Carotid Ultrasound Screening for Coronary Heart Disease: Results Based on a meta-analysis of 18 Studies and 44,861 Subjects

Abstract

Background

Carotid artery ultrasound is a possible screening test for future coronary heart disease (CHD) events to select individuals for preventive treatment.

Objectives

To assess the screening performance of carotid artery intima-media thickness (IMT) and carotid plaque in the identification of individuals with CHD.

Methods

meta-analysis of case-control and cohort studies, reporting carotid IMT or plaque in individuals with and without CHD. Screening performance (detection rates [DRs] for specified false-positive rates [FPRs]) was assessed from the relative Gaussian distributions of IMT among individuals with and without CHD and from the proportion of affected and unaffected individuals with plaque.

Results

Eighteen studies, involving 2920 individuals with CHD (mean age range 46–73 years) and 41,941 without (aged 44–73 years) were included in the meta-analysis. For plaque the DR was 62% for an FPR of 30%; likelihood ratio (2.1 [95% CI 1.6–2.4]). For IMT, the DR was 65% for the same 30% FPR (IMT cut-off ≥0.82 mm); likelihood ratio 2.2 (1.9–2.5). The results were similar in case-control and cohort studies.

Conclusion

Neither carotid plaque nor IMT has a CHD screening performance that is sufficiently discriminatory between affected and unaffected individuals to be a worthwhile screening test.

Introduction

Carotid artery ultrasound is a non-invasive imaging method used to measure the thickness of the wall of carotid arteries and the presence or absence of atherosclerotic plaque. Observational studies have shown that increased carotid intima-media thickness (IMT) or the presence of plaque is associated with angiographic coronary heart disease (CHD) and the risk of coronary death and myocardial infarction^1,2 raising interest in using such measurements in screening to select individuals for the investigation and treatment of CHD.^3,4

Observational studies have tended to present the risk of CHD as an odds ratio for a higher versus a lower IMT result,³ which demonstrates whether or not there is an association between IMT and disease, but does not permit a quantitative assessment of carotid ultrasound as a screening test. There is therefore uncertainty over the value of carotid ultrasound in screening and whether the measurement of IMT or plaque or a combination of the two is the more discriminatory variable in screening.

This prompted us to carry out a meta-analysis of case-control and cohort studies that provided information on both carotid IMT and plaque in individuals with and without CHD, to estimate the screening performance of these measurements in predicting CHD.

Methods

Data sources were identified using a search of Medline and EMBASE databases and a manual search of the citation lists of the relevant publications and reviews. Keywords for the Medline search were carotid ultrasound or carotid intima-media or carotid plaque and CHD or coronary artery disease or ischaemic heart disease or myocardial infarction. All studies published in English up until March 2008 were considered eligible provided they reported either the carotid ultrasound IMT and the presence or absence of carotid plaque in affected individuals with either a confirmed CHD event (death or myocardial infarction) or symptoms of ischaemic chest pain (angina) supported by a confirmed coronary artery stensois of ≧ 50% on a coronary angiogram. Control subjects were unaffected individuals from the general population without a history of vascular disease or subjects undergoing coronary angiography with normal coronary arteries in whom either carotid ultrasound IMT and plaque measurements were reported. Studies (i) with fewer than 20 affected or unaffected individuals and (ii) with controls who were patients with ischaemic chest pain and mild coronary artery disease (generally <50% coronary stenosis on coronary angiography) were excluded. The initial search generated 463 potentially relevant studies which reduced to 74 on review of the abstracts and 18 on review of the full publication^5–23 (in one cohort study the results on IMT were published separately from those on plaque).

Statistical Analysis

The published studies included in the meta-analysis generally reported the arithmetic mean IMT values and standard deviation. For each study, the IMT and standard deviation values were log 10 transformed because of the positive skew of the IMT distributions, using published formulae for the mean and standard deviation of the log-Gaussian distribution²⁴ (see Statistical Appendix). Probability plots of log-transformed data, from two studies that reported individuals data points, showed a reasonably good fit to a Gaussian distribution.^9,13

A summary mean (and standard deviation) IMT was calculated for affected and unaffected individuals, weighted by 1/standard error² in affected and unaffected individuals, respectively, using a random effects model. Screening performance was estimated from the relative Gaussian distributions of IMT in affected and unaffected individuals for all studies combined. IMT is a quantitative variable, so detection rates (DRs) were estimated at specified false-positive rates (FPRs) and the corresponding IMT cut-offs were determined. Plaque is a qualitative variable (either present or absent), so a summary DR (the proportion of unaffected individuals with plaque) and a summary FPR (the proportion of unaffected individuals with plaque) were calculated for all studies, weighted by 1/standard error² using a random effects model. Heterogeneity was assessed by Cochranes Q-test for log mean IMT and plaque in cases and controls and meta-regression was used to account for possible sources of heterogeneity. STATA (Version 10) was used for all analyses.

Results

Table 1 summarizes the characteristics of the 18 studies used in the meta-analysis that provided information on carotid IMT or plaque or both in a total of 2920 CHD-affected individuals and 41,941 unaffected individuals, listed in order of increasing age of participants (65% were men). Table 2 shows the reported IMT results in each study and the proportion of participants with plaque in affected and unaffected individuals. Web Table 2 (www.wolfson.qmul.ac.uk/epm/publications/webtables) gives the corresponding log IMT values for each study and the summary weighted estimates for all studies combined.

Table 1

Studies included in the meta-analysis according to age of participants

			Number		Mean age		Description of
First author	% men	Design	CHD affected	Unaffected	CHD affected	Unaffected	Coronary heart disease (CHD)	Intima-media thickness (IMT)	Plaque
Vrtovec⁵	100	Case control	30	30	46	44	Myocardial infarction	Maximum of left and right common carotid artery^†	> 1.0 mm
Lorenz⁶	49	Cohort	227	4825	50	50	Myocardial infarction	Mean of left and right common carotid artery	Not measured
Cerne⁷	85	Case control	100	70	54	52	Angina and angiographic stenosis >50%	Maximum of right common carotid artery	>2.0 mm
Chambless⁸^*	43	Cohort	290	12,551	57	54	Death or myocardial infarction	Mean of multiple measurements from common carotid artery^†	Not measured
Geroulakas⁹	100	Case contro	75	34	58	54	Angina and angiographic stenosis >50%	Mean of multiple measurements from common carotid artery	Not measured
Kablack-Ziembicka¹⁰	78	Case control	463	95	59	56	Angina or myocardial infarction and angiographic stenosis >50%	Mean of multiple measurements from common carotid artery^†	>1.3 mm
Hunt¹¹^†	43	Cohort	399	11,976	57	54	Death or myocardial infarction	–	>1.5 mm
Alan¹²	56	Case control	180	53	59	54	Angina and angiographic stenosis >50%	Mean of left and right common carotid artery	≥ 1.3 mm
Wald¹³	75	Case control	55	45	59	55	Myocardial infarction and angiographic stenosis >50%	Mean of left and right common carotid artery	> 1.0 mm
Rosvall¹⁴	65	Cohort	113	5050	61	57	Death or myocardial infarction	Mean of left and right common carotid artery	>1.2 mm
Sun¹⁵	56	Case control	78	69	62	61	Angina and angiographic stenosis >50%	Mean of multiple measurements from common carotid artery	Not measured
Balbirini¹⁶	80	Case control	107	43	64	62	Angina and angiographic stenosis >50%	Mean and maximum of left and right common carotid artery	>2.5 mm
Ebrahim¹⁷	53	Case control	102	673	66	66	Myocardial infarction	Mean of left and Right Common carotid artery^†	≥ 1.2 mm
Kato¹⁸	59	Case control	72	46	67	65	Angina and angiographic stenosis >50%	Maximum of left and right common carotid artery	Not measured
Teragawa¹⁹	63	Case control	56	25	67	62	Angina and angiographic stenosis >50%	Mean of multiple measurements from common carotid artery	Not measured
Morito²⁰	59	Case control	75	41	70	65	Angina and angiographic stenosis >50%	Mean of left and right common carotid artery	>1.1 mm
Raso²¹	59	Case control	37	33	71	71	Angina and angiographic stenosis >50%	Maximum of left and right common carotid artery	Not measured
Iqlesias del Sol²²	42	Cohort	194	2073	72	70	Myocardial infarction	Maximum of left and right common carotid artery	Not measured
O'Leary²³	39	Cohort	267	4209	73	73	Myocardial infarction	Maximum of left and right common carotid artery	Not measured

Studies also measured IMT in carotid bifurcation and internal carotid artery

†

Studies by Chambless and Hunt were separate reports from the same ARIC cohort

Table 2

Intima-media thickness and plaque in studies of coronary heart disease (affected) and unaffected individuals

	Affected				Unaffected
First author	Number	IMT (mm) Mean	SD	% with plaque	Number	IMT (mm) Mean	SD	% with plaque
Vrtovec⁵	30	0.80	0.19	63	30	0.57	0.11	20
Lorenz⁶	227	0.78^*	0.13	–	4825	0.69^*	0.12	–
Cerne⁷	100	1.00	0.30	50	70	0.70	0.20	8.6
Hunt^{11 †}	399	–	–	58	1 1,976	–	–	34
Chambless^{8 †}	290	0.70	0.16	–	12,551	0.63	0.16	–
Geroulakas⁹	75	0.91	0.18	–	40	0.71	0.16	–
Alan¹²	180	0.82	0.10	37	53	0.57	0.10	11
Wald¹³	55	0.91	0.28	75	45	0.75	0.18	18
Kablack-Ziembicka^{10 ‡}	463	1.32	0.28	–	95	1.01	0.19	–
Rosvall¹⁴	113	0.84	0.15^§	65	5050	0.77	0.15^§	43
Sun¹⁵	78	1.10^*	0.40	–	69	0.90^*	0.20	–
Balbirini¹⁶	107	0.89	0.13	71	43	0.71	0.18	44
Ebrahim¹⁷	102	0.84	0.17	79	673	0.79	0.18	54
Teraqawa¹⁹	56	1.09	0.37	–	25	0.79	0.20	–
Kato¹⁸	25	0.99	0.19	–	84	0.84	0.16	–
Morito^{20 ‡}	75	0.90	0.36	–	41	0.77	0.20	–
Raso²¹	37	1.10	0.24	–	33	0.84	0.23	–
Iglesias del Sol²²	194	1.17	0.29	–	2073	1.03	0.22	–
O'Leary²³	267	1.10^*	0.19	–	4209	1.01^*	0.19	–

IMT, intima-media thickness; SD, standard deviation

Median in place of mean

†

Data from ARIC study; IMT reported by Chambless,⁸ plaque reported by Hunt¹¹

‡

Kablack-Ziembicka¹⁰ and Morito²⁰ specified a plaque cut-off but did not report prevalence in affected and unaffected individuals

Standard deviation of cases and controls (figures not given separately)

Where SD missing the SD was calculated directly from the proportion of subjects in categories of IMT (see Methods)

Figure 1 shows forest plots of carotid IMT (and 95% CI) in affected (upper plot) and unaffected individuals (lower plot) for each study, ranked according to age of participant and for all studies combined. Median IMT values are shown (the antilog of the individual and pooled weighted mean log IMT values. The median IMT in affected (mean age, 62 years) was 0.92 mm (95% CI 0.83–1.01) and that in unaffected (mean age, 58 years) was 0.76 mm (0.68–0.84). Median IMT in affected individuals was similar in the 12 case-control studies (0.93 mm [0.83–1.05]) and in the six cohort studies (0.89 mm [0.73–1.08]). In unaffected individuals they were also similar, 0.74 mm (0.68–0.81) and 0.8 mm (0.66–0.97), respectively.

Figure 1

Forest plots of carotid IMT (and 95% CI) in affected (upper plot) and unaffected individuals (lower plot) for each study, ranked according to age of participant and for all studies combined

Figure 2 shows forest plots of the proportion of CHD-affected individuals with plaque (the DR) and unaffected individuals with plaque (FPR) in each study together with the weighted summary estimates for all studies combined. The overall summary screening performance yielded a DR of 62% (95% CI 50–75) for an FPR of 30% (18–39%). There were only two cohort studies that reported plaque, limiting the extent to which differences between case-control and cohort studies could be examined. The proportion in affected individuals in case-control studies was 63% (47–77%) and in cohort studies was 60% (54–66%) and in unaffected individuals, 24% (8–46%) and 38% (30–48%), respectively.

Figure 2

Forest plots of the proportion of CHD affected individuals with plaque (detection rate) and unaffected individuals with plaque (false-positive rate) in each study ranked according to age of participant, together with the weighted summary estimates for all studies combined. CHD, Coronary heart disease

There was heterogeneity in the proportion of affected and unaffected individuals with plaque (P < 0.001) and in the median IMT in affected and unaffected individuals between studies (P < 0.001), justifying the use of a random effects model in the analysis. Sensitivity analyses were conducted to examine the influence of individual studies on the pooled estimates. Each study was removed in turn and the analysis repeated on all but the removed study. The overall results were not materially affected by any one study. Meta-regression was used to examine age and other sources of variation between studies such as differences in the methods used to measure IMT (the mean of multiple IMT measurements was used in 11 studies and the maximum IMT in 7) and the definition of plaque (<2 mm in 8 studies and >2 mm in 2), differences in study design (case control in 12 studies and cohort in 6), CHD outcome used (death or myocardial infarction in 7 studies and angina supported by an angiographic stenosis of >50% in 11) and the proportion of men and women in each study. With the exception of age, there were no statistically significant effects observed in meta-regression analyses of IMT or plaque on any of these variables.

IMT increased with age by 1.4% per year (95% CI 0.7–2.1, P < 0.001). In unaffected individuals the increase was greater than in affected (1.7% versus 1.1%); however, this difference was not statistically significant (P = 0.372). The presence of plaque increased by about 1.5 percentage points per year (0.3–2.8, P = 0.018). Again, the increase was greater in unaffected individuals (2 versus 1 percentage point), but the difference was not statistically significant (P = 0.359). There was no effect of age on the variance for either measure.

Figure 3 shows relative frequency distributions of IMT in affected and unaffected individuals (mean age 59 years). The DRs are shown for a range of FPRs, together with the IMT cut-offs (mm) that specify these FPRs; DRs of 19%, 30% and 47% for FPRs of 5%, 10% and 20%, respectively.

Figure 3

Relative frequency distributions of IMT in affected and unaffected individuals (mean age 59), together with the detection rates and IMT cut-offs specified by a 5%, 10% and 20% false-positive rate. IMT, intima-media thickness

Table 3 gives the screening performance (DR, FPR and the corresponding likelihood ratios) using plaque and IMT for all studies combined. The IMT-based DR is estimated for the summary FPR observed with plaque (30%) to allow a direct comparison between IMT and plaque. The screening performance for the two ultrasound variables was similar.

Table 3

Comparison of screening performance between intima-media thickness and plaque from meta-analysis of 1 8 studies

	FPR (%)	DR (%)	Cut-off (mm) (IMT only)	Likelihood ratio (95% CI)
Plaque	30	62	–	2.1 (1.6–3.4)
IMT	30	65	0.82	2.2 (1.9–2.5)

IMT, intima-media thickness; DR, detection rate

Discussion

The results of this meta-analysis show that carotid IMT and carotid plaque each have similar screening performance in identifying patients with CHD. At an FPR of 30% (the summary value using plaque) the DR for plaque was 62% and for IMT it was 65% (IMT cut-off ≥0.82 mm). Both ultrasound variables therefore yield a likelihood ratio of about 2; indicating about a two-fold increase in risk in those with a screen-positive result, compared with an untested population.

This is the first meta-analysis to provide a quantitative assessment on the value of carotid ultrasound as a screening test for CHD. Previous observational studies and meta-analysis of such studies have focused attention on the association between plaque or IMT and CHD.⁴ These generally presented results as a relative risk of CHD for a very high IMT (say the top fifth of the distribution) compared with a very low IMT (the bottom fifth of the distribution). This comparison is useful for demonstrating whether an association is present but not for assessing its value as a screening test, because the groups being compared are mutually exclusive and most people in the middle of the distribution, where most cases tend to occur, are not considered in the analysis.²⁵ When IMT is examined using overlapping distributions, as in our analysis, the probability of a positive test result is estimated relative to the whole sample population rather than only those in the tails of the distribution. This permits an assessment of the DR for specified FPRs. It also allows comparison of IMT with plaque and with other risk factors which are used as screening tests for cardiovascular disease. For example, at age 55 and an FPR of 5%, the DR for IMT in our meta-analysis was about 19%, compared with previously published estimates of 15% using serum cholesterol,²⁵ 13% for diastolic blood pressure,²⁶ 13% for serum homocysteine²⁷ and about 10% for C-reactive protein.²⁸

There was a suggestion from the results of our meta-analysis that screening performance may decline with increasing age. While we lacked the power to confirm this pattern, we cannot exclude the possibility that screening may be better at younger ages. For plaque, for example, the estimated DR at age 45 was 41% for an FPR of 12% (likelihood ratio 3.3) and at age 65 the DR was 72% for an FPR of 43% (likelihood ratio 1.6). While a three-fold concentration in risk in a 45-year old person with a positive carotid ultrasound result may be seen as a worthwhile means of targeting preventive treatment, the background risk in this age group is relatively low (about 1 in 1000). The effect of screening would be to identify a group with a risk of 1 in 300 (3.3/1000) and this would be at the cost of missing over half of all cases of CHD (DR about 41%). Whatever the preventive remedy offered, screening is unlikely to be useful; if the preventive remedy were hazardous (e.g. coronary artery bypass surgery) the difference in risk between positive and negative results would be too small to justify screening and if it were safe (e.g. statins) there would be little reason to withhold treatment from so many people who will develop CHD simply because of negative screening test results. The conclusion contrasts with ultrasound screening for abdominal aortic aneurysms likely to rupture, which is very discriminatory. Here, a screen-positive result (>6 cm aortic diameter on ultrasound) is associated with a likelihood ratio of 143 for aortic rupture (DR 86% for a 0.6% FPR); a level of discrimination that has reasonably been judged sufficient to justify population screening.²⁹

Study characteristics other than age did not materially influence the results from this analysis. In particular, studies that classified CHD as ‘hard’ disease events (death and non-fatal MI) or ‘soft’ angiographic endpoints (angina and evidence of coronary artery disease on angiography), showed no statistically significant differences in ultrasound measurements. For example, IMT was, respectively, 0.94 mm (0.88–1.00) and 0.90 mm (0.76–1.06) in affected individuals and 0.74 mm (0.66–0.84) and 0.77 mm (0.67–0.89) in unaffected individuals. Similarly, there were no statistically significant differences in the results from studies of case-control or cohort study design or from studies that adopted multiple (mean) or single (maximal) measures of IMT and different size cut-offs to define plaque. Our results are therefore reasonably robust across different study designs and measurement methods. Substantial publication bias is unlikely because a standard statistical assessment of publication bias (the regression asymmetry test)³⁰ showed no basis for concern in either the studies reporting carotid IMT (P = 0.83 for affected and P = 0.64 for unaffected) or plaque (P = 0.39 and 0.82).

Neither carotid plaque nor IMT has a CHD screening performance that is sufficiently discriminatory between affected and unaffected individuals to be a worthwhile population screening test. Carotid ultrasound screening for CHD is improved by combining information on IMT and plaque (this increases the DR by about 10 percentage points at an FPR of 10%, as shown in an accompanying paper in this edition of the Journal).¹³ If this were combined with other screening tests of comparable discrimination (possibly CT calcium score³¹ or measures of arterial stiffness³²), then there may be a case for screening younger people (about age 45) to select individuals for preventive treatment. Further work is needed to assess this and the screening performance of other such tests, alone and in combination with carotid ultrasound, to determine if this would be worthwhile.

Footnotes

Statistical Appendix

If a continuous variable X has mean μ and standard deviation σ and X follows a log-Gaussian distribution and on the natural logarithm scale has mean m and standard deviation s, i.e. In (X) ~ N(m, s²), then

and

Treating equations (1) and (2) as simultaneous equations and solving for m and s gives

and

To convert a natural logarithm scale to a log₁₀ scale, equations (3) and (4) are divided by ln(10).

The following example, using data from the paper by Vrtovec et al.⁵ illustrates the method used for this transformation. The mean IMT was reported as μ = 0.80 mm and standard deviation σ = 0.19 mm in affected individuals. From equation (3), the mean on the log₁₀ scale is given as

and from equation (4) the standard deviation on the log₁₀ scale is given as

References

Lorenz

, Marcus

, Bots

, Rosvall

, Sitzer

. Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis. Circulation 2007; 115: 459–67.

Greenland

, Abrams

, Aurigemma

. Prevention Conference V: beyond secondary prevention: identifying the high-risk patient for primary prevention: noninvasive tests of therosclerotic burden: writing group III. Circulation 2000; 101: e16–22.

Stein

, Korcarz

DVM

, Hurst

. Use of carotid ultrasound to identify subclinical vascular disease and evaluate cardiovascular disease risk: a consensus statement from the american society of echocardiography carotid intima-media thickness task force. J Am Soc Echocardiogr 2008; 21: 93–111.

Sonecha

, Henein

. The role of intima-medial thickness (IMT) in clinical cardiovascular practice. Int J Cardiovasc Imaging 2005; 21: 505–7.

Vrtovec

, Keber

, Gadzijev

, Bardorfer

, Keber

. Carotid intima-media thickness of young coronary patients. Coron Artery Dis 1999; 10: 407–11.

Lorenz

, von Kegler

, Steinmetz

, Markus

, Sitzer

. Cartoid intima-media thickening indicates a higher vascular risk across a wide range. Stroke 2006; 37: 87–92.

Cerne

, Kranjec

. Atherosclerotic burden of coronary and peripheral arteries in patients with first clinical manifestation of coronary artery disease. Heart Vessels 2002; 16: 217–26.

Chambless

, Heiss

, Folsom

. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: The Atherosclerosis Risk in Communities (ARICJ study, 1987-1993. Am J Epidemiol 1997; 146: 483–94.

Geroulakas

, O'Gorman

, Kalodiki

, Sheridan

, Nicolaides

. The carotid intima-media thickness as a marker of the presence of severe symptomatic coronary artery disease. Eur Heart J 1994; 15: 781–5.

10.

Kablack-Ziembicka

, Tracz

, Przewlocki

, Pieniazek

, Sokolowski

, Konieczynska

. Association of increased carotid intima-media thickness with the extent of coronary heart disease. Heart 2004; 90: 1286–90.

11.

Hunt

, Sharrett

, Chambless

, Folsom

, Evans

, Heiss

. Acoustic shadowing on B-Mode ultrasound of the carotid artery predicts CHD. Ultrasound Med Biol 2001;27: 357–65

12.

Alan

, Ulgen

, Ozturk

. Relation between coronary artery disease, risk factors and intima-media thickness of carotid artery, arterial distensibility, and stiffness index. Angiology 2003; 54: 261–7.

13.

Wald

, Bestwick

, Morton

. Combining carotid intima-media thickness with carotid plaque on screening for coronary heart disease. J Med Screen 2009; 16: 155–9.

14.

Rosvall

, Janzon

, Berglund

, Engstrom

, Hedblad

. Incident coronary events and case fatality in relation to common carotid intima-media thickness. J Intern Med 2005; 257: 430–7.

15.

Sun

, Takasu

, Yamamoto

. Assessment of aortic atherosclerosis and carotid atherosclerosis in coronary artery disease. Jpn Circ J 2000; 64: 745–9.

16.

Balbirini

, Buttitta

, Limbruno

, Petronio

, Baglini

, Strata

. Usefulness of carotid intima-media thickness measurement and peripheral B-Mode ultrasound scan in the clinical screening of patients with coronary artery disease. Angiology 2000; 51: 269–79.

17.

Ebrahim

, Papacosta

, Whinchup

. Carotid plaque, intima media thickness, cardiovascular risk factors, and prevalent cardiovascular disease in men and women. Stroke 1999; 30: 841–50.

18.

Kato

, Aihara

, Kikuya

. Risk factors and predictors of coronary arterial lesions in Japanese hypertensive patients. Hypertens Res 2004; 24: 3–11.

19.

Teragawa

, Kato

, Kurokawa

, Yamagata

, Matsuura

, Chayama

. Usefulness of flow-mediated dilation of the brachial artery and/or the intima-media thickness of the carotid artery in predicting coronary narrowing in patients suspected of having coronary artery disease. Am J Cardiol 2001; 88: 1147–51.

20.

Morito

, Inoue

, Urata

. Increased carotid artery plaque score is an independent predictor of the presence and severity of coronary artery disease. J Cardiol 2008; 51: 25–32.

21.

Raso

, van Popele

, Schalekamp

, van der Cammen

. Intima-media thickness of the common carotid arteries is related to coronary aterosclerosis and left ventricular hypertrophy in older adults. Angiology 2002; 53: 569–74.

22.

Iglesias del Sol

, Bots

, Grobbee

, Hofman

, Witteman

JCM

. Carotid intima-media thickness at different sites: relation to incident myocardial infarction. Eur Heart J 2002; 23: 934–40.

23.

O'Leary

, Polak

, Krnmal

, Manolio

, Burke

, Wolfson

. Carotid-artery intima and media thickness as a risk factor for myocardial infarction and stroke in older adults. N Engl J Med 1999; 340: 14–22.

24.

Aitcheson

, Brown

JAC

. The Lognormal Distribution. USA: Cambridge University Press, 1969

25.

Wald

, Hackshaw

, Frost

. When can a risk factor be used as a worthwhile screening test? BMJ 1999; 319: 1562–5.

26.

Law

, Wald

, Morris

. Lowering blood pressure to prevent myocardial infarction and strokes: a new preventive strategy. Health Technol Assess 2003; 7 (No. 31): 29–38.

27.

Wald

, Law

, Morris

. The dose-response relation between serum homocysteine and cardiovascular disease: implications for treatment and screening. Eur J Cardiovasc PrevRehabil 2004; 11: 250–3.

28.

Shah

, Casas

, Cooper

. Critical approasial of CRP measurement for the prediction of coronary heart disease events: new data and systematic review of 31 prospective cohorts. Int J Epidemiol 2009; 38: 217–31.

29.

Law

, Morris

, Wald

. Screening for abdominal aortic aneurysms. J Med Screen 1994; 1: 110–6.

30.

Egger

, Davey Smith

, Schneider

, Minder

. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–34.

31.

Arad

, Goodman

, Roth

, Newsteain

, Gurci

. Coronary calcification, coronary disease risk factors, C-Reative Protein, and atherosclerotic cardiovascular disease events: the St Francis Heart Study J Am Coll Cardiol 2005; 46: 158–65.

32.

Liao

, Arnett

, Tyroler

. Arterial stiffness and the development of hypertension. Hypertension 1999; 34: 201–6.