent to the critical care out reach team

and pharmacy

Successes

• Multidisciplinary system wide approach to the introduction of the AKI care pathway:

MDT team including Medical Director, Renal Physician, Critical Care Physician/outreach team, Consultant Biochemist, Pharmacist

• MDT action plan agreed prior to system wide introductionIncluding:1. Introduction of e-alerts on ICE, with a message linking to AKI care

bundle on Trust intranet.2. Standard Trust documentation locally adapted from London AKI

network3. Daily list of all AKI alerts sent from the laboratory to out reach team

and pharmacy4. Communication prior to introduction via various streams- all Trust

user ‘message of the day’, population of AKI care bundles on Trust intranet, teaching via grand round/outreach team training, foundation teaching, departmental training

5. AKI observation charts included in patient casenotes6. Agreed plan for ongoing clinical audit

Successes have been a cohesive system- wide implementation plan with clinical support at all levels across the Organisation and agreed multi-disciplinary processes.

Challenges, outcomes and forward planning

• An initial baseline retrospective audit has been undertaken based on case note review of patients with AKI alerts identified on ICE

• The audit indicated the requirement for additional clinical education to enhance performance and documentation of observations and review in accordance with guidelines

• The out reach team now review all patients identified with AKI stage 3

• Ongoing action plan includes:Development to include AKI alert on Extramed system (bed bereau)Bespoke training packagesFurther planned presentation at Trust wide meetingsDepartmental auditPlanned extension to out reach review

There has not been a cost to the Trust to introduce the alert system, however, the clinical resource to deliver the requirements of the care bundle has not been objectively assessed

References1 NICE clinical guideline 169 (2013):Acute kidney injury2 London AKI Network manual

AKI Stage Staging criteria

Stage 1 Absolute increase serum creatinine of 26umol/L within 48 hr orFold increase 1.5-1.9 (ref period 7 days)

Stage 2 Fold increase 2.0-2.9 (ref period 90 days)

Stage 3 Fold increase 3 (ref period 90 days) orS creatinine 354umol/L with a delta of 44umol/L in 48hr

BK(P)1

AKI Detection at Gloucestershire Hospitals NHS Foundation Trust

Dr Preetham Boddana [email protected]

Hospital setting

Gloucestershire Hospitals NHS Foundation Trust comprises of two busy district general hospitals serving a catchment population of over 612,000 people. Our Lab services is PAS and our AKI software is provided by Hewlett Packard (HP).



How is AKI result sent to clinicians? Results seen on computer screen with AKI flag for all adult acute admissions and in patients who have AKI when renal bloods checked.

Are any results telephoned out from the labs? No

Do the results require end users to acknowledge the result? N/A

Have you employed additional staff to see patients who trigger an AKI result e.g. AKI outreach teams, or nephrologists with a dedicated AKI role Yes 1. Nephrologist leading AKI project2. Outreach team

Successes

What has gone well?

Increased awareness amongst all specialties about the importance of AKI detection and implementing AKI care bundle.

How can others learn from this?

Team work and importance of PDSA testing ramps for AKI care bundle.

Did you introduce AKI detection in isolation or in tandem with other AKI improvement strategies?

In tandem with other AKI improvement strategies-

Challenges

What have been the most difficult challenges to overcome and how may others learn from these?We need to improve further -95% reliability

Lessons learnt after implementation?AKI common and largely avoidable. Management is straightforward “good clinical care”Has there been a cost to introduce your system?Yes-£12000 for AKI software.Has there been a large increase in workload for clinicians/nephrologists Yes

AKI Learning

Safety Café

Briefings

E-Learningmodule for

junior doctors on AKI

AKI covered during

induction to trust

Briefings on fluid balance

to nurses during

handover

Current Creatinine level

= A

Creatinine within the last 48 hr

= B

A-B >= 26

Current creatinine = 354 umol/L or greater and more than 44 umol/L or greater than the reference creatinine

ADDTEST AKIStage2

A/C = 1.5-1.9

Lowest creatinine within 90 days = CIf, and only if, no creatinine within 90 days then Lowest creatinine in previous 91-365days

= C*

IF No Creatinine within 48 hours

A/C = 2.0-2.9

ADDTEST AKIStage1

A/C > = 3.0

ADDTEST AKIStage3

AKI FLAG

Medication Review

Early Warning Score

Repeat Creatinine

Fluid Balance Review

Senior Review

AKI Detection at Harrogate and District NHS Foundation TrustNuthar Jassam

[email protected]

Hospital settingHarrogate and District NHS Foundation Trust (HDFT) is a 300 beds acute hospital which provides secondary health care services. This includes all surgical specialities and community health services for the population of North Yorkshire and North East Leeds. Biochemical results are conveyed to clinicians via ICE system that integrated with the lab LIMS system ( LabCentre, CliniSys).

Details of AKI detection algorithmere in HDFT we

have implemented AKI alert system for

In HDFT, we have implemented an AKI alert system for in-patients only. The alert system is consistent of 1. A real time delta check of 50%. [Setting: in-patients, OP and primary

care]2. We have also incorporated algorithms into our LIMS that use

presenting and comparator ( baseline) serum creatinine values to generate AKI staging alerts. [Setting: In-patient only]. Baseline creatinine is calculated from a reverse eGFR of 75 ml/min/1.73 m2. See the algorithm below. (figure 1)

3. Absolute rise in serum creatinine > 354µmol/L. [Setting: In-patient, OP, Primary care]

Figure 1: HDFT AKI alert system is based on Derby’s alert system

Details of AKI alertThe alert became a development after a discussion with our nephrologists who were keen to implement a sensitive AKI Alert system. With this in mind, certain locations (e.g. renal wards, ITU and neonatal ITU) were excluded. To further refine the system, where no baseline value is available, we used ideal creatinine values to alert clinicians to the risk of AKI, and also include comments to request a repeat measurement in 24 hours to determine the trajectory of the biomarker.

• Creatinine results indicative of AKI risk or injury are reported to clinicians via ICE with a comment stating the stage of AKI.

• During the working day, AKI alert is consistent of a comment, AKI stage reporting and phoning out AKI stage 2 &3 .

• In OOH, only increase creatinine value to > 354µmol/L and delta increase of 50% are phoned out to inpatient only ( AKI staging not available in OOH ).

• Increase creatinine> 354µmol/L are phoned out to in-patient, OP and community.

• Delta 50% increase is applied to all patients, therefore, community AKI can be identified during biochemical results authorisation. Patient at risk of AKI are phoned out at the discretion of the clinical biochemist on duty during in hours only .

Successes• Our nephrologists have reported a decrease in the rate of referral• With the aim to promote AKI awareness, the introduction of AKI in

HDFT was presented in tandem with a teaching programme, using electronic learning material.

• Our data showed that 90% of identified AKI cases are in stage 1 ( figure 3) . In the majority of cases, deterioration can be avoided by timely withdrawals of nephrotoxic drugs and adequate rehydration plan. Therefore , AKI comments on the patient’s record guides the clinician to a downloadable IV fluid chart and a summary of AKI guidelines on a single sheet located on the intranet. Although no formal audit has been performed yet, a rapid resolution to the identified AKI cases has been observed.

Challenges • Our current LIMS system cannot accurately identify baseline

creatinine or AKI stage. The limitations has been ameliorated by the implementation of an AKI alert system that combines IT identification step with manual stage calculation.

• Lack of resources to report or phone out AKI stage in OOH. The risk has been ameliorated by phoning out 50% increase of creatinine.

• Barriers to integrating the AKI e-learning package into an established education programme.

• IT limitations restricted the option of connecting the AKI e-alert to the electronic prescribing system to block the prescription of nephrotoxic drugs in patients identified with risk of AKI.

• Progressing with community AKI alert was hindered by limitations in the current IT system, and lack of education programmes for GPs.

Figure 2: delta check increase of creatinine identifies only 33% of total AKI cases

Other outputs• Local audit showed that most of AKI-stage 1 cases are pre-renal or

due to nepherotoxic drugs administration. Hence a corrective action was built into the alert system by providing a link to local IV fluid guidance, and list of nepherotoxic drugs that should be stopped once a patient has been identified at risk of AKI.

• AKI comments that are added to patient report guides clinicians to required action plan based on the AKI stage reported.

• Alert system based on delta increase of 50% only detected 33% of total AKI cases ( figure 2).

• The current alert allows collection of AKI incidents, work is underway.

Derby Alert system 57

Delta check of 50%,

28

0

10

20

30

40

50

60

1 2

Figure 2

AKI 1

AKI 2

AKI 3

Figure 3: Delta increase in creatinine in Aki stage 1, 2 and 3.


AKI Detection at Hull and East Yorkshire Hospitals NHS Trust

Dr Adil Hazara, Dr Steve Holding, Dr Sarah Naudeer, Prof Sunil Bhandari

Hospital setting

Hull and East Yorkshire Hospitals NHS Trust is a large teaching hospital and is a tertiary care referral centre for renal services covering a population of 1.2 million. Routine clinical biochemistry service is provided to local population of 720 000.

Details of AKI detection algorithm• Pathology use an integrated Blood Sciences LIMS (LabCentre v1.8b,

Clinisys).• Algorithm developed in-house on Pathology LIMS (LabCentre, Clinisys).

Algorithm design• Uses: Current creatinine CC

Last creatinine LCBaseline creatinine BC

• Compares CC with both LC & BC to check for:• Increase of ≥ 25µmol/L in last 48 hours.• Increase to >1.5x baseline within one year.

• The 12 month window is designed for high clinical sensitivity but will give false positive results in some chronic disease patients.

Baseline creatinine (BC):• Updated at every visit (regardless of source of sample).

• Lowest value from LC or BC provided this was within the last 6 months.• We use 6 months to increase the probability that the next

sample will fall within the 12 month window.

• When the AKI flag is positive BC becomes LC.• This results in the flag being added to no more than two

consecutive reports unless creatinine continues to rise.

• When BC is greater than 6 months old it reverts to LC at the next visit.• This results in a “loss” of the last 6 months data.• Direct calculation of the lowest or median result for a set

period is not practicable with the current LIMS system.

• 7.1% of patient requests do not have a BC.


AKI alerts are reported alongside creatinine results via the Trust PAS system (PatientCentre v4.2.1135, iSoft).

Successes

We have been able to create an ‘AKI pathway’. This enables hospital non-specialists to correctly stage AKI patients and suggests initial management (see diagram). AKI pathway is available through a link on our hospital’s intranet.

We take into account any serum creatinine values from the community (i.e those requested by General Practitioners) to determine patient’s baseline.

Challenges

Availability of personnel: Writing the algorithm itself was a challenge. It was developed in-house by the pathology department. Our pathology LIMS (Labcentre, Clinisys) did not originally have this feature built-in.

Determination of baseline creatinine: Clinicians can usually look at large numbers of serial creatinines and estimate a baseline. To some extent, this requires application of clinical judgement. Ideally we would directly calculate the baseline creatinine from historical results, adapting the period used to the results available. It is more difficult to program a computer to determine a baseline for a given patient that is consistent with the guidelines. The current AKI definition is not easily converted to a programmable algorithm. The resulting algorithm is therefore a compromise solution.

Increased workload: it has been challenging to measure impact of AKI e-alert system on workload; database of consultation requests for nephrology input has not been formalised. A retrospective audit is underway to answer the above question.

Other outputs

AKI e-alert system has enabled us to determine the incidence of community-acquired AKI (cAKI). We have also estimated 30-day mortality, duration of hospital stay and re-admission rates in survivors.

Our work was accepted for oral presentation* at the American Society of Nephrology (ASN) conference at Atlanta, USA (November 2013).

*Hazara AM, Bhandari S. Community-Acquired Acute Kidney Injury: Hospital Admissions and Outcomes [Abstract]. J Am Soc Nephrol 24, 2013: 2A

AKI Detection : Kent, Surrey and Sussex AHSN AKI EQ teamEd Kingdon and Kay Mackay on behalf of the KSS AHSN Enhancing Quality AKI Clinical reference group

ASPH Ian White, Erica Heppleston BSUH Peter Anderson, Emma Hamza DVH Muhammad Javaid Hannah Rogers ESHT Matt Lees Emma Tate EKUHT Ian John Robin Ufton MTW Lee Baldwin Alan Dando

Medway Sanjay Suman Sarah Leng RSCH Mike Carrareto Trudi Mansfield SASH Raad Makadsi + Jonathan Parr WSHT Lui Forni Simon Higgs

Hospital(s) setting

Hospital setting : 9 acute trusts serving c 4.2M populationHeterogeneous pathology IT infrastructure LIMSiSoft Apex Laboratory Information Management Systemi-lab (APEX) version 5.8.1002 Medway SASHiSoft iLabTP (Telepath) version 1.9 DVH, MTW, WSHT (Chichester only)iSoft iLaboratory version 5.6 ESHT

Clinisys Winpath laboratory information management system (LIMS) BSUH, ASPH and RSCHBespoke system WSHT (Worthing only)

QlikView

Results reporting Sunquest- ICE is used in Surrey and Sussex Kent has different installations at different Trusts with a historical mixture of ICE, PlumTree (dart OCM) and Indigo (tQuest and review).


Location of detection algorithm: a) LIMS: All Kent trusts, ESHT b) ICE ASPH and RSCH (installation Spring 2014), BSUH (Summer 2014)c) In bespoke mechanism run by SITS (Sussex IT Services) SASH +WSHT

moving to for the data we upload to Clarity on behalf of SASH andDescription of the algorithm methodology Agreement for universal implementation sof ACB/RA algorithm June 2014www.acb.org.uk/docs/appendix-a-algorithm

Variation in baseline creatinine value used in pilots• Lowest creatinine in the previous 90 days• Lowest creatinine in the previous 365 days• Mean out-patient creatinine in last 365 days (Siew et al)

Handling of patients without a baseline valueExcludedExchange of information between ICE installations or back-calculation of Cr from MDRD not in place at present

Details of AKI alertHow is AKI result sent to clinicians?1. QlikView export (onwards to CCOT and Kent nephrologists) EKUHT and Medway2. Passive alerts within LIMS: ESHT MTW

Are any results telephoned out from the labs? No

Do the results require end users to acknowledge the result? No

Have you employed additional staff to see patients who trigger an AKI result EKUHT- CCOT review all AKIN 2 cases within the trustSASH and BSUH – business cases under consideration for additional staffing for CCOT

Successes

What has gone well? • Engagement- All AHSN EQ trusts now have a clinical lead and programme lead for AKI.

This is underpinned in some trusts by CQUIN payments

• De-mystifying IT infrastructure- The programme and clinical leads are aware of their trust’s systems and what they can do. Before EQ-AKI 3 out of 4 Kent trusts were unaware of the potential to identify AKI cases within LIMS.

• Structures for sharing information and collaboration: The CRG met regularly to plan developments and advertise local successes. Links formed locally have assisted implementation. A collaborative educational day in Feb 2014 was a big success.

• http://nww.enhancingqualitycollaborative.nhs.uk/index.php?option=com_docman&task=cat_view&gid=164&Itemid=71

• New calculation script for AKI algorithm to sun in Sunquest-ICE: The ICE installation for ASPH, RSCH and Frimley Park now includes a calculation script for the KDIGO AKI algorithm. This will shortly be installed in BSUH. The next ICE release will include an alert module. Anonymised details of AKI cases identified are exported to IT 3rd

party and standardised for comparison. This methodology could be employed to notify UKRR of cases.


• Agreed process measures for Quality Improvement Senior review, physiological scoring, urinalysis, medication review, renal imaging U+Es in acute admissions, U+Es repeated within 24 hours of AKIN ID Discharge summary identifies AKI event (NEW)

• Collaboration with Informatics 3rd party http://www.clarity.co.uk/

• Collation of process measures and Secondary User Survey outcomes Potential to ascertain cases on basis of CVVHF +/or IHD HRGs

• Benchmarking

Challenges

What have been the most difficult challenges to overcome and how may others learn from these?• Engaging acute trusts and primary care organisations without data is a problem. HES data

on AKI based on HRGs vastly underestimates AKI incidence. If engagement is needed to generate data using new or existing systems delays may ensue

• Finding a low-cost solution applicable across 9 acute trusts: LIMS manufacturers were not very interested in this area historically and there have been delays in completion of a middle-ware-based AKI alert

What still needs work ?• Implementation of middle ware AKI alert at all interested sites• Improvement of basic care processes across all specialities.

Lessons learnt after implementation?• Passive alerts without links to supporting resources and/or supporting clinicians may not achieve hoped-for change

Has there been a cost to introduce your system?• Calculation script development costs £17k, Modification of each ICE installation c £1k

AKI Detection at Kingston Hospital NHS Foundation Trust

Emma Ashley [email protected]

Hospital setting

Kingston Hospital is a district general hospital based 12 miles from central London. Thehospital supports approximately 320,000 people in the surrounding area includingKingston, Richmond and Putney.

The laboratory provides a comprehensive diagnostic service covering ClinicalBiochemistry, POCT, specific protein analysis and Immunology. The departmentreceives both inpatient samples and those from 600 GP’s operating from approximately90 surgeries. It provides a 24 hour service, 365 days a year.

The number of requests received into the laboratory in the past year is 450522. OurLIMS is Winpath provided by Clinisys, and results are reported into Powerchart providedby Cerner.


Following consultation with acute medical and anaesthetist colleagues, the AKI alertwas designed using the KDIGO stage 1 criteria. Introduction of the alert is in line withthe NICE CG169 guidelines on AKI. The detection algorithm is built into the LIMS rulesbase and it applies to requests from A&E and inpatients only. The algorithmmethodology is as follows:


The AKI alert sent to clinicians via WinPath Ward Enquiry is shown below. It is alsoavailable on the Care Records Summary (CRS) software. The AKI checklist in KingstonHospital’s ‘Blue Book’, which contains guidelines for the management of commonmedical emergencies, highlights the definition and stages of AKI as described by theKDIGO criteria. It also describes the AKI care bundle, with the recommended laboratoryinvestigations such as liver function for hepatorenal syndrome, creatine kinase forrhabdomyolysis.

The action limits for phoning are shown in the table below. The AKI alert does notrequire the end user to acknowledge the result. The trust has not employed additionalstaff to assess patients who trigger an AKI alert.

Successes

Overall, the AKI alert has been implemented successfully in line with the bluebook instructions as an improvement strategy. However, it is unknown whetherthe introduction of this system has improved patient outcomes; a retrospectiveaudit is required to review data before and after implementation of the alerts.

One major learning point from this design and implementation process is torecommend appointing a formal AKI group before setting up an AKI alert to allowfor effective and timely collaboration between different disciplines.

Challenges

There has been no cost in the introduction of the system, except for the timerequired to add the rule base to the LIMS system and the subsequent testingprocess.

A current technical issue involves the occasional unavoidable shutdown of theauthorisation status checker software. If this occurs, the rule cannot use the mostrecent baseline creatinine value and the AKI alert may appear inappropriately.

In order to see the AKI alert on the CRS platform, the user is required to click onthe creatinine result with an asterisk. This may potentially be a concern as thealert is not immediately visible and could be missed.

As there is no requirement to acknowledge receipt of the AKI alert, it is difficult toknow whether they are acted upon and whether the patient’s management ismodified accordingly. Furthermore, GPs are not currently informed of AKI alertsadded to their patient’s reports during their hospital stay. This is an subject thatrequires consideration in the future.

Other outputs

The AKI detection system has allowed us to extract AKI alert data. From16/07/2013 to 19/05/2014, the AKI comment was appended to 2176 reports.These 2176 alerts accounted for 1414 individual patients. For 394 patients the AKIalert was used more than once; this may have been due to multiple admissions ofa patient on separate occasions or due to fluctuations in creatinine resultsthroughout their hospital stay.

The pie chart shows the number of AKI alerts the top 10 wards received.Histogram 1 shows the spread of the age of patients and histogram 2 shows therange of creatinine results. The eGFR values ranged from >90 to 3mL/min/1.73m2.

This data has not yet been fully analysed to determine AKI incidence, patientoutcomes or to measure the effect of introducing AKI detection.

Analyte To phone if ≥: Unit Comment

Urea (adults)Urea (< 16 years)

30(>10)

mmol/L Phone if unknown or urea has increased by ≥ 10 mmol/L since last result

CreatinineCreatinine (< 16 years)

400(>200)

µmol/L Phone unless known

635

220

203

181

136

115

108

99

9182

A/E ITU

Acute Assessment Unit Bronte - Medical Unit

Hamble - Medical Unit Blyth

Hardy Cambridge Ward

Derwent - Medical Unit Astor Ward

0

100

200

300

400

500

600

700

800

900

<50 51 - 110 111 - 200 201 - 400 401 - 600 >600

Histogram 2

Creatinine (μmol/L)

Fre

qu

en

cy

0

100

200

300

400

500

600

700

800

18 - 40 41 - 60 61 - 70 71 - 80 81 - 90 91+Age

Fre

qu

en

cy

Histogram 1

AKI Detection at Lancashire Teaching Hospitals NHS Foundation TrustJ. Armer, S. Haslam, N. Hunt, P. Denny, R. Allcock, M. Myers , R. Coward, H. Shetty, J. Anderton

Contact e-mail: ( [email protected] )

Hospital setting

Lancashire Teaching Hospitals provides district general services to a localpopulation of 370 000 people at two hospital sites; Royal PrestonHospital and Chorley & South Ribble District General Hospital. Specialistservices are provided to 1.5 million people across Lancashire and SouthCumbria. We are a regional specialist centre for Oncology, Renal,Neurosurgery, Neurology, Major Trauma, Plastic surgery and burns.

Laboratory services are provided 24/7 at both hospital sites.

In the laboratory, the LIMS is Sunquest. Our hospital information system is Quadramed CPR.


Location of detection algorithm: LIMS (Sunquest)

Description of the algorithm methodology :Used for in-patients only. The system looks for the previous creatinine result (within the last 12months). If the current result has increased by ≥26 µmol/L compared tothe previous result, the ‘AKI’ comment is added to the current result nthe LIMS (Sunquest).

Description of how baseline creatinine value is selected (and with this definition how many patients do not have a baseline value):Using our present LIMS, we are only able to look for the previous creatinine result on each patient. This is used as the baseline. If the patient has not had a creatinine result within the last 12 months, the algorithm cannot be applied.


How is AKI result sent to clinicians? The comment can be viewed with the creatinine results in the patient’s file on Quadramed. See screen shot example below.

Are any results telephoned out from the labs? NoDo the results require end users to acknowledge the result? No

Successes

We introduced an AKI alert system and an AKI protocol in October 2012. We did this using our current LIMS without any additional cost implications.

Challenges • Our system is not sophisticated; however we have introduced an AKI

alerting system to detect a rise in creatinine without any additional cost.

• We plan to introduce a more sophisticated alerting mechanism at the end of 2014 when we install a new LIMS (Swisslab)

• Ensuring that clinicians are aware of AKI, what to do if they see an alert and the AKI protocol.

• To overcome this, a training programme was rolled out at the same time as the introduction of the alerts.

Other outputsThe table below shows an audit of 4 days of in-patient U&E requests and AKI alerts in May 2014. The averages are displayed in the table. Fewer in-patients have a U&E request at the weekend. However, the number of AKI alerts is comparable to weekdays.

An audit of the response to AKI alerts was performed in Summer 2013. This showed poor documentation of an AKI alert in the patient notes and variable adherence to the AKI protocol. For example, 30 patients identified via an AKI alert were on nephrotoxic drugs; these drugs were stopped in only 20 patients.

Further clinical audits of AKI are ongoing.

No. of in-patients

with a U&E

request

No. of AKI alerts % AKI alerts% AKI alerts

(excluding

the renal

ward)

Weekday 491 18 3.7 2.1

Weekend 345 16 4.7 3.1

AKI Detection at University Hospitals of Leicester NHS Trust

Richard Baines ([email protected])

Hospital setting

University Hospitals of Leicester NHS Trust comprises of threeindependently sited hospitals with an approximate total of 1900inpatient beds.

Laboratory data:• Laboratory software system (LIMS) – i-lab • LIMS provided by CSC for inpatient wards and Sunquest (for ICE) for

ED and GP’s• Results reporting system – i-lab, ICM and ICE


The detection algorithm is located within i-lab

Algorithm methodology

Stage 1An increase of 26 umol/L from baseline within 48 hours OR 1.5 – 1.9 foldincrease from baseline. Baseline taken as the lowest creatinine valuewithin the past 7 days.Stage 2A 2.0-2.9 fold increase from baseline. Baseline taken as the lowestcreatinine value within the past 90 days.Stage 3A >3.0 fold increase from baseline. Baseline taken as the lowestcreatinine value within the past 90 days OR a creatinine result of >354with an increase of 44umol/L from baseline within the past 48 hours.AKINAIf there is no a creatinine result within the last 7 days for Stage 1 or 180days for stage 2 and 3 then AKI stage will not be calculated and acomment will appear as follows:-No recent creatinine result available. A rising creatinine even withinrange may indicate AKI.All AKI stage 3 results will be telephoned out to the requestingward/clinician. A comment is attached to these results as follows:-AKI stage 3 is defined as an increase in creatinineof >3 times baseline ORa creatinine>355 umol/l with an increase of >45 umol/l over baseline

Baseline creatinine established as the lowest recorded value in theprevious 6 months.


The AKI result appears on i-lab below the serum creatinine and all newcases of AKI-3 are telephoned to the host team.

No acknowledgement of the AKI result is currently required – but actiontaken is recorded by the AKI outreach team as part of an ongoing audit.

Additional staff have been employed by the ITU outreach team in orderfor them to visit all cases of AKI-3 within the trust. The clinical team issupported by additional administrative staff who maintain the database.

Successes

For the first time UHL has a fully automated AKI alert process. Thetriggering of a visit from the AKI outreach team should reduce theincidence of avoidable AKI.

The alert process is one aspect of AKI improvement work. Other aspectsare separate e-learning packages aimed at colleagues in secondary andmore latterly primary care.

Challenges

The major problem with the algorithm has been establishing a baseline –extending the time interval searching for baselines meant that AKINAcould be reclassified as AKI-3. Minor problems relate to extracting thosewith ESRD triggering an alert on post- and pre-dialysis bloods.

The major challenge has been conceiving an action to be triggered by thealert despite limited resources.

Establishing the time period to be searched for the baseline creatininemay improve capture of more AKI-3. Work needs to continue on how towork most effectively on those that trigger as AKI-1 and AKI-2. This isparticularly so in primary care.

After implementation it has become ever more clear that we need towork effectively with colleagues in primary care to reduce overallincidence of AKI.

There was a small cost in purchasing the detection algorithm from i-lab.

The impact of the alert system on the workload of nephrologists has notbeen formally assessed but anecdotally no major increase has beennoted.

Other outputs

The database generated will enable us to calculate AKI incidence andoutcomes such as in-hospital length of stay and mortality. We have notedthat a significant number of AKI alerts are in patients receiving end of lifecare

AKI Detection at Royal Liverpool and Broadgreen University Hospitals NHS Trust

T. Hine¹, C. Hill¹ and S. Ahmed² (Correspondence: [email protected])

Dept. of Clinical Chemistry¹ and Nephrology² , Royal Liverpool University Hospital, UK

Hospital settingThe Royal Liverpool and Broadgreen University Hospitals NHS Trust (RLBUHT)is the major adult acute university teaching hospital for Merseyside and Cheshire covering a population of 450,000 and offers high quality treatment and diagnostics across more than 20 specialties. The tertiary renal unit at RLBUH covers a large population of 1.2 million including a large outreach population at Whiston and Warrington. The blood sciences laboratory department produces over 400,000 results per year.

We have incorporated an algorithm into our laboratory information management system (LIMS; TelePath (iSOFT)) to issue real-time alerts to help with detection of AKI. Additionally, incorporating the alert into the ward order communications (ICE) software, to be acknowledged by designated nephrology clinicians, ensures the condition is addressed appropriately, thus potentially improving patient outcome.

Details of AKI detection algorithmAKI algorithm is used in LIMS (Telepath) and alerts and comments are reported to ICE. We currently only generate AKI alerts on in-patients, excluding some locations such as renal wards and ITU.

Details of AKI alertAKI alerts are generated on Telepath and transmitted to ICE and are flagged for acknowledgement. This is auditable.There are dedicated AKI specialist nurses as well as nephrology staff who can review all patients flagged with an AKI alert on ICE.

Successes The e-AKI alert system has been a success in terms of over all quality improvements including-a. Early identification of AKI resulting in earlier response and therefore

reduction in progression of AKI cases.b. Service innovation and development: We have developed a hospital

AKI team with the appointment of a consultant with AKI interest and three AKI nurse practitioners to provide seven day care.

c. We have also established local audit tools and took part in the NHS Kidney Care National Audit project, ‘AKI proof of concept’ to identify shared best practices in 2013 using our AKI alert system.

Challenges 1. Education and motivation of junior doctors , nurses , and clinicians in

using the alert system. The key approach we adopted is education and training through workshops and teaching sessions.

2. For patients admitted with raised serum creatinine (CR) without a comparator within 365 days, we have developed an age and sex based equation based on 137,000 GP CR results to derive a reference CR. This enhances the sensitivity of the algorithm.

3. The algorithm was also improved to increase specificity in stage 3 alerts- to eliminate false alerts on patients with CKD but also to ensure detection of AKI in CKD patients (table 1).

4. The next challenge will be to implement an AKI alert system for detection in GP and outpatients.. We have established a pilot programme for a community AKI service.

Output and Lessons Learned: 1. The AKI alert system and algorithm has enabled collection of data

from Telepath (iSoft). A recent audit of AKI alerts identified 6% of CR results generating an alert in a 3 month period.

2. We have not seen a large increase of work load for clinicians/nephrologist s over the 2 years since the alert system was put in place . It has helped clinicians to act early and we have observed a fall of total number of stage 3 AKI in our hospital (average of 50-60 stage 3 AKI in 2012 vs. 35-40 in Nov- Dec 2013)

3. There has not been any additional cost other than time commitment from personnel involved in developing the project

4. We identified an increase both in AKI alerts and in severity of alerts during the weekend period compared to a week day.

5. A recent audit of AKI alerts (November, 2013) showed that 88% of AKI alerts were true AKI.

* Not confirmed on data review e.g. CKD, dialysis, too little data to assess

Acknowledgement: Laura Hill & Tharun Zacharia, Renal Unit , RLUH

C

Figure 1. ICE alerts. (A) The screen available to nephrology staff to view the alerts (B) the alert incorporated into the patient’s ICE record and (C) the acknowledgement screen that is auditable.

Table 1. Serum creatinine criteria to trigger an AKI alert

A

B

Population based alerts total 21 ( stage 1 AKI) 10 ( stage 2 AKI) 7 ( stage 3 AKI)

Confirmed population based alerts 10 8 6

Total number of

patients with AKI alerts

Confirmed AKI

with audit*

In-hospital

mortality

Overall mortality

AKI stage 1 215 182 18.7% 23.1%

AKI stage 2 74 69 17.4% 33.3%

AKI stage 3 36 32 43.7% 50%

AKI Detection (Not live yet!) at Luton and Dunstable Hospital, Luton and Dunstable University Hospital NHS Foundation Trust

R Leyland, D Housley. Contact email: [email protected]

Hospital settingThe Luton and Dunstable Hospital is a large district general hospital with approximately 600 beds providing several specialist services including Neonatal Intensive Care, Cardiac services, Stroke care and paediatric services. The hospital also provides emergency services with a busy A&E department serving a locality of 350,000 people.The laboratory receives between 2000-3000 specimens a day with samples coming from hospital patients and the community.

The laboratory uses CERNER PathNet v15.0 and the hospital reporting system is Sunquest ICE.

Please note AKI alerts are not live yet, they are currently being tested and a plan for implementation is being drawn up.

Details of AKI detection algorithmThe detection algorithm is written into Sunquest ICE and consists of three rules, one for each of the stages of AKI.

AKI Stage 1 alert: change in serum creatinine from previous result >/=1.5AKI Stage 2 alert: change in serum creatinine from previous result >/=2.0AKI Stage 3 alert: change in serum creatinine from previous result >/=3.0

The algorithm uses the patient’s previous serum creatinine result (irrespective of when this was) as the baseline value.

The alerts can not overwrite each other but they are in a hierarchy. This means that the highest ranking alert (Stage 3 > stage 2 > stage 1) will be shown at the top of the patient’s record, but when opened up all lower ranking alerts will be present.

Unknown how many patient’s do not have a previous serum creatinineand therefore a baseline value. A comment will be added to all creatinine results above the top of the reference range to make requestors aware of this limitation.

Details of AKI alertHow is AKI result sent to clinicians?The AKI alert is displayed at the top of the patient record on ICE as a coloured strip relating to the severity of the alert

Are any results telephoned out from the labs?No, but AKI alerts are all sent via ICEMail (e-mail internal to ICE system) to the patient safety lead nurse and team.

Do the results require end users to acknowledge the result?Yes – the user will be logged onto ICE in order to view results and then need to acknowledge the alert in order for it to be removed therefore providing an audit trail.

Have you employed additional staff to see patients who trigger an AKI result e.g. AKI outreach teams, or nephrologists with a dedicated AKI roleNo. The alerts are being introduced as part of a CQUIN project which will provide some resource to check alerts are acted on. Existing critical care out reach and nephrology services will be used for referrals as required.

SuccessesWhat has gone well?Setting up the rule outside of the laboratory LIMS system.How can others learn from this?Think outside the box, if you can’t use your LIMS for the alerts consider what other software the hospital uses to manage blood tests results. Did you introduce AKI detection in isolation or in tandem with other AKI improvement strategies?The AKI detection alerts will be introduced as part of a wider CQUIN

initiative for the prevention, detection and management of AKI in the Trust.

Challenges What have been the most difficult challenges to overcome and how may others learn from these?The most difficult challenge has been ensuring procedures and education are in place to cover the prevention and management of AKIWhat still needs work ?The support documentation and education of staff for the prevention and management of AKI will be an on going process lead by the medical division. Training staff how to clear alerts from ICE.Lessons learnt after implementation? N/AHas there been a cost to introduce your system? No direct costs. However cost of staff time for both of laboratory staff and clinical staff has not been calculated. Has there been a large increase in workload for clinicians/nephrologists and has this been measured objectively? N/A – not live yet but this will be audited.

Serum Creatinine (SCrea) result generated, R2

AKI rule checkPrevious SCrea (R1) available?

YES

Change from baseline = R2/R1

Previous SCrea (R1) is ‘baseline’ result

≥ 1.5

≥ 2.0

≥ 3.0

NO

Screa≥ 300 umol/L

Result phoned according to

routine clinical biochemistry telephoning procedures

Stage 3 AKI alert

Stage 2 AKI alert

Stage 1 AKI alert

< 1.5 No Alert

SCrea≥ 120 umol/L

Automated comment to

suggest repeat SCrea if

concerned about AKI

Screa ≥ 120 umol/L

AKI Detection at Maidstone and Tunbridge Wells NHS Trust

Jenny Ireland [email protected] and Mark Turner [email protected]

Hospital settingMaidstone and Tunbridge Wells NHS Trust (MTW), has two acute

hospitals; Maidstone Hospital and the Tunbridge Wells Hospital at

Pembury, both with laboratories on site.

Laboratory software (LIMS) system, iLab TP version 2.0, CSC.

Results reporting systems

Hospital: users have access via web browser into Telepath

database, Patient Administration System (PAS), iSOFT Patient

Centre and paper reports issued by laboratory to requesting

clinician

GP surgeries: Pathology Messaging Enabler Programme (PMEP)

end an electronic copy of results to surgeries. GP’s can also view

hospital generated results via ICE desktop or when requested

received a copy paper report.


AKI detection algorithm is within the LIMS (telepath)

CSC have written programme base on ACB algorithm

http://www.renal.org/Clinical/GuidelinesSection/AcuteKidneyInjury.

aspx

The delta period is used to establish Stage 1 in the absence of a

reference creatinine result (see below). The recommended period

for this is 48 hours, though this value can be modified if future

guidelines change.

The reference serum creatinine is the lowest creatinine value

recorded within three months of the event (again, this value is user

definable and can be changed in response to changed guidelines).

Details of AKI alertAKI stage is calculated and reported with Renal (Sodium,

Potassium and Creatinine) results to clinicians for AE and In-

patients. We do not calculate or report AKI results for GP or Out

patients renal requests. (see example user LMIS views and report

over page)

Telephone alert limits to contact clinicians urgently are based

on Creatinine or Urea results. Currently our Trust IT systems do

not require end users to acknowledge the results. I am not aware

of any staff being employed within MTW to see patients who trigger

an AKI result (e.g. AKI outreach teams, or nephrologists with a

dedicated AKI role.) Our Trust has a satellite renal unit from Kent

and Canterbury Hospital.

Examples :Web browser view

Graph AKI in LMIS (Telepath)

Laboratory Paper report


AKI Detection at Broomfield Hospital Mid Essex Hospital TrustDr Suki Sankaralingham, Dr R Saldana Chaparro, Jason Button, Dr S Abeygunasekara, Dr Anthony Chan

[email protected]

Hospital setting

Broomfield Hospital is a 300 + bed acute hospital with renal dialysis service serving population of 300000. The incidence of stage 3 AKI is 19/1000 hospital admissions. Pathology department currently uses Winpath 5.32 service pack 25, this is supplied by Clinisys. We also use Anglia ICE for results messaging to GP location, but they also have access to Review which is supplied and supported by Indigo 4.


The location of the rule is in the Winpath system, it is a rule based alerting setup, during the authorisation of the results the result rules are run, it starts with looking at whether the requester is hospital based or GP based, if the person is GP based nothing more happens. If the requestor is hospital based, Winpath looks back a maximum of 180 days to find the last result of the patient. If the new result is more than 26 umol/L greater than the last result a comment is added below the creatinine result and a guide is added to the lab number as well. Also the last lab result is added to the lab number for comparison.


How is AKI result sent to clinicians? E-Alert message of AKI diagnosis published on pathology review system along side the abnormal creatinine results.Are any results telephoned out from the labs? NoDo the results require end users to acknowledge the result? NoHave you employed additional staff to see patients who trigger an AKI result e.g. AKI outreach teams, or nephrologists with a dedicated AKI roleNoPlease include screenshots if possible to illustrate

Successes

What has gone well? How can others learn from this?It is possible to have an electronic alert despite limitations of the

laboratory software system.


AKI alert warning message includes advice to refer to AKI guidelines which is available on the Trust intranet.

Challenges

What have been the most difficult challenges to overcome and how may others learn from these?Implementation of e-Alert had not been difficult in view of the theenthusiastic effort put in by the AKI e-Alert committee!!

What still needs work ?To fine tune the e-Alert to detect AKI in patients without a recent creatinine to compare with. To implement a system which enables calculation of a theoretical baseline creatinine.

Lessons learnt after implementation?AKI warning message changed in order to be more visually noticeable amongst the plethora of lab results.

Has there been a cost to introduce your system? No.

Has there been a large increase in workload for clinicians/nephrologists and has this been measured objectively?Not been measured as yet.

Other outputs

e.g.Has your AKI detection system allowed local collection of AKI incidence and outcomes etc.?Yes.Have you measured the effect of introducing AKI detection ?YesPlease include conference abstract presentations or publications relevant to your detection system

Abstract presented as oral presentation at CRRT 2014 San Diego.

AKI Detection at James Cook University Hospital (South Tees NHS Foundation Trust): Progress and obstacles in developing e-alert system

Jonathan Murray1 ([email protected]), Steve Kardasz1, Stewart Pattman2, John Frater3 & David Bottoms4

1Renal Medicine, James Cook University Hospital, 2 Chemical Pathology, James Cook University Hospital, 3Chemical Pathology, University Hospital of North Tees, 4North East Pathology Network Manager

Hospital settingSouth Tees NHS Foundation Trust primarily comprises 2 main sites:-

James Cook University Hospital (JCUH) in Middlesbrough The Friarage Hospital in Northallerton.

Serves a population of approximately 1.5 million

receives approximately 82000 emergency admissions per annum

JCUH provides a wide range of general and specialty services including

major trauma, spinal injuries, cancer services, neurosurgery, cardiothoracic & vascular surgery and renal medicine

• Laboratory Information Management System (LIMS) iLab (CSC)• Results Reporting System WebICE (Sunquest)

Details of AKI detection algorithmWhilst working on NHS Kidney Care (2012) & Renal Registry (2013) AKIprojects began to develop local AKI e-alert & management bundles

Collaboration renal, chemical pathology & information technology

Focused most on how to determine reference creatinine measurements

we felt optimising this aspect of the algorithm was vital since any calculation comparing current creatinine against non-representative reference creatinine was going to be unreliable, skewed or misleading

When no creatinine taken 7 days, considered using ‘median creatinine’ from days -8 to -365 (as suggested by Association of Clinical Biochemists) but speculated adopting ‘median creatinine’ in such circumstances

may often miss triggering patients with significant AKI:-

Proposed location of detection algorithm within results reporting / ICE system. Specification agreed between nephrologists and biochemists and ordered by North East Pathology Network Manager (January 2014):-

Above final specification order followed much dialogue between Regional Pathology Manager and ICE system company indications were that test system could be delivered promptly but despite prompting +++ nothing delivered to date now considering embedding algorithm within LIMS (iLab) system

Algorithm Methodology

If no value 1 yr use population reference range (age & sex matched)

Details of AKI alertPlan to display alerts within ICE reporting system traffic light colours

Aspiration for results to require end user acknowledgement At night considering utilisation of ‘Hospital at Night’ alert system

(clinical tasks are centrally triaged and communicated to on-call doctors via hand-held mobile devices)

Given delays setting up detection algorithm, no process yet developed to mandate end user acknowledgement

Considering mandatory AKI alert icon workstation on each ward No plans at present to telephone alerts to clinician

New specialist nurse role considered – to assist AKI outreach & education

Successes & Challenges√ Collaboration across specialties & region√ Other AKI improvement strategies (care bundles) finalised & readyΧ Hamstrung by dependence on specific IT systemΧ Unable to provide data at this stage to support proposal that using

median creatinine for reference creatinine may often miss triggering significant AKI (especially in high risk group who suffer recurrent AKI)


AKI Detection in NHS Lanarkshire

Ian M Godber and Malcolm Hand - [email protected]

Hospital setting

The NHS Lanarkshire (NHSL) catchment is approximately 600,000 patients located to the South and East of Glasgow. It includes 3 acute hospitals at:Wishaw General Hospital (633 Beds)Monklands District General Hospital (473 beds)Hairmyres Hospital (415 beds)It also serves 104 local health centres, with 350 General Practitioners operating across NHS Lanarkshire as a whole.Total number of laboratory requests for investigations per annum from all three sites is over 850,000.

The laboratories operate a single Laboratory Information Management System (LIMS) installed in 2012 by InterSystems (TrakCare Lab). Clinician view laboratory results on an InterSystems web results viewer in the acute sector, and through a download into INPS Vision in General Practice.


NHSL, like most health boards in Scotland at present does not actively alert clinicians to AKI, however as a pilot, to assess the impact we programmed the LIMS system to send all requests which failed a set AKI algorithm to a reporting queue for 1 week during February 2014. The rule was set up to trap patients using the following criteria:•Check back all episode within 48hrs for absolute increase of 26.5 if results are entered or authorised for the test, based upon collection date/time (else receipt date/time if collection does not exist)AND/OR•Check back to earliest episode within a maximum of 7days for % increase of 50% (or 1.5 x baseline), if results are entered or authorised for the test set, based upon collection date/time (else receipt date/time if collection does not exist)

Details of AKI alert – Pilot Population

This snapshot excluded all of the ITU’s within NHSL as well as all GP and Outpatient requests.

Approx. 205 results were flagged up. Of these results, the data was excluded if it was felt to be non patient related or not relating to an appropriate inpatient (eg theatres).

There were 34 results from patients undergoing HD and x1 result from a patient undergoing PD- not all of these patients were in a renal ward 1. The HD and PD results were subsequently excluded as AKI algorithm would be meaningless. Also, looking through the results, a number of patients seem to have been flagged up because of an inappropriate outlying low creatinine value, or, in some, probably a degree of malnutrition, this was fairly easy to spot when looking at the results but not representing AKI, so these were also excluded from the results below but all would have been picked up as AKI stage 1

AKI grade/stage 1 (creat >26 or creat inc by 1.5-1.9x baseline creatinine)5o patients per week

AKI grade/stage 2 (creat increase by 2-2.9 x reference creatinine)13 patients week

AKI grade/stage 3 (creat >3x reference value or >354)9 patients per week

Conclusion

From this data we suggest that there should be a written note on all lab reports which pick up a possible AKI stage 1 to enable a clinical review to see if these patients do have AKI or not.

We would suggest that all AKI 2 and 3 should be ‘phoned to the ward where the sample originated, and some sort of structured review triggered. Would need to think of how we exclude HD or PD patients which may be problematic if they cant be easily identified.

The spread of results would suggest that this is a medical and surgical problem, so ideally we do need a pan health board approach.

Challenges

We now need to work with InterSystems to produce an automated algorithm and reporting structure. InterSystems are currently also working with NHS Wales on this enhancement.

Any new enhancement must include the following functionality:

•The requirement to suppress an AKI alert if the serum creatinine is falling. •The possibility of different alert messages being sent to different clinical locations•In the event that more than one serum creatinine test item in used in future, it should be possible to include more than one in calculations. •The AKI alert should be in its own separate comment field – not combined with other biochemistry comments•Exclusions need to be available e.g. Renal dialysis patients

AKI Detection at University Hospital of North Staffordshire NHS TrustDr Chris Thompson, Mrs Alexandra Yates UHNS, Stoke on Trent ST4 6QG

[email protected]

Hospital settingUHNS is a teaching hospital with a local catchment area of 750,000 and a tertiary referral catchment of 2.35 million patients. The emergency department has between 300-400 attendances per day. We have a PFI build completed 2013 with on site pathology services using Clinisys Pathmanager as the LIMS software. Our Patient administration system has been developed locally and integrates several systems into a Clinical Information System CIS. Assessment in March 2012 of baseline data from audit of the reporting of AKI to requestors using existing electronic and manual systems, (automated comments, alert flags in ICM biochemistry reporting system and telephone calls) in place via biochemistry laboratory were not adequate in detecting and highlighting AKI in line with KDIGO guidelines. The results: 375 patients were defined as having AKI stage 1 (76%), 2 (13%), 3 (10%), of these cases, clinicians were alerted by telephone in only 61 cases. 115 had an automatic comment added stating the creatinine had change by 25%. 208 baseline creatinines had a concentration above the reference range and were hence flagged as high. In addition, renal registrars recorded only 19% of patients with AKI2/3 were referred or discussed with the renal team.

AKI patients needed to be easily recognisable by teams with prompts for action, a new E-alert programme was developed on the Laboratory IT system, Pathmanagercommissioned by UHNS from Clinisys.

Details of AKI detection algorithmThe local AKI detection algorithm is embedded within the Clinisys Pathmanagersoftware. The primary data with Creatinine values for all tests performed by the biochemistry department is automatically put through an integrated programme as part of the laboratory data system Clinisys Pathmanager. This programme commissioned by UHNS and has been providing data since March 2012. This raw data was then put on the Clinical Information System as an e-alert.

The Clinisys Pathmanager programme filters as follows:Firstly the system Calculates AKI score on change in Creatinine from within 30 days if only 2 Creatinine values are known in this time. Eg. Creat 100 on 02/01/2012 and Creatinine 151 on 30/01/2012 = AKI 1.But if two samples within 7 days thenCalculates AKI Score in 48 hours time frame between lowest and highest Creatinine in this time or increase in baseline over just 48 hours of more than 26micromol/l or if Creatinine >354 a >44micromol/l rise.Eg. Creat 151 30/01/2012 and Creat 178 on 31/01/12 = AKI1Calculate AKI score in 7 day time frame between lowest and highest Creatinine in this time.Eg. Creat 151 30/01/2012 and Creat 302 on 04/2/2012 = AKI2 (not AKI3 from baseline on 02/1/12) •Score is highest value 1-3 from above to be classified as belowoAKI1 150%-199% increase in Creatinine within 7days or 26micromol/l increase within 48hoursoAKI2 200%-299% increase in Creatinine within 7 days oAKI3 >300% increase in baseline Creatinine over 7 days or if Creatinine >354 a >44micromol increase in Creatinine in 48 hours.

Details of AKI alertIn addition to previously described electronic flags in biochemistry reporting tools. The E-alert system was launched in March 2013 pulling information from AKI detection algorithm in Pathmanager run as a daily programme filtered for patients in dialysis clinics placing the information in to the clinical information system CIS as shown in the screen shots. The E-alert appears in CIS in 3 ways first in a ward view which enables each ward to identify any AKI patient with having to access individual data. The alert can also be accessed as individual patient search. Clicking an alert in screen, links into details of alert see second screen shot.

All AKI 1-3 results also appear in a list format. The E-alert results remain on the CIS for 2weeks. Alerts are accompanied with advice which prompts further actions:AKIN Stage 1 - Refer to clinical guidelines. Check U&Es within 24 hours.AKIN Stage 2 - Refer to clinical guidelines. Check U&Es within 24 hours. Dose adjust medication. Consider urgent ultrasoundAKIN Stage 3 - Refer to clinical guidelines. Check U&Es within 24 hours. Dose adjust medication. Consider Urgent ultrasound within 24 hours. Refer to On-Call Renal Registrar.AKI 3 result are phoned to requesting clinician\location. No additional staff were employed to deliver the alert it is entirely automated.

Successes Since introduction of e-alert referral rate to Nephrology for AKI3 patients has increased to 50%.The audit of patients identified has led to a successful business case for AKI outreach nurse to review all AKI 2/3 patients to improve morbidity associated with AKI.. There are plans for launching AKI nurse role, AKI pathway and datixreporting with RCA for all hospital acquired AKI 3 linked to AKI E-alert list

Challenges UHNS participated in the NHS kidney AKI Care Proof of concept audit and had a AKI3 rate of 13.5 patients per 1000 hospital admissions. We now have 2 years data which we plan to map to interventions and look at affect of different events up to January 2015. This time frame will also allow us to assess change to ACB agreed AKI algorithm which is due to be launched locally in near future

Previous Poster presentation at RCPE AKI concensus conference.

AKI Detection at the Royal Cornwall Hospitals Trust (RCHT)AS Mallard1, AL Barton1, A Patterson1, SC Fleming1, KR Wallace2, JD Stratton2, PA Johnston2, S Dickinson2 and RG Parry2

1Clinical Chemistry Department 2Renal Unit

Hospital settingThe Royal Cornwall Hospitals NHS Trust is the principal provider of acute care services in the county of Cornwall. It serves a population of around 450,000 people, a figure often doubled by holidaymakers during holiday periods. The Trust is responsible for the provision of services at three sites (comprising approximately 660 beds): • Royal Cornwall Hospital, Treliske, Truro • West Cornwall Hospital (WCH), Penzance • St Michael's Hospital, Hayle.

We have one laboratory based at the Treliske site which serves all three sites, plus seven community hospitals and 85 GP practices. This provides a 24 hour, 365 days a year service utilising Roche Modular analysers for all routine work and Integras/E170 analysers for the urgent/out-of-hours work. The workload comprises around 40% hospital and 60% GP requests. We currently handle around 6000 samples a day and produce an average of 8200 creatinine results a month.

The laboratory utilises Winpath5 v5.32 SP22 (Build 114), produced by CliniSys Solutions Ltd. Results are available via the IMS Maxims system.

Details of AKI detection algorithmThe initial AKI detection system is within the delta-check area of Winpath; the rule looks back for a result within the last 365 days of the current result and goes to the fail queue based on the following parameters - an absolute creatinine change of +26/-26 umol/L or acreatinine % change of +50%/-100%. All flagged results are sent to our Clinical Approval queue and the Duty Biochemist (we have four in total) assesses these and codes them from AKIN criteria, against a relevant previous creatinine result. A test code AKI1, AKI2 or AKI3 is added which automatically generates a comment on the report stating “Significant increase in serum creatinine, probable AKI, stage [1, 2 or 3]” and refers the user to the trust’s local guidelines.If no creatinine result within the preceding year is available our algorithm does not identify the abnormality as an episode of AKI. We would only identify such episodes if other abnormal results within the same request were flagged to the Clinical Approval queue.

Details of AKI alertFor patients with an AKI2 code, the result is phoned to the requesting location; for patients with an AKI3 code, the result is phoned to the requesting location, and the renal team is notified by email (up to 16:00 Mon-Fri) or via bleep/phone from 16:00 – 17:30 and Saturday mornings to 13:00. Only from 1st May this year have all results had to be acknowledged in Maxims.Once alerted the Renal Registrar/Consultant contacts the patient location to ascertain that appropriate action is being taken according to the local guidelines (right). This policy was introduced alongside education for junior doctors and ward staff.

Successes Prior to AKI flagging the laboratory produced daily lists (Mon-Fri) of patients with creatinine levels >= 354 mmol/L for a number of months.Feedback from the renal team is that the introduction of flagging AKI is a very useful adjunct resulting in patients now being seen earlier in the pathway. No additional staff have been employed to date and there has not been a large increase in work for the renal team as, on average, nine AKI cases are identified per day (5 AKI1, 2-3 AKI2 and 1-2 AKI3).

Challenges At times it can be difficult for the Duty Biochemists to establish the most appropriate baseline creatinine to use – it is not always pertinent to use the most recent result when a patient has been admitted as they could have had a GP-taken sample earlier that day/week which has already shown an increasing level. We have also started non-reportable coding for community patients and auditing follow-up requests/admissions.

One drawback is that the system is not automated and requires Duty Biochemist input, with no coding between 13:00 on Saturday until Monday morning (or Tuesday if a bank holiday).

We also have not had a failsafe mechanism in place to ensure that the delta-check was actually running. We do intermittent downloads of AKI data for audit purposes and plot incidence – see graph. Numbers decreased in autumn 2013/winter 2014 which was unexpected – it transpired that the delta-check rules were deleted within our Winpath system in early September 2013 and we did not detect this until late February 2014. See poster (if accepted) at this year’s Focus meeting.

Other outputsThe first six-month’s data has been published (ref 1). There were 1,906 AKI reports in 1,518 patients: 56.3% AKI1, 26.9% AKI2 and 16.8% AKI3. Median age 78 years, 51% were male. Median length of stay (LoS) was 8-9 days for AKI patients vs 2 days for non-AKI. Overall mortality for AKI patients was 21.4% vs 2.3% in non-AKI patients, but mortality increased with staging -18.0 to 22.0 to 31.4% respectively. However, mortality was even higher (42.4%) in patients who progressed whilst an in-patient, 50% being AKI1 to 2 patients – their LoS also increased to a median 18 days.

References/poster presentations1. KR Wallace, AS Mallard, JD Stratton, PA Johnston, S Dickinson and RG Parry. Use of an electronic alert to identify patients with acute kidney injury. Clinical Medicine 2014, Vol 14, No 1: 22–62. KR Wallace, AS Mallard, J Stratton, PA Johnston, S Dickinson and RG Parry, “Real time Reporting of AKI: Six Month’s Experience”, poster at: The Royal College of Physicians of Edinburgh UK Consensus Conference: Management of Acute Kidney Injury: the role of fluids, e-alerts and biomarkers, 16/17th November 2012; and The British Renal Society meeting, Manchester, May 20133. AS Mallard, KR Wallace, RG Parry, AL Barton, SC Fleming and A Patterson, “Real time Reporting of Acute Kidney Injury (AKI): A Year’s Experience”, poster at ACB National Meeting, York, April 20134. Barton AL, Mallard AS, Patterson A and Fleming SC, “Identification of Acute Kidney Injury (AKI) in Primary Care”, poster at ACB National Meeting, York, April 2013 5. Jarvis S, Wallace K, Mallard AS and Parry RG, “ASA Grade as a Predictor of Post Operative AKI”, poster at the British Renal Society meeting, Manchester, May 20136. Coupe A, Barton AL, Mallard AS, Johnston P, and Parry RG, “Community Acquired Acute Kidney Injury (cAKI): Incidence and Outcomes”, poster at the British Renal Society meeting, Manchester, May 2013; the American Society of Nephrology meeting, Atlanta GA, November 2013; and UK Kidney Week, Glasgow, April/May 2014

AKI by category per month

0

50

100

150

200

250

300

350

400

Dec

-11

Jan-

12

Feb-1

2

Mar

-12

Apr-1

2

May

-12

Jun-

12

Jul-1

2

Aug-1

2

Sep-1

2

Oct-1

2

Nov

-12

Dec

-12

Jan-

13

Feb-1

3

Mar

-13

Apr-1

3

May

-13

Jun-

13

Jul-1

3

Aug-1

3

Sep-1

3

Oct-1

3

Nov

-13

Dec

-13

Jan-

14

Feb-1

4

Mar

-14

Apr-1

4

No

. c

as

es

AKI3

AKI2

AKI1

AKI Detection at Pathology at Wigan and Salford

Wigan Wrightington & Leigh NHS Foundation Trust, Salford Royal NHS Foundation Trust

Denise Darby, Mark Guy, Rob Nipah, Stephen Gulliford

Contact email: [email protected]

Hospital settingPathology at Wigan and Salford (PAWS) was formed in 2013 following themerger of the Pathology Directorates of Salford Royal NHS FoundationTrust and Wigan, Wrightington and Leigh NHS Foundation Trust.

Salford Royal NHS Foundation Trust (SRFT) provides comprehensiveservices to the 220,000 population of Salford as well as being a regionaland national tertiary referral center. It is a large teaching Trust providingover 1 million hospital and community contacts for patients includingemergency and elective inpatient services (in approximately 850 beds),outpatient and daycase services.

Wigan, Wrightington and Leigh NHS Foundation Trust (WWL) is a majoracute trust serving the people of Wigan and Leigh. The Trust has threehospital sites: Royal Albert Edward Infirmary which is an acute hospitalin Wigan; Wrightington Hospital which is a Specialist Orthopaedic Centreand Leigh Infirmary which provides predominately elective surgery,endoscopy and outpatient services. There are approximately 850 bedsacross the 3 sites.

• PAWS uses the Telepath LIMS system

• SRFT uses the Allscripts electronic patient record for results reporting.

• WWL uses Concerto electronic patient record for results reporting.


Location of detection algorithm: LIMS (Telepath)

Description of the algorithm methodology A test called ‘AKI Score’ has been added into the U&E set and isautomatically requested each time a creatinine is requested. The result isnumerical score corresponding to the AKI Stage:

• Stage 1: an absolute increase of ≥26 µmol/L or a 1.5-1.9 fold increasecompared to baseline.

• Stage 2: a 2.0-2.9 fold increase compared to baseline

• Stage3: a creatinine ≥354 µmol/L or ≥3 fold increase compared tobaseline

Results are only reported for in-patients.

Description of how baseline creatinine value is selected For stage 1 the baseline is the lowest creatinine in the last 48 hours forthe absolute increase and the last 7 days for the relative increase. Forstages 2 and 3 the baseline is the lowest creatinine in the last 90 days.

(NB – We are currently awaiting a software update, following which thebaseline will be calculated in line with the UK consensus algorithm)


How is AKI result sent to clinicians?NB – The e-alert system has not yet gone live with AKI scores not currentlybeing reported to the EPR.

Once the e-alert system goes live, the AKI Score will be reported as anumeric result (corresponding to AKI stage) below the creatinine. This wasfavoured rather than adding the AKI stage as a comment to the creatinineresult, as in our experience most clinicians tend to view results incumulative format on EPR during which the comments are not directlyvisible.

Are any results telephoned out from the labs? New AKI Score of 3

Do the results require end users to acknowledge the result? No

Have you employed additional staff to see patients who trigger an AKI result At WWL an AKI Specialist Nurse has been employed.

Successes • A multi-disciplinary working group was established across both trusts

which produced guidelines for the investigation and management of AKIprior to the e-alert system being launched.

• Telepath LIMS system had already developed a software algorithmwhich could cope with the complexities of the KDIGO scoring system.

Challenges • Cost of purchasing software• Implementation across multiple Trusts with different EPR systems• Changes to the recommended algorithm – software update required.

Other outputsAt WWL, prior to the implementation of the Telepath e-alert software, apilot study was performed whereby a LIMS search was performed for all in-patients that had a creatinine ≥354 µmol/L. The notes and results of allthese patients were manually reviewed to determine if they fulfilled thecriteria for AKI and if so guidance was given on further management.Success of this pilot study has led to the appointment of an AKI Specialistnurse. The introduction of the Telepath e-alert system will improve thedetection process and enable more time to be spent ensuring patients withAKI are managed appropriately.

• Gulliford SR, Sloan J. Acute physicians should take the lead in developing acute kidney injury services in district general hospitals. Acute Med. 2014;13(1):49-50.

AKI Detection at University College London Hospitals NHS Foundation Trust

Nick Flynn, Anne Dawnay ([email protected])

Hospital settingUniversity College London Hospitals NHS Foundation Trust (UCLH) is an 846 bed NHS trust providing acute and specialist services from several sites in central London. Services include A&E, 4 ITUs, cancer (Macmillan Cancer Centre), cardiac (the Heart Hospital), neurology (NHNN Queen Square), and many other specialisms, as well as general surgery and medicine. UCLH does not provide specialist renal services. The laboratory at UCLH also serves general practice in the London boroughs of Camden, Islington and Westminster.

The laboratory at UCLH uses CliniSys WinPath v5.32 (CliniSys, Chertsey, UK).

Details of AKI detection algorithmA real-time automated delta check in CliniSys WinPath flags a 50% increase in creatinine to > 50 µmol/L from the most recent result within a 90 day period. The e-alert includes all patients served by the laboratory –inpatients, outpatients and primary care; no results are excluded based on patient location.

Details of AKI alert•The comment ‘?AKI – creatinine increase > 50% from previous’ with a link to the London Acute Kidney Injury Network website (www.londonaki.net) is automatically added to results that fail the delta check.•Delta check e-alerts with trigger creatinine > 100 µmol/L or from patients < 18 years result in the reflex addition of an ‘Acute Kidney Injury Flag’ test that always flags as abnormal (to increase the prominence of the alert within electronic records) and are phoned to the requesting clinician as part of normal laboratory critical results procedures.•A list of all delta check AKI e-alerts from inpatients at University College Hospital are emailed to Intensive Treatment Unit (ITU) outreach twice a day via an automated scheduled query of the pathology data repository. This email contains patient demographics, location and previous creatinine results.•In addition, all creatinine results > 300 µmol/L are retrospectively reviewed twice daily by the duty biochemist and phoned if AKI is suspected.

Successes Our e-alert shows that a simple, automated delta-check can detect and flag AKI in real-time, 24 hours a day, 365 days a year, at little extra cost, without any human input and using existing laboratory software. As delta checks are a simple feature of many laboratory information management systems, this is a system most laboratories could easily replicate.

It is not unusual to see an occasional low creatinine result, with the subsequent creatinine result triggering a false positive delta check. If baseline is defined as the nadir creatinine, this would result in multiple false positive AKI e-alerts. We therefore advise against using nadir creatinine for baseline estimation in automated e-alert systems.

Challenges As with all screening techniques, the AKI e-alert system does not have 100% specificity or sensitivity. We acknowledge that there are likely to be missed cases of AKI. This could occur if there are no previous creatinineresults within a 90 day period or if successive creatinine increases are < 50% of the previous value. Although some of these cases are detected if creatinine rises above 300 µmol/L, some may be missed completely.

Similarly, 27 of 90 (30%) delta check e-alerts with a trigger creatinine > 50 µmol/L were not due to AKI.

A more sophisticated algorithm and an improved creatinine assay is required to improve detection and reduce false positives and negatives. Given our method requires 10 minutes of IT time to set up it has been pretty good value for money. The most difficult challenge has been overcoming fear of staff being overwhelmed.

Other outputsWe have audited the electronic alert to determine its specificity and to describe characteristics of the delta check and patient outcomes.1

• From 11930 creatinine requests over a 12 day period, 63 of 90 (70%) delta check e-alerts were due to AKI, identifying 61 episodes of AKI. Thirty four of 54 (63%) creatinine results > 300 μmol/L were due to AKI, identifying a further 10 episodes of AKI. The positive predictive value for AKI of a delta check e-alert was greater when the trigger creatinine was > 100 umol/L (PPV 89%) or when the absolute change in creatinine was > 50 μmol/L (PPV 93%).

• 120 day mortality after the first e-alert or creatinine result > 300 umol/L for each patient was 3/39 (8%) in patients with false positive alerts, 4/38 (11%) in AKIN 1, 4/15 (27%) in AKIN 2 and 9/18 (50%) in AKIN 3. Of the 54 patients with AKI who survived to 120 days, 43 (80%) recovered their baseline renal function (serum creatinine within 20% of baseline) and 4 (7%) did not; 7 (13%) had insufficient data to assess recovery.

We have presented audits of our e-alert as poster presentations at several conferences including the World Congress of Nephrology 2013, ACB Focus 2013 and Renal Association Congress 2013.

1. Flynn N, Dawnay A. A simple electronic alert for acute kidney injury. Ann Clin Biochem. 2014 [Epub ahead of print] doi:10.1177/0004563214534832

>50% increase in creatinine from the most recent result within a 180

day period

Creatinine > 300 μmol/L

Automatic comment appended to creatinine result:‘?AKI – creatinine increase > 50% from previous’

with a link to the LAKIN website (londonaki.net)

Retrospective review twice daily by duty biochemistResult phoned (if not already done and AKI suspected)

Result phoned by BMS as part of normal lab critical result procedures

Acute Kidney Injury Flag test reflexed

If creatinine < 100 umol/L

and patient > 18 years

If creatinine ≥100 umol/L

or patient < 18 years

AKI Detection in WalesR G Roberts ([email protected])1; Prof. John Geen2,3; John Tovey4; Stephen Winder2; Prof. Aled Phillips5; Mumtaz Jaffery6; Paul Gausden6

1.Hywel Dda University Health Board. 2. Cwm Taf University Health Board. 3. University of South Wales. 4. Aneurin Bevan University Health Board. 5. University of Cardiff School of Medicine. 6. InterSystems Corporation

Hospital setting

The detection of AKI and generation of alerts in Wales has been integrated with the introduction of the All Wales LIMS. This system is currently live in 3 out of 6 Local Health Boards and the AKI alert function has been live since March 2014.

The LIMS system is TrakCare Lab provided by InterSystems Corporation.The results reporting is via electronic link to the Welsh Clinical Portal, a bespoke product designed by the National Wales Informatics Service.Results to Primary Care are currently sent via GP-link software provided by Sunquest Information Systems.


The detection algorithm is located within the LIMS; it is based on the recently agreed UK National algorithm (e-Alerts for AKI meeting, Derby, July 2013) .http://www.acb.org.uk/docs/appendix-a-algorithm

The algorithm is based on KDIGO, AKIN definitions of AKI. It considers changes in creatinine against a baseline value; the criteria for evaluation are indicated below.If no comparator creatinine is available on the information system within 365 days then warning is generated on results above the reference range: “raised creatinine ?AKI, consider also CKD and repeat in 24 hr if indicated “Paediatric patients are also included; the rules and messages generated in Wales will be the same, although the background algorithm will generate a CKD stage 3 code for paediatric results which are > 3 times the age related reference range.

In Wales the decision was taken not to indicate the AKI stage in the alert message, but the staging information can be gathered from the LIMS using the message code rather than the expansion.Also, repeat alerts are generated only if the current creatinine is significantly higher than the previous result (at present, 6% higher,2 x c.v.value from creatinine IQC data).

Details of AKI alertThe AKI alert is sent out as a report comment which is presented beneath the relevant test results. It is currently a passive alert and this has prompted some sites also to telephone the alert to the patient location.

Successes

The alerts were introduced onto the LIMS on the date that the third Wales Health Board went live on the system. There were no significant problems following introduction. Feedback from clinicians was positive, but the passive nature of the alerts means that no definitive evidence exists if and when they were acted upon.Although there was ample communication regarding the introduction of AKI detection, it was introduced in isolation in two Health Boards. In the third Health Board the alert is presented with a hyperlink which opens the Board’s AKI management guidelines:

Challenges

The most difficult challenge was to work with the LIMS supplier to ensure that the highly complex algorithm was correctly configured and also allowed adequate flexibility to ensure that any future changes were relatively easy to implement. We were fortunate to have a team including scientists and IT experts who were able to devote adequate time to this.The functionality was provided by InterSystems as part of their commitment to implement UK nationally agreed requirements onto their product.Future work will concentrate on the nature of the e-alert and follow up:To extend the hyperlink guideline access to all Health BoardsTo configure reporting such that AKI alerts require acknowledgement from the requesting clinician/teamFollowing an acknowledged alert, to monitor whether an AKI Management Care Bundle has been initiated. This might include an interruptive alert if further tests are requested on the patient.ing clinician/team

Other outputs

An AKI Steering Group has been established under the direction of the Welsh Renal Clinical Network. A Service Evaluation Project “AKI in Wales: Incidence Management and Outcome” is in process of approval.Data will be gathered from LIMS includingDate and time of AKI alertAKI stageLocation of patient; specialty of requesting clinicianSubsequent renal function results for duration of acute episodeThis data will be used to access the patient record to evaluate clinical impact, response and patient outcomes. Baseline results will be compared with future information to assess how and when improvements in management and outcomes are achieved.

Alert messages Criteria for generation

Acute Kidney Injury alert: rising creatinine within last 48hrs.

Current creatinine is >26 umol/L higher than lowest previous result within 48hrs

Acute Kidney Injury alert: rising creatinine within last 7 days

Current creatinine is >50% higher than lowest result within 7 days.

Acute Kidney injury alert: creatinine increase over baseline value

Current creatinine is >50% higher than the median of results which exist between 8 – 365 days

Acute Kidney Injury (AKI) Programme Board

AKI Detection at East Lancashire Hospitals NHS Trust Dr Kathryn Brownbill, Consultant Biochemist ([email protected]) and Dr Ian Stanley, Consultant Anaesthetist

Hospital setting East Lancashire Hospitals NHS Trust encompasses the Royal Blackburn and Burnley General Hospital s and covers a population of approximately 550,000. Much of the acute hospital work is centred on the Royal Blackburn site and as such Clinical Laboratory Medicine has centralised services for Blood Sciences, Microbiology and Cellular Pathology on that site. A satellite blood sciences laboratory exists on the Burnley General site, serving clinical services on that site, including a level 3 NICU, Womens and Newborn centre and Urgent Care Centre. GP Direct Access work is performed at the centralised laboratory. The laboratory information software (LIMS) system is Telepath with the Sunquest ICE order comms and results reporting system.

Details of AKI detection algorithm • AKI detection algorithm is located within the LIMS system • Description of the algorithm methodology :

Details of AKI alert • AKI alert is sent to clinician via ICE desktop (see example screenshot

below). • A daily list of all AKI alerts is sent to the critical care out reach team and

pharmacy

Successes • Multidisciplinary system wide approach to the introduction

of the AKI care pathway: MDT team including Medical Director, Renal Physician, Critical Care Physician/outreach team, Consultant Biochemist, Pharmacist

• MDT action plan agreed prior to system wide introduction Including: 1. Introduction of e-alerts on ICE, with a message linking to AKI care

bundle on Trust intranet. 2. Standard Trust documentation locally adapted from London AKI

network 3. Daily list of all AKI alerts sent from the laboratory to out reach team

and pharmacy 4. Communication prior to introduction via various streams- all Trust

user ‘message of the day’, population of AKI care bundles on Trust intranet, teaching via grand round/outreach team training, foundation teaching, departmental training

5. AKI observation charts included in patient casenotes 6. Agreed plan for ongoing clinical audit Successes have been a cohesive system- wide implementation plan with clinical support at all levels across the Organisation and agreed multi-disciplinary processes.

Challenges, outcomes and forward planning • An initial baseline retrospective audit has been undertaken based on

case note review of patients with AKI alerts identified on ICE • The audit indicated the requirement for additional clinical education

to enhance performance and documentation of observations and review in accordance with guidelines

• The out reach team now review all patients identified with AKI stage 3

• Ongoing action plan includes: Development to include AKI alert on Extramed system (bed bereau) Bespoke training packages Further planned presentation at Trust wide meetings Departmental audit Planned extension to out reach review There has not been a cost to the Trust to introduce the alert system, however, the clinical resource to deliver the requirements of the care bundle has not been objectively assessed References 1 NICE clinical guideline 169 (2013):Acute kidney injury 2 London AKI Network manual

AKI Stage Staging criteria

Stage 1 Absolute increase serum creatinine of 26umol/L within 48 hr or Fold increase 1.5-1.9 (ref period 7 days)

Stage 2 Fold increase 2.0-2.9 (ref period 90 days)

Stage 3 Fold increase 3 (ref period 90 days) or S creatinine 354umol/L with a delta of 44umol/L in 48hr

mailto:[email protected]

Documents

AKI Detection at Barking, Havering and Redbridge NHS Trustresults to primary care. AKI alerts went live in our Trust in November 2013 for all creatinine requests. Details of AKI detection