The

n e w e ng l a n d j o u r na l

of

m e dic i n e

original article

Long-Term Effects of Intensive Glucose

Lowering on Cardiovascular Outcomes
The ACCORD Study Group*

A bs t r ac t
Background
The members of the writing group
(Hertzel C. Gerstein, M.D., McMaster University and Hamilton Health Sciences,
Hamilton, ON, Canada; Michael E. Miller,
Ph.D., Wake Forest University School of
Medicine, Winston-Salem, NC; Saul Genuth, M.D., Case Western Reserve University, Cleveland; Faramarz Ismail-Beigi,
M.D., Ph.D., Case Western Reserve University, Cleveland; John B. Buse, M.D.,
Ph.D., University of North Carolina, Chapel Hill; David C. Goff, Jr., M.D., Ph.D.,
Wake Forest University School of Medicine, Winston-Salem, NC; Jeffrey L. Probstfield, M.D., University of Washington,
Seattle; William C. Cushman, M.D.,
Memphis Veterans Affairs Medical Center, Memphis; Henry N. Ginsberg, M.D.,
Columbia University College of Physicians and Surgeons, New York; J. Thomas Bigger, M.D., Columbia University College of Physicians and Surgeons, New
York; Richard H. Grimm, Jr., M.D., Ph.D,
University of Minnesota, Berman Center
for Outcomes and Clinical Research,
Minneapolis; Robert P. Byington, Ph.D.,
Wake Forest University School of Medicine, Winston-Salem, NC; Yves D. Rosenberg, M.D., National Heart, Lung, and
Blood Institute, Bethesda, MD; and William T. Friedewald, M.D., Columbia University College of Physicians and Surgeons, New York) assume responsibility
for the content of this article. Address
reprint requests to Dr. Gerstein at McMaster University, Department of Medicine, HSC 3V38, 1200 Main St. W., Hamilton, ON L8N 3Z5, Canada, or at gerstein@
mcmaster.ca.
* Members of the Action to Control Cardiovascular Risk in Diabetes (ACCORD)
Study Group are listed in the Supplementary Appendix, available at NEJM.org.

Intensive glucose lowering has previously been shown to increase mortality among
persons with advanced type 2 diabetes and a high risk of cardiovascular disease. This
report describes the 5-year outcomes of a mean of 3.7 years of intensive glucose lowering on mortality and key cardiovascular events.
Methods

We randomly assigned participants with type 2 diabetes and cardiovascular disease

or additional cardiovascular risk factors to receive intensive therapy (targeting a
glycated hemoglobin level below 6.0%) or standard therapy (targeting a level of 7 to
7.9%). After termination of the intensive therapy, due to higher mortality in the
intensive-therapy group, the target glycated hemoglobin level was 7 to 7.9% for all
participants, who were followed until the planned end of the trial.
Results

Before the intensive therapy was terminated, the intensive-therapy group did not
differ significantly from the standard-therapy group in the rate of the primary outcome (a composite of nonfatal myocardial infarction, nonfatal stroke, or death from
cardiovascular causes) (P=0.13) but had more deaths from any cause (primarily cardiovascular) (hazard ratio, 1.21; 95% confidence interval [CI], 1.02 to 1.44) and fewer
nonfatal myocardial infarctions (hazard ratio, 0.79; 95% CI, 0.66 to 0.95). These
trends persisted during the entire follow-up period (hazard ratio for death, 1.19;
95% CI, 1.03 to 1.38; and hazard ratio for nonfatal myocardial infarction, 0.82; 95%
CI, 0.70 to 0.96). After the intensive intervention was terminated, the median glycated hemoglobin level in the intensive-therapy group rose from 6.4% to 7.2%, and
the use of glucose-lowering medications and rates of severe hypoglycemia and other
adverse events were similar in the two groups.
Conclusions

As compared with standard therapy, the use of intensive therapy for 3.7 years to target
a glycated hemoglobin level below 6% reduced 5-year nonfatal myocardial infarctions but increased 5-year mortality. Such a strategy cannot be recommended for
high-risk patients with advanced type 2 diabetes. (Funded by the National Heart,
Lung and Blood Institute; ClinicalTrials.gov number, NCT00000620.)

N Engl J Med 2011;364:818-28.

Copyright 2011 Massachusetts Medical Society.

818

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

intensive glucose lowering and cardiovascular events

ype 2 diabetes mellitus is a strong,

independent risk factor for cardiovascular
disease and death,1 and many epidemiologic analyses have identified a progressive relationship between hyperglycemia and these outcomes.2-5 The Action to Control Cardiovascular
Risk in Diabetes (ACCORD) trial was designed to
determine whether a strategy of targeting normal glycated hemoglobin levels (i.e., <6.0%) would
reduce the risk of serious cardiovascular events
in middle-aged and elderly people with type 2
diabetes mellitus, glycated hemoglobin levels of
7.5% or more, and additional cardiovascular risk
factors.6 However, on the basis of a mean of 3.5
years worth of data, the independent data and
safety monitoring board recommended termination of the intensive glucose-lowering regimen
because of the finding of higher mortality in the
intensive-therapy group. Therefore, we applied the
approaches that were used in the standard control
group to participants assigned to the intensivetherapy group, for up to 17 months of additional
follow-up. We report the clinical outcomes at
5 years of follow-up in response to a mean of 3.7
years of an intensive glycemia strategy.

Termination of Intensive Regimen

and Assessment of Outcomes

Recruitment occurred in two phases, from January to June 2001 and from February 2003 to October 2005. On February 5, 2008, participants were
informed of the decision to discontinue the intensive glucose-lowering regimen, after a mean treatment period of 3.7 years. Participants in the intensive-therapy group subsequently were switched
to standard glycemic therapy, and their target
glycated hemoglobin level of less than 6% was
changed to a target level of 7 to 7.9%. Since participants had also been assigned to receive treatment either to control lipid levels or to lower
blood pressure,8,9 they continued to be followed
at least every 4 months until the originally
planned end of the trial (June 2009). Thus, data
on clinical outcomes, including the primary outcome (a composite of nonfatal myocardial infarction, nonfatal stroke, or death from cardiovascular causes) and death from any cause (a secondary
outcome), continued to be collected and adjudicated for an additional 17 months by a central
committee whose members were unaware of studygroup assignments.
Intervention Effects

Me thods
Study Design

The design and major results of the trial have

been published previously.6,7 Briefly, we recruited
male and female volunteers from 77 clinical centers in the United States and Canada. The participants were 40 to 79 years of age, had type 2
diabetes mellitus and a glycated hemoglobin level of 7.5% or more, and had previous evidence of
cardiovascular disease or risk factors for cardiovascular disease. Participants were randomly assigned to receive either intensive glucose-lowering therapy targeting a glycated hemoglobin level
of less than 6.0% or standard glucose-lowering
therapy targeting a level of 7 to 7.9%. All participants received counseling about lifestyle and education about the management of diabetes. Glucose-lowering drugs were chosen from a common
formulary according to the participants studygroup assignment and response to therapy.7 Glycated hemoglobin levels were audited regularly
according to treatment group and study center, and
feedback was provided to facilitate the attainment of the target glycated hemoglobin levels.

The effects of the glycemic intervention during a

mean of 3.5 years (until December 10, 2007),
which provided the basis for the data and safety
monitoring boards recommendation to discontinue the intensive regimen, have been reported
previously.6 Here we report the effect of the intervention during an additional 0.2 years (i.e., until
February 5, 2008), which was when participants
were informed of the change in approach. We also
report on outcomes that occurred before February 5, 2008, that were not reported to the coordinating center as of December 10, 2007. Therefore,
we used intention-to-treat analyses to report on
the effect of a mean of 3.7 years of an intensive
glycemic intervention on cardiovascular disease,
followed by a mean of 1.2 years of standard glycemic therapy. Also reported are the effects of the
glycemic intervention until the transition date and
until the end of the overall trial for both the
blood-pressure and lipid trials. The treatment effects of the lipid and blood-pressure interventions were reported separately.8,9 All primary and
secondary outcomes were adjudicated centrally
by two adjudicators who were unaware of treat-

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

819

The

n e w e ng l a n d j o u r na l

ment-group assignments, and in addition, deaths

were reviewed by two diabetes experts (who were
unaware of treatment-group assignments) to determine whether they were due to hypoglycemia.
The ACCORD trial was sponsored by the National Heart, Lung, and Blood Institute (NHLBI), and
the protocol (available with the full text of this
article at NEJM.org) was approved by an NHLBI
review panel and by the ethics committee at each
center. All participants provided written informed
consent. All authors vouch for the accuracy and
completeness of the reported data. The donors of
medications and devices had no role in the study
design, data accrual and analysis, or manuscript
preparation.
Statistical Analysis

All statistical analyses were conducted at the coordinating center with the use of S-Plus software,
version 8.0 (Insightful), or SAS software, version
9.2 (SAS Institute). Baseline characteristics of the
participants were summarized with the use of
means, standard deviations, and percentages. Median glycated hemoglobin levels were calculated
monthly (by calendar month) to show the effect
of the switch from intensive therapy to the standard approach. Exposure to glucose-lowering
drugs was summarized as the number of participants who were prescribed a medication at the last
visit before the transition date and at the trial
termination. The incidence of key safety outcomes
was expressed as the percentage of events per
follow-up year, taking into account censoring of
follow-up data. KaplanMeier estimates were used
to calculate the percentage of participants who had
an event during follow-up.
Primary and secondary outcomes were analyzed with the use of Cox proportional-hazards
regression analyses according to the intention-totreat principle, and between-group comparisons
of the outcomes were performed with the use of
hazard ratios and 95% confidence intervals derived from these models. These analyses were
performed for events occurring from randomization until the date of transition (February 5,
2008) and from randomization until the final
visit (between the beginning of March and the
end of June 2009). An additional post hoc analysis was performed for the primary outcome and
death from any cause with the use of data from
the post-transition phase only.
For analyses of outcomes, data from participants without final follow-up data were censored
820

of

m e dic i n e

as of the time of their last completed 4-month

visit in both the intensive-therapy and standardtherapy groups. Data on mortality for participants
in the United States who were not followed for
the full follow-up period and who were not
known to be deceased were censored as of the
most recent date they were known to be alive or
January 1, 2008, on the basis of the National
Death Index.
Silent myocardial infarctions were identified
on the basis of electrocardiograms obtained every
2 years and were considered to have occurred at
the midpoint of the dates between the electrocardiogram showing a new myocardial infarction and
the previous electrocardiogram. Information from
electrocardiograms obtained after the transition
date were not known to the data and safety monitoring board or investigators at the time of the
transition. Therefore, new silent myocardial infarctions detected after the transition date that would
have been assigned to the period before transition were deemed to have occurred on the date of
transition; this occurred for 29 participants.
Cox models for the primary outcome contained a term representing study-group assignments plus terms accounting for the following
prespecified stratifying variables: assignment to
the blood-pressure trial or lipid trial; assignment
to the intensive blood-pressure intervention in the
blood-pressure trial; assignment to receive fibrate
in the lipid trial; the seven clinical center networks; and the presence or absence of previous
cardiovascular disease. For all secondary outcomes, an a priori decision was made to drop the
clinical center networks from this model, because
fewer events were expected than for the primary
outcomes. The consistency of the effect of the
study-group assignment on death from any cause
and on the primary outcome in the blood-pressure trial and the lipid trial was assessed with the
use of statistical tests of interactions between the
treatment effect and the subgroup within the Cox
models.
Unless otherwise indicated, nominal P values,
unadjusted for the multiple tests performed for
this report or for monitoring by the data and
safety monitoring board, are reported. Since we
conducted 46 statistical tests of hypotheses related to secondary end points and subgroups,
there was a 91% chance (i.e., 1-[1-0.05]46) that at
least one of these tests would be significant at an
alpha level of 0.05, assuming independence between tests.

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

intensive glucose lowering and cardiovascular events

The effect of the study-group assignment on

the primary outcome or mortality after the transition in participants who had not had a primary
outcome and who were alive at the transition date
was explored with the use of KaplanMeier curves
and Cox regression models. Further post hoc exploratory analyses to identify factors associated
with higher mortality in the intensive-therapy
group have examined baseline characteristics,10
the achieved glycated hemoglobin level and the
rapidity of its decline,11 and hypoglycemic
events.12,13

R e sult s
Baseline characteristics of the participants have
been reported previously.6 Table 1 shows these
characteristics before the transition date and at
the final visit. Figure 1 in the Supplementary Appendix (available at NEJM.org) shows the completeness of follow-up in the two study groups.
Median glycated hemoglobin levels before the
transition date in the intensive-therapy and standard-therapy groups were 6.4% and 7.5%, respectively. After the transition date, therapy was relaxed (i.e., fewer drugs or lower doses were used)
for a particular indication at least as often in the
intensive-therapy group as in the standard-therapy
group. For example, at the first post-transition
visit, relaxation of therapy was indicated in 94%
of participants in the intensive-therapy group and
69% of those in the standard-therapy group. At
the final visit, median glycated hemoglobin levels
were 7.2% in the intensive-therapy group and 7.6%
in the standard-therapy group (Fig. 2 in the Supplementary Appendix). By the final visit, the numbers of participants who were receiving metformin, secretagogues, thiazolidinediones, insulin,
and combination therapy with insulin and oral
agents were similar in the two groups (Table 1 in
the Supplementary Appendix). Rates of severe
hypoglycemia and other adverse events within
the two groups were similar after the transition
(Table 2 in the Supplementary Appendix).
Figure 1 shows the incidence of the primary
outcome and death from any cause from randomization until the time of transition, from randomization until the end of the whole study, and
from the transition date until the termination of
the trial. Figure 2 shows the effect of intensive
glucose-lowering therapy on all the major outcomes, from randomization until the end of
the active treatment period and until the end of

the study. Before the transition, the incidence

of the primary outcome among the participants
in the intensive-therapy group was 2.0% per year,
as compared with an incidence of 2.2% per year
among the participants in the standard-therapy
group (hazard ratio, 0.90; 95% confidence interval [CI], 0.78 to 1.03; nominal P=0.132, and
P=0.134 after adjustment for repeat testing by the
data and safety monitoring board) and remained
nonsignificant throughout the entire period of
observation (hazard ratio, 0.91; 95% CI, 0.81 to
1.03; P=0.12).
The intensive therapy had different effects on
two of the key components of this primary outcome. At the time of the transition, the rate of
nonfatal myocardial infarction in the intensivetherapy group was lower than that in the standard-therapy group (1.08% vs. 1.35%; hazard ratio, 0.79; 95% CI, 0.66 to 0.95; P=0.01), but the
rate of death from cardiovascular causes was nonsignificantly higher (0.71% vs. 0.55%; hazard ratio, 1.27%; 95% CI, 0.99 to 1.63; P=0.07). These
divergent effects were retained at the end of the
study, with a rate of nonfatal myocardial infarction in the intensive-therapy group that was lower
than that in the standard-therapy group (1.18 vs.
1.42; hazard ratio, 0.82; 95% CI, 0.70 to 0.96;
P=0.01) and a rate of death from cardiovascular
causes that was higher (0.74 vs. 0.57; hazard ratio, 1.29; 95% CI, 1.04 to 1.60; P=0.02).
Finally, at the time of the transition, there was
a 21% higher rate of death from any cause in the
intensive-therapy group than in the standardtherapy group (1.42 vs. 1.16; 95% CI, 1.02 to 1.44;
nominal P=0.030 and P=0.036 after adjustment
for repeat testing by the data and safety monitoring board) and a 19% higher rate at the end of
the study (1.53 vs. 1.27; 95% CI, 1.03 to 1.38;
P=0.02) (Fig. 1 and 2). The causes of death are
listed in Table 2. There was no clear difference
between study groups in any other predefined
cardiovascular outcomes.
Table 3 lists the annual incidence of the primary and secondary outcomes in the two treatment groups after the transition date, and Figures
1C and 1F show the corresponding KaplanMeier
curves for the primary outcome and death from
any cause. Hazard ratios in the post-transition
period were not significantly different from those
in the pretransition period for either the primary
outcome (ratio of pretransition to post-transition
hazard ratios, 0.95; 95% CI, 0.72 to 1.26; P=0.72)
or death from any cause (ratio of pretransition

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

821

The

n e w e ng l a n d j o u r na l

of

m e dic i n e

Table 1. Characteristics of the Participants at Baseline, before the Transition, and after the Transition.*
Characteristic

Intensive Therapy
Baseline
(N=5057)

Mean weight kg

93.618.6

Standard Therapy

Pretransition Post-Transition
(N=5052)
(N=4429)
96.720.7

95.620.9

861 (17.1)

699 (15.8)

32.25.5

33.56.2

33.16.2

106.713.9

109.615.3

Systolic

136.216.9

Diastolic

74.810.6

Baseline
(N=5051)

93.719.8

93.820.1

426 (8.5)

452 (10.1)

32.25.5

32.55.9

32.56.0

109.115.3

106.813.8

107.514.8

107.714.9

127.717.2

129.317.4

136.417.2

128.517.4

128.516.5

67.310.8

68.310.5

75.010.7

68.110.8

67.710.1

8.31.1

6.61.0

7.41.2

8.31.1

7.71.1

7.81.2

8.1

6.4

7.2

8.1

7.5

7.6

174.955.9

117.045.0

143.455.4

175.756.5

153.453.4

154.955.2

104.933.9

90.834.0

87.933.1

104.933.8

90.734.5

87.933.7

Women

47.213.0

48.613.2

49.513.4

46.912.2

47.613.6

49.013.9

Men

38.49.5

40.110.8

40.410.4

38.89.7

39.610.8

40.511.2

183.441.9

164.842.3

163.142.1

183.241.6

167.244.1

164.042.2

160.3125.0

189.4148.8

175.5151.5

166.0114.8

Weight gain of >10 kg since baseline

no. (%)
Body-mass index
Waist circumference (cm)

93.618.6

Pretransition Post-Transition
(N=5052)
(N=4483)

Blood pressure mm Hg

Glycated hemoglobin %
Mean
Median
Fasting serum glucose mg/dl
Cholesterol g/dl
Low-density lipoprotein
High-density lipoprotein

Total
Median triglycerides mg/dl

190.9148.4 154.6102.0

Potassium mg/dl

4.50.4

4.40.4

4.40.4

4.50.7

4.40.5

4.40.5

Serum creatinine mg/dl

0.90.2

1.10.4

1.10.4

0.90.2

1.10.4

1.10.4

52 (1.0)

27 (0.6)

77 (1.5)

21 (0.5)

Alanine aminotransferase >3 times ULN

no. (%)

* Baseline measurements include all participants who underwent randomization. Pretransition measurements are the last measurements
made before the transition, on February 5, 2008, for participants with at least one measurement. Post-transition measurements are the last
measurements made for participants for whom measurements were made during the post-transition period. Plusminus values are means
SD except where otherwise noted. To convert the values for glucose to millimoles per liter, multiply by 0.05551. To convert the values for
cholesterol to millimoles per liter, multiply by 0.02586. To convert the values for triglycerides to millimoles per liter, multiply by 0.01129. To
convert the values for potassium to millimoles per liter, multiply by 0.2558. To convert the values for creatinine to micromoles per liter,
multiply by 88.4. ULN denotes upper limit of the normal range.
Body-mass index is the weight in kilograms divided by the square of the height in meters.

to post-transition hazard ratios, 1.06; 95% CI,

0.76 to 1.46; P=0.74). There was a possible difference in the effect of the intensive therapy on
the pretransition primary outcome among participants with a baseline glycated hemoglobin level
of 8% or less as compared with those with a
level of more than 8% (P=0.03 for interaction)
(Fig. 3 in the Supplementary Appendix).
A total of 4733 participants were randomly
assigned to receive either intensive or standard
therapy to lower their blood pressure, and 5518
participants were randomly assigned to a statin
plus either fenofibrate or placebo for control of
822

low-density lipoprotein cholesterol. No significant

interactions were noted between the glucoselowering study and the blood-pressure study for
the primary outcome, or between the glucoselowering study and the lipid study for either the
primary outcome or death from any cause. However, there was evidence of an interaction between
the intensive glucose-lowering group and the intensive blood-pressurelowering group with respect to death from any cause both before the
transition (P=0.03 for interaction) and at the end
of the trial (P=0.05 for interaction) (Fig. 4 in the
Supplementary Appendix). Before the transition,

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

0
0

Years since Randomization

0 1 2 3 4 5 6 7 8

Intensive

Years since Randomization

18
21

20

40

60

80

0 1 2 3 4 5 6 7 8

Standard

Intensive

Years since Randomization

10

20

20

40

60

80

4611
4552

Standard 4414
Intensive 4427

No. at Risk

20

40

60

80

Standard

Intensive

Standard

4197
4218

Years since Transition

10

Hazard ratio, 1.15 (95% CI, 0.871.51)

F Death from Any Cause after Transition

100

Intensive

Years since Transition

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.
Line

Combo

4-C

H/T

SIZE
4 col
broadside

JOB: 36409

ISSUE: 03-03-11

AUTHOR, PLEASE NOTE:

Figure has been redrawn and type has been reset.
Please check carefully.

TYPE:

Figure 1. KaplanMeier Curves for the Primary Outcome and Death

fromGerstein
Any Cause.
RETAKE:
1st
AUTHOR:
2nd causes. Panels A and D show the incidence rates from ranThe primary outcome was a composite of nonfatal myocardial infarction, nonfatal stroke, or death from cardiovascular
FIGURE: 1 of 2
3rd
domization until the time of transition, Panels B and E show the rates from randomization until the end of
the
trial,
and Panels C and F show the rates for the post-transition periRevised
od. Plots for the post-transition period (Panels C and F) are included
fortsdescriptive purposes only; they cannot be used to infer any effect of the intensive therapy in this period.
ARTIST:

Standard 5123 5017 5006 4918 4127 2494 842 477 266
Intensive 5128 5066 4992 4855 4053 2479 814 496 263

0 1 2 3 4 5 6 7 8

Standard

Intensive

100

No. at Risk

10

20

Standard 5123 5071 5006 3807 2217 528 518

Intensive 5128 5066 4992 3767 2190 551 539

20

40

60

80

100

Hazard ratio, 1.19 (95% CI, 1.031.38)

E Death from Any Cause until End of Study

No. at Risk

Participants with Events (%)

Hazard ratio, 1.21 (95% CI, 1.021.44)

Participants with Events (%)

D Death from Any Cause before Transition

Participants with Events (%)

10

Hazard ratio, 0.94 (95% CI, 0.741.21)

Standard 4742
Intensive 4690

14
19

10

Standard

100

No. at Risk

Years since Randomization

20

60

60

80

20

Standard 5123 4912 4729 4580 3774 2251 729 407 217
Intensive 5128 4911 4743 4594 3750 2277 734 457 239

0 1 2 3 4 5 6 7 8

Intensive

Standard

100

No. at Risk

10

20

C Primary Outcome after Transition

Standard 5123 4912 4729 3533 2001 457 436

Intensive 5128 4911 4743 3544 2001 498 483

20

40

60

80

100

Hazard ratio, 0.91 (95% CI, 0.811.03)

B Primary Outcome until End of Study

Participants with Events (%)

No. at Risk

Participants with Events (%)

Hazard ratio, 0.90 (95% CI, 0.781.03)

Participants with Events (%)

A Primary Outcome before Transition

intensive glucose lowering and cardiovascular events

823

The

Outcome

n e w e ng l a n d j o u r na l

Intensive

Standard

of

m e dic i n e

P Value for
Interaction

Hazard Ratio (95% CI)

no. of events (%)

Primary outcome
Before transition
Until end of study
Nonfatal myocardial infarction
Before transition
Until end of study
Nonfatal stroke
Before transition
Until end of study
Death from cardiovascular causes
Before transition
Until end of study
Death from any cause
Before transition
Until end of study
Primary outcome, revascularization,
or hospitalization for CHF
Before transition
Until end of study
Fatal myocardial infarction, nonfatal myocardial
infarction, or unstable angina
Before transition
Until end of study
Fatal or nonfatal stroke
Before transition
Until end of study
Fatal or nonfatal CHF
Before transition
Until end of study

380 (2.0)
503 (2.1)

414 (2.2)
543 (2.2)

0.90 (0.781.03)
0.91 (0.811.03)

0.13
0.12

207 (1.1)
287 (1.2)

257 (1.4)
344 (1.4)

0.79 (0.660.95)
0.82 (0.700.96)

0.01
0.01

72 (0.4)
82 (0.3)

72 (0.4)
94 (0.4)

0.99 (0.721.38)
0.87 (0.651.17)

0.98
0.35

140 (0.7)
187 (0.7)

109 (0.6)
144 (0.6)

1.27 (0.991.63)
1.29 (1.041.60)

0.07
0.02

283 (1.4)
391 (1.5)

232 (1.2)
327 (1.3)

1.21 (1.021.44)
1.19 (1.031.38)

0.03
0.02

931 (5.3)
1159 (5.2)

955 (5.4)
1229 (5.5)

0.96 (0.881.06)
0.93 (0.861.01)

0.43
0.09

439 (2.3)
580 (2.4)

490 (2.6)
627 (2.6)

0.88 (0.771.00)
0.90 (0.811.01)

0.05
0.08

78 (0.4)
91 (0.4)

80 (0.4)
106 (0.4)

0.97 (0.711.33)
0.86 (0.651.13)

0.85
0.27

189 (1.0)
232 (0.9)

158 (0.8)
212 (0.9)

1.19 (0.961.47)
1.09 (0.911.32)

0.11
0.35

0.50

1.00

2.00

Intensive Therapy Standard Therapy

Better
Better

Figure 2. Hazard Ratios for the Prespecified Primary and Secondary Outcomes.
The effect of intensive glucose-lowering therapy is shown from randomization until the time of transition and from randomization until
the end of the trial. Squares represent hazard ratios, and horizontal bars represent 95% confidence intervals. CHF denotes congestive
heart failure.

this interaction was characterized by a marginally higher mortality rate in the intensive glucoselowering group than in the standard glucoselowering group among participants also assigned
to the intensive blood-pressurelowering group
(hazard ratio, 1.45; 95% CI, 1.00 to 2.12; P=0.05)
but not among those also assigned to the standard blood-pressurelowering group (hazard ratio, 0.78; 95% CI, 0.52 to 1.18; P=0.24).

hemoglobin level of at least 7.5%, and who had a

high risk of cardiovascular disease. Our findings
indicate that in a high-risk population such as
this, a mean of 3.7 years of intensive therapy consisting of multiple glucose-lowering methods to
target normal glycated hemoglobin levels (i.e.,
below 6.0%) does not result in a significantly
lower number of major cardiovascular events after 5 years than does an approach that uses similar methods to target levels that are more typically achieved in persons in the United States and
Discussion
Canada (i.e., 7 to 7.9%). Indeed, the intensive apThe ACCORD trial involved persons who had had proach led to more deaths. Effects on the primary
diabetes for a median of 10 years, with a glycated outcome were similar during the 3.7-year glucose-

824

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

intensive glucose lowering and cardiovascular events

Table 2. Causes of Death.*

During Pretransition
Period

Cause of Death

Intensive
Therapy

Standard
Therapy

From Randomization
until End of Study
Intensive
Therapy

Standard
Therapy

number (percent)
Any

283 (5.5)

232 (4.5)

391 (7.6)

327 (6.4)

Cardiovascular disease
Unexpected or presumed cardiovascular disease

89 (1.7)

78 (1.5)

124 (2.4)

103 (2.0)

Fatal myocardial infarction

20 (0.4)

12 (0.2)

24 (0.5)

14 (0.3)

Fatal congestive heart failure

26 (0.5)

20 (0.4)

32 (0.6)

25 (0.5)

Fatal procedure for cardiovascular disease

11 (0.2)

5 (0.1)

14 (0.3)

7 (0.1)

4 (0.1)

12 (0.2)

6 (0.1)

18 (0.4)

Fatal arrhythmia
Fatal procedure for noncardiovascular disease

1 (<0.1)

4 (0.1)

Fatal stroke

9 (0.2)

12 (0.2)

Other cardiovascular disease

2 (<0.1)
13 (0.3)

4 (0.1)
17 (0.3)

8 (0.2)

10 (0.2)

11 (0.2)

10 (0.2)

Cancer

69 (1.3)

70 (1.4)

102 (2.0)

101 (2.0)

Condition other than cancer or cardiovascular disease

57 (1.1)

40 (0.8)

84 (1.6)

60 (1.2)

Undetermined

11 (0.2)

12 (0.2)

12 (0.2)

21 (0.4)

Identified through National Death Index

6 (0.1)

1 (<0.1)

6 (0.1)

1 (<0.1)

* Data within categories are not mutually exclusive, and persons who were classified as having more than one possible
cause of death are listed in the relevant categories.
This condition was a component of the outcome of fatal cardiovascular disease.

lowering period and the entire 5-year follow-up

period; effects on mortality also were similar
during the two periods. Similar effects on the
primary outcome and mortality were noted in
most of the predefined subgroups. The nominally
positive tests for interaction with respect to the
primary outcome and baseline glycated hemoglobin levels and with respect to death from any
cause and the blood-pressure intervention may
well have been due to chance, since a large number of statistical tests were performed. No inferences can be made about the effect of the intervention during the post-transition period, because
between-group differences during this period
alone are likely to have been driven by betweengroup differences in the characteristics of participants who survived and were followed during
this period.
Reasons for the higher mortality in the intensive-therapy group during the pretransition period remain unclear. Because of the equivalent
rates of hypoglycemia in the post-transition period, severe hypoglycemia cannot be implicated.

Additional analyses reported elsewhere12 also do

not implicate severe hypoglycemia. According to
other analyses, the degree of reduction in glycated hemoglobin levels cannot be implicated.11
Further analyses should explore possible explanations, such as the role of various drugs, drug
combinations, or drug interactions; weight gain;
the relatively short intervention period (3.7 years);
and the observed interaction between the bloodpressure and glycemia trials with respect to
mortality.
Strengths of our study include the randomized
trial design, large sample, wide variety of clinics,
frequent follow-up, high rate of complete followup, high rate of adherence to the study assignment, and adjudication of all events by a central
committee that was unaware of the study-group
assignments. The clinical relevance of the results is highlighted by the following facts: the
approach used commonly available drugs, glycemia was managed within the context of good
control of blood pressure and lipid levels, the
recruited participants were representative of many

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

825

The

n e w e ng l a n d j o u r na l

of

m e dic i n e

Table 3. Incident Event Rates after the Transition Date.*

Outcome

Intensive Therapy
no. of
patients (%)
123 (2.8)

Nonfatal myocardial infarction

Nonfatal stroke

Primary outcome

Standard Therapy

% per year

no. of
patients (%)

% per year

2.35

129 (2.9)

2.47

80 (1.8)

1.51

87 (2.0)

1.64

10 (0.2)

0.18

22 (0.5)

0.40

47 (1.0)

0.84

35 (0.7)

0.62

Secondary outcome

Death
Cardiovascular causes

108 (2.3)

1.92

95 (2.0)

1.67

Primary outcome, revascularization, or hospitalization

for heart failure

Any cause

227 (5.8)

4.97

274 (7.0)

6.04

Fatal or nonfatal myocardial infarction or unstable

angina

141 (3.2)

2.74

138 (3.2)

2.70

Fatal or nonfatal stroke

13 (0.3)

0.24

26 (0.6)

0.47

Fatal or nonfatal congestive heart failure

43 (1.0)

0.80

54 (1.2)

0.99

* Data are for descriptive purposes only and cannot be used to infer any effect of the intervention during the post-transition period alone; therefore, statistical tests are not included.
The primary outcome was a composite of nonfatal myocardial infarction, nonfatal stroke, or death from cardiovascular
causes.

people with diabetes who are currently receiving

care in ambulatory settings, and several organizations have recommended glycemic targets of
6.5% or lower.
These findings are most applicable to middleaged and older patients with a long duration of
diabetes, a high risk of cardiovascular disease,
and hyperglycemia and should be interpreted in
light of the specific features of the ACCORD trial.
For example, the ACCORD trial excluded people
with glycated hemoglobin levels below 7.5%.
Moderate heterogeneity with respect to subgroups predefined by the glycated hemoglobin
level at baseline (Fig. 3 in the Supplementary
Appendix) suggests that participants whose glycated hemoglobin level at baseline was 8% or
lower may have had a better response to therapy
than participants with higher glycated hemoglobin levels. Although this hypothesis is clearly
not proved by the ACCORD trial, it is supported
by a recent epidemiologic analysis of the cardiovascular effect of glucose lowering in a cohort of
people with type 2 diabetes.14
The ACCORD trial explicitly tested whether
targeting a glycated hemoglobin level below 6%
by means of a large menu of glucose-lowering
agents is superior to targeting a glycated hemoglobin level of 7 to 7.9%. Therefore, our findings
826

should be interpreted in relation to these therapies and target glycated hemoglobin levels. Furthermore, targeting normal glycated hemoglobin
levels (i.e., <6.0%) required the use of multiple
combinations of glucose-lowering medications
in ways that are not used in standard care. For
example, 42% of participants in the intensivetherapy group were receiving three or more
classes of oral agents, either alone (17%) or in
combination with insulin (25%), whereas such
combinations were used in 19% of the participants in the standard-therapy group (Table 1 in
the Supplementary Appendix). Whether these
unconventional combinations were responsible
for the results and whether similar findings
would have been observed with newer glucoselowering therapies, different drug combinations,
or different target glycated hemoglobin levels is
unknown.
Finally, people with newly diagnosed diabetes
may have a different response to intensive glucoselowering therapy. A large trial involving people
with newly diagnosed type 2 diabetes, in which
normal glucose levels were targeted and a median
glycated hemoglobin level of 7% (as opposed to
7.9%) was achieved, showed a neutral cardiovascular effect after 10 years but a reduced rate of myocardial infarction and death after 20 years.15

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

intensive glucose lowering and cardiovascular events

In summary, the results of the ACCORD trial

show that in persons who have a high risk of cardiovascular disease and suboptimally controlled,
long-standing diabetes, with good blood-pressure
and lipid control, an intensive therapeutic approach targeting normal glycated hemoglobin
levels with the use of multiple medications is
associated with higher mortality than is a standard approach targeting higher glycated hemoglobin levels. The higher risk of death from any
cause and from cardiovascular causes in the intensive-therapy group means that a therapeutic
approach that targets glycated hemoglobin levels
below 6% cannot be generally recommended in
this population. Thus, the results of the ACCORD
trial suggest a lower limit for glycemic targets,
achieved with the use of multiple combinations
of currently available approaches.
Supported by the National Heart, Lung, and Blood Institute
(contracts N01-HC-95178, N01-HC-95179, N01-HC-95180, N01HC-95181, N01-HC-95182, N01-HC-95183, N01-HC-95184,
IAA#Y1-HC-9035, and IAA#Y1-HC-1010), and partially supported by the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute on Aging, and the National
Eye Institute and by General Clinical Research Centers at many
sites; substudies within the ACCORD trial on cost-effectiveness
and health-related quality of life were supported by the Centers
for Disease Control and Prevention. The following companies
provided study medications, equipment, or supplies: Abbott
Laboratories, Amylin Pharmaceutical, AstraZeneca, Bayer
HealthCare, Closer Healthcare, GlaxoSmithKline, King Pharmaceuticals, Merck, Novartis, Novo Nordisk, Omron Healthcare,
Sanofi-Aventis, and Schering-Plough.
Dr. Bigger reports receiving consulting fees and travel support from Merck and Roche and patent fees and royalties from
the Massachusetts Institute of Technology for risk-stratification
software; Dr. Buse, receiving consulting fees from Novo Nordisk,
Amylin, Becton Dickinson, Eli Lilly, HoffmannLa Roche (now
Genentech), Glyco-Mark, Wyeth, Daiichi Sankyo, Bristol-Myers
Squibb, Bayhill Therapeutics, LipoScience, MannKind, Veritas,
MicroIslet, GlaxoSmithKline, Abbott, Exsulin, and GI Dynamics
References
1. Huxley R, Barzi F, Woodward M. Excess risk of fatal coronary heart disease
associated with diabetes in men and
women: meta-analysis of 37 prospective
cohort studies. BMJ 2006;332:73-8.
2. Selvin E, Coresh J, Golden SH, Brancati FL, Folsom AR, Steffes MW. Glycemic
control and coronary heart disease risk in
persons with and without diabetes: the
Atherosclerosis Risk in Communities
study. Arch Intern Med 2005;165:1910-6.
3. Levitan EB, Liu S, Stampfer MJ, et al.
HbA1C measured in stored erythrocytes
and mortality rate among middle-aged
and older women. Diabetologia 2008;51:
267-75.
4. Selvin E, Steffes MW, Zhu H, et al.
Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults.
N Engl J Med 2010;362:800-11.

and grant support from Amylin, Novo Nordisk, Medtronic, Eli

Lilly, Novartis, Tolerex, Osiris, Halozyme, Pfizer, HoffmannLa
Roche, InterKrin, Merck, Sanofi-Aventis, Dexcom, Johnson
& Johnson, Bristol-Myers Squibb, Fujisawa, and the American
Academy of Family Practice Foundation, holding stock in Insulet,
and providing expert testimony for Novo Nordisk; Dr. Cushman,
receiving consulting fees from Novartis, Takeda, Sanofi-Aventis,
Bristol-Myers Squibb, King, Daiichi-Sankyo, Gilead, Theravance,
Pharmacopeia, and Sciele and institutional grant support to the
Memphis Veterans Affairs Medical Center from Novartis, GlaxoSmithKline, and Merck; Dr. Genuth, receiving consulting fees
from Merck and Daiichi Sankyo and holding stock in Novartis
and Johnson & Johnson; Dr. Gerstein, receiving consulting fees
from Sanofi-Aventis, GlaxoSmithKline, Eli Lilly, Novo Nordisk,
AstraZeneca, Bristol-Myers Squibb, Roche, Medtronic, Merck,
Bayer, Bioavail, and Janssen-Ortho, institutional grant support
to McMaster University from Sanofi-Aventis, GlaxoSmithKline,
Novo Nordisk, Merck, Pronova, Roche, Eli Lilly, and Boehringer
Ingelheim, and lecture fees from Sanofi-Aventis, GlaxoSmithKline, Solvay, Boehringer Ingelheim, Servier, Bayer, Eli Lilly,
Novo Nordisk, and Takeda; Dr. Ginsberg, being a member of the
board of Merck and Schering-Plough and the global advisory
board of Bristol-Myers Squibb/AstraZeneca and receiving consulting fees from GlaxoSmithKline, Merck, Bristol-Myers
Squibb, AstraZeneca, Regeneron/Sanofi-Aventis, Abbott, Roche,
Isis/Genzyme, Novartis, and Pfizer, institutional grant support
to the Columbia University College of Physicians and Surgeons
from Merck, Roche, Isis/Genzyme, and AstraZeneca, and payment from Pfizer for development of an educational presentation; Dr. Goff, being a member of the data and safety monitoring board for Takeda and receiving institutional grant support
to the Wake Forest University School of Medicine from Merck;
Dr. Grimm, being a member of the board of Pfizer, receiving
consulting fees from Pfizer, Merck, and Novartis, personal and
institutional grants to the University of Minnesota from Pfizer,
Merck, and Novartis, lecture fees from Pfizer, Merck, Novartis,
Forest, Schering-Plough, and Takeda, and travel support from
Takeda and Roche, and attending the AstraZeneca symposium
at the Cleveland Clinic and investigator meetings for Merck,
Novartis, and Pfizer; Dr. Ismail-Beigi, receiving consulting fees
from Eli Lilly; Dr. Miller, receiving consulting fees from Roche;
and Dr. Probstfield, receiving institutional grant support to the
University of Washington School of Medicine from SanofiAventis, Boehringer Ingelheim, and Abbott. No other potential
conflict of interest relevant to this article was reported.
Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.

5. Gerstein HC, Swedberg K, Carlsson J,

et al. The hemoglobin A1c level as a progressive risk factor for cardiovascular
death, hospitalization for heart failure, or
death in patients with chronic heart failure: an analysis of the Candesartan in
Heart failure: Assessment of Reduction in
Mortality and Morbidity (CHARM) program. Arch Intern Med 2008;168:1699-704.
6. The Action to Control Cardiovascular
Risk in Diabetes Study Group. Effects of
intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;358:2545-59.
7. Gerstein HC, Riddle MC, Kendall DM,
et al. Glycemia treatment strategies in the
Action to Control Cardiovascular Risk in
Diabetes (ACCORD) trial. Am J Cardiol
2007;99(12A):34i-43i.
8. The ACCORD Study Group. Effects of
combination lipid therapy in type 2 diabe-

tes mellitus. N Engl J Med 2010;362:156374. [Erratum, N Engl J Med 2010;362:1748.]

9. Idem. Effects of intensive blood-pressure control in type 2 diabetes mellitus.
N Engl J Med 2010;362:1575-85.
10. Calles-Escandn J, Lovato LC, SimonsMorton DG, et al. Effect of intensive compared with standard glycemia treatment
strategies on mortality by baseline subgroup characteristics: the Action to Control
Cardiovascular Risk in Diabetes (ACCORD)
trial. Diabetes Care 2010;33:721-7.
11. Riddle MC, Ambrosius WT, Brillon
DJ, et al. Epidemiologic relationships between A1C and all-cause mortality during
a median 3.4-year follow-up of glycemic
treatment in the ACCORD trial. Diabetes
Care 2010;33:983-90.
12. Bonds DE, Miller ME, Bergenstal RM,
et al. The association between symptom-

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.

827

intensive glucose lowering and cardiovascular events

atic, severe hypoglycaemia and mortality

in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study.
BMJ 2010;340:b4909.
13. Miller ME, Bonds DE, Gerstein HC, et
al. The effects of baseline characteristics,
glycaemia treatment approach, and gly-

828

cated haemoglobin concentration on the

risk of severe hypoglycaemia: post hoc
epidemiological analysis of the ACCORD
study. BMJ 2010;340:b5444.
14. Greenfield S, Billimek J, Pellegrini F,
et al. Comorbidity affects the relationship
between glycemic control and cardiovas-

cular outcomes in diabetes: a cohort

study. Ann Intern Med 2009;151:854-60.
15. Holman RR, Paul SK, Bethel MA,
Matthews DR, Neil HA. 10-Year follow-up
of intensive glucose control in type 2 diabetes. N Engl J Med 2008;359:1577-89.
Copyright 2011 Massachusetts Medical Society.

n engl j med 364;9 nejm.org march 3, 2011

The New England Journal of Medicine

Downloaded from nejm.org on December 28, 2014. For personal use only. No other uses without permission.
Copyright 2011 Massachusetts Medical Society. All rights reserved.