CVD prevention & management: a new approach for primary care Rod Jackson School of Population Health University of Auckland New Zealand
Why bother about CVD in 1°care? In a population of 10,000 primary care patients, every year there are about: 10 coronary & stroke deaths 1 diabetic death 1 breast cancer death 1 prostate cancer death 1 suicide every year 1 road traffic death (1 cervical cancer death every 5 years) NZHIS annual mortality statistics
Blood pressure and CHD Law & Wald BMJ 2002;324:1570-6
PSC.
Reduction in stroke with combination BP lowering therapy in PROGRESS, regardless of baseline BP
There is no such thing as hypertension
CHD and SBP or Total cholesterol Blood pressure Systolic blood pressure (mmHg) Risk of coronary disease “Hyper- tension” “Hyperchol- esterolaemia” Total cholesterol (mmol/l) Cholesterol
Reduction in CV events with cholesterol lowering in Heart Protection Study, regardless of baseline cholesterol
There is no such thing as hypercholesterolaemia
Smoking and the risk of stroke Source: Bonita, 1999 Odds Ratio
‘Diabetes’ & body mass index
There is no such thing as obesity
Stroke, CHD, CVD & blood glucose Asia Pacific Cohort Studies Collaboration
HbA1c and microalbuminuria: Auckland, NZ Metcalf et al (unpublished) excl. diagnosed diabetics
There is no such thing as non- insulin dependant diabetes
Message Number 1: there is no such thing as hypertension or hypercholesterolaemia or obesity or type 2 diabetes and we all have CHD
a new paradigm: ‘risk factors’ ‘CVD risk factors interact’
Impact of multiple risk factors on CVD risk Jackson et al. Lancet :434-41
Relative Risk and 95% CI 34% 25% 0%5%10%15%20% Few or no participants had a history of stroke Estimated 5 year stroke event rate Treatment Control Most or all participants had a history of stroke or TIA 1.4% 5.1% Relative Reduction in strokesAbsolute Reduction in strokes / 5 years Absolute Effects Relative Effects ‘The bigger the CVD risk the bigger the benefit’: trials of BP lowering & stroke
15% 5 yr risk NZ threshold for CVD risk drugs
Message Number 2: Measure risk, not risk factors
Estimating clinical risk: Framingham Heart Study Sex Age Diabetes Smoking BP TC HDL (LVH) Anderson et al. Am Heart J. 1991;121:293-8
45 yr old man BP 150/90 mmHg non smoker TC 6.0 mmol/L HDLC 1.2 mmol/L new ‘diabetes’ 60 yr old man BP 150/90 mmHg smoker TC 6.0 mmol/L HDLC 1.0 mmol/L No ‘diabetes’ 5 yr CVD risk ≈ 10% 5 yr CVD risk ≈ 25% Are lipid +/or BP-lowering drugs indicated?
Clinical risk: short-term vs life-time?
Lifetime risk is clinically irrelevant The risk of death is 1 / person (100%) What’s clinically relevant is when it happens The lifetime CVD risk chart
Who should we treat? Everybody - because we all have CHD BUT the intensity of treatment should be directly proportional to the clinical risk and to the costs of treatment
Clinical risk treatment thresholds? $$$$$$$$$$$$$$$$$$$$$$$ At the clinical (absolute) risk that is affordable to individuals or populations Cheaper interventions should be initiated at lower risk levels
risk threshold for high cost treatment SBP treatment threshold for equal Rx benefit Clinical CVD risk (% per yr) low high Patient 1 Patient 2 Patient risk threshold for low cost treatment
Treatment goals? Based on clinical risk and the ‘costs’ of lowering risk
CVD risk threshold for drug treatment SBP target for equal Rx benefit Clinical CVD risk (% per yr) low high Patient 1 Patient 2 Patient CVD risk target for treatment
Message Number 3: Treat risk, not risk factors
The polypill Aspirin Statin Diuretic ± ACEI ± BB ± CCB metformin?
PREDICT: a clinical decision support system for CVD & diabetes risk assessment & management PREDICT is a computer programme that calculates CVD risk & provides E-B management recommendations
(Please note – dates are not representative as this is a test case) Workflow: Individual Patient Tracking
Sample Report –Group Data
Patient populations All clinical data is made non- identifiable with encrypted NHI and sent via secure internet connection for analyses Combining information on patients Stored anonymous CVD risk profiles Practice/PHO/DHB population needs assessment & service planning
patient-specific outcomes: hospital admissions, deaths Electronic medical record Enrolled population patient-specific CVD risk factor profiles NHI NHI (encrypted) Making new risk prediction charts
patient-specific outcomes: hospital admissions, deaths Electronic medical record Enrolled population patient-specific CVD risk factor profiles NHI NHI (encrypted) Link with encrypted NHI Making new prediction charts
Risk groups in first 30,878 patients from PREDICT
Results: estimated 5-year incidence of CVD event For prior CVD 5-year risk is: *Framingham score Mean est. 5-year incidence for Hx CVD is 28.4% (95%CI 26.3 to 30.4)
Results: events in risk groups in first 30,878 patients from PREDICT 47% 26% 63% of events occur in 21% of the people (high risk) 16% 11%
The potential magnitude of the population evidence base One assessment per practitioner every other day for 46 weeks/year = 115 per year A practitioner can assess all appropriate patients in less than 5 years 1000 practitioners could assess more than 100,000 patients per year ‘one every other day is ok’ ‘one every other day is ok’
Message Number 4: The next revolution in medicine will be electronic, not genomic The future is already here, its just not widely distributed It will be led by primary care
metabolic syndrome: ‘metabollocks!’
Relative stroke risk and usual Blood Pressure diastolic blood pressure (mmHg) PSC Lancet 1995;346: (45 prospective studies: 450,000 people 13,000 events) Relative Risk
Relative stroke risk and usual Blood Pressure diastolic blood pressure (mmHg) PSC Lancet 1995;346: (45 prospective studies: 450,000 people 13,000 events) Relative Risk DBP > 100 mmHg DBP > 95 mmHg DBP > 90 mmHg DBP > 80 mmHg