1. No category

Articles
A novel risk score to predict cardiovascular disease risk in
national populations (Globorisk): a pooled analysis of
prospective cohorts and health examination surveys
Kaveh Hajifathalian*, Peter Ueda*, Yuan Lu, Mark Woodward, Alireza Ahmadvand†,Carlos A Aguilar-Salinas†, Fereidoun Azizi†, Renata Cifkova†,
Mariachiara Di Cesare†, Louise Eriksen†, Farshad Farzadfar†, Nayu Ikeda†, Davood Khalili†, Young-Ho Khang†, Vera Lanska†, Luz León-Muñoz†,
Dianna Magliano†, Kelias P Msyamboza†, Kyungwon Oh†, Fernando Rodríguez-Artalejo†, Rosalba Rojas-Martinez†, Jonathan E Shaw†,
Gretchen A Stevens†, Janne Tolstrup†, Bin Zhou†, Joshua A Salomon, Majid Ezzati, Goodarz Danaei
Summary
Background Treatment of cardiovascular risk factors based on disease risk depends on valid risk prediction equations.
We aimed to develop, and apply in example countries, a risk prediction equation for cardiovascular disease (consisting
here of coronary heart disease and stroke) that can be recalibrated and updated for application in different countries
with routinely available information.
Methods We used data from eight prospective cohort studies to estimate coefficients of the risk equation with
proportional hazard regressions. The risk prediction equation included smoking, blood pressure, diabetes, and total
cholesterol, and allowed the effects of sex and age on cardiovascular disease to vary between cohorts or countries. We
developed risk equations for fatal cardiovascular disease and for fatal plus non-fatal cardiovascular disease. We
validated the risk equations internally and also using data from three cohorts that were not used to create the equations.
We then used the risk prediction equation and data from recent (2006 or later) national health surveys to estimate the
proportion of the population at different levels of cardiovascular disease risk in 11 countries from different world
regions (China, Czech Republic, Denmark, England, Iran, Japan, Malawi, Mexico, South Korea, Spain, and USA).
Findings The risk score discriminated well in internal and external validations, with C statistics generally 70% or
more. At any age and risk factor level, the estimated 10 year fatal cardiovascular disease risk varied substantially
between countries. The prevalence of people at high risk of fatal cardiovascular disease was lowest in South Korea,
Spain, and Denmark, where only 5–10% of men and women had more than a 10% risk, and 62–77% of men and
79–82% of women had less than a 3% risk. Conversely, the proportion of people at high risk of fatal cardiovascular
disease was largest in China and Mexico. In China, 33% of men and 28% of women had a 10-year risk of fatal
cardiovascular disease of 10% or more, whereas in Mexico, the prevalence of this high risk was 16% for men and 11%
for women. The prevalence of less than a 3% risk was 37% for men and 42% for women in China, and 55% for men
and 69% for women in Mexico.
Interpretation We developed a cardiovascular disease risk equation that can be recalibrated for application in different
countries with routinely available information. The estimated percentage of people at high risk of fatal cardiovascular
disease was higher in low-income and middle-income countries than in high-income countries.
Funding US National Institutes of Health, UK Medical Research Council, Wellcome Trust.
Introduction
The fact that treatment for cardiometabolic risk factors
such as blood pressure and cholesterol should be based on
disease risk, instead of the levels of individual risk factors,
is now widely accepted.1–3 Risk-based treatment is included
in clinical guidelines in many countries,4,5 although debate
continues on the appropriate threshold for treatment.
Risk-based multidrug treatment and counselling has also
been assessed as a cost-effective intervention for reduction
of the burden of non-communicable diseases worldwide.6
As part of the global response to non-communicable
diseases, countries have agreed to a target of 50% coverage
of multidrug treatment and counselling for people aged
40 years and older who are at high risk of cardiovascular
disease, including coronary heart disease and stroke.7
Risk-based treatment depends on prediction of
cardiovascular disease risk for each individual, which is
most accurately done with risk prediction equations
(often via risk charts or web-based risk calculators).3,8–10 To
measure progress towards the global non-communicable
disease treatment target, information about the number
of people in each country who are at high risk of
cardiovascular disease will be needed, which will require
an appropriate risk prediction equation and nationally
representative data for risk factors. A risk prediction
equation (or risk score) estimates a person’s risk of
cardiovascular disease during a specific period (eg,
10 years) based on their levels of risk factors and the
average cardiovascular disease risk in the population.
The risk score has a set of coefficients, usually hazard
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Lancet Diabetes Endocrinol 2015
Published Online
March 26, 2015
http://dx.doi.org/10.1016/
S2213-8587(15)00081-9
See Online/Comment
http://dx.doi.org/10.1016/
S2213-8587(15)00002-9
*Joint first authors
†These authors contributed
equally and are listed
alphabetically
Department of Global Health
and Population, Harvard School
of Public Health, Boston, MA,
USA (K Hajifathalian MD,
P Ueda PhD, Y Lu MSc,
G Danaei ScD); Department of
Internal Medicine, Cleveland
Clinic, Cleveland, OH, USA
(K Hajifathalian); The George
Institute for Global Health,
Nuffield Department of
Population Health, University
of Oxford, Oxford, UK
(M Woodward PhD); The George
Institute for Global Health,
University of Sydney, Sydney,
NSW, Australia (M Woodward);
Department of Epidemiology,
Johns Hopkins University,
Baltimore, MD, USA
(M Woodward); MRC-PHE
Centre for Environment and
Health (A Ahmadvand MD,
M Di Cesare PhD, B Zhou MSc,
M Ezzati FMedSci), and
Department of Epidemiology
and Biostatistics, School of
Public Health (A Ahmadvand,
M Di Cesare, B Zhou, M Ezzati),
Imperial College London,
London, UK; NonCommunicable Diseases
Research Center, Endocrinology
and Metabolism Population
Sciences Institute, Tehran
University of Medical Sciences,
Tehran, Iran (A Ahmadvand,
F Farzadfar MD); Department of
Endocrinology and
Metabolism, Instituto Nacional
de Ciencias Médicas y Nutrición,
Salvador Zubirán,
1
Articles
Mexico City, Mexico
(C A Aguilar-Salinas PhD);
Research Institute for
Endocrine Sciences, Shahid
Beheshti University of Medical
Sciences, Tehran, Iran
(F Azizi MD, D Khalili PhD);
Department of Preventive
Cardiology, Thomayer Teaching
Hospital, Prague, Czech
Republic (R Cifkova MD,
J Tolstrup PhD); National
Institute of Public Health,
University of Southern
Denmark, Copenhagen,
Denmark (L Eriksen MSc);
Center for International
Collaboration and Partnership,
National Institute of Health
and Nutrition, Tokyo, Japan
(N Ikeda PhD); Institute of
Health Policy and
Management, Seoul National
University College of Medicine,
Seoul, South Korea
(Y-H Khang MD); Statistical
Unit, Institute for Clinical and
Experimental Medicine,
Prague, Czech Republic
(V Lanska PhD); Department of
Preventive Medicine and Public
Health, School of Medicine,
Universidad Autónoma de
Madrid/Idipaz, Madrid, Spain
(L León-Muñoz PhD,
F Rodríguez-Artalejo MD); CIBER
of Epidemiology and Public
Health, Madrid, Spain
(L León-Muñoz,
F Rodríguez-Artalejo); Baker IDI
Heart and Diabetes Institute,
Melbourne, VIC, Australia
(D Magliano PhD, J E Shaw MD);
WHO, Malawi Country Office,
Lilongwe, Malawi
(K P Msyamboza PhD); Division
of Health and Nutrition Survey,
Korea Centers for Disease
Control and Prevention,
Cheongwon-gun, South Korea
(K Oh PhD); Centro de
Investigación en Salud
Poblacional, Instituto Nacional
de Salud Publica, Mexico City,
Mexico (R Rojas-Martinez PhD);
Department of Health
Statistics and Information
Systems, WHO, Geneva,
Switzerland (G A Stevens DSc);
and Department of
Epidemiology, Harvard School
of Public Health, Boston, MA,
USA (G Danaei)
Correspondence to:
Dr Goodarz Danaei, Department
of Global Health and Population ,
Harvard School of Public Health,
Boston, MA 02115, USA
[email protected]
See Online for appendix
2
ratios (HRs), each of which quantify the proportional
effect of the risk factor on cardiovascular disease risk.
Strong evidence from cohort pooling and multicountry
studies suggests that HRs for major cardiovascular
disease risk factors are similar in so-called Western
populations (North America, west Europe, Australia, and
New Zealand) and Asian populations, and over time in
the same population,11–13 although more data are needed
from Africa and Latin America. By contrast, average
cardiovascular disease risk differs substantially between
populations and over time because of differences in
mean risk factor levels and other determinants of
cardiovascular disease, such as access to and quality of
health care, and environmental, genetic, psychosocial,
and foetal and early childhood factors.
Therefore, risk prediction equations developed in one
population cannot be satisfactorily applied to other
populations, or even used in the same country years after
they were originally developed, because of changes in
average risk factor levels and disease risks.14,15 This
challenge is dealt with by recalibration of the model,
whereby the average risk factor levels and disease risks
are reset to the levels of the target population. For
example, the Framingham risk score has been updated
several times and recalibrated for application in different
countries, with mixed results.16–20 The Systematic
Coronary Risk Evaluation (SCORE) risk score, which
used European cohorts to estimate coefficients,9 has also
been recalibrated and applied to various European
populations with variable results.15,21,22 WHO developed a
series of regional cardiovascular disease risk charts in
2007.3 However, the coefficients of the risk prediction
equation used to develop these charts were taken from
epidemiological studies of one risk factor at a time—ie,
the coefficients for different risk factors were not derived
from the same regression model or even from a consistent
set of cohorts.23 Furthermore, the risk charts were only
produced at the regional and not country level, even
though the determinants of cardiovascular disease and
mortality vary across countries within the same region.
In this paper, we describe a risk prediction equation
that can be recalibrated and updated for use in different
countries with routinely available information.
Additionally, our risk prediction equation allows the age
pattern of cardiovascular disease risk to vary by sex and
across populations, and further allows the proportional
effect of risk factors to vary by age; neither of these
features have been included in previous risk scores.
Methods
Data sources
To estimate the coefficients of the risk prediction
equation, we pooled individual-level data from eight pro­
spective cohorts (Atherosclerosis Risk in Communities,
Cardio­vascular Health Study, Framingham Heart Study
original cohort, Framingham Heart Study offspring
cohort, Honolulu Heart Program, Multiple Risk Factor
Intervention Trial, Puerto Rico Heart Health Program,
and Women’s Health Initiative Clinical Trial; appendix
p 5). We pooled data from multiple cohorts because it
enhances statistical power, which in turn allows inclusion
of interaction terms between age or sex and risk factors,
and because pooling reduces the effect of between-cohort
variation on coefficients. We included participants who at
baseline were aged 40 years or older because there were
few events in younger participants. Patients were included
if they did not have a history of coronary heart disease or
stroke, were not missing data for the risk factors used in
the risk prediction equation, and did not have biologically
implausible risk factor levels (figure 1). Data were regarded
as implausible if total cholesterol was less than 1·75 mmol/L
or more than 20 mmol/L; systolic blood pressure was
less than 70 mm Hg or more than 270 mm Hg; and BMI
was more than 80 kg/m².
In our primary analysis, we developed a risk score for
fatal cardiovascular disease only. Although both fatal and
non-fatal cardiovascular disease are important for clinical
and public health applications, national data for average
death rates are much more reliable than those for disease
incidence, even in high-income countries. Therefore, a
risk score based on mortality can be more easily
recalibrated than one that includes both fatal and non-fatal
cardiovascular disease, an approach that was also used for
SCORE.9 We also present risk scores for fatal plus nonfatal cardiovascular disease for application in countries
with high-quality data for total cardiovascular disease event
rates. We used event data as defined by each cohort’s event
adjudication committee. Fatal cardiovascular disease was
defined as death from ischaemic heart disease or sudden
cardiac death (ICD10 codes I20–I25) or death from stroke
(ICD10 codes I60–I69); fatal plus non-fatal cardiovascular
disease was defined as fatal cardiovascular disease or nonfatal myocardial infarction (ICD10 codes I21–I22) or stroke.
For external validation, we used data from three cohorts
from countries other than the USA that had not been
used in the development of the risk prediction equation:
the Scottish Heart Health Extended Cohort, the Tehran
Lipid and Glucose Study, and the Australian Diabetes,
Obesity and Lifestyle cohort (appendix p 7). We did the
validation analysis for participants who were 40–80 years
old at baseline.
Statistical analysis
We used Cox proportional hazards regression to estimate
the coefficients of the risk scores. The models were
stratified by cohort and sex because the age and sex
patterns of cardiovascular disease incidence and mortality
can differ across populations—eg, events generally
happen at younger ages in low-income and middleincome countries than in high-income countries.24
Importantly, this formulation of the risk prediction
equation, described in detail in the appendix (pp 2–4), also
allows recalibration of the risk equation for each age and
sex group in any country with age-and-sex-specific mean
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Articles
65 560 participants from eight cohorts
44 409 male participants from seven cohorts*
21 151 female participants from five cohorts*†
11 086 excluded
2568 had history of coronary heart disease or stroke
at baseline
4203 younger than 40 years
4291 had missing data on age, systolic blood
pressure, total cholesterol, diabetes, or smoking
24 had extreme risk factor values‡
4345 excluded
1326 had history of coronary heart disease or stroke
at baseline
2327 younger than 40 years
662 had missing data on age, systolic blood
pressure, total cholesterol, diabetes, or smoking
30 had extreme risk factor values‡
33 323 eligible
16 806 eligible
During the first 15 years of follow-up
1703 fatal cardiovascular disease events
355 strokes
2525 non-fatal cardiovascular disease events
720 strokes
During the first 15 years of follow-up
562 fatal cardiovascular disease events
130 strokes
1252 non-fatal cardiovascular disease events
579 strokes
Figure 1: Flowchart of inclusion and exclusion of cohort participants
*The Honolulu Heart Program, Multiple Risk Factor Intervention Trial, and Puerto Rico Heart Health Program include only men, and the Women’s Health Initiative
Clinical Trial includes only women. †Only a subset of participants in the Women’s Health Initiative Clinical Trial had cholesterol and fasting glucose measurements by
design. ‡Extreme risk factor values indicate implausible data (total cholesterol <1·75 or >20 mmol/L, systolic blood pressure <70 or >270 mm Hg, and BMI >80 kg/m²).
risk factor levels and age-and-sex-specific cardiovascular
disease death rates. Some of the cohorts had a very long
follow-up time (eg, >50 years in the original Framingham
cohort); after such a long period of time, baseline risk
factor data might have little value in prediction of
cardiovascular disease events. Therefore, we used data
from a maximum follow-up of 15 years, after which we
deemed all participants administratively censored.
The risk factors in the model were systolic blood
pressure, serum total cholesterol, diabetes, and smoking
(age and sex were included in the risk score as a part of
event rate [appendix]). We defined diabetes using fasting
(≥126 mg/dL or 7 mmol/L), random (≥200 mg/dL or
11·1 mmol/L), or postprandial (≥225 mg/dL or
12·5 mmol/L) plasma glucose concentrations,25
depending on the available data in each cohort, or use of
insulin or oral hypoglycaemic drugs. We used total
cholesterol because it can be measured more easily and
with less cost than HDL or LDL cholesterol, and is
therefore measured more often in low-income and
middle-income countries.26 We considered using BMI in
addition to the other risk factors, but did not include it in
the final model because it did not improve risk
prediction. We allowed the coefficients of all risk factors
to vary by age by including interaction terms, because
the findings of prospective studies11,27,28 have shown that
cardiovascular disease HRs often decrease with age. We
also included interaction terms between sex and diabetes
and sex and smoking because evidence suggests the
existence of sex differences in the proportional effects of
these risk factors on cardiovascular disease.29,30
For validation of the risk prediction equation, we used
Harrell’s C statistic,31 which measures the ability of the
risk score to assign increased risk to individuals with
short time to event, a property known as discrimination.
We also assessed the calibration of the risk score using
the Hosmer-Lemeshow χ² test to compare the predicted
number of events during 10 years of follow-up with the
number of events seen (corrected for censoring with the
Kaplan-Meier estimator), stratified by deciles of risk. We
validated the model in three ways. First, we did internal
validation in the pooled cohorts used to estimate the risk
factor coefficients. Second, we iteratively withheld each
one of the eight cohorts from the Cox model and used
the other seven cohorts to estimate the coefficients; we
then validated the obtained model against the withheld
cohort. Finally, we validated the model against three
cohorts from outside the USA that had not been used to
estimate the risk prediction equation.
Application in national populations
We used the risk score to estimate the 10-year risk of fatal
cardiovascular disease in 11 countries with recent (2006
or later) nationally representative health examination
surveys in different regions of the world (China, Czech
Republic, Denmark, England, Iran, Japan, Malawi,
Mexico, South Korea, Spain, and USA; appendix p 6). We
first recalibrated the risk score for each country by
replacing the age-and-sex-specific average risk factor
levels from the cohorts with those observed in each
country’s health examination survey, and replacing the
age-and-sex-specific hazard of cardiovascular disease
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
3
Articles
1
2a
2b
3
4
5
Calculate the
individual’s risk factors’
distance from the
population mean by
subtracting
age-and-sex-specific
population mean risk
factor levels* from the
individual’s risk factor
levels (equation 2 in
appendix)
Calculate the main
effect of HR
coefficients
using the distances
of risk factors from
the population
mean of each of the
risk factors and the
corresponding
regression
coefficients
from the
proportional hazards
model (table 1;
equation 2 in
appendix)
Calculate the age interaction of HR
coefficients using the distances of
risk factors from the population
mean of each of the risk factors,
the age of the individual, and the
corresponding regression
coefficients from the proportional
hazards model (table 1; equation 2
in appendix)
Exponentiate
the sum of the
main effects
and the age
interaction
components for
each risk factor
to obtain HR for
that risk factor
Multiply the joint HR
with age-and-sex-specific
cardiovascular disease
death rate of the country†
to obtain the
cardiovascular disease
hazard for the year
(equation 2 in appendix)
Calculate
the survival
for the year
from the
cardiovascular
disease
hazard
(equation 3
in appendix)
Calculate the cumulative
survival and cumulative
cardiovascular risk
(equation 3 in appendix)
Update age each year
Multiply the
HRs of all risk
factors to
obtain the joint
HR (equation 2
in appendix)
Risk
factor
distance
from
population
mean
Risk
factor
distance
from
population
mean
Cardiovascular
disease death
rate† (for
the agesex-group
and country)
One-year
cardiovascular
disease
risk ψi(t)
Survival
exp (–ψi(t))
Cumulative Cumulative
survival
cardio∏exp (–ψi(t)) vascular
disease
risk
1–∏exp
(–ψi(t))
Step
Risk
factor
level at
baseline
Year 0
Population
mean risk
factor
level*(for
the agesex-group
and
country)
SBP
(10 mm Hg) 13·0 – 12·4
TC
Age 60 years (mmol/L)
5·5 – 4·8
Sex Male
Diabetes
Country USA (yes/no)
0 – 0·18
Year 2011
Smoking
1 – 0·19
(yes/no)
Main
term of
HR coefficient
Age 69 years
Year 2020
HRs
× 0·3412
+
0·6
× –0·0024
×
60
1·12
0·7
× 0·5776
+
0·7
× –0·0063
×
60
1·15
–0·18
× 1·4557
+
–0·18
× –0·0101
×
60
0·86
0·81
× 1·7657
+
0·81
× –0·0174
×
60
1·79
+
0·6
× –0·0024
×
61
1·12
+
0·7
× –0·0063
×
61
1·14
+
–0·18
× –0·0101
×
61
0·86
+
0·81
× –0·0174
×
61
1·77
··
··
··
··
··
··
··
··
··
··
··
··
+
0·6
× –0·0024
× 69
1·11
+
0·7
× –0·0063
× 69
1·10
+
–0·18 × –0·0101
× 69
0·87
+
0·81
× 69
1·58
Age 61 years
Year 2012
Year 9
Age
0·6
Year 1
··
··
··
Age
interaction
of HR
coefficient
··
··
··
× –0·0174
Joint
HR
6
Repeat step 2b–6 for
each of the remaining
years of risk prediction.
Age and cardiovascular
death rate are updated
annually
1·98 ×
0·00238
0·00470
0·99532
0·99532
0·00468
1·94 ×
0·00246
0·00478
0·99524
0·99057
0·00943
··
··
··
1·68
··
··
··
×
0·00326
··
··
··
0·00549
··
··
··
0·99452
··
··
··
··
··
··
0·95057
7
0·049
The cumulative risk
at the end of year 9
is the 10 year fatal
cardiovascular disease
risk (equation 3)
Figure 2: Procedure for recalibration and application of the (fatal) cardiovascular disease risk score
Figure shows the steps for application of the risk prediction equation for fatal cardiovascular disease to individuals in different countries through recalibration. The example used is a 60-year-old man
in the USA in 2011 with risk factor levels as shown in the figure. For recalibration of the fatal plus non-fatal cardiovascular disease risk score, coefficients from the corresponding model in the table and
rates for total cardiovascular disease events should be used. SBP=systolic blood pressure. TC=total cholesterol. HR=hazard ratio. *Mean risk factor levels for men aged 60–64 years in the USA in 2011
were obtained from the National Health and Nutrition Examination Survey (NHANES) survey 2011–2012. †Age-and-sex-specific cardiovascular disease death rates are available from national health
and statistics offices or from WHO. Because risk prediction needs information on average death rates for the whole prediction period, death rates need to be projected into the future, done here by use
of linear extrapolation of the natural logarithm of age-and-sex-specific deaths rates.
4
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Articles
with cardiovascular disease death rates from WHO
(appendix p 9).32 The detailed recalibration procedure for
country application is described in the appendix (pp 2–4),
and shown in figure 2.
We used the recalibrated risk scores to generate
country-specific risk charts for the eight countries for
which the health examination surveys covered the full
age range of 40–84 years. We used the recalibrated risk
scores and individual-level data from the health
examination surveys to estimate the distributions of
10-year risk of fatal cardiovascular disease in each
country. For presentation purposes, we used risk groups
of less than 3%, 3–6%, 7–9%, 10–14%, and 15% or more,
which—if a third of cardiovascular disease events are
assumed to be fatal, as is true in many high-income
countries—are nearly equivalent to less than 10%,
10–20%, 20–30%, 30–45%, and 45% or more for fatalplus-non-fatal cardiovascular disease.33–35 If case-fatality
rate is higher than a third—eg, in low-income countries
with little access to treatment—the fatal cardiovascular
disease cutoffs correspond to lower risks of fatal plus
non-fatal cardiovascular disease. To account for complex
survey design, we used sample weights for the country
results if they were available.
All analyses were done with Stata 11.0. The study
protocol was approved by the institutional review board
at the Harvard T.H. Chan School of Public Health
(Boston, MA, USA). The Globorisk risk prediction tool
will be available online in 2015.
Role of the funding source
The funders of the study had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. KH, PU, MDC, BZ, and GD had
full access to all the data in the study and GD had final
responsibility for the decision to submit for publication.
Results
We included 50 129 participants in the estimation of the
proportional hazards models (figure 1). The mean age at
baseline was 55 years (SD 9) and a third of eligible
participants were women (figure 1; appendix p 5).
During 15 years of follow-up, 4228 (12·7%) men and 1814
(10·8%) women had a first cardiovascular disease event;
of these events, 1703 in men and 562 in women were
fatal. Women had a lower risk of fatal cardiovascular
disease (10-year risk was 2·3% in women vs 3·9% in
men, p<0·0001). Smoking, diabetes, and increased
systolic blood pressure and serum total cholesterol were
associated with increased risk of fatal cardiovascular
disease and fatal plus non-fatal cardiovascular disease.
The magnitude of the association for all risk factors
decreased with age in both models (table), and the
associations of diabetes and smoking with cardiovascular
disease were stronger in women than in men. The age
and sex results are consistent with findings of previous
pooled analyses of prospective cohorts.11,27–30
Coefficient
HR†
Main effect
Age interaction term*
Systolic blood pressure
(per 10 mm Hg)
0·3412 (p<0·0001)
–0·0024 (p=0·0022)
Total cholesterol (per 1 mmol/L)
0·5776 (p<0·0001)
–0·0063 (p=0·00034)
1·18 (1·13–1·22)
Diabetes
1·4557 (p<0·0001)
–0·0101 (p=0·048)
2·21 (1·93–2·52)
Fatal cardiovascular disease
Diabetes and female
0·4430 (p=0·00045)
Smoking
1·7657 (p<0·0001)
Smoking and female
0·2649 (p=0·034)
··
1·20 (1·18–1·22)
1·56 (1·22–1·99)
–0·0174 (p=0·00079)
··
1·85 (1·66–2·06)
1·30 (1·02–1·67)
Fatal plus non-fatal cardiovascular disease
Systolic blood pressure
(per 10 mm Hg)
0·1077 (p=0·00039)
Total cholesterol (per 1 mmol/L)
0·5533 (p<0·0001)
–0·0059 (p<0·0001)
Diabetes
1·0721 (p<0·0001)
–0·0078 (p=0·029)
Diabetes and female
0·3902 (p<0·0001)
Smoking
2·0872 (p<0·0001)
Smoking and female
0·3384 (p<0·0001)
0·0002 (p=0·67)
1·13 (1·12–1·14)
··
1·77 (1·61–1·94)
1·48 (1·27–1·72)
–0·0250 (p<0·0001)
··
1·19 (1·17–1·22)
1·63 (1·52–1·74)
1·40 (1·23–1·60)
Data are logHR (p value) or HR (95% CI). HR=hazard ratio. *We included an interaction term between age and all risk
factors because the HRs for effects on cardiovascular disease decrease with age;11,27,28 therefore the HR at any age
depends on the main effect and interaction terms. †HRs for systolic blood pressure, total cholesterol, diabetes, and
smoking are shown at median age of event, which is 66 years for fatal cardiovascular disease and 64 years for fatal plus
non-fatal cardiovascular disease in the selected cohorts; HRs for diabetes and smoking are for men, and its interaction
with sex shows the additional risk among women.
Table: Coefficients and HR from the Cox proportional hazard model used to parameterise the risk scores
for fatal cardiovascular disease and fatal plus non-fatal cardiovascular disease
Our fatal cardiovascular disease risk score performed
well in internal validation, with a C statistic of 71% (95% CI
70–73) and calibration χ² of 15·5 (df 8, p=0·051). The
median C statistic was 73·5%, with a range of 60–78%
when data from different cohorts were held back and used
for validation. Predicted and observed risks of fatal
cardiovascular disease were similar across different
cohorts and deciles of risk (figure 3), and the median χ²
was 14·9 (df 8, p=0·061) with a range of 7·7 (df 8, p=0·46)
to 39·4 (df 8, p<0·0001) across the eight cohorts (appendix
p 7). For external validation, when applied in the Scottish
Heart Health Extended Cohort, the C statistic for the fatal
cardiovascular disease risk score was 74% (71–77) and
calibration χ² was 12·0 (df 8, p=0·15). For the Tehran Lipid
and Glucose Study, the C statistic was 83% (79–86) and
calibration χ² was 12·9 (df 8, p=0·12). For the Australian
Diabetes, Obesity and Lifestyle cohort, the C statistic was
84% (82–87) and calibration χ² was 44·3 (df 8, p<0·0001).
The less than optimal calibration of the Australian
Diabetes, Obesity and Lifestyle cohort happened mainly in
the highest decile of predicted risk (average observed risk
of 13·1% and average predicted risk of 22·4%). χ² was 15·0
(df 7, p=0·036) when we excluded this category.
When we applied the risk score in national populations,
we saw that, at any age and risk factor level, the estimated
10-year risk of fatal cardiovascular disease varied
substantially between countries. As shown in figure 4, risk
was lowest in Japan, South Korea, Spain, Denmark, and
England, and highest in China and Mexico for both sexes
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
For the risk prediction tool see
www.globorisk.org
For more on the National
Health and Nutrition
Examination Survey see http://
www.cdc.gov.nchs/nhanes.htm
5
Articles
Observed 10-year risk
0·3
ARIC
HHP
CHS
MRFIT
FHS
PRHHP
FHS-OFF
WHICT
0·2
0·3
0·1
0
0
0
0·1
0
0·2
Predicted 10-year risk
0·3
0·3
Figure 3: Observed and predicted 10-year risk of a fatal cardiovascular
disease event in risk score validation, by cohort and deciles of risk
Each point represents one decile of risk of one cohort. ARIC=Atherosclerosis Risk
In Communities. CHS=Cardiovascular Health Study. FHS=Framingham Heart
Study Original Cohort. FHS-OFF=Framingham Heart Study Offspring Cohort.
HHP=Honolulu Heart Program. MRFIT=Multiple Risk Factor Intervention Trial.
PRHHP=Puerto Rico Heart Health Program. WHICT=Women’s Health Initiative
Clinical Trial.
(and in Czech Republic and Iran, for which risk charts are
not shown because national surveys for these countries
only included participants aged ≤65 years). For example, a
non-smoking 65-year-old man with diabetes, a systolic
blood pressure of 140 mm Hg, and a total cholesterol of
6 mmol/L would have an estimated 10-year risk of fatal
cardiovascular disease of 5% in Japan versus 24% in China;
the risks would be 9% in Japan and 36% in China if he
smoked. Similarly, the 10-year risk of fatal cardiovascular
disease for a 65-year-old woman with the same risk factor
profile would be 3% in Japan or Spain versus 33% in China
if she did not smoke, and 6% versus 58% if she smoked.
Total (fatal plus non-fatal) 10-year cardiovascular
disease risk would be higher than shown in figure 4, by
the ratio of total-to-fatal cardiovascular disease event
rates. For example, total risk for a person aged 60 years
and living in a high-income country, where about a third
of all cardiovascular disease events are fatal, would be
roughly three times higher than shown in the figure.
Total-to-fatal cardiovascular disease ratio tends to
decrease with age—ie, more fatal cardiovascular disease
events at older ages.33,36–38 Total-to-fatal cardiovascular
disease ratio is likely to be decreased in low-income and
middle-income countries where more cardiovascular
disease events are fatal because of reduced access to
health care and treatment, and might be closer to two.
We identified substantial differences in fatal cardio­
vascular disease risk distributions across countries
(figure 5). In South Korea, 77% of men aged 40–84 years
had a predicted 10-year risk of fatal cardiovascular disease
of less than 3%, and only 7% had a predicted risk of 10%
or more. South Korea was followed by Spain (67% of men
with <3% fatal cardiovascular disease risk, and 9% of
men with ≥10% risk) and Denmark (62% of men with
<3% risk and 10% of men with ≥10% risk) in terms of
having a large proportion of men in low-risk groups. The
same three countries also had the largest proportions of
6
women in the low-risk group: in South Korea, 82% of
women had less than 3% risk of fatal cardiovascular
disease and 7% of women had 10% or more risk; in Spain
80% of women had less than 3% risk and 6% of women
had 10% or more risk; and in Denmark, 79% of women
had less than 3% risk and 5% of women had a risk of
10% or more. Overall, more people in Japan were in the
medium-to-high risk groups than in these three countries
because Japan has an older population. A larger
proportion of men and women were at 10% or more risk
of fatal cardiovascular disease in the USA than in
Denmark, South Korea, and Spain. In fact, the prevalence
of women at high risk in the USA (11%) and England
(10%) was more similar to Mexico (11%) than to other
high-income countries used as examples here. The
proportion of people at high risk of fatal cardiovascular
disease was largest in China, where 33% of men and
28% of women had a predicted 10-year risk of 10% or
more. These proportions translate to nearly 170 million
Chinese men and women aged between 40–84 years who
are at high risk of fatal cardiovascular disease. China was
followed by Mexico, in which 16% of men and 11% of
women were at 10% or more risk. The prevalence of less
than a 3% risk was 37% for men and 42% for women in
China, and 55% for men and 69% for women in Mexico.
When we separately analysed the risk distributions for
people younger than 65 years and older than 65 years
(which allowed the inclusion of three additional countries
with data only available for individuals <65 years of age),
the prevalence of high risk of fatal cardiovascular disease
was consistently highest in low-income and middleincome countries and in the Czech Republic (figure 5).
More than 90% of 65–84-year-old Chinese men and
women had a 10-year fatal cardiovascular disease risk of
10% or more; more than 80% of these men and 70% of
women in China had a risk of 15% or more. By contrast,
the proportion of older Japanese people with a 10 year
fatal cardiovascular disease risk of 10% or more was 28%
for men and 17% for women, and in Spain, the
proportions were 29% for men and 17% for women.
Discussion
Our risk-prediction equation and the risk charts that can
be generated with it have clinical and public health
applications. It can be used to identify individual patients
who are at high risk of cardiovascular disease and thus
need treatment, and to estimate the number of such
people in a country, which is needed to measure progress
towards the global non-communicable disease treatment
goal. In addition to the risk score and risk charts, our
results suggested a substantially increased prevalence of
people at high cardiovascular disease risk (eg, ≥10%
10-year risk of fatal cardiovascular disease) in low-income
and middle-income countries compared with highincome countries. Our findings are in line with those of
the 2014 report from the Prospective Urban and Rural
Epidemiological study, which noted that, at any level of
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
10
7
5
3
6
4
3
2
3
2
2
1
2
2
1
1
2
1
1
<1
9
6
4
3
5
3
2
2
3
2
1
1
180
160
140
120
180
160
140
120
180
160
140
120
180 2
160 1
140 1
120 <1
180 1
160 1
140 <1
120 <1
<1%
4
17
13
9
7
180 15
160 11
140 8
120 6
3
25
19
14
10
23
17
13
10
180
160
140
120
29 31
22 24
17 18
12 14
5
1%
2
1
1
1
3
2
1
1
4
3
2
1
7
5
3
2
12
8
6
4
6
3
2
1
1
4
3
2
1
5
4
2
2
8
6
4
3
13
9
7
5
2%
7
3
2
1
1
5
3
2
1
7
4
3
2
10
7
5
3
15
11
8
5
19 22 24
14 16 18
10 12 13
7
8 9
27
20
15
11
43 45
34 36
27 29
21 22
37
30
23
18
180
160
140
120
41
33
26
20
39
31
24
19
59
50
41
33
180
160
140
120
34
25
19
14
43
34
26
20
60
49
40
32
81
72
62
52
82
73
63
54
37
28
21
15
41 45
31 35
23 26
17 19
46 49 53
36 39 42
28 30 33
21 23 25
62 64 67
51 54 56
42 44 46
33 35 37
Smoker
79 80
70 71
60 61
50 51
5
3
2
1
7
5
3
2
17
12
8
5
6
9
5
4
2
30
22
16
11
3
5
3
2
1
7
5
3
2
5
6
7
8 10 14
5 7 9
3 4 6
2 3 4
11 14 18
8 10 12
5 6 8
3 4
5
15 19 23
10 13 16
7 9 11
5 6
7
≥15%
4
6
4
2
2
9
6
4
3
10 12
7
8
5
6
3
4
39 43
29 33
21 24
15 17
23 27 31
16 19 22
11 13 16
8 9 11
34
25
18
13
46 50 54 59
35 39 43 47
27 30 33 36
20 22 25 27
58 62 65 69
47 50 54 57
37 40 43 46
29 31 34 36
77 79 81 83
66 69 71 73
56 58 60 63
46 48 50 52
94
88
79
69
95
90
81
72
99
97
93
87
96
91
83
74
99
97
94
88
48
36
26
19
63
50
38
28
55
42
31
22
68
55
43
32
38
27
18
12
46
33
23
15
51
38
27
19
62
48
36
26
74
61
48
36
3
4
5
7
40
45
50
55
60
65
70
75
80
Age
(years)
29
22
16
12
43
34
27
21
31
24
18
13
45
36
29
22
34
26
20
15
48
38
30
24
3
2
1
1
5
4
3
2
3
2
1
1
6
4
3
2
2
1
1
1
4
3
2
1
8
5
4
2
3
4
5
6
1 1 1 1
<1 <1 1 1
<1 <1 <1 1
<1 <1 <1 <1
1 1 2
1 1 1
<1 1 1
<1 <1 <1
2
1
1
1
4
3
2
1
9 10 12 14
6 7 9 10
4 5 6 7
3 4 4 5
16 18 20 23
12 13 15 17
9 10 11 12
6 7 8 9
26
20
15
11
41
33
26
20
64
54
45
37
7
2
1
1
<1
3
2
1
1
5
3
2
1
9
6
4
3
16
11
8
6
25
19
14
10
36
28
21
16
50
40
32
25
28
21
15
11
40
31
24
18
55
45
36
28
31
23
17
13
43
34
26
20
57
47
38
30
Smoker
74 76
65 66
55 56
46 47
3
1
1
1
<1
2
1
1
1
4
3
2
1
9
6
4
3
4
2
1
1
<1
3
2
1
1
5
4
2
2
5
3
2
1
1
4
2
2
1
7
5
3
2
10 12
7 9
5 6
3 4
16 18 21
11 13 15
8
9 11
6
6 8
25
19
14
10
37
29
22
17
53
43
34
27
73
63
54
44
Non-diabetic
Non-smoker
65 67 68 69
56 57 58 59
46 47 48 49
38 39 40 41
Total cholesterol (mmol/L)
6
15 19 24 31 39
10 13 16 21 27
6
8 11 14 18
4
5
7 9 12
31
21
14
10
31 37 44
22 27 32
15 18 22
10 13 16
42
31
22
16
57
44
33
25
73 78 82 85
61 66 70 75
49 53 58 62
38 42 46 50
82 85 88 90
72 75 78 81
60 64 67 71
49 52 56 59
93
86
77
67
Smoker
99 99
96 97
92 93
86 86
20 25
14 17
9
11
6
8
26
18
12
8
36
27
19
13
51
39
29
22
69
56
45
34
79
68
57
46
92
84
75
65
98
96
91
85
Diabetic
Non-smoker
92 93 94 94
86 87 88 89
78 79 80 81
68 70 71 72
16 19
11 13
8
9
5
6
10–14%
7
11
7
5
3
11 14
8 10
5 6
3 4
14
10
7
4
5–9%
5
7
4
3
2
9
6
4
3
9 12
6 8
4 5
3 4
4
3–4%
3
4
2
2
1
6
4
2
2
8
5
3
2
12 14 17 20 24
8 10 12 14 17
6 7 8 10 12
4 5 6
7 8
26
19
14
10
42
32
24
18
55
44
34
26
75
64
53
43
92
85
77
67
Women
19 22 25 29 33
14 16 18 21 24
10 11 13 15 17
7 8 9 11 12
30
23
17
12
40
31
24
18
57
47
38
30
78
68
58
49
Non-diabetic
Non-smoker
60 62 63 64
51 52 53 54
42 43 44 45
34 35 36 37
China
78
68
58
49
6
3
2
1
1
5
3
2
1
8
6
4
3
15
10
7
5
7
4
3
2
1
6
4
3
2
10
7
5
3
17
12
8
6
24 27
17 20
12 14
9 10
35 38
26 29
19 22
14 16
47 50
37 40
28 31
22 23
60 62
49 51
40 42
31 33
77
67
57
48
66
55
45
36
68
57
47
38
70
60
49
40
73
62
51
42
3
1
1
1
<1
2
2
1
1
5
3
2
1
10
7
5
3
19
13
9
7
31
23
17
12
37 41
28 32
21 24
16 17
4
2
1
1
1
3
2
1
1
6
4
3
2
5
3
2
1
1
4
3
2
1
7
5
3
2
12 14
8 10
6 7
4 5
6
3
2
1
1
5
3
2
1
9
6
4
3
17
12
8
6
21 24 28
15 18 20
11 13 15
8 9 10
34
26
19
14
7
5
3
2
1
7
4
3
2
11
8
5
3
20
14
10
7
32
23
17
12
45
35
26
20
Smoker
91 92
84 85
76 77
66 67
3
4
2
2
1
6
4
2
2
10
7
5
3
4
5
3
2
1
7
5
3
2
5
7
4
3
2
9
6
4
3
13 15
9 11
6 7
4 5
19 22 26
13 16 18
9 11 13
6 8 9
31 35 40
23 26 30
16 19 22
12 14 16
45 50 54
35 39 43
26 29 33
20 22 25
61 64 68
49 53 56
39 42 45
30 33 35
75 78 80
65 67 70
54 57 59
44 46 49
91
83
75
65
Diabetic
Non-smoker
85 86 87 88
76 78 79 80
67 68 69 71
57 58 59 61
45 49 52 55 59
36 39 41 45 48
28 30 32 35 38
21 23 25 27 29
63
53
43
34
84
75
66
56
Men
35
26
18
13
50
38
29
21
63
51
40
30
75
63
52
41
84
74
64
53
93
87
79
70
6
7
9 11
6 7
4 5
2 3
12 15
8 10
5 7
3 4
19 23
13 16
9 11
6 7
30
22
15
11
45
34
25
18
59
47
36
27
71
60
48
38
82
72
61
51
93
86
78
69
Articles
Systolic blood pressure (mm Hg)
(Figure 4 continues on next
page)
7
8
1
1
<1
<1
1
<1
<1
<1
1
1
1
<1
1
180 1
160 <1 <1
140 <1 <1
120 <1 <1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
4
1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
<1%
1
1
1
<1
2
2
1
1
2
1
1
1
180
160
140
120
5
6
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
2
1
1
1
3
2
2
1
6
5
3
2
12
9
7
5
1%
<1
<1
<1
<1
<1
<1
<1
<1
3
2
1
1
6
4
3
2
5
4
3
2
5
3
2
2
180
160
140
120
11
8
6
5
10
8
6
4
23
18
14
11
180 10
160 7
140 5
120 4
180
160
140
120
2%
7
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
2
1
7
5
4
3
12
9
7
5
2
1
1
1
3
2
1
1
5
4
3
2
2
1
1
1
2
1
1
1
3
2
2
1
6
4
3
2
4
3–4%
3
6
2
1
1
<1
2
1
1
1
2
2
1
1
3
2
1
1
4
3
2
1
7
5
3
2
2
1
1
1
2
2
1
1
4
3
2
1
8
5
4
3
3
4
5
2
1
1
<1
2
1
1
1
2
2
1
1
3
2
1
1
5
3
2
2
6
2
1
1
1
2
2
1
1
3
2
1
1
4
2
2
1
5
4
3
2
9 10
6 7
4 5
3 4
≥15%
1
1
1
1
<1 <1
<1 <1
1
1
1
1
<1 1
<1 <1
2
1
1
<1
2
1
1
1
3
2
2
1
7
5
3
2
7
3
2
1
1
3
2
1
1
4
3
2
1
4
3
2
1
6
4
3
2
11
8
6
4
16 18 20
12 13 15
9 10 11
7 7 8
27 29 30 32
21 22 24 25
16 17 18 19
12 13 14 15
3
3
2
1
1
4
2
1
1
4
3
2
1
5
4
2
2
8
5
4
3
74
65
55
46
76
66
56
47
4
4
2
2
1
5
3
2
1
6
4
2
2
6
4
3
2
9
6
4
3
5
5
3
2
1
6
4
2
2
7
5
3
2
8
5
4
2
7
9
6
4
2
9
6
4
3
40
45
50
55
60
65
70
75
80
Age
(years)
2
1
1
1
3
2
1
1
5
4
3
2
2
1
1
1
3
2
2
1
6
4
3
2
3
2
1
1
4
3
2
1
6
5
3
2
9 10 10
6 7 8
5 5 6
3 4 4
3
<1
<1
<1
<1
4
5
6
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1 1
<1 1 1 1
<1 <1 <1 1
<1 <1 <1 <1
1 1 1 2
1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
1
1
1
<1
2
2
1
1
4
3
2
2
8
6
4
3
15 16 17 18
11 12 13 14
9 9 10 11
7 7 7 8
30
24
19
15
Non-smoker
31 32 33
25 25 26
19 20 21
15 16 16
Total cholesterol (mmol/L)
6
7
4
3
2
7
5
3
2
9 11
6 7
4 5
3 3
9 11
6 8
4 5
3 4
11 12 14
7 9 10
5 6 7
4 4 5
18 20 22
13 15 16
9 11 12
7 8 9
28 30 33 35
21 23 25 27
16 17 19 21
12 13 14 15
44 46 49 51
35 37 39 41
28 29 31 33
21 23 24 26
Smoker
72 73
62 64
53 54
44 45
14 16
10 11
7
8
5
6
26
19
14
11
42
33
26
20
71
61
51
43
Diabetic
Non-smoker
53 55 56 57
44 45 46 47
36 37 38 39
29 30 31 31
14 15
10 11
7
8
5
6
26
20
15
11
52
43
35
28
10–14%
7
2
1
1
1
2
2
1
1
3
2
1
1
3
2
1
1
4
3
2
1
7
5
4
3
12 13
9 10
7
7
5
5
5–9%
5
1 1 1
<1 1 1
<1 <1 1
<1 <1 <1
1 1 1
1 1 1
<1 <1 1
<1 <1 <1
1 1
1 1
1 1
<1 <1
2
1
1
<1
2
2
1
1
5
3
2
2
9 10 11
7 7 8
5 5 6
4 4 4
20 22
16 17
12 13
9 10
Smoker
37 38 39 40
30 30 31 32
24 24 25 26
19 19 20 20
17 18 19
13 14 15
10 10 11
7 8 8
36
29
23
18
Non-diabetic
Women
7
1
1
<1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
4
3
12
9
6
5
19
15
11
8
34
27
21
17
22
17
13
10
23
18
14
10
Smoker
38 39
31 32
24 25
19 20
3
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
4
3
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
9
7
5
3
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
1
2
2
1
1
3
2
2
1
5
3
2
2
8
6
4
3
12 13 14
9 10 11
7
7 8
5
5 6
21
16
12
9
37
30
24
19
Non-diabetic
6
2
1
1
<1
3
2
1
1
4
2
2
1
5
3
2
2
7
5
3
2
11
8
6
4
16
12
9
6
7
2
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
8
5
4
3
12
9
6
5
17
13
9
7
25 26
19 20
15 16
11 12
40 42
33 34
26 27
20 21
Men
3
1
1
<1
<1
2
1
1
<1
2
1
1
1
3
2
2
1
5
3
2
2
32
25
19
14
3
2
1
1
3
2
1
1
5
3
2
2
7
5
3
2
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
1
3
2
1
1
4
3
2
1
6
4
3
2
6
2
1
1
<1
3
2
1
1
4
3
2
1
6
4
3
2
8
6
4
3
13
9
7
5
17 18 20
12 13 15
9 10 11
7 7 8
9 10 12
7
7 8
5
5 6
3 4 4
15
11
8
6
27 28 30
21 22 23
16 17 18
12 13 14
47
39
31
25
Non-smoker
49 50 51
40 41 42
32 33 34
26 26 27
7
3
2
1
1
4
3
2
1
5
3
2
1
7
5
3
2
9
7
5
3
15
11
8
6
22
16
12
9
33
26
20
15
52
43
35
28
27
20
15
11
17 19
12 14
9 10
6 7
24
18
14
10
3
2
1
1
1
4
2
2
1
4
3
2
1
6
4
3
2
4
3
2
1
1
5
3
2
1
6
4
2
2
8
5
4
2
5
4
2
2
1
6
4
3
2
7
5
3
2
9
6
4
3
9 10 12
6
7 9
4
5 6
3 4 4
15
11
8
6
22
17
13
9
35 37 40
28 29 31
22 23 24
17 18 19
Smoker
57 58 59
47 48 50
38 40 41
31 32 33
Diabetic
31
24
18
13
44
35
27
21
62
52
43
35
6
5
3
2
1
7
7
4
3
2
8 10
5 7
3 4
2 3
9 11
6 7
4 5
3 3
11 13
8 9
5 6
4 4
14 16
10 12
7 8
5 6
21 24
15 17
11 13
8 9
29
22
16
12
42
33
26
20
61
51
42
34
(Figure 4 continued from
previous page)
Non-smoker
24 25 25 26
19 20 20 21
15 15 16 16
12 12 12 13
Denmark
Articles
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Systolic blood pressure (mm Hg)
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
3
2
2
1
2
1
1
1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
4
3
2
2
1
2
1
1
1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
<1%
9
7
5
4
8
6
5
4
180
160
140
120
180
160
140
120
23
18
14
11
23
18
14
11
180
160
140
120
5
6
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
2
2
1
1
4
3
2
2
10
8
6
4
2%
7
<1
<1
<1
<1
1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
3
2
1
1
5
3
2
2
11
8
6
5
36
29
23
18
2
1
1
1
4
3
2
1
7
5
4
3
2
2
1
1
5
3
2
2
7
5
4
3
4
3–4%
3
5
6
1
1
<1
<1
1
1
1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
5
4
3
2
8
6
4
3
5–9%
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1
<1 1 1
<1 <1 <1
<1 <1 <1
1 1 1
1 1 1
<1 <1 1
<1 <1 <1
1 1 2
1 1 1
1 1 1
<1 <1 <1
2
1
1
1
4
3
2
1
6
4
3
2
7
2
1
1
1
3
2
1
1
6
4
3
2
10
7
5
4
24
18
14
11
52
43
35
28
3
4
2
1
1
1
3
2
1
1
4
3
2
1
8
6
4
3
5
6
1 1
1 1
<1 <1
<1 <1
1 2
1 1
1 1
<1 <1
2
1
1
<1
2
1
1
1
3
2
2
1
7
5
3
2
≥15%
1
1
<1 <1
<1 <1
<1 <1
1
1
1
1
<1 <1
<1 <1
26 28
20 22
16 17
12 13
55 56
46 47
37 38
30 31
7
2
1
1
<1
2
2
1
1
3
2
1
1
3
2
1
1
4
3
2
1
9
6
5
3
39 41
31 32
24 25
19 20
43
34
27
21
45
36
28
22
72 73 74 75
62 63 64 66
52 53 54 56
43 44 45 46
Smoker
3
2
1
1
<1
3
2
1
1
3
2
1
1
4
3
2
1
5
4
3
2
11
8
6
4
4
2
1
1
1
3
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
13
9
7
5
5
3
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
7
5
4
2
7
5
3
2
1
7
5
3
2
7
5
3
2
8
6
4
3
40
45
50
55
60
65
70
75
80
Age
(years)
31
25
20
15
32
26
20
16
33
26
21
16
2
1
1
1
3
2
1
1
5
4
3
2
2
2
1
1
3
2
1
1
6
4
3
2
3
<1
<1
<1
<1
4
5
6
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
<1 <1 <1 <1
2
1
1
1
3
2
1
1
4
3
2
1
6
5
3
2
9 10 11
7 8 8
5 6 6
4 4 5
1 1 2
1 1 1
<1 1 1
<1 <1 <1
2
1
1
<1
2
2
1
1
4
3
2
2
9
6
5
3
15 16 17 18
11 12 13 14
8 9 10 10
6 7 7 8
30
24
19
15
Non-smoker
Total cholesterol (mmol/L)
6
4
3
2
1
6
4
2
2
6
4
3
2
7
5
3
2
8 10
6 7
4 5
3 3
14 16 18
10 12 13
8 9 10
5 6 7
18 19 21 23 25
13 14 16 17 19
10 11 12 13 14
7
8 9 9 10
37
29
23
17
70
60
51
42
Diabetic
11 12 13
8 9 10
6 7
7
4 5
5
25
19
15
11
54
44
36
29
Non-smoker
1
1
1
1
<1 1
<1 <1
1
1
1
<1
2
2
1
1
5
4
3
2
9
7
5
4
22
17
13
10
10–14%
1
1
1
<1
2
1
1
<1
2
1
1
1
2
2
1
1
3
2
1
1
6
4
3
2
9
6
5
3
17 18
13 14
10 11
8 8
51
42
34
27
Women
37 38 39
30 31 32
24 24 25
19 19 20
Smoker
15 15 16
11 12 13
8 9 10
6 7 7
35
28
22
18
Non-diabetic
25 26
20 20
15 16
12 12
1%
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
2
2
1
1
4
3
2
1
9
7
5
4
24
19
15
12
Non-smoker
England
7
1
1
<1
<1
1
1
1
<1
2
2
1
1
3
2
2
1
4
3
2
1
7
5
4
3
12
9
7
5
19
14
11
8
34
27
21
17
38
31
24
19
39
32
25
20
Smoker
3
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
4
3
2
1
1
1
3
2
2
1
5
3
2
1
6
4
3
2
9
7
5
3
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
<1
3
2
1
1
4
3
2
1
5
3
2
2
8
6
4
3
13 14 15
10 10 12
7
8 9
5
6 6
20 21 23
15 16 17
12 13 13
9
9 10
37
30
24
19
Non-diabetic
41
33
27
21
6
2
1
1
<1
3
2
1
1
4
3
2
1
6
4
3
2
6
5
3
2
11
8
5
4
7
3
2
1
1
3
2
1
1
5
4
2
2
7
5
3
2
8
5
4
3
12
9
6
4
17 19
13 14
9 10
7 8
24 26
19 20
14 15
11 11
40
32
26
20
49
40
32
26
50 51
41 42
33 34
27 27
Non-smoker
3
1
1
<1
<1
1
1
1
<1
2
2
1
1
4
2
2
1
5
3
2
2
2
1
1
1
4
3
2
1
5
4
2
2
7
5
3
2
4
5
1 2
1 1
<1 1
<1 <1
2
1
1
<1
3
2
1
1
4
3
2
1
6
4
3
2
9 10 11
6
7 8
5
5 6
3 4 4
6
2
1
1
1
3
2
1
1
5
3
2
1
6
4
3
2
8
5
4
3
13
9
7
5
16 18 19 21
12 13 15 16
9 10 11 12
7
7 8 9
26 28 29 31
20 21 23 24
15 16 17 18
12 12 13 14
48
39
31
25
Men
7
3
2
1
1
4
2
1
1
6
4
3
2
8
5
4
2
9
6
4
3
15
11
8
5
23
17
13
10
33
25
20
15
52
43
35
28
Smoker
3
2
1
1
1
3
2
1
1
5
4
2
2
7
5
3
2
4
3
2
1
1
4
3
2
1
7
4
3
2
5
4
2
2
1
5
3
2
1
8
5
4
2
9 10
6 7
4 5
3 3
9 10 12
6
7 8
4
5 6
3
3 4
15 16 18
11 12 13
8 9 10
5 6 7
24 26 28
18 20 22
14 15 16
10 11 12
35 37 39
27 29 30
21 22 24
16 17 18
33
26
19
14
43
34
27
21
6
5
3
2
1
6
4
3
2
7
7
4
3
2
8
5
3
2
10 13
7 8
5 6
3 4
12 15
9 10
6 7
4 5
14 16
10 11
7 8
5 6
21 23
15 17
11 13
8 9
31
24
18
13
41
32
25
19
56 58 59 60 62
47 48 49 51 52
38 39 41 42 43
31 32 33 34 35
Diabetic
Articles
(Figure 4 continued from
previous page)
9
Systolic blood pressure (mm Hg)
10
3
2
1
1
1
1
1
1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
4
2
2
1
1
1
1
1
<1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
<1%
5
4
3
2
5
4
3
2
180
160
140
120
15
12
9
7
15
11
9
7
5
2
1
1
1
3
2
2
1
6
5
3
3
6
1
<1
<1
<1
1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
1%
<1
<1
<1
<1
1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
2
1
1
1
3
2
2
1
6
4
3
2
15 16
12 12
9 10
7
7
2%
7
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
4
3
2
1
7
5
4
3
16
13
10
8
24
19
15
11
24
19
15
12
Smoker
2
1
1
1
3
2
2
1
5
4
3
2
2
1
1
1
2
2
1
1
3
2
2
1
6
4
3
2
3
3–4%
4
5
6
2
1
1
<1
2
1
1
1
2
1
1
1
2
2
1
1
3
2
1
1
4
3
2
1
6
5
3
3
11
8
6
5
7
2
1
1
1
3
2
1
1
5
3
2
2
8
6
4
3
3
4
5
2
1
1
<1
2
1
1
1
2
1
1
1
3
2
1
1
3
2
2
1
5
4
3
2
6
2
1
1
1
3
2
1
1
3
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
9 10
7 7
5 5
4 4
≥15%
1
1
1
1
<1 <1
<1 <1
1
2
1
1
1
1
<1 <1
1
2
1
1
1
1
<1 <1
2
1
1
1
2
2
1
1
4
3
2
1
7
5
4
3
7
3
2
1
1
3
2
1
1
3
2
1
1
4
3
2
1
5
3
2
2
7
5
3
2
11
8
6
4
16 17 18
12 13 14
9 10 11
7 7 8
53
44
36
29
54
45
36
29
Smoker
55
46
37
30
3
3
2
1
1
4
2
2
1
4
3
2
1
5
3
2
1
6
4
3
2
8
6
4
3
4
4
3
2
1
5
3
2
1
5
3
2
1
6
4
3
2
7
5
3
2
10
7
5
4
14 16
11 12
8
9
6
6
31
24
18
14
56
47
38
31
5
5
3
2
1
6
4
3
2
6
4
3
2
7
5
3
2
8
5
4
3
9
6
4
3
7
9
6
4
2
40
45
50
55
60
65
70
75
80
Age
(years)
2
1
1
1
2
2
1
1
4
3
2
1
6
5
3
3
2
1
1
1
3
2
1
1
4
3
2
2
7
5
4
3
2
1
1
1
3
2
1
1
3
2
2
1
5
4
3
2
8
6
4
3
3
4
5
6
<1 1 1 1
<1 <1 <1 1
<1 <1 <1 <1
<1 <1 <1 <1
1 1 1 1
<1 1 1 1
<1 <1 <1 1
<1 <1 <1 <1
1 1 2
1 1 1
<1 1 1
<1 <1 <1
1
1
1
<1
2
1
1
1
3
2
2
1
6
4
3
2
10 10 11 12
7 8 8 9
6 6 6 7
4 4 5 5
19 19 20 20
15 15 15 16
11 12 12 12
9 9 9 10
Non-smoker
Total cholesterol (mmol/L)
6
7
4
3
2
8 10
5 7
3 4
2 3
7
5
3
2
8 10
6 7
4 5
3 3
9 11
6 8
5 5
3 4
11 12 14
8 9 10
6 6 7
4 5 5
17 19 20
13 14 15
9 10 11
7 8 8
25 26 28 29
19 20 21 23
14 15 16 17
11 12 12 13
52
43
35
28
Diabetic
36 37 38 39
29 30 31 31
23 24 24 25
18 19 19 20
Non-smoker
14 15
11 11
8
9
6
7
35
28
23
18
10–14%
2
1
1
1
3
2
1
1
2
2
1
1
3
2
1
1
3
2
2
1
4
3
2
2
7
5
4
3
12
9
7
5
25 26
20 20
15 16
12 12
5–9%
1 1 1
<1 1 1
<1 <1 1
<1 <1 <1
1 1 2
1 1 1
<1 1 1
<1 <1 <1
1 1 2
1 1 1
<1 1 1
<1 <1 <1
1 2
1 1
1 1
<1 <1
2
1
1
1
3
2
1
1
5
4
3
2
9 10 10
7 7 8
5 6 6
4 4 4
23
18
14
11
Non-diabetic
Women
7
1
1
1
<1
2
1
1
<1
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
8
6
5
3
13
10
7
5
21
16
13
10
24
19
15
12
25
20
15
12
Smoker
2
1
1
1
3
2
1
1
3
2
2
1
4
3
2
1
6
5
3
2
2
2
1
1
3
2
2
1
4
3
2
1
5
4
3
2
7
5
4
3
10 11
7 8
5 6
4 4
3
4
5
1
1 2
1
1 1
<1
1 1
<1 <1 <1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
6
4
3
2
9
6
5
3
14 14 15
10 11 12
8
8 9
6
6 7
23
18
14
11
Non-diabetic
6
3
2
1
1
3
2
1
1
4
3
2
1
5
3
2
2
6
4
3
2
8
6
4
3
12
9
6
5
16
12
9
7
7
3
2
1
1
4
3
2
1
5
3
2
2
6
4
3
2
7
5
3
2
9
7
5
3
13
9
7
5
17
13
10
8
25 26
20 21
16 16
12 13
Men
3
1
1
<1
<1
2
1
1
<1
2
2
1
1
3
2
2
1
4
3
2
2
7
5
4
3
11
8
6
4
4
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
5
4
3
2
8
6
4
3
5
2
1
1
1
3
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
9
6
5
3
12 13
9 10
7 7
5 5
6
3
2
1
1
3
2
1
1
5
3
2
1
6
4
3
2
7
5
4
2
10
7
5
4
15
11
8
6
18 19 20 21
13 14 15 16
10 11 12 12
8 8 9 9
31 32 33 34
25 25 26 27
19 20 21 21
15 16 16 17
Non-smoker
7
4
2
1
1
4
3
2
1
6
4
3
2
7
5
3
2
8
6
4
3
11
8
6
4
16
12
9
7
22
17
13
10
34
28
22
17
Smoker
3
3
2
1
1
4
2
2
1
5
3
2
2
6
4
3
2
8
6
4
3
11
8
6
4
4
4
2
2
1
5
3
2
1
6
4
3
2
8
5
4
2
5
5
3
2
1
6
4
3
2
8
5
4
2
9
6
4
3
9 11
7 8
5 5
3 4
13 14
9 10
7 8
5 5
17 18 20
12 14 15
9 10 11
7
7 8
24 25 27
19 20 21
14 15 16
11 11 12
38 39 40
30 31 32
24 25 26
19 20 20
Diabetic
24
18
13
10
30
24
18
14
42
34
27
22
6
7
4
3
2
7
9
6
4
2
8 10
5 7
3 4
2 3
10 12
7 8
4 6
3 4
11 13
8 9
5 6
4 4
13 15
9 10
6 7
4 5
16 18
12 13
9 10
6 7
22
16
12
9
29
22
17
13
41
33
26
21
(Figure 4 continued from
previous page)
180
160
140
120
Non-smoker
Japan
Articles
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Systolic blood pressure (mm Hg)
8
6
4
3
3
2
1
1
1
1
1
<1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
4
7
5
4
3
4
3
2
1
2
2
1
1
180
160
140
120
180 1
160 <1
140 <1
120 <1
180 1
160 <1
140 <1
120 <1
3
180
160
140
120
180 1
160 1
140 1
120 <1
<1
<1
<1
<1
180
160
140
120
180
160
140
120
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
<1%
4
3
2
2
16
12
9
7
30
24
19
15
180 15
160 12
140 9
120 7
180
160
140
120
5
1%
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
2
1
1
1
3
2
2
1
5
3
2
2
6
1
<1
<1
<1
1
1
1
<1
1
1
1
<1
2
1
1
1
4
3
2
1
5
4
3
2
9 10
7
7
5
5
4 4
17 18
13 14
10 11
8
8
Non-smoker
31 31 32
24 25 26
19 20 20
15 15 16
Mexico
2%
7
1
1
<1
<1
2
1
1
<1
2
1
1
<1
2
2
1
1
4
3
2
1
6
4
3
2
11
8
6
4
19
15
11
8
33
27
21
16
27
21
16
13
2
1
1
1
2
1
1
1
4
2
2
1
6
4
3
2
4
3–4%
3
13
9
7
5
21
15
11
8
6
3
2
1
1
4
2
2
1
3
2
2
1
5
4
2
2
23
17
13
10
40
32
25
19
3
2
1
1
3
2
1
1
5
3
2
2
9
6
4
3
3
4
5
2
2
1
1
4
2
2
1
4
2
2
1
6
4
3
2
6
3
2
1
1
5
3
2
1
5
3
2
1
7
5
3
2
7
4
3
2
1
6
4
3
2
6
4
3
2
9
6
4
3
10 12 14
7 8 10
5 6
7
4 4 5
≥15%
1
2
1
1
1
1
<1 <1
2
1
1
1
2
2
1
1
4
3
2
1
8
5
4
3
27 30
21 23
16 17
12 13
44 46
35 37
28 29
22 23
68
58
48
39
84
75
66
56
85
76
67
57
26 29 33 36
20 22 25 27
14 16 18 20
10 12 13 15
41 44 47 51
32 35 37 40
24 26 29 31
18 20 22 24
61 63 66 68
50 53 55 57
41 43 45 47
32 34 36 38
Smoker
82 83
73 74
63 64
53 54
3
5
3
2
1
7
4
3
2
7
4
3
2
10
7
5
3
15 18 21
10 12 15
7 8 10
5 6 7
4
6
4
2
2
9
6
4
2
5
7
40
45
50
55
60
65
70
75
80
Age
(years)
25
19
15
11
27
21
16
12
28
22
17
13
2
1
1
1
3
2
1
1
5
4
3
2
3
2
1
1
4
3
2
1
6
4
3
2
3
4
5
6
1 1 1 1
<1 1 1 1
<1 <1 <1 1
<1 <1 <1 <1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
4
2
9 10 11
6 7 8
4 5 6
3 4 4
1 1 2
1 1 1
<1 1 1
<1 <1 <1
2
1
1
1
3
2
1
1
5
3
2
2
8
5
4
3
12 13 14 16
9 10 11 12
7 7 8 9
5 5 6 6
24
18
14
11
38
30
24
19
Non-smoker
39 40 41
31 32 33
25 26 26
20 20 21
Total cholesterol (mmol/L)
6
8 10 14
5 7 9
3 4 6
2 3 4
11 14 17
7 9 12
5 6 8
3 4 5
8 10 12 15
5
7 8 10
4
5 6 7
2
3 4 5
12
8
6
4
17 20 23 26 30
12 14 16 19 22
9 10 12 14 16
6
7 8 10 11
24
17
13
9
38
29
22
17
59
48
39
31
81
72
62
52
Diabetic
15 17 19
11 12 14
8 9 10
6 6
7
25
19
14
11
42
34
26
20
Non-smoker
64 65 66
54 55 56
45 46 47
36 37 38
12 13
8
9
6
7
4
5
21
16
12
9
38
30
24
18
63
53
44
36
10–14%
7
3
2
1
1
5
3
2
1
4
3
2
1
6
4
3
2
8 10
6
7
4
5
3
3
11
8
6
4
5–9%
5
2
1
1
1
3
2
1
1
3
2
1
1
4
3
2
1
7
5
4
2
9 10
6 7
5 5
3 4
1 1
1 1
<1 1
<1 <1
2
1
1
<1
2
1
1
1
3
2
1
1
5
4
3
2
8
6
4
3
19
14
10
8
29 31 32
23 24 25
17 18 20
13 14 15
Smoker
46 47 48 49
37 38 39 40
30 31 32 33
24 25 25 26
14 16 17
11 12 13
8 9 9
6 6 7
26
20
15
12
45
36
29
23
Non-diabetic
Women
7
2
1
1
1
3
2
1
1
4
3
2
1
5
4
3
2
8
6
4
3
12
9
6
5
17
13
10
7
30
23
18
14
42
34
27
21
34
26
20
16
36
28
22
17
Smoker
47 48
38 39
31 32
24 25
3
2
1
1
<1
2
1
1
1
4
3
2
1
5
3
2
2
8
6
4
3
4
2
1
1
1
3
2
1
1
5
3
2
1
6
4
3
2
5
3
2
1
1
4
2
2
1
6
4
3
2
7
5
3
2
10 11
7 8
5 6
3 4
12 14 16
9 10 11
6
7 8
5
5 6
18 19 21
13 15 16
10 11 12
7
8 9
32
25
19
15
46
37
30
24
Non-diabetic
6
4
2
2
1
5
3
2
1
7
5
3
2
9
6
4
3
13
9
6
5
7
5
3
2
1
6
4
3
2
9
6
4
3
10
7
5
3
15
11
8
5
18 20
13 15
9 11
7 8
23 25
17 19
13 14
10 11
38 40
30 31
23 24
18 19
49 50
40 41
32 33
26 27
Men
3
2
1
1
<1
2
2
1
1
4
3
2
1
6
4
3
2
10
7
5
3
15
11
8
6
22
17
13
9
40
32
25
19
57
48
39
32
27 29
20 22
15 16
11 12
4
2
1
1
1
3
2
1
1
5
3
2
2
5
3
2
1
1
4
3
2
1
6
4
3
2
8
6
4
3
11 13
8 9
6 7
4 5
7
5
3
2
62
52
43
35
6
4
3
2
1
5
3
2
1
8
5
4
2
10
7
5
3
15
11
8
5
7
5
3
2
1
6
4
3
2
10
7
4
3
12
8
6
4
18
13
9
6
24
18
13
9
31
24
18
13
Smoker
67 69
58 59
48 49
39 40
70
60
50
41
71
61
52
43
3
4
3
2
1
5
4
2
2
9
6
4
3
11
8
5
4
4
6
4
2
1
7
5
3
2
5
8
5
3
2
9
6
4
2
11 13
7 9
5 6
3 4
13 16
9 11
6 8
4 5
17 20 23
12 14 16
9 10 12
6
7 8
24 27 30
18 20 22
13 15 17
9 11 12
32 35 38
25 27 29
19 20 22
14 15 17
30
22
16
11
37
28
21
15
44
34
26
20
6
7
10 13
6 8
4 5
3 3
11 14
7 10
5 6
3 4
16 20
11 14
8 9
5 6
19 22
13 16
9 11
6 8
26
19
14
10
33
25
19
14
41
32
24
18
52 54 57 59 62
42 44 46 49 51
33 35 37 39 41
26 27 29 31 33
66
56
47
38
Diabetic
44 47 49
35 37 39
28 29 31
22 23 24
17 19 22
12 14 16
9 10 12
7 7 8
24
18
14
10
42
34
26
20
Non-smoker
58 59 61
49 50 51
40 41 42
32 33 34
Articles
(Figure 4 continued from
previous page)
11
Systolic blood pressure (mm Hg)
12
2
1
1
1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
4
2
1
1
1
180 1
160 1
140 <1
120 <1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
<1%
4
3
2
2
4
3
2
1
180
160
140
120
10
8
6
4
9
7
5
4
180
160
140
120
180 20
160 15
140 12
120 9
2
2
1
1
5
4
3
2
5
1%
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
6
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1 1
<1
1
<1 <1
2
2
1
1
5
3
2
2
11 11
8 8
6 6
5
5
2%
7
1
<1
<1
<1
1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
3
2
1
1
5
4
3
2
12
9
7
5
22
17
14
11
Smoker
32 33
25 26
20 21
16 16
2
1
1
1
4
3
2
2
8
6
4
3
2
2
1
1
5
3
2
2
9
7
5
4
4
3–4%
3
6
1
1
1
<1
2
1
1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
5
4
3
2
10
7
5
4
2
1
1
1
3
2
1
1
6
4
3
2
3
2
1
1
1
2
1
1
1
3
2
1
1
4
3
2
1
8
6
4
3
5
6
1 2
1 1
<1 1
<1 <1
1
1
1
<1
2
1
1
<1
2
1
1
1
3
2
2
1
7
5
4
3
≥15%
4
1
1
<1 1
<1 <1
<1 <1
1
1
1
1
<1 <1
<1 <1
1
1
1
1
<1 1
<1 <1
1
1
1
<1
2
2
1
1
5
4
3
2
7
2
1
1
1
2
2
1
1
3
2
1
1
3
2
1
1
5
3
2
2
9
6
5
3
13 15 16
10 11 12
7 8 9
5 6
7
26 28 30 31
20 22 23 24
16 17 18 19
12 13 13 14
68
58
49
40
70
60
50
41
43 45 47 50
34 36 38 40
27 28 30 32
21 22 23 25
Smoker
66 67
56 57
46 47
38 39
3
2
1
1
1
3
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
4
3
2
1
1
4
2
1
1
4
3
2
1
5
3
2
1
7
5
3
2
5
4
3
2
1
5
3
2
1
5
3
2
1
6
4
3
2
8
6
4
3
7
7
5
3
2
8
5
3
2
7
5
3
2
8
6
4
3
40
45
50
55
60
65
70
75
80
Age
(years)
2
2
1
1
4
3
2
2
8
6
4
3
2
1
1
1
3
2
1
1
5
4
3
2
2
1
1
1
3
2
1
1
6
4
3
2
9 10
7 7
5 5
4 4
3
<1
<1
<1
<1
4
5
6
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
<1 <1 <1 <1
1 1 1 1
<1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
1 1
1 1
1 1
<1 <1
2
1
1
1
4
3
2
1
7
5
4
3
16 17 18 19
12 13 14 15
9 10 11 11
7 7 8 8
33
26
21
16
Non-smoker
34 35 36
27 28 28
21 22 23
17 17 18
Total cholesterol (mmol/L)
6
5
3
2
1
6
4
2
2
6
4
3
2
7
5
3
2
9 11
6 8
5 5
3 4
11 13 14 16 18
8
9 11 12 13
6
7 8 9 10
4
5
5 6 7
21 23 25 27 30
16 17 19 21 23
12 13 14 15 17
9 10 11 12 13
41
32
25
20
64
54
45
37
Diabetic
Non-smoker
47 48 49 51
38 39 40 42
31 32 33 34
25 25 26 27
11 12
8
9
6
7
4
5
25
19
15
11
46
37
30
24
10–14%
7
2
1
1
<1
2
1
1
1
2
1
1
1
2
2
1
1
3
2
2
1
6
4
3
2
11
8
6
4
20 21
15 16
11 12
9 9
34 35
27 28
21 22
17 17
5–9%
5
1 1 1
<1 <1 1
<1 <1 <1
<1 <1 <1
1 1 1
<1 1 1
<1 <1 <1
<1 <1 <1
1 1 1
1 1 1
<1 <1 1
<1 <1 <1
1 1 2
1 1 1
<1 1 1
<1 <1 <1
2
1
1
1
4
3
2
1
7
5
4
3
16 17 19
13 13 14
10 10 11
7 8 8
31
25
19
15
Non-diabetic
Women
7
1
1
<1
<1
1
1
1
<1
2
1
1
1
2
2
1
1
4
2
2
1
6
5
3
2
11
8
6
4
21
16
12
9
37
29
23
18
3
1
<1
<1
<1
1
1
1
<1
2
1
1
<1
2
2
1
1
3
2
2
1
7
5
3
2
11
8
6
4
22
17
13
10
40
32
26
20
Non-diabetic
25
19
15
11
2
1
1
1
2
2
1
1
3
2
2
1
5
3
2
2
8
6
4
3
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
4
3
12 13
9 10
7 7
5 5
23
18
14
10
Smoker
41 42
33 34
26 27
21 22
6
2
1
1
<1
3
2
1
1
3
2
1
1
4
3
2
1
5
4
3
2
9
7
5
3
7
2
1
1
1
3
2
1
1
4
3
2
1
5
3
2
2
6
4
3
2
11
8
6
4
15 16
11 12
8 9
6
7
26 28
20 21
15 16
12 13
43 45
35 36
28 29
22 23
Men
3
1
<1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
8
6
4
3
14
10
8
6
28
22
17
13
51
42
34
27
32 33
25 26
19 20
15 15
2
1
1
1
3
2
1
1
4
3
2
1
6
4
3
2
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
<1
2
1
1
1
3
2
1
1
5
3
2
2
9 10
7 7
5 5
3 4
6
2
1
1
<1
3
2
1
1
3
2
2
1
5
3
2
1
6
5
3
2
12
8
6
4
15 17 19
12 13 14
9 9 10
6 7 8
30
23
18
14
7
2
2
1
1
4
2
2
1
4
3
2
1
6
4
3
2
8
5
4
3
13
9
7
5
20
15
11
8
35
28
21
16
64
54
45
36
66
56
46
38
3
2
1
1
1
3
2
1
1
4
3
2
1
5
4
2
2
7
5
4
3
13
10
7
5
21
16
12
9
25
19
14
10
27
21
15
11
30
22
17
13
4
3
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
5
4
2
1
1
5
3
2
1
6
4
3
2
8
5
4
2
9 10
6 7
4 5
3 3
6
5
3
2
1
7
4
3
2
7
5
3
2
7
6
4
3
2
8
6
4
2
9
6
4
3
9 11
6 8
4 5
3 4
12 13
8 10
6 7
4 5
15 17 19 21
11 12 14 15
8 9 10 11
6 6
7 8
23
17
13
9
37 39 42 44 46
29 31 33 35 37
23 24 26 27 29
18 19 20 21 22
Smoker
60 61 63
50 52 53
41 43 44
33 34 35
Diabetic
Non-smoker
52 53 55 56
43 44 45 46
35 36 37 38
28 29 30 30
(Figure 4 continued from
previous page)
Non-smoker
20 21 22
16 16 17
12 13 13
10 10 10
South Korea
Articles
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Systolic blood pressure (mm Hg)
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
3
2
2
1
1
1
1
1
1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
4
3
2
1
1
180 1
160 1
140 1
120 <1
180 1
160 <1
140 <1
120 <1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
3
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
180
160
140
120
<1%
6
4
3
2
5
4
3
2
180
160
140
120
180 14
160 11
140 9
120 7
5
1%
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
2
1
1
1
3
2
2
1
6
5
3
3
6
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
2
1
1
1
3
3
2
1
6
5
4
3
Non-smoker
15 15 16
11 12 12
9
9 9
7
7
7
Spain
2%
7
<1
<1
<1
<1
<1
<1
<1
<1
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
4
3
2
2
7
5
4
3
16
12
10
8
Smoker
23 24
18 19
14 15
11 12
2
1
1
1
3
2
2
1
6
4
3
2
2
1
1
1
3
2
2
1
6
5
3
2
4
6
1
1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
2
2
1
1
4
3
2
1
7
5
4
3
12
9
7
5
2
2
1
1
5
3
2
2
9
6
5
3
3
2
2
1
1
3
2
2
1
6
4
3
2
5
6
1 1
<1 1
<1 <1
<1 <1
1 1
1 1
<1 1
<1 <1
1 2
1 1
1 1
<1 <1
2
1
1
1
3
2
1
1
5
4
3
2
9 10
7 8
5 6
4 4
≥15%
4
<1 1
<1 <1
<1 <1
<1 <1
1
1
<1 <1
<1 <1
<1 <1
1
1
1
1
<1 <1
<1 <1
1
2
1
1
1
1
<1 <1
2
1
1
1
4
3
2
1
8
6
4
3
7
1
1
1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
3
2
11
8
6
5
17 18 19
13 14 15
10 10 11
7 8 8
3
1
1
1
<1
2
1
1
<1
2
2
1
1
3
2
2
1
5
3
2
2
9
6
4
3
15
11
8
6
26
20
15
12
51
42
34
27
Diabetic
Non-smoker
36 37 37 38
29 29 30 31
23 23 24 25
18 18 19 20
15 16
11 12
9
9
7
7
35
28
22
17
10–14%
7
1
1
<1
<1
1
1
1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
2
8
6
4
3
12
9
7
5
25 25
19 20
15 16
12 12
5–9%
5
<1 1
<1 <1
<1 <1
<1 <1
3–4%
3
<1
<1
<1
<1
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1
<1 1 1
<1 <1 <1
<1 <1 <1
1 1 1
1 1 1
<1 1 1
<1 <1 <1
1
1
1
<1
3
2
1
1
5
4
3
2
10 10 11
7 8 8
5 6 6
4 4 5
23
18
14
11
Non-diabetic
Women
54
45
37
30
56
46
38
30
4
2
1
1
<1
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
10
7
5
4
5
2
2
1
1
3
2
1
1
4
2
2
1
5
3
2
1
7
5
3
2
7
4
3
2
1
5
3
2
1
6
4
2
2
7
5
3
2
9
6
4
3
40
45
50
55
60
65
70
75
80
2
1
1
1
2
2
1
1
4
3
2
1
7
5
4
3
2
1
1
1
3
2
1
1
4
3
2
2
7
5
4
3
2
2
1
1
3
2
1
1
5
3
2
2
8
6
4
3
3
<1
<1
<1
<1
4
5
6
<1 1 1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
<1 <1 <1 <1
1 1 1 2
1 1 1 1
<1 <1 1 1
<1 <1 <1 <1
1
1
1
<1
2
1
1
1
3
2
2
1
6
4
3
2
11 12 13 14
9 9 10 10
6 7 7 8
5 5 5 6
22
17
13
10
Non-smoker
23 23 24
18 18 19
14 14 15
11 11 12
Total cholesterol (mmol/L)
6
3
2
1
1
4
3
2
1
4
3
2
1
6
4
3
2
8
5
4
3
11 12 14
8 9 10
6 6 7
4 5 5
17 18 20 21
12 13 15 16
9 10 11 12
7
7 8 9
27 29 30 32
21 22 24 25
16 17 18 19
12 13 14 15
Smoker
52 53
43 44
35 36
28 29
Age
(years)
7
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
3
2
2
1
5
4
3
2
9
6
5
3
14
11
8
6
25
20
15
12
Smoker
28 29
22 23
18 18
14 14
3
1
<1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
3
2
2
1
5
4
3
2
9
7
5
4
2
1
1
1
3
2
1
1
4
3
2
1
5
3
2
2
7
5
4
3
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
<1
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
10 11
7 8
5 6
4 4
15 16 17
12 13 13
9
9 10
7
7 8
27
22
17
13
Non-diabetic
6
2
1
1
<1
3
2
1
1
4
2
2
1
5
3
2
1
5
4
3
2
8
6
4
3
12
9
7
5
7
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
6
4
3
2
9
7
5
3
13
10
7
5
19 20
14 15
11 11
8 9
30 31
24 25
19 19
15 15
Men
3
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
7
5
3
2
12
9
6
5
2
1
1
1
3
2
1
1
4
3
2
1
5
4
3
2
9
6
4
3
4
5
1 1
1 1
<1 1
<1 <1
2
1
1
<1
3
2
1
1
4
2
2
1
5
3
2
2
8
5
4
3
13 14
9 10
7 8
5 6
6
2
1
1
<1
3
2
1
1
4
3
2
1
5
4
2
2
6
4
3
2
10
7
5
4
15
11
8
6
20 21 23 24
15 16 17 18
12 13 13 14
9 9 10 11
36
29
23
18
7
2
2
1
1
4
2
1
1
5
3
2
1
6
4
3
2
7
5
4
3
11
8
6
4
17
12
9
7
25
20
15
11
3
2
1
1
1
3
2
1
1
4
3
2
1
6
4
3
2
7
5
4
2
11
8
6
4
4
3
2
1
1
4
3
2
1
5
4
2
2
7
5
3
2
5
4
2
1
1
5
3
2
1
7
5
3
2
9
6
4
3
8 10
6 7
4 5
3 3
13 14
9 10
7 7
5 5
17 19 21
13 14 15
10 10 11
7 8 8
27 29 30
21 22 24
16 17 18
12 13 14
25
18
14
10
34
27
21
16
6
5
3
2
1
6
4
3
2
7
6
4
3
2
8
5
4
2
8 10
6 7
4 5
2 3
10 12
7 9
5 6
3 4
11 13
8 9
6 7
4 5
16 18
12 13
8 10
6 7
22
17
13
9
32
25
19
15
Smoker
44 45 46 47 49
36 37 38 39 40
28 29 30 31 32
23 23 24 25 25
Diabetic
Non-smoker
37 38 39 40
30 30 31 32
24 24 25 26
19 19 20 20
Articles
(Figure 4 continued from
previous page)
13
Systolic blood pressure (mm Hg)
14
Systolic blood pressure (mm Hg)
8
6
5
3
5
4
3
2
3
2
1
1
2
1
1
1
1
1
1
<1
1
1
<1
<1
1
<1
<1
<1
<1
<1
<1
<1
4
8
6
4
3
5
3
3
2
3
2
1
1
2
1
1
1
3
180
160
140
120
180
160
140
120
180 1
160 <1
140 <1
120 <1
<1
<1
<1
<1
180
160
140
120
180 1
160 1
140 <1
120 <1
<1
<1
<1
<1
180
160
140
120
180
160
140
120
180
160
140
120
5
1%
1
<1
<1
<1
1
1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
3
2
2
1
6
4
3
2
9
7
5
4
6
1
<1
<1
<1
1
1
<1
<1
1
1
1
<1
2
1
1
1
2
2
1
1
4
3
2
1
6
5
3
2
9
7
5
4
19 20
15 16
12 12
9 9
2%
7
1
1
<1
<1
1
1
1
<1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
2
7
5
4
3
10
7
6
4
21
16
13
10
30
24
19
14
2
1
1
1
2
2
1
1
3
2
1
1
4
3
2
1
6
5
3
2
3
2
1
1
3
2
1
1
4
3
2
1
5
3
2
2
7
5
4
3
4
3–4%
3
6
3
2
1
1
3
2
1
1
4
2
2
1
5
3
2
1
6
4
3
2
8
6
4
3
22
17
13
10
2
2
1
1
3
2
1
1
4
3
2
1
6
4
3
2
9
7
5
3
3
5
2
2
1
1
3
2
1
1
4
2
2
1
5
4
2
2
7
5
3
2
6
3
2
1
1
4
3
2
1
5
3
2
1
6
4
3
2
7
4
3
2
1
5
3
2
1
6
4
3
2
7
5
3
2
8 10
6
7
4
5
3
3
10 12 13
8 9 10
5 6
7
4 4
5
≥15%
4
1
2
1
1
1
1
<1 <1
2
1
1
1
2
2
1
1
4
2
2
1
5
4
3
2
8
6
4
3
25 26
19 20
15 16
11 12
17 18 20
13 14 15
9 10 11
7 8 8
23
18
14
10
3
4
3
2
1
6
4
3
2
7
4
3
2
9
6
4
3
12
8
6
4
17
12
9
6
26
20
15
11
13 15 18
9 11 13
6 7 9
4 5 6
4
6
4
2
2
8
5
3
2
5
7
40
45
50
55
60
65
70
75
80
Age
(years)
26
21
16
13
27
22
17
13
29
22
18
14
2
1
1
1
3
2
1
1
5
4
2
2
7
5
4
3
2
2
1
1
3
2
1
1
4
3
2
1
7
5
3
2
9
6
5
3
3
4
5
6
1 1 1 1
<1 1 1 1
<1 <1 <1 1
<1 <1 <1 <1
2
1
1
1
3
2
1
1
3
2
2
1
6
4
3
2
8
6
4
3
9 10 11
7 7 8
5 5 6
4 4 4
1 1
1 1
<1 1
<1 <1
2
1
1
1
2
2
1
1
4
3
2
1
6
4
3
2
8
6
4
3
14 15 16 17
10 11 12 13
8 8 9 10
6 6 7 7
26
20
16
12
Non-smoker
Total cholesterol (mmol/L)
6
8 10 13
5 7 9
3 4 6
2 3 4
10 12 15
6 8 10
4 5 7
3 3 4
8 10 13 16
6
7 9 11
4
5 6 7
2
3 4 5
11
7
5
3
14 16 18 21
10 11 13 15
7 8 9 11
5 6 7 8
19 21 24 27
14 16 18 20
10 11 13 15
7 8 9 11
29 31 34 37
22 24 26 28
16 18 20 21
12 13 15 16
43
34
27
21
64 65 67 69
54 55 57 58
44 46 47 49
36 37 38 40
Smoker
35 37 39 41
27 29 30 32
21 22 24 25
16 17 18 19
62
52
43
35
Diabetic
44 46 47 49
36 37 38 40
29 30 31 32
23 24 24 25
Non-smoker
14 15
10 11
8
8
6
6
21
16
12
9
43
35
28
22
10–14%
7
3
2
1
1
4
3
2
1
4
3
2
1
5
4
3
2
7
5
3
2
9
7
5
3
12 14
9 10
7
7
5
5
5–9%
5
1 1 2
1 1 1
<1 1 1
<1 <1 <1
2
1
1
<1
2
1
1
1
3
2
1
1
4
3
2
1
6
4
3
2
9 10 11
7 8 8
5 6 6
4 4 5
16 17
12 13
9 10
7 8
31 32 33
25 26 27
19 20 21
15 16 16
Smoker
14 14 15
10 11 12
8 8 9
6 6 7
29
23
18
14
Non-diabetic
Women
7
2
1
1
<1
3
2
1
1
4
3
2
1
5
3
2
2
8
6
4
3
10
7
5
4
12
9
7
5
18
13
10
8
30
23
18
14
34
27
21
16
35
28
22
17
Smoker
3
2
1
1
<1
3
2
1
1
4
2
2
1
5
3
2
1
8
5
4
3
4
2
1
1
1
3
2
1
1
5
3
2
1
6
4
3
2
5
3
2
1
1
4
3
2
1
6
4
3
2
7
5
3
2
9 11
6 7
4 5
3 4
10 11 13
7
8 9
5
6 7
4
4 5
13 14 15
9 10 11
7
7 8
5
6 6
19 20 21
14 15 16
11 12 12
8
9 9
32
26
20
16
Non-diabetic
6
4
2
1
1
6
4
2
2
7
5
3
2
8
6
4
3
12
9
6
4
7
5
3
2
1
7
5
3
2
9
6
4
3
10
7
5
3
14
10
7
5
15 16
11 12
8 9
5 6
17 18
12 14
9 10
7
7
23 24
17 18
13 14
10 11
36 37
29 30
23 24
18 18
Non-smoker
20
15
11
8
17
12
9
6
11
8
5
4
10
6
4
3
8
5
3
2
5
3
2
1
12 14 16 18
9 10 11 13
6
7 8 9
5
5 6
7
15
10
7
5
9
6
4
3
8
5
3
2
6
4
3
2
4
2
2
1
3
2
1
1
<1
3
2
1
1
4
3
2
1
5
4
2
2
4
2
1
1
1
4
2
2
1
5
3
2
2
6
4
3
2
5
3
2
1
1
5
3
2
1
6
4
3
2
8
5
4
2
Men
6
7
23
17
13
9
16 17 19 21
12 13 14 16
9 10 10 12
6
7 8 9
11 13
8 9
5 6
4 4
31
24
18
14
24 26 27 29
19 20 21 23
14 15 16 17
11 12 12 13
9
7
5
3
47
38
31
24
Smoker
36
29
22
17
3
4
3
2
1
7
4
3
2
9
6
4
3
10
7
5
3
4
6
4
2
1
5
7
5
3
2
9 11
6 7
4 5
2 3
11 13
7 9
5 6
3 4
12 15
9 10
6 7
4 5
16 19 22
12 14 16
8 10 11
6
7 8
20 22 25
15 16 19
11 12 14
8 9 10
23 26 28
18 19 21
13 14 16
10 11 12
33 35
26 27
20 21
15 16
29
21
15
11
31
23
17
13
33
25
19
14
41
32
25
19
6
7
10 12
6 8
4 5
2 3
14 18
9 12
6 8
4 5
16 20
11 13
7 9
5 6
18 21
12 15
9 10
6 7
25
18
13
9
28
21
15
11
30
23
17
13
39
30
24
18
51 53 54 56 57
42 43 45 46 48
34 35 36 37 39
27 28 29 30 31
Diabetic
41 43 44 46
33 35 36 37
27 28 29 30
21 22 23 23
Men
Women
<1%
19
15
11
9
180 18
160 14
140 11
120 8
Non-smoker
Figure 4: Country risk charts
for 10-year risk of fatal
cardiovascular disease
To establish a person’s risk,
first identify the column that
represents the person’s sex,
smoking, and diabetes status.
Then identify the cell
representing the person’s age,
total cholesterol and systolic
blood pressure levels. Risk
charts are not shown for Czech
Republic, Iran, and Malawi
because their health
examination surveys did not
include participants aged
65 years or older.
USA
Articles
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Articles
Examination Survey in the USA, 9% of men and 7% of
women had a history of coronary heart disease or stroke.
Fourth, although WHO uses demographic and
Men
Women
South Korea
Spain
Spain
South Korea
Denmark
Japan
England
Denmark
Japan
England
USA
USA
Mexico
Mexico
Malawi
Czech Republic
China
Malawi
Iran
Spain
Japan
Japan
Spain
South Korea
Denmark
Denmark
England
England
South Korea
USA
USA
Mexico
Mexico
China
China
South Korea
South Korea
Spain
Spain
Denmark
Denmark
England
England
USA
Japan
Mexico
USA
Japan
Mexico
China
China
0
20
40
60
80
100
Population in risk category (%)
<3%
3–6%
7–9%
10–14%
40–84 years
Iran
China
65–84 years
Czech Republic
40–64 years
risk factor exposure, the rates of major CVD events were
higher in low-income and middle-income countries than
in high-income countries.39
Methodologically, our risk score is consistent with the
derivations and recalibrations of the Framingham risk
score, SCORE, and the American College of Cardiology/
American Heart Association Pooled Cohort Risk
Equations,10,15–22 including the use of fatal cardiovascular
disease in SCORE to allow recalibration in different
countries (panel).40–42 Nonetheless, we also developed a
risk score for fatal-plus-non-fatal cardiovascular disease,
for use in countries with data for cardiovascular disease
incidence. Our reformulation of the risk model to include
age-and-sex-specific death rates is epidemiologically well
established43 and allows the variation of age and sex
patterns of cardiovascular disease risk across countries
and over time to be taken into account. Our risk score
also overcomes the methodological limitation of the
present WHO risk charts,3 the coefficients of which are
based on separate analyses of individual risk factors
rather than coefficients estimated in a regression model
that includes all risk factors together.44 Other strengths of
our study are the use of multiple high-quality prospective
cohorts; good performance in external validation; and
application to individual-level, nationally representative
data from countries in different world regions to estimate
the prevalence of high-risk individuals, instead of
summary statistics used in a previous global analysis.3
The use of individual-level data accounts for the fact that
the associations between risk factors vary across countries
and time.45
Our study also has some limitations. First, although use
of multiple cohorts is an improvement compared with a
single cohort, as used in some other risk prediction
equations, our cohorts were all from US territories.
However, these cohorts included groups from diverse
ethnic origins. Furthermore, abundant evidence from
multicountry studies11–13 suggests that the proportional
effects of cardiovascular risk factors are similar across
Western and Asian populations. Ideally, we would like to
replicate our analysis with more cohorts, including
cohorts from different world regions to reconfirm this
similarity. Second, although we developed risk scores for
both fatal cardiovascular disease and fatal-plus-non-fatal
cardiovascular disease, our country applications could
only estimate the risk of fatal cardiovascular disease
because cardiovascular disease incidence rates are not
available for most countries. This unavailability of data
underscores the need for monitoring cardiovascular
disease incidence, for example through data linkage or
sentinel sites, as has been done for cancer registries.
Third, in addition to people with hazardous risk factor
profiles, cardiovascular disease risk is also high in people
who have had a previous cardiovascular event. Information
is needed about individual patients with such disease
histories and their prevalence in the population. For
example, in the 2011–12 National Health and Nutrition
0
20
40
60
80
100
Population in risk category (%)
≥15%
Figure 5: Distributions of 10-year risk of fatal cardiovascular disease by country, sex, and age group
Results for Czech Republic, Iran, and Malawi are shown only for people aged 40–64 years because older individuals
were not enrolled in the national health examination surveys in these countries. Raw data are shown in the appendix.
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
15
Articles
epidemiological methods to make valid and comparable
estimates of death rates by underlying cause,32 potential
exists for inconsistent or incomparable assignment of
underlying cause of death at the time of certification.
Furthermore, in countries without vital registration,
cardiovascular disease death rates must be estimated from
partial information with demographic and epidemiological
methods. Finally, lifetime risk can be an alternative
relevant measure of risk for young individuals who might
be assigned a low 10-year risk of cardiovascular disease
despite having increased risk factors.46
Randomised trials have established the benefits of risk
factor treatment in people with high absolute disease risk,
which has lead multidrug therapy to become a cornerstone
of cardiovascular disease prevention that is included in
clinical guidelines in many countries. With many
antihypertensive drugs and some statins coming off-patent
Panel: Research in context
Systematic review
We assessed risk scores for cardiovascular diseases reported in a 2010 review article40 and in
a systematic review of 102 models from 84 studies published between Jan 1, 1999, and
Feb 24, 2009.41 We then searched PubMed for articles published between Feb 25, 2009,
and Sept 10, 2014, using combinations of the terms “risk score” or “risk scores” or “risk
prediction” and “cardiovascular disease”, “coronary heart disease” or “stroke” in the
publication title. We identified studies published in English in which a risk score was
developed for a healthy adult population free from cardiovascular disease at baseline; the
study outcome was fatal or non-fatal coronary heart disease or stroke; and a risk score
model was developed and validated in the population. We also searched the reference lists
of the identified studies. We reviewed the identified articles and assessed the
generalisability of the risk scores between populations. Many of the reviewed articles
aimed to improve prediction by adding new risk factors to the model; recalibrate a risk
score for validation in a new population; or develop a new risk score for a specific
population. A few risk scores were devised for use across different populations. The SCORE
model9 predicts risk of death from cardiovascular diseases and provides separate risk scores
for the high-risk and low-risk countries of Europe. The American College of Cardiology/
American Heart Association Pooled Cohort Risk Equations10 provide ethnic origin-and-sexspecific risk scores for the USA. WHO has developed risk charts for each specific WHO
subregion, although cardiovascular risk patterns might differ between countries within the
same subregion and the coefficients for the risk factors in the model were not derived from
the same regression model or even the same set of epidemiological studies.3 Finally, the
INTERHEART Modifiable Risk score was developed from a multicountry case-control study
of myocardial infarction, unlike other models that are based on prospective cohorts.42
Interpretation
Our novel risk prediction equation for cardiovascular disease fills the need for a unified
risk score that can be used in different countries. The risk score can be recalibrated and
updated for use in different populations and years with routinely available information,
and allows for the effects of sex and age on cardiovascular risk to vary between
countries. When the risk score was recalibrated and applied to several example
countries, low-income and middle-income countries had increased prevalence of people
with high 10-year risk of fatal cardiovascular disease compared with high-income
countries. Our prediction equation helps with global implementation of risk-based
treatment of cardiovascular risk factors, and can be used to measure progress towards
the global non-communicable disease treatment target by estimating the number of
people in each country who are at high risk.
16
and available at reduced cost, these treatments are deemed
essential medicines for non-communicable disease
prevention worldwide.47 However, access to these
treatments remains low in low-income and middle-income
countries.48 The obstacles to large-scale risk-based
prevention are both technical (ie, absence of guidelines for
risk estimation and treatment) and related to health
systems (including financial, infrastructure, and health
personnel barriers to primary care access). Our risk
prediction equation helps to overcome the technical
barriers for global application of risk stratification and will
provoke debate about the needs of health systems by
allowing calculation of the coverage of risk-based treatment
in different countries.
Contributions
GD, ME, and, MW conceived the study. AA, CAA-S, FA, RC, MDC, LE,
FF, NI, DK, Y-HK, VL, LL-M, DM, KPM, KO, FR-A, RR-M, JES, GAS, JT,
BZ collected and managed risk factor survey or external cohort data. KH,
PU, and YL analysed cohort and survey data and prepared results. KH,
PU, ME, and GD wrote the manuscript with input from all other
co-authors. GD and ME oversaw the research. GD is the study guarantor.
Declaration of interests
MW reports fees from Amgen and Sanofi for being a research project
consultant, and fees from Novartis for being an advisory board member.
All other authors report no competing interests.
Acknowledgments
This study was funded by the US National Institutes of Health (NIH;
NIDDK: 1R01-DK090435), UK Medical Research Council, and Wellcome
Trust. Data for prospective cohorts were obtained from the National
Heart Lung and Blood Institute (NHLBI) Biologic Specimen and Data
Repository Information Coordinating Center. GAS and KPM are staff
members of WHO. This study does not necessarily reflect the opinions
or views of the cohorts used in the analysis, WHO, or the NHLBI. This
research also uses data from China Health and Nutrition Survey (CHNS).
We thank the National Institute of Nutrition and Food Safety, China
Center for Disease Control and Prevention, Carolina Population Center
(5 R24 HD050924), the University of North Carolina at Chapel Hill, the
NIH (R01-HD30880, DK056350, R24 HD050924, and R01-HD38700), and
the Fogarty International Center of the NIH for financial support for the
CHNS data collection and analysis files from 1989 to 2011. We thank the
China-Japan Friendship Hospital, Ministry of Health of China for
support for CHNS 2009. Access to individual records of the National
Health and Nutrition Survey was obtained under the Grant-in-Aid for
Scientific Research from the Japan Society for the Promotion of Science
(grant number: 24590785). We thank Pablo Perel and Rod Jackson for
their comments on an earlier version of the manuscript.
References
1 Jackson R, Lawes CM, Bennett DA, Milne RJ, Rodgers A. Treatment
with drugs to lower blood pressure and blood cholesterol based on an
individual’s absolute cardiovascular risk. Lancet 2005; 365: 434–41.
2 JBS3 Board. Joint British Societies’ consensus recommendations for the
prevention of cardiovascular disease. Heart 2014; 100 (suppl 2): ii1–67.
3 WHO. Prevention of cardiovascular disease: guidelines for
assessment and management of cardiovascular risk. Geneva:
World Health Organization, 2007.
4 Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA
guideline on the treatment of blood cholesterol to reduce
atherosclerotic cardiovascular risk in adults: a report of the American
College of Cardiology/American Heart Association Task Force on
Practice Guidelines. J Am Coll Cardiol 2013; 63: 2889–934.
5 NICE. Lipid modification: cardiovascular risk assessment and the
modification of blood lipids for the primary and secondary
prevention of cardiovascular disease. London: National Institute for
Health and Care Excellence, 2014.
6 Lim SS, Gaziano TA, Gakidou E, et al. Prevention of cardiovascular
disease in high-risk individuals in low-income and middle-income
countries: health effects and costs. Lancet 2007; 370: 2054–62.
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
Articles
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
WHO. Global action plan for the prevention and control of
noncommunicable diseases 2013–2020. Geneva: World Health
Organization, 2013.
D’Agostino RB Sr, Grundy S, Sullivan LM, Wilson P, for the CHD
Risk Prediction Group. Validation of the Framingham coronary
heart disease prediction scores: results of a multiple ethnic groups
investigation. JAMA 2001; 286: 180–87.
Conroy RM, Pyorala K, Fitzgerald AP, et al. Estimation of ten-year
risk of fatal cardiovascular disease in Europe: the SCORE project.
Eur Heart J 2003; 24: 987–1003.
Goff DC Jr, Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA
guideline on the assessment of cardiovascular risk: a report of
the American College of Cardiology/American Heart Association
Task Force on Practice Guidelines. Circulation 2014;
129 (25 suppl 2): S49–73.
Singh GM, Danaei G, Farzadfar F, et al. The age-specific
quantitative effects of metabolic risk factors on cardiovascular
diseases and diabetes: a pooled analysis. PLoS One 2013; 8: e65174.
Woodward M, Huxley H, Lam TH, Barzi F, Lawes CM, Ueshima H.
A comparison of the associations between risk factors and
cardiovascular disease in Asia and Australasia.
Eur J Cardiovasc Prev Rehabil 2005; 12: 484–91.
Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiable
risk factors associated with myocardial infarction in 52 countries
(the INTERHEART study): case-control study. Lancet 2004;
364: 937–52.
Cook NR, Paynter NP, Eaton CB, et al. Comparison of the
Framingham and Reynolds Risk scores for global cardiovascular
risk prediction in the multiethnic Women’s Health Initiative.
Circulation 2012; 125: 1748–56.
Neuhauser HK, Ellert U, Kurth BM. A comparison of Framingham
and SCORE-based cardiovascular risk estimates in participants of
the German National Health Interview and Examination Survey
1998. Eur J Cardiovasc Prev Rehabil 2005; 12: 442–50.
Liu J, Hong Y, D’Agostino RB Sr, et al. Predictive value for the
Chinese population of the Framingham CHD risk assessment tool
compared with the Chinese Multi-Provincial Cohort Study.
JAMA 2004; 291: 2591–99.
Khalili D, Hadaegh F, Soori H, Steyerberg EW, Bozorgmanesh M,
Azizi F. Clinical usefulness of the Framingham cardiovascular risk
profile beyond its statistical performance: the Tehran Lipid and
Glucose Study. Am J Epidemiol 2012; 176: 177–86.
Wu Y, Liu X, Li X, et al. Estimation of 10-year risk of fatal and
nonfatal ischemic cardiovascular diseases in Chinese adults.
Circulation 2006; 114: 2217–25.
Suka M, Sugimori H, Yoshida K. Application of the updated
Framingham risk score to Japanese men. Hypertens Res 2001;
24: 685–89.
Bhopal R, Fischbacher C, Vartiainen E, Unwin N, White M,
Alberti G. Predicted and observed cardiovascular disease in South
Asians: application of FINRISK, Framingham and SCORE models to
Newcastle Heart Project data. J Public Health (Oxf) 2005; 27: 93–100.
Sans S, Fitzgerald AP, Royo D, Conroy R, Graham I. Calibrating the
SCORE cardiovascular risk chart for use in Spain. Rev Esp Cardiol
2007; 60: 476–85.
Simmonds MC, Wald NJ. Risk estimation versus screening
performance: a comparison of six risk algorithms for cardiovascular
disease. J Med Screen 2012; 19: 201–05.
Mendis S, Lindholm LH, Mancia G, et al. World Health
Organization (WHO) and International Society of Hypertension
(ISH) risk prediction charts: assessment of cardiovascular risk for
prevention and control of cardiovascular disease in low and
middle-income countries. J Hypertens 2007; 25: 1578–82.
Moran AE, Forouzanfar MH, Roth GA, et al. The global burden of
ischemic heart disease in 1990 and 2010: the global burden of
disease 2010 study. Circulation 2014; 129: 1493–501.
Yano K, Kagan A, McGee D, Rhoads GG. Glucose intolerance and
nine-year mortality in Japanese men in Hawaii. Am J Med 1982;
72: 71–80.
Farzadfar F, Finucane MM, Danaei G, et al. National, regional, and
global trends in serum total cholesterol since 1980: systematic
analysis of health examination surveys and epidemiological studies
with 321 country-years and 3·0 million participants. Lancet 2011;
377: 578–86.
27 Prospective Studies Collaboration. Blood cholesterol and vascular
mortality by age, sex, and blood pressure: a meta-analysis of
individual data from 61 prospective studies with 55 000 vascular
deaths. Lancet 2007; 370: 1829–39.
28 Lawes CM, Bennett DA, Parag V, et al. Blood pressure indices and
cardiovascular disease in the Asia Pacific region: a pooled analysis.
Hypertension 2003; 42: 69–75.
29 Peters SA, Huxley RR, Woodward M. Diabetes as a risk factor for
stroke in women compared with men: a systematic review and
meta-analysis of 64 cohorts, including 775 385 individuals and
12 539 strokes. Lancet 2014; 383: 1973–80.
30 Huxley RR, Woodward M. Cigarette smoking as a risk factor for
coronary heart disease in women compared with men: a systematic
review and meta-analysis of prospective cohort studies. Lancet 2011;
378: 1297–305.
31 Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models:
issues in developing models, evaluating assumptions and adequacy,
and measuring and reducing errors. Stat Med 1996; 15: 361–87.
32 WHO. WHO methods and data sources for global causes of death
2000–2011. Geneva: World Health Organization, 2013.
33 Smolina K, Wright FL, Rayner M, Goldacre MJ. Incidence and
30-day case fatality for acute myocardial infarction in England in
2010: national-linked database study. Eur J Public Health 2012;
22: 848–53.
34 Degano IR, Elosua R, Marrugat J. Epidemiology of acute coronary
syndromes in Spain: estimation of the number of cases and trends
from 2005 to 2049. Rev Esp Cardiol 2013; 66: 472–81.
35 Feigin VL, Lawes CM, Bennett DA, Barker-Collo SL, Parag V.
Worldwide stroke incidence and early case fatality reported in
56 population-based studies: a systematic review. Lancet Neurol
2009; 8: 355–69.
36 Mathers C, Truelsen T, Begg S, Satoh T. Global burden of ischaemic
heart disease in the year 2000. Global Burden of Disease 2000
working paper. Geneva: World Health Organization, 2004.
37 Hammar N, Alfredsson L, Rosen M, Spetz CL, Kahan T, Ysberg AS.
A national record linkage to study acute myocardial infarction
incidence and case fatality in Sweden. Int J Epidemiol 2001;
30 (suppl 1): S30–34.
38 Krumholz HM, Normand SL, Wang Y. Trends in hospitalizations
and outcomes for acute cardiovascular disease and stroke,
1999–2011. Circulation 2014; 130: 966–75.
39 Yusuf S, Rangarajan S, Teo K, et al. Cardiovascular risk and events
in 17 low-, middle-, and high-income countries. N Engl J Med 2014;
371: 818–27.
40 Cooney MT, Dudina A, D’Agostino R, Graham IM. Cardiovascular
risk-estimation systems in primary prevention: do they differ? Do
they make a difference? Can we see the future? Circulation 2010;
122: 300–10.
41 Matheny M, McPheeters ML, Glasser A, et al. Systematic review of
cardiovascular disease risk assessment tools. Rockville (MD):
Agency for Healthcare Research and Quality (US), 2011.
42 McGorrian C, Yusuf S, Islam S, et al. Estimating modifiable coronary
heart disease risk in multiple regions of the world: the INTERHEART
Modifiable Risk Score. Eur Heart J 2011; 32: 581–89.
43 Fibrinogen Studies Collaboration. Measures to assess the
prognostic ability of the stratified Cox proportional hazards model.
Stat Med 2009; 28: 389–411.
44 Ezzati M, Lopez AD, Rodgers A, Murray CJL, eds. Comparative
quantification of health risks: global and regional burden of
diseases attributable to selected major risk factors, vols 2 and 2.
Geneva: World Health Organization, 2004.
45 Gregg EW, Cheng YJ, Cadwell BL, et al. Secular trends in
cardiovascular disease risk factors according to body mass index in
US adults. JAMA 2005; 293: 1868–74.
46 Berry JD, Dyer A, Cai X, et al. Lifetime risks of cardiovascular
disease. N Engl J Med 2012; 366: 321–29.
47 Hogerzeil HV, Liberman J, Wirtz VJ, et al. Promotion of access to
essential medicines for non-communicable diseases: practical
implications of the UN political declaration. Lancet 2013;
381: 680–89.
48 Yusuf S, Islam S, Chow CK, et al. Use of secondary prevention
drugs for cardiovascular disease in the community in high-income,
middle-income, and low-income countries (the PURE Study):
a prospective epidemiological survey. Lancet 2011; 378: 1231–43.
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00081-9
17
Comment
Cardiovascular disease is the leading cause of death
worldwide.1 Many national and international guide­
lines endorse the use of risk prediction models to
guide individualised decision-making for lifestyle
recommendations and medical treatments as part of
primary and even secondary prevention. Well known
examples are the Pooled Cohort Equations (PCE) of the
American Heart Association, the SCORE model in various
European countries, and QRisk in the UK. Risk scores
to predict cardiovascular disease risk are abundant, as
shown by a comprehensive review of the literature done
in 2009, which identified more than 100 cardiovascular
disease risk models.2 Although such a review has not
been repeated since, based on the general increase in
literature about risk models, this number is likely to
have doubled over the past 6 years. The findings of the
comprehensive review2 and others showed that most
cardiovascular disease prediction models are never
validated for predictive accuracy in individuals outside
the population they were developed for.3 Moreover,
most cardiovascular disease prediction models are
developed from single-country cohort or registry
studies, which are, generally, from North American or
European countries. However, cardiovascular disease
burden is also rapidly increasing in low-income and
middle-income countries, including those in in Asia,
Africa, and Latin America.1 Therefore, now is the time
for either cardiovascular disease prediction models
to be developed from and validated in datasets from
these countries, or for existing cardiovascular disease
prediction models to be tailored or recalibrated to these
populations.
Such development, validation and recalibration is
done comprehensively by Kaveh Hajifathalian and
colleagues in The Lancet Diabetes & Endocrinology.4 At
first glance, the investigators seem to have developed
and validated yet another cardiovascular disease
risk score—called Globorisk—using only eight North
American cohort studies that include slightly more than
50 000 participants. QRisk, by contrast, was developed
from data for more than 1 million participants.
However, Hajifathalian and colleagues clearly aimed
to rigorously develop and externally validate a novel
prediction model that can easily be tailored to other
countries around the world if some basic country-
specific data are available (data for overall cardiovascular
disease risk and population averages of the predictors
used in the model). Nowadays, such data is available
from many nationwide registries. The investigators
demonstrate how to do this for 11 countries from
different continents, including Africa, Asia, and Latin
America. In addition to using rigorous methods for
model development and validation, which were done
according to the most recent guidelines in prediction
research,5,6 the investigators also report their results
in a way that is compliant with the recently launched
TRIPOD statement for transparent reporting of
prediction modelling research.7,8
Further aspects of the study4 are also worthy of
consideration. The authors tailored their model to
various countries based on individual participant data
from recent national surveys. They developed countryspecific risk charts for predicting risk of cardiovascular
disease in individuals (shown in their figure 4) and
country-specific 10 year cardiovascular disease burden
(shown in their figure 5). How these predictions would
compare with those of other well-known recalibrated
cardiovascular disease prediction models such as
SCORE, QRisk, and PCE would be interesting to know.
The Globorisk model was developed such that countryspecific recalibration was feasible with few data.
However, if individual participant-level data from the
target setting are available, existing models can often
be similarly recalibrated, which would allow for headto-head comparisons between models per country.
Such direct comparisons will further help health-care
professionals and policy-makers in these countries
to decide which model to promote,9 which puts the
findings of Hajifathalian and colleagues in an even
broader perspective.
Recalibration of prediction models to other settings
can be done by adjustment at three levels: the baseline
cardiovascular disease risk, the average predictor
values, and the predictor-outcome associations.
Hajifathalian and colleagues recalibrated with respect
to the first two and argued that major cardiovascular
disease predictor-outcome associations are similar in
Northern American, European, and Asian populations,
although they recognise that more data are needed
from Africa and Latin America. If individual participant
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00002-9
Hybrid Medical Animation/Science Photo Library
Prediction of cardiovascular disease worldwide
Lancet Diabetes Endocrinol 2015
Published Online
March 26 date, 2015
http://dx.doi.org/10.1016/
S2213-8587(15)00002-9
See Online/Articles
http://dx.doi.org/10.1016/
S2213-8587(15)00081-9
1
Comment
data were available for these countries, recalibration of
the predictor-outcome associations would be possible,
which might yield more precise country-specific risk
charts and cardiovascular disease burden estimates.
This recalibration would, in turn, allow for enhanced
individualised prediction by the Globorisk model
and thus improved guidance for administration of
preventive CVD strategies.2
The country-specific predictions for estimated 10 year
cardiovascular disease burden are striking, particularly
the large proportion of high-risk individuals in China,
Mexico, Czech Republic, and Iran. A next step would
be to quantify the positive effects on a population
level if the Globorisk model and subsequent riskbased preventative management were used in these
countries. By use of so-called population-level linkedevidence models,10 estimates of country-specific
10 year cardiovascular disease-risk groups can be
combined with known effect sizes from randomised
trials of various treatments (eg, lipid-lowering and
blood-pressure-lowering drugs), supplemented with
treatment adherence figures, to quantify the expected
decrease in cardiovascular disease burden per country
within 10 years. These predictions might further help,
and indeed convince, decision-makers across the world
to decide on wide-scale introduction of risk-based
management for cardiovascular disease.
2
*Karel G M Moons, Ewoud Schuit
Julius Center for Health Sciences and Primary Care, University
Medical Center Utrecht, Utrecht 3508GA, Netherlands (KGMM,
ES); and Stanford Prevention Research Center, Stanford University,
Stanford, CA, USA (ES)
[email protected]
We declare no competing interests.
1
WHO. Global status report on noncommunicable diseases 2010. Geneva:
World Health Organization, 2011.
2 Matheny M, McPheeters ML, Glasser A, et al. Systematic review of
cardiovascular disease risk assessment tools. Evidence synthesis no. 85.
Rockville, MD: Agency for Healthcare Research and Quality, 2011.
3 Siontis GC, Tzoulaki I, Castaldi PJ, Ioannidis JP. External validation of new
risk prediction models is infrequent and reveals worse prognostic
discrimination. J Clin Epidemiol 2015; 68: 25–34.
4 Hajifathalian K, Ueda P, Lu Y, et al. A novel risk score to predict
cardiovascular disease risk in national populations (Globorisk): a pooled
analysis of prospective cohorts and health examination surveys.
Lancet Diabetes Endocrinol 2015; published online March 26. http://dx.doi.
org/10.1016/S2213-8587(15)00081-9.
5 Harrell FE Jr. Regression modeling strategies: with applications to linear
models, logistic regression, and survival analysis. Springer; 2001.
6 Steyerberg EW. Clinical Prediction Models: A practical approach to
development, validation, and updating. New York: Springer, 2009.
7 Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent Reporting of a
multivariable prediction model for Individual Prognosis Or Diagnosis
(TRIPOD): The TRIPOD Statement. Ann Intern Med 2015; 162: 55–63.
8 Moons KG, Altman DG, Reitsma JB, et al. Transparent reporting of a
multivariable prediction model for individual prognosis or diagnosis
(TRIPOD): explanation and elaboration. Ann Intern Med 2015; 162: W1–W73.
9 Collins GS, Moons KG. Comparing risk prediction models. BMJ 2012;
344: e3186.
10 Briggs A, Sculpher K. Decision modelling for health economic
interventions. Oxford: Oxford University Press, 2006.
www.thelancet.com/diabetes-endocrinology Published online March 26, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00002-9