3. Major Tertiary Health Provider in Northeast
Melbourne
3 Campuses
- The Austin Hospital
- Heidelberg Repatriation Hospital
- Royal Talbot Rehabilitation Centre
Major Services
- Liver and Gastro-Intestinal Transplantation
- Spinal Cord Injuries
- Oncology
- Victorian Respiratory Services
- Olivia Newton John Cancer Centre
4. 93,000 Inpatient Admissions
900 Beds
73,000 Emergency Attendances
57,000 Placement
Days for Entry Level
to Practice Students
in 17 disciplines
176,000 Outpatients
176,000 Outpatients
(360 clinics)
8,000 staff
>26,000 Surgical Operations
Large Nursing
and Medical Post Graduate
Education Program
6. Austin Health – Strategic Priorities
Build Capacity in Systems Redesign to Improve Quality, Value and
Efficiency
Provide Contemporary Clinical and Business Information Systems
that Support Excellence in Decision Making, Patient Care and
Accountability
Continually Enhance Information Technology and Communication
Systems
9. EMR Adoption Model
EMR Adoption Model
Structure Ensures
Objectivity
2012 Self Assessment
for Austin Health =
Stage 2
2014 Self Assessment
for Austin Health =
Stage 5
10. Transformational Change
More than 70% of all major transformation efforts fail.
Why? Because organisations do not take a consistent,
holistic approach to changing themselves, nor do they
engage their workforce effectively.
Kotter 1995
11. Prevalence of Diabetes
Prevalence of diabetes in Australia is estimated at 7%
(AusDiab)1
23% for people older than 75 years2
40% of diabetes undiagnosed3
1 Diabetes Care 2002; 25: 829-834
2 Dunstan DW, Zimmet PZ, Welborn TA, et al. Diabetes Care 2002; 25: 829-834
3 Diabetes Care 32:287–294, 2009
.
12. Background – Inpatient hyperglycaemia
0
2
4
6
8
10
12
Known diabetes New hyperglycaemia
(FPG >7mmol/L)
Uncertain glycaemic
status (FPG, 5.6–
6.9mmol/L)
Normoglycaemia (FPG, <
5.6mmol/L)
Mortality(%)
p=0.04
13. Diabetes in the Surgical Units
• Comparison of Length of Stay data for Surgical
patients from 2009 to 2013:
– Ave LOS is 6.91 days
– Patients with a coded diagnosis of Diabetes 10.61
days
– Diabetes patients stay 53% longer
• Comparison of Readmission rates
– Diabetes patients have higher readmission rate –
but may be due to other reasons
14. Goal: patient
safety &
consistent
practice
Clinical System
supports Clinical
practice
Clinician Led
Use evidence based
protocols
BUT
Who needs the
intervention?
Diabetes Management Project
15. Diabetes Discovery Project
• Aim
– To investigate the prevalence of diabetes (diagnosed and
undiagnosed) at Austin Health via routine HbA1c testing in
inpatients using the CERNER Millennium Health IT System
– To identify inpatients with poor glycaemic control (HbA1c≥
8.5%, 69 mmol/mol)
• Hypothesis
– Information technology tools such as CERNER Millennium
aid the identification of patients with undiagnosed and
patients with poor glycaemic control
16. Diabetes management Technical build
• Early & broad identification via Hb A1c Auto ordering
• Notification of poorly controlled and New diabetes
patients via:
Medical History via Problems/Alerts & Message Centre
• Reports for division of duty of care by Hba1c ranges
• Standardised evidence based ordering –BMJ action sets
• Powerplans & BMJ subscription (initially)
• Diabetes Educator Referrals
• Communication to GP community via Discharge Summary
17. Automated ordering of HbA1c
Austin Health Admissions (July 2013 to Jan 2014)
Inclusion criteria:
≥ 54 years
Acute admissions
Austin campus
Exclusion criteria:
Day cases
Palliative care
Psychiatry
18. Automated Ordering of HbA1c
Austin Health Admissions (July 2013 to Jan 2014)
Inclusion criteria:
≥ 54 years
Acute admissions
Austin campus
Exclusion criteria:
Day cases
Palliative care
Psychiatry
Automated CERNER order for HbA1c% generated if no result within 3 months
19. Change Management
Workflow for Medical staff
Who sees which patients?
How do they know who they need to see ?
Alerts and Notifications
Tools to support Medical staff workflow
HbA1c Results Extract Report
Improve communication with GPs via discharge summary documentation
Clinical Guideline translated to a PowerPlan
Nursing Workflows
Patient Access List : Referrals , Meds to be administered, Path to be collected
Diabetes education team – e-referral workflows
Task List
Reports
Documenting outcomes
25. Outcomes
N = 5083 patients, 6716 admissions (June 2013 to Jan 2014)
History of diabetes
HbA1c <6.5% HbA1c ≥ 6.5%
Known Diabetes
N= 1453
No history of diabetes
HbA1c <6.5%
No Diabetes
N=3359
HbA1c ≥ 6.5%
New Diabetes
N=271
28. Conclusions
34% of all inpatients > 54 years have diabetes
• 5% of inpatients have undiagnosed diabetes
• 29% known diabetes
Higher HbA1c is associated with
• increased admission rates
• Longer length of stay in surgical patients
Routine inpatient HbA1c testing using CERNER addresses a currently
missed opportunity to identify patients with newly diagnosed
diabetes and poor glycaemic control.
.
29. Evolving Changes to practice
Inclusion of Mental Health patients – with different auto
ordering criteria
Refinement of parameters – who sees which patients
General medicine Outpatient Clinic- follow-up of poorly
controlled patients post discharge
Ongoing education in diabetes management to junior medical
staff
Research in ICU – using HbA1c results – changes to protocols
The impact of early identification and treatment of poor
glycaemic control on patient outcomes requires further study
30. Acknowledgements
Cerner Corporation
University of Melbourne – Endocrinology Unit at Austin Health
Austin Health - Clinical Systems Projects Unit & Business
Intelligence Unit
Health Shared Services
BMJ – Action Sets
Notes de l'éditeur
Austin Health Named as Lead Agency in 2009
Aim to Implement a Common System Design Victorian State Build
Undertaking the Project as Part of a Group, with each Organisation’s project running in parallel and working to a common schedule with shared domain with multiple Health Services
UK Experience – One size will NOT necessarily fit all
▪▪▪▪Brennan 2009
Ordering of pathology, radiology and direct interface into the systems that receive the orders
Results viewing and endorsement
Discharge summaries
Discharge prescriptions
Alerts and allergy management
Medication ordering and administration management
Medication reconciliation
Drug interaction alerts
Fluid balance charts
E-referral
Austin Health Named as Lead Agency in 2009
Aim to Implement a Common System Design Victorian State Build
Undertaking the Project as Part of a Group, with each Organisation’s project running in parallel and working to a common schedule with shared domain with multiple Health Services
UK Experience – One size will NOT necessarily fit all
▪▪▪▪Brennan 2009
Not an IT Project
Transform the way we do business in health
‘Can Do’ Organisational culture
Focus of the organisation for 3-5 years
Governing principle - Patient and patient Safety First
Established sound Governance Structure
Established strong working relationships with all external stakeholders - Cerner Corp, Health Shared Serivces , peninsula Health, Eastern Health and RVEEH
What was difficult
Engaging clinicans and finding something of interest to them to be the catalyst for change.
Linking the impleemnttaion to more tangible benefits in the clinical setting
As everyone in this room is well aware diabetes poses an increasing challenge to healthcare provision. In 2011 366 million people reported to have diabetes and is estimated to increase to 552 million by 2030(1). The prevalence of diabetes in the Australian community is 7.4%, rising to 23% for those aged above 75 years.
Previous audits, including from Austin, estimate prevalence of inpatient diabetes at ≈20%*
Inpatient mortality
This is likely to be an underestimated as many are undiagnosed at the time.
AIHW data indicate that people with diabetes have longer lengths of stay, being about 2 days longer than people without diabetes.
Inpatient hyperglycaemia is associated with poor hospital outcomes. In several settings, hyperglycaemia has been associated with increased morbidity and mortality
Reasons for increased morbidity and mortality may be related to poor immune response, delayed healing, inflammation and thrombosis associated with hyperglycaemia as well as a higher rate of co-morbidities in this patient group
A. Length of Stay by Age Group shows the average length of stay for patients with and without diabetes by age group. The average length of stay is 6.91 days for patients without diabetes diagnosis, and 10.61 days for patients with diabetes. The relative length of stay for diabetes patients is 153% (i.e. they stay, on average 53% longer). This is highly significant. There are a couple of filters on this pivot table that allow you to select specific calendar years and clinical units.
B. Change in Length of Stay Over Time shows how the length of stay has changed over time. You will see that over time the gap in average length of stay has decreased, from a relative LOS of 183% in 2009 to 127% in 2013. Again this is a very significant change.
C. Length of Stay by Specialty shows the same data, but split by clinical unit. Some units (e.g. Gastroenterology, Cardiac Surgery) appear insensitive to a diabetes diagnosis.
D. Length of Stay by Admission Urgency shows how length of stay varies by admission urgency (Emergency vs Elective) and across campus.
E. Unplanned Readmissions within 30 days shows the rate on unplanned readmissions (defined as emergency admissions via ED) within 30 days of the hospital discharge for the surgical admission including any subacute component. Diabetes patients have a significantly higher readmission rate, but the readmission reason may not be related to the original surgical admissio
Not an IT Project
Transform the way we do business in health
‘Can Do’ Organisational culture
Focus of the organisation for 3-5 years
Governing principle - Patient and patient Safety First
Established sound Governance Structure
Established strong working relationships with all external stakeholders - Cerner Corp, Health Shared Serivices , peninsula Health, Eastern Health and RVEEH
Discern Rule logic- Hb A1c order placement upon admission
Patients aged >=54 years (Mental Health patients >29 years)
Patients without Hb A1c results within 90 days
Inpatients in acute locations
Excluded Day Surgery, ED Short Stay, Oncology & Dialysis Day Stay locations
Hb A1c orders placed in ‘pending dispatch’ status with Endocrine lead consultant as ordering doctor
Hb A1c test ‘nets’ with other blood orders for collection
Post Auto Ordering - Duplicate checking warning presented within 90 days if another Hb A1c being ordered manually
1Diabetes Problems = 200+ Snomed Terms evaluated during rule processing
Evaluation of Abnormal HbA1c% Results via Discern Rule
Outcome 1 - Results 6.5- 8.0% w/out ‘Diabetes’ Documented in Medical Hx
“Possible New Diabetes Pt” Alert applied
Message sent to Generic Gen-Med Diabetes Inbox
Generic Inbox managed by Gen Med Clinical Unit
Outcome 2 - Results >=8.1%
‘Possible Poorly Controlled Diabetes’ Alert applied
Managed by Endocrine Unit via Alerts or HbA1c Report
Auto applied alerts part of Patient’s on-going Problems/Alerts Profile
“Possible New Diabetes Patient” Alert
Inactivated & replaced with Snomed CT ‘Diabetes’ Medical History if pertinent after patient assessment
Medical History(Snomed CT & alerts) created for all future care
Inactive alerts are still auditable on General Alerts report
“Possible Poorly Controlled” Alert
Not inactivated unless incorrect
Snomed CT ‘Diabetes’ Medical History if pertinent after patient assessment
Medical History(Snomed CT & alerts) created for all future care
Auditable on General Alerts report
Creates a list of patients with Hb A1c results utilised for a more targeted treatment
Available to all clinical roles through explorer menu – Used primarily by Endocrine & General Med
Run daily by Endocrine Unit for 8.0% and above
User determines
Result and Date range
Clinical Units and Locations
Report and CSV outputs
Enhanced to show “Medical – Diabetes” alert presence
Indicator that this patient is ‘known’
BMJ Action Sets built as PowerPlans where possible within the licencing
Home Unit to use PowerPlan to guide treatment of patients with HbA1C readings > 6.4% and < 8.5%
NB Extracts only: Power Plans include different medication regimes and are quite extensive
Link to the Action set is within the PowerPlan
Implemented Diabetes Referrals April 2014
Utilised same ‘Referrals’ template used for many disciplines at Austin health
Diabetes Edu Referral and Review Orders/Tasks
Referrals placed by any clinician or self referred
Reviews placed by Diabetes Edu CNCs
Discipline designed list of referral and review reasons
Referral/Review Task list (MPTL) managed Diabetes Edu CNC
‘Phone” icon on Patient Access List (PAL) for all users indicating referrals/reviews are outstanding for patients
Link to the Action set is within the PowerPlan
Implemented Diabetes Referrals April 2014
Utilised same ‘Referrals’ template used for many disciplines at Austin health
Diabetes Edu Referral and Review Orders/Tasks
Referrals placed by any clinician or self referred
Reviews placed by Diabetes Edu CNCs
Discipline designed list of referral and review reasons
Referral/Review Task list (MPTL) managed Diabetes Edu CNC
‘Phone” icon on Patient Access List (PAL) for all users indicating referrals/reviews are outstanding for patients
Link to the Action set is within the PowerPlan
Launched in Diabetes Week July 2013
6 months data analysis in progress
Preliminary Findings
8892 Admissions analysed (patients over 54, multiple day stay)
6721 HbA1C orders (70% autogenerated)
1791 patients had HbA1C > 6.5%
380 (21% ) was not previously known
34 Type 1 Diabetes
1,295 Type 2 Diabetes
10 Other (gestational diabetes, steroid induced etc)
Launched in Diabetes Week July 2013
6 months data analysis in progress
Preliminary Findings
8892 Admissions analysed (patients over 54, multiple day stay)
6721 HbA1C orders (70% autogenerated)
1791 patients had HbA1C > 6.5%
380 (21% ) was not previously known
34 Type 1 Diabetes
1,295 Type 2 Diabetes
Other (gestational diabetes, steroid induced etc)
Medical – 1986, 959, 187
Surgical – 1373, 494, 84
Separate manual audits conducted of units with high prevalence of diabetes – oncology, respiratory and stroke to obtain characteristics of patients with diabetes
Gen med had 1072 patients 360 (33%) had know diabetes and 62 (5.8%) had previously undiagnosed diabetes
All acute admissions, >1 day, 6 months (7/2013 to 1/2014), age>54 years
6716 admissions in total
4388 medical, 28% HbA1c ≥6.5% (n=1234)
2328 surgical, 21% HbA1c ≥ 6.5% (n=483)
34% of all inpatients ≥54 years have diabetes
29% known diabetes
5% undiagnosed diabetes
Higher HbA1c is associated with increased readmission rates
Higher HbA1c is associated with longer length of stay in surgical patients