356
ABA Medica1IScientific Statement
Science Advisory
An Updated Coronary Risk Profile
A Statement for Health Professionals
Keaven M. Anderson, PhD; Peter W.F. Wilson, MD;
Patricia M. Odell, PhD; and William B. Kannel, MD, MPH
C oronary heart disease (CHD) continues to be
the cause of the greatest number of deaths
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017
among adult Americans.' Although there is
a downward trend in cardiovascular mortality rates,
morbidity and mortality rates remain high and are of
great concern to clinicians and health officials. Using
a simple worksheet, a patient's 5- and 10-year CHD
risks can be estimated. The components of the profile
were selected because they are objective and strongly
and independently related to CHD and because they
can be measured through simple office procedures
and laboratory results.
In the past, investigators with the Framingham
Heart Study developed CHD risk equations for use
by clinicians in predicting the development of coro-
nary disease in individuals free of disease.2 Early
efforts reflected the experience of study investigators
from 1950 to the mid-1960s.34 In addition to age and
gender, risk factors included systolic blood pressure
(SBP), serum cholesterol, cigarette smoking, glucose
intolerance, and left ventricular hypertrophy (LVH).
A handbook5 containing CHD risk tables based on
Framingham equations6 was published in 1973. An
even simpler approximation of the equations, by
which CHD risk could be estimated by following
instructions on a pocket-sized card, soon followed.7
The equations presented here have several advan-
tages over previous versions. The data base from
which they are derived is larger and more recent. In
particular, more data for individuals older than 60
years are available. In addition, the influence of high
density lipoprotein (HDL) cholesterol, which has
been measured in the Framingham Heart Study since
1968, is reflected in these equations. Measurement of
the ratio of total cholesterol to HDL cholesterol has
been found to be superior to measurement of serum
cholesterol as a predictor of CHD.8
At the baseline examination for this study (1968-
1975), all members of the original Framingham co-
hort were more than 50 years old. To provide current
estimates of CHD risk over a large age range, data
Requests for reprints should be sent to the Office of Scientific
Affairs, American Heart Association, 7320 Greenville Avenue,
Dallas, TX 75231.
from the original cohort have been combined with
data from the second-generation study population,
the Framingham Offspring Cohort, for which 12
years of follow-up have recently been completed.
Together, these groups span the ages of 12-82 years;
however, only persons 30-74 years old have been
included in this study. Those 75 years of age or older
were excluded because of possible differences in risk
factors in this older group and its potentially large
influence in the algorithm determination. Tables 1-4
detail the crude incidence rates of CHD and the
distributions of risk factors among the study popula-
tion by age. From the experience of this group during
a 12-year period, estimates of CHD risk have been
produced that reflect the approximate combined
impacts of total and HDL cholesterol, SBP or dia-
stolic blood pressure (DBP), cigarette smoking, dia-
betes mellitus, and LVH as measured by electrocar-
diography (ECG-LVH).
The derivation and uses of the worksheet are
detailed in the remainder of this article.
Methods
Population and Risk Factors
Every member of the original and offspring cohorts
who met three basic criteria were included in the study.
Requirements for inclusion were 1) age 30-74 years at
the time of the baseline examination; 2) measurements
available for SBP and DBP, cigarette smoking status,
total and HDL cholesterol, and diagnoses (yes or no) of
diabetes and ECG-LVH (when information on diabe-
tes or LVH was not available, diagnoses were pre-
sumed to be negative); and 3) freedom from cardiovas-
cular disease (stroke, transient ischemia, CHD
[includes angina pectoris, coronary insufficiency (unsta-
ble angina), myocardial infarction, and sudden death],
congestive heart failure, and intermittent claudication)
until time of risk factor measurement.
Definitions of risk factors and end points are those
considered standard in the Framingham study.9 In
the original cohort, diabetes was diagnosed if a casual
whole blood glucose measurement was 150 mg/dl or
above or the individual was being treated with insulin
or oral hypoglycemics. In the offspring study, a more
 Anderson et al Updated Coronary Risk Profile 357

TABLE 1. Prevalence of Dichotomous Risk Factors and Crude Coronary Heart Disease Incidence by Age for Men
Age (yr)
30-39 40-49 50-59 60-69 70-74 30-74
CHD incidence (n) (%) 41 (5) 74 (11) 143 (20) 98 (29) 29 (26) 385 (15)
Cigarette smoking (n) (%) 340 (45) 284 (43) 301 (42) 102 (30) 28 (25) 1,055 (41)
Diabetes (n) (%) 11 (1) 23 (4) 69 (10) 63 (19) 17 (15) 183 (7)
ECG-LVH (n) (%) 1 (0.1) 4 (0.6) 2 (0.3) 17 (5) 4 (4) 28 (1)
Total (n) 759 655 725 340 111 2,590
CHD, coronary heart disease; ECG-LVH, left ventricular hypertrophy by electrocardiography.

recent definition, requiring treatment or a fasting Let T denote the time until CHD from the begin-
glucose level of 140 mg/dl or above from plasma ning of follow-up, ,u a location parameter, and a o-

measurement, was used. Measurements of risk fac-
tors for the original cohort were taken from the first
examination cycle in which HDL cholesterol levels
were measured. In most cases (87.7%), this was
examination 11 (1968-1971); for some cohort mem-
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017
scale parameter for log(T). Throughout, log( ) de-
notes the natural logarithm function. Assume that ,u
and or depend on risk factors, as will be described,
and that

bers, it was examination 10 or 12. Follow-up was log (T)-4

performed through the 17th examination cycle, a
span of approximately 12 years. For the offspring
cohort, risk factor measurements were from the first
examination cycle (1971-1975), whereas follow-up
was performed through the third examination cycle,
approximately 12 years later. The study included
5,573 persons (2,983 women and 2,590 men).
Risk factors are age (years), female (1, woman; 0,
man), SBP [average of two office measurements
(mm Hg)], DBP [average of two office measurements
(mm Hg)], cholesterol [total serum cholesterol mea-
sured by the Abell-Kendall method (mg/dl)], HDL
cholesterol [determined after heparin-manganese
precipitation (mg/dl)], smoking (1, cigarette smoking
or quit within past year; 0, otherwise), diabetes [1,
diabetes; 0, otherwise (conservative definition is
treatment with insulin or oral agents or having a
fasting glucose of 140 mg/dl or above10)], and ECG-
LVH (1, definite; 0, otherwise).
cr
follows an extreme value distribution. This implies
that T follows the Weibull distribution, which is often
used in analyzing studies of times until event.
The location parameter ,u is assumed to be a sum
of the products of the risk factors multiplied by their
corresponding coefficients; for example,
g =,f30±+,1 xfemale+832xlog (age)+f33xlog (SBP)....
The exact form taken by this function is presented in
"Results." Next, assume that
log (u)=tJO+l1X/g
This functional form for cr, as well as others, is
discussed in detail elsewhere13 and provides a highly
statistically significant improvement in model fit over
the standard model in which is held constant. Thus,
o-

the unknown parameters to be estimated are PO, ....

Statistical Modeling
A parametric regression model was used for risk
estimation. Like the logistic model previously
used,4,5,11 it provides a simple formula for estimating
probabilities of disease given risk factor levels. It has
the additional advantage of allowing computation for
variable durations of follow-up. The standard accel-
erated failure time model12 has been used with two
variations, which are described below.
Pk, 00, and 01.
The models are estimated by the maximum likeli-
hood method, which is implemented with a special-
ized computer program written by one of the authors.
This type of model can also be estimated with PROC
NLIN from SAS Institute, Cary, N.C. To reduce the
apparently undue influence of the earliest CHD
events, those events occurring during the first 4 years
of follow-up are coded as such rather than as occur-

TABLE 2. Prevalence of Dichotomous Risk Factors and Crude Coronary Heart Disease Incidence by Age for Women
Age (yr)
30-39 40-49 50-59 60-69 70-74 30-74
CHD incidence (n) (%) 4 (1) 32 (5) 105 (12) 68 (15) 32 (20) 241 (8)
Cigarette smoking (n) (%) 357 (45) 284 (41) 391 (45) 101 (22) 26 (16) 1,159 (39)
Diabetes (n) (%) 4 (0.5) 13 (2) 58 (7) 55 (12) 24 (15) 154 (5)
ECG-LVH (n) (%) 2 (0.3) 0 (0) 3 (0.4) 3 (0.7) 7 (4) 15 (0.5)
Total (n) 798 693 867 462 163 2,983
CHD, coronary heart disease; ECG-LVH, left ventricular hypertrophy by electrocardiography.
 358 AHA Statement Circulation Vol 83, No 1, January 1991

TABLE 3. Distribution of Continuous Risk Factors by Age for Men

Age (yr)
Measurement 30-39 40-49 50-59 60-69 70-74 30-74
Total cholesterol
5% percentile 144 161 161 160 149 153
Median 197 213 214 216 208 210
95% percentile 264 279 295 292 281 283
HDL cholesterol
5% percentile 27 29 28 29 31 28
Median 42 42 44 44 43 43
95% percentile 64 66 68 72 66 67
Total/HDL cholesterol
5% percentile 2.8 2.9 2.8 3.0 3.0 2.9
Median 4.6 5.0 4.9 4.9 4.8 4.8
95% percentile 7.9 8.1 8.3 7.8 7.6 8.0
SBP (mm Hg)
5% percentile 107 109 110 116 112 109
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017

Median 122 125 130 141 142 128

95% percentile 148 157 173 181 181 168
DBP (mm Hg)
5% percentile 68 69 70 69 65 69
Median 81 83 84 84 82 82
95% percentile 98 103 105 104 96 102
Total points
5% percentile -2 5 11 15 19 1
Median 6 13 18 23 24 15
95% percentile 15 22 26 32 32 27
Total 759 655 725 340 111 2,590
HDL, high density lipoprotein; SBP, systolic blood pressure; DBP, diastolic blood pressure.

ring at the exact time of onset. This methodology is tions of some components of the equation are given
the subject of a separate report.14 after the calculations.
Results Systolic Blood Pressure Equation
Results are presented in two forms; the first is an There are some differences in equation calcula-
updated version of the 1973 report equations6 that tions for men and women, but both calculations begin
incorporates new data on HDL cholesterol with the in the same way. Compute an interim number a that
present study data described above. Then, an easy- is based on risk factor measurements.
to-calculate point-scoring algorithm that is based on a=11.1122-0.9119xlog (SBP)-0.2767x
one of the equations is given. It is hoped that the
algorithm will be easy to use. Predicted risk may be smoking-0.7181 x log (cholesterol/HDL) -
calculated with SBP or DBP. Because of the high 0.5865 x ECG-LVH (1)
correlation between these two measurements, both
cannot be included; the redundancy leads to difficulty The next step, computing a second interim value m, is
in interpretation. Although in general results are different for men and women. For men, compute
similar, separate equations are given for SBP and m =a - 1.4792x log (age) -0.1759x diabetes (2a)
DBP to accommodate strong user preferences. The
equation incorporating SBP is recommended be- For women, compute
cause SBP is more accurately determined, has a m=a-5.8549+1.8515x
wider range of values, and is a stronger predictor of
CHD, particularly in the elderly. [log (age/74)]2-0.3758 x diabetes (2b)
Framingham Equations Next, for both sexes, compute
Computation with the estimated equation is de- ttu=4.4181 +m (3)
scribed below. In this section, SBP is used; the
equation incorporating DBP is given next. Explana- o=exp (-0.3155-0.2784xm) (4)
 Anderson et al Updated Coronary Risk Profile 359

TABLE 4. Distribution of Continuous Risk Factors by Age for Women

Age (yr)
Measurement 30-39 40-49 50-59 60-69 70-74 30-74
Total cholesterol
5% percentile 141 152 173 179 183 151
Median 181 200 229 245 246 212
95% percentile 244 268 306 320 305 295
HDL cholesterol
5% percentile 36 35 35 36 35 35
Median 55 56 58 56 54 56
95% percentile 81 86 88 87 85 86
Total/HDL cholesterol
5% percentile 2.2 2.3 2.4 2.5 2.7 2.3
Median 3.2 3.5 3.9 4.3 4.5 3.7
95% percentile 5.6 6.2 6.9 7.3 7.2 6.7
SBP (mm Hg)
5% percentile 97 100 105 112 118 100
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017

Median 114 119 130 142 150 123

95% percentile 137 153 170 178 186 168
DBP (mm Hg)
5% percentile 61 64 65 64 64 63
Median 73 78 81 82 82 79
95% percentile 89 97 100 100 102 98
Total points
5% percentile -16 -3 4 8 10 -12
Median -6 5 12 16 18 8
95% percentile 5 15 22 26 27 22
Total 798 693 867 462 163 2,983
HDL, high density lipoprotein; SBP, systolic blood pressure; DBP, diastolic blood pressure.

Finally, choose the number of years for which you ,u = -0.4972+4.4181 =3.921
wish to predict (from 4 to 12) and call it t. Compute
log (t)-p. o-=exp [-0.3155-0.2784x(-0.4972)]=0.8377
u= (5) We let t be 10 years and compute for a man

u= log (10)-3.685
The predicted probability for t is
p=1-exp (-eu) (6)
0.894*61.546
As an example, consider a 55-year-old individual p=1-exp (-e-1546)=0.192
with SBP of 130 mm Hg, total cholesterol of 240
mg/dl, and HDL cholesterol of 45 mg/dl who smokes For a woman, compute
cigarettes. We assume that neither diabetes nor
ECG-LVH has been diagnosed. First, we compute log (10)-3.921
0.8377
a=11.1122-0.9119xlog (130)-0.2767-
0.7181 xlog (240/45)=5.1947 p=1-exp (-e- 1.932).=0.135
For a man, then compute Diastolic Blood Pressure Equation
m=5.1947-1.4792xlog (55)= -0.7329 The equations incorporating DBP instead of SBP
are precisely analogous to those given above.
,u= -0.7329+4.4181=3.685
a = 11.0938-0.8670 x log (DBP) -0.2789 x smoking-
o-=exp [-0.3155-0.2784x(-0.7329)]=0.894
For a woman, compute 0.7142xlog (cholesterol/HDL)-
m=5.1947-5.8549+1.8515x [log (55/74)]2=-0.4972 0.7195 x ECG-LVH (7)
 360 AHA Statement Circulation Vol 83, No 1, January 1991
Again, the next step is different for men and women.
For men, compute
m-=a-1.6346xlog (age)-0.2082xdiabetes (8a)
For women, compute
m =a -6.5306+2.1059 x [log (age/74)]2-
0.4055 x diabetes (8b)
Next, compute
,u=4.4284+m (9)
cr=exp (-0.3171-0.2825 xm) (10)
Proceed as with the SBP equation to compute
predicted probabilities. For example, if DBP is 90
mm Hg, p=0.22 for a man and p=0.16 for a woman.
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017
Model Selection
There are several points to consider in the deriva-
tions of the models. First, the natural logarithms of
continuous covariates were used rather than actual
values. Likelihood analysis indicates that such use
improves the fit of the model; one reason is that
extremely large covariate measurements now receive
less emphasis, allowing risk to change more slowly for
very large values than for the range of the bulk of the
data. This explains the uneven covariate intervals in
the risk prediction worksheet.
Second, the addition of a quadratic term for age
for women produces a significantly improved model
fit. This term accommodates a rapid increase in risk
at younger ages and little change at older ages. The
maximum age effect in women was arbitrarily forced
to occur at the oldest age in the population; other-
wise, this would have occurred at a slightly younger
age. This change is not statistically significant. Be-
cause of the quadratic form of the function, extrap-
olation to older and younger age groups is particu-
larly questionable.
Third, the ratio of total cholesterol to HDL cho-
lesterol was used because no improved fit was found
when the covariates were used separately. Note that
the equations do not contain a cholesterol-age inter-
action term as did some previous prediction mod-
els.1' When serum cholesterol-age and HDL choles-
terol-age interaction terms were added to the model,
their contributions were negligible. Thus, it appears
that including the ratio of total cholesterol to HDL
cholesterol in the equations eliminates the need for
such interaction terms.
Fourth, except for differences in aging effects, no
significantly different effects were found for men and
women. A separate coefficient for diabetes was fit for
women based on significant differences between men
and women in previous studies.15 Although female
smokers were not found to be at increased risk in
some previous Framingham reports,2 the offspring
cohort provides evidence of such a relation. Thus, no
distinction has been made between smoking effects
for men and those for women. Using the number of
cigarettes rather than a simple yes-or-no code for
cigarette smoking did not provide an improvement in
model fit. This may be largely due to the fact that
older smokers in this study reported smoking fewer
cigarettes than did younger smokers. Presumably,
this confounded possible dose effects with age
effects.
Finally, the estimated effect of ECG-LVH is very
large but has a large standard error because of the
small prevalence of the condition at baseline.
Point Score Algorithm
Equations 1-6 were used to devise a worksheet
that enables users to estimate their CHD risk by
assigning a point score to each risk factor (Table 5).
Clinicians or patients can insert the appropriate
points and add as indicated to obtain a good approx-
imation of CHD risk during a 5- or 10-year period.
For comparison, average risk values, by age and sex,
for the Framingham population are also given.
If the values used in the sample computations are
used in the worksheet, results are as given in Table 6.
These results are a close approximation of those
obtained with Equations 1-6 (19.2% for a man and
13.5% for a woman).
Discussion
The CHD risk equations and point score predic-
tion probability algorithm attempt to elucidate the
multifactorial etiology of CHD and produce possible
comparisons and interpretations for clinicians and
their patients at risk. The point-scoring technique
allows estimation of an individual's 5- and 10-year
risks of CHD, whereas the equations may be used for
4- to 12-year estimations. These calculations are
based on the Framingham experience for individuals
free of cardiovascular disease at baseline, and such
calculations are not appropriate for individuals with
coronary disease.
Generalization of the Framingham equations to
the population at large is always a matter of concern
and should be done cautiously. However, it has been
demonstrated repeatedly that the Framingham risk
model is effective in predicting heart disease in other
large population samples in the United States.16-19
For example, the 1973 equations were used in the
Multiple Risk Factor Intervention Trial to predict
the number of cases of CHD to be expected during
the course of the trial.20 The trial was administered
during the early 1970s, when there was a sharp
decrease in CHD mortality rates in the United States
and extensive intervention against risk factors was
being practiced. Hence, as could be anticipated, the
estimates were high for both the usual-care and the
special-intervention groups. But they were systemat-
ically high, and the relative weightings of risk factors
distinguished low- from high-risk individuals.
Nonetheless, cautions should be observed when
using the equations or worksheet. First, the equa-
tions can, of course, be used only with all risk factors
 Anderson et al Updated Coronary Risk Profile 361

TABLE 5. Framingham Heart Study Coronary Heart Disease Risk Prediction Chart
1. Find points for eachrisk factor
Age (if female) (yr) Age (if male) (yr) HDL cholesterol
Age Points Age Points Age Points Age Points HDL Points HDL Points
30 -12 41 1 30 -2 48-49 9 25-26 7 67-73 -4
31 -11 42-43 2 31 -1 50-51 10 27-29 6 74-80 -5
32 -9 44 3 32-33 0 52-54 11 30-32 5 81-87 -6
33 -8 45-46 4 34 1 55-56 12 33-35 4 88-96 -7
34 -6 47-48 5 35-36 2 57-59 13 36-38 3
35 -5 49-50 6 37-38 3 60-61 14 39-42 2
36 -4 51-52 7 39 4 62-64 15 43-46 1
37 -3 53-55 8 40-41 5 65-67 16 47-50 0
38 -2 56-60 9 42-43 6 68-70 17 51-55 -1
39 -1 61-67 10 44-45 7 71-73 18 56-60 -2
40 0 68-74 11 46-47 8 74 19 61-66 -3
Total cholesterol (mg/dl) Systolic blood pressure (mm Hg)
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017

Points
Chol Points Chol Points SBP Points SBP Points Other factors Yes No
139-151 -3 220-239 2 98-104 -2 150-160 4 Cigarette smoking 4 0
152-166 -2 240-262 3 105-112 -1 161-172 5 Diabetes
167-182 -1 263-288 4 113-120 0 173-185 6 Male 3 0
183-199 0 289-315 5 121-129 1 Female 6 0
200-219 1 316-330 6 130-139 2 ECG-LVH 9 0
140-149 3
2. Addpoints for all risk factors
+ + + + + +=
(Age) (Total chol) (HDL) (SBP) (Smoking) (Diabetes) (ECG-LVH) (Total)
Note: Minus points subtract from total.
3. Look up risk corresponding to point total
Probability (%) Probability (%) Probability (%) Probability (%)
Points 5 yr 10 yr Points 5 yr 10 yr Points 5 yr 10 yr Points 5 yr 10 yr
.1 <1 <2 9 2 5 17 6 13 25 14 27
2 1 2 10 2 6 18 7 14 26 16 29
3 1 2 11 3 6 19 8 16 27 17 31
4 1 2 12 3 7 20 8 18 28 19 33
5 1 3 13 3 8 21 9 19 29 20 36
6 1 3 14 4 9 22 11 21 30 22 38
7 1 4 15 5 10 23 12 23 31 24 40
8 2 4 16 5 12 24 13 25 32 25 42
4. Compare with average 10-year risk
Age Probability (%) Age Probability (%) Age Probability (%)
(yr) Women Men (yr) Women Men (yr) Women Men
30-34 <1 3 45-49 5 10 60-64 13 21
35-39 <1 5 50-54 8 14 65-69 9 30
40-44 2 6 55-59 12 16 70-74 12 24
HDL, high density lipoprotein; SBP, systolic blood pressure; ECG-LVH, left ventricular hypertrophy by electrocardiography.

measured. These were chosen because they can be
measured objectively and because they make inde-
pendent contributions to the risk equations. Other
important risk influences have not been included in
the equations, but they should not be underesti-
mated. For example, heredity is a factor of major
importance, but information on this subject is dif-
ficult to quantify or even obtain accurately. Individ-
uals with family histories of heart disease should view
increased risk factors with particular concern and
deal with them more vigorously than should persons
without such backgrounds. Obesity is another exam-
 362 AHA Statement Circulation Vol 83, No 1, January 1991
TABLE 6. Sample Worksheet Results
Points
Risk factor Level Man Woman
Age 55 12 8 Dallas, Tex, 1973
HDL cholesterol 45 1 1
Total cholesterol 240 3 3
SBP 130 2 2
Smoking 1 4 4 WB, Gordon T (eds): The Framingham Study: An Epidemio-
Diabetes 0 0 0 Government Printing Office No. 426-1301/1345, Washington,
ECG-LVH 0 0 0 D.C., 1971
Total 22 18
10-year risk (%) 21 14
HDL, high density lipoprotein; SBP, systolic blood pressure;
ECG-LVH, left ventricular hypertrophy measured by electrocar-
diography.
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017
ple of a significant risk factor in long-term predic-
tions21; however, in shorter-term studies such as this
one, its effects tend to be mediated by other risk
factors. Hence, it is not included in the equations.
Second, the predictions may not be appropriate for
individuals with extremely elevated risk factors such
as malignant hypertension and severe diabetes mel-
litus or extremely high cholesterol and HDL choles-
terol levels that place them in the top or bottom few
percentiles of the distributions.
Third, the equations may not be directly applicable
to populations with very low CHD incidence rates.19
Although these equations are considered more accu-
rate than earlier versions when incidence rates are
low, this has not been tested extensively. Hence, the
equations may be inappropriate for use with popula-
tions from countries or ethnic groups that have CHD
incidence rates that are much lower or higher than
the range presented here.18
When generalization seems reasonable, estimation
of CHD risk can be useful in projecting patient prog-
ress in clinics at which preventive cardiology is the goal,
such as those concentrating on lipid or blood pressure
influence. Risk factor scores and calculations can be
discussed with patients and provide a framework for
intervention. Clinicians can predict possible rewards in
the form of improved risk profiles for patients who
make the appropriate although often difficult changes
in smoking, eating, and exercise habits.
References
1. National Center for Health Statistics: Births, marriages,
divorces, and deaths for May 1989. Monthly Vital Statistics
Report, vol 38, no. 5. US Dept of Health and Human Services
publication No. (PHS) 89-11200. Hyattsville, Md, 1989
2. Kannel WB, McGee D, Gordon T: A general cardiovascular
risk profile: The Framingham Study. Am J Cardiol 1976;38:
46-51
3. Truett J, Cornfield J, Kannel W: A multivariate analysis of the
risk of coronary heart disease in Framingham. J Chronic Dis
1967;20:511-524
4. Walker SH, Duncan DB: Estimation of the probability of an
event as a function of several independent variables. Biomet-
rika 1967;54:167-179
5. American Heart Association: Coronary Risk Handbook: Esti-
mating the Risk of Coronary Heart Disease in Daily Practice.
6. Gordon T, Sorlie P, Kannel WB: Coronary heart disease,
atherothrombotic brain infarction, intermittent claudica-
tion-A multivariate analysis of some factors related to their
incidence: Framingham Study, 16-year followup, in Kannel
logical Investigation of Cardiovascular Disease, section 27. US
7. Brittain E: Probability of coronary heart disease developing.
WestJ Med 1982;136:86-89
8. Castelli WP, Garrison RJ, Wilson PW, Abbott RD, Kalous-
dian S, Kannel WB: Incidence of coronary heart disease and
lipoprotein cholesterol levels: The Framingham Study. JAMA
1986;256:2835-2838
9. Shurtleff D: Some characteristics related to the incidence of
cardiovascular disease and death: Framingham study, 16-year
follow-up, in Kannel WB, Gordon T (eds): The Framingham
Study: An Epidemiological Investigation of Cardiovascular Dis-
ease, section 26. US Government Printing Office No. 0-414-
297, Washington, D.C.,1971
10. National Diabetes Data Group: Classification and diagnosis of
diabetes mellitus and other categories of glucose intolerance.
Diabetes 1983;28:1039-1057
11. Abbott RD, McGee D: The probability of developing certain
cardiovascular diseases in eight years at specified values of
some characteristics: The Framingham Study, in Kannel WB,
Wolf PA, Garrison RJ (eds): The Framingham Study: An
Epidemiological Investigation of Cardiovascular Disease, section
37. National Technical Information Service No. PB 87-221644/
A5. Springfield, Va, 1987
12. Kalbfleisch JD, Prentice RL: The Statistical Analysis of Failure
Time Data. New York, John Wiley & Sons, Inc, 1980
13. Carroll RJ, Ruppert D: Transformation and Weighting in
Regression. New York, Chapman and Hall, 1988
14. Anderson KM: A non-proportional hazards Weibull acceler-
ated failure time model. Biometrics 1991 (in press)
15. Kannel WB, McGee DL: Diabetes and cardiovascular disease:
The Framingham study. JAAL4 1979;241:2035-2038
16. Brand RJ, Rosenman RH, Sholtz RI, Friedman M: Multivari-
ate prediction of coronary heart disease in the Western
Collaborative Group Study compared to the findings of the
Framingham Study. Circulation 1976;53:348-355
17. Leaverton PE, Sorlie PD, Kleinman JC, Dannenberg AL,
Ingster-Moore L, Kannel WB, Cornoni-Huntley JC: Repre-
sentativeness of the Framingham risk model for coronary
heart disease mortality: A comparison with a national cohort
study. J Chronic Dis 1987;40:775-784
18. McGee D, Gordon T: The results of the Framingham study
applied to four other US-based epidemiologic studies of
cardiovascular disease, in Kannel WB, Gordon T (eds): The
Framingham Study: An Epidemiological Investigation of Cardio-
vascular Disease, section 31. US Dept of Health, Education,
and Welfare publication No. 76-1083. Bethesda, Md, US
Government Printing Office, 1976
19. Pooling Project Research Group: Relationship of blood pres-
sure, serum cholesterol, smoking habit, relative weight and
ECG abnormalities to incidence of major coronary events:
Final report of the pooling project. J Chronic Dis 1978;31:
201-306
20. The Multiple Risk Factor Intervention Trial Group: Statistical
design considerations in the NHLI Multiple Risk Factor
Intervention Trial (MRFIT). J Chronic Dis 1977;30:261-275
21. Hubert HB, Feinleib M, McNamara PM, Castelli WP: Obesity
as an independent risk factor for cardiovascular disease: A
26-year follow-up of participants in the Framingham Heart
Study. Circulation 1983;67:968-977
 An updated coronary risk profile. A statement for health professionals.
K M Anderson, P W Wilson, P M Odell and W B Kannel

Circulation. 1991;83:356-362
doi: 10.1161/01.CIR.83.1.356
Downloaded from http://circ.ahajournals.org/ by guest on February 16, 2017

Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1991 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/83/1/356.citation

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Circulation is online at:

http://circ.ahajournals.org//subscriptions/