Content uploaded by Nancy C Tkacs
Author content
All content in this area was uploaded by Nancy C Tkacs on Apr 06, 2015
Content may be subject to copyright.
Insulin resistance is a component of several health
disorders, most notably impaired glucose tolerance
and type 2 diabetes mellitus. Insulin-resistant indi-
viduals have an impaired biological response to the
usual action of insulin; that is, they have reduced
insulin sensitivity. Various methods are used to assess
insulin sensitivity both in individuals and in study
populations. Validity, reproducibility, cost, and
degree of subject burden are important factors for
both clinicians and researchers to consider when
weighing the merits of a particular method. This arti-
cle describes several in vivo methods used to assess
insulin sensitivity and presents the advantages and
disadvantages of each.
Key words: insulin sensitivity, insulin resistance,
minimal model, HOMA, QUICKI, glucose clamp
Insulin sensitivity is defined as the effectiveness of
insulin in reducing blood glucose by directly promoting
glucose uptake into muscle and fat cells as well as by
increasing hepatic glycogen storage and reducing
hepatic glucose production. A significantly reduced level
of insulin sensitivity is called insulin resistance (in which
cells are less responsive to this effect of insulin so that
higher insulin concentrations are required to achieve
euglycemia with a given glucose load; Cefalu, 2000).
Insulin sensitivity varies both acutely and chronically.
Prolonged sedentary activity, stress with concomitant
counterregulatory hormone release, and infection are
known factors that can temporarily reduce insulin sensi-
tivity. When insulin sensitivity is significantly reduced on
a chronic basis (such as occurs with obesity and type 2
diabetes), this chronic insulin resistance can have serious
pathological consequences (Expert Committee on the
Diagnosis and Classification of Diabetes Mellitus
[Expert Committee], 2003; Grundy et al., 2005; Reaven,
1988, 1993).
Mechanisms of Insulin Action
Whole-body glucose disposal is both insulin
mediated and non–insulin mediated. Although an
exhaustive description of all of the factors mediating
glucose homeostasis is beyond the scope of this arti-
cle, a brief overview of insulin-mediated glucose dis-
posal is shown in Figure 1. The primary action of
insulin is to enable the rapid uptake of glucose into
muscle and fat cells. An additional glucose-lowering
action of insulin is mediated by its effect on the liver.
Methods of Measuring Insulin Sensitivity
Kimberly K. Trout, PhD, RN, CNM
Carol Homko, PhD, RN, CDE
Nancy C. Tkacs, PhD, RN
BIOLOGICAL RESEARCH FOR NURSING
Vol. 8, No. 4, April 2007, 305-318
DOI: 10.1177/1099800406298775
Copyright © 2007 Sage Publications
Kimberly K. Trout, PhD, RN, CNM, is an assistant professor
at Villanova University College of Nursing, Villanova,
Pennsylvania. Carol Homko, PhD, RN, CDE, is an assistant
research professor in the Department of Medicine, Temple
University School of Medicine, Philadelphia, Pennsylvania.
Nancy C. Tkacs, PhD, RN, is an associate professor at the
University of Pennsylvania School of Nursing and the
Institute for Diabetes, Obesity and Metabolism, University
of Pennsylvania School of Medicine, Philadelphia, Pennsy-
lvania. Address for correspondence: Kimberly K. Trout,
PhD, RN, CNM, Villanova University College of Nursing,
St. Mary’s Hall, 800 Lancaster Avenue, Villanova, PA
19085; e-mail: kimberly.trout@villanova.edu.
This work was supported in part by a grant from the
National Institutes of Health, National Institute of Nursing
Research (1-F31-NR008179 to K.K.T.).
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
For individuals in the FED state, insulin promotes
hepatic glucose utilization for glycogen synthesis
and glycolysis. In addition, insulin inhibits hepatic
glucose production by the metabolic pathways of
glycogenolysis and gluconeogenesis. Thus, insulin
facilitates hepatic glucose uptake by promoting glu-
cose utilization and by preventing glucose produc-
tion. The action of insulin to promote glucose uptake
and utilization by cells results in decreased levels of
blood glucose (Ganong, 2003).
Insulin is produced by the β cells of the pancreas.
Preproinsulin originates in the endoplasmic reticu-
lum and then is cleaved by microsomal enzymes to
proinsulin. Proinsulin is packaged into secretory
granules, where it is further cleaved by enzymes,
resulting in the production of equimolar amounts of
insulin and C-peptide (Lingappa & Farey, 2000).
Although there is no known physiologic function for
C-peptide, its measurement in plasma, as well as that
of proinsulin, is useful as a marker of endogenous
insulin secretion.
Glucose is the primary secretagogue of insulin secre-
tion. This process is initiated when glucose enters the
βcells via the GLUT 2 transporter, which is immediately
followed by glucose phosphorylation by the enzyme glu-
cokinase. Glucose-6-phosphate is then metabolized via
the glycolytic pathway and the Krebs cycle, increasing β
cell adenosine triphosphate. The increasing intracellular
adenosine triphosphate/adenosine diphosphate ratio sig-
nals for closure of membrane potassium channels, lead-
ing to depolarization of the cell’s membrane potential.
β cell depolarization allows voltage-dependent calcium
channels to open; calcium enters the cells, producing
exocytosis of secretory granules and insulin release
(Liang & Matschinsky, 1994). Once insulin is released
and reaches its target cell, it binds to specific glycopro-
tein receptors that are located in the cell membrane. In
muscle and fat cells, binding of insulin to its receptor
stimulates the mobilization of GLUT 4 glucose trans-
porters. The usual location of the GLUT 4 transporter
protein is within specialized storage vesicles in the cell;
however, when insulin binds to its receptor, a cascade of
306 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
Figure 1. Insulin action in the postprandial state. As an anabolic hormone, insulin works on target tissues to promote storage
of the products of digestion and absorption of a mixed meal. The target tissues referred to here are liver, muscle, and adipose.
Glucose uptake into muscle and fat cells is mediated by the insulin-dependent glucose transporter GLUT 4. Glucose uptake into
liver cells is promoted by insulin-stimulated glucose utilization for synthesis of glycogen and for glycolysis. The acetyl CoA result-
ing from glycolysis may ultimately be used for synthesis of triglyceride, a longer-term energy storage vehicle than glycogen. The
actions of insulin on all three of these target tissues contribute to decreased plasma glucose in the postabsorptive state. Glucose-
6-P ==glucose-6-phosphate; GLUT 4 ==insulin-sensitive glucose-transporter 4; LDL ==low-density lipoprotein; LPL ==lipoprotein
lipase; VLDL ==very-low-density lipoprotein.
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
intracellular signaling events results in the translocation
of GLUT 4 transporter proteins to the plasma mem-
brane. An increase in the number of glucose transporters
on the plasma membrane allows increased transport of
glucose into the cell. Disruption of some of the complex
intracellular signaling pathways that enable the translo-
cation of the GLUT 4 protein is seen as an important
mechanism in the pathogenesis of insulin resistance.
However, decreased insulin sensitivity can be caused by
prereceptor, receptor, and/or postreceptor defects.
Enhancement of insulin sensitivity is also possible, par-
ticularly after exercise. It is associated, in part, with
increased insulin-receptor binding and increased traf-
ficking of GLUT 4 transporters to the cell surface
(Cefalu, 2000; Lingappa & Farey, 2000; Tulchinsky &
Little, 1994).
Pathologic Manifestations of
Insulin Resistance
Insulin resistance is a central component of type 2
diabetes mellitus, a chronic disease with multiple
metabolic effects. Estimates are that the prevalence of
type 2 diabetes will increase to approximately 250
million people worldwide by the year 2020 (O’Rahilly
& Savill, 1997). Impairment in regulating glucose
homeostasis is central to this disease. Individuals with
type 2 diabetes are so insulin resistant that the βcells
of the pancreas ultimately are unable to compensate
for the increased insulin needs, thus failing to produce
enough insulin to promote adequate glucose uptake
into the insulin-resistant cells. This failure results in
prolonged hyperglycemia (Expert Committee, 2003).
Type 1 diabetes is not primarily characterized as a dis-
ease of insulin resistance but is, instead, characterized by
absent β cell function due to autoimmune destruction of
the βcells. It is apparent, however, that insulin resistance
can also appear to some degree with type 1 diabetes (and
not necessarily only with commonly attributed comor-
bidities, such as obesity; Baron, Laakso, Brechtel, &
Edelman, 1991; DeFronzo, Hendler, & Simonson, 1982;
Greenbaum, 2002). Strowig and Raskin (2005) noted
that “insulin clamp procedures performed on nonobese
type 1 diabetic subjects under a variety of glycemic con-
ditions demonstrated increased hepatic glucose produc-
tion and reduced insulin clearance compared with nondi-
abetic subjects” (p. 1562).
Individuals with type 1 diabetes are dependent on the
exogenous administration of insulin to sustain life.
Identifying any factors that might cause variations in
insulin sensitivity would enable patients to make
more judicious insulin dosage adjustments. The well-
characterized “dawn phenomenon,” which necessitates
increased insulin dosage adjustments in the early morn-
ing hours to maintain euglycemia, is an example
(consisting partially of a decreased insulin sensitivity
component; Schmidt, Hadji-Georgopoulos, Rendell,
Margolis, & Kowarski, 1981; Scheiner & Boyer, 2005).
Chronic insulin resistance itself (even with ade-
quate βcell compensation that achieves euglycemia)
is seen as pathogenic. Insulin resistance syndrome,
metabolic syndrome, and syndrome X are all terms
that describe a syndrome in which insulin resistance
is just one component (Reaven, 1993). This syn-
drome was first characterized by Reaven (1993); the
clinical presentation includes hypertension, hyperin-
sulinemia, and atherogenic dyslipidemia. A recent
statement by the American Heart Association (AHA)
and the National Heart, Lung, and Blood Institute
(NHLBI) described these elements of the clinical pre-
sentation as key factors that are part of a constellation
that increases one’s risk of developing cardiovascular
disease and type 2 diabetes (Grundy et al., 2005). In
light of the total burden of morbidity and mortality
resulting from the sequelae of both diabetes mellitus
and the insulin resistance syndrome, establishing
accurate, reproducible, and cost-effective methods of
quantifying insulin sensitivity is critical.
Experimental studies of insulin sensitivity must
include in their designs the many factors that can cause
insulin sensitivity to vary. For example, exercise will
improve insulin sensitivity (Kirwan et al., 2000; Kraniou,
Cameron-Smith, Misso, Collier, & Hargreaves, 2000;
Nishida et al., 2001). Obesity (Saydah, Byrd-Holt, &
Harris, 2002), chronic heart failure (Doehner et al.,
2002), and corticosteroid use (Leipala, Raivio, Sarnesto,
Panteleon, & Fellman, 2002) decrease insulin sensitiv-
ity. Assessing the factors that influence the degree of
insulin sensitivity present in both type 1 and type 2 dia-
betes is an important area of research that will poten-
tially lead to better blood glucose control in affected
individuals. Quantifying the degree of insulin sensitivity
in nondiabetic individuals can help to identify those who
may be at risk for diabetes (such as those individuals
with impaired glucose tolerance) and could also help
determine treatment strategies (such as metformin or
thiazolidinedione use) for those who are not overtly dia-
betic but exhibit higher levels of insulin resistance.
Several different methods have been used to evalu-
ate insulin sensitivity. The most common methods are
Trout et al. / Methods of Measuring Insulin Sensitivity 307
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
described here, as well as the relative advantages and
disadvantages of each from both clinical and research
perspectives. Ideally, the method chosen for a particu-
lar study should be driven by its appropriateness for
the research question. However, factors such as degree
of subject burden and cost often play a large role in
determining which method is chosen for a particular
study. The highly rigorous euglycemic hyperinsuline-
mic clamp and minimal model analysis of the fre-
quently sampled intravenous glucose tolerance test
(FSIGT) are the standards to which other methods are
often compared. This article discusses the concordance
rates of other methods with clamp studies or minimal
model values. As Wallace and Matthews (2002) noted
in their review, a degree of caution is always warranted
when comparing studies because there can be slight
variations in protocols and techniques, even within the
same method. The studies selected for review in this
article serve to illustrate the method under discussion
and involve at-risk populations in which accurate
assessment of insulin resistance has particular clinical
relevance for that study population.
Methods of Insulin Sensitivity Assessment
Euglycemic Hyperinsulinemic Clamp
Background. The euglycemic hyperinsulinemic
clamp is often referred to in the literature as the gold
standard for assessing insulin sensitivity (Gungor,
Saad, Janofsky, & Arslanian, 2004; Uwaifo et al.,
2002). It allows researchers to quantify βcell sensi-
tivity to glucose and the sensitivity of body tissues
(muscle, fat, and liver) to insulin. The clamp test was
first developed by Andres, Swerdloff, Pozefsky, and
Coleman in 1965 in response to perceived limitations
of the prevailing tests, the oral glucose tolerance test
and the rapid intravenous glucose tolerance test
(Andres et al., 1965). Andres et al. identified poor
reproducibility of results, even within the same indi-
vidual, as the main problem of these tests.
Procedure. The euglycemic hyperinsulinemic
clamp is performed in the morning following a 12-hr
fast. The procedure for the clamp begins with the
insertion of two intravenous lines. One is placed in an
antecubital vein for glucose and insulin infusion. A
second is placed in retrograde fashion in a hand vein
while the forearm is placed in a heated box. Heating
the forearm opens arteriovenous anastomoses that
bypass the capillary bed, making it possible to insert
the catheter in the retrograde direction and collect arte-
rialized venous blood. This line is used for drawing
samples for glucose determination. In Andres et al.’s
(1965) original description of the test, they actually
used an arterial line, sampling whole blood from the
brachial artery.
The goal of the insulin infusion is to raise the
plasma insulin concentration to a plateau approxi-
mately 100 µU/ml of plasma over basal insulin levels
to approximate usual postprandial levels and to main-
tain it at that plateau for a period of 2 to 4 hr. The
insulin infusate is prepared with regular insulin in iso-
tonic saline, to which 2 ml of the subject’s blood or 5
cc of 25% albumin is added to prevent the absorption
of the insulin by the plastic IV bag. DeFronzo, Tobin,
and Andres (1979) noted that “the priming infusion
(of insulin) can be adjusted upward or downward on a
proportional basis to achieve any desired level of
plasma insulin concentration” (p. E217). The objec-
tive is to raise insulin to a level that will suppress
hepatic glucose production (Liu, Gardner, & Barrett,
1993; Murayama, Kawai, Watanabe, Yoshikawa, &
Yamashita, 1989).
To maintain this plateau and to keep the individual
euglycemic, variable amounts of 20% dextrose must
be infused (DeFronzo et al., 1979). Insulin dosage is
at a fixed level and the amount of dextrose infused
will depend on the subject’s insulin sensitivity.
Maintaining plasma glucose concentrations at
approximately 90 ±2 mg/dl (euglycemia) requires
frequent blood sampling (every 5-15 min) for plasma
glucose concentrations, which are measured with a
precise glucose analyzer. Samples for insulin levels
are generally obtained every 30 to 60 min throughout
the clamp procedure.
There are a number of ways to control the glucose
infusion rate during a clamp study. The most popular
and simplest way is manual infusion adjustments
based on operator experience. Generally, the rate and
amount of glucose infused is determined by trial and
error by the principal investigator. Alternatively, a
mathematical algorithm can be used. To use this
method, after the first 20 min of the test, glucose
measurements are obtained at 5-min intervals; the
rate of glucose infusion is adjusted according to the
formula noted in Figure 2.
Irrespective of the method used to determine rates
of glucose infusion, the results of the test are then
analyzed by computing Mvalues for each 20-min
308 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
interval of the test. A value of Mreflects the amount
of glucose metabolized (DeFronzo et al., 1979) and is
reported in values of mg/(kg ×min). Mis calculated
using the following formula:
M=INF – UC – SC,
where INF is the glucose infusion rate, UC is equal to
the amount of glucose lost in the urine, and SC is a
glucose distribution space correction. To determine
insulin sensitivity, an M/I ratio is calculated. Iis the
plasma insulin response, and an M/I ratio is a “mea-
sure of the quantity of glucose metabolized per unit of
insulin concentration and is thus a reasonable index of
tissue sensitivity to insulin” (DeFronzo et al., 1979,
p. E218). In summary, if an individual is relatively
insulin sensitive, larger amounts of glucose will need
to be infused (∼6-12 mg/kg/min) for a given amount
of insulin to keep the individual’s blood glucose in a
euglycemic range. If an individual is relatively insulin
resistant, smaller amounts of glucose will need to
be infused for a given amount of insulin to achieve
euglycemia.
Limitations/variations. Several variations in the
euglycemic hyperinsulinemic clamp have been intro-
duced, with differences in technique specifically
designed to answer particular research questions. The
Trout et al. / Methods of Measuring Insulin Sensitivity 309
(Gd-Gi) x 10 x (0.19 x body wt in kg)
Initially, begun at 4 min after the initiation of the insulin infusion at 2.0 mg/(kg
min), then:
where Si = the glucose concentration at any specified point of time i
Gd = the desired glucose concentration
Gi = the glucose concentration at any specified point of time i
Ginf = the glucose infusate concentration in mg per ml
PF = an infusion pump factor
SMi = the setting for the metabolic component of the infusion rate
Gb = the basal glucose concentration
(measured) plasma glucose level from the goal glucose level. Note: The
Ginf x 15
FMi = a correction factor that compensates for the deviation of the actual
FMi-1 for the initial (10 min) plasma glucose concentration is assumed to be
1.00; after that time the computed FMi-1 is used (DeFronzo et al., 1979).
x PF + [(SMi-2) x Gb/Gi) x FMi-1)]Si =
Figure 2. Formula for glucose infusion for the euglycemic clamp.
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
hyperglycemic clamp is additionally able to assess
insulin secretion, as this technique evaluates the
insulin response to sustained hyperglycemia (Elahi,
1996). Another variation of the clamp is the hyperin-
sulinemic hypoglycemic clamp, particularly useful
if the research question involves hypoglycemia and
counterregulatory responses (Kinsley, Widom, &
Simonson, 1995). Variations that involve the infusion
of isotope-labeled glucose allow the researcher to
distinguish the contribution of hepatic glucose output
to insulin resistance (Kirwan et al., 2000). Overall,
the original euglycemic hyperinsulinemic clamp and
its variations offer exquisite precision for the para-
meters that it measures. On the other hand, the clamp
tests are all labor intensive, technically difficult to
perform, and expensive. A researcher needs to gain
experience by working with someone proficient in
administering clamp tests before attempting to use
this method. Another limitation of the euglycemic
hyperinsulinemic clamp is that insulin sensitivity is
measured only under a steady-state condition, and
therefore, the test does not realistically portray
dynamic conditions such as those occurring after
normal meals.
FSIGT Minimal Model Analysis
Background. The FSIGT was developed by
Richard Bergman in 1979 and is based on what is
called the minimal model of glucose and insulin kinet-
ics. The minimal model has been described as “a
mathematical description of physiological reality”
(Bergman & Lovejoy, 1997, p. 10). Bergman devel-
oped the FSIGT and minimal model when he realized
that there were major limitations with the conven-
tional intravenous glucose tolerance test (IVGTT) that
was being used at the time. First, one could derive no
separation of glucose and insulin kinetics from the
results of the conventional IVGTT. Second, there
were not enough data-sampling points to enable a
meaningful analysis. Bergman analyzed seven differ-
ent possible mathematical models before settling on
the current model to best represent glucose and insulin
kinetics (Bergman, Ider, Bowden, & Cobelli, 1979;
Bergman & Lovejoy, 1997). The model is applied to
observed glucose and insulin values obtained from the
FSIGT to obtain an insulin sensitivity index (SI). The
SIvalue of the minimal model represents the fractional
disappearance of glucose per insulin concentration
unit over time; results are expressed as min–1/µU/ml.
In the original, unmodified, glucose-only FSIGT,
the mean SIfor nondiabetic subjects was 5.1 ×10–4
min–1/µU/ml, with a range of 1.8 ×10–4 min–1/µU/ml
to 9.6 ×10–4 min–1/µU/ml (Bergman & Lovejoy, 1997).
Procedure. Patient preparation requires that the
subject fast for 10 to 12 hr prior to beginning the test.
Subjects who are on subcutaneous insulin injections
typically discontinue their evening dose of insulin
and are placed on an intravenous insulin drip
overnight (Ward et al., 1991). Subjects on continuous
subcutaneous insulin infusion pumps (in which only
rapid-acting insulin is used) typically will discon-
tinue their pump use a few hours prior to initiation of
the FSIGT and may not require an intravenous insulin
infusion prior to the test.
Two intravenous lines must be inserted prior to the
beginning of the FSIGT, one for blood sampling and
one for the infusion of glucose or glucose and insulin.
When insulin or tolbutamide are also infused, the test
is sometimes referred to as the modified FSIGT
because the original test did not require the additional
infusion of insulin or tolbutamide and relied on only
the subject’s own endogenous insulin secretion
(Ward et al., 1991). Insulin administration is neces-
sary for using the test in subjects with type 1 diabetes
(who have no endogenous insulin secretion;
Finegood, Hramiak, & Dupre, 1990). Tolbutamide
can be used to augment insulin secretion in subjects
with type 2 diabetes or in subjects without diabetes.
Baseline blood samples are collected prior to initiat-
ing the infusion. At time =0, an infusion of 50% dex-
trose in water is typically infused over a 1- to 2-min
period, with a typical dosage of 300 mg/kg of body
weight. The blood-sampling protocol varies, but
approximately 26 to 30 blood samples are obtained
(Bergman & Lovejoy, 1997). There is some contro-
versy about the optimal frequency of sampling, with
researchers obviously wanting to acquire the minimal
number of samples needed to obtain the necessary
data (Steil, Murray, Bergman, & Buchanan, 1994;
Steil, Vollund, Kahn, & Bergman, 1993). The 26-
sample protocol requires samples for plasma insulin
and glucose to be obtained at regular intervals (shown
in Table 1) in relation to the glucose injection.
Typically, at time =20 min, an intravenous insulin
injection is given over a 30-s period at a dosage of
0.02 to 0.03 U/kg of body weight (although some
research protocols have given insulin at larger
dosages over a 2- to 5-min infusion for those persons
310 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
who are more insulin resistant; Bergman & Lovejoy,
1997; Ward et al., 1991).
The minimal model analysis consists of several dis-
crete phases. The first 7 to 10 min after glucose injec-
tion constitutes a phase in which glucose mixing in
the circulation occurs. The glucose bolus also stimu-
lates the subject’s pancreas to secrete insulin in
response to the glucose challenge, which is activated
when the elevated glucose reaches the pancreatic
βcells. At the same time, the elevated glucose level
causes a cessation of glucose production by the liver.
After the initial mixing phase, and up to about 20 min,
glucose disposal is almost entirely glucose mediated
(a parameter referred to in the model as glucose effec-
tiveness, or SG). After insulin is injected at time =20
min, glucose disposal is primarily insulin mediated.
There are typically two insulin peaks that occur sub-
sequent to the glucose injection: one peak of glucose-
stimulated endogenous insulin secretion and one peak
from the exogenous insulin infusion. In someone with
type 1 diabetes, there would be only one insulin peak
after the injection. The disposition of glucose in rela-
tion to the amount of insulin reflects the SI(insulin
sensitivity value). One can also report the acute
insulin response to glucose (AIRg), which assesses the
adequacy of insulin secretion in the individual. A dis-
position index (DI) can be calculated as the product of
AIRg and SI. The DI reflects the total capacity of the
individual to achieve insulin-mediated glucose dis-
posal and reflects the adequacy of both insulin secre-
tion (AIRg) and action (SI; Bergman et al., 1979;
Bergman & Lovejoy, 1997).
There are advantages to using the FSIGT and the
minimal model for research. It is highly reliable and
reproducible, and results from the FSIGT have been
validated internationally in more than 240 different
studies since its inception in 1979 (Bergman &
Lovejoy, 1997). It is not as labor intensive or as
Trout et al. / Methods of Measuring Insulin Sensitivity 311
Table 1. Sample Worksheet for Blood Sampling for Frequently Sampled Intravenous Glucose Tolerance Test
Blood Glucose at
Sample Relative Estimated Exact Time 5 cc Purple Bedside (PRN and as
Number Time (min) Time (hr) (hr)a(Study Blood) Notes Indicated)b
–60 IVs placed X
1 –10 X
2–5 X
3 0 X Dextrose (1 min)
42 X
53 X
64 X
75 X
86 X
98 X
10 10 X
11 12 X
12 14 X
13 16 X
14 18 X
15 20 X Insulin (30 s)
16 22 X
17 25 X
18 30 X
19 40 X
20 50 X
21 70 X
22 100 X
23 140 X
24 180 X
NOTE: PRN =as needed; IV =intravenous.
a. An X may indicate that the exact time is the same as the estimated time.
b. X indicates glucose value obtained.
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
expensive to perform as the euglycemic hyperinsu-
linemic clamp, and SIresults from the FSIGT have
good concordance rates with the euglycemic hyper-
insulinemic clamp (Bergman & Lovejoy, 1997). A
computer software program, MINMOD Millenium
(MINMOD, Inc., Los Angeles, CA), has been specif-
ically designed to allow easy analysis of the FSIGT
parameters after input of the glucose and insulin val-
ues obtained during the FSIGT. The FSIGT also
allows clinicians and researchers to identify the indi-
vidual parameters that constitute glucose disposal in
a particular individual. For example, if someone has
a low SIbut a normal or high AIRg, this individual is
relatively insulin resistant but is able to maintain
glucose homeostasis because of adequate insulin
secretion. The ability to identify and separate distinct
components of glucose disposal is a significant
advantage of the FSIGT and minimal model.
Limitations/variations. On the other hand, the
FSIGT is not as technically simple or inexpensive as
several of the oral or fasting specimen-only methods,
which somewhat limits its utility for large epidemio-
logic studies. Also, FSIGT parameters are sometimes
difficult to obtain in subjects with markedly decreased
insulin sensitivity, as demonstrated in a study of
women with gestational diabetes mellitus (GDM;
Buchanan, 1997). In a study that compared minimal
model parameters with clamp indices, Buchanan
(1997) found that approximately 10% of women with
GDM had an SIthat was undetectable when the
insulin FSIGT was modified with an insulin dosage of
0.03 U/kg body weight. Buchanan suggested that in
subjects with markedly decreased insulin sensitivity, a
higher insulin dose may be needed to achieve mean-
ingful results. However, using a higher insulin dose
increases the risk of the subject’s developing hypo-
glycemia during the test. Therefore, further research is
needed to determine the optimum insulin dosages dur-
ing the FSIGT for insulin-resistant individuals.
Despite this limitation, correlation of the minimal
model SIwith clamp-derived glucose infusion
rate:insulin ratios (GIR/I) resulted in a Pearson’s cor-
relation coefficient of r=.58 (Buchanan, 1997).
Katz and colleagues (2000) also noted an artifact
of this method in type 2 diabetic subjects. They found
that “for 7 of the 15 diabetic subjects, minimal model
analysis generated large negative values...imply-
ing that rises in insulin somehow cause glucose lev-
els to increase in these subjects” (p. 2404). Other
studies comparing the FSIGT (in populations not at
the most extreme levels of insulin resistance) with the
glucose clamp technique have demonstrated
Pearson’s rvalues ranging from .74 to .91 (Beard,
Bergman, Ward, & Porte, 1986; Bergman & Lovejoy,
1997; Bergman, Prager, Volund, & Olefsky, 1987;
Finegood, Pacini, & Bergman, 1984).
Oral Glucose Tolerance Test (OGTT)
Background. Other than simple fasting plasma glu-
cose values, the OGTT is the method most frequently
used by clinicians to establish the diagnosis of diabetes
(Expert Committee, 2003). There are several variations
of this test in terms of the oral dose of glucose and
sampling times. For nonpregnant individuals, the 75-g
OGTT is most often used. A diagnosis of diabetes is
conferred if an individual has a plasma glucose level
≥200 mg/dl as measured 2 hr after the ingestion of a
75-g glucose load. If an individual has a value in the
range of 140 to 199 mg/dl post–glucose load, she or he
is designated as having impaired glucose tolerance. In
pregnant women, a 50-g glucose load is given to
screen for GDM between 24 and 28 weeks gestation
(a period when the diabetogenic effects of various preg-
nancy hormones begin to manifest). A value of less
than 140 mg/dl at 1 hr post–glucose ingestion is con-
sidered normal. If the woman has a value that equals or
exceeds 140 mg/dl, then she should undergo a 100-g,
3-hr OGTT for diagnostic purposes.
Procedure. The 3-hr OGTT requires that the
patient fast overnight (8-14 hr) before the test, having
followed a diet for the previous 3 days that was unre-
stricted and contained ≥150 g of carbohydrates daily.
During the administration of the OGTT, the patient is
seated and cannot smoke. Normal values for the 3-hr
OGTT are fasting, <95mg/dl; 1-hr value, <180 mg/dl;
2-hr value, <155 mg/dl; and 3-hr value, <140 mg/dl.
Diagnosis of GDM is conferred when two or more of
the above values are reached or exceeded (American
Diabetes Association [ADA], 2004).
Limitations/variations. The OGTT is technically quite
simple to perform and certainly lower in cost than the
euglycemic hyperinsulinemic clamp or the FSIGT. These
considerations have made the OGTT the glucose-
challenge test of choice in clinical situations (ADA, 2004;
Expert Committee, 2003). However, there are some
problems with the OGTT that make it less desirable for
312 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
use in research situations. First, there is variability in the
rate of gastric emptying and glucose absorption from
the gastrointestinal tract, causing some imprecision from
the start. An extreme example of delayed gastric empty-
ing can be found in diabetic individuals with gastropare-
sis (Coulie & Camilleri, 2000). This variability can par-
tially account for poorly reproducible results even within
the same individual (ADA, 2004). Second, glucose mea-
surements in the standard OGTT do not give adequate
information regarding the dynamics of glucose and
insulin action.
The OGTT is a relatively crude measure of glu-
cose tolerance. It does not measure the components
of insulin sensitivity and insulin secretion. Knowing
the value of these parameters may be especially help-
ful in terms of projecting a pregnant woman’s risk of
developing diabetes later in life because women with
GDM are at greater risk of developing other forms of
diabetes in the future (Expert Committee, 2003).
In light of this limitation, attempts have been made
to obtain indices from OGTT data that might better
reflect βcell function and insulin sensitivity (Matsuda
& DeFronzo, 1999).
The procedure devised by Matsuda and DeFronzo
(1999) entails giving an individual a 75-g glucose
load after a 10- to 12-hr overnight fast. Blood
samples are then obtained at –30, –15, 0, 30, 60, 90,
and 120 min post–glucose ingestion for determina-
tion of plasma glucose and insulin concentrations.
The results are then subjected to the following equa-
tion to determine a whole-body insulin sensitivity
index:
10,000
√(FPG x FPI) x (Mean OGTT glucose concentration
x mean OGTT insulin concentration)
where 10,000 represents a constant that enables one
to achieve results in the range of 0 to 12, FPG is fast-
ing plasma glucose (mg/dl), and FPI is fasting plasma
insulin (µU/ml). Matsuda achieved a concordance of
r=.73 between the euglycemic hyperinsulinemic
clamp and the above formula. Obtaining insulin val-
ues in addition to glucose values and more frequent
sampling points are essential components of this for-
mula. However, even when subjected to this analysis,
the variability in the rate of gastric emptying and
glucose absorption from the gastrointestinal tract
remains a limitation of the OGTT.
Homeostasis Model Assessment (HOMA)
The original homeostasis model assessment
(HOMA, HOMA1, HOMA-R, or HOMA-IR) is a
relatively simple mathematical index for assessing
insulin resistance, which is calculated by using fast-
ing insulin and glucose values. The formula for the
HOMA model is
(G0×I0)/22.5,
where G0is the fasting plasma glucose value (mea-
sured in mmol/L), I0 is the fasting plasma insulin value
(measured in mU/L), and 22.5 is a constant (Matthews
et al., 1985). The HOMA has a range of approximately
2 to 15 (Yeni-Komshian, Carantoni, Abbasi, &
Reaven, 2000), with higher scores indicating increas-
ing insulin resistance. For example, in the San Antonio
Heart Study, the mean HOMA was 2.1 among individ-
uals with normal glucose tolerance, 4.3 in individuals
with impaired glucose tolerance, and 8.3 in people
with diabetes (Haffner, Miettinen, & Stern, 1997).
The advantages and limitations of the HOMA model
are essentially identical with those of another model, the
quantitative insulin sensitivity check index model
(QUICKI). Thus, we discuss the advantages and limi-
tations of both together below. The original HOMA
model may have one slight extra advantage in that it is
somewhat easier to perform the calculation to derive
the HOMA result (not requiring a computer or scien-
tific calculator).
Levy, Matthews, and Hermans (1998) have subse-
quently developed a computerized model for HOMA
that is available for use by researchers, which they
conclude is more precise in its physiologic modeling
of insulin sensitivity than the simple algebraic equa-
tion. The simple equation eliminates some terms
from the more complex equation from which it was
derived. The computer model is commonly referred
to as “HOMA2” (Wallace, Levy, & Matthews, 2004).
Information about the use of the HOMA2 can
be obtained from www.OCDEM.ox.ac.uk. Despite
the increased precision of the HOMA2 model,
Wallace and colleagues (2004) noted that it “has not
been as widely used as the approximation formulae”
(p. 1487).
Researchers may encounter reports of other stud-
ies with slight variations from the original expression
noted in the HOMA. For example, Kondo, Nomura,
Nakaya, Ito, and Ohguro (2005) examined the effects
Trout et al. / Methods of Measuring Insulin Sensitivity 313
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
of exercise on risk factors for cardiovascular disease
such as inflammatory markers and insulin resistance.
The formula they used for HOMA (which they
referred to as HOMA-R) was
fasting glucose ×fasting insulin/405.
This formula is different from the one originally pub-
lished by Matthews and colleagues (1985) simply
because glucose in this equation is measured in
mg/dl. To convert from mmol/L to mg/dl, one simply
multiplies by a factor of 18, thus explaining the 405
in the denominator of this equation.
Wallace and colleagues (2004) noted that a mean
of three insulin samples obtained at 5-min intervals
(for the fasting insulin value) is more accurate than a
single fasting sample (because of the pulsatile nature
of insulin secretion). In actual practice, however, a
single sample is usually obtained.
Studies comparing HOMA and clamp. Lansang,
Williams, and Carroll (2001) recently determined the
correlation between the HOMA model and the glucose
clamp method in a population of hypertensive patients
and normal controls. None of the subjects had other
significant comorbidities, and patients with diabetes or
renal disease were specifically excluded. Subjects on
angiotensin-converting enzyme inhibitors discontinued
their use of medication 3 months prior to study entry,
and those subjects on other antihypertensives discon-
tinued their medications 1 month prior to study entry.
After fasting overnight, subjects underwent euglycemic
hyperinsulinemic clamp studies, according to the
method devised by Andres and colleagues (1965) and
DeFronzo and colleagues (1979). HOMA values were
calculated according to the formula devised by
Matthews and colleagues (1985). There was a signifi-
cant negative correlation between clamp M/I values and
log-transformed HOMA indices in both hypertensive
(p<.0001) and normal subjects (p=.002), although the
Pearson’s rvalues of –.67 in the hypertensive group and
–.58 in the normal group were modest. These authors
concluded that the HOMA is a reasonable alternative to
the euglycemic hyperinsulinemic clamp and that it is
more practical for epidemiologic studies (Lansang
et al., 2001). Cohen and colleagues (2006) validated the
HOMA in obese women during gestation. They com-
pared the HOMA insulin sensitivity index and glucose
utilization rates (GRd) during hyperinsulinemic eug-
lycemic clamps during the second and third trimesters
of pregnancy and postpartum. Correlations between the
GRd and the HOMA-derived metabolic parameters were
significant, with r2=.435 (p=.003). Kirwan, Huston-
Presley, Kalhan, and Catalano (2001) also found
HOMA to be a good predictor of total insulin sensitiv-
ity throughout gestation.
QUICKI
The QUICKI was first proposed by Katz and col-
leagues (2000) as a relatively easy mathematical
index for assessing insulin sensitivity. QUICKI is
defined by the following formula:
1/log I0+ log G0,
where I0=fasting plasma insulin value (in µU/ml) and
G0=fasting plasma glucose value (in mg/dl). Katz and
colleagues found that, in the study population in which
they developed the QUICKI, the mean QUICKI value
for nonobese subjects was 0.382 ±0.007, for obese
subjects was 0.331 ±0.010, and for diabetic subjects
was 0.304 ±0.007 (Katz et al., 2000). The major
advantage of both the QUICKI and HOMA models is
that they both require only one blood draw from a fast-
ing patient. They thus do not require extensive techni-
cal expertise and constitute a much lower cost per
subject when compared with the euglycemic hyperin-
sulinemic clamp or the FSIGT, making the QUICKI
and HOMA models much more practical for use in
large-scale epidemiologic studies and for clinical situ-
ations. However, the major disadvantage is that both of
these methods fail to provide information about the
stimulated glucose and insulin systems. Essentially,
they provide information only about what is occurring
with homeostatic mechanisms in the fasting state,
largely reflecting insulin’s effect on hepatic glucose
production. Neither method adequately addresses the
peripheral action of insulin when stimulated with a
glucose challenge (either oral or intravenous).
Studies comparing QUICKI and other methods.
One study evaluated the HOMA, QUICKI, and fast-
ing insulin measures and how well each correlated
with the SIvalue of the minimal model in prepubertal
children (Cutfield, Jefferies, Jackson, Robinson, &
Hofman, 2003). Cutfield et al.’s (2003) study popula-
tion consisted of three groups of prepubertal children:
twins, children who were born prematurely (<36
weeks gestational age), and those who were born
314 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
small for gestational age (less than the 10th percentile
for gestational age). After overnight fasts, the
children underwent modified FSIGTs according to
Bergman et al.’s (1979) minimal model. Baseline
insulin and glucose samples were used to determine
HOMA and QUICKI indices. The QUICKI index did
not correlate well with SI(r=.2, p=.02), and the cor-
relation between the HOMA index and SI(r=–.4,
p<.001) was no better than the correlation of SIwith
fasting values of insulin alone. The HOMA and
QUICKI indices even failed to detect changes in
insulin sensitivity in subgroups of the children who
had known conditions with reduced insulin sensitiv-
ity (such as obesity). Declines in SIwere robustly
detected by the minimal model in these children, thus
validating the FSIGT and minimal model in these
high-risk groups. The authors concluded that neither
HOMA nor QUICKI should be used to evaluate
insulin sensitivity and insulin resistance in prepuber-
tal children (Cutfield et al., 2003).
However, Gungor and colleagues (2004) found
significant concordance of both the HOMA and the
QUICKI with euglycemic hyperinsulinemic clamp
studies in a mixed population of 156 children. The
study included prepubertal, pubertal nonobese, and
pubertal obese females affected with polycystic ovary
syndrome (PCOS) with both normal glucose toler-
ance and impaired glucose tolerance. The correlation
between clamp-derived insulin sensitivity and the
HOMA index was r=.91, and the correlation
between the QUICKI and the clamp-derived insulin
sensitivity was r=.78. The overall conclusions were
that fasting glucose and insulin levels subjected to the
HOMA and QUICKI analyses were “valuable surro-
gates” for the clamp estimates in nondiabetic children
(Gungor et al., 2004).
Another study found that the HOMA and QUICKI
models were unsatisfactory in accurately assessing
insulin resistance in women with PCOS (Diamanti-
Kandarakis, Kouli, Alexandraki, & Spina, 2003).
Insulin resistance is a core feature of this disorder, and
treatment of a woman’s insulin resistance often amelio-
rates other symptoms, such as infertility. Diamanti-
Kandarakis and colleagues (2003) evaluated 59 women
who had been diagnosed with PCOS. The women were
divided fairly equally across body mass index ranges:
There was a lean group, an overweight group, and an
obese group. After 10- to 12-hr overnight fasts, the
women underwent euglycemic hyperinsulinemic
clamp studies according to the method developed by
DeFronzo and colleagues (1979). HOMA determina-
tions were calculated according to the method devised
by Matthews and colleagues (1985), and QUICKI
determinations were calculated according to the
method devised by Katz and colleagues (2000). The
authors found no significant correlations between
the Mvalue obtained from the clamp studies and the
HOMA (r=.1, p=nonsignificant [ns]) or the QUICKI
(r=.1, p=ns).
Other Notable Methods
Fasting (8-12 hr) insulin levels are often used as a
clinical marker for insulin resistance. The AHA/
NHLBI statement (Grundy et al., 2005) noted that the
European Group for the Study of Insulin Resistance
uses the criterion of insulin levels greater than the
75th percentile of the population as indicative of
insulin resistance. When measuring insulin levels, lab-
oratory tests cannot distinguish between endogenous
and exogenously administered insulin. For this rea-
son, C-peptide or proinsulin assays are more often
used as valid markers of endogenous insulin secretion
for those individuals taking insulin. Another clinical
marker of insulin resistance in this population is the
total daily insulin dose, with more insulin-resistant
individuals obviously requiring higher total daily
doses of insulin, with notably higher insulin:carbohy-
drate ratios (Wallace & Matthews, 2002).
McAuley’s index is another method that is based
on fasting plasma values, in this case, fasting plasma
insulin (in µU/ml) and fasting triglycerides (in
mmol/L; McAuley et al., 2001). The formula for
McAuley’s index is
exp [2.63 – 0.28ln (insulin) – 0.31ln (triglycerides)].
Otterdoom and colleagues (2005) found that
McAuley’s index (r=.61) correlated better with M/I
clamp values than the HOMA (r=–.53), QUICKI
(r=.52), or fasting insulin values alone (r=–.56) in
a stable renal transplant population.
Conclusions
Developing valid, reliable, cost-effective methods of
assessing insulin sensitivity is a major scientific chal-
lenge. The two methods that are considered superior in
Trout et al. / Methods of Measuring Insulin Sensitivity 315
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
terms of validity and reliability are the euglycemic
hyperinsulinemic clamp and the FSIGT. Unfortunately,
both of these methods are relatively expensive to per-
form, the former more so than the latter. The OGTT,
HOMA, and QUICKI methods are relatively inexpen-
sive and technically easy; however, they do not provide
nearly the same degree of information as the clamp and
the FSIGT, even though some studies have found good
concordance rates between these and the more reliable
tests. Poor concordance, however, has been reported
in prepubertal children and women with PCOS.
McAuley’s index shows promise as a method requiring
only fasting plasma samples; thus, it is expected that
more studies may seek to validate this method with dif-
ferent populations in the future.
The FSIGT is technically easier and less
expensive to perform than the euglycemic hyperin-
sulinemic clamp and has good concordance with
clamp-derived insulin sensitivity values, although its
usefulness may be limited in those individuals who
have the highest insulin resistance (Bergman &
Lovejoy, 1997). Minimal model analysis of the
FSIGT is able to provide some additional information
that is not available from the euglycemic hyperinsu-
linemic clamp. The FSIGT is able to differentiate
between glucose-mediated glucose disposal (SG) and
total insulin-mediated glucose disposal (DI). Also,
the FSIGT is able to differentiate between the acute
insulin response to glucose (AIRg) and the peripheral
action of insulin to affect glucose uptake (SI). Its use
in more than 240 studies has validated it as an
effective method of assessing insulin sensitivity and
insulin resistance (Bergman & Lovejoy, 1997).
Cefalu (2000) noted that the FSIGT is “more practi-
cal” for use with larger populations, as it is not as
technically complex or expensive to perform as the
glucose clamp (p. 71).
With the current diabetes epidemic, it is expected
that more research will be generated using all of
these methods. Assessing insulin sensitivity across
different ethnic backgrounds, clinical conditions, and
environmental states may help to elucidate factors that
will contribute to better diabetes control or avoidance.
References
American Diabetes Association. (2004). Standards of medical
care in diabetes. Diabetes Care,27(Suppl. 1), S15-S35.
Andres, R., Swerdloff, R., Pozefsky, T., & Coleman, D. (1965).
Manual feedback technique for the control of blood glucose
concentration. In L.T. Skeggs (Ed.), Automation in analytical
chemistry (pp. 486-491). New York: Mediad, Inc.
Baron, A. D., Laakso, M., Brechtel, G., & Edelman, S. V. (1991).
Mechanism of insulin resistance in insulin-dependent diabetes
mellitus: A major role for reduced skeletal muscle blood flow.
Journal of Clinical Endocrinology and Metabolism,73, 637-643.
Beard, J. C., Bergman, R. N., Ward, W. K., & Porte, D. (1986).
The insulin sensitivity index in nondiabetic man: Correlation
between clamp-derived and IVGTT-derived values. Diabetes,
35, 362-369.
Bergman, R. N., Ider,Y. Z., Bowden, C. R., & Cobelli, C. (1979).
Quantitative estimation of insulin sensitivity. American
Journal of Physiology,236, E667-E677.
Bergman, R. N., & Lovejoy, J. C. (Eds.). (1997). The minimal
model approach and determinants of glucose tolerance (Vol.
7). Baton Rouge: Louisiana State University Press.
Bergman, R. N., Prager, R., Volund, A., & Olefsky, J. M. (1987).
Equivalence of the insulin sensitivity index in man derived by
the minimal model method and the euglycemic glucose
clamp. Journal of Clinical Investigation,79, 790-800.
Buchanan, T. A. (1997). Measurement of insulin sensitivity in
pregnancy with glucose clamps and the minimal model. In R.
N. Bergman & J. C. Lovejoy (Eds.), The minimal model
approach and determinants of glucose tolerance (pp. 323-
343). Baton Rouge: Louisiana State University Press.
Cefalu, W. T. (2000). Insulin resistance. In J. L. Leahy, N. G.
Clark, & W. T. Cefalu (Eds.), Medical management of dia-
betes mellitus (pp. 57-75). New York: Marcel Dekker.
Cohen, O., Epstein, G. S., Weisz, B., Homko, C. J., & Sivan, E.
(2006). Longitudinal assessment of insulin sensitivity in preg-
nancy: Validation of the homeostasis model assessment.
Clinical Endocrinology,64, 640-644.
Coulie, B., & Camilleri, M. (2000). Gastrointestinal complica-
tions of diabetes mellitus. In J. L. Leahy, N. G. Clark, & W. T.
Cefalu (Eds.), Medical management of diabetes mellitus (pp.
387-406). New York: Marcel Dekker.
Cutfield, W. S., Jefferies, C. A., Jackson, W. E., Robinson, E. M.,
& Hofman, P. L. (2003). Evaluation of HOMA and QUICKI
as measures of insulin sensitivity in prepubertal children.
Pediatric Diabetes,4, 119-125.
DeFronzo, R. A., Hendler, R., & Simonson, D. (1982). Insulin
resistance is a prominent feature of insulin-dependent dia-
betes. Diabetes,31, 795-801.
DeFronzo, R. A., Tobin, J. D., & Andres, R. (1979). Glucose
clamp technique: A method for quantifying insulin secre-
tion and resistance. American Journal of Physiology,237,
E214-E223.
Diamanti-Kandarakis, E., Kouli, C., Alexandraki, K., & Spina, G.
(2003). Failure of mathematical indices to accurately assess
insulin resistance in lean, overweight, or obese women with
polycystic ovary syndrome. Journal of Clinical Endocrinology
& Metabolism,89, 1273-1276.
Doehner, W., Rauchhaus, M., Godsland, I. F., Egerer, K.,
Niebauer, J., Sharma, R., et al. (2002). Insulin resistance in
moderate chronic heart failure is related to hyperleptinaemia,
but not to norepinephrine or TNF-alpha. International
Journal of Cardiology,83, 73-81.
Elahi, D. (1996). In praise of the hyperglycemic clamp. Diabetes
Care,19, 278-286.
316 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
Expert Committee on the Diagnosis and Classification of
Diabetes Mellitus. (2003). Report of the Expert Committee on
the Diagnosis and Classification of Diabetes Mellitus.
Diabetes Care,26(Suppl. 1), S5-S20.
Finegood, D. T., Hramiak, I. M., & Dupre, J. (1990). A modified
protocol for estimation of insulin sensitivity with the minimal
model of glucose kinetics in patients with insulin-dependent
diabetes. Journal of Clinical Endocrinology & Metabolism,
70, 1538-1549.
Finegood, D. T., Pacini, G., & Bergman, R. N. (1984). The
insulin sensitivity index: Correlation in dogs between values
determined from the intravenous glucose tolerance test and
the euglycemic clamp. Diabetes,33, 362-368.
Ganong, W. F. (2003). Review of medical physiology. New York:
McGraw-Hill.
Greenbaum, C. J. (2002). Insulin resistance in type 1 diabetes.
Diabetes/Metabolism Research and Reviews,18, 192-200.
Grundy, S. M., Cleeman, J. I., Daniels, S. R., Donato, K. A.,
Eckel, R. H., Franklin, B. A., et al. (2005). Diagnosis and
management of the metabolic syndrome: An American Heart
Association/National Heart, Lung, and Blood Institute scien-
tific statement. Circulation,112, 2735-2752.
Gungor, N., Saad, R., Janofsky, J., & Arslanian, M. D. (2004).
Validation of surrogate estimates of insulin sensitivity and
insulin secretion in children and adolescents. Journal of
Pediatrics,144, 47-55.
Haffner, S. M., Miettinen, H., & Stern M. P. (1997). The homeo-
static model in the San Antonio Heart Study. Diabetes Care,
20,1087-1088.
Katz, A., Nambi, S. S., Mather, K., Baron, A. D., Follmann, D.
A., Sullivan, G., et al. (2000). Quantitative insulin sensitivity
check index: A simple, accurate method for assessing insulin
sensitivity in humans. Journal of Clinical Endocrinology &
Metabolism,85, 2402-2410.
Kinsley, B. T., Widom, B., & Simonson, D. C. (1995).
Differential regulation of counterregulatory hormone secre-
tion and symptoms during hypoglycemia in IDDM. Diabetes
Care,18, 17-26.
Kirwan, J. P., Del Aguila, L. F., Hernandez, J. M., Williamson, D.
L., O’Gormna, D. J., Lewis, R., et al. (2000). Regular exercise
enhances insulin activation of IRS-1 associated PI-3 kinase in
human skeletal muscle. Journal of Applied Physiology,88,
797-803.
Kirwan, J. P., Huston-Presley, L., Kalhan, S. C., & Catalano, P.
M. (2001). Clinically useful estimates of insulin sensitivity
during pregnancy. Diabetes Care,24, 1602-1607.
Kondo, N., Nomura, M., Nakaya,Y., Ito, S., & Ohguro, T. (2005).
Association of inflammatory marker and highly sensitive C-
reactive protein with aerobic exercise capacity, maximum
oxygen uptake and insulin resistance in healthy middle-aged
volunteers. Circulation Journal,69, 452-457.
Kraniou, Y., Cameron-Smith, D., Misso, M., Collier, G., &
Hargreaves, M. (2000). Effects of exercise on GLUT-4 and
glycogenin gene expression in human skeletal muscle.
Journal of Applied Physiology,88, 794-796.
Lansang, M. C., Williams, G. H., & Carroll, J. S. (2001).
Correlation between the glucose clamp technique and the
homeostasis model assessment in hypertension. American
Journal of Hypertension,14, 51-53.
Leipala, J. A., Raivio, K. O., Sarnesto, A., Panteleon, A., &
Fellman, V. (2002). Intrauterine growth restriction and post-
natal steroid treatment effects on insulin sensitivity in preterm
neonates. Journal of Pediatrics,141, 472-476.
Levy, J. C., Matthews, D. R., & Hermans, M. P. (1998). Correct
Homeostasis Model Assessment (HOMA) evaluation uses the
computer program. Diabetes Care,21, 2191-2192.
Liang, Y., & Matschinsky, F. M. (1994). Mechanisms of action of
nonglucose insulin secretagogues. Annual Review of
Nutrition,14, 59-81.
Lingappa, V. R., & Farey, K. (2000). Physiological medicine: A
clinical approach to basic medical physiology. New York:
McGraw-Hill.
Liu, Z., Gardner, L. B., & Barrett, E. J. (1993). Insulin and glu-
cose suppress hepatic glycogenolysis by distinct enzymatic
mechanisms. Metabolism,42, 1546-1551.
Matsuda, M., & DeFronzo, R. A. (1999). Insulin sensitivity
indices obtained from oral glucose tolerance testing. Diabetes
Care,22, 1462-1470.
Matthews, D. R., Hosker, J. P., Rudenski, A. S., Naylor, B. A.,
Treacher, D. F., & Turner, R. C. (1985). Homeostasis model
assessment: Insulin resistance and beta-cell function from
fasting plasma glucose and insulin concentrations in man.
Diabetalogia,28, 412-419.
McAuley, K. A., Williams, S. M., Mann, J. I., Walker, R. J.,
Lewis-Barned, N. J., Temple, L. A., et al. (2001). Diagnosing
insulin resistance in the general population. Diabetes Care,
24, 460- 464.
Murayama, Y., Kawai, K., Watanabe, Y., Yoshikawa, H., &
Yamashita, K. (1989). Insulin and glucagon secretion are sup-
pressed equally during both hyper- and euglycemia by moder-
ate hyperinsulinemia in patients with diabetes mellitus. Journal
of Clinical Endocrinology and Metabolism,68, 925-931.
Nishida, Y., Higaki, Y., Tokuyama, K., Fujimi, K., Kiyonaga, A.,
Shindo, M., et al. (2001). Effect of mild exercise training on glu-
cose effectiveness in healthy men. Diabetes Care,24, 1008-1013.
O’Rahilly, S., & Savill, J. (1997). Science, medicine, and the
future. Non-insulin dependent diabetes mellitus: The gather-
ing storm. British Medical Journal,314, 955-959.
Otterdoom, L. H., DeVries, A. P. J., vanSon, W. J., van der Heide,
J. J. H., Ploeg, R. J., Gansevoort, R. T., et al. (2005).
Validation of insulin resistance indexes in a stable renal trans-
plant population. Diabetes Care,28, 2424-2429.
Reaven, G. M. (1988). Role of insulin resistance in human dis-
ease. Diabetes,37, 1595-1607.
Reaven, G. M. (1993). Role of insulin resistance in human dis-
ease (syndrome X): An expanded definition. Annual Review of
Medicine,44, 121-131.
Saydah, S. H., Byrd-Holt, D., & Harris, M. I. (2002). Projected
impact of implementing the results of the diabetes prevention
program in the U.S. population. Diabetes Care,25, 1940-1945.
Scheiner, G., & Boyer, B. (2005). Characteristics of basal insulin
requirements by age and gender in type-1 diabetes patients
using insulin pump therapy. Diabetes Research and Clinical
Practice,69, 14-21.
Schmidt, M. I., Hadji-Georgopoulos, A., Rendell, M., Margolis,
S., & Kowarski, A. (1981). The dawn phenomenon, an early
morning glucose rise: Implications for diabetic intraday blood
glucose variation. Diabetes Care,4, 579-584.
Trout et al. / Methods of Measuring Insulin Sensitivity 317
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
Steil, G. M., Murray, J., Bergman, R. N., & Buchanan, T. A.
(1994). Repeatability of insulin sensitivity and glucose effec-
tiveness from the minimal model: Implications for study
design. Diabetes,43, 1365-1371.
Steil, G. M., Vollund, A., Kahn, S. E., & Bergman, R. N. (1993).
Reduced sample number for calculation of insulin sensitivity
and glucose effectiveness from the minimal model: Suitability
for use in population studies. Diabetes,42, 250-256.
Strowig, S. M., & Raskin, P. (2005). The effect of rosiglitazone
on overweight subjects with type 1 diabetes. Diabetes Care,
28, 1562-1567.
Tulchinsky, D., & Little, A. B. (Eds.). (1994). Maternal-fetal
endocrinology (2nd ed.). Philadelphia: W. B. Saunders.
Uwaifo, G. I., Parikh, S. J., Keil, M., Elberg, J., Chin, J., & Yanovski, J.
A. (2002). Comparison of insulin sensitivity, clearance, and secretion
estimates using euglycemic and hyperglycemic clamps in children.
Journal of Clinical Endocrinology & Metabolism,87, 2899-2905.
Wallace, T. M., Levy, J. C., & Matthews, D. R. (2004). Use and
abuse of HOMA modeling. Diabetes Care,27, 1487-1494.
Wallace, T. M., & Matthews, D. R. (2002). The assessment of
insulin resistance in man. Diabetic Medicine,19, 527-534.
Ward, G. M., Weber, K. M., Walters, I. M., Aitken, P. M., Lee, B.,
Best, J. D., et al. (1991). A modified minimal model analysis
of insulin sensitivity and glucose-mediated glucose disposal
in insulin-dependent diabetes. Metabolism-Clinical and
Experimental,40, 4-9.
318 BIOLOGICAL RESEARCH FOR NURSING Vol. 8, No. 4, April 2007
at UNIV OF PENNSYLVANIA on April 6, 2015brn.sagepub.comDownloaded from
