Upload
timothy-nichols
View
214
Download
0
Embed Size (px)
Citation preview
Broad societal determinants of CVD and health
Dubai 6/1/2006
Issues for the Pure study
• Health transition (Britain as an example)• Gender differences during health transition • At what age does the influence of broad social
factors on adult disease start?• (Income inequality, social capital, social
comparison)• Is a prospective study necessary?• Contingent causes, proximal causes and
understanding mechanisms
Preston. Population Studies, 1975
Economic development and health: selection of the Pure study countries may determine the answers
Health transition, coronary heart disease and stroke in Britain
Different trajectories in men and women over time, and socio-economic patterning changing
CVD trends (from Lawlor et al BMJ 2001)
0
1000
2000
3000
4000
5000
6000
1921 1931 1941 1951 1961 1971 1981 1991 1998
CHD - M CHD - F Stroke - M Stroke - F
Age-adjusted mortality rates from heart disease by sex, US 1900-98
0
100
200
300
400
500
600
700
800
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990
rate
per
100
,000 male
female
Coronary Heart Disease and stroke share risk factors
• Raised blood pressure• High blood cholesterol• Smoking• Insulin resistance / diabetes• Physical inactivity
The paradox
• So – if coronary heart disease and stroke share risk factors –
• Why are the secular trends different?– And why no epidemic of stroke during the 20th
century?
What caused the epidemic of coronary heart disease?
• Increased intake of dietary fat • Increase in smoking • Increase in obesity?• Decrease in physical activity?
BUT
• Secular decline in blood pressure
Stroke: two major pathologiesHaemorrhage Ischaemia
which are rather different…
haemorrhage Ischaemia
but look the same to doctors…
Isch
aem
ia:h
aem
orrh
age
ratio
Year
Ischaemia:haemorrhage ratio, autopsy data
Fitted values
1920 1940 1960 1980 2000
0.25
0.5
1.0
2.0
4.0
Lawlor et al. Lancet 2002
Ratio of ischaemia to haemorrhage,1930 to 2000
Mor
talit
y / 1
00 0
00
Year
Total stroke rate Estimated haemorrhagic rate
Estimated ischaemic rate
1940 1960 1980 2000
10
20
40
80
160
Stroke trends by sub-type
Lawlor et al. Lancet 2002
Risk factors by stroke subtype
Haemorrhagic
Blood pressure +++
Smoking +/-
Cholesterol –
Obesity +
Ischaemic
Blood pressure ++
Smoking +++
Cholesterol +++
Obesity ++
Changing social distribution of CVD risk factors
Moteiro CA. Bull WHO 2004;82:940-6
Age of influence of socio-environmental factors for blood
pressure
Blood pressure trends
• Marked declines among 18 year olds from 1948-1968 in UK
• Clear cohort effects in declines in blood pressure in U.S.
Early life influences on adult blood pressure
• Adverse social circumstances
• Low birthweight
• Poor height gain in early childhood (especially in leg length)
• Artificial feeding
• Infancy sodium intake
• ? Low physical activity / obesity ?
Lawlor and Davey Smith. Current Opinion in Nephrology and Hypertension 2005;14:259-64
Childhood social class and stroke subtype:
Manual vsManual vs+
non-manual non-manual
Haemorrhagic 2.84 (1.12-7.20) 3.22 (1.15-9.03)
Ischaemic 1.25 (0.77-2.03) 0.92 (0.53-1.61)+risk factor adjustedHart and Davey Smith; J Epidemiol Community Health 2003
Height quintile RR associated with 10 cm increase in height
1 2 3 4 5
All stroke
Relative rate 1 0.96 0.82 0.77 0.70 0.80 (0.73 to 0.88)
Haemorrhagic
Relative rate 1 1.16 0.93 0.66 0.54 0.70 (0.51 to 0.97)
Ischaemic
Relative rate 1 1.16 1.17 1.02 0.91 0.87 (0.71 to 1.06)
Stroke by height: Paisley and Renfrew study
McCarron et al. J Epidemiol Community Health 2001;55:404-405
Stroke subtype by number of siblings
0-2 3-4 5-6 7+ per sib per
sib+
Haemorrhagic 1 1.63 1.79 2.33 1.111.11
Ischaemic 1 0.86 1.00 1.36 1.03 1.01
+adjusted SEP and risk factorsHart and Davey Smith, J Epidemiol Community Health 2003
Systolic blood pressure mmHg (mean, 95% CI)
Systolic blood pressurea
Diastolic blood pressure
Diastolic blood pressurea
Dehydration admission
100.4 101.4 62.4 62.7
No dehydration admission
98.9 98.9 56.5 56.5
Difference 1.5 (-5.8, 8.9) 2.5 (-4.3, 9.3) 5.9 (0.6, 11.3) 6.2 (1.04, 11.4)
P difference 0.7 0.47 0.03 0.02
a adjusted for sex, age, height and BMI at the time of blood pressure measurement, birthweight and gestation, maternal education
Davey Smith et al JECH, in press
Income inequality, social capital, social comparison
Comparing Canada and the U.S.
Ross, et al., BMJ (2000)
0.15 0.19 0.23 0.27Median Share of Income
200
300
400
500
600
Rat
e pe
r 10
0,00
0 P
opul
atio
n
Working Age (25-64) Mortality by Median ShareU.S. and Canadian Metropolitan Areas
U.S. cities (n=282) with weighted linear fit (from Lynch et al. 1998)Canadian cities with weighted linear fit (n=53) (slope not significant)
Mortality Rates Standardized to the Canadian Popluation in 1991
FlorenceSC
JacksonvilleNC
SiouxCityIA
MonroeLA
ChicagoIL
TuscaloosaAL
Prince George
WashingtonDC
Oshawa
NewYorkNY
LosAngelesCA
NewOrleansLA
BryanTX
McallenTX
SiouxCityIA
AppletonWI
PortsmouthNH
PineBluffAR
FlorenceSC
MonroeLA
AugustaGA
Montreal
Vancouver
Toronto
Barrie
Oshawa
Shawinigan
GS3 Aug. 6, 1999 2:40:20 PM
Canada Paradox?
0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28
Median Share of Income
200
300
400
500
600
Wo
rkin
g A
ge
(2
5-6
4)
Mo
rta
lity
US AUSSWEUKCAN
New York
Stockholm
Sydney
Income Inequality and Working-Age Mortality528 Metropolitan Areas in Five Countries, 1990/91
Toronto
London
Ross, et al (unpublished)
American “exceptionalism”?
Income inequality in 1949 versus all-cause age-adjusted (2000 standard) mortality in 1949-51
WVWA
VA
VT
UT
TX
TN SD
SC
RI
PA
OR
OK
OH
ND
NC
NY
NM
NJ
NH
NV
NE
MT
MO
MS
MN
MI
MA
MD
ME
LA
KY
KSIA
INIL
ID
HI
GA
FL
DE
CT
CO CA
AR
AZ
AK
ALR2 = 0.011
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Ag
e-ad
just
ed m
ort
alit
y p
er 1
00,0
00
Income inequality in 1959 versus all-cause age-adjusted (2000 standard) mortality in 1959-61
WYWI
WV
WA
VAVT
UT TX
TN
SD
SC
RI
PA
OR
OK
OH
ND
NC
NY
NM
NJ
NH
NV
NE
MT
MO
MS
MN
MIMA
MD
MELA
KY
KSIA
IN
IL
ID
HI
GA
FL
DE
CT
COCA
AR
AZ
AKAL
R2 = 0.0034
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Ag
e-ad
just
ed m
ort
alit
y p
er 1
00,0
00
1950 1960
Income inequality in 1969 versus all-cause age-adjusted (2000 standard) mortality in 1969
WY
WI
WV
WA
VA
VT
UT
TX
TN
SD
SC
RI
PA
OR
OK
OH
ND
NC
NY
NM
NJNH
NV
NE
MT
MO
MS
MN
MI
MA
MDME
LA
KY
KSIA
INIL
ID
HI
GA
FL
DE
CTCO
CA
AR
AZ
AK
AL
R2 = 0.0466
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Ag
e-ad
just
ed m
ort
alit
y p
er 1
00,0
00
1970
Income inequality in 1979 versus all-cause age-adjusted (2000 standard) mortality in 1979
WY
WI
WV
WA
VA
VT
UT
TXTN
SD
SC
RI
PA
OR
OK
OH
ND
NC NY
NM
NJ
NH
NV
NE
MTMO
MS
MN
MI
MA
MD
ME
LA
KY
KSIA
INIL
ID
HI
GA
FL
DE
CTCO
CA
AR
AZ
AK
AL
R2 = 0.1496
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Ag
e-ad
just
ed m
ort
alit
y p
er 1
00,0
00
1980
Income inequality in 1999 versus all-cause age-adjusted (2000 standard) mortality in 1999
WY
WI
WV
WA
VA
VT
UT
TX
TN
SD
SC
RI
PA
OR
OK
OH
ND
NC
NY
NMNJ
NH
NV
NEMT
MO
MS
MN
MI
MA
MDME
LA
KY
KS
IA
IN
IL
ID
HI
GA
FL
DE
CTCO
CA
AR
AZAK
ALR2 = 0.1917
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Age
-adj
uste
d m
orta
lity
per
100,
000
Income inequality in 1989 versus all-cause age-adjusted (2000 standard) mortality in 1989
WY
WI
WV
WA
VA
VT
UT
TX
TN
SD
SC
RI
PA
OR
OKOH
ND
NC NY
NM
NJ
NH
NV
NE MT
MO
MS
MN
MI
MA
MD
ME
LA
KY
KSIA
INIL
ID
HI
GA
FL
DE
CT
CO
CA
AR
AZ
AK
AL
R2 = 0.3357
650
700
750
800
850
900
950
1000
1050
1100
1150
1200
1250
1300
1350
1400
1450
1500
1550
1600
30 32 34 36 38 40 42 44 46 48 50 52
Gini coefficient
Age
-adj
uste
d m
orta
lity
per
100,
000
1990
2000
Material conditions or social comparison?
Check-In Seat Leg
Room Food Service
Sleeping
Health Relevant Material Conditions?
What generates health inequalities after a
long airline flight?
Or: seeing Martin McKee in business class?
Measures of social position can give some insight ..
Unadjusted Odds Ratios and percent reduction of elevated risk on adjustment for
mortality by caste and standard of living indexInfant
(<1 year)
Caste Standard of living index
Other Caste 1.00 - Top fifth 1.00 -
SC 1.36 64% Fourth fifth 1.67 13%
ST 1.44 68% Third fifth 2.13 15%
OBC 1.27 56% Second fifth 2.64 16%
No Caste 1.39 44% Bottom fifth 3.09 15%
Subramanian et al, AJPH, in press
Unadjusted Odds Ratios and percent reduction of elevated risk on adjustment for
mortality by caste and standard of living index
Young adult
(19-44 years)
Caste Standard of living index
Other Caste 1.00 - Top fifth 1.00 -
SC 1.29 97% Fourth fifth 1.66 -2%
ST 1.87 47% Third fifth 1.91 0%
OBC 1.16 94% Second fifth 2.11 2%
No Caste 1.02 200% Bottom fifth 3.00 4%
Unadjusted Odds Ratios and percent reduction of elevated risk on adjustment for
mortality by caste and standard of living index
Middle-aged adult
(45-64 years)
Caste Standard of living index
Other Caste 1.00 - Top fifth 1.00 -
SC 1.23 87% Fourth fifth 1.19 -16%
ST 1.28 89% Third fifth 1.40 -10%
OBC 1.08 113% Second fifth 1.65 -6%
No Caste 1.14 43% Bottom fifth 1.91 -7%
Is a prospective study necessary?
Bias
Illustrations of how bias can generate associations between self-reported
exposures and outcomes
Relative rates for CHD outcomes according to stress
Incident Incident CHDStress angina ischaemia mortality
High 2.63 0.58 0.96
Medium 1.36 0.88 0.95
Low 1.00 1.00 1.00
p for trend <0.001 0.10 0.69Macleod et al BMJ 2001
Reporting tendency, angina and psychological stress
Perceived stress Mean reporting tendency score
Incident angina
Odds ratios
Adjustment A Adjustment B
High 0.77 2.63 2.28
Medium 0.52 1.36 1.27
Low 0.41 1.00 1.00
p for trend <0.001 <0.001 0.003
Macleod J et al. JECH 2002;56:76-77
Job control
Odds ratio for incident “Rose” angina
Low control 2.02 (1.22-3.34); Intermediate control 1.44 (0.86-2.39);High control 1.00
Bosma et al. BMJ 1997;314:558-65.
Fully subjective
High Exposure 2.66 Medium Exposure 1.37 Low Exposure 1.00
High Exposure 2.02 Medium Exposure 1.44 Low Exposure 1.00
Partially subjective
High Exposure 1.20 Medium Exposure 1.03 Low Exposure 1.00
High Exposure 1.65 Medium Exposure 1.13 Low Exposure 1.00
Fully objective
High Exposure 0.67 Medium Exposure 1.03 Low Exposure 1.00
High Exposure 1.17 Medium Exposure 1.16 Low Exposure 1.00
Macleod J et al. BMJ 2002;324:1247-51.Bosma H et al. BMJ 1997;314:558-65.Stansfeld SA et al. Int J Epidemiology 2002;31:248-255.
Contingent causes, proximal causes and understanding
mechanisms
TIME
CB
OUTCOMEMEASURE
Contingent links
Context dependentunderlying factors
TIMELUNGCANCER
Contingent links
Social position Smoking
TIMEOUTCOMEMEASURE
Contingent links
Context dependentunderlying factors BMI Insulin
resistance