1. Automated Patient-Specific Reporting for Chronic Disease Management in the Clinical Laboratory Dr Glenn Edwards MBBS, MD, FRCPA Medical Director, St John of God Pathology Perth, Western Australia glenn.edwards@sjog.org.au Disclosures Former CEO & Medical Director, Pacific Knowledge Systems Owns stock in PKS

2. Cost of Chronic Disease 2007 National Diabetes Fact Sheet Total: 23.6 million children and adults in the United States (7.8% of population) Cost of Diabetes (Direct + Indirect) $174 billion: Total costs of diagnosed diabetes in the US in 2007 Including undiagnosed diabetes, pre-diabetes, and gestational diabetes: $218 billion. Heart Disease The leading cause of death in US. (> 650,000 deaths/yr = 27%) Direct + Indirect cost: $475.3 billion in 2009 Chronic Kidney Disease 1 in 3 adults at increased risk; 1 in 7 adults have some sign of CKD Risk of death from CVD is 20 x need for dialysis or transplantation Cost (Medicare): $20.7 billion

3. Interpretative Reporting Testing guidelines improve outcomes and cost …but poor compliance and considerable variation Physicians want higher quality information and advice more engagement with laboratory/clinical pathologists support for compliance Current status vast majority of lab tests not interpreted by lab most discussion around esoteric testing RippleDown shown to support automated commenting in clinical lab Current lab role: Reactive, Result-focussed Our goal:Pro-active, Patient-focussed

4. Automated support for pathologist opinions LabWizard (RippleDown-based reporting application) Interfaced to LabTrak LIS Incremental knowledge acquisition based on live cases Supports pathologist opinion writing Focus on chronic disease Common, high volume tests (lipids, glucose, HbA1c, GTT, etc) Opinions designed to Reinforce management according to guidelines Detect guideline non-compliance Identify clinical management priorities & variation Make specific recommendations to correct non-compliance NOT canned comments – must be close match to manual task

5. Guidelines Guidelines for the Assessment of Absolute Cardiovascular Risk (National Heart Foundation of Australia, 2009) Position Statement on Lipid Management (National Heart Foundation of Australia and Cardiac Society of Australia and New Zealand, 2005) Diabetes Management in General Practice, 2009/10 (Diabetes Australia and Royal Australian College of General Practitioners) Chronic Kidney Disease (CKD) Management in General Practice (Kidney Health Australia, Melbourne, 2007) (Local practice) (Local opinion)

6. Key compliance issues identifying patients known to be at increased absolute risk for CVD identifying patients, not known to be at increased risk, who require risk assessment providing an internet URL as information to doctors identifying low- and high-risk individuals for CHD recommending LDL-cholesterol target levels relevant to risk category identifying non-compliance with LDL-cholesterol targets recommending statin therapy for patients at increased risk for CHD identifying patients with various hyperlipidaemias recommending target levels for triglycerides and HDL-cholesterol recommending triglyceride-lowering therapy for resistant hypertiglyceridaemia identifying patients at risk for Familial Hypercholesterolaemia identifying patients who require testing for secondary causes of hyperlipidaemia Hypothyroidism, Liver disease, Renal disease identifying patients at risk for diabetes recommending follow up testing for abnormal glucose tests On current or previous results identifying patients who require recurrent follow up (eg impaired glucose tolerance) recommending timing and selection of appropriate tests for follow up testing In diabetes, identifying non-compliance and recommending corrective action with: annual HbA1c testing for people with diabetes annual urine microalbumin testing for people with diabetes annual assessment of lipid profile in people with diabetes more frequent HbA1c testing in people with poor glycaemic control annual eGFR testing in people at increased risk for chronic kidney disease

7. LabWizard case view

8. Patient-specific report

10. Patient-specific reports Archive: 8,695 total patient reports Unique opinions: 2,406 Opinions given on 5 or fewer occasions: 2,188 (91%)

11. Specific, low frequency comments Results in diabetic range noted. Absolute Risk for CVD: ASSESS. LDL remains ABOVE target for high risk individuals. If diabetes not confirmed clinically, suggest glucose tolerance test. Also suggest check urine microalbumin:creatinine ratio (ACR). Given : 5 Total cholesterol > 7.5 mmol/L. Absolute Risk for CVD: INCREASED. LDL is ABOVE target for high risk individuals. Consider statin therapy. Aim for LDL 4.9 mmol/L. Suggest review for clinical stigmata of FH (eg tendon xanthomata, corneal arcus), review family history, and repeat full lipid profile. Also suggest check TSH to exclude sub-clinical hypothyroidism. Given : 5 Absolute Risk for CVD: ASSESS IF INDICATED (www.heartfoundation.org.au). LDL is now WITHIN target for high risk individuals. Given : 4 Absolute Risk for CVD: ASSESS. LDL is WITHIN target for high risk individuals. Dyslipidaemic pattern, with raised triglycerides and low HDL. Aim for triglycerides 1.0 mmol/L. Suggest check fasting glucose. Given : 4

12. Incremental Knowledge Acquisition

13. Auto-validation control

14. Auto-validation

15. Ambiguous data

16. Key findings Acceptance by pathologist group Patient specific Addressed entire patient record Doctor specific Modified for diabetes specialists, cardiologists and nephrologists Guideline concordance Near-natural language Opinions closely match to pathologists level of uncertainty Ambiguous or missing data Customise commentary Manual validation of relevant cases Filtered by clinical relevance, rather than “abnormal” values AV customised according to need Minimal impact on TAT

17. Conclusions Rippledown-based expert systems facilitate comprehensive provision of pathologist opinions Applicable to all areas of clinical pathology Support best clinical practice in chronic disease Not only “esoteric” tests Focus of clinical interpretation Trends Guideline/best practice compliance Therapy Care planning … not just diagnosis Restores clinical dialogue for lab/clinical pathologist

18. Automated Patient-Specific Reporting for Chronic Disease Management in the Clinical Laboratory Dr Glenn Edwards MBBS, MD, FRCPA St John of God Pathology Perth, Western Australia glenn.edwards@sjog.org.au

19. Conclusions

20. The myriad guidelines…. > 15% Absolute Risk of CVD in 5 years Diabetes and age > 60yrs Diabetes with microalbuminuria Males : ACR > 2.5 mg/mmol Females : ACR > 3.5 mg/mmol Moderate or severe CKD eGFR < 45 mL/min/1.73m 2 Known diagnosis of Familial Hypercholesterolaemia Hypertension SBP >= 180 mmHg DBP >= 110 mmHg Serum total cholesterol > 7.5 mmol/L

21. Rules by type

22. Non-compliance with guidelines in chronic disease Impact of education campaign Stern E. et al Int J Clin Pract. 2005 Oct;59(10):1126-30 ComplianceBeforeAfter Annual HbA1c60%85% HbA1c < 7%38%50% Annual eye clinic65%75% Annual microalbumin27%37%

23. “Non-compliance” Total DMs292 (13%) % of DMs HbA1c overdue27.009.25 ACR overdue88.0030.14 Total pts with overdue90.0030.82