JACC: CARDIOVASCULAR IMAGING VOL. 5, NO.

9, 2012

© 2012 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 1936-878X/$36.00

PUBLISHED BY ELSEVIER INC. http://dx.doi.org/10.1016/j.jcmg.2012.01.022

Yield of Screening for Coronary Artery Calcium

in Early Middle-Age Adults Based on the
10-Year Framingham Risk Score
The CARDIA Study
Tochi M. Okwuosa, DO,* Philip Greenland, MD,† Hongyan Ning, MD, MS,†
Kiang Liu, PHD,† Donald M. Lloyd-Jones, MD, SCM†
Detroit, Michigan; and Chicago, Illinois

O B J E C T I V E S The purpose of this study was to assess the prevalence and distribution of coronary
artery calcium (CAC) across Framingham Risk Score (FRS) strata and therefore determine FRS levels at which
asymptomatic, young to early middle-age individuals could potentially benefit from CAC screening.

B A C K G R O U N D High CAC burden is associated with increased risk of coronary events beyond the
FRS. Expert panel recommendations for CAC screening are based on data obtained in middle-age and
older individuals.

M E T H O D S We included 2,831 CARDIA (Coronary Artery Risk Development in Young Adults) study
participants with an age range of 33 to 45 years. The number needed to screen ([NNS] number of people
in each FRS stratum who need to be screened to detect 1 person with a CAC score above the specified
cut point) was used to assess the yield of screening for CAC. CAC prevalence was compared across FRS
strata using a chi-square test.

R E S U L T S CAC scores ⬎0 and ⱖ100 were present in 9.9% and 1.8% of participants, respectively. CAC
prevalence and amount increased across higher FRS strata. A CAC score ⬎0 was observed in 7.3%, 20.2%,
19.1%, and 44.8% of individuals with FRSs of 0 to 2.5%, 2.6% to 5%, 5.1% to 10%, and ⬎10%, respectively
(NNS ⫽ 14, 5, 5, and 2, respectively). A CAC score of ⱖ100 was observed in 1.3%, 2.4%, and 3.5% of those
with FRSs of 0 to 2.5%, 2.6% to 5%, and 5.1% to 10%, respectively (NNS ⫽ 79, 41, and 29, respectively),
but in 17.2% of those with an FRS ⬎10% (NNS ⫽ 6). Similar trends were observed when findings were
stratified by sex and race.

C O N C L U S I O N S In this young to early middle-age cohort, we observed concordance between

CAC prevalence/amount and FRS strata. Within this group, the yield of screening and possibility of
identifying those with a high CAC burden (CAC score of ⱖ100) is low in those with an FRS of ⱕ10%, but
considerable in those with an FRS ⬎10%. (J Am Coll Cardiol Img 2012;5:923–30) © 2012 by the
American College of Cardiology Foundation

From the *Division of Cardiology, Wayne State University School of Medicine, Detroit, Michigan; and the †Departments of
Preventive Medicine and Medicine, Northwestern University, Feinberg School of Medicine, Chicago, Illinois. Work was
supported (or partially supported) by contracts from University of Alabama at Birmingham, Coordinating Center, N01-HC-
95095; University of Alabama at Birmingham, Field Center, N01-HC-48047; University of Minnesota, Field Center and Diet
Reading Center (Year 20 Exam), N01-HC-48048; Northwestern University, Field Center, N01-HC-48049; Kaiser
Foundation Research Institute, N01-HC-48050; University of California, Irvine, Echocardiography Reading Center (Year 5 &
10), N01-HC-45134; Harbor-UCLA Research Education Institute, Computed Tomography Reading Center (Year 15 Exam),
N01-HC-05187; Wake Forest University (Year 20 Exam), N01-HC-45205; New England Medical Center (Year 20 Exam),
N01-HC-45204 from the National Heart, Lung, and Blood Institute. All authors have reported that they have no relationships
relevant to the contents of this paper to disclose.
Manuscript received October 3, 2011; revised manuscript received January 2, 2012, accepted January 20, 2012.
924 Okwuosa et al. JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012

Yield of CAC Screening in Early Middle-Age Adults SEPTEMBER 2012:923–30

C
AC is associated with an increased risk of In the younger to early middle-age asymptomatic
coronary heart disease (CHD) events and biracial cohort of the CARDIA (Coronary Artery
provides incremental risk prediction be- Risk Development in Young Adults) study, we
yond the Framingham Risk Score (FRS) sought to ascertain the prevalence and distribution
(1). Coronary artery calcium (CAC) increases with of CAC across Framingham risk categories, strati-
age and is associated with traditional risk factor fied by sex and race. These associations can then
burden. In addition, higher CAC burden (CAC form the basis for determining the yield of CAC
score ⱖ100) carries a greater risk of CHD events screening, and therefore the FRS ranges for which
(1), and compared with traditional cardiovascular CAC scoring might be beneficial in risk assessment.
risk factors alone, CAC scoring improves risk clas- Findings from this study may facilitate further risk
sification for the prediction of CHD events (2). stratification for young, asymptomatic individuals
predicted to be at low or intermediate 10-year risk
See page 931 by age and traditional risk factors.

To date, some expert panels have recommended

METHODS
testing for CAC in intermediate-risk individuals
(FRS predicted 10-year risk 10% to 20%) (1,3) and
The CARDIA study is a multicenter, prospective
state that although it could be reasonable to screen
cohort study designed to investigate the evolution
for CAC in low- to intermediate-risk (FRS 6% to
of CHD risk in young adults. Details of the study
10%) individuals, it is reasonable to do so in those
design, as well as inclusion/exclusion criteria and
at intermediate risk (FRS 10% to 20%) (4). Others
baseline characteristics, were described previously
suggest that there is more harm than benefit result-
(16). Briefly, the CARDIA study enrolled 5,115 black
ing from CAC measurement in intermediate-
and white participants (55% women), ranging from 18
ABBREVIATIONS risk individuals (5), and yet other panels
to 30 years of age, during the period 1985 to 1986
AND ACRONYMS suggest a benefit of widespread CAC
from 4 U.S. urban areas (Birmingham, Alabama;
screening in all asymptomatic men 45 to
CAC ⴝ coronary artery calcium Oakland, California; Chicago, Illinois; and Minneap-
75 years of age and asymptomatic women
CHD ⴝ coronary heart disease
olis, Minnesota). The institutional review boards at all
55 to 75 years of age, except for those
the study sites approved the study protocol, and
FRS ⴝ Framingham Risk Score defined as very low risk based on the
written informed consent was obtained from all study
NNS ⴝ number needed to screen absence of any traditional cardiovascular
participants. The study was designed to include ap-
risk factors (6). With the exception of a
proximately balanced numbers of participants by age,
study by Taylor et al. (7), most of the studies cited
sex, race, and education level.
by these consensus panels examining associations
For the current study, we included men and women
between CAC and CHD events included partici-
with measured coronary calcium at the year 15 (our
pants with mean ages older than 50 years, likely
study baseline) examination (n ⫽ 3,043), when the
because of lower power to detect CHD events in
mean age was approximately 40 years. From this
the younger population. Thus, expert recommenda- number, we excluded 170 participants with diabetes
tions are even less clear about screening for CAC in because they are considered high-risk under current
the younger population even though this population National Cholesterol Education Program Adult
is still at risk of the development of CAC and CHD Treatment Panel III guidelines (17), and we focused
events (7). on evaluating the yield of screening in individuals at
We previously observed, based on the distribu- lower risk. Additionally, 41 participants were excluded
tion of CAC relative to FRS, in the Multi-Ethnic due to missing FRS equation covariates.
Study of Atherosclerosis cohort (mean age, 60.9 Risk factor measurements. Data on cigarette smok-
years), that there might be minimal benefit to ing status, age, race, socioeconomic measures, dia-
screening for clinically significant levels of CAC in betes history, and medication use were obtained by
very low risk persons with an FRS of ⱕ5% (8). participant self-report (16). Current smoking was
However, for young and early middle-age persons, an defined as at least 5 cigarettes per week almost every
appropriate FRS threshold above which CAC screen- week for at least 3 months. Family history of
ing might be useful is unclear. Although some studies myocardial infarction was obtained using a self-
examined the relationship between CAC distribution administered questionnaire. Blood pressure was
and FRS (9 –15), none were performed in individuals measured 3 times with a random-zero device, and
younger than 50 years of age. the average of the last 2 measurements was used.
JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012 Okwuosa et al. 925
SEPTEMBER 2012:923–30 Yield of CAC Screening in Early Middle-Age Adults

Body mass index was calculated by dividing weight test was used to compare the prevalence of CAC
in kilograms by the square of the height in meters. categories across FRS 10-year risk strata for the
The CARDIA study physical activity history ques- participants included in this study, then after strat-
tionnaire was used to assess physical activity, which ification by sex and race. All analyses performed for
was coded as exercise units (18). Venous blood the current study (CARDIA year 15 examination;
samples were obtained from participants after a age range, 33 to 45 years) were also repeated in
12-h fast. Plasma triglycerides and total and high- secondary analyses using data from the CARDIA
density lipoprotein cholesterol were determined year 20 examination (age range, 38 to 50 years).
using an enzymatic assay by Northwest Lipids NNS was defined as the number of people who
Research Laboratory (Seattle, Washington). Low- need to be screened to identify 1 individual with a
density lipoprotein cholesterol was then derived CAC score above the pre-specified CAC cut point
using the Friedewald equation (19). in each FRS category. It was calculated by dividing
Agatston CAC measurement and scoring were the total number of participants by the number of
previously described (20). The presence of CAC people with a CAC score ⬎0 (or ⱖ100) in each
was defined as having a positive, nonzero Agatston FRS stratum. The CAC amount was represented
score determined from the average of 2 scans. by median CAC scores in FRS groups.
Because of the young age of the participants, each
scan set with at least 1 nonzero score was reviewed
RESULTS
and verified by an expert investigator who was
blinded to the scan scores. There was reasonable Baseline characteristics. Our study sample consisted
agreement between scans (kappa ⫽ 0.79, with only of a total of 2,832 black and white participants
3.6% discordance). For this study, CAC scores were (mean age, 40.3 years [range, 33 to 45 years]; 53%
categorized as ⬎0 or ⱖ100. The prevalence of women). With the exception of body mass index
advanced CAC (CAC score ⱖ300 or 400) was too and some measures of socioeconomic status, there
low in this cohort because of the younger age of the were significant differences in most of the tradi-
participants. As such, we made use of a lower cut tional risk factors (including FRS) between those
point (CAC score ⱖ100, previously shown to be with a CAC score of 0 versus a CAC score ⬎0 and
associated with increased risk of CHD events) (1) a CAC score ⬍100 versus a CAC score ⱖ100
in our definition of high CAC burden. Concurrent (Table 1). Race and physical activity were significantly
FRS 10-year risk of CHD was calculated and different for the CAC score of 0 versus ⬎0 categories,
stratified as follows: 0 to 2.5%, 2.6% to 5%, 5.1% to but not CAC score ⬍100 versus ⱖ100 categories.
10%, and ⬎10%. Further stratification of FRS Ninety percent of individuals with a CAC score ⱖ100
categories for those with an FRS ⬎10% would not and an FRS ⬎10% smoked, so that cigarette smoking
have been meaningful due to the relative youth and was the prevalent cardiovascular risk factor among this
therefore low-risk composition of our study cohort. subset of our study population.
Statistical analysis. All analyses were performed us- Of 1,501 women in our study, 76 had a CAC
ing SAS software, version 9.2 (SAS Institute, Cary, score ⬎0. Of these, 66 were premenopausal and 10
North Carolina). A 2-tailed p value ⬍0.05 was were post-menopausal (data not shown). Among
considered statistically significant. The Framing- pre-menopausal women, 4.9% had the presence of
ham 10-year risk estimates for all participants were any CAC versus 6.2% of postmenopausal women
calculated using the risk prediction functions from (p ⫽ 0.45 for comparison of CAC prevalence
the National Cholesterol Education Program Adult between the 2 groups).
Treatment Panel III guidelines (17) based on an Distribution of CAC prevalence, amount, and NNS
update from the Framingham methodology re- compared across FRS strata. Table 2 shows the distri-
ported by Wilson et al. (21). The covariates in- bution of CAC scores ⬎0 and ⱖ100 across FRS
cluded in the FRS calculation were age, total and strata. Overall, CAC scores ⬎0 and ⱖ100 were
high-density lipoprotein cholesterol levels, current present in 9.9% and 1.8% of participants, respectively.
smoking status, systolic blood pressure, and the use Among individuals with CAC, median CAC scores
of antihypertensive medication. Baseline character- increased with higher FRS. As expected, the preva-
istics were compared according to FRS 10-year risk lence of CAC scores ⬎0 and ⱖ100 increased across
strata and by CAC categories using general linear greater FRS strata (Fig. 1) (both p for trend ⬍0.01).
models for continuous variables and cross- Consequently, the NNS (signifying the number of
tabulations for categorical variables. A chi-square individuals who need to be screened to detect 1 person
926 Okwuosa et al. JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012

Yield of CAC Screening in Early Middle-Age Adults SEPTEMBER 2012:923–30

Table 1. Baseline Characteristics Stratified by CAC Score (N ⴝ 2,832)

CAC Score Categories

Characteristics 0 (n ⴝ 2,553) >0 (n ⴝ 279) p Value <100 (n ⴝ 2,781) >100 (n ⴝ 51) p Value
Age, yrs 40.1 ⫾ 3.6 42 ⫾ 3.1 ⬍0.01 40.2 ⫾ 3.6 43.1 ⫾ 2.6 ⬍0.01
Female, % 56 27.2 ⬍0.01 53.6 27.5 ⬍0.01
Black race, % 45.8 34.4 ⬍0.01 44.9 33.3 0.10
SBP, mm Hg 112.4 ⫾ 14.1 118.5 ⫾ 16.1 ⬍0.01 112.9 ⫾ 14.3 120.8 ⫾ 18.4 ⬍0.01
DBP, mm Hg 74.1 ⫾ 11.1 78.1 ⫾ 12.6 ⬍0.01 74.4 ⫾ 11.2 81.4 ⫾ 14.7 ⬍0.01
Total cholesterol, mg/dl 183.8 ⫾ 34.3 198.4 ⫾ 40.6 ⬍0.01 185.0 ⫾ 34.9 199.6 ⫾ 50.6 ⬍0.01
HDL, mg/dl 51.1 ⫾ 14.3 46.1 ⫾ 14.3 ⬍0.01 50.7 ⫾ 14.3 45.9 ⫾ 15.8 0.02
LDL, mg/dl 112.3 ⫾ 30.6 127 ⫾ 37.8 ⬍0.01 113.5 ⫾ 31.4 125.4 ⫾ 43.9 0.01
Current smoking, % 19.3 32.3 ⬍0.01 20.2 39.2 ⬍0.01
Hypertension treatment, % 6.0 11.1 ⬍0.01 6.4 11.8 0.12
Lipid treatment, % 1.7 4.7 ⬍0.01 1.8 11.8 ⬍0.01
Family history of heart attack, % 19.6 26.9 ⬍0.01 19.9 45.1 ⬍0.01
FRS, % 1.3 ⫾ 2.5 3.5 ⫾ 4.8 ⬍0.01 1.4 ⫾ 2.7 5.2 ⫾ 6.9 ⬍0.01
BMI, kg/m2 28.3 ⫾ 6.2 28.7 ⫾ 5.9 0.28 28.3 ⫾ 6.2 28.2 ⫾ 6.3 0.88
Physical activity (intensity score) 352 ⫾ 282.8 403.4 ⫾ 310.8 ⬍0.01 357.0 ⫾ 285.6 365.6 ⫾ 314.3 0.83
Education, % 0.01 0.14
Less than high school 0.3 0.7 0.3 2.0
High school 20.1 28 20.7 27.5
College 58 51.3 57.4 52.9
Graduate school 21.7 20.1 21.6 17.7
Marital status: married, % 54.3 50.9 0.29 54.0 52.9 0.89
Income, % 0.38 0.29
⬍$25,000 14 16.3 14.0 21.6
$25,000–$50,000 24.2 27.1 24.6 15.7
$50,000–$75,000 23.2 20.2 22.9 21.6
⬎$75,000 38.7 36.5 38.4 41.2
Health insurance, % 87.8 85.7 0.30 87.7 86.3 0.77
Values are mean ⫾ SD or %. Baseline refers to CARDIA year 15 examinations.
BMI ⫽ body mass index; CAC ⫽ coronary artery calcium; DBP ⫽ diastolic blood pressure; FRS ⫽ Framingham Risk Score; HDL ⫽ high-density lipoprotein; LDL ⫽ low-density lipoprotein;
SBP ⫽ systolic blood pressure.

with a CAC score ⬎0 [or ⱖ100]) decreased with a When data were stratified by sex, the general
higher FRS. In each CAC category, the NNS was pattern of distribution of CAC scores ⬎0 and
higher for lower than higher FRS strata (Table 2). For ⱖ100 across FRS strata remained the same, with a
example, among those with a CAC score of ⱖ100, the higher prevalence of CAC scores ⬎0 and ⱖ100
NNS was 79 for participants with an FRS of 0 to 2.5% across FRS strata (Table 3). The prevalence of
and 6 for those with an FRS ⬎10%. CAC scores ⬎0 and ⱖ100 was higher in men than

Table 2. CAC Prevalence, Amount, and NNS Compared With FRS Categories (N ⴝ 2,832)

FRS Categories

0%–2.5% 2.6%–5% 5.1%–10% >10%

CAC Score Group (n ⴝ 2,372) (n ⴝ 287) (n ⴝ 115) (n ⴝ 58) p Value
Median CAC score (IQR)* 17.9 (5.1–64.5) 16.6 (5.6–52.2) 18.9 (11.9–73.5) 58.8 (20.6–215.7)
CAC score ⬎0, % (n ⫽ 279) 7.3 20.2 19.1 44.8 ⬍0.01
NNS 13.7 5.0 5.2 2.2
CAC score ⱖ100, % (n ⫽ 51) 1.3 2.4 3.5 17.2 ⬍0.01
NNS 79 41 28.8 5.8

*Among those with a CAC score ⬎0.

IQR ⫽ interquartile range, NNS ⫽ number needed to screen (number of people who need to be screened to identify 1 person with a CAC score above a pre-specified
cut point in each Framingham risk score category); other abbreviations as in Table 1.
JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012 Okwuosa et al. 927
SEPTEMBER 2012:923–30 Yield of CAC Screening in Early Middle-Age Adults

women. Likewise, the overall median CAC scores

100
(among those with CAC) were higher in men.
90
Further stratification by race revealed that al-
though the overall prevalence of CAC was greater 80
in white compared with black participants, the overall

CAC Prevalence (%)

70
median CAC scores were higher in black than in p for difference across strata for both CAC thresholds: < 0.01
60
white participants (Table 4). As with the overall CAC > 0
distribution in Table 2, the prevalence of CAC scores 50
CAC ≥ 100
⬎0 and ⱖ100 increased with higher FRSs. 40
The observed prevalence and NNS patterns were 30
similar when CARDIA year 20 examination data
20
were analyzed in the same fashion.
10
DISCUSSION 0
0 - 2.5% 2.6 - 5% 5.1 - 10% >10%

We report the prevalence of CAC scores ⬎0 and Estimated 10-year FRS Categories
ⱖ100 relative to FRS strata in a cohort of young to
Figure 1. CAC Score Compared With FRS
early middle-age black and white men and women
without diabetes. There was significant concor- Prevalence of coronary artery calcium (CAC) scores ⬎0 and ⱖ100 compared across
10-year Framingham Risk Score (FRS) strata in the CARDIA study. There was signifi-
dance between CAC prevalence/amount and FRS cant concordance between CAC prevalence/amount and FRS such that prevalence of
such that the prevalence of CAC scores ⬎0 and CAC scores ⬎0 and ⱖ100 were low in the lower FRS strata and increased with
ⱖ100 and median CAC scores were low in the higher FRSs.
lower FRS strata and increased with higher FRSs.
Correspondingly, the NNS to detect CAC scores
⬎0 or ⱖ100 was lower with higher FRSs. Findings persons who will have events into the high-risk
were similar when stratified by sex and race. category (2). In addition, CAC is useful for guiding
Potential implications. The FRS is a useful tool for and monitoring effects of therapy and for motivat-
predicting coronary events, but fails to identify a ing patients in lifestyle and/or drug therapy for
significant number of individuals who will have cardiovascular risk factor modification (4). As such,
events (22,23). Clinical trial data showing reduced CAC testing is a topic of discussion for different
event rates due to CAC screening are lacking. consensus panels.
Nevertheless, recent data showed that compared In defining FRS thresholds for CAC screening,
with no CAC testing, randomization to CAC expert panels have generally focused on individuals
screening was associated with improved coronary 50 years of age and older and differ in their
artery disease risk factor control without increased recommendations for what constitutes a reasonable
downstream medical testing (24). In addition, FRS threshold at which to screen for CAC (1,3– 6).
CAC predicts CHD events independent of the In the current study, we attempted to determine
FRS (1,23) and appropriately reclassifies low-risk FRS thresholds at which screening for the presence

Table 3. CAC Prevalence and Amount Compared With FRS Categories, Stratified by Sex (N ⴝ 2,832)

FRS Categories

0%–2.5% 2.6%–5% 5.1%–10% >10%

CAC Score Group (n ⴝ 2,372) (n ⴝ 287) (n ⴝ 115) (n ⴝ 58) p Value
Males (n ⫽ 1,327) n ⫽ 927 n ⫽ 241 n ⫽ 106 n ⫽ 53
Median CAC score (IQR)* 20.6 (5.6–64.6) 16.6 (5.6–41.8) 18.9 (10.8–76.5) 60.6 (20.6–215.7)
CAC score ⬎0, % (n ⫽ 203) 11.9 20.8 18.9 43.4 ⬍0.01
CAC score ⱖ100, % (n ⫽ 37) 1.9 2.5 3.8 17.0
Females (n ⫽ 1,505) n ⫽ 1,445 n ⫽ 46 n⫽9 n⫽5
Median CAC score (IQR)* 15.0 (4.2–61.7) 19.1 (4.9–57.9) 30.6 (15.9–45.4) 33.0 (16.4–571.1)
CAC score ⬎0, % (n ⫽ 76) 4.4 17.4 22.2 60.0 ⬍0.01
CAC score ⱖ100, % (n ⫽ 14) 0.8 2.2 0.0 20.0

*Among those with a CAC score ⬎0.

Abbreviations as in Table 2.
928 Okwuosa et al. JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012

Yield of CAC Screening in Early Middle-Age Adults SEPTEMBER 2012:923–30

Table 4. CAC Prevalence and Amount Compared With FRS Categories, Stratified by Race (N ⴝ 2,832)

FRS Categories

0%–2.5% 2.6%–5% 5.1%–10% >10%

CAC Score Group (n ⴝ 2,372) (n ⴝ 287) (n ⴝ 115) (n ⴝ 58) p Value
White (n ⫽ 1,566) n ⫽ 1,307 n ⫽ 159 n ⫽ 64 n ⫽ 36
Median CAC score (IQR)* 16.5 (5.1–63.3) 22.4 (6.6–53.5) 16.8 (9.8–31.8) 24.1 (10.0–161.2)
CAC score ⬎0, % (n ⫽ 183) 9.3 19.5 20.3 47.2 ⬍0.01
CAC score ⱖ100, % (n ⫽ 34) 1.7 3.1 1.6 16.7
Black (n ⫽ 1,266) n ⫽ 1,065 n ⫽ 128 n ⫽ 51 n ⫽ 22
Median CAC score (IQR)* 34.6 (4.7–64.6) 13.2 (3.7–30.0) 45.4 (12.5–108.2) 76.9 (36.3–215.7)
CAC score ⬎0, % (n ⫽ 96) 4.8 21.1 17.7 40.9 ⬍0.01
CAC score ⱖ100, % (n ⫽ 17) 0.8 1.6 5.9 18.2

*Among those with a CAC score ⬎0.

Abbreviations as in Table 2.

of CAC, and especially a high CAC burden, might with 2.6% to 5% and ⬃2-fold for FRSs of 2.6% to
be useful in young to early middle-age individuals 5% versus 0 to 2.5%. Putting our findings in context,
based on distribution of CAC by FRS strata. it should be noted that in the MASS (Multicentre
Compared with the presence of any CAC, a high Aneurysm Screening Study) (28), which used abdom-
CAC burden (CAC score ⱖ100) has been associ- inal ultrasound to evaluate the benefit of screening for
ated with greater risk (⬎2-fold and as high as a abdominal aortic aneurysms, the NNS to prevent 1
7-fold increase in multivariable-adjusted relative death secondary to abdominal aortic aneurysm was
risk) for CHD events (1,6,25,26). We therefore 20.4 among those screened.
focus our discussion for the current study on deter- The prevalence and NNS data from our study
mining possible FRS screening thresholds for CAC suggest a low yield of screening for CAC scores
scores ⱖ100. ⱖ100 in those young individuals identified as being
We used the NNS as a tool to aid our prevalence at lower 10-year risk of CHD events (FRSs ⱕ10%).
data in determining potential thresholds for CAC Thus, in this population, an FRS of 10% might
screening across FRS strata. The NNS is an exten- represent a logical threshold for CAC screening in
sion of the concept of the number needed to treat younger adults. This is in agreement with some
and is typically defined as the number of people consensus guidelines (1,3) that suggest that persons
who need to be screened to prevent 1 death or 1 at intermediate 10-year risk of CHD events (FRSs
adverse event (27). As in a previous study by our of 10% to 20%) are more likely to benefit from
group (8), we defined the NNS as the number of screening for CAC to aid further risk factor inter-
people who need to be screened to detect 1 person ventions, especially in situations in which there is
with CAC above a specified cut point in each FRS uncertainty regarding the use of drug therapy.
stratum. The prevalence and NNS data from our According to the guidelines, those with an FRS
previous study (of multiethnic men and women 45 ⬎20% are considered to be at high risk of CHD
to 84 years of age) suggested a low yield of screen- events and should be appropriately managed with
ing for clinically significant levels of CAC in drug therapy and lifestyle modifications (17). Also
individuals with an FRS ⱕ5%. in support of expert panels (1,6), our study suggests
In the current study, the prevalence of CAC that decisions regarding CAC measurement should
scores ⱖ100 was low (⬍5%) among all FRS pre- be made in the context of traditional cardiovascular
dicted strata ⬍10%, and considerably higher (⬃17%) risk factors rather than in isolation. As such, data
in those with an FRS ⬎10%. Correspondingly, the from our study support the avoidance of radiation
NNS was much higher (NNS ⬎28) in participants exposure, discovery of incidental findings requiring
with an FRS of 0 to 2.5%, 2.6% to 5%, and 5.1% to follow-up computed tomography scans, as well as
10% compared with those with an FRS ⬎10% time, money, and effort spent on CAC measure-
(NNS ⫽ 6). Furthermore, the relative difference in ment for clinical guidance in young, low-risk pa-
NNS for CAC scores ⱖ100 was reasonably high tients with an FRS ⬍10%.
(5-fold) for FRSs ⬎10% versus 5.1% to 10%. This Other findings. Consistent with other studies, strat-
relative difference for adjacent FRS strata was much ification by sex and race revealed the prevalence and
less (1.4-fold) for FRSs of 5.1% to 10% compared amount of CAC to be higher in men than in
JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012 Okwuosa et al. 929
SEPTEMBER 2012:923–30 Yield of CAC Screening in Early Middle-Age Adults

women and CAC prevalence to be higher in whites U.S. young adults. For the same reason, there were
compared with blacks (29 –33). Contrary to expec- fewer participants with a CAC score of ⱖ100 in
tations, however, median CAC scores were higher each FRS category when stratified by sex and race.
in black than in white participants, likely due to the Finally, we did not separate the intermediate FRS
skewed distribution of data as a result of fewer (10% to 20%) from the high FRS (⬎20%) risk
participants in the higher FRS categories. Because groups because, of 58 individuals with an FRS
of the young age of the cohort, we did not stratify ⬎10% in this young cohort, only 6 persons had an
our data by age. Not surprisingly in this young FRS ⬎20% (3 of whom had the presence of any
cohort, cigarette smoking was the most predomi- CAC).
nant risk factor among those with CAC scores of
ⱖ100 and FRSs ⬎10%. This represents individuals
already at higher risk of CHD/cardiovascular dis- CONCLUSIONS
ease events based on FRS for whom smoking
cessation should be emphasized as a modifiable risk In this young to early middle-age nondiabetic,
factor, especially if CAC screening revealed signif- asymptomatic cohort, there was concordance be-
icant CAC burden. tween CAC prevalence/amount and FRS strata.
Study limitations. The very low number of CHD Our study suggests that in this group of relatively
events in this young cohort to date precluded young individuals, the yield of screening for high
validation of our suggested FRS cut points for CAC CAC burden (CAC score ⱖ100) among low-risk
screening using event data. Furthermore, due to the persons with an FRS of ⱕ10% is low. However,
relative youth of our cohort, we had few participants CAC testing might be considered in younger per-
with high CAC burden. As such, we used a lower sons with an FRS of ⱖ10%.
cut point for high CAC burden (CAC score ⱖ100)
Reprint requests and correspondence: Dr. Donald M.
and could not examine NNS and FRS distributions Lloyd-Jones, Department of Preventive Medicine and
relative to advanced CAC burden (CAC scores Bluhm Cardiovascular Institute, Northwestern Univer-
ⱖ300 or 400). Our study, however, is likely repre- sity, 680 North Lake Shore Drive, Suite 1400, Chicago,
sentative of the distribution of CAC burden among Illinois 60611. E-mail: [email protected].

REFERENCES Committee on Cardiovascular Imag- Coronary calcium independently

ing and Intervention, Council on Car- predicts incident premature coronary
1. Greenland P, Bonow RO, Brundage
diovascular Radiology and Interven- heart disease over measured cardio-
BH, et al. ACCF/AHA 2007 clinical
tion, and Committee on Cardiac vascular risk factors: mean three-
expert consensus document on coro-
Imaging, Council on Clinical Cardi- year outcomes in the Prospective
nary artery calcium scoring by com-
ology. Circulation 2006;114:1761–91. Army Coronary Calcium (PACC)
puted tomography in global cardiovas-
4. Greenland P, Alpert JS, Beller GA, et project. J Am Coll Cardiol 2005;46:
cular risk assessment and in evaluation
al. 2010 ACCF/AHA guideline for 807–14.
of patients with chest pain: a report of
the American College of Cardiology assessment of cardiovascular risk in 8. Okwuosa TM, Greenland P, Ning H,
Foundation Clinical Expert Consen- asymptomatic adults: a report of the et al. Distribution of coronary artery
sus Task Force (ACCF/AHA Writ- American College of Cardiology calcium scores by Framingham 10-
ing Committee to Update the 2000 Foundation/American Heart Associa- year risk strata in the MESA (Multi-
Expert Consensus Document on tion Task Force on Practice Guide- Ethnic Study of Atherosclerosis): po-
Electron Beam Computed Tomogra- lines. J Am Coll Cardiol 2010;56: tential implications for coronary risk
phy) developed in collaboration with e50 –103. assessment. J Am Coll Cardiol 2011;
the Society of Atherosclerosis Imag- 5. U.S. Preventive Services Task Force. 57:1838 – 45.
ing and Prevention and the Society Using nontraditional risk factors in 9. Nucifora G, Schuijf JD, van Werkhoven
of Cardiovascular Computed To- coronary heart disease risk assessment: JM, et al. Prevalence of coronary ar-
mography. J Am Coll Cardiol 2007; U.S. Preventive Services Task Force tery disease across the Framingham
49:378 – 402. recommendation statement. Ann In- risk categories: coronary artery cal-
2. Polonsky TS, McClelland RL, Jor- tern Med 2009;151:474 – 82. cium scoring and MSCT coronary
gensen NW, et al. Coronary artery 6. Naghavi M, Falk E, Hecht HS, et al. angiography. J Nucl Cardiol 2009;
calcium score and risk classification From vulnerable plaque to vulnerable 16:368 –75.
for coronary heart disease prediction. patient–Part III: Executive summary 10. Desai MY, Nasir K, Braunstein JB, et
JAMA 2010;303:1610 – 6. of the Screening for Heart Attack al. Underlying risk factors incremen-
3. Budoff MJ, Achenbach S, Blumenthal Prevention and Education (SHAPE) tally add to the standard risk estimate
RS, et al. Assessment of coronary Task Force report. Am J Cardiol in detecting subclinical atherosclerosis
artery disease by cardiac computed 2006;98:2H–15H. in low- and intermediate-risk middle-
tomography: a scientific statement 7. Taylor AJ, Bindeman J, Feuerstein I, aged asymptomatic individuals. Am
from the American Heart Association Cao F, Brazaitis M, O’Malley PG. Heart J 2004;148:871–7.
930 Okwuosa et al. JACC: CARDIOVASCULAR IMAGING, VOL. 5, NO. 9, 2012

Yield of CAC Screening in Early Middle-Age Adults SEPTEMBER 2012:923–30

11. Achenbach S, Nomayo A, Couturier Obes Relat Metab Disord 2000;24: the Multi-Ethnic Study of Athero-
G, et al. Relation between coronary 1475– 87. sclerosis (MESA). Arch Intern Med
calcium and 10-year risk scores in 19. Friedewald WT, Levy RI, Fredrickson 2007;167:2437– 42.
primary prevention patients. Am J DS. Estimation of the concentration of 27. Rembold CM. Number needed to
Cardiol 2003;92:1471–5. low-density lipoprotein cholesterol in screen: development of a statistic for
12. Hoffmann U, Massaro JM, Fox CS, plasma, without use of the preparative disease screening. BMJ 1998;317:
Manders E, O’Donnell CJ. Defining ultracentrifuge. Clin Chem 1972;18: 307–12.
normal distributions of coronary ar- 499 –502. 28. Ashton HA, Buxton MJ, Day NE, et
tery calcium in women and men 20. Loria CM, Liu K, Lewis CE, et al. al. The Multicentre Aneurysm
(from the Framingham Heart Study). Early adult risk factor levels and sub- Screening Study (MASS) into the ef-
Am J Cardiol 2008;102:1136 – 41, sequent coronary artery calcification: fect of abdominal aortic aneurysm
1141.e1. the CARDIA study. J Am Coll Car- screening on mortality in men: a ran-
13. Sung J, Lim SJ, Choe Y, et al. Com- diol 2007;49:2013–20. domised controlled trial. Lancet 2002;
parison of the coronary calcium score 21. Wilson PW, D’Agostino RB, Levy D, 360:1531–9.
with the estimated coronary risk. Belanger AM, Silbershatz H, Kannel 29. Bild DE, Detrano R, Peterson D, et
Coron Artery Dis 2008;19:475–9. WB. Prediction of coronary heart dis- al. Ethnic differences in coronary cal-
14. Greenland P, LaBree L, Azen SP, Do- ease using risk factor categories. Cir- cification: the Multi-Ethnic Study of
herty TM, Detrano RC. Coronary ar- culation 1998;97:1837– 47. Atherosclerosis (MESA). Circulation
tery calcium score combined with 22. Brindle P, Beswick A, Fahey T, Ebra- 2005;111:1313–20.
Framingham score for risk prediction in him S. Accuracy and impact of risk 30. Doherty TM, Tang W, Detrano RC.
asymptomatic individuals [published assessment in the primary prevention Racial differences in the significance
correction appears in JAMA 2004;291: of cardiovascular disease: a systematic of coronary calcium in asymptomatic
563]. JAMA 2004;291:210 –5. review. Heart 2006;92:1752–9. black and white subjects with coronary
15. Pletcher MJ, Tice JA, Pignone M, 23. Murphy TP, Dhangana R, Pencina risk factors. J Am Coll Cardiol 1999;
McCulloch C, Callister TQ, Browner MJ, Zafar AM, D’Agostino RB. Per- 34:787–94.
WS. What does my patient’s coronary formance of current guidelines for 31. Lee TC, O’Malley PG, Feuerstein I,
artery calcium score mean? Combin- coronary heart disease prevention: op- Taylor AJ. The prevalence and sever-
timal use of the Framingham-based ity of coronary artery calcification on
ing information from the coronary
risk assessment. Atherosclerosis 2011; coronary artery computed tomography
artery calcium score with information
216:452–7. in black and white subjects. J Am Coll
from conventional risk factors to esti-
mate coronary heart disease risk. 24. Rozanski A, Gransar H, Shaw LJ, et Cardiol 2003;41:39 – 44.
BMC Med 2004;2:31. al. Impact of coronary artery calcium 32. Newman AB, Naydeck BL, Whittle J,
scanning on coronary risk factors and Sutton-Tyrrell K, Edmundowicz D,
16. Friedman GD, Cutter GR, Donahue downstream testing the EISNER
RP, et al. CARDIA: study design, Kuller LH. Racial differences in cor-
(Early Identification of Subclinical
recruitment, and some characteristics onary artery calcification in older
Atherosclerosis by Noninvasive Imag-
of the examined subjects. J Clin Epi- adults. Arterioscler Thromb Vasc Biol
ing Research) prospective randomized
demiol 1988;41:1105–16. 2002;22:424 –30.
trial. J Am Coll Cardiol 2011;57:
17. Executive Summary of The Third Re- 1622–32. 33. Tang W, Detrano RC, Brezden OS,
port of The National Cholesterol Ed- et al. Racial differences in coronary
25. Vliegenthart R, Oudkerk M, Song B,
ucation Program (NCEP) Expert calcium prevalence among high-risk
van der Kuip DA, Hofman A, Witte-
Panel on Detection, Evaluation, And adults. Am J Cardiol 1995;75:
man JC. Coronary calcification detected
Treatment of High Blood Cholesterol by electron-beam computed tomogra- 1088 –91.
In Adults (Adult Treatment Panel phy and myocardial infarction. The
III). JAMA 2001;285:2486 –97. Rotterdam Coronary Calcification
18. Schmitz KH, Jacobs DR Jr., Leon Study. Eur Heart J 2002;23:1596 – 603. Key Words: coronary artery
AS, Schreiner PJ, Sternfeld B. Phys- 26. Lakoski SG, Greenland P, Wong calcium y coronary heart disease
ical activity and body weight: asso- ND, et al. Coronary artery calcium y Framingham Risk Score y
ciations over ten years in the CAR- scores and risk for cardiovascular
DIA study. Coronary Artery Risk events in women classified as “low number needed to screen y risk
Development in Young Adults. Int J risk” based on Framingham risk score: factors.