Effectiveness of sensor-augmented insulin-pump therapy in type 1 diabetes.

Abstract

Recently developed technologies for the treatment of type 1 diabetes mellitus include a variety of pumps and pumps with glucose sensors.
In this 1-year, multicenter, randomized, controlled trial, we compared the efficacy of sensor-augmented pump therapy (pump therapy) with that of a regimen of multiple daily insulin injections (injection therapy) in 485 patients (329 adults and 156 children) with inadequately controlled type 1 diabetes. Patients received recombinant insulin analogues and were supervised by expert clinical teams. The primary end point was the change from the baseline glycated hemoglobin level.
At 1 year, the baseline mean glycated hemoglobin level (8.3% in the two study groups) had decreased to 7.5% in the pump-therapy group, as compared with 8.1% in the injection-therapy group (P<0.001). The proportion of patients who reached the glycated hemoglobin target (<7%) was greater in the pump-therapy group than in the injection-therapy group. The rate of severe hypoglycemia in the pump-therapy group (13.31 cases per 100 person-years) did not differ significantly from that in the injection-therapy group (13.48 per 100 person-years, P=0.58). There was no significant weight gain in either group.
In both adults and children with inadequately controlled type 1 diabetes, sensor-augmented pump therapy resulted in significant improvement in glycated hemoglobin levels, as compared with injection therapy. A significantly greater proportion of both adults and children in the pump-therapy group than in the injection-therapy group reached the target glycated hemoglobin level. (Funded by Medtronic and others; ClinicalTrials.gov number, NCT00417989.)

Full text

Page 1

The new england journal of medicine
10.1056/nejmoa1002853 nejm.org
1
original article
Effectiveness of Sensor-Augmented Insulin-
Pump Therapy in Type 1 Diabetes
Richard M. Bergenstal, M.D., William V. Tamborlane, M.D.,
Andrew Ahmann, M.D., John B. Buse, M.D., Ph.D., George Dailey, M.D.,
Stephen N. Davis, M.D., Carol Joyce, M.D., Tim Peoples, M.A.,
Bruce A. Perkins, M.D., M.P.H., John B. Welsh, M.D., Ph.D.,
Steven M. Willi, M.D., and Michael A. Wood, M.D., for the STAR 3 Study Group*
From the International Diabetes Center
at Park Nicollet, Minneapolis (R.M.B.);
Yale University, New Haven, CT (W.V.T.);
Oregon Health and Science University,
Portland (A.A.); University of North Car-
olina School of Medicine, Chapel Hill
(J.B.B.); Scripps Institute, La Jolla (G.D.),
and Medtronic, Northridge (T.P., J.B.W.)
— both in California; University of Mary-
land School of Medicine, Baltimore
(S.N.D.); Memorial University of New-
foundland, Health Science Centre, St.
John’s, NL, Canada (C.J.); Toronto Gen-
eral Hospital, Toronto (B.A.P.); Children’s
Hospital of Philadelphia, Philadelphia
(S.M.W.); and Helen DeVos Children’s
Hospital, Grand Rapids, MI (M.A.W.).
Address reprint requests to Dr. Bergen-
stal at the International Diabetes Center
at Park Nicollet, Minneapolis, MN 55416,
or at richard.bergenstal@parknicollet.com.
*Members of the Sensor-Augmented
Pump Therapy for A1C Reduction (STAR)
3 Study Group are listed in the Appendix.
This article (10.1056/NEJMoa1002853) was
published on June 29, 2010, at NEJM.org.
N Engl J Med 2010.
Copyright © 2010 Massachusetts Medical Society.
Abstract
Background
Recently developed technologies for the treatment of type 1 diabetes mellitus include
a variety of pumps and pumps with glucose sensors.
Methods
In this 1-year, multicenter, randomized, controlled trial, we compared the efficacy of
sensor-augmented pump therapy (pump therapy) with that of a regimen of multiple
daily insulin injections (injection therapy) in 485 patients (329 adults and 156 chil-
dren) with inadequately controlled type 1 diabetes. Patients received recombinant
insulin analogues and were supervised by expert clinical teams. The primary end
point was the change from the baseline glycated hemoglobin level.
Results
At 1 year, the baseline mean glycated hemoglobin level (8.3% in the two study groups)
had decreased to 7.5% in the pump-therapy group, as compared with 8.1% in the
injection-therapy group (P<0.001). The proportion of patients who reached the gly-
cated hemoglobin target (<7%) was greater in the pump-therapy group than in the
injection-therapy group. The rate of severe hypoglycemia in the pump-therapy group
(13.31 cases per 100 person-years) did not differ significantly from that in the in-
jection-therapy group (13.48 per 100 person-years, P = 0.58). There was no significant
weight gain in either group.
Conclusions
In both adults and children with inadequately controlled type 1 diabetes, sensor-
augmented pump therapy resulted in significant improvement in glycated hemoglo-
bin levels, as compared with injection therapy. A significantly greater proportion of
both adults and children in the pump-therapy group than in the injection-therapy
group reached the target glycated hemoglobin level. (ClinicalTrials.gov number,
NCT00417989.)
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Downloaded from www.nejm.org on June 30, 2010 . For personal use only. No other uses without permission.

Page 2

The new england journal of medicine
10.1056/nejmoa1002853 nejm.org
2
I
tus,1-3 and diabetes practitioners are continuously
challenged to optimize glucose control while mini-
mizing severe hypoglycemia and weight gain. In-
sulin pumps and systems for continuous glucose
monitoring represent technologies designed to as-
sist patients with type 1 diabetes in safely reach-
ing glycemic goals. Among adults, the use of an
insulin pump has been shown to reduce glycated
hemoglobin levels without an increased risk of hy-
poglycemia, as compared with a regimen of mul-
tiple daily insulin injections, but results in chil-
dren have been inconsistent.4 Recent studies have
suggested that patients who used sensor-augment-
ed pump therapy with adherence to continuous
glucose monitoring had improved glycated hemo-
globin levels without an increased rate of hypo-
glycemia.5-7 Similarly, in a multicenter trial of con-
tinuous glucose monitoring in patients with type 1
diabetes, sponsored by the Juvenile Diabetes Re-
search Foundation (JDRF) (ClinicalTrials.gov num-
ber, NCT00406133), the use of a continuous glu-
cose-monitoring device was effective in reducing
glycated hemoglobin levels among patients who
were 25 years of age or older but not among pa-
tients under the age of 25 years.8
Sensor-augmented pump therapy integrates
these two technologies into one system and al-
lows patients and clinicians to monitor treatment
and response through Internet-based software.
Whether, and to what extent, switching directly to
sensor-augmented pump therapy might improve
metabolic control in patients with type 1 diabetes
who were previously unable to reach glycemic tar-
gets with a regimen of multiple daily injections
and conventional blood-glucose monitoring is un-
known. In this unmasked, randomized, controlled
trial, called Sensor-Augmented Pump Therapy for
A1C Reduction (STAR) 3, we evaluated the use
of sensor-augmented pump therapy and injec-
tion therapy at 30 diabetes centers in the United
States and Canada for 1 year.9
mproved glycemic control can reduce
the microvascular and macrovascular compli-
cations associated with type 1 diabetes melli-
Methods
Patients
Patients with type 1 diabetes were eligible if they
were between the ages of 7 and 70 years, had re-
ceived multiple daily injections that included a
long-acting analogue insulin during the previous
3 months, had a glycated hemoglobin level be-
tween 7.4% and 9.5%, and had been under the
care of the principal investigator or a referring
physician for at least 6 months. Patients were re-
quired to have access to a computer and to have
a history of testing blood glucose an average of
four or more times per day for the previous 30
days. Exclusion criteria were the use of insulin-
pump therapy within the previous 3 years, a his-
tory of at least two severe hypoglycemic events in
the year before enrollment, the use of a pharma-
cologic noninsulin treatment for diabetes during
the previous 3 months, and pregnancy or the in-
tention to become pregnant. All patients provided
written informed consent.
Treatments
Patients were randomly assigned to receive either
sensor-augmented pump therapy (pump therapy)
or a regimen of multiple daily injections (injection
therapy) with the use of a block design, stratified
according to age group: adults (19 to 70 years of
age) or children (7 to 18 years of age). Levels of
glycated hemoglobin and blood glucose in the two
study groups and sensor glucose values in the
pump-therapy group were disclosed to investiga-
tors, caregivers, and patients in order to optimize
glycated hemoglobin levels and to minimize the
risk of severe hypoglycemia.
The pump-therapy group used a device that in-
tegrates an insulin pump with continuous glucose
monitoring (MiniMed Paradigm REAL-Time Sys-
tem, Medtronic) (see Fig. 1 in the Supplementary
Appendix, available with the full text of this ar-
ticle at NEJM.org). Before randomization, all pa-
tients received training in intensive diabetes man-
agement, including carbohydrate counting and
the administration of correction doses of insulin.
Patients were first placed on insulin-pump ther-
apy for 2 weeks, and then glucose sensors were
introduced. During the 5 weeks after randomiza-
tion, patients in the pump-therapy group com-
pleted online insulin-pump training and attended
additional visits for insulin-pump and sensor
training. This group used insulin aspart (NovoLog
or NovoRapid, Novo Nordisk). The injection-ther-
apy group used both insulin glargine (Lantus,
Sanofi-Aventis) and insulin aspart under the guid-
ance of the treating clinician. Sensor glucose val-
ues were collected for 1-week periods at baseline,
6 months, and 1 year in the two study groups. In
the injection-therapy group, a device for continu-
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Downloaded from www.nejm.org on June 30, 2010 . For personal use only. No other uses without permission.

Page 3

Sensor-Augmented Insulin-Pump Therapy
10.1056/nejmoa1002853 nejm.org
3
ous glucose monitoring that collected but did
not display data (Guardian REAL-Time Clinical,
Medtronic) was used.
All patients were seen at 3, 6, 9, and 12 months
after randomization and used diabetes-manage-
ment software (CareLink Therapy Management
System for Diabetes–Clinical, Medtronic). Between
visits, communication with clinicians was initi-
ated at the discretion of the patient.
At follow-up clinic visits, glucose data were
reviewed, therapy was adjusted, glycated hemo-
globin was measured, and data on adverse events
were collected. Severe hypoglycemia was defined
as an episode requiring assistance and was con-
firmed by documentation of a blood glucose value
of less than 50 mg per deciliter (2.8 mmol per liter)
or recovery with restoration of plasma glucose.
Quest Clinical Trials Laboratory measured glycated
hemoglobin by means of immunoturbidimetry.
Study Oversight
The institutional review board at each study site
approved the protocol, and the conduct of the study
was consistent with the Good Clinical Practice
provisions of the Declaration of Helsinki and lo-
cal regulatory requirements. Data management
and statistical analyses were conducted by Par-
exel International, an independent clinical research
organization, which transferred all data to the
sponsor, Medtronic. Novo Nordisk supplied all
insulin aspart used in the study, and LifeScan,
Bayer Healthcare, and Becton Dickinson supplied
blood glucose meters.
All authors had access to the data, wrote the
first draft of the manuscript with editorial assis-
tance from representatives of the sponsor, sub-
sequently revised the manuscript, and made the
decision to submit the manuscript for publication.
All authors vouch for the accuracy and complete-
ness of the data and analyses. The STAR 3 steer-
ing committee was responsible for the study de-
sign and methods.9
The study was conducted in accordance with
the original trial protocol, with the following ex-
ceptions: the eligibility cutoff level for glycated
hemoglobin was lowered from 7.5% to 7.4%, the
exclusion criteria were changed from no previous
use of insulin-pump therapy to no such use with in
the previous 3 years, the sample size was in-
creased from 336 patients at 25 centers to 552 pa-
tients at 30 centers, results on the Hypoglycemia
Fear Survey were moved from a secondary end
point to a tertiary end point, the Telemetered
Glucose-Monitoring System (Medtronic) was re-
placed with the MiniLink transmitter (Medtronic),
and three visits during the 5 weeks after ran-
domization were removed from the schedule for
the injection-therapy group. (The trial protocol
is available at NEJM.org.)
Statistical Analysis
The primary outcome was the change from base-
line in the glycated hemoglobin level at 1 year.
Severe rates of hypoglycemia were analyzed as a
secondary outcome. We calculated that the en-
rollment of 495 patients would provide a power
of 90% to detect an absolute difference of 0.35
percentage points in the primary outcome. Analy-
ses were performed in the intention-to-treat pop-
ulation, defined as patients who underwent at least
one measurement of glycated hemoglobin after
randomization, with the last observation carried
forward for the imputation of missing data.
Differences in the change in glycated hemo-
globin levels were analyzed with the use of analy-
sis of covariance (ANCOVA) with three categorical
variables (study group, pooled investigative site,
and sex) and four continuous variables (age, dura-
tion of diabetes, body-mass index, and baseline
glycated hemoglobin level) as fixed effects. The
proportion of patients reaching glycated hemo-
globin targets was analyzed with the use of a
logistic-regression model with two categorical
variables (study group and sex) and the above-
mentioned continuous variables as fixed effects.
The effect of sensor use on glycated hemoglobin
levels was analyzed with the use of an ANCOVA
model with three categorical variables (sensor-use
categories, pooled investigative site, and sex) and
the above-mentioned continuous variables as fixed
effects. Changes in weight were analyzed with
the use of a general linear model with adjust-
ment for baseline weight. Incidences of adverse
events were compared with the use of a logistic-
regression model, and rates of adverse events per
100 person-years were compared with the use of
an ANCOVA model; both models had two cate-
gorical variables (study group and sex) and the
above-mentioned continuous variables as fixed
effects. The area under the curve is the product
of the magnitude and duration of the sensor-
measured glucose level above or below a specified
cutoff level. Higher values for this calculation in-
dicate more numerous, severe, or protracted gly-
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Downloaded from www.nejm.org on June 30, 2010 . For personal use only. No other uses without permission.

Page 4

The new england journal of medicine
10.1056/nejmoa1002853 nejm.org
4
cemic events. The areas under the curve that were
calculated from continuous glucose monitoring at
1 year were compared with the use of an ANCOVA
model with three categorical variables (study group,
pooled investigative site, and sex) and four con-
tinuous variables (age, duration of diabetes, body-
mass index, and baseline area under the curve)
as fixed effects. Baseline characteristics were com-
pared with the use of a two-sample t-test for
continuous variables and either the chi-square test
or Fisher’s exact test for categorical variables.
Analyses were conducted with SAS software,
version 9.2 (SAS Institute). All reported P values
are two-sided; a P value of less than 0.05 was con-
sidered to indicate statistical significance for com-
parisons of the primary outcome, baseline char-
acteristics, and safety.
Results
Study Recruitment and Baseline
Characteristics
From January 2007 through December 2008, a to-
tal of 495 patients underwent randomization; fol-
low-up data on glycated hemoglobin levels were
missing for 10 patients, who were not included in
the primary analysis. Of the remaining 485 pa-
tients, 4 (1%) were lost to follow-up, 32 (7%) dis-
continued the study or were withdrawn, and 6 (1%)
did not provide 1-year results for glycated hemo-
globin, leaving 443 patients in the primary analy-
sis (Fig. 2 in the Supplementary Appendix). Base-
line characteristics were similar in the two study
groups, except for weight (P = 0.02) and student
status (P = 0.02) among adult patients (Table 1).
Table 1. Baseline Characteristics of the Patients.*
CharacteristicSensor-Augmented Pump TherapyInjection Therapy
All Patients
(N = 244)
Adults
(N = 166)
Children
(N = 78)
All Patients
(N = 241)
Adults
(N = 163)
Children
(N = 78)
Mean age — yr32.2±17.541.9±12.3 11.7±3.031.5±16.540.6±12.0 12.7±3.1
Male sex — no. (%)140 (57)94 (57)46 (59) 134 (56)93 (57) 41 (53)
Race or ethnic group — no. (%)†
Hispanic7 (3)5 (3)2 (3)7 (3) 3 (2) 4 (5)
White221 (91)151 (91)70 (90)222 (92)153 (94) 69 (88)
Other16 (7)10 (6)6 (8) 12 (5)7 (4)5 (6)
Nonsmoking — no. (%)220 (90) 142 (86)78 (100)217 (90)139 (85) 78 (100)
No alcohol use — no. (%) 133 (55)56 (34)77 (99)127 (53)50 (31) 77 (99)
Employment status — no. (%)‡
Disabled3 (1)3 (2)0 4 (2)4 (2)0
Employed or volunteer140 (57) 139 (84)1 (1) 128 (53)127 (78)1 (1)
Homemaker 8 (3)8 (5)07 (3) 7 (4)0
Retired8 (3) 8 (5)0 8 (3) 8 (5)0
Student87 (36)9 (5)§ 78 (100)99 (41)21 (13)§78 (100)
Unemployed6 (2)6 (4)07 (3)7 (4)0
Country of residence — no. (%)
United States216 (89) 138 (83)78 (100)211 (88)133 (82)78 (100)
Canada28 (11)28 (17)0 30 (12) 30 (18)0
Interval since diagnosis of diabetes — yr15.2±12.520.2±12.24.7±3.115.4±12.0 20.2±11.75.4±3.7
Glycated hemoglobin — %8.3±0.5 8.3±0.58.3±0.68.3±0.58.3±0.58.3±0.5
Weight — kg71.9±25.380.8±15.9§ 49.0±17.973.0±21.8 85.1±18.5§51.6±19.3
Body-mass index¶25.3±6.027.4±4.420.2±3.825.6±5.628.4±5.720.6±4.5
* Plus–minus values are means ±SD. Percentages may not total 100 because of rounding.
† Race or ethnic group was self-reported.
‡ Patients could select more than one answer in this category.
§ P<0.05 for the comparison between the pump-therapy group and the injection-therapy group.
¶ The body-mass index is the weight in kilograms divided by the square of the height in meters.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Downloaded from www.nejm.org on June 30, 2010 . For personal use only. No other uses without permission.

Page 5

Sensor-Augmented Insulin-Pump Therapy
10.1056/nejmoa1002853 nejm.org
5
Primary and Secondary Outcomes
At 1 year, the baseline mean glycated hemoglo-
bin level (8.3% in the two study groups) had de-
creased to 7.5% in the pump-therapy group (ab-
solute reduction, 0.8±0.8 percentage points), as
compared with 8.1% in the injection-therapy group
(absolute reduction, 0.2±0.9 percentage points), for
a between-group difference in the pump-therapy
group of –0.6 percentage points (95% confidence
interval [CI], –0.7 to 0.4; P<0.001) (Fig. 1). Among
adults, the absolute reduction in the mean gly-
cated hemoglobin level was 1.0±0.7 percentage
points in the pump-therapy group and 0.4±0.8
percentage points in the injection-therapy group,
for a between-group difference in the pump-ther-
apy group of –0.6 percentage points (95% CI, –0.8
to 0.4; P<0.001). Among children, there was an
absolute reduction in glycated hemoglobin of
0.4±0.9 percentage points in the pump-therapy
group and an increase of 0.2±1.0 percentage points
in the injection-therapy group, for a between-group
difference favoring the pump-therapy group of –0.5
percentage points (95% CI, –0.8 to –0.2; P<0.001),
with adjustment for the statistical model.
Post hoc analyses that included only data from
observed patients and used multiple imputation
of missing values yielded similar results (Table 1
in the Supplementary Appendix). In both adults
and children in the pump-therapy group, glycated
hemoglobin levels fell rapidly from baseline to
3 months and remained lower than levels in the
injection-therapy group for the remainder of the
study (Fig. 1). An increased frequency of sensor
use was associated with a greater reduction in
glycated hemoglobin levels at 1 year (P = 0.003 with
adjustment for the baseline glycated hemoglobin
level) (Fig. 2).
The numbers of patients who reached a gly-
cated hemoglobin value of 7% or less were 67 of
244 patients (27%) in the pump-therapy group and
23 of 241 patients (10%) in the injection-therapy
group (P<0.001); these numbers included 57 of
166 adults (34%) in the pump-therapy group and
19 of 163 adults (12%) in the injection-therapy
group (P<0.001) and 10 of 78 children (13%) in the
pump-therapy group and 4 of 78 children (5%)
in the injection-therapy group (P = 0.15) (Fig. 3).
In a post hoc analysis that used glycated hemo-
globin targets recommended by the American Dia-
betes Association for children between the ages
of 6 and 12 years (<8%) and adolescents between
the ages of 13 and 19 years (<7.5%),10 a total of
35 of the 80 children and adolescents (44%) in the
pump-therapy group and 16 of the 80 (20%) in
the injection-therapy group reached these targets
at 1 year (P = 0.005). Among adults, weight in-
creased by 2.4 kg in the pump-therapy group and
by 1.8 kg in the injection-therapy group (P = 0.19).
Glycated Hemoglobin (%)
9.0
8.0
8.5
7.5
7.0
0.0
0369
12
Month
B Age ≥19 Yr
9.0
A All Patients
Glycated Hemoglobin (%)
8.0
8.5
7.5
7.0
0.0
0369
12
Month
C Age 7–18 Yr
9.0
Glycated Hemoglobin (%)
8.0
8.5
7.5
7.0
0.0
0
369
12
Month
Injection therapy
Injection therapy
Sensor-augmented
pump therapy
Sensor-augmented
pump therapy
Sensor-augmented
pump therapy
Injection therapy
*
*
*
*
*
*
*
*
*
*
*
*
Figure 1. Glycated Hemoglobin Levels at 3, 6, 9, and 12 Months
in All Patients and in Subgroups According to Age.
Values are means ±SE. Asterisks denote P<0.001 for all comparisons be-
tween pump therapy and injection therapy at each time point.
Copyright © 2010 Massachusetts Medical Society. All rights reserved.
Downloaded from www.nejm.org on June 30, 2010 . For personal use only. No other uses without permission.