well-targeted primary prevention of cardiovascular disease: an underused high-value intervention?

Transcription

1 well-targeted primary prevention of cardiovascular disease: an underused high-value intervention? Rod Jackson University of Auckland, New Zealand October 2015

2

3

4 Lancet 1999; 353: Findings: Contribution to changing CHD mortality varied, but in populations in which mortality decreased, coronary-event rates contributed two thirds and case fatality one third. Interpretation: Over the decade studied ( through ), the 37 populations in the WHO MONICA Project showed substantial contributions from changes in survival, but the major determinant of decline in CHD mortality is whatever drives changing coronary-event rates.

5 Rose s prevention paradox the whole population risk axiom a large number of people at small risk may give rise to more cases in a population than a small number of people at high risk

6 rationale for a population-based approach: lowering blood pressure & stroke events 95% population 75% strokes 5% population 25% strokes RR of stroke MacMahon J Hyp 1996;14 (Suppl 6)

7 population-based approach Distribution shifting and/or high-risk hypertension treatment approach distribution Usual DBP shifting: (mmhg) BP or TC population- wide treatment of high BP or TC: only if high

8 has the low-hanging fruit of population-based CVD prevention all been picked? high-risk strategies population strategies

9 Rose s high-risk axiom all policy (including treatment) decisions should be based on absolute measures of risk

10 rationale for a population-based approach: lowering blood pressure & stroke events 95% population 75% strokes 5% population 25% strokes RR of stroke MacMahon J Hyp 1996;14 (Suppl 6)

11 relative risk of stroke relative stroke risk and usual Blood Pressure (45 prospective studies: 450,000 people 13,000 events) diastolic blood pressure (mmhg) PSC Lancet 1995;346:

12 BP & absolute IHD mortality risk by age PSC. Lancet 2002; 360:

13 BP & absolute IHD mortality risk by age PSC. Lancet 2002; 360:

14 5-yr absolute CVD risk (%) clinical impact of a single risk factor depends on combined effect of multiple risk factors * * 50 yr old woman

15 5-yr absolute CVD risk (%) patients with high absolute risk benefit most from treatment

16 CVD events avoided per 1000 treated avoidable CVD events per 1000 treated by baseline combined risk and extent of systolic blood pressure-lowering 5 year absolute CVD risk (%) BPLTTC. Lancet 2014; 384:

17 Vascular deaths avoided per 1000 treated Figure 5 avoidable vascular deaths per 1000 treated by baseline combined risk and extent of LDL lowering with statins 5 year absolute CVD risk (%) CTTC. Lancet 2012; 380:

18 rationale for a population-based approach: lowering blood pressure & stroke events 95% population 75% strokes 5% population 25% strokes RR of stroke MacMahon J Hyp 1996;14 (Suppl 6)

19 rationale for high-risk approach: treating high absolute risk patients & CHD events 15% population 75% CHD events? Absolute Voss et al. Int J Epidemiol 2002;31:

20 treat absolute risk not single risk factors

21 hypertension, hypercholesterolemia (and type-2 diabetes?) are not clinically relevant diagnoses

22 only absolute risk is clinically relevant

23 how can you measure a patient s absolute CVD risk?

24 NZ risk charts for estimating patients absolute risk (based on Framingham) Jackson R. Br Med J. 2000;320:

25 Framingham cohort 5,215 US White men and women Aged 30 to to 1983 (12 yrs) BP, Smoking, DM, TC, HDL CHD events

26 how often do you use the CVD risk charts? NZ GPs 1999: (n=500, resp. rate=83%)* *after 5 years of intensive nation-wide education & distribution of multiple risk charts Arroll et al. NZ Family Physician 2002:29:177-83

27 how relevant is a US CVD risk prediction study from the 1970s to a multi-ethnic NZ populations in the 21 st century?

28 PREDICT in PHOs: electronic decision support for CVD risk prediction & management 2002

29

30

31

32 PREDICT was designed to: Patient population Electronic medical record in primary care electronic decision support get current best evidence on risk & management into clinical practice

33 & to simultaneously generate new evidence on risk from clinical practice Patient population Electronic medical record in primary care NHI PREDICT: electronic decision support NHI patient-specific outcomes: hospital admissions, deaths encrypted NHI patient-specific CVD risk factor profiles

34 Cumulative numbers PREDICT 1 care recruitment , , , , ,594 32, ,

35 National mortality database linked National hospitalisation database National Virtual diabetes register PREDICT in PHOs web-based platform in primary care Regional laboratory (TestSafe) database National drug dispensing database by e-nhi National PHO enrolment database

36 1 prevention cohort by ethnicity aged years: Men Women Total (205,274) 114,463 90,811 European/other 74,002 57,757 Maori 14,142 12,583 Pacific 16,372 13,490 Indian 9,947 6,981 with no hx of CVD, renal disease or AF

37 1 prevention cohort events by type non fatal fatal All CVD (4,595) 4, MI 1, Other CHD 1, Stroke TIA PVD CHF Coronary procedures Peripheral procedures ,000 person-years follow-up; average 3 years & range 0-10 years

38 0 5 Kaplan-Meier Observed event rate observed vs predicted risk: Framingham score Framingham score Predicted event rate: Framingham Men 1 prevention score

39 0 5 Kaplan-Meier Observed event rate observed vs predicted risk: PREDICT score PREDICT score Predicted event rate: new score Men 1 prevention score

40 CVD events during follow-up in PREDICT population years, by clinical history Population events during f/u 83% % 47% 4% 17% 36% with prior CVD renal dis. CHF, AF other

41

42

43

44 CVD events during follow-up in PREDICT population years, by clinical history Population events during f/u 17% % 13% 47% 4% 17% 17% 19% with prior CVD renal dis. CHF, AF other

45 rationale for high-risk approach: treating high absolute risk patients & CHD events 15% population 75% CHD events? Absolute Voss et al. Int J Epidemiol 2002;31:

46 CVD events during follow-up in PREDICT population years, by clinical history Population events during f/u 34% population 81% CHD events 17% 66% % 47% 4% 17% 17% 19% with prior CVD renal dis. CHF, AF other

47 CVD events during follow-up in PREDICT population years, by clinical history Population events during f/u 30% population 74% CHD events 13% 13% 70% 4% 47% 17% 10% 26% with prior CVD renal dis. CHF, AF with diabetes no CVD or diabetes

48 Rose s prevention paradox the whole population risk axiom a large number of people at small risk may give rise to more cases in a population than a small number of people at high risk

49 well-targeted primary prevention of cardiovascular disease: an underused high-value intervention? Rod Jackson University of Auckland, New Zealand October 2015

50 BL proportion on treatment (%) vascular risk management: Auckland under-treatment over-treatment CVD No CVD: >=20% risk No CVD: >=15% risk No CVD: 10-15% risk No CVD: <10% risk 0 months Mehta et al. Eur J Prevent Cardiol. 2014;21:

51

52 CVD events during follow-up in PREDICT population years, by clinical history Population 70% events during f/u 13% 13% 4% 47% 17% 7% 29% with prior CVD renal dis. CHF, AF with diabetes no CVD or diabetes

53 CVD events during follow-up in PREDICT population years, by clinical history Population 73% events during f/u 13% 14% 46% 17% 37% with prior CVD with diabetes no CVD or diabetes

54 CVD events by history of CVD in NZ: (PREDICT n=270,000) % population events % 10% 40% 0 no CVD hx CVD hx unpublished

55 CVD events by history of CVD in NZ: (PREDICT n=270,000) % % population % events % 10% 40% 0 no CVD hx CVD hx unpublished

56 BL vascular risk management: Auckland % under-treatment over-treatment CVD No CVD: >=20% risk No CVD: >=15% risk No CVD: 10-15% risk No CVD: <10% risk 0 months Mehta et al. Eur J Prevent Cardiol. 2014;21:

57 CVD events avoided per 1000 treated % of population & numbers how should we choose treatment thresholds? yr risk (%)

58 % of population & numbers who should have their risk predicted? <10% 10-15% 15-20% 15-20% 5 yr risk (%) choose the most efficient way of identifying those meeting treatment criteria (e.g. by age, sex, medical hx)

59 5-yr combined CVD risk (%) combined risk of CVD: effect of increasing blood pressure & other CVD risk factors Reference: 50 yr old females Jackson et al. Lancet :434-41

60 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

61 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

62 relative risk of death age-specific mortality in men according to SBP & age: relative risk of death Age (years) Geoffrey Rose. BMJ 1981;282:

63 relative risk of death absolute risk of death age-specific mortality in men according to SBP & age: (a) relative & (b) absolute risk Age (years) Age (years) Geoffrey Rose. BMJ 1981;282:

64 relative & absolute benefits from treating hypertension according to age & Age (yr) presence of CV-renal abnormality Cardiovascular -renal abnormality Relative treatment effectiveness (%) Lives saved per 100 treated (absolute) < > Geoffrey Rose. BMJ 1981;282:

65 5 year CVD risk (%) absolute risk of CVD associated with increasing blood pressure & other CVD risk factors Reference: 50 yr old females Jackson et al. Lancet :434-41

66 5 year CVD risk (%) absolute risk of CVD associated with increasing blood cholesterol & other CVD risk factors Reference: 50 yr old females Jackson et al. Lancet :434-41

67 absolute CVD risk & glycaemia: HbA1c ± other CVD risk factors Ref: non- smoking, non-diabetic woman, 55 years, TC:HDL <4.5, SBP <140 mmhg Epic Norfolk unpublished

68

69 CVD events by history of CVD in NZ: (PREDICT n=270,000) % population events % 10% 40% 0 no CVD hx CVD hx unpublished

70 CVD events by history of CVD in NZ: (PREDICT n=270,000) % % population % events % 10% 40% 0 no CVD hx CVD hx unpublished

71 blood pressure & IHD mortality by age PSC. Lancet 2002; 360:

72 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

73 differences in cardiovascular risk in different ethic groups & different regions

74 using the original Framingham functions in a Chinese population

75 using recalibrated Framingham functions in a Chinese population

76

77 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

78 CVD events avoided per 1000 treated % of population & numbers how should we choose treatment thresholds? yr risk (%)

79 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

80 % of population & numbers who should have their risk predicted? <10% 10-15% 15-20% 15-20% 5 yr risk (%) choose the most efficient way of identifying those meeting treatment criteria (e.g. by age, sex, medical hx)

81 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

82 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

83 35 yr old male Overweight smoker Non diabetic BP 140/ 80 mmhg TC 6.0 mmol/l HDLC 1.0 mmol/l TC/HDL = yr CVD risk = 4% but long-term risk

84 Why predict short-term CVD risk?

85 Blood pressure lowering & stroke Progress Lancet 2001; 358:

86 Lipid lowering & CVD HPS Lancet 2002; 360: 7 22

87 Lifetime risk of CVD to age 95 yrs: Framingham 3.2% of men 4.5% of women Lloyd-Jones et al. Circ 2006;113:

88 Predicting life-time CVD risk No D iabet es Nonsmoker Smoker Ratio of Total Cholesterol:HDL Diabet es Nonsmoker Smoker Ratio of Total Cholesterol:HDL / / /95 140/85 120/75 AGE /95 140/85 120/75 180/ / /95 140/85 120/75 AGE /95 140/85 120/75 180/ /95 140/85 120/75 AGE / /95 140/85 120/75 180/ /95 140/85 120/75 AGE / /95 140/85 120/75.is clinically irrelevant Ratio of Total Cholesterol:HDL Ratio of Total Cholesterol:HDL

89 35 yr old male Overweight smoker Non diabetic BP 140/ 80 mmhg TC 6.0 mmol/l HDLC 1.0 mmol/l TC/HDL = yr CVD risk = 4% but long-term risk

90 key questions on CVD risk prediction why treat predicted (combined) risk rather than individual risk factors? which prediction tool for which population? how should we choose treatment thresholds? who should have their risk predicted? should predicted risk be modified for patients already on treatment? should we treat short-term or long-term risk? how should we communicating risk? heart age & heart forecast

91 Drs Sue Wells & Andrew Kerr (UoA)

92

93 5 year CVD risk (%)

94 5 year CVD risk (%)

95 5 year CVD risk (%)

96 5 year CVD risk (%)

97

98

99

100 400,0 deaths from heart disease & other causes: NZ 350,0 300,0 250,0 Total Cancer ASR Ischaemic Heart Disease ASR Cerebrovascular Disease ASR Chronic lower respiratory diseases ASR Other forms of heart disease ASR Pneumonia and Influenza ASR Diabetes ASR Motor vehicle accidents ASR Intentional self-harm ASR Homicide ASR 200,0 150,0 100,0 50,0 0,

101 % population-based approach Distribution shifting and/or high-risk hypertension treatment approach Usual DBP (mmhg) distribution shifting: BP or TC population- wide treatment of high BP or TC: only if high

102

103 relative risk of stroke relative stroke risk and usual Blood Pressure (45 prospective studies: 450,000 people 13,000 events) DBP > 90 mmhg DBP > 80 mmhg DBP > 95 mmhg 50% of 60 year olds DBP > 100 mmhg 5% of 60 year olds diastolic blood pressure (mmhg) PSC Lancet 1995;346:

104

105

106

107

108

109

110