1. Julia Critchley
Professor of Epidemiology
St George’s, University of London UK
June 13 2013
Thanks: Simon Capewell, Martin O’Flaherty,
Peter Phillimore, Susanne Logstrup, Sophie
O’Kelly, Muriel Mioulet, Lars Ryden, Ilaria
Leggeri, Robin Ireland, Philip James, Hilary
Graham, Maddy Bajekal, Margaret Whitehead,
Peter Whincup, Earl Ford, Pedro Marques-
Vidal, Rosalind Raine, Sarah Wild, Ann Capewell
Funding: EU, MRC, BHF, NIHR
Inequalities in CVD risk factors
Causes, Consequences & Challenges

2. Presentation Outline
• Coronary Heart Disease (CHD)
• Mortality trends
• Risk factor trends
• Socio-economic circumstance (SEC)
• IMPACTsec: explaining recent mortality trends by
SEC
2

3. Trends in age-standardised mortality rates,
Men 65+: Q1(affluent) vs Q5(deprived) 1982-2006
Total mortality, per 100,000 CHD mortality, per 100,000
3
0.0
1000.0
2000.0
3000.0
4000.0
5000.0
6000.0
7000.0
8000.0
9000.0
10000.0
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
Q1
Q5
National
0.0
500.0
1000.0
1500.0
2000.0
2500.0
3000.0
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
Q1
Q5
National

4. Systolic Blood Pressure (mm Hg):
trends by deprivation quintiles
Men 55+ Women 55+
130
132
134
136
138
140
142
144
146
148
150
1994
1995
1996
1998*
2001*
2003*
2006*
2008*
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
130
132
134
136
138
140
142
144
146
148
150
1994
1995
1996
1998*
2001*
2003*
2006*
2008*
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
4

5. Diabetes (%): trends by deprivation quintiles
Men 55+
0%
5%
10%
15%
20%
1994
1998
2003
2006
National IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5
Women 55+
0%
5%
10%
15%
20%
1994
1998
2003
2006
National IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5
5

6. Obesity (%): trends by deprivation quintiles
Men 55+ Women 55+
5%
10%
15%
20%
25%
30%
35%
40%
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
5%
10%
15%
20%
25%
30%
35%
40%
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
6

7. IMPACTsec
explaining recent UK mortality trends
+ stratification by socio-economic circumstances
(sec)
7

8. 0
5000
10000
15000
20000
25000
q1 q2 q3 q4 q5
Expected
Observed
CHD Mortality fall 2007 vs 2000
by IMD quintiles
Target DPPs = 38, 068
8
Affluent Deprived

9. IMPACT model
Explaining the CHD mortality fall 1981-2000
• 70% mortality fall due to
risk factor
reductions
• 40%: due to evidence-
based therapies
Unal, Critchley & Capewell Circulation 2004 109(9) 1101

10. IMPACT model
Explaining the CHD mortality fall 1981-2000
• 70% mortality fall due to
risk factor
reductions
• 40%: due to evidence-
based therapies
-80000
-60000
-40000
-20000
0
Unal, Critchley & Capewell Circulation 2004 109(9) 1101
68,230 fewer
CHD deaths
1981 2000

11. IMPACT model
Explaining the CHD mortality fall 1981-2000
• 70% mortality fall due to
risk factor
reductions
• 40%: due to evidence-
based therapies
-80000
-60000
-40000
-20000
0
Risk Factors worse +13%
Risk Factors better -70%
Treatments -40%
Unal, Critchley & Capewell Circulation 2004 109(9) 1101
68,230 fewer
CHD deaths
1981 2000

12. IMPACT model
Explaining the CHD mortality fall 1981-2000
• 70% mortality fall due to
risk factor
reductions
• 40%: due to evidence-
based therapies
-80000
-60000
-40000
-20000
0
Risk Factors worse +13%
Obesity (increase) +3.5%
Diabetes (increase) +4.8%
Physical activity (less) +4.4%
Risk Factors better -70%
Smoking -41%
Cholesterol -9%
Population BP fall -9%
Deprivation -3%
Other factors -8%
Treatments -40%
AMI treatments -8%
Secondary prevention -11%
Heart failure -12%
Angina: CABG & PTCA -4%
Angina: Aspirin etc -5%
Hypertension therapies -3%
Unal, Critchley & Capewell Circulation 2004 109(9) 1101
68,230 fewer
CHD deaths
1981 2000

13. -80000
-60000
-40000
-20000
0
2000 2007
-20k
-18k
+4k
IMPACTsec: CHD Deaths
prevented England 2000-2007

14. -80000
-60000
-40000
-20000
0
38,000 fall
(~90% explained)
2000 2007
-20k
-18k
+4k
IMPACTsec: CHD Deaths
prevented England 2000-2007
Bajekal et al. Plos Medicine 2012 9:6

15. -80000
-60000
-40000
-20000
0
Risk Factors worse + 11%
BMI (increase) + 2%
Diabetes (increase) + 9%
Risk Factors better -49%
Smoking - 4%
Cholesterol - 6%
SBP fall - 33%
Physical inactivity - 1%
Fruit & Veg - 5%
Treatments uptake change -52%
AMI/NSTEACS - 1%
2’ post MI - 9%
2’ post-revasc - 2%
Stable Angina - 13%
Heart failure - 10%
Hypertension therapies - 4%
Hyperlipidemia Rx - 12%
Unexplained 11%2000 2007
-20k
-18k
+4k
IMPACTsec: CHD Deaths
prevented England 2000-2007
38,000 fall
(~90% explained)
Bajekal et al. Plos Medicine 2012 9:6

16. Distribution of deaths prevented by
socio-economic group (IMD quintiles)
-10%
0%
10%
20%
30%
40%
50%
60%
IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5 England
PercentoftotalDPPs
Treatments Risk factors Unexplained
Affluent Deprived
16

17. Main overall messages
• 35% CHD mortality decline accelerated after 2000
• Nationally, the proportion of mortality decline explained
by Treatments and Risk Factors
changed from 40%:60% (in 1980-2000 period)
to treatments 55%: 45% risk factors (since 2000)
17

18. Main overall messages
• CHD mortality decline accelerated post-2000 (35% fall)
• Nationally, the proportion of mortality decline explained
by Treatments and Risk Factors
changed from 40%:60% (in 1980-2000 period)
to 55%:45% (since 2000)
• No SEC gradients in most treatment uptakes
• Bigger than expected falls in population Blood Pressure
• % unexplained by model:
small in deprived, larger in affluent
(range 2% Q5 to 20% Q1). Why?
18

19. UK Risk Factor
trends 2000-2007
19

20. UK Trends in
NHS burden
2000-2007
20

21. Smoking: trends by deprivation quintiles
Men 55+
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
Women 55+
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
21

22. Total Cholesterol (mmol/l):
trends by deprivation quintiles
Men 55+ Women 55+
5.0
5.5
6.0
6.5
7.0
1994
1998
2003
2006
2008
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
5.0
5.5
6.0
6.5
7.0
1994
1998
2003
2006
2008
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
22

23. ‹#›

24. 24Trends in unstable angina admissions. Hospital admission rates by
age, sex and socio-economic circumstance quintile (SEC) in 1999
and 2007. Figure 2a: Men, 2b: Women. Source: Hospital Episode Statistics

25. 25
Trends in heart failure admissions. Hospital admission rates by age, sex and
socio-economic circumstance quintile (SEC) in 1999 and 2007.
a: Men, b: Women. Data source: Hospital Episode Statistics

26. Treatment trends
2000-2007
generally equitable
(comprehensive
national health service)
26

27. CVD PREVENTION
27

28. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Birth
AtheromaArtery Thrombosis
Survival

29. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Birth
AtheromaArtery Thrombosis
Survival

30. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Birth
First Stroke or
Heart Attack
AtheromaArtery Thrombosis
Survival

31. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Sudden
Death
(common)
LuckyTypical
decline
NO Symptoms
Symptoms
Birth
First Stroke or
Heart Attack
Survival

32. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Sudden
Death
(common)
LuckyTypical
decline
NO Symptoms
Symptoms
Birth
Secondary prevention Health services
First Stroke or
Heart Attack
Survival

33. CVD process: in an individual
Capewell et al 2009
Middle Age Age (years) Youth
100%
0%
Sudden
Death
(common)
Typical
decline
NO Symptoms
Symptoms
Birth
Primary
Prevention
Secondary prevention
First Stroke or
Heart Attack
Health services
Disease
Promotion
Survival
Lucky

34. CVD Prevention in a POPULATION
Capewell et al 2009
70 Age (years)
100%
0%
60
Primary
Prevention
First Stroke or
Heart Attack
80
Marketing
Survival

35. Capewell et al 2009
70 Age (years)
100%
0%
60
More
Marketing
Primary
Prevention First Stroke or
Heart Attack
80
CVD Prevention in a POPULATION
Survival

36. Capewell et al 2009
70 Age (years)
100%
0%
60
Marketing
Eg  tobacco control
EFFECTIVE
Primary
Prevention
DELAYED First Stroke
or Heart Attack
80
CVD Prevention in a POPULATION
Survival

37. Capewell et al 2009
70 Age (years)
100%
0%
60
HEALTH PROTECTION
Eg by tobacco or salt
legislation
EFFECTIVE
Primary
Prevention
First Stroke or Heart
Attack PREVENTED
80
CVD Prevention in a POPULATION
Survival

38. CVD prevention strategies
• High Risk Individual approach
• Population-based approach

39. 130 160110
Systolic
BP0
30
20
10
Prevalence
%
Blood Pressure
distribution in the
population
CVD prevention approaches
120 140 150

40. 130 160110
Systolic
BP
0
30
20
10
Prevalence
%
Blood Pressure
distribution in the
population
CVD prevention: High risk individual
approach
120
SBP >140 mmHg
140 150

41. 130 160110
0
30
20
10
Prevalence
%
Blood Pressure
distribution in the
population
120
BP >140 mmHg
Medications
CVD prevention: High risk individual
approach
Systolic
BP 140 150

42. 130 160110
0
30
20
10
Prevalence
%
Shifting Blood
Pressure distribution
Population-based CVD prevention strategy
120
Systolic
BP 140 150

43. 130 160110
0
30
20
10
Prevalence
%
Shifting Blood
Pressure distribution
120
Systolic
BP 140 150
Population-based CVD prevention strategy

44. 130 160110
0
30
20
10
Prevalence
%
Shifting Blood
Pressure distribution
120
Fewer BP >140 mmHg
Less treatments
Systolic
BP 140 150
Population-based CVD prevention strategy

45. – Farmers’ subsidies to stop dairy & beef ,
& increase fruit & berry production (Finland)
– Support food reformulation (All)
Whole-population approach for
preventing CVD: successful policies

46. – Farmers’ subsidies to stop dairy & beef ,
start fruit & berry production (Finland)
– Support food reformulation (All)
– Banning transfats (Denmark, Switzerland, Austria)
– Slashing dietary salt (Finland)
– Promoting smoke-free public spaces
(Ireland, UK ,Italy etc)
Whole-population approach for
preventing CVD: successful policies

47. Ireland: modelling reductions in
cardiovascular risk factors
Primary Prevention
Population Approach
 Risk Factors in everyone
Versus
High Risk strategy
using statin& bloodpressuremedications
BMC Public Health 2007 7 117

48. -1,500
-1,300
-1,100
-900
-700
-500
-300
-100
CHD prevention in Ireland 1985-2000:
Population v. High Risk
Strategies
Deaths prevented or postponed (Sensitivity analysis )
C h o l e s t e r o l
Population
diet
change
Population
secular BP
trends
High
Risk
Statins
Treating
High
Risk
Diet
change
in
CHD
patients
Blood
Pressure
BMC Public Health 2007 7
BMC Public Health.
2007; 7:117.

49. -1,500
-1,300
-1,100
-900
-700
-500
-300
-100
CHD prevention in Ireland 1985-2000:
Population v. High Risk
Strategies
Deaths prevented or postponed (Sensitivity analysis )
C h o l e s t e r o l
Population
diet
change
Population
secular BP
trends
High
Risk
Statins
Treating
High
Risk
Diet
change
in
CHD
patients
Blood
Pressure
BMC Public Health 2007 7 117

50. CVD PREVENTION
& INEQUALITIES
50

51. Will CVD prevention
widen health inequalities?
Capewell & Graham PLoS Medicine 2010

52. UK Department of Health programme:
NHS Health Checks
The UK high risk
approach
for preventing CVD
Capewell & Graham PLoS Medicine 2010

53. UK Department of Health programme:
NHS Health Checks
– All adults aged 40+ screened for CVD risk
– If 20%+ risk CVD event in the next ten
years, treat with:
• lifestyle advice plus
• tablets to reduce cholesterol & blood pressure
The UK high risk
approach
for preventing CVD
Capewell & Graham PLoS Medicine 2010

54. Tudor Hart’s “Inverse Care Law”
Tugwell’s “staircase effect”
J Tudor Hart . The inverse care law. Lancet 1971; 1; 405. P Tugwell; BMJ 2006; 332; 358
Evidence that high risk
approach
may increase social inequalities
Capewell & Graham PLoS Medicine 2010

55. Tudor Hart’s “Inverse Care Law”
• The availability of good medical care tends to
vary inversely with actual need
Tugwell’s “staircase effect” Disadvantage can
occur at every stage:
– Health beliefs, health behaviour, presentation
participation, persistence or adherence
J Tudor Hart . The inverse care law. Lancet 1971; 1; 405. P Tugwell; BMJ
2006; 332; 358
Evidence that high risk
approach
may increase social inequalities

56. Prescribing gradients
Long term adherence
Smoking cessation
Nutrition interventions in individuals
Oldroyd J. JECH 2008; 62:573. Thomsen R W, Br J Clin Pharm. 2005; 60;534;
Ashworth, M, QJof Amb Care Management: 2008; 31; 220;
Vrijens B, BMJ 2008;336:1114; Morisky D. Clin Hypertension 2008; 10; 348
Johnell K BMC Public Health 2005, 5:17 Chaudhry HJ. Current Ather.
Rep 2008; 10; 19; Bouchard MH, Br J Clin Pharmacol. 2007 63(6): 698
Evidence that high risk
approach
may increase social inequalities

57. Kivimaki, Marmot et al Lancet 2008
15 year risk of CHD death
• calculated in British men aged 55
• quantified the benefits of decreasing risk factors uniformly
across population
[systolic blood pressure 10mmHg
total cholesterol 2mmol/l & glucose  1 mmol/l ]
Evidence that whole POPULATION CVD
prevention reduces social inequalities

58. Kivimaki, Marmot et al Lancet 2008
15 year risk of CHD death
• calculated in British men aged 55
• quantified the benefits of decreasing risk factors uniformly
across population
[systolic blood pressure 10mmHg
total cholesterol 2mmol/l & glucose  1 mmol/l ]
• Would reduce the absolute mortality gap between affluent &
deprived by 70%
Evidence that whole POPULATION CVD
prevention reduces social inequalities

59. Diet interventions
Folic acid fortification of cereals (USA population1996)
Dowd IJE 2008; 37(5):1059
Evidence that whole POPULATION CVD
prevention reduces social inequalities

60. Diet interventions
Folic acid fortification of cereals (USA population1996)
Blood folate levels: Social gradients   70%
Dowd IJE 2008; 37(5):1059
Evidence that whole POPULATION CVD
prevention reduces social inequalities

61. Smoking
• cigarette price increases more effective in
deprived groups Townsend BMJ 1994; 309; 923
“increase in tobacco price may have the potential
to reduce smoking related health inequalities”
Main Meta-analysis. BMC Public Health 2008; 8; 178
Evidence that whole POPULATION CVD
prevention reduces social inequalities

62. ♥High Risk Strategies
to screen & treat individuals
typically widen social inequalities
CVD prevention
& health inequalities
VERDICT
Capewell & Graham PLoS Medicine 2010

63. ♥High Risk Strategies
to screen & treat individuals
typically widen social inequalities
♥Population wide policy interventions
usually narrow the inequalities gap
CVD prevention
& health inequalities
VERDICT
Capewell & Graham PLoS Medicine 2010

64. RESERVE
SLIDES
64

65. 65

66. WHO
Commission
on
Social
Determinants
of Health
2008

67. Life expectancy at birth (men)
Glasgow, Scotland (deprived suburb) 54
India 61
Philippines 65
Lithuania 66
Poland 71
Mexico 72
Cuba 75
US 75
UK 76
WHO Commission on Social Determinants of Health 2008

68. Life expectancy at birth (men)
Glasgow, Scotland (deprived suburb) 54
India 61
Philippines 65
Lithuania 66
Poland 71
Mexico 72
Cuba 75
US 75
UK 76
Glasgow, Scotland (affluent suburb) 82
WHO Commission on Social Determinants of Health 2008

70. • Improve conditions of daily life
• Tackle the inequitable distribution of
power, money & resources
• Measure & understand the problem
and assess the impact of action
WHO Commission on
Social Determinants of
Health
Three overarching recommendations:
http://www.euro.who.int/socialdeterminants/publications/publications

71. • CHD Mortality trends Global & UK
• UK trends by socio-economic circumstance (SEC):
• CHD Mortality trends; risk factor trends
• IMPACTsec: explaining recent UK mortality trends
• DISCUSSION 1: Main research messages for L&G
UK trends to 2020
• Risk factors; Treatments; CHD mortality
71

72. 72
Major risk factors:
projecting trends to 2020

73. Men
0
10
20
30
110
120
4
5
6
1994
1998
2003
2006
2008
2020
Total cholesterol mmol/l
24
26
28
30
1994
2008
2020
BMI kg/m
2
Q1
Q2
Q3
Q4
Q5
Policy-targets
Low-risk
Major risk factors: projecting trends to 2020
73

74. Men
0
10
20
30
110
120
4
5
6
1994
1998
2003
2006
2008
2020
Total cholesterol mmol/l
24
26
28
30
1994
2008
2020
BMI kg/m
2
Q1
Q2
Q3
Q4
Q5
Policy-targets
Low-risk
Major risk factors: projecting trends to 2020
74

75. Men
0
10
20
30
40
Smoking prevalence (%)
0
10
20
30
40
Diabetes prevalence (%)
0
10
20
30
40
50
60
70
80
90
Physical inactivity (%)
110
120
130
140
150
SBP mmHg
4
5
6
1994
1998
2003
2006
2008
2020
Total cholesterol mmol/l
24
26
28
30
1994
2008
2020
BMI kg/m
2
Q1
Q2
Q3
Q4
Q5
Policy-targets
Low-risk
Major risk factors: projecting trends to 2020 75

76. Men Women
0
10
20
30
40
Smoking prevalence (%)
0
10
20
30
40
Diabetes prevalence (%)
0
10
20
30
40
50
60
70
80
90
Physical inactivity (%)
110
120
130
140
150
SBP mmHg
4
5
6
1994
1998
2003
2006
2008
2020
Total cholesterol mmol/l
24
26
28
30
1994
2008
2020
BMI kg/m
2
Q1
Q2
Q3
Q4
Q5
Policy-targets
Low-risk
0
10
20
30
40
Smoking prevalence (%)
0
10
20
30
40
Diabetes prevalence (%)
0
10
20
30
40
50
60
70
80
90
Physical inactivity (%)
110
120
130
140
150
SBP mmHg
4
5
6
1994
1998
2003
2006
2008
2020
Total cholesterol mmol/l
24
26
28
30
1994
2008
2020
BMI kg/m
2
Q1
Q2
Q3
Q4
Q5
Policy-targets
Low-risk
Major risk factors: projecting trends to 2020 76

77. IMPACTsec 2020
Future scenarios to 2020
77

78. IMPACTsec 2020
Future scenarios to 2020: What will happen to
CHD inequalities if recent trends continue, or
targets met, or low-risk levels achieved?
• 3 scenarios
• Project past trajectories forwards (from mid-1990s)
• Health targets met for ALL social groups
• ALL social groups reach the healthy low-risk
profile
78

79. IMPACTsec 2020
Future scenarios to 2020: What will happen to
CHD inequalities if recent trends continue, or
targets met, or low-risk levels achieved?
• 3 scenarios
• Project past trajectories forwards (from mid-1990s)
• Health targets met for ALL social groups
• ALL social groups reach the healthy low-risk
profile
• Only Risk Factor trends considered
(assumed treatment uptakes remain equitable)
79

80. IMPACTsec Deaths potentially prevented in 2020
3000
6000
9000
12000
Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5
Men Women
Trends continue Policy targets Low-risk
80
Men

81. IMPACTsec Deaths potentially prevented in 2020
3000
6000
9000
12000
Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5
Men Women
Trends continue Policy targets Low-risk
81

82. IMPACTsec Deaths potentially prevented in 2020
3000
6000
9000
12000
Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5
Men Women
Trends continue Policy targets Low-risk
82

83. IMPACTsec Deaths potentially prevented in 2020
3000
6000
9000
12000
Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5
Men Women
Trends continue Policy targets Low-risk
83
20,000 Fewer deaths 50,000 fewer deaths 75,000 fewer deaths

84. Policy implications of 2020 ‘scenarios’
• Actuaries
• Falls in overall mortality rates anticipated
• Relative inequalities in CHD mortality likely to persist
• Risk factor trajectories suggest that SEC differentials
will continue to widen (unless stronger policy actions)
84

85. Policy implications of 2020 ‘scenarios’
• Actuaries
• Falls in overall mortality rates anticipated
• Relative inequalities in CHD mortality likely to persist
• Risk factor trajectories suggest that SEC differentials
will continue to widen (unless stronger policy actions)
• Health Policy Interventions
• Regulation & Legislation would have biggest impact
on reducing risk factor levels, especially in disadvantaged
(Capewell, NICE)
• ?Target disadvantaged groups, aim to level-up to
‘best’, ‘proportionate universalism’?
85

86. DISCUSSION 2
Burning questions for Istanbul?
86

87. DISCUSSION 2
Burning questions for Istanbul?
87
Trends in
CHD,
Cardiovascular
disease &
common
cancers
(NCDs) ?
Increasing
gap
between
rich & poor:
effects on
CHD?
Why recent
population
falls in
cholesterol
or in
blood
pressure?
Future
effects of
tobacco
legislation?
Transfat
bans?

88. Key questions (Istanbul)
1. What are the key policy changes or new drug developments that
would affect mortality trends in the near future?
2. How will the gap in life expectancy between SEC change over,
say next 30 years?
3. How will the gap in life expectancy between genders change
over, say next 30 years?
4. What are the key drivers for the future fall in mortality for
different SEC?
5. How would these mortality drivers affect different SEC?
88

89. IMPACTsec team
• Legal & General – Dr Madhavi Bajekal, Dr Shaun
Scholes, Hande Love
• UCL – Prof Rosalind Raine,
• Liverpool University – Prof Simon Capewell, Dr Martin
O’Flaherty, Dr Nat Hawkins
• Advisory Group
89

90. Which factors explained ‘excess’ CHD
mortality in Q5 compared to Q1?
(PROVISIONAL ANALYSIS)
90
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Excess CHD deaths:
Q5 vs Q1, 2007
5.8k
2.6k
Higher treatments
uptake in Q1 + 26%
Lower risk factor
levels in Q1 + 58%
% Unexplained by model + 16%
1.6k
Q1 Q5

91. 91
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Excess CHD deaths:
Q5 vs Q1, 2007
5.8k
2.6k
Treatments uptake + 26%
AMI/NSTEACS + 4%
2’ post MI + 5%
2’ post-revasc + 1%
Stable Angina + 7%
Heart failure + 9%
1’ statin/hypertension + 0%
Risk Factors + 58%
BMI + 2%
Diabetes + 11%
Smoking + 26%
Cholesterol (better Q5) - 3%
SBP fall + 12%
Physical inactivity + 1%
Fruit & Veg + 10%
Unexplained 16%1.6k
Q1 Q5
Which factors explained ‘excess’ CHD
mortality in Q5 compared to Q1?
(PROVISIONAL ANALYSIS)

92. • Treatments (numbers)
a) Eligible patients – Acute MI, Angina, Heart Failure
b) Treatment uptake: 10 (eg statins, BP ) + 20 prevention
medication+ surgical interventions (CABG, PTCA)
c) Case fatality rate by diagnosis (one year)
d) Estimated mortality red’n due to treatment (one year)
Calculation
• DPP for one year = a*b*c*d
• Change DPP= DPP final – DPP base
Parameters and calculation
92

93. Treatments 1- eligible patient counts
Condition Hospital Community
STEMI 60% AMI admissions
(HES)
NSTEMI + Unstable
Angina
(or NSTEACS)
40% AMI admissions +
100% UA admissions
(HES)
2’ post-revascularisation
(CABG/PTCA/PCI)
2000-2007 counts from
HES, adjusted for
survival
2’ post-MI GPRD
Chronic angina (no MI) GPRD
Heart Failure 50% admissions (HES) 50% prevalence (GPRD)
1’ lipid lowering drugs HSfE
1’ hypertensive treatment HSfE
93

94. Treatments 2 – medical therapies
Condition Drugs Interventions Other
STEMI MINAP PPCI & thrombolysis
(MINAP)
CPR – H (MINAP)
CPR – cmty (no change)
NSTEMI + Unstable
Angina
MINAP CABG/PCI (HES) CPR – H (MINAP)
2’ post-
revascularisation
(CABG/PTCA)
GPRD Rehabilitation (Bethell +
National Audit)
2’ post-MI GPRD Rehabilitation (Bethell +
National Audit)
Chronic angina (no MI) GPRD
Heart Failure - H NHS Heart Failure
Survey, 2005
Heart Failure - Cmty GPRD
1’ lipid lowering drugs HSfE (no CHD)
1’ hypertensive
treatment
HSfE (no CHD)
94

95. Treatments 3 - challenges
• Low or Negative Tx DPPs – because numbers admitted
higher in 2000 than 2007 eg for HF
– Applied treatment uptake rates in 2000 to patient counts in
2007. Difference gives net DPPs
• Patient overlaps
• Mants & Hicks adjustment to calc benefits from drug
combinations
95

96. Risk Factor methods

97. Parameters and calculation
Continuous measures
– Systolic BP, BMI, total cholesterol, fruit and vegetable
consumption
• Regression approach:
(deaths base year)* (1-e (∆ in RF*beta coeff))
Binary measures
– Smoking, diabetes, physical activity
• PARF (population attributable risk factor) approach:
(deaths base year)* (PARF final – PARF base)
97

98. Risk Factors 1 - description
Risk factor HSfE survey years Description
Current cigarette smoking 2000-7 Self-reported status
SBP (mmHg) 2000-7, except 2004 Mean of the 2nd and 3rd readings
Fruit and vegetable
consumption
2001-7 Measured in portions per day
Body Mass Index 2000-7 Weight (kg)/ height squared (m2)
Total cholesterol (mmol/l) 1998, 2003, 2006 Subdivisible into those on lipid-lowering
drugs and those not on drugs
Physical (in) activity 1998, 2003, 2006 % not doing 30+ minutes of moderate or
vigorous leisure –time activity at least 5
days/ per week.
Diabetes 1998, 2003, 2006 Doctor diagnosed diabetes, excl
pregnancy
98

99. Risk factors 2 - challenges
• Sample sizes per year too small for precise estimates by
age (7), sex (2) and SEC (5)
• Fixed gradient approach used for SEC estimates
– Pooled survey data (2000-07) by SEC and calc scaling factor
for each SEC relative to the national level for each age/sex groups
– Applied SEC scaling factors to England values in 2000 & 2007
(or nearest available year)
– Where end-points 7+ years, adjusted down change estimates
(eg *7/8 for 1998, 2006 est)
99

100. Other model parameters
Treatments
• Case fatality rates: Canadian IMPACT
• RRR due to Tx: Canadian IMPACT + updated to most
recent published (Nat Hawkins)
Risk Factors
• Beta coeffs: US IMPACT + Cholesterol (ICELAND) + new
Fruit and Veg (Duchet)
• Relative risk – US IMPACT
100

101. Treatment uptake - summary
• Uptake of all drugs increased in all SEC groups between
2000 and 2007.
• No socioeconomic gradients in the uptake of treatment -
NHS delivering equitable service.
• Treatments in the community increased most: eg drug
treatment for angina + 2’ prevention doubled, from around
30% to over 60%.
• However, uptake levels were still below optimal and need
to increase further.
101

102. Main overall messages
• CHD mortality decline accelerated post-2000 (35% fall)
• Nationally, the proportion mortality decline explained
by Treatments and Risk Factors
changed from 40:60 (in 1980-2000 period)
to 55:45 (in post-2000 period)
• No SEC gradient in Treatment uptake
• Bigger than expected falls in population Blood Pressure
• % unexplained by model small in deprived, larger in
affluent (range 2% Q5 to 20% Q1). Why?
102

103. LIMITATIONS: With hindsight..
• Extend time period – minimum 10 years (1997-2007)
• Micro-level GPRD data:
– for accurate drug combination uptake; compliance over year;
– Eligible patients tightly defined, eg excluding MIs recorded
10+ years ago, unconfirmed diagnosis?
• Include upstream risk factors – eg psychosocial
status – to explain the ‘unexplained’ part of the model?
• Refine sensitivity analysis method: Credible limits
for risk factor DPPs narrower than Tx – counter-intuitive!
103

104. 104
β Coefficients = % fall in CHD mortality per
unit decrease in risk factors
(from meta-analyses & cohorts , Ford et al, NEJM 2007 356 : 2388
Cholesterol lowering PSC 2007 Reduction in CHD deaths
 0.1mmol/l mean pop cholesterol   5%
Fruit & Veg Duchet J Nutrition 2006
1 portion/day   4%
Blood pressure PSC Lancet 2003
 1 mm Hg Systolic BP   3.5% (log -0.035)
Obesity Bogers, 2008
1 Kg/M2  BMI   2.5%
Diabetes InterHEART, 2004
1%  diabetic population   2%
Smoking InterHEART, 2004
1%  Smoking prevalence   1%
Physical Activity InterHEART, 2004
1%  inactive population   0.3%
Ford et al NEJM 2007

105. Population risk factor change 1980/2000:
Impact on CHD Mortality US example
3mmHg fall in systolic BP in women aged 55-64
CHD deaths Beta Risk Factor Deaths
in 1980 coefficient reduction prevented
1980-2000 or postponed (DPP)
a x β x c = a*(1-(EXPβ x c))
26,350 x -0.035 x 3 = 2700 DPP
SOURCES
Mortality Oxford PSC NHANES
statistics meta-analyses surveys
105
Ford et al NEJM 2007

106. AMI: Thrombolysis & Aspirin, Men 55-64 years
Patients Treatment Relative Case Deaths prevented
eligible uptake risk Fatality or postponed (DPP)
reduction
a x b x c x d = a x b x c x d
102,280 X 21% x 0.26 x 0.054 = 303
SOURCES
HES MINAP Estess & FTT US/??
statistics audits Meta-analyses
106
Treating individual CHD patients - impact
on population CHD mortality: US example
Ford et al NEJM 2007

107. Trends in AMI admissions 1999-2006
per 100,000 by deprivation quintiles
Males (3 year average) Females (3 year average)
1999 2000 2001 2002 2003 2004 2005 2006
q1
q2
q3
q4
q5
0.0
50.0
100.0
150.0
200.0
250.0
300.0
350.0
1999 2000 2001 2002 2003 2004 2005 2006
Q1
Q2
Q3
Q4
Q5
107

108. GPRD Data fuzziness: MI & Angina
(not to scale)
Schematic diagram
2006 – counts (pop prev in
brackets)
Post MI
(A)
Other unspecified
IHD
(C)
Angina in the
community
(B)
Post MI
6,109 (2.3%)
Other unspecified
IHD
9,222
Angina in the
community
15,658 (3.4%)
(D
)
(E)
(F)
(G
)
2,686 6,348
9,048
17,553
108

109. Fruit & veg (portions per day):
trends by deprivation quintiles
Men 55+ Women 55+
2.5
3.0
3.5
4.0
4.5
5.0
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
2.5
3.0
3.5
4.0
4.5
5.0
2002
2003
2004
2005
2006
2007
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
109

110. Meeting physical activity
recommendations (%):
trends by deprivation quintiles
Men 55+ Women 55+
5%
10%
15%
20%
25%
1998
2003
2006
2008
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
5%
10%
15%
20%
25%
1998
2003
2006
2008
National IMDQ1 IMDQ2
IMDQ3 IMDQ4 IMDQ5
110

111. Change in key risk factor levels
Men 25+ Age-standardised rates, by IMD quintiles
5.20
5.25
5.30
5.35
5.40
5.45
5.50
5.55
5.60
5.65
IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5
mmol/l
Total Cholesterol: 1998 v 2006
1998
2006
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5
ASRper100
Diabetes: 1998 v 2006
111

112. Risk factor trends by SEC: adults aged 55+
Annual % Δ Men Women
Significant
decrease
across
all SEC
groups
Smoking 
SBP 
Total cholesterol 
Smoking  (x Q4)
SBP 
Total cholesterol 
Significant
increase
across all
SEC groups
Obesity
Diabetes
Obesity (x Q2)
Diabetes
Mixed picture
by SEC
Phys activity increase: Q1-Q3
Fruit & veg increase: Q3
Phys activity increase: Q1-Q4
Fruit & veg increase: Q3-Q4
112Q1 = least deprived; Q5 = most deprived Scholes, SSM 2010

113. UK 1984-2004
Overall age-adjusted CHD mortality rates
0.00
100
200
300
400
500
600
700
800
900
year
CHDratesper10^6
men
women
2.7% per year (rate of decline)
3.7%
5.4%
1.7%
3.5%
4.2%

114. Men 65 – 74 years
• JoinPoint Analysis
• 3 periods identified
• annual percent change
increased
UK Age-specific
CHD mortality rates
1984-2004
Deaths per 105
1600
600

115. Rates Flattening in
Men 45-54 years
• 4 periods identified
• annual percent change
decreased
Men 65 – 74 years
• JoinPoint Analysis
• 3 periods identified
• annual percent change
increased
UK Age-specific
CHD mortality rates
1984-2004
Deaths per 105
40
20
Deaths per 105
1600
600

116. Treatment trends
2000-2007
116

117. Myocardial infarction treatments
Trends & socioeconomic gradients

118. • No socioeconomic gradients
• Most therapies already high uptake
• Clopidogrel uptake doubled
Myocardial infarction treatments
Trends & socioeconomic gradients

119. Secondary prevention
Trends & socioeconomic gradients

120. • No socioeconomic gradients
• Overall treatment levels approximately doubled
• Absolute treatment uptake 60 to 70%
Secondary prevention
Trends & socioeconomic gradients

121. Chronic angina treatments
Trends & socioeconomic gradients

122. • Socioeconomic gradients – greater uptake in most
deprived
• Overall treatment levels approximately tripled
• Treatment levels lower than MI or secondary prevention
Chronic angina treatments
Trends & socioeconomic gradients

123. Change in treatment: Statin &
ACE-I/ARB uptake Men 55-74
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5 Eng
Statins
2000
2007
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
IMDQ1 IMDQ2 IMDQ3 IMDQ4 IMDQ5 Eng
ACE-I/ARB
2000
2007
Affluent Deprived Affluent Deprived