ANNALS OF FAMILY MEDICINE
VOL. 5, NO. 5
RISK OF DIABETES IN YOUNG ADULTS
hood of having undiagnosed diabetes.
There are few
measures that assess the risk of developing diabetes.
Some risk scores for the likelihood of having undi-
agnosed diabetes have been tested in populations
other than the ones in which they were created and
have unfortunately not worked as well.
ing the importance of identifying individuals at risk
for developing diabetes, a strategy for assessing risk of
developing diabetes in young adults has many beneﬁ ts,
including targeted interventions for young adults at
high risk. Thus, the purpose of this study was to evalu-
ate how well a risk score for developing diabetes that
was created with a middle-aged population performs in
a cohort of young adults.
This study is based on an analysis of the Coronary
Artery Risk Development in Young Adults (CARDIA),
a population-based observational study of participants
aged 18 to 30 years recruited in 1985-1986. Partici-
pants were recruited in 4 communities: Birmingham,
Alabama; Chicago, Illinois; Minneapolis, Minnesota;
and Oakland, California. Recruitment was stratiﬁ ed
by race (black and white), age (18 to 24 years, and 25
to 30 years), and education (less than high school,
and high school or more). Second (1987-1988), third
(19 9 0 -19 91), fo u r th (19 92-19 93), ﬁ fth (19 95-19 96), and
sixth examinations (2000-2001) have been completed
in the cohort. The public use data set used for this
study, however, only includes information from the
ﬁ rst 5 examinations.
For the progression to diabetes analyses, all individu-
als had no indication of diabetes at baseline. This cohort
was comprised of 2,543 persons. A total of 100 persons
out of 2,543 developed diabetes within the 10 years.
Diabetes was deﬁ ned by self-report in response to
the question, “Has a doctor or nurse ever said you
had diabetes (high sugar in blood or urine)?” and by
a fasting plasma glucose of ≥126 mg/dL. Although
this biomarker deﬁ nition deviates from the deﬁ nition
in place at baseline (≥140 mg/dL), we believed that it
was important to use a current deﬁ nition of diabetes,
whether diagnosed or not. This deﬁ nition is also con-
sistent with the diabetes risk score used in this study.
Development of diabetes was deﬁ ned as having diabe-
tes at year 10 (examination 5).
Diabetes Risk Score
The risk score used in this study predicts the develop-
ment of diabetes, not the risk of having undiagnosed
It was created from an analysis of individu-
als aged 45 to 64 years in Atherosclerosis Risk in Com-
munities (ARIC) study and is based on the metabolic
Among individuals without diagnosed
diabetes or fasting plasma glucose ≥126 mg/dL at base-
line, a scoring strategy was developed that included
large waist circumference (>102 cm in men and >88
cm for women), raised blood pressure (>130/85 mm
Hg or antihypertensive medications), low high-density
lipoprotein cholesterol levels (<40 mg/dL for men and
<50 mg/dL for women), high triglyceride levels (>150
mg/dL), body mass index (BMI) of greater than 30
, and hyperglycemia. Each of the characteristics
are worth 1 point except for hyperglycemia, which can
be worth 2 points if fasting glucose is ≥102 mg/dL or 5
points when fasting glucose ≥111 mg/dL. A score of ≥4
puts an individual at high risk for development of dia-
betes, either diagnosed or undiagnosed.
This particular risk score was chosen for several
reasons. First, it has moderate sensitivity (68%) and
speciﬁ city (75%). The area under the receiver operat-
ing characteristic (ROC) curve was 0.78. Second, it
is computed in a reasonably straightforward manner
without having to use coefﬁ cients from the ARIC
cohort that may be speciﬁ c to that cohort.
Family history of diabetes has been shown to be a pre-
dictor of development of diabetes.
We deﬁ ned family
history as either a parent having diagnosed diabetes, or
a parent or sibling having diagnosed diabetes.
We used MedC alc sof tware
to compute ROC curve
analyses in an effort to evaluate the ability of the diabe-
tes risk score, as well as other variables, including family
history of diabetes and BMI, to predict development of
diabetes in 10 years. We speciﬁ cally examined the use-
fulness of family history as an alternative to the diabetes
risk score, because family history was not included in
the risk score. We also examined the predictive ability
of BMI by itself, because recent evidence showed that
BMI was as predictive of having undiagnosed diabetes as
the Cambridge Risk Score.
The parsimonious beneﬁ t
of prediction by means of one easily accessible variable
(eg, BMI) instead of a 6-variable measure would be sub-
stantial. BMI was evaluated in a continuous manner as
well as in a 3-category classiﬁ cation (<25, 25-29.9, ≥30).
To compute the beneﬁ ts of adding family history of
diabetes to BMI, we needed to provide a point score for
the new variable. Thus, we scored 1 point for BMI <25,
2 points for BMI 25-29.99, 3 points for BMI ≥30, and 1
point for family history of diabetes.
Finally, we stratiﬁ ed the CARDIA cohort by race to
examine the utility of the diabetes risk score within dif-