Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Abstract

Objective:

Carotid artery intima-media thickness (CIMT) is a non-invasive marker of subclinical atherosclerosis and a predictor of coronary heart disease (CHD). This study aimed (1) to recalibrate the Framingham Risk Score (FRS) using Sri Lankan population data and (2) to evaluate the association between the recalibrated FRS models and carotid artery CIMT measurements.

Design, setting and participants:

A sample of 356 participants aged 40–74 with no CHD history was selected from a tertiary hospital in Sri Lanka. The first published FRS equation, β-coefficients, 10-year CHD-free survival rates (separately for all ages (model 1) and for 40–74 years (model 2)), and local risk factor prevalence were used for recalibration. CIMT was measured in mm by ultrasonography, and a composite CIMT score was derived.

Main outcome measure:

Association between recalibrated FRS models and CIMT.

Results:

The mean age of the sample was 58.7 ± 10.1 years (52.5% male). The original FRS (oFRS) categorised more participants into higher 10-year-CHD risk groups than the recalibrated FRS (rFRS) models. Among males, 30.5% and among females, 68.0% had consistent classifications across all models. CIMT-values differed significantly by risk category for both oFRS and rFRS models (P<.05), with rFRS models showing higher CIMT-values. The composite carotid scores (ACA-CIMT and ACA-Max) were positively correlated with all FRS models (P=.001). CIMT values were higher in recalibrated models, with model 1 showing higher values than model 2 in males.

Conclusions:

The recalibrated FRS models provided lower overall CHD risk estimates while maintaining stronger associations with CIMT than the original FRS, supporting their improved applicability for CHD risk prediction in the Sri Lankan population.

Keywords

Recalibration Framingham score CIMT Sri Lankan population

Introduction

Accurate risk estimation for coronary heart disease (CHD) is crucial for evidence-based treatment and prevention strategies in primary care. Cardiovascular risk prediction tools, such as the Framingham Risk Score (FRS), are recommended by cardiology guidelines for evaluating CHD prevention needs.^1–3 The FRS, initially designed to estimate the likelihood of developing CHD over ten years based on the Framingham Heart Study data, was expanded in 2008 to include additional cardiovascular events.⁴ FRS has been shown to overestimate CHD risk in certain populations, including those from Asia and Europe.^5–8 Recalibrating the FRS (rFRS) with local risk factors has yielded better risk predictions.^6,7,9,10 Non-invasive methods like measuring carotid intima-media thickness (CIMT) serve as significant indicators of early atherosclerosis, revealing individuals at risk of future CHD events before classical risk factors manifest.¹¹ Contrary relationships between FRS and CIMT have been reported probably due to not utilising region-specific CIMT values or recalibrated FRS.^12,13 The rFRS improved reliability for predicting CHD risk among Sri Lankans due to the rising burden of CHD morbidity and mortality, necessitating precise data for effective healthcare planning. Therefore, this study aimed to recalibrate the Framingham Risk Score using Sri Lankan population data and to examine how the recalibrated risk estimates correlate with carotid artery intima–media thickness (CIMT) across different CHD-risk categories.

Methods

Study design

The source population of this cross-sectional study (n = 356) was obtained from a larger comparative study carried out prospectively to determine the association between CHD and CIMT values (in mm) at a tertiary private hospital in Colombo, Sri Lanka.¹⁴

Study population

The study included participants aged 40–74 years who had regular health check-ups. Out of 476 participants initially screened, 65 were diagnosed with CHD and 55 who had signs of an ongoing infection were excluded; 356 participants without CHD were included in the analysis.¹⁴ Inclusion criteria included no evidence of angina-type chest pain, significant ECG changes, positive treadmill test, echocardiographic wall hypokinesia, or elevated Troponin I and T levels. Participants were excluded if they had a history of stroke, malignancy, carotid endarterectomy, connective tissue disorder, or ongoing infection.¹⁴ Baseline risk factors included age, sex, blood pressure, cholesterol, fasting glucose, and smoking status. The flow chart of the selection of the study sample is demonstrated in Figure S1.

Data collection

An interviewer-administered questionnaire (Supplemental material 2) was developed in English, translated into Sinhala and Tamil, and back-translated into English.^15,16 The questionnaire gathered demographic data, medical history, clinical characteristics, and biochemical results. Blood pressure readings were classified based on the Framingham Risk Score guidelines.⁴ Diabetes was defined as a fasting glucose >140 mg/dl or a known diagnosis.⁴

Biochemical studies

Following a 12-h fasting period, 5 ml of venous blood was collected under aseptic conditions for biochemical tests, including full blood count, fasting blood sugar, and lipid profile after obtaining informed consent.

Ultrasound measurements of CIMT

Carotid Doppler ultrasound, performed using a 7.5-MHz linear probe, assessed CIMT(in mm) in three segments: distal common carotid artery (CCA), carotid bulb (CB), and proximal internal carotid artery (ICA), with measurements taken at the end-diastolic phase synchronised to the ECG R-wave. Plaque was defined as a structure ≥0.5 mm or >50% of surrounding IMT or with thickness >1.5 mm.¹⁷ The mean and maximum CIMT were calculated in mm for each segment, and the composite CIMT score (mean ACA CIMT) was the average of these values across the CCA, CB, and ICA in both arteries.¹⁴

Calculation of 10-year CHD risk using the original FRS

Data on age, total cholesterol level (mg/dL), HDL cholesterol level (mg/dL), having diabetes, smoking status, and blood pressure (systolic and diastolic) levels were used to predict the 10-year CHD risk among the participants using the original FRS (oFRS).⁴ The scores for the different categories of each variable considered for females and males are given in Table S1. The scores for the different variables were summed to give the oFRS. The estimated 10-year CHD risk based on the FRS for females and males is given in Table S2.

Recalibration of FRS

Recalibration of the oFRS was done in several steps.

Step-01:

Published equations from Wilson et al. (1998) were used to calculate the Framingham functions [5]. The recalibration was done separately for males and females.⁴ The β-coefficients used in these functions for males and females are given in Table S3.

The function ‘G’ (given in equation (1)) is calculated for the population based on regression coefficients, derived from the original Framingham population, and multiplied by population characteristics such as mean age and proportions of a characteristic in a population, such as the proportion with diabetes, etc. In this instance, sample values were used by applying the sample characteristics to the coefficients given for males and females separately (Table S3). For age, the mean age of the sample was taken, and for the other variables, the proportion of the sample falling within the cutoff values for different categories (see Table 2) as given by Wilson et al. (1998) was applied.

For men;

\begin{aligned} G (men) = & (0 .04826 * mean age) - (0 .65945 * P (T < 160)) + (0 .00 * P (T C 160 – 199)) + (0.17692 * P (TC 200 – 239)) \\ + (0.50539 * P (T C 240 – 279)) + (0.65713 * P (T C \geq 280) + (0.49744 * P (H D L < 35)) \\ + (0.24310 * P (H D L 35 – 44)) + (0.00 * P (H D L 45 – 49)) - (0.05107 * P (H D L 50 – 59)) \\ - (0.48660 * P (H D L \geq 60)) - (0.00226 * P (BP optimal)) + (0.00 * P (BP normal)) \\ + (0.28320 * P (BP high normal)) + (0.52168 * P (BP stage I hypertension)) \\ + (0.61859 * P (BP stage II hypertension)) + (0.42839 * P (diabetes present)) \\ + (0.00 * P (diabetes not present)) + (0.52337 * P (smoker)) + (0.00 * P (not smoker)) \end{aligned}

(1)

where P(characteristic) is the probability of that characteristic for males in the sample. Function G(male) is 3.451.

For females;

\begin{aligned} G (female) = & (0.337666 * mean age) – (0.002686 * mean ag e^{2}) - (0.26138 * P (T C < 160)) + (0.00 * P (T C 160 – 199)) \\ + (0.20771 * P (T C 200 – 239)) + (0.24385 * P (T C 240 – 279)) + 0.53513 * P (T C \geq 280) \\ + (0.84312 * P (H D L < 35)) + (0.37796 * P (H D L 35 – 44)) + (0.19785 * P (H D L 45 – 49)) \\ - (0.00 * P (H D L 50 – 59)) - (0.42951 * P (H D L \geq 60)) - (0.53363 * P (BP optimal)) \\ + (0.00 * P (BP normal)) - (0.06773 * P (BP high Normal)) + (0.26288 * P (BP stage I hypertension)) \\ + (0.46573 * P (BP stage II Hypertension)) + (0.59626 * P (diabetes present)) \\ + (0.00 * P (diabetes not present)) + (0.29246 * P (smoker)) + (0.00 * P (not smoker)) \end{aligned}

(1)

where P(characteristic) is the probability of that characteristic for females in the sample. Function G (female) is 10.656.

Step 02:

In this step, a function L[f(L)] was calculated for each individual using the same regression coefficients by applying the regression coefficients to values of individual characteristics. For example, for a 40-year-old male, having total cholesterol between 160 and 199 mg/dL, a HDL level of 45–49 mg/dL, a high normal blood pressure, being non-diabetic and a smoker, the equation will be

\begin{aligned} f (L) = & (0 .04826 * 40) - (0 .00 * P (TC 160 – 199)) + (0 .00 * P (HDL 45 – 49)) - + (0 .28320 * P (BP high normal)) \\ + (0 .61859 * P (BP stage II hypertension)) + (0 .52337 * P (smoker)) \end{aligned}

(3)

where P(characteristic) is coded as 1 if present; 0 otherwise.

Likewise, f(L) was derived for each female as well.

Step 03:

For each individual, the results obtained from the two steps above were combined. For each individual, function (A) was calculated as

A = L - G

(4)

Then, for each individual, function (B) was calculated as

B = e^{A}

(5)

Step 04:

For each individual, the 10-year probability of fatal or non-fatal CHD events (P₁₀) was calculated as

P_{10} = 1 - [S (t)]^{B}

(6)

where S(t) is the CHD-free survival rate.

To obtain the CHD-free survival function, two models were developed following the recalibration method described by Ranasinghe et al. (2024).¹⁸ In the first model (rFRS 1), cardiovascular event data for all ages in the Colombo district were used for males and females separately. In the second model (rFRS 2), data for the 40–74-year-old general population of the Colombo district were used for males and females separately, corresponding to the age range of the original Framingham cohort. The number of coronary heart disease (CHD) events and population data used for both models were obtained from the Epidemiology Unit, Ministry of Health, Sri Lanka (2020). Event rates and 10-year CHD-free survival probabilities were calculated using the method outlined by Ranasinghe et al. (2024).¹⁸ The detailed calculations and parameter estimates are presented in Supplementary Table 4.

Data analysis

Descriptive statistics (mean ± standard deviation and frequencies) were used to summarise demographic, clinical, and biochemical characteristics. Composite CIMT values (ACA and ACA-Max) for the original Framingham Risk Score (oFRS) and the two recalibrated models (rFRS 1 and rFRS 2) were compared across 10-year CHD risk categories (<10%, 10–20%, >20%) using one-way ANOVA. The original FRS equation (Wilson et al., 1998), β-coefficients, 10-year CHD-free survival rates, and local prevalence data were used for recalibration. Data were analysed using SPSS version 27.0 (Chicago, IL). Pearson correlation coefficients were calculated to assess the linear association between mean composite carotid intima–media thickness (ACA and ACA-Max) and each of the three Framingham Risk Score models (oFRS, rFRS 1, and rFRS 2). Statistical significance for all analyses, including correlations, was set at P < 0.05. Distributions of CIMT and FRS values were summarised using mean ± SD to protect participant anonymity.

Results

The mean(±SD) age of the study sample was 58.71 ± 10.04 years (male 60.11±9.63; female 57.16±10.27 years; P=.005). The majority of the sample were males (52.5%), were married (74.2%) and had an educational level of more than General Certificate of Education Advanced Level (GCE A/L) (84%). Most participants (45.8%) were employed in the private sector. Almost 94% of the study population had a monthly family income above Rs. 51,863.00; 58.4% of the study population did not have health insurance coverage (Table 1).

Table 1.

Socio-demographic characteristics of the study population.

Variable	Non-CHD^a group [n=356] N (%)
Sex
Male	187 (52.5)
Female	169 (47.5)
Marital status
Married	264 (74.2)
Never married	70 (19.7)
Divorced	12 (3.4)
Widow	10 (3.7)
Educational level^b
Schooling years 5 to 10	10 (2.8)
Passed GCE O/L	47 (13.2)
Passed GCE A/L	199 (55.8)
Higher education	100 (28.2)
[Vocational/Diploma/Degree/post graduate]
Employment status^b
Public sector employee	74 (20.8)
Private sector employee	163 (45.8)
Employers	81 (22.7)
Own account workers	38 (10.7)
Occupation^c
^dGroup 1:
Managers	85 (23.8)
Professionals	83 (23.3)
Technicians and associate professionals	78 (21.9)
^dGroup 2:
Clerical support workers	46 (12.9)
Service and sales workers	40 (11.2)
Craft and related trades workers	6 (1.7)
^dGroup 3: Armed forces personnel	6 (1.7)
Plant and machine operators, and assemblers	3 (0.8)
Elementary occupations	9 (2.7)
Monthly family income (SLR)^e
Less than 23,518	2 (0.7)
23,519 to 36,445	2 (0.7)
36,446 to 51,862	19 (5.3)
51,863 to81, 371	235 (66.1)
And more than 81,372	98 (27.5)
Availability of insurance scheme
No insurance	208 (58.4)
Private insurance	80 (22.5)
Government insurance	68 (19.1)

CHD refers to coronary heart disease.

Department of Census and Statistics, Sri Lanka 2018.

Department of Census and Statistics, Sri Lanka 2018 and Sri Lanka Labour Demand Survey. Department of Census and Statistics. Ministry of National Policies and Economic Affairs 2017.

Occupations were categorised into three groups based on ‘International standard classification of occupation (isco - 88)’, Department of Census and Statistics Sri Lanka, 2011.

Based on income quintiles of Household Income and Expenditure Survey, Sri Lanka −2016

[Note: GCE O/L refers to General Certificate of Education Ordinary Level, and GCE A/L refers to General Certificate of Education Advance Level.]

Based on the original Framingham definitions, the categorisations included mean age, total cholesterol level(mg/dl), high-density lipoprotein cholesterol level(mg/dl), blood pressure, diabetes status and current smoker by sex (Table 2). The mean(±SD) age of the study sample was 58.7±10.1 years (male [mean±SD]; 60.1±9.6 years: female [mean±SD]: 57.2±10.3 years) (Table 2).

Table 2.

Prevalence of risk factor levels expressed according to the original Framingham risk score definitions.

Variable	Males (n=187)	Females (n=169)	All (n=356)
Age, mean (±SD)	61.1±9.6	57.2±10.3	58.7±10.1
Age categories n (%)
40–44	17 (9.1)	28 (16.6)	45 (12.6)
45–49	20 (10.7)	26 (15.4)	46 (12.9)
50–54	18 (9.6)	12 (7.1)	30 (8.4)
55–59	14 (7.5)	34 (20.1)	48 (13.5)
60–64	56 (29.9)	8 (4.7)	64 (18.0)
65–69	15 (8.0)	37 (21.9	52 (14.6)
70–74	47 (25.1)	24 (14.2)	71 (19.9)
Total cholesterol (mg/dl) n (%)
<160	20 (10.7)	22 (13.1)	42 (11.8)
160–199	125 (66.8)	117 (69.2)	242 (67.9)
200–239	28 (14.9)	16 (9.5)	44 (12.4)
240–279	11 (5.8)	10 (5.9)	21 (5.9)
≥ 280	3 (1.6)	4 (2.4)	7 (1.9)
High-density lipoprotein cholesterol (mg/dl) n (%)
<35	28 (15.0)	15 (8.9)	43 (12.1)
35–44	57 (30.5)	43 (25.4)	100 (28.1)
45–49	55 (29.4)	43 (25.4)	98 (27.5)
50–59	30 (16.0)	51 (30.2)	81 (22.8)
≥ 60	17 (9.1)	17 (10.1)	34 (9.6)
Blood pressure* n (%)
Optimal	48 (25.7)	70 (41.4)	118 (33.1)
Normal	14 (7.5)	11 (6.5)	25 (7.0)
High normal	63 (33.7)	50 (29.6)	113 (31.7)
Stage I hypertensive	56 (29.9)	36 (21.3)	92 (25.8)
Stage II-IV hypertensive	6 (3.2)	2 (1.2)	8 (2.2)
Having diabetes** n (%)	40 (21.4)	9 (5.3)	49 (13.8)
Current smoker n (%)	45 (24.1)	4 (2.3)	49 (13.8)

Blood pressure (mmHg) categories: Optimal systolic<120, diastolic < 80; Normal: systolic 120–129, diastolic 80–84; high normal: systolic 130–139, diastolic 85–89; Stage I hypertensive: systolic 140–159, diastolic 90–99; Stage II–IV hypertensive; systolic ≥160, diastolic ≥100.

Diabetes was defined according to the original Framingham risk score definition of fasting glucose >140 mg/dl or known diabetes.

Among 187 males, all three risk scores assigned similar categories to 57 (30.5%) (47 in the low-risk category and none in the intermediate and 10 in high-risk categories); 11 of those who were classified as having low-risk in the oFRFS and rFRS model 1 were classified as having intermediate-risk in rFRS model 2. Eleven males classified as low-risk in rFRS model 2 and rFRS model 1 were classified as having intermediate-risk in oFRS; 20 males classified as having intermediate-risk on oFRS and rFRS model 1 were classified as having low-risk in rFRS model 2; overall 81 subjects were classified as having intermediate-risk in the oFRS. Overall, 48 males were classified as having high-risk by the oRFS of whom10 were classified as having high-risk by all three risk scores (Table 3).

Table 3.

Comparison of 10-year coronary heart disease risk between the three risk scores (original Framingham Risk Score (oFRS). recalibrated model 1 (rFRS 1) and recalibrated model (rFRS 2)).

Original FRS risk categories	Recalibrated FRS^b risk based on model 1	Recalibrated FRS^c risk based on model 2			Total (n within original FRS risk category) <% within sex>
Original FRS risk categories	Recalibrated FRS^b risk based on model 1	<10% (low risk)	10–20% (intermediate risk)	>20% (high risk)	Total (n within original FRS risk category) <% within sex>
Males
<10% (1ow risk) (n=58)	<10% (low risk) (n)	47	11	0	58 <31.1>
	(Row %)	(81.0)	(19.0)	(0.0)
	[column % within original FRS risk category]	[100]	[100]	[0.0]
	{% of total}	{25.1}	{5.9}	{0.0}

	10–20% (intermediate risk)	0	0	0
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}

	>20% (high risk)	0	0	0
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}
10–20% (intermediate risk) (n=81)	<10% (low risk) (n)	11	47	0	81 <43.3>
	(Row %)	(19.0)	(81.0)	(0.0)
	[column % within original FRS risk category]	[35.5]	[100]	[0.0]
	{% of total}	{5.9}	{25.1}	{0.0}

	10–20% (intermediate risk)	20	0	3
	(Row %)	(87.0)	(0.0)	(13.0)
	[column % within original FRS risk category]	[64.5]	[0.0]	[100]
	{% of total}	{10.7}	{0.0}	{1.6}

	>20% (high risk)	0	0	0
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}
>20% (high risk) (n=48)	<10% (low risk) (n)	0	2	0	48 <25.7>
	(Row %)	(0.0)	(100)	(0.0)
	[column % within original FRS risk category]	[0.0]	[66.7]	[0.0]
	{% of total}	{0.0}	{1.1}	{0.0}

	10–20% (intermediate risk)	7	1	28
	(Row %)	(19.5)	(0.03)	(77.8)
	[column % within original FRS risk category]	[100]	[33.3]	[73.7]
	{% of total}	{3.7}	{0.1}	{15.0}

	>20% (high risk)	0	0	10
	(Row %)	(0.0)	(0.0)	(100)
	[column % within original FRS risk category]	[0.0]	[0.0]	[26.3]
	{% of total}	{0.0}	{0.0}	{5.3}
Females
<10% (1ow risk) (n=124)	<10% (low risk) (n)	109	15	0	124 <73.4>
	(Row %)	(96.0)	(4.0)	(0.0)
	[column % within original FRS risk category]	[100]	[100]	[0.0]
	{% of total}	{64.5}	{8.9}	{0.0}

	10–20% (intermediate risk)	0	0	0
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}
	>20% (high risk)	0	0	0

	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}
10–20% (intermediate risk) (n=43)	<10% (low risk) (n)	21	13	0	43 <25.4>
	(Row %)	(61.8)	(38.9)	(0.0)
	[column % within original FRS risk category]	[95.5]	[72.2]	[0.0]
	{% of total}	{12.4}	{7.7}	{0.0}

	10–20% (intermediate risk)	1	5	3
	(Row %)	(11.1)	(55.5)	(33.0)
	[column % within original FRS risk category]	[4.5]	[27.8]	[100]
	{% of total}	{0.01}	{3.0}	{1.8}

	>20% (high risk)	0	0	0
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}
>20% (high risk) (n=2)	<10% (low risk) (n)	0	0	0	2 <1.2>
	(Row %)	(0.0)	(0.0)	(0.0)
	[column % within original FRS risk category]	[0.0]	[0.0]	[0.0]
	{% of total}	{0.0}	{0.0}	{0.0}

	10–20% (intermediate risk)	0	0	1
	(Row %)	(0.0)	(0.0)	(100)
	[column % within original FRS risk category]	[0.0]	[0.0]	[50.0]
	{% of total}	{0.0}	{0.0}	{0.6}

	>20% (high risk)	0	0	1
	(Row %)	(0.0)	(0.0)	(100)
	[column % within original FRS risk category]	[0.0]	[0.0]	[50.0]
	{% of total}	{0.0}	{0.0}	{0.6}

Ford ES, Giles WH, Mokdad AH. The distribution of 10-year risk for coronary heart disease among US adults: findings from the National Health and Nutrition Examination Survey III. J Am Coll Cardiol. 2004; 43:1791–1796. ^bModel 1: Calculation of S(t)s for males and females in Colombo district for 2020 using all age groups. ^cModel 2: Calculation of S(t)s for males and females in Colombo district for 2020 using 40–74 year age group.

Among 169 females, all three risk scores assigned similar risk categories to 115 (68.0%) (109 in the low-risk category, 5 in the intermediate-risk category and 01 in the high-risk category). Of 124 classified as having low-risk by the rFRS model 1, 15 were classified as having intermediate-risk by rFRS model 2 (4%). Of the 43 females classified as having intermediate-risk by oFRS, 21 (48.9%) were classified as having low-risk by rFRS model 1 and rFRS model 2 (Table 3 and Figure S2).

Table 4 presents the mean ± SD values of composite carotid intima–media thickness (ACA CIMT) and maximum CIMT (ACA-Max CIMT) for each 10-year CHD-risk category across the original and recalibrated Framingham Risk Score (FRS) models, illustrating the distribution of observed CIMT values by risk level. The mean composite ACA CIMT values (mm) for different 10-year CHD-risk categories of the original (oFRS) and both the recalibrated FRS (rFRS) models were significantly different in both the sexes (Table 4). Among males, the mean ACA CIMT values in each risk category for different risk categories were significantly different from each other in all three models. The mean ACA CIMT values in the different risk categories of the rFRS models were greater than those in oFRS; except for the <10% risk group, the mean ACA CIMT values (mm) in rFRS model 2 was lower than that of rFRS model 1. Similar results were observed in males for ACA-Max CIMT; however, all ACA-Max CIMT values in different risk categories in rFRS model 2 were less than that of rFRS in model 1.

Table 4.

Mean Carotid Intima Media Thicknesses (CIMT) of composite scores (ACA and ACA-Max) by original (oFRS) and recalibrated Framingham risk scores (rFRS) by sex.

	ACA CIMT ^d Mean±SD (mm)	ACA-Max CIMT^e Mean±SD (mm)
Males
Original FRS^f
<10% (low risk) [n=58]	0.685±0.11^a	0.778±0.09^a
10–20% (intermediate risk) [n=81]	0.741±0.07^b	0.821±0.08^b
>20% (high risk) [n=48]	0.783±0.07^c	0.858±0.09^c
P-value^g	F ₂₁₈₆=13.147	F ₂₁₈₆=10.297
P-value^g	P<.0001	P<.0001
^hRecalibrated FRS^f based on Model 1
<10% (low risk) [n=118]	0.716±0.09^a	0.807±0.09^a
10–20% (intermediate risk) [n=59]	0.771±0.07^b	0.844±0.09^b
>20% (high risk) [n=10]	0.812±0.06^c	0.880±0.11^c
P-value^g	F ₂₁₈₆=9.385	F ₂₁₈₆=7.134
P-value^g	P<.001	P=.001
ⁱRecalibrated FRS^f based on Model 2
<10% (low risk) [n=85]	0.727±0.11^a	0.791±0.10^a
10–20% (intermediate risk) [n=61]	0.751±0.07^b	0.832±0.09^b
>20% (high risk) [n=41]	0.791±0.08^c	0.861±0.07^c
P-value^g	F ₂₁₈₆=6.529	F ₂₁₈₆=6.138
P-value^g	P=.002	P=.003
Females
Original FRS^f
<10% (low risk) [n=124]	0.646±0.10	0.767±0.09
10–20% (intermediate risk) [n=43]	0.720±0.06	0.843±0.06
>20% (high risk) [n=2]	Not calculated	Not calculated
P-value	t ₁₆₅= 4.79	t ₁₆₅= 5.07
P-value	P=.001	P=.001
^hRecalibrated FRS^f based on Model 1
<10% (low risk) [n=158]	0.685±0.11	0.789±0.09
10–20% (intermediate risk) [n=10]	0.770±0.09	0.860±0.09
>20% (high risk) [n=1]	Not calculated	Not calculated
P-value	t ₁₆₅= 4.19	t ₁₆₅= 4.18
P-value	P=.001	P=.001
ⁱRecalibrated FRS^f based on Model 2
<10% (low risk) [n=131]	0.675±0.11^a	0.779±0.09^a
10–20% (intermediate risk) [n=33]	0.741±0.06^b	0.850±0.06^b
>20% (high risk) [n=5]	0.820±0.04^c	0.900±0.07^c
P-value^g	F ₂₁₆₈=6.861	F ₂₁₆₈=7.754
P-value^g	P=.001	P=.001

Based on Ford ES, Giles WH, Mokdad AH. The distribution of 10-year risk for coronary heart disease among US adults: findings from the National Health and Nutrition Examination Survey III. J Am Coll Cardiol. 2004; 43:1791–1796.

Based on one-way ANOVA; Post Hoc test (Tukey).

Composite Carotid Intima Media Thickness (CIMT) [abbreviated as ACA for all carotid arteries] value (in mm) defined as the average value of all six segments of both the left and right sides. [Σ (left/right CIMT of common carotid artery (CCA)+ left / right CIMT of carotid bulb(CB)+ left/right CIMT of internal carotid artery(ICA)/6].

The maximum CIMT (ACA-Max) (in mm) was defined as the average of all maximum intima-media thickness measurements (IMT) in the three right and left segments of the carotid artery (CCA, CB, and ICA).

a, b

Means having the same letter as a superscript are not significantly different.

Model-01: Calculation of S(t)s for males and females in Colombo district (all age groups) in 2020.

Model-02: Calculation of S(t)s for males and females in Colombo district (age 40–74 years) in 2020.

Among females, there were significant differences in the mean ACA composite CIMT values and ACA-Max CIMT values between the different risk categories in all three models. Similar to the males, the mean ACA composite CIMT was higher in the recalibrated models as compared to oFRS; the CIMT values in the different risk categories of rFRS model 2 was lower than that of rFRS model 1 except in the low-risk category. The ACA-Max values for different risk categories were higher in rFRS model 1 as compared to rFRS model 2 (Table 4).

In both males and females, mean CIMT values increased progressively from the low- to the high-risk categories, with all differences reaching statistical significance (P < .01). The recalibrated models generally demonstrated higher mean CIMT values than the original FRS, with Model 1 yielding slightly greater values than Model 2. The mean ACA-CIMT cut-off values for risk prediction were higher in both recalibrated models than in the original FRS, with Model 1 showing the highest mean values as reflected in Table 4. These findings indicate a clear, positive association between higher predicted CHD risk and increased carotid arterial wall thickness in this study population.

ACA and ACA-Max values were positively correlated with oFRS and both models of rFRS (P=.001 for all) (Table 5). Both rFRS models had a higher corelation with ACA and ACA-Max values than oFRS.

Table 5.

Correlation between the mean Carotid Intima Media Thicknesses (CIMT) of composite scores (ACA and ACA-Max) and original (oFRS) and recalibrated Framingham risk scores (rFRS).

Risk model	ACA^a		ACA-Max^b
Risk model	Pearson correlation coefficient	P-value	Pearson correlation coefficient	P-value
oFRS	0.474	P=0.001	0.342	P=.001
rFRS Model 1^c	0.538	P=.001	0.413	P=.001
rFRS Model 2^d	0.502	P=.002	0.401	P=.006

Composite Carotid Intima Media Thickness (CIMT) [abbreviated as ACA for all carotid arteries] value (in mm) defined as the average value of all six segments of both the left and right sides. [Σ (left/right CIMT of common carotid artery (CCA) + left/right CIMT of carotid bulb (CB)+ left/right CIMT of internal carotid artery (ICA)/6].

The maximum CIMT (ACA-Max) was defined as the average of all maximum intima-media thickness measurements (IMT) in the three right and left segments of the carotid artery (CCA, CB, and ICA).

Model-01: Calculation of S(t)s for males and females in Colombo district (all age groups) in 2020.

Model-02: Calculation of S(t)s for males and females in Colombo district (age 40–74 years) in 2020.

Discussion

Principal findings

Global primary prevention guidelines in cardiology recommend using risk prediction scores to estimate atherosclerotic cardiovascular disease (ASCVD) risk as an initial step in preventive treatment.^3,19 The Framingham Risk Score (FRS) is widely used to assess coronary heart disease (CHD) risk, though its outcomes have been inconsistent across regions, necessitating recalibration for specific populations.¹⁹ This study applied both the original FRS (1998) and a recalibrated version (rFRS) to a Sri Lankan cohort using two models. The original FRS was preferred over the revised 2008 version, as it focuses on CHD, aligning with the study's focus, and the revised FRS tends to overestimate CHD risk in Asian and South Asian populations.^6–8^,20,21

The first recalibrated model used the CHD-free survival rate across all age groups, as done by Ranasinghe et al.,¹⁸ while the second focused on the 40–74 years age group, relevant to the study's demographic. The rFRS models classified a higher proportion of males as low-risk compared to the original FRS (model 1: 63.1% vs. oFRS: 31.1%). For females, 93.5% were classified as low-risk by model 1, compared to 73.4% in the oFRS. Model 2 also classified a larger proportion of participants as low-risk. This recalibration resulted in fewer individuals in intermediate- and high-risk categories, likely due to the lower CHD-free survival rate used in model 2. The rFRS consistently indicated lower risk for both genders, aligning with findings from rural India.⁸ The original FRS often overestimates risk in non-Western populations, as seen in studies from Australia and India.^20,22

Comparison with other studies

Recalibrating the FRS enhances CHD risk prediction accuracy, reducing overestimation and the number of individuals requiring treatment, thus optimising statin use and improving healthcare capacity, especially in resource-limited settings.²³ While all-cause mortality and cardiovascular death rates have been documented in Sri Lanka, studies on CHD-free survival rates remain scarce.²⁴ Our previous research showed that composite carotid intima-media thickness (CIMT) outperforms segment-specific CIMTs in predicting CHD risk in Sri Lanka.¹⁴ This study demonstrated significantly higher mean CIMT values in higher risk categories, as predicted by both the original and recalibrated FRS models, a trend observed in US populations.¹⁷ Integrating CIMT with FRS has been suggested to improve cardiovascular risk prediction.²⁵ However, the FRS may still have limited predictive accuracy in diverse populations, as shown in studies from Korea, France, and Nigeria.²⁶ The observed increase in mean ACA-CIMT cut-off values in both recalibrated models, particularly Model 1, supports their improved alignment with subclinical atherosclerosis indicators compared with the original FRS. The study achieved its dual objectives by successfully recalibrating the FRS for a Sri Lankan population and demonstrating that the recalibrated models better aligned with CIMT-based measures of subclinical atherosclerosis compared to the original equation.

Strengths and limitations

The main strengths of this study include the use of locally derived recalibration parameters and the comparison of two distinct models against CIMT, a validated surrogate marker of subclinical atherosclerosis.

This study has several limitations. First, the sample size (n = 356) was relatively modest, although adequate to detect statistically significant associations between recalibrated Framingham Risk Score (FRS) estimates and carotid intima–media thickness (CIMT). Second, CIMT measurements were obtained using a high-resolution B-mode ultrasound system without an automated edge-tracking function, which may introduce minor observer variability. However, all measurements were performed by a single experienced ultrasonographer following a standardised protocol to minimise measurement error. Third, as a cross-sectional study conducted in a hospital-based cohort from a private tertiary facility, the findings may be subject to selection bias and may not fully represent the broader Sri Lankan population.

In addition, the study could not be validated against prospective CHD-event data, which limits our ability to confirm whether CIMT improves long-term CHD risk prediction. Subgroup analyses by sex were restricted due to small sample sizes, and technical constraints prevented assessment of carotid plaque burden – an important factor for future investigations. Moreover, given the rapid changes in cardiovascular risk profiles and CHD incidence in developing countries such as Sri Lanka, repeated recalibrations using multiple population-based cohorts will be essential for accurate, locally relevant risk prediction tools. Future longitudinal studies with larger, community-based samples and automated CIMT tracking systems are recommended to validate and strengthen these findings.

Implications and conclusions

Both rFRS models classified more persons in the <10% risk category compared to the oFRS among both males and females. The mean ACA-CIMT cut-off values for risk prediction were higher in both rFRS models compared to the oFRS. These findings suggest that recalibration of the FRS improves its applicability to South Asian populations by better reflecting local risk factor distributions. Further studies with longitudinal follow-up are needed to validate and refine the recalibrated models for clinical use in Sri Lanka.

Supplemental Material

sj-tif-1-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-tif-1-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-docx-2-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-docx-2-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-docx-3-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-docx-3-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-docx-4-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-docx-4-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-docx-5-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-docx-5-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-docx-6-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-docx-6-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-tif-7-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-tif-7-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Supplemental Material

sj-doc-8-cvd-10.1177_20480040251405685 - Supplemental material for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study

Supplemental material, sj-doc-8-cvd-10.1177_20480040251405685 for Recalibration of the Framingham coronary heart disease risk score for a selected Sri Lankan population and its association with carotid artery intima-media thickness: A cross-sectional study by Visula Abeysuriya, Prakash Priyadharshan, Lal Gotabaya Chandrasena and Ananda Rajitha Wickremasinghe in JRSM Cardiovascular Disease

Footnotes

Acknowledgements:

The authors thank their colleagues at the Division of Cardiology, Nawaloka Hospital PLC, Colombo, Sri Lanka.

ORCID iD

Visula Abeysuriya

Ethics approval

Ethics approval for the study was obtained from the Ethics Review Committee of the Faculty of Medicine, University of Kelaniya (reference number: P/118/06/2019).

Consent to participate

All participants who provided informed written consent for participation were included in the study.

Consent for publication

Not applicable

Authors’ contributions

VA and ARW were the lead authors responsible for the manuscript's preparation, study design, data collection, entry, statistical analysis, interpretation, and writing. PP and LGC contributed to data collection, assisted with statistical analysis, and reviewed the manuscript. All authors approved the final version and are accountable for the accuracy and integrity of the work.

Visula Abeysuriya conceived and designed the study, collected the data, performed the statistical analysis, and drafted the manuscript.

Prakash Priyadharshan provided clinical oversight, supervised CIMT measurements, and contributed to the interpretation of cardiovascular data.

Lal Gotabaya Chandrasena contributed to participant recruitment and coordination of the study.

Ananda Rajitha Wickremasinghe supervised the overall study design, provided critical revisions for important intellectual content, and approved the final version for submission.

All authors reviewed and approved the final version of the manuscript and agree to be accountable for all aspects of the work.

Visula Abeysuriya and Ananda Rajitha Wickremasinghe contributed equally to this work as joint first authors.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability and materials

The datasets generated and/or analysed during the current study are not publicly available due to privacy concerns and confidentiality agreements but are available from the corresponding author upon reasonable request.

Guarantor

Dr Visula Abeysuriya is the guarantor for this work and accepts full responsibility for the integrity of the data, the accuracy of the analysis, and the decision to submit for publication.

Supplemental material

Supplemental material for this article is available online.

References

Piepoli

Hoes

Agewall

, et al. 2016 European Guidelines on cardiovascular disease prevention in clinical practice: the Sixth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of 10 societies and by invited experts) Developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR). Atherosclerosis 2016; 252: 207–274.

Catapano

Graham

De Backer

, et al. 2016 ESC/EAS guidelines for the management of dyslipidaemias. Eur Heart J 2016; 37: 2999–3058l.

Pylypchuk

Wells

Kerr

, et al. Cardiovascular disease risk prediction equations in 400 000 primary care patients in New Zealand: a derivation and validation study. Lancet 2018; 391: 1897–1907.

Wilson

PWF

D’Agostino

Levy

, et al. Prediction of coronary heart disease using risk factor categories. Circulation 1998; 97: 1837–1847.

Liu

Hong

D’Agostino

, et al. Predictive value for the Chinese population of the Framingham CHD risk assessment tool compared with the Chinese Multi-provincial Cohort Study. J Am Med Assoc 2004; 291: 2591–2599.

Empana

. Are the Framingham and PROCAM coronary heart disease risk functions applicable to different European populations? The PRIME Study. Eur Heart J 2003; 24: 1903–1911.

Chow

Joshi

Celermajer

, et al. Recalibration of a Framingham risk equation for a rural population in India. J Epidemiol Community Health (1978) 2009; 63: 379–385.

Hua

McDermott

Lung

, et al. Validation and recalibration of the Framingham cardiovascular disease risk models in an Australian Indigenous cohort. Eur J Prev Cardiol 2017; 24: 1660–1669.

Conroy

. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J 2003; 24: 987–1003.

10.

Marrugat

D’Agostino

Sullivan

, et al. An adaptation of the Framingham coronary heart disease risk function to European Mediterranean areas. J Epidemiol Community Health (1978) 2003; 57: 634–638.

11.

Zyriax

Dransfeld

Windler

. Carotid intima–media thickness and cardiovascular risk factors in healthy volunteers. Ultrasound Journal 2021; 13: 17.

12.

Touboul

Labreuche

Vicaut

, et al. Carotid intima-media thickness, plaques, and Framingham risk score as independent determinants of stroke risk. Stroke 2005; 36: 1741–1745.

13.

Abeysuriya

Perera

BPR

Wickremasinghe

. Regional and demographic variations of carotid artery intima and media thickness (CIMT): a systematic review and meta-analysis. PLoS One 2022; 17: e0268716.

14.

Abeysuriya

Wijesinha

NAI

Priyadharshan

, et al. Composite carotid intima-media thickness as a risk predictor of coronary heart disease in a selected population in Sri Lanka. PLoS One 2022; 17: e0271986.

15.

Bots

Groenewegen

Anderson

, et al. Common carotid intima-media thickness measurements do not improve cardiovascular risk prediction in individuals with elevated blood pressure: the USE-IMT collaboration. Hypertension 2014; 63: 1173–1181.

16.

Riley

Guthold

Cowan

, et al. The World Health Organization STEPwise approach to noncommunicable disease risk-factor surveillance: methods, challenges, and opportunities. Am J Public Health 2016; 106: 74.

17.

Touboul

Hennerici

Meairs

, et al. Mannheim carotid intima-media thickness and plaque consensus (2004-2006-2011). Cerebrovasc Dis 2012; 34: 290–296.

18.

Ranasinghe

Ekanayake

EMDS

Ranasinghe

, et al. Recalibration of Framingham risk for a local population of Sri Lanka. BMC Public Health 2024; 24: 1–9.

19.

Sivaswamy

Sud

, et al. Calibration and discrimination of the Framingham risk score and the pooled cohort equations. CMAJ 2020; 192: E442–E449.

20.

Liu

. 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.

21.

Hense

Schulte

Löwel

, et al. Framingham risk function overestimates risk of coronary heart disease in men and women from Germany--results from the MONICA Augsburg and the PROCAM cohorts. Eur Heart J 2003; 24: 937–945.

22.

Barzi

Patel

, et al. Cardiovascular risk prediction tools for populations in Asia. J Epidemiol Community Health (1978) 2007; 61: 115–121.

23.

Sud

Sivaswamy

Chu

, et al. Population-Based recalibration of the Framingham risk score and pooled cohort equations. J Am Coll Cardiol 2022; 80: 1330–1342.

24.

Kasturiratne

Ediriweera

De Silva

, et al. Patterns and predictors of mortality in a semi-urban population-based cohort in Sri Lanka: findings from the Ragama Health Study. BMJ Open 2020; 10: e038772.

25.

Sumayin Ngamdu

Adewale

Mallawaarachchi

, et al. Association between the Framingham risk score and carotid artery intima-media thickness in patients with human immunodeficiency virus. Am J Cardiol 2020; 127: 156–162.

26.

Youn

Lee

Kim

, et al. Normative values and correlates of mean common carotid intima-media thickness in the Korean rural middle-aged population: the atherosclerosis RIsk of rural areas iN Korea general population (ARIRANG) study. J Korean Med Sci 2011; 26: 365–371.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB

0.07 MB

0.02 MB

0.01 MB

0.02 MB

2.02 MB

0.67 MB