Comparison of the Framingham Risk Score, UKPDS Risk Engine, and SCORE for Predicting Carotid Atherosclerosis and Peripheral Arterial Disease in Korean Type 2 Diabetic Patients.

Abstract

To compare the predictability of the Framingham Risk Score (FRS), United Kingdom Prospective Diabetes Study (UKPDS) risk engine, and the Systematic Coronary Risk Evaluation (SCORE) for carotid atherosclerosis and peripheral arterial disease in Korean type 2 diabetic patients.
Among 1,275 registered type 2 diabetes patients in the health center, 621 subjects with type 2 diabetes participated in the study. Well-trained examiners measured the carotid intima-media thickness (IMT), carotid plaque, and ankle brachial index (ABI). The subject's 10-year risk of coronary heart disease was calculated according to the FRS, UKPDS, and SCORE risk scores. These three risk scores were compared to the areas under the curve (AUC).
The odds ratios (ORs) of all risk scores increased as the quartiles increased for plaque, IMT, and ABI. For plaque and IMT, the UKPDS risk score provided the highest OR (95% confidence interval) at 3.82 (2.36, 6.17) and at 6.21 (3.37, 11.45). For ABI, the SCORE risk estimation provided the highest OR at 7.41 (3.20, 17.18). However, no significant difference was detected for plaque, IMT, or ABI (P = 0.839, 0.313, and 0.113, respectively) when the AUCs of the three risk scores were compared. When we graphed the Kernel density distribution of these three risk scores, UKPDS had a higher distribution than FRS and SCORE.
No significant difference was observed when comparing the predictability of the FRS, UKPDS risk engine, and SCORE risk estimation for carotid atherosclerosis and peripheral arterial disease in Korean type 2 diabetic patients.

Full text

Page 1

doi:10.4082/kjfm.2011.32.3.189Korean J Fam Med. 2011;32:189-196
Vol. 32, No. 3 Mar 2011 | 189
Korean J Fam Med
Comparison of the Framingham Risk
Score, UKPDS Risk Engine, and SCORE
for Predicting Carotid Atherosclerosis and
Peripheral Arterial Disease in Korean Type
2 Diabetic Patients
Original
Article
Hye-Ran Ahn, Min-Ho Shin, Woo-Jun Yun1, Hye-Yeon Kim2, Young-Hoon Lee3,
Sun-Seog Kweon4, Jung-Ae Rhee, Jin-Su Choi, Seong-Woo Choi5,*
Department of Preventive Medicine, Chonnam National University Medical School, 1Gwangju-Jeonnam Regional
Cardiocerebrovascular Center, Chonnam National University Hospital, Gwangju, 2Department of Preventive
Medicine, Seonam University College of Medicine, Namwon, 3Jeonbuk Regional Cardiocerebrovascular Center,
Wonkwang University Hospital, Iksan, 4Jeonnam Regional Cancer Center, Chonnam National University Hwasun
Hospital, Hwasun, 5Chungnam Regional Cardiocerebrovascular Center, Chungnam National University Hospital,
Daejeon, Korea
Background: To compare the predictability of the Framingham Risk Score (FRS), United Kingdom Prospective Diabetes
Study (UKPDS) risk engine, and the Systematic Coronary Risk Evaluation (SCORE) for carotid atherosclerosis and
peripheral arterial disease in Korean type 2 diabetic patients.
Methods: Among 1,275 registered type 2 diabetes patients in the health center, 621 subjects with type 2 diabetes
participated in the study. Well-trained examiners measured the carotid intima-media thickness (IMT), carotid plaque,
and ankle brachial index (ABI). The subject’s 10-year risk of coronary heart disease was calculated according to the FRS,
UKPDS, and SCORE risk scores. These three risk scores were compared to the areas under the curve (AUC).
Results: The odds ratios (ORs) of all risk scores increased as the quartiles increased for plaque, IMT, and ABI. For plaque
and IMT, the UKPDS risk score provided the highest OR (95% confidence interval) at 3.82 (2.36, 6.17) and at 6.21 (3.37,
11.45). For ABI, the SCORE risk estimation provided the highest OR at 7.41 (3.20, 17.18). However, no significant difference
was detected for plaque, IMT, or ABI (P = 0.839, 0.313, and 0.113, respectively) when the AUCs of the three risk scores were
compared. When we graphed the Kernel density distribution of these three risk scores, UKPDS had a higher distribution
than FRS and SCORE.
Conclusion: No significant difference was observed when comparing the predictability of the FRS, UKPDS risk engine, and
SCORE risk estimation for carotid atherosclerosis and peripheral arterial disease in Korean type 2 diabetic patients.

Keywords: Risk Assessment; Peripheral Arterial Disease; Carotid Artery Thrombosis; Diabetes Mellitus, Type 2
Received: February 16, 2011, Accepted: March 21, 2011
*Corresponding Author: Seong-Woo Choi
Tel: 82-43-713-8464, Fax: 82-43-713-8905, E-mail: jcsw74@hanmail.net
Korean Journal of Family Medicine
Copyright © 2011 The Korean Academy of Family Medicine
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://
creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium,
provided the original work is properly cited.

Page 2

Hye-Ran Ahn, et al: Comparison of Risk Scores for Predicting Atherosclerosis
190 | Vol. 32, No. 3 Mar 2011
Korean J Fam Med
Terefore we investigated the association between these risk scores
(FRS, UKPDS, and SCORE) and surrogate measures of CVD,
such as carotid plaque, CCA-IMT, and ABI. Also we compared the
predictability of these risk scores for carotid plaque, CCA-IMT and
ABI in Korean type 2 diabetic patients.
METHODS
1. Study Subjects
Between October 2008 and June 2009, out of 1,275 registered
type 2 diabetes patients in the public health center of Seo-gu,
Gwangju and Gokseng-gun, Jeollanamdo, Korea, 709 subjects with
type 2 diabetes chose to participate in this study. Te response rate
for this study was 55.6%. Seventy-eight subjects who had a history
of CVD, and 10 subjects who did not provide a blood sample
were excluded from the study. In total, 621 patients participated
in this study. This study was conducted in accordance with the
Declaration of Helsinki guideline. Te study protocol was approved
by the Institutional Review Board of Chonnam National University
Hospital (IRB number for Seo-gu, Gwangju: I-2008-11-135, IRB
number for Gokseng-gun, Jeollanamdo: I-2009-07-069), and
informed consent was obtained from each subject.
Well-trained examiners interviewed the patients using a
questionnaire that included items regarding cigarette smoking,
consumption of alcohol, physical activity, duration of diabetes,
diabetes complications, and hypertension medication.
Weight was measured to the nearest 0.1 kg while the subjects
wore light clothes and height was measured to the nearest 0.1 cm in
stocking feet. Abdominal circumference was measured to the nearest
0.1 cm at expiration through a horizontal plane around the abdomen
at the level of the midpoint between the lowest rib and the iliac
crest. Blood pressure was measured twice with a standard mercury
sphygmomanometer afer resting for at least 5 minutes.
After an overnight fast, venous blood was sampled and se rum
was separated on-site and stored at -70oC until analysis. Con-
centrations of total cholesterol, high-density lipoprotein cholesterol,
and triglycerides were analyzed using enzymatic methods. All sera
were measured using an automatic analyzer (HITACHI-7600,
Hitachi Ltd., Tokyo, Japan). LDL cholesterol was calculated using the
equation proposed by Friedewald et al.,16) except when triglycerides
exceeded 400 mg/dL, in which case the data were treated as
INTRODUCTION
In the 20th century, cardiovascular disease (CVD) became the
main cause of mortality and morbidity in Western populations, and
the global prevalence of diabetes was estimated at 171 million in
2000.1) Te risk of coronary artery disease is six times higher in type
2 diabetes patients than in the general population.2,3) Therefore, to
control CVD risk factors and lower the disease burden of diabetes,
a method devised to estimate the risk of CVD in diabetes patients is
highly necessary.
Several surrogate measures for CVD that can estimate subclinical
atherosclerosis have been developed. Common carotid artery
intima-media thickness (CCA-IMT) is known to be correlated
with coronary artery disease, stroke, and several other risk factors.4-6)
Carotid plaque has different pathogenic characteristics compared
to IMT; however, both IMT and carotid plaque share a common
association with atherosclerosis and ischemic heart.7,8) The ankle–
brachial index (ABI) is an easy method for measuring peripheral
arterial disease (PAD) and is commonly used to screen peripheral
vascular disease in patients with diabetes.9)
Wilson et al.10) estimated the risk of CVD using data from the
Framingham heart study. The Framingham Risk Score (FRS) was
developed to predict the incident risk of CVD according to age, low-
density lipoprotein (LDL) levels, high-density lipoprotein (HDL)
levels, smoking, and hypertension. Because very few diabetes patients
were included in this previous study, some uncertainty remains
according the accuracy of the FRS to predict CVD risk in diabetes
patients.11) On the other hand, some European studies have reported
that the FRS overestimates CVD risk in the general European
population.12-14) To correct this overestimation, the Systematic
Coronary Risk Evaluation (SCORE) project was initiated to develop
an appropriate risk estimation method for the general European
population.15) In 2001, the United Kingdom Prospective Diabetes
Study (UKPDS) risk engine was published based on data from
5,102 newly diagnosed type 2 patients who were followed-up for an
average of 10.4 years. Te UKPDS includes the duration of diabetes
and glycated hemoglobin (HbA1c) level as variables, thus allowing it
address the risk of CVD specifcally in diabetes patients.
Even though these risk scores were developed and used for
controlling CVD risk factors, we do not know which risk score would
beter predict future CVD events in Korean type 2 diabetic patients.

Page 3

Hye-Ran Ahn, et al: Comparison of Risk Scores for Predicting Atherosclerosis
Vol. 32, No. 3 Mar 2011 | 191
Korean J Fam Med
was defned as the CCA-IMT.
Afer at least 5-minute rest, the ABI was automatically calculated
in the supine position using aVP-1000 system (Colin Co, Komaki,
Japan), with cufs around both the arms and ankles. If any of the ABIs
were less than 0.9, the patient was defned as having PAD.
3. FRS, UKPDS, and SCORE Risk Equations
The FRS was calculated using the equation for estimating 10-
year CHD risk.10) Te UKPDS risk engine (ver. 2.0) for 10-year fatal
CHD risk was downloaded from the website,17) and used to analyze
the data. Te 10-year fatal CHD SCORE risk score for the high risk
group was used to analyze our data, as previously described.15) Te
components of these risk equations are listed in Table 1.
4. Statistical Analysis
All statistics, except the receiver operating characteristic (ROC)
curve, were performed using SPSS ver. 15.0 (SPSS Inc., Chicago, IL,
USA). Carotid plaque was dichotomized according to whether or
not there was carotid plaque. We defined abnormal CCA-IMT as
the 75th percentile of CCA-IMT, and CCA-IMT was dichotomized
according to the 75th percentile of CCA-IMT (≤ 0.83 mm or >
0.83 mm). ABI was dichotomized according to ABI > 0.9 or ABI >
0.9.18) Logistic regression was used to compare odds ratios (ORs) for
FRS, UKPDS, and SCORE risk scores in relation to carotid plaque,
CCA-IMT, and ABI data. Te areas under the curve (AUC) of the
ROC curves were compared using Stata ver. 10 (StataCorp, College
Station, TX, USA), and the distribution of the FRS, UKPDS, and
SCORE risk scores was graphed using kernel density estimation in
Stata ver. 10.
RESULTS
1. General Characteristics
Te general and biochemical characteristics of the 621 subjects
(194 men and 427 women) are detailed in Table 2. The mean
(± standard deviation [SD]) age at diabetic diagnosis was 59.2 ±
12.6 years old, and the mean diabetic duration was 8.9 ± 8.0 years.
The mean systolic blood pressure (BP) was 130.3 ± 17.8 mmHg,
and the mean diastolic BP was 71.8 ± 9.8 mmHg. The mean total
cholesterol was 195.0 ± 43.0 mg/dL, and the mean HDL cholesterol
was 47.9 ± 11.9 mg/dL. Te mean HbA1c level was 7.5 ± 1.6%, and
missing. HbA1c levels were analyzed via high-performance liquid
chromatography using a VARIANT II system (BIO-RD, Hercules,
CA, USA).
2. Carotid Plaque, CCA-IMT, and ABI
Well-trained medical doctors performed ultrasonographic
scanning of the carotid arteries using high-resolution mode B
ultrasound (SONOACE 9900, Medison, Korea) with an electrical
linear array transducer (7.5 MHz). The IMT was defined as the
distance from the leading edge of the first echogenic line to the
second echogenic line, which indicated the media-adventitia
interface. Images of the thickest point at the carotid bulb and the
thickest point within 10 mm from the carotid bulb to the CCA
were saved as CCA-IMT and then measured using Sigma Scan Pro
ver. 5.0.0 (SPSS Inc., Chicago, IL, USA). The examiners evaluated
the CCA, carotid bulb, and internal carotid artery to determine the
amount of carotid plaque. Protrusions into the lumen that were 100%
thicker than the nearest area were defned as plaque. If the plaque was
the thickest point, the distance to the nearest point without plaque
Table 1. Components of the FRS, UKPDS risk engine, and SCORE
risk equations.
FRSUKPDS risk engine
SCORE risk
equations
Sex SexSex
Age Age at diabetic
diagnosis
Age
Total cholesterol
or LDL
Total cholesterolTotal cholesterol
HDLHDL-
Systolic blood
pressure
Systolic blood
pressure
Systolic blood
pressure
DiabetesDiabetes duration-
Smoking SmokingSmoking
-Glycated
hemoglobin
-
-Atrial fibrillation-
-Ethnicity-
FRS: Framingham Risk Score, UKPDS: United Kingdom Pros-
pective Diabetes Study, SCORE: Systematic Coronary Risk Evalua-
tion, HDL: high-density lipoprotein, LDL: low-density lipoprotein.

Page 4

Hye-Ran Ahn, et al: Comparison of Risk Scores for Predicting Atherosclerosis
192 | Vol. 32, No. 3 Mar 2011
Korean J Fam Med
the mean CCA-IMT was 0.73 ± 0.15 mm. Te number of subjects
whose CCA-IMT was thicker than 75 percentile (0.83 mm) was
149 (24.7%). Te number of current smokers was 95 (15.3%). Te
number of subjects with one more carotid plaques in their carotid
arteries was 287 (47.0%), and 69 subjects (11.2%) had ABIs < 0.9.
Diabetic duration, height, weight, diastolic BP and fasting plasma
glucose are higher in males than in females. Age, age at diabetic
diagnosis, BMI, abdomen circumference, total cholesterol, HDL
cholesterol and LDL cholesterol are higher in females. Male subjects
were more likely to smoke, be more present of carotid artery plaque,
and have an ABI of less than 0.9.
2. Comparison of ORs
Te ORs of risk scores according to quartiles are listed in Table
3. Te frst quartile of the scores was set as the reference (OR = 1.0).
The ORs of all risk scores increased as the quartiles increased for
carotid plaque, CCA-IMT, and ABI. For carotid plaque, the UKPDS
risk score provided the highest OR (95% confdence interval [CI])
at 3.82 (2.36, 6.17). For CCA-IMT, the UKPDS risk score provided
the highest OR (95% CI) at 6.21 (3.37, 11.45). For ABI, the SCORE
risk estimation provided the highest OR (95% CI) at 7.41 (3.20,
17.18).
Table 2. General characteristics of subjects.
VariableTotalMaleFemale P-value
Total number 621 (100.0)194 (31.2) 427 (68.8) -
Age (y) 68.1 ± 10.3 65.2 ± 10.4 69.4 ± 10.1 <0.001
Age at diabetic diagnosis (y) 59.2 ± 12.6 55.1 ± 12.7 61.2 ± 12.1 <0.001
Diabetic duration (y) 8.9 ± 8.0 10.1 ± 8.4 8.3 ± 7.8 0.007
Height (cm) 154.5 ± 8.6 163.7 ± 6.6 150.4 ± 5.7 <0.001
Weight (kg)58.3 ± 9.7 61.5 ± 9.856.9 ± 9.4 <0.001
Body mass index (kg/m2)24.4 ± 3.7 23.0 ± 3.2 25.1 ± 3.8 <0.001
Abdomen circumference (cm)88.7 ± 9.5 86.5 ± 8.1 89.7 ± 9.9 <0.001
Systolic blood pressure (mmHg)130.3 ± 17.8 131.8 ± 19.2 129.6 ± 17.1 0.161
Diastolic blood pressure (mmHg)71.8 ± 9.8 73.7 ± 10.2 70.9 ± 9.5 0.001
Total cholesterol (mg/dL)195.0 ± 43.0 183.7 ± 43.4 200.1 ± 42.0 <0.001
Triglycerides (mg/dL)182.5 ± 87.4 180.7 ± 118.8 183.3 ± 68.8 0.739
HDL cholesterol (mg/dL) 47.9 ± 11.9 46.1 ± 12.5 48.7 ± 11.6 0.011
LDL cholesterol (mg/dL) 111.4 ± 34.6 102.9 ± 32.9 115.1 ± 34.8 <0.001
Fasting plasma glucose (mg/dL)137.6 ± 50.3 145.5 ± 61.9 134.0 ± 43.7 0.008
Glycated hemoglobin (%) 7.5 ± 1.6 7.5 ± 1.9 7.4 ± 1.5 0.675
CCA-IMT (mm) 0.73 ± 0.15 0.74 ± 0.17 0.73 ± 0.14 0.650
CCA-IMT > 75 percentile (0.83 mm) 149 (24.7) 45 (23.9) 104 (25.1)0.767
Current smoker 95 (15.3) 78 (40.2) 17 (4.0)<0.001
Carotid plaque 287 (47.0) 102 (54.0) 185 (43.9)0.022
ABI ≤ 0.9 69 (11.2) 30 (15.5) 39 (9.3) 0.022
Values are given as the mean ± standard deviation or number (%).
HDL: high-density lipoprotein, LDL: low-density lipoprotein, CCA: common carotid artery, IMT: intima-media thickness, PAD: peripheral
arterial disease, ABI: ankle-brachial index.

Page 5

Hye-Ran Ahn, et al: Comparison of Risk Scores for Predicting Atherosclerosis
Vol. 32, No. 3 Mar 2011 | 193
Korean J Fam Med
respectively).
4. Kernel Density Distribution
The Kernel density distributions of the FRS, UKPDS, and
SCORE risk scores are showed in Figure 1. Te Kernel density of the
FRS, UKPDS, and SCORE is highest at 2.0, 15.0, and 1.7, respec-
tively. UKPDS showed a higher risk score distribution than FRS or
SCORE.
DISCUSSION
We compared the predictability of the FRS, UKPDS risk engine,
and SCORE risk estimation for carotid atherosclerosis and PAD in
Korean type 2 diabetic patients. Te ORs of all risk scores increased
as the quartiles increased for carotid plaque, CCA-IMT, and ABI.
When we compared the predictability of these three risk scores
according to the carotid plaque, CCA-IMT, and ABI, no signifcant
diference was detected.
3. Comparison of AUCs
Comparisons of the AUC for the FRS, UKPDS, and SCORE risk
scores according to carotid plaque, CCA-IMT, and ABI are shown
in Table 4. No significant difference was observed among scores
for carotid plaque, CCA-IMT, or ABI (P = 0.839, 0.313, and 0.113,
Table 3. Comparison of odds ratio among FRS, UKPDS, SCORE.
Carotid plaque†
CCA-IMT‡
ABI§
FRS
Q1 (<1.2, ref) 1.001.001.00
Q2 (1.2-5.4)1.21 (0.76, 1.94) 0.84 (0.45, 1.56)1.25 (0.48, 3.26)
Q3 (5.5-14.6)2.18 (1.37, 3.46)**1.51 (0.86, 2.65)2.10 (0.87, 5.08)
Q4 (>14.6) 2.97 (1.86, 4.74)*3.49 (2.05, 5.96)*5.45 (2.43, 12.21)*
P for trend<0.001<0.001 <0.001
UKPDS
Q1 (<12.5, ref)1.001.00 1.00
Q2 (12.5-19.1)1.62 (1.01, 2.59)** 1.89 (0.98, 3.67)0.57 (0.22, 1.48)
Q3 (19.2-28.4)2.29 (1.43, 3.68)**3.01 (1.59, 5.68)** 1.20 (0.54, 2.69)
Q4 (>28.4)3.82 (2.36, 6.17)* 6.21 (3.37, 11.45)*3.72 (1.85, 7.49)*
P for trend <0.001<0.001 <0.001
SCORE
Q1 (<1.8, ref) 1.001.00 1.00
Q2 (1.8-3.4) 1.66 (1.04, 2.66)** 2.13 (1.14, 3.97)**1.25 (0.48, 3.26)
Q3 (3.5-6.8) 2.14 (1.34, 3.42)**2.12 (1.13, 3.97)**2.10 (0.87, 5.08)
Q4 (>6.8)3.81 (2.36, 6.15)*5.23 (2.90, 9.43)*7.41 (3.20, 17.18)*
P for trend<0.001<0.001<0.001
FRS: Framingham Risk Score, UKPDS: United Kingdom Prospective Diabetes Study, SCORE, Systematic Coronary Risk Evaluation, CCA:
common carotid artery, IMT: intima-media thickness, PAD: peripheral arterial disease, ABI: ankle-brachial index.
*P < 0.001. **P < 0.05. †Carotid plaque was dichotomized according to whether there was plaque or not. ‡CCA-IMT was dichotomized
according to 75 percentile of IMT (≤ 0.83 mm or > 0.83 mm). §ABI was dichotomized according to ABI ≤ 0.9 or ABI > 0.9.
Table 4. Comparison of AUCs among FRS, UKPDS, SCORE.
FRSUKPDS SCOREP-value
Carotid plaque* 0.6330.6430.634 0.839
CCA-IMT†
0.674 0.700 0.674 0.313
ABI‡
0.692 0.6820.7250.113
AUC: area under the curve, FRS: Framingham Risk Score, UKPDS:
United Kingdom Prospective Diabetes Study, SCORE: Systematic
Coronary Risk Evaluation, CCA: common carotid artery, IMT:
intima-media thickness, ABI: ankle-brachial index.
*Carotid plaque was dichotomized according to whether there was
plaque or not. †CCA-IMT was dichotomized according to 75
percentile of IMT (≤ 0.83 mm or > 0.83 mm). ‡ABI was dichoto-
mized according to ABI ≤ 0.9 or ABI > 0.9.